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. The old version
59 remains supported, as it has a much smaller footprint.
61 * "machinectl pull-tar" and related call may now do verification of
62 downloaded images using SUSE-style .sha256 checksum files in addition
63 to the already existing support for validating using Ubuntu-style
66 * sd-bus gained support for a new sd_bus_message_appendv() call which
67 is va_list equivalent of sd_bus_message_append().
69 * sd-boot gained support for validating images using SHIM/MOK.
71 * The SMACK code learnt support for "onlycap".
73 * systemd-mount --umount is now much smarter in figuring out how to
74 properly unmount a device given its mount or device path.
76 * The code to call libnss_dns as a fallback from libnss_resolve when
77 the communication with systemd-resolved fails was removed. This
78 fallback was redundant and interfered with the [!UNAVAIL=return]
79 suffix. See nss-resolve(8) for the recommended configuration.
81 * systemd-logind may now be restarted without losing state. It stores
82 the file descriptors for devices it manages in the system manager
83 using the FDSTORE= mechanism. Please note that further changes in
84 other components may be required to make use of this (for example
85 Xorg has code to listen for stops of systemd-logind and terminate
86 itself when logind is stopped or restarted, in order to avoid using
87 stale file descriptors for graphical devices, which is now
88 counterproductive and must be reverted in order for restarts of
89 systemd-logind to be safe. See
90 https://cgit.freedesktop.org/xorg/xserver/commit/?id=dc48bd653c7e101.)
92 * All kernel install plugins are called with the environment variable
93 KERNEL_INSTALL_MACHINE_ID which is set to the machine ID given by
94 /etc/machine-id. If the file is missing or empty, the variable is
95 empty and BOOT_DIR_ABS is the path of a temporary directory which is
96 removed after all the plugins exit. So, if KERNEL_INSTALL_MACHINE_ID
97 is empty, all plugins should not put anything in BOOT_DIR_ABS.
99 Contributions from: Adrian Heine né Lang, Aggelos Avgerinos, Alexander
100 Kurtz, Alexandros Frantzis, Alexey Brodkin, Amir Pakdel, Amir Yalon,
101 Anchor Cat, Anthony Parsons, Bastien Nocera, Benjamin Gilbert, Charles
102 Plessy, Chris Chiu, Chris Lamb, Christian Brauner, Daniel Drake,
103 Danielle Church, Daniel Molkentin, Daniel Rusek, Daniel Wang, Davide
104 Cavalca, David Herrmann, David Michael, Dax Kelson, Dimitri John
105 Ledkov, Djalal Harouni, Dušan Kazik, Elias Probst, Evgeny Vereshchagin,
106 Federico Di Pierro, Felipe Sateler, Felix Zhang, Franck Bui, Gary
107 Tierney, George McCollister, Hans de Goede, hecke, Hendrik Westerberg,
108 Hristo Venev, Ian Wienand, Insun Pyo, Ivan Shapovalov, James Cowgill,
109 James Hemsing, Janne Heß, Jan Synacek, Jason Reeder, João Paulo Rechi
110 Vita, John Paul Adrian Glaubitz, Jörg Thalheim, Josef Andersson, Josef
111 Gajdusek, Julian Mehne, Kai Krakow, Krzysztof Jackiewicz, Lars
112 Karlitski, Lennart Poettering, Lluís Gili, Lucas Werkmeister, Lukáš
113 Nykrýn, Łukasz Stelmach, Mantas Mikulėnas, Marcin Bachry, Marcus
114 Cooper, Mark Stosberg, Martin Pitt, Matija Skala, Matt Clarkson,
115 Matthias Greiner, Matthijs van Duin, Max Resch, Michael Biebl, Michal
116 Koutný, Michal Sekletar, Michal Suchanek, Nate Clark, Nathaniel
117 R. Lewis, Neil Brown, Pascal S. de Kloe, Pat Riehecky, Patrik Flykt,
118 Paul Kocialkowski, Peter Hutterer, Philip Withnall, Piotr Szydełko,
119 Rafael Fontenelle, Ray Strode, Richard Maw, Roelf Wichertjes, Ronny
120 Chevalier, Sarang S. Dalal, Sjoerd Simons, slodki, Stefan Schweter,
121 Susant Sahani, Ted Wood, Thomas Blume, Thomas Haller, Thomas
122 H. P. Andersen, Timothée Ravier, Tobias Jungel, Tobias Stoeckmann, Tom
123 Gundersen, Tom Yan, Torstein Husebø, Umut Tezduyar Lindskog,
124 userwithuid, Waldemar Brodkorb, WaLyong Cho, Yusuke Nojima, Yu
125 Watanabe, Zbigniew Jędrzejewski-Szmek
131 * This version requires at least gperf 3.1 for building, 3.0 is not
134 * The "hybrid" control group mode has been modified to improve
135 compatibility with "legacy" cgroups-v1 setups. Specifically, the
136 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
137 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
138 cgroups-v1 hierarchy), the only externally visible change being that
139 the cgroups-v2 hierarchy is also mounted, to
140 /sys/fs/cgroup/unified. This should provide a large degree of
141 compatibility with "legacy" cgroups-v1, while taking benefit of the
142 better management capabilities of cgroups-v2.
144 * The default control group setup mode may be selected both a boot-time
145 via a set of kernel command line parameters (specifically:
146 systemd.unified_cgroup_hierarchy= and
147 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
148 default selected on the configure command line
149 (--with-default-hierarchy=). The upstream default is "hybrid"
150 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
151 this will change in a future systemd version to be "unified" (pure
152 cgroups-v2 mode). The third option for the compile time option is
153 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
154 distributions to default to "hybrid" mode for release distributions,
155 starting with v233. We recommend "unified" for development
156 distributions (specifically: distributions such as Fedora's rawhide)
157 as that's where things are headed in the long run. Use "legacy" for
158 greatest stability and compatibility only.
160 * Note one current limitation of "unified" and "hybrid" control group
161 setup modes: the kernel currently does not permit the systemd --user
162 instance (i.e. unprivileged code) to migrate processes between two
163 disconnected cgroup subtrees, even if both are managed and owned by
164 the user. This effectively means "systemd-run --user --scope" doesn't
165 work when invoked from outside of any "systemd --user" service or
166 scope. Specifically, it is not supported from session scopes. We are
167 working on fixing this in a future systemd version. (See #3388 for
168 further details about this.)
170 * DBus policy files are now installed into /usr rather than /etc. Make
171 sure your system has dbus >= 1.9.18 running before upgrading to this
172 version, or override the install path with --with-dbuspolicydir= .
174 * All python scripts shipped with systemd (specifically: the various
175 tests written in Python) now require Python 3.
177 * systemd unit tests can now run standalone (without the source or
178 build directories), and can be installed into /usr/lib/systemd/tests/
179 with 'make install-tests'.
181 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
182 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
185 * Support for the %c, %r, %R specifiers in unit files has been
186 removed. Specifiers are not supposed to be dependent on configuration
187 in the unit file itself (so that they resolve the same regardless
188 where used in the unit files), but these specifiers were influenced
189 by the Slice= option.
191 * The shell invoked by debug-shell.service now defaults to /bin/sh in
192 all cases. If distributions want to use a different shell for this
193 purpose (for example Fedora's /sbin/sushell) they need to specify
194 this explicitly at configure time using --with-debug-shell=.
196 * The confirmation spawn prompt has been reworked to offer the
199 (c)ontinue, proceed without asking anymore
200 (D)ump, show the state of the unit
201 (f)ail, don't execute the command and pretend it failed
203 (i)nfo, show a short summary of the unit
204 (j)obs, show jobs that are in progress
205 (s)kip, don't execute the command and pretend it succeeded
206 (y)es, execute the command
208 The 'n' choice for the confirmation spawn prompt has been removed,
209 because its meaning was confusing.
211 The prompt may now also be redirected to an alternative console by
212 specifying the console as parameter to systemd.confirm_spawn=.
214 * Services of Type=notify require a READY=1 notification to be sent
215 during startup. If no such message is sent, the service now fails,
216 even if the main process exited with a successful exit code.
218 * Services that fail to start up correctly now always have their
219 ExecStopPost= commands executed. Previously, they'd enter "failed"
220 state directly, without executing these commands.
222 * The option MulticastDNS= of network configuration files has acquired
223 an actual implementation. With MulticastDNS=yes a host can resolve
224 names of remote hosts and reply to mDNS A and AAAA requests.
226 * When units are about to be started an additional check is now done to
227 ensure that all dependencies of type BindsTo= (when used in
228 combination with After=) have been started.
230 * systemd-analyze gained a new verb "syscall-filter" which shows which
231 system call groups are defined for the SystemCallFilter= unit file
232 setting, and which system calls they contain.
234 * A new system call filter group "@filesystem" has been added,
235 consisting of various file system related system calls. Group
236 "@reboot" has been added, covering reboot, kexec and shutdown related
237 calls. Finally, group "@swap" has been added covering swap
238 configuration related calls.
240 * A new unit file option RestrictNamespaces= has been added that may be
241 used to restrict access to the various process namespace types the
242 Linux kernel provides. Specifically, it may be used to take away the
243 right for a service unit to create additional file system, network,
244 user, and other namespaces. This sandboxing option is particularly
245 relevant due to the high amount of recently discovered namespacing
246 related vulnerabilities in the kernel.
248 * systemd-udev's .link files gained support for a new AutoNegotiation=
249 setting for configuring Ethernet auto-negotiation.
251 * systemd-networkd's .network files gained support for a new
252 ListenPort= setting in the [DHCP] section to explicitly configure the
253 UDP client port the DHCP client shall listen on.
255 * .network files gained a new Unmanaged= boolean setting for explicitly
256 excluding one or more interfaces from management by systemd-networkd.
258 * The systemd-networkd ProxyARP= option has been renamed to
259 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
260 renamed to ReduceARPProxy=. The old names continue to be available
263 * systemd-networkd gained support for configuring IPv6 Proxy NDP
264 addresses via the new IPv6ProxyNDPAddress= .network file setting.
266 * systemd-networkd's bonding device support gained support for two new
267 configuration options ActiveSlave= and PrimarySlave=.
269 * The various options in the [Match] section of .network files gained
270 support for negative matching.
272 * New systemd-specific mount options are now understood in /etc/fstab:
274 x-systemd.mount-timeout= may be used to configure the maximum
275 permitted runtime of the mount command.
277 x-systemd.device-bound may be set to bind a mount point to its
278 backing device unit, in order to automatically remove a mount point
279 if its backing device is unplugged. This option may also be
280 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
281 on the block device, which is now automatically set for all CDROM
282 drives, so that mounted CDs are automatically unmounted when they are
283 removed from the drive.
285 x-systemd.after= and x-systemd.before= may be used to explicitly
286 order a mount after or before another unit or mount point.
288 * Enqueued start jobs for device units are now automatically garbage
289 collected if there are no jobs waiting for them anymore.
291 * systemctl list-jobs gained two new switches: with --after, for every
292 queued job the jobs it's waiting for are shown; with --before the
293 jobs which it's blocking are shown.
295 * systemd-nspawn gained support for ephemeral boots from disk images
296 (or in other words: --ephemeral and --image= may now be
297 combined). Moreover, ephemeral boots are now supported for normal
298 directories, even if the backing file system is not btrfs. Of course,
299 if the file system does not support file system snapshots or
300 reflinks, the initial copy operation will be relatively expensive, but
301 this should still be suitable for many use cases.
303 * Calendar time specifications in .timer units now support
304 specifications relative to the end of a month by using "~" instead of
305 "-" as separator between month and day. For example, "*-02~03" means
306 "the third last day in February". In addition a new syntax for
307 repeated events has been added using the "/" character. For example,
308 "9..17/2:00" means "every two hours from 9am to 5pm".
310 * systemd-socket-proxyd gained a new parameter --connections-max= for
311 configuring the maximum number of concurrent connections.
313 * sd-id128 gained a new API for generating unique IDs for the host in a
314 way that does not leak the machine ID. Specifically,
315 sd_id128_get_machine_app_specific() derives an ID based on the
316 machine ID a in well-defined, non-reversible, stable way. This is
317 useful whenever an identifier for the host is needed but where the
318 identifier shall not be useful to identify the system beyond the
319 scope of the application itself. (Internally this uses HMAC-SHA256 as
320 keyed hash function using the machine ID as input.)
322 * NotifyAccess= gained a new supported value "exec". When set
323 notifications are accepted from all processes systemd itself invoked,
324 including all control processes.
326 * .nspawn files gained support for defining overlay mounts using the
327 Overlay= and OverlayReadOnly= options. Previously this functionality
328 was only available on the systemd-nspawn command line.
330 * systemd-nspawn's --bind= and --overlay= options gained support for
331 bind/overlay mounts whose source lies within the container tree by
332 prefixing the source path with "+".
334 * systemd-nspawn's --bind= and --overlay= options gained support for
335 automatically allocating a temporary source directory in /var/tmp
336 that is removed when the container dies. Specifically, if the source
337 directory is specified as empty string this mechanism is selected. An
338 example usage is --overlay=+/var::/var, which creates an overlay
339 mount based on the original /var contained in the image, overlayed
340 with a temporary directory in the host's /var/tmp. This way changes
341 to /var are automatically flushed when the container shuts down.
343 * systemd-nspawn --image= option does now permit raw file system block
344 devices (in addition to images containing partition tables, as
347 * The disk image dissection logic in systemd-nspawn gained support for
348 automatically setting up LUKS encrypted as well as Verity protected
349 partitions. When a container is booted from an encrypted image the
350 passphrase is queried at start-up time. When a container with Verity
351 data is started, the root hash is search in a ".roothash" file
352 accompanying the disk image (alternatively, pass the root hash via
353 the new --root-hash= command line option).
355 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
356 be used to dissect disk images the same way as systemd-nspawn does
357 it, following the Bootable Partition Specification. It may even be
358 used to mount disk images with complex partition setups (including
359 LUKS and Verity partitions) to a local host directory, in order to
360 inspect them. This tool is not considered public API (yet), and is
361 thus not installed into /usr/bin. Please do not rely on its
362 existence, since it might go away or be changed in later systemd
365 * A new generator "systemd-verity-generator" has been added, similar in
366 style to "systemd-cryptsetup-generator", permitting automatic setup of
367 Verity root partitions when systemd boots up. In order to make use of
368 this your partition setup should follow the Discoverable Partitions
369 Specification, and the GPT partition ID of the root file system
370 partition should be identical to the upper 128bit of the Verity root
371 hash. The GPT partition ID of the Verity partition protecting it
372 should be the lower 128bit of the Verity root hash. If the partition
373 image follows this model it is sufficient to specify a single
374 "roothash=" kernel command line argument to both configure which root
375 image and verity partition to use as well as the root hash for
376 it. Note that systemd-nspawn's Verity support follows the same
377 semantics, meaning that disk images with proper Verity data in place
378 may be booted in containers with systemd-nspawn as well as on
379 physical systems via the verity generator. Also note that the "mkosi"
380 tool available at https://github.com/systemd/mkosi has been updated
381 to generate Verity protected disk images following this scheme. In
382 fact, it has been updated to generate disk images that optionally
383 implement a complete UEFI SecureBoot trust chain, involving a signed
384 kernel and initrd image that incorporates such a root hash as well as
385 a Verity-enabled root partition.
387 * The hardware database (hwdb) udev supports has been updated to carry
388 accelerometer quirks.
390 * All system services are now run with a fresh kernel keyring set up
391 for them. The invocation ID is stored by default in it, thus
392 providing a safe, non-overridable way to determine the invocation
395 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
396 options for bind mounting arbitrary paths in a service-specific
397 way. When these options are used, arbitrary host or service files and
398 directories may be mounted to arbitrary locations in the service's
401 * Documentation has been added that lists all of systemd's low-level
402 environment variables:
404 https://github.com/systemd/systemd/blob/master/ENVIRONMENT.md
406 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
407 whether a specific socket file descriptor matches a specified socket
410 * systemd-firstboot has been updated to check for the
411 systemd.firstboot= kernel command line option. It accepts a boolean
412 and when set to false the first boot questions are skipped.
414 * systemd-fstab-generator has been updated to check for the
415 systemd.volatile= kernel command line option, which either takes an
416 optional boolean parameter or the special value "state". If used the
417 system may be booted in a "volatile" boot mode. Specifically,
418 "systemd.volatile" is used, the root directory will be mounted as
419 tmpfs, and only /usr is mounted from the actual root file system. If
420 "systemd.volatile=state" is used, the root directory will be mounted
421 as usual, but /var is mounted as tmpfs. This concept provides similar
422 functionality as systemd-nspawn's --volatile= option, but provides it
423 on physical boots. Use this option for implementing stateless
424 systems, or testing systems with all state and/or configuration reset
425 to the defaults. (Note though that many distributions are not
426 prepared to boot up without a populated /etc or /var, though.)
428 * systemd-gpt-auto-generator gained support for LUKS encrypted root
429 partitions. Previously it only supported LUKS encrypted partitions
430 for all other uses, except for the root partition itself.
432 * Socket units gained support for listening on AF_VSOCK sockets for
433 communication in virtualized QEMU environments.
435 * The "configure" script gained a new option --with-fallback-hostname=
436 for specifying the fallback hostname to use if none is configured in
437 /etc/hostname. For example, by specifying
438 --with-fallback-hostname=fedora it is possible to default to a
439 hostname of "fedora" on pristine installations.
441 * systemd-cgls gained support for a new --unit= switch for listing only
442 the control groups of a specific unit. Similar --user-unit= has been
443 added for listing only the control groups of a specific user unit.
445 * systemd-mount gained a new --umount switch for unmounting a mount or
446 automount point (and all mount/automount points below it).
448 * systemd will now refuse full configuration reloads (via systemctl
449 daemon-reload and related calls) unless at least 16MiB of free space
450 are available in /run. This is a safety precaution in order to ensure
451 that generators can safely operate after the reload completed.
453 * A new unit file option RootImage= has been added, which has a similar
454 effect as RootDirectory= but mounts the service's root directory from
455 a disk image instead of plain directory. This logic reuses the same
456 image dissection and mount logic that systemd-nspawn already uses,
457 and hence supports any disk images systemd-nspawn supports, including
458 those following the Discoverable Partition Specification, as well as
459 Verity enabled images. This option enables systemd to run system
460 services directly off disk images acting as resource bundles,
461 possibly even including full integrity data.
463 * A new MountAPIVFS= unit file option has been added, taking a boolean
464 argument. If enabled /proc, /sys and /dev (collectively called the
465 "API VFS") will be mounted for the service. This is only relevant if
466 RootDirectory= or RootImage= is used for the service, as these mounts
467 are of course in place in the host mount namespace anyway.
469 * systemd-nspawn gained support for a new --pivot-root= switch. If
470 specified the root directory within the container image is pivoted to
471 the specified mount point, while the original root disk is moved to a
472 different place. This option enables booting of ostree images
473 directly with systemd-nspawn.
475 * The systemd build scripts will no longer complain if the NTP server
476 addresses are not changed from the defaults. Google now supports
477 these NTP servers officially. We still recommend downstreams to
478 properly register an NTP pool with the NTP pool project though.
480 * coredumpctl gained a new "--reverse" option for printing the list
481 of coredumps in reverse order.
483 * coredumpctl will now show additional information about truncated and
484 inaccessible coredumps, as well as coredumps that are still being
485 processed. It also gained a new --quiet switch for suppressing
486 additional informational message in its output.
488 * coredumpctl gained support for only showing coredumps newer and/or
489 older than specific timestamps, using the new --since= and --until=
490 options, reminiscent of journalctl's options by the same name.
492 * The systemd-coredump logic has been improved so that it may be reused
493 to collect backtraces in non-compiled languages, for example in
494 scripting languages such as Python.
496 * machinectl will now show the UID shift of local containers, if user
497 namespacing is enabled for them.
499 * systemd will now optionally run "environment generator" binaries at
500 configuration load time. They may be used to add environment
501 variables to the environment block passed to services invoked. One
502 user environment generator is shipped by default that sets up
503 environment variables based on files dropped into /etc/environment.d
504 and ~/.config/environment.d/.
506 * systemd-resolved now includes the new, recently published 2017 DNSSEC
509 * hostnamed has been updated to report a new chassis type of
510 "convertible" to cover "foldable" laptops that can both act as a
511 tablet and as a laptop, such as various Lenovo Yoga devices.
513 Contributions from: Adrián López, Alexander Galanin, Alexander
514 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
515 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
516 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
517 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
518 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
519 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
520 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
521 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
522 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
523 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
524 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
525 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
526 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
527 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
528 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
529 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
530 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
531 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
532 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
533 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
534 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
535 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
536 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
537 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
538 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
539 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
540 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
547 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
548 RestrictAddressFamilies= enabled. These sandboxing options should
549 generally be compatible with the various external udev call-out
550 binaries we are aware of, however there may be exceptions, in
551 particular when exotic languages for these call-outs are used. In
552 this case, consider turning off these settings locally.
554 * The new RemoveIPC= option can be used to remove IPC objects owned by
555 the user or group of a service when that service exits.
557 * The new ProtectKernelModules= option can be used to disable explicit
558 load and unload operations of kernel modules by a service. In
559 addition access to /usr/lib/modules is removed if this option is set.
561 * ProtectSystem= option gained a new value "strict", which causes the
562 whole file system tree with the exception of /dev, /proc, and /sys,
563 to be remounted read-only for a service.
565 * The new ProtectKernelTunables= option can be used to disable
566 modification of configuration files in /sys and /proc by a service.
567 Various directories and files are remounted read-only, so access is
568 restricted even if the file permissions would allow it.
570 * The new ProtectControlGroups= option can be used to disable write
571 access by a service to /sys/fs/cgroup.
573 * Various systemd services have been hardened with
574 ProtectKernelTunables=yes, ProtectControlGroups=yes,
575 RestrictAddressFamilies=.
577 * Support for dynamically creating users for the lifetime of a service
578 has been added. If DynamicUser=yes is specified, user and group IDs
579 will be allocated from the range 61184..65519 for the lifetime of the
580 service. They can be resolved using the new nss-systemd.so NSS
581 module. The module must be enabled in /etc/nsswitch.conf. Services
582 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
583 any resources allocated by the service will be cleaned up when the
584 service exits. They also have ProtectHome=read-only and
585 ProtectSystem=strict enabled, so they are not able to make any
586 permanent modifications to the system.
588 * The nss-systemd module also always resolves root and nobody, making
589 it possible to have no /etc/passwd or /etc/group files in minimal
590 container or chroot environments.
592 * Services may be started with their own user namespace using the new
593 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
594 under which the service is running are mapped. All other users are
597 * Support for the cgroup namespace has been added to systemd-nspawn. If
598 supported by kernel, the container system started by systemd-nspawn
599 will have its own view of the cgroup hierarchy. This new behaviour
600 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
602 * The new MemorySwapMax= option can be used to limit the maximum swap
603 usage under the unified cgroup hierarchy.
605 * Support for the CPU controller in the unified cgroup hierarchy has
606 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
607 options. This controller requires out-of-tree patches for the kernel
608 and the support is provisional.
610 * Mount and automount units may now be created transiently
611 (i.e. dynamically at runtime via the bus API, instead of requiring
612 unit files in the file system).
614 * systemd-mount is a new tool which may mount file systems – much like
615 mount(8), optionally pulling in additional dependencies through
616 transient .mount and .automount units. For example, this tool
617 automatically runs fsck on a backing block device before mounting,
618 and allows the automount logic to be used dynamically from the
619 command line for establishing mount points. This tool is particularly
620 useful when dealing with removable media, as it will ensure fsck is
621 run – if necessary – before the first access and that the file system
622 is quickly unmounted after each access by utilizing the automount
623 logic. This maximizes the chance that the file system on the
624 removable media stays in a clean state, and if it isn't in a clean
625 state is fixed automatically.
627 * LazyUnmount=yes option for mount units has been added to expose the
628 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
631 * /efi will be used as the mount point of the EFI boot partition, if
632 the directory is present, and the mount point was not configured
633 through other means (e.g. fstab). If /efi directory does not exist,
634 /boot will be used as before. This makes it easier to automatically
635 mount the EFI partition on systems where /boot is used for something
638 * When operating on GPT disk images for containers, systemd-nspawn will
639 now mount the ESP to /boot or /efi according to the same rules as PID
640 1 running on a host. This allows tools like "bootctl" to operate
641 correctly within such containers, in order to make container images
642 bootable on physical systems.
644 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
646 * Two new user session targets have been added to support running
647 graphical sessions under the systemd --user instance:
648 graphical-session.target and graphical-session-pre.target. See
649 systemd.special(7) for a description of how those targets should be
652 * The vconsole initialization code has been significantly reworked to
653 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
654 support unicode keymaps. Font and keymap configuration will now be
655 copied to all allocated virtual consoles.
657 * FreeBSD's bhyve virtualization is now detected.
659 * Information recorded in the journal for core dumps now includes the
660 contents of /proc/mountinfo and the command line of the process at
661 the top of the process hierarchy (which is usually the init process
664 * systemd-journal-gatewayd learned the --directory= option to serve
665 files from the specified location.
667 * journalctl --root=… can be used to peruse the journal in the
668 /var/log/ directories inside of a container tree. This is similar to
669 the existing --machine= option, but does not require the container to
672 * The hardware database has been extended to support
673 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
676 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
677 specify the click rate for mice which include a horizontal wheel with
678 a click rate that is different than the one for the vertical wheel.
680 * systemd-run gained a new --wait option that makes service execution
681 synchronous. (Specifically, the command will not return until the
682 specified service binary exited.)
684 * systemctl gained a new --wait option that causes the start command to
685 wait until the units being started have terminated again.
687 * A new journal output mode "short-full" has been added which displays
688 timestamps with abbreviated English day names and adds a timezone
689 suffix. Those timestamps include more information than the default
690 "short" output mode, and can be passed directly to journalctl's
691 --since= and --until= options.
693 * /etc/resolv.conf will be bind-mounted into containers started by
694 systemd-nspawn, if possible, so any changes to resolv.conf contents
695 are automatically propagated to the container.
697 * The number of instances for socket-activated services originating
698 from a single IP address can be limited with
699 MaxConnectionsPerSource=, extending the existing setting of
702 * systemd-networkd gained support for vcan ("Virtual CAN") interface
705 * .netdev and .network configuration can now be extended through
708 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
709 Segmentation Offload, Generic Receive Offload, Large Receive Offload
710 can be enabled and disabled using the new UDPSegmentationOffload=,
711 TCPSegmentationOffload=, GenericSegmentationOffload=,
712 GenericReceiveOffload=, LargeReceiveOffload= options in the
713 [Link] section of .link files.
715 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
716 Port VLAN ID can be configured for bridge devices using the new STP=,
717 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
718 section of .netdev files.
720 * The route table to which routes received over DHCP or RA should be
721 added can be configured with the new RouteTable= option in the [DHCP]
722 and [IPv6AcceptRA] sections of .network files.
724 * The Address Resolution Protocol can be disabled on links managed by
725 systemd-networkd using the ARP=no setting in the [Link] section of
728 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
729 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
730 encode information about the result and exit codes of the current
731 service runtime cycle.
733 * systemd-sysctl will now configure kernel parameters in the order
734 they occur in the configuration files. This matches what sysctl
735 has been traditionally doing.
737 * kernel-install "plugins" that are executed to perform various
738 tasks after a new kernel is added and before an old one is removed
739 can now return a special value to terminate the procedure and
740 prevent any later plugins from running.
742 * Journald's SplitMode=login setting has been deprecated. It has been
743 removed from documentation, and its use is discouraged. In a future
744 release it will be completely removed, and made equivalent to current
745 default of SplitMode=uid.
747 * Storage=both option setting in /etc/systemd/coredump.conf has been
748 removed. With fast LZ4 compression storing the core dump twice is not
751 * The --share-system systemd-nspawn option has been replaced with an
752 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
753 this functionality is discouraged. In addition the variables
754 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
755 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
756 individual namespaces.
758 * "machinectl list" now shows the IP address of running containers in
759 the output, as well as OS release information.
761 * "loginctl list" now shows the TTY of each session in the output.
763 * sd-bus gained new API calls sd_bus_track_set_recursive(),
764 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
765 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
766 tracking objects in a "recursive" mode, where a single client can be
767 counted multiple times, if it takes multiple references.
769 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
770 sd_bus_get_exit_on_disconnect(). They may be used to to make a
771 process using sd-bus automatically exit if the bus connection is
774 * Bus clients of the service manager may now "pin" loaded units into
775 memory, by taking an explicit reference on them. This is useful to
776 ensure the client can retrieve runtime data about the service even
777 after the service completed execution. Taking such a reference is
778 available only for privileged clients and should be helpful to watch
779 running services in a race-free manner, and in particular collect
780 information about exit statuses and results.
782 * The nss-resolve module has been changed to strictly return UNAVAIL
783 when communication via D-Bus with resolved failed, and NOTFOUND when
784 a lookup completed but was negative. This means it is now possible to
785 neatly configure fallbacks using nsswitch.conf result checking
786 expressions. Taking benefit of this, the new recommended
787 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
789 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
791 * A new setting CtrlAltDelBurstAction= has been added to
792 /etc/systemd/system.conf which may be used to configure the precise
793 behaviour if the user on the console presses Ctrl-Alt-Del more often
794 than 7 times in 2s. Previously this would unconditionally result in
795 an expedited, immediate reboot. With this new setting the precise
796 operation may be configured in more detail, and also turned off
799 * In .netdev files two new settings RemoteChecksumTx= and
800 RemoteChecksumRx= are now understood that permit configuring the
801 remote checksumming logic for VXLAN networks.
803 * The service manager learnt a new "invocation ID" concept for invoked
804 services. Each runtime cycle of a service will get a new invocation
805 ID (a 128bit random UUID) assigned that identifies the current
806 run of the service uniquely and globally. A new invocation ID
807 is generated each time a service starts up. The journal will store
808 the invocation ID of a service along with any logged messages, thus
809 making the invocation ID useful for matching the online runtime of a
810 service with the offline log data it generated in a safe way without
811 relying on synchronized timestamps. In many ways this new service
812 invocation ID concept is similar to the kernel's boot ID concept that
813 uniquely and globally identifies the runtime of each boot. The
814 invocation ID of a service is passed to the service itself via an
815 environment variable ($INVOCATION_ID). A new bus call
816 GetUnitByInvocationID() has been added that is similar to GetUnit()
817 but instead of retrieving the bus path for a unit by its name
818 retrieves it by its invocation ID. The returned path is valid only as
819 long as the passed invocation ID is current.
821 * systemd-resolved gained a new "DNSStubListener" setting in
822 resolved.conf. It either takes a boolean value or the special values
823 "udp" and "tcp", and configures whether to enable the stub DNS
824 listener on 127.0.0.53:53.
826 * IP addresses configured via networkd may now carry additional
827 configuration settings supported by the kernel. New options include:
828 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
829 PrefixRoute=, AutoJoin=.
831 * The PAM configuration fragment file for "user@.service" shipped with
832 systemd (i.e. the --user instance of systemd) has been stripped to
833 the minimum necessary to make the system boot. Previously, it
834 contained Fedora-specific stanzas that did not apply to other
835 distributions. It is expected that downstream distributions add
836 additional configuration lines, matching their needs to this file,
837 using it only as rough template of what systemd itself needs. Note
838 that this reduced fragment does not even include an invocation of
839 pam_limits which most distributions probably want to add, even though
840 systemd itself does not need it. (There's also the new build time
841 option --with-pamconfdir=no to disable installation of the PAM
844 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
845 capability is now also dropped from its set (in addition to
846 CAP_SYS_MKNOD as before).
848 * In service unit files it is now possible to connect a specific named
849 file descriptor with stdin/stdout/stdout of an executed service. The
850 name may be specified in matching .socket units using the
851 FileDescriptorName= setting.
853 * A number of journal settings may now be configured on the kernel
854 command line. Specifically, the following options are now understood:
855 systemd.journald.max_level_console=,
856 systemd.journald.max_level_store=,
857 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
858 systemd.journald.max_level_wall=.
860 * "systemctl is-enabled --full" will now show by which symlinks a unit
861 file is enabled in the unit dependency tree.
863 * Support for VeraCrypt encrypted partitions has been added to the
864 "cryptsetup" logic and /etc/crypttab.
866 * systemd-detect-virt gained support for a new --private-users switch
867 that checks whether the invoking processes are running inside a user
868 namespace. Similar, a new special value "private-users" for the
869 existing ConditionVirtualization= setting has been added, permitting
870 skipping of specific units in user namespace environments.
872 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
873 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
874 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
875 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
876 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
877 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
878 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
879 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
880 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
881 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
882 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
883 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
884 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
885 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
886 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
887 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
888 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
889 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
890 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
891 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
892 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
893 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
894 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
895 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
896 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
897 Jędrzejewski-Szmek, Zeal Jagannatha
899 — Santa Fe, 2016-11-03
903 * In service units the various ExecXYZ= settings have been extended
904 with an additional special character as first argument of the
905 assigned value: if the character '+' is used the specified command
906 line it will be run with full privileges, regardless of User=,
907 Group=, CapabilityBoundingSet= and similar options. The effect is
908 similar to the existing PermissionsStartOnly= option, but allows
909 configuration of this concept for each executed command line
912 * Services may now alter the service watchdog timeout at runtime by
913 sending a WATCHDOG_USEC= message via sd_notify().
915 * MemoryLimit= and related unit settings now optionally take percentage
916 specifications. The percentage is taken relative to the amount of
917 physical memory in the system (or in case of containers, the assigned
918 amount of memory). This allows scaling service resources neatly with
919 the amount of RAM available on the system. Similarly, systemd-logind's
920 RuntimeDirectorySize= option now also optionally takes percentage
923 * In similar fashion TasksMax= takes percentage values now, too. The
924 value is taken relative to the configured maximum number of processes
925 on the system. The per-service task maximum has been changed to 15%
926 using this functionality. (Effectively this is an increase of 512 →
927 4915 for service units, given the kernel's default pid_max setting.)
929 * Calendar time specifications in .timer units now understand a ".."
930 syntax for time ranges. Example: "4..7:10" may now be used for
931 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
934 * The InaccessableDirectories=, ReadOnlyDirectories= and
935 ReadWriteDirectories= unit file settings have been renamed to
936 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
937 applied to all kinds of file nodes, and not just directories, with
938 the exception of symlinks. Specifically these settings may now be
939 used on block and character device nodes, UNIX sockets and FIFOS as
940 well as regular files. The old names of these settings remain
941 available for compatibility.
943 * systemd will now log about all service processes it kills forcibly
944 (using SIGKILL) because they remained after the clean shutdown phase
945 of the service completed. This should help identifying services that
946 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
947 systemd-logind's configuration a similar log message is generated for
948 processes killed at the end of each session due to this setting.
950 * systemd will now set the $JOURNAL_STREAM environment variable for all
951 services whose stdout/stderr are connected to the Journal (which
952 effectively means by default: all services). The variable contains
953 the device and inode number of the file descriptor used for
954 stdout/stderr. This may be used by invoked programs to detect whether
955 their stdout/stderr is connected to the Journal, in which case they
956 can switch over to direct Journal communication, thus being able to
957 pass extended, structured metadata along with their log messages. As
958 one example, this is now used by glib's logging primitives.
960 * When using systemd's default tmp.mount unit for /tmp, the mount point
961 will now be established with the "nosuid" and "nodev" options. This
962 avoids privilege escalation attacks that put traps and exploits into
963 /tmp. However, this might cause problems if you e. g. put container
964 images or overlays into /tmp; if you need this, override tmp.mount's
965 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
968 * systemd now supports the "memory" cgroup controller also on
971 * The systemd-cgtop tool now optionally takes a control group path as
972 command line argument. If specified, the control group list shown is
973 limited to subgroups of that group.
975 * The SystemCallFilter= unit file setting gained support for
976 pre-defined, named system call filter sets. For example
977 SystemCallFilter=@clock is now an effective way to make all clock
978 changing-related system calls unavailable to a service. A number of
979 similar pre-defined groups are defined. Writing system call filters
980 for system services is simplified substantially with this new
981 concept. Accordingly, all of systemd's own, long-running services now
982 enable system call filtering based on this, by default.
984 * A new service setting MemoryDenyWriteExecute= has been added, taking
985 a boolean value. If turned on, a service may no longer create memory
986 mappings that are writable and executable at the same time. This
987 enhances security for services where this is enabled as it becomes
988 harder to dynamically write and then execute memory in exploited
989 service processes. This option has been enabled for all of systemd's
990 own long-running services.
992 * A new RestrictRealtime= service setting has been added, taking a
993 boolean argument. If set the service's processes may no longer
994 acquire realtime scheduling. This improves security as realtime
995 scheduling may otherwise be used to easily freeze the system.
997 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
998 value. This may be used for requesting that the system manager inside
999 of the container reports start-up completion to nspawn which then
1000 propagates this notification further to the service manager
1001 supervising nspawn itself. A related option NotifyReady= in .nspawn
1002 files has been added too. This functionality allows ordering of the
1003 start-up of multiple containers using the usual systemd ordering
1006 * machinectl gained a new command "stop" that is an alias for
1009 * systemd-resolved gained support for contacting DNS servers on
1010 link-local IPv6 addresses.
1012 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
1013 its caches. A method call for requesting the same operation has been
1014 added to the bus API too, and is made available via "systemd-resolve
1017 * systemd-resolve gained a new --status switch. If passed a brief
1018 summary of the used DNS configuration with per-interface information
1021 * resolved.conf gained a new Cache= boolean option, defaulting to
1022 on. If turned off local DNS caching is disabled. This comes with a
1023 performance penalty in particular when DNSSEC is enabled. Note that
1024 resolved disables its internal caching implicitly anyway, when the
1025 configured DNS server is on a host-local IP address such as ::1 or
1026 127.0.0.1, thus automatically avoiding double local caching.
1028 * systemd-resolved now listens on the local IP address 127.0.0.53:53
1029 for DNS requests. This improves compatibility with local programs
1030 that do not use the libc NSS or systemd-resolved's bus APIs for name
1031 resolution. This minimal DNS service is only available to local
1032 programs and does not implement the full DNS protocol, but enough to
1033 cover local DNS clients. A new, static resolv.conf file, listing just
1034 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
1035 now recommended to make /etc/resolv.conf a symlink to this file in
1036 order to route all DNS lookups to systemd-resolved, regardless if
1037 done via NSS, the bus API or raw DNS packets. Note that this local
1038 DNS service is not as fully featured as the libc NSS or
1039 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
1040 used to deliver link-local address information (as this implies
1041 sending a local interface index along), LLMNR/mDNS support via this
1042 interface is severely restricted. It is thus strongly recommended for
1043 all applications to use the libc NSS API or native systemd-resolved
1046 * systemd-networkd's bridge support learned a new setting
1047 VLANFiltering= for controlling VLAN filtering. Moreover a new section
1048 in .network files has been added for configuring VLAN bridging in
1049 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
1051 * systemd-networkd's IPv6 Router Advertisement code now makes use of
1052 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
1053 now be acquired without relying on DHCPv6. Two new options
1054 UseDomains= and UseDNS= have been added to configure this behaviour.
1056 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
1057 renamed IPv6AcceptRA=, without altering its behaviour. The old
1058 setting name remains available for compatibility reasons.
1060 * The systemd-networkd VTI/VTI6 tunneling support gained new options
1061 Key=, InputKey= and OutputKey=.
1063 * systemd-networkd gained support for VRF ("Virtual Routing Function")
1064 interface configuration.
1066 * "systemctl edit" may now be used to create new unit files by
1067 specifying the --force switch.
1069 * sd-event gained a new function sd_event_get_iteration() for
1070 requesting the current iteration counter of the event loop. It starts
1071 at zero and is increased by one with each event loop iteration.
1073 * A new rpm macro %systemd_ordering is provided by the macros.systemd
1074 file. It can be used in lieu of %systemd_requires in packages which
1075 don't use any systemd functionality and are intended to be installed
1076 in minimal containers without systemd present. This macro provides
1077 ordering dependencies to ensure that if the package is installed in
1078 the same rpm transaction as systemd, systemd will be installed before
1079 the scriptlets for the package are executed, allowing unit presets
1082 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
1083 been added to simplify packaging of generators.
1085 * The os-release file gained VERSION_CODENAME field for the
1086 distribution nickname (e.g. VERSION_CODENAME=woody).
1088 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
1089 can be set to disable parsing of metadata and the creation
1090 of persistent symlinks for that device.
1092 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
1093 to make them available to logged-in users has been reverted.
1095 * Much of the common code of the various systemd components is now
1096 built into an internal shared library libsystemd-shared-231.so
1097 (incorporating the systemd version number in the name, to be updated
1098 with future releases) that the components link to. This should
1099 decrease systemd footprint both in memory during runtime and on
1100 disk. Note that the shared library is not for public use, and is
1101 neither API not ABI stable, but is likely to change with every new
1102 released update. Packagers need to make sure that binaries
1103 linking to libsystemd-shared.so are updated in step with the
1106 * Configuration for "mkosi" is now part of the systemd
1107 repository. mkosi is a tool to easily build legacy-free OS images,
1108 and is available on github: https://github.com/systemd/mkosi. If
1109 "mkosi" is invoked in the build tree a new raw OS image is generated
1110 incorporating the systemd sources currently being worked on and a
1111 clean, fresh distribution installation. The generated OS image may be
1112 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
1113 UEFI PC. This functionality is particularly useful to easily test
1114 local changes made to systemd in a pristine, defined environment. See
1115 HACKING for details.
1117 * configure learned the --with-support-url= option to specify the
1118 distribution's bugtracker.
1120 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
1121 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
1122 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
1123 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
1124 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
1125 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
1126 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
1127 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
1128 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
1129 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
1130 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
1131 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
1132 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
1133 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
1134 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
1135 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
1136 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
1137 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
1138 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
1140 — Berlin, 2016-07-25
1144 * DNSSEC is now turned on by default in systemd-resolved (in
1145 "allow-downgrade" mode), but may be turned off during compile time by
1146 passing "--with-default-dnssec=no" to "configure" (and of course,
1147 during runtime with DNSSEC= in resolved.conf). We recommend
1148 downstreams to leave this on at least during development cycles and
1149 report any issues with the DNSSEC logic upstream. We are very
1150 interested in collecting feedback about the DNSSEC validator and its
1151 limitations in the wild. Note however, that DNSSEC support is
1152 probably nothing downstreams should turn on in stable distros just
1153 yet, as it might create incompatibilities with a few DNS servers and
1154 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
1155 automatically whenever we detect such incompatible setups, but there
1156 might be systems we do not cover yet. Hence: please help us testing
1157 the DNSSEC code, leave this on where you can, report back, but then
1158 again don't consider turning this on in your stable, LTS or
1159 production release just yet. (Note that you have to enable
1160 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
1161 and its DNSSEC mode for host name resolution from local
1164 * systemd-resolve conveniently resolves DANE records with the --tlsa
1165 option and OPENPGPKEY records with the --openpgp option. It also
1166 supports dumping raw DNS record data via the new --raw= switch.
1168 * systemd-logind will now by default terminate user processes that are
1169 part of the user session scope unit (session-XX.scope) when the user
1170 logs out. This behavior is controlled by the KillUserProcesses=
1171 setting in logind.conf, and the previous default of "no" is now
1172 changed to "yes". This means that user sessions will be properly
1173 cleaned up after, but additional steps are necessary to allow
1174 intentionally long-running processes to survive logout.
1176 While the user is logged in at least once, user@.service is running,
1177 and any service that should survive the end of any individual login
1178 session can be started at a user service or scope using systemd-run.
1179 systemd-run(1) man page has been extended with an example which shows
1180 how to run screen in a scope unit underneath user@.service. The same
1181 command works for tmux.
1183 After the user logs out of all sessions, user@.service will be
1184 terminated too, by default, unless the user has "lingering" enabled.
1185 To effectively allow users to run long-term tasks even if they are
1186 logged out, lingering must be enabled for them. See loginctl(1) for
1187 details. The default polkit policy was modified to allow users to
1188 set lingering for themselves without authentication.
1190 Previous defaults can be restored at compile time by the
1191 --without-kill-user-processes option to "configure".
1193 * systemd-logind gained new configuration settings SessionsMax= and
1194 InhibitorsMax=, both with a default of 8192. It will not register new
1195 user sessions or inhibitors above this limit.
1197 * systemd-logind will now reload configuration on SIGHUP.
1199 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
1200 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
1201 enable. Also, support for the "io" cgroup controller in the unified
1202 hierarchy has been added, so that the "memory", "pids" and "io" are
1203 now the controllers that are supported on the unified hierarchy.
1205 WARNING: it is not possible to use previous systemd versions with
1206 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
1207 is necessary to also update systemd in the initramfs if using the
1208 unified hierarchy. An updated SELinux policy is also required.
1210 * LLDP support has been extended, and both passive (receive-only) and
1211 active (sender) modes are supported. Passive mode ("routers-only") is
1212 enabled by default in systemd-networkd. Active LLDP mode is enabled
1213 by default for containers on the internal network. The "networkctl
1214 lldp" command may be used to list information gathered. "networkctl
1215 status" will also show basic LLDP information on connected peers now.
1217 * The IAID and DUID unique identifier sent in DHCP requests may now be
1218 configured for the system and each .network file managed by
1219 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
1221 * systemd-networkd gained support for configuring proxy ARP support for
1222 each interface, via the ProxyArp= setting in .network files. It also
1223 gained support for configuring the multicast querier feature of
1224 bridge devices, via the new MulticastQuerier= setting in .netdev
1225 files. Similarly, snooping on the IGMP traffic can be controlled
1226 via the new setting MulticastSnooping=.
1228 A new setting PreferredLifetime= has been added for addresses
1229 configured in .network file to configure the lifetime intended for an
1232 The systemd-networkd DHCP server gained the option EmitRouter=, which
1233 defaults to yes, to configure whether the DHCP Option 3 (Router)
1236 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
1237 systemd-socket-activate and installed into /usr/bin. It is now fully
1240 * systemd-journald now uses separate threads to flush changes to disk
1241 when closing journal files, thus reducing impact of slow disk I/O on
1242 logging performance.
1244 * The sd-journal API gained two new calls
1245 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
1246 can be used to open journal files using file descriptors instead of
1247 file or directory paths. sd_journal_open_container() has been
1248 deprecated, sd_journal_open_directory_fd() should be used instead
1249 with the flag SD_JOURNAL_OS_ROOT.
1251 * journalctl learned a new output mode "-o short-unix" that outputs log
1252 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
1253 UTC). It also gained support for a new --no-hostname setting to
1254 suppress the hostname column in the family of "short" output modes.
1256 * systemd-ask-password now optionally skips printing of the password to
1257 stdout with --no-output which can be useful in scripts.
1259 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
1260 (devices tagged with ID_MAKER_TOOL) are now tagged with
1261 "uaccess" and are available to logged in users.
1263 * The DeviceAllow= unit setting now supports specifiers (with "%").
1265 * "systemctl show" gained a new --value switch, which allows print a
1266 only the contents of a specific unit property, without also printing
1267 the property's name. Similar support was added to "show*" verbs
1268 of loginctl and machinectl that output "key=value" lists.
1270 * A new unit type "generated" was added for files dynamically generated
1271 by generator tools. Similarly, a new unit type "transient" is used
1272 for unit files created using the runtime API. "systemctl enable" will
1273 refuse to operate on such files.
1275 * A new command "systemctl revert" has been added that may be used to
1276 revert to the vendor version of a unit file, in case local changes
1277 have been made by adding drop-ins or overriding the unit file.
1279 * "machinectl clean" gained a new verb to automatically remove all or
1280 just hidden container images.
1282 * systemd-tmpfiles gained support for a new line type "e" for emptying
1283 directories, if they exist, without creating them if they don't.
1285 * systemd-nspawn gained support for automatically patching the UID/GIDs
1286 of the owners and the ACLs of all files and directories in a
1287 container tree to match the UID/GID user namespacing range selected
1288 for the container invocation. This mode is enabled via the new
1289 --private-users-chown switch. It also gained support for
1290 automatically choosing a free, previously unused UID/GID range when
1291 starting a container, via the new --private-users=pick setting (which
1292 implies --private-users-chown). Together, these options for the first
1293 time make user namespacing for nspawn containers fully automatic and
1294 thus deployable. The systemd-nspawn@.service template unit file has
1295 been changed to use this functionality by default.
1297 * systemd-nspawn gained a new --network-zone= switch, that allows
1298 creating ad-hoc virtual Ethernet links between multiple containers,
1299 that only exist as long as at least one container referencing them is
1300 running. This allows easy connecting of multiple containers with a
1301 common link that implements an Ethernet broadcast domain. Each of
1302 these network "zones" may be named relatively freely by the user, and
1303 may be referenced by any number of containers, but each container may
1304 only reference one of these "zones". On the lower level, this is
1305 implemented by an automatically managed bridge network interface for
1306 each zone, that is created when the first container referencing its
1307 zone is created and removed when the last one referencing its zone
1310 * The default start timeout may now be configured on the kernel command
1311 line via systemd.default_timeout_start_sec=. It was already
1312 configurable via the DefaultTimeoutStartSec= option in
1313 /etc/systemd/system.conf.
1315 * Socket units gained a new TriggerLimitIntervalSec= and
1316 TriggerLimitBurst= setting to configure a limit on the activation
1317 rate of the socket unit.
1319 * The LimitNICE= setting now optionally takes normal UNIX nice values
1320 in addition to the raw integer limit value. If the specified
1321 parameter is prefixed with "+" or "-" and is in the range -20..19 the
1322 value is understood as UNIX nice value. If not prefixed like this it
1323 is understood as raw RLIMIT_NICE limit.
1325 * Note that the effect of the PrivateDevices= unit file setting changed
1326 slightly with this release: the per-device /dev file system will be
1327 mounted read-only from this version on, and will have "noexec"
1328 set. This (minor) change of behavior might cause some (exceptional)
1329 legacy software to break, when PrivateDevices=yes is set for its
1330 service. Please leave PrivateDevices= off if you run into problems
1333 * systemd-bootchart has been split out to a separate repository:
1334 https://github.com/systemd/systemd-bootchart
1336 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
1337 merged into the kernel in its current form.
1339 * The compatibility libraries libsystemd-daemon.so,
1340 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
1341 which have been deprecated since systemd-209 have been removed along
1342 with the corresponding pkg-config files. All symbols provided by
1343 those libraries are provided by libsystemd.so.
1345 * The Capabilities= unit file setting has been removed (it is ignored
1346 for backwards compatibility). AmbientCapabilities= and
1347 CapabilityBoundingSet= should be used instead.
1349 * A new special target has been added, initrd-root-device.target,
1350 which creates a synchronization point for dependencies of the root
1351 device in early userspace. Initramfs builders must ensure that this
1352 target is now included in early userspace.
1354 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
1355 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
1356 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
1357 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
1358 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
1359 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
1360 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
1361 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
1362 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
1363 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
1364 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
1365 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
1366 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
1367 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
1368 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
1369 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
1370 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
1371 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
1372 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
1373 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
1374 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
1375 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
1376 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
1377 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
1378 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
1381 — Fairfax, 2016-05-21
1385 * The systemd-resolved DNS resolver service has gained a substantial
1386 set of new features, most prominently it may now act as a DNSSEC
1387 validating stub resolver. DNSSEC mode is currently turned off by
1388 default, but is expected to be turned on by default in one of the
1389 next releases. For now, we invite everybody to test the DNSSEC logic
1390 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
1391 service also gained a full set of D-Bus interfaces, including calls
1392 to configure DNS and DNSSEC settings per link (for use by external
1393 network management software). systemd-resolved and systemd-networkd
1394 now distinguish between "search" and "routing" domains. The former
1395 are used to qualify single-label names, the latter are used purely
1396 for routing lookups within certain domains to specific links.
1397 resolved now also synthesizes RRs for all entries from /etc/hosts.
1399 * The systemd-resolve tool (which is a client utility for
1400 systemd-resolved) has been improved considerably and is now fully
1401 supported and documented. Hence it has moved from /usr/lib/systemd to
1404 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
1407 * The coredump collection logic has been reworked: when a coredump is
1408 collected it is now written to disk, compressed and processed
1409 (including stacktrace extraction) from a new instantiated service
1410 systemd-coredump@.service, instead of directly from the
1411 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
1412 processing large coredumps can take up a substantial amount of
1413 resources and time, and this previously happened entirely outside of
1414 systemd's service supervision. With the new logic the core_pattern
1415 hook only does minimal metadata collection before passing off control
1416 to the new instantiated service, which is configured with a time
1417 limit, a nice level and other settings to minimize negative impact on
1418 the rest of the system. Also note that the new logic will honour the
1419 RLIMIT_CORE setting of the crashed process, which now allows users
1420 and processes to turn off coredumping for their processes by setting
1423 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
1424 and all forked processes by default. Previously, PID 1 would leave
1425 the setting at "0" for all processes, as set by the kernel. Note that
1426 the resource limit traditionally has no effect on the generated
1427 coredumps on the system if the /proc/sys/kernel/core_pattern hook
1428 logic is used. Since the limit is now honoured (see above) its
1429 default has been changed so that the coredumping logic is enabled by
1430 default for all processes, while allowing specific opt-out.
1432 * When the stacktrace is extracted from processes of system users, this
1433 is now done as "systemd-coredump" user, in order to sandbox this
1434 potentially security sensitive parsing operation. (Note that when
1435 processing coredumps of normal users this is done under the user ID
1436 of process that crashed, as before.) Packagers should take notice
1437 that it is now necessary to create the "systemd-coredump" system user
1438 and group at package installation time.
1440 * The systemd-activate socket activation testing tool gained support
1441 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
1442 and --seqpacket switches. It also has been extended to support both
1443 new-style and inetd-style file descriptor passing. Use the new
1444 --inetd switch to request inetd-style file descriptor passing.
1446 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
1447 variable, which takes a boolean value. If set to false, ANSI color
1448 output is disabled in the tools even when run on a terminal that
1451 * The VXLAN support in networkd now supports two new settings
1452 DestinationPort= and PortRange=.
1454 * A new systemd.machine_id= kernel command line switch has been added,
1455 that may be used to set the machine ID in /etc/machine-id if it is
1456 not initialized yet. This command line option has no effect if the
1457 file is already initialized.
1459 * systemd-nspawn gained a new --as-pid2 switch that invokes any
1460 specified command line as PID 2 rather than PID 1 in the
1461 container. In this mode PID 1 is a minimal stub init process that
1462 implements the special POSIX and Linux semantics of PID 1 regarding
1463 signal and child process management. Note that this stub init process
1464 is implemented in nspawn itself and requires no support from the
1465 container image. This new logic is useful to support running
1466 arbitrary commands in the container, as normal processes are
1467 generally not prepared to run as PID 1.
1469 * systemd-nspawn gained a new --chdir= switch for setting the current
1470 working directory for the process started in the container.
1472 * "journalctl /dev/sda" will now output all kernel log messages for
1473 specified device from the current boot, in addition to all devices
1474 that are parents of it. This should make log output about devices
1475 pretty useful, as long as kernel drivers attach enough metadata to
1476 the log messages. (The usual SATA drivers do.)
1478 * The sd-journal API gained two new calls
1479 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
1480 that report whether log data from /run or /var has been found.
1482 * journalctl gained a new switch "--fields" that prints all journal
1483 record field names currently in use in the journal. This is backed
1484 by two new sd-journal API calls sd_journal_enumerate_fields() and
1485 sd_journal_restart_fields().
1487 * Most configurable timeouts in systemd now expect an argument of
1488 "infinity" to turn them off, instead of "0" as before. The semantics
1489 from now on is that a timeout of "0" means "now", and "infinity"
1490 means "never". To maintain backwards compatibility, "0" continues to
1491 turn off previously existing timeout settings.
1493 * "systemctl reload-or-try-restart" has been renamed to "systemctl
1494 try-reload-or-restart" to clarify what it actually does: the "try"
1495 logic applies to both reloading and restarting, not just restarting.
1496 The old name continues to be accepted for compatibility.
1498 * On boot-up, when PID 1 detects that the system clock is behind the
1499 release date of the systemd version in use, the clock is now set
1500 to the latter. Previously, this was already done in timesyncd, in order
1501 to avoid running with clocks set to the various clock epochs such as
1502 1902, 1938 or 1970. With this change the logic is now done in PID 1
1503 in addition to timesyncd during early boot-up, so that it is enforced
1504 before the first process is spawned by systemd. Note that the logic
1505 in timesyncd remains, as it is more comprehensive and ensures
1506 clock monotonicity by maintaining a persistent timestamp file in
1507 /var. Since /var is generally not available in earliest boot or the
1508 initrd, this part of the logic remains in timesyncd, and is not done
1511 * Support for tweaking details in net_cls.class_id through the
1512 NetClass= configuration directive has been removed, as the kernel
1513 people have decided to deprecate that controller in cgroup v2.
1514 Userspace tools such as nftables are moving over to setting rules
1515 that are specific to the full cgroup path of a task, which obsoletes
1516 these controllers anyway. The NetClass= directive is kept around for
1517 legacy compatibility reasons. For a more in-depth description of the
1518 kernel change, please refer to the respective upstream commit:
1520 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
1522 * A new service setting RuntimeMaxSec= has been added that may be used
1523 to specify a maximum runtime for a service. If the timeout is hit, the
1524 service is terminated and put into a failure state.
1526 * A new service setting AmbientCapabilities= has been added. It allows
1527 configuration of additional Linux process capabilities that are
1528 passed to the activated processes. This is only available on very
1531 * The process resource limit settings in service units may now be used
1532 to configure hard and soft limits individually.
1534 * The various libsystemd APIs such as sd-bus or sd-event now publicly
1535 expose support for gcc's __attribute__((cleanup())) C extension.
1536 Specifically, for many object destructor functions alternative
1537 versions have been added that have names suffixed with "p" and take a
1538 pointer to a pointer to the object to destroy, instead of just a
1539 pointer to the object itself. This is useful because these destructor
1540 functions may be used directly as parameters to the cleanup
1541 construct. Internally, systemd has been a heavy user of this GCC
1542 extension for a long time, and with this change similar support is
1543 now available to consumers of the library outside of systemd. Note
1544 that by using this extension in your sources compatibility with old
1545 and strictly ANSI compatible C compilers is lost. However, all gcc or
1546 LLVM versions of recent years support this extension.
1548 * Timer units gained support for a new setting RandomizedDelaySec= that
1549 allows configuring some additional randomized delay to the configured
1550 time. This is useful to spread out timer events to avoid load peaks in
1551 clusters or larger setups.
1553 * Calendar time specifications now support sub-second accuracy.
1555 * Socket units now support listening on SCTP and UDP-lite protocol
1558 * The sd-event API now comes with a full set of man pages.
1560 * Older versions of systemd contained experimental support for
1561 compressing journal files and coredumps with the LZ4 compressor that
1562 was not compatible with the lz4 binary (due to API limitations of the
1563 lz4 library). This support has been removed; only support for files
1564 compatible with the lz4 binary remains. This LZ4 logic is now
1565 officially supported and no longer considered experimental.
1567 * The dkr image import logic has been removed again from importd. dkr's
1568 micro-services focus doesn't fit into the machine image focus of
1569 importd, and quickly got out of date with the upstream dkr API.
1571 * Creation of the /run/lock/lockdev/ directory was dropped from
1572 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
1573 been available for many years. If you still need this, you need to
1574 create your own tmpfiles.d config file with:
1576 d /run/lock/lockdev 0775 root lock -
1578 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
1579 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
1580 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
1581 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
1582 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
1583 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
1584 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
1585 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
1586 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
1587 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
1588 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
1589 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
1590 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
1591 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
1592 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
1593 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
1594 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
1595 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
1596 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
1598 — Berlin, 2016-02-11
1602 * A number of properties previously only settable in unit
1603 files are now also available as properties to set when
1604 creating transient units programmatically via the bus, as it
1605 is exposed with systemd-run's --property=
1606 setting. Specifically, these are: SyslogIdentifier=,
1607 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
1608 EnvironmentFile=, ReadWriteDirectories=,
1609 ReadOnlyDirectories=, InaccessibleDirectories=,
1610 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
1612 * When creating transient services via the bus API it is now
1613 possible to pass in a set of file descriptors to use as
1614 STDIN/STDOUT/STDERR for the invoked process.
1616 * Slice units may now be created transiently via the bus APIs,
1617 similar to the way service and scope units may already be
1618 created transiently.
1620 * Wherever systemd expects a calendar timestamp specification
1621 (like in journalctl's --since= and --until= switches) UTC
1622 timestamps are now supported. Timestamps suffixed with "UTC"
1623 are now considered to be in Universal Time Coordinated
1624 instead of the local timezone. Also, timestamps may now
1625 optionally be specified with sub-second accuracy. Both of
1626 these additions also apply to recurring calendar event
1627 specification, such as OnCalendar= in timer units.
1629 * journalctl gained a new "--sync" switch that asks the
1630 journal daemon to write all so far unwritten log messages to
1631 disk and sync the files, before returning.
1633 * systemd-tmpfiles learned two new line types "q" and "Q" that
1634 operate like "v", but also set up a basic btrfs quota
1635 hierarchy when used on a btrfs file system with quota
1638 * tmpfiles' "v", "q" and "Q" will now create a plain directory
1639 instead of a subvolume (even on a btrfs file system) if the
1640 root directory is a plain directory, and not a
1641 subvolume. This should simplify things with certain chroot()
1642 environments which are not aware of the concept of btrfs
1645 * systemd-detect-virt gained a new --chroot switch to detect
1646 whether execution takes place in a chroot() environment.
1648 * CPUAffinity= now takes CPU index ranges in addition to
1651 * The various memory-related resource limit settings (such as
1652 LimitAS=) now understand the usual K, M, G, ... suffixes to
1653 the base of 1024 (IEC). Similar, the time-related resource
1654 limit settings understand the usual min, h, day, ...
1657 * There's a new system.conf setting DefaultTasksMax= to
1658 control the default TasksMax= setting for services and
1659 scopes running on the system. (TasksMax= is the primary
1660 setting that exposes the "pids" cgroup controller on systemd
1661 and was introduced in the previous systemd release.) The
1662 setting now defaults to 512, which means services that are
1663 not explicitly configured otherwise will only be able to
1664 create 512 processes or threads at maximum, from this
1665 version on. Note that this means that thread- or
1666 process-heavy services might need to be reconfigured to set
1667 TasksMax= to a higher value. It is sufficient to set
1668 TasksMax= in these specific unit files to a higher value, or
1669 even "infinity". Similar, there's now a logind.conf setting
1670 UserTasksMax= that defaults to 4096 and limits the total
1671 number of processes or tasks each user may own
1672 concurrently. nspawn containers also have the TasksMax=
1673 value set by default now, to 8192. Note that all of this
1674 only has an effect if the "pids" cgroup controller is
1675 enabled in the kernel. The general benefit of these changes
1676 should be a more robust and safer system, that provides a
1677 certain amount of per-service fork() bomb protection.
1679 * systemd-nspawn gained the new --network-veth-extra= switch
1680 to define additional and arbitrarily-named virtual Ethernet
1681 links between the host and the container.
1683 * A new service execution setting PassEnvironment= has been
1684 added that allows importing select environment variables
1685 from PID1's environment block into the environment block of
1688 * Timer units gained support for a new RemainAfterElapse=
1689 setting which takes a boolean argument. It defaults to on,
1690 exposing behaviour unchanged to previous releases. If set to
1691 off, timer units are unloaded after they elapsed if they
1692 cannot elapse again. This is particularly useful for
1693 transient timer units, which shall not stay around longer
1694 than until they first elapse.
1696 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
1697 default now (the kernel default is 16). This is beneficial
1698 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
1699 allows substantially larger numbers of queued
1700 datagrams. This should increase the capability of systemd to
1701 parallelize boot-up, as logging and sd_notify() are unlikely
1702 to stall execution anymore. If you need to change the value
1703 from the new defaults, use the usual sysctl.d/ snippets.
1705 * The compression framing format used by the journal or
1706 coredump processing has changed to be in line with what the
1707 official LZ4 tools generate. LZ4 compression support in
1708 systemd was considered unsupported previously, as the format
1709 was not compatible with the normal tools. With this release
1710 this has changed now, and it is hence safe for downstream
1711 distributions to turn it on. While not compressing as well
1712 as the XZ, LZ4 is substantially faster, which makes
1713 it a good default choice for the compression logic in the
1714 journal and in coredump handling.
1716 * Any reference to /etc/mtab has been dropped from
1717 systemd. The file has been obsolete since a while, but
1718 systemd refused to work on systems where it was incorrectly
1719 set up (it should be a symlink or non-existent). Please make
1720 sure to update to util-linux 2.27.1 or newer in conjunction
1721 with this systemd release, which also drops any reference to
1722 /etc/mtab. If you maintain a distribution make sure that no
1723 software you package still references it, as this is a
1724 likely source of bugs. There's also a glibc bug pending,
1725 asking for removal of any reference to this obsolete file:
1727 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
1729 Note that only util-linux versions built with
1730 --enable-libmount-force-mountinfo are supported.
1732 * Support for the ".snapshot" unit type has been removed. This
1733 feature turned out to be little useful and little used, and
1734 has now been removed from the core and from systemctl.
1736 * The dependency types RequiresOverridable= and
1737 RequisiteOverridable= have been removed from systemd. They
1738 have been used only very sparingly to our knowledge and
1739 other options that provide a similar effect (such as
1740 systemctl --mode=ignore-dependencies) are much more useful
1741 and commonly used. Moreover, they were only half-way
1742 implemented as the option to control behaviour regarding
1743 these dependencies was never added to systemctl. By removing
1744 these dependency types the execution engine becomes a bit
1745 simpler. Unit files that use these dependencies should be
1746 changed to use the non-Overridable dependency types
1747 instead. In fact, when parsing unit files with these
1748 options, that's what systemd will automatically convert them
1749 too, but it will also warn, asking users to fix the unit
1750 files accordingly. Removal of these dependency types should
1751 only affect a negligible number of unit files in the wild.
1753 * Behaviour of networkd's IPForward= option changed
1754 (again). It will no longer maintain a per-interface setting,
1755 but propagate one way from interfaces where this is enabled
1756 to the global kernel setting. The global setting will be
1757 enabled when requested by a network that is set up, but
1758 never be disabled again. This change was made to make sure
1759 IPv4 and IPv6 behaviour regarding packet forwarding is
1760 similar (as the Linux IPv6 stack does not support
1761 per-interface control of this setting) and to minimize
1764 * In unit files the behaviour of %u, %U, %h, %s has
1765 changed. These specifiers will now unconditionally resolve
1766 to the various user database fields of the user that the
1767 systemd instance is running as, instead of the user
1768 configured in the specific unit via User=. Note that this
1769 effectively doesn't change much, as resolving of these
1770 specifiers was already turned off in the --system instance
1771 of systemd, as we cannot do NSS lookups from PID 1. In the
1772 --user instance of systemd these specifiers where correctly
1773 resolved, but hardly made any sense, since the user instance
1774 lacks privileges to do user switches anyway, and User= is
1775 hence useless. Moreover, even in the --user instance of
1776 systemd behaviour was awkward as it would only take settings
1777 from User= assignment placed before the specifier into
1778 account. In order to unify and simplify the logic around
1779 this the specifiers will now always resolve to the
1780 credentials of the user invoking the manager (which in case
1781 of PID 1 is the root user).
1783 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
1784 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
1785 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
1786 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
1787 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
1788 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
1789 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
1790 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
1791 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
1792 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
1793 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
1794 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
1795 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
1796 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
1799 — Berlin, 2015-11-18
1803 * systemd now depends on util-linux v2.27. More specifically,
1804 the newly added mount monitor feature in libmount now
1805 replaces systemd's former own implementation.
1807 * libmount mandates /etc/mtab not to be regular file, and
1808 systemd now enforces this condition at early boot.
1809 /etc/mtab has been deprecated and warned about for a very
1810 long time, so systems running systemd should already have
1811 stopped having this file around as anything else than a
1812 symlink to /proc/self/mounts.
1814 * Support for the "pids" cgroup controller has been added. It
1815 allows accounting the number of tasks in a cgroup and
1816 enforcing limits on it. This adds two new setting
1817 TasksAccounting= and TasksMax= to each unit, as well as a
1818 global option DefaultTasksAccounting=.
1820 * Support for the "net_cls" cgroup controller has been added.
1821 It allows assigning a net class ID to each task in the
1822 cgroup, which can then be used in firewall rules and traffic
1823 shaping configurations. Note that the kernel netfilter net
1824 class code does not currently work reliably for ingress
1825 packets on unestablished sockets.
1827 This adds a new config directive called NetClass= to CGroup
1828 enabled units. Allowed values are positive numbers for fixed
1829 assignments and "auto" for picking a free value
1832 * 'systemctl is-system-running' now returns 'offline' if the
1833 system is not booted with systemd. This command can now be
1834 used as a substitute for 'systemd-notify --booted'.
1836 * Watchdog timeouts have been increased to 3 minutes for all
1837 in-tree service files. Apparently, disk IO issues are more
1838 frequent than we hoped, and user reported >1 minute waiting
1841 * 'machine-id-commit' functionality has been merged into
1842 'machine-id-setup --commit'. The separate binary has been
1845 * The WorkingDirectory= directive in unit files may now be set
1846 to the special value '~'. In this case, the working
1847 directory is set to the home directory of the user
1848 configured in User=.
1850 * "machinectl shell" will now open the shell in the home
1851 directory of the selected user by default.
1853 * The CrashChVT= configuration file setting is renamed to
1854 CrashChangeVT=, following our usual logic of not
1855 abbreviating unnecessarily. The old directive is still
1856 supported for compat reasons. Also, this directive now takes
1857 an integer value between 1 and 63, or a boolean value. The
1858 formerly supported '-1' value for disabling stays around for
1861 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
1862 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
1863 RootDirectory= properties can now be set for transient
1866 * The systemd-analyze tool gained a new "set-log-target" verb
1867 to change the logging target the system manager logs to
1868 dynamically during runtime. This is similar to how
1869 "systemd-analyze set-log-level" already changes the log
1872 * In nspawn /sys is now mounted as tmpfs, with only a selected
1873 set of subdirectories mounted in from the real sysfs. This
1874 enhances security slightly, and is useful for ensuring user
1875 namespaces work correctly.
1877 * Support for USB FunctionFS activation has been added. This
1878 allows implementation of USB gadget services that are
1879 activated as soon as they are requested, so that they don't
1880 have to run continuously, similar to classic socket
1883 * The "systemctl exit" command now optionally takes an
1884 additional parameter that sets the exit code to return from
1885 the systemd manager when exiting. This is only relevant when
1886 running the systemd user instance, or when running the
1887 system instance in a container.
1889 * sd-bus gained the new API calls sd_bus_path_encode_many()
1890 and sd_bus_path_decode_many() that allow easy encoding and
1891 decoding of multiple identifier strings inside a D-Bus
1892 object path. Another new call sd_bus_default_flush_close()
1893 has been added to flush and close per-thread default
1896 * systemd-cgtop gained support for a -M/--machine= switch to
1897 show the control groups within a certain container only.
1899 * "systemctl kill" gained support for an optional --fail
1900 switch. If specified the requested operation will fail of no
1901 processes have been killed, because the unit had no
1902 processes attached, or similar.
1904 * A new systemd.crash_reboot=1 kernel command line option has
1905 been added that triggers a reboot after crashing. This can
1906 also be set through CrashReboot= in systemd.conf.
1908 * The RuntimeDirectory= setting now understands unit
1909 specifiers like %i or %f.
1911 * A new (still internal) library API sd-ipv4acd has been added,
1912 that implements address conflict detection for IPv4. It's
1913 based on code from sd-ipv4ll, and will be useful for
1914 detecting DHCP address conflicts.
1916 * File descriptors passed during socket activation may now be
1917 named. A new API sd_listen_fds_with_names() is added to
1918 access the names. The default names may be overridden,
1919 either in the .socket file using the FileDescriptorName=
1920 parameter, or by passing FDNAME= when storing the file
1921 descriptors using sd_notify().
1923 * systemd-networkd gained support for:
1925 - Setting the IPv6 Router Advertisement settings via
1926 IPv6AcceptRouterAdvertisements= in .network files.
1928 - Configuring the HelloTimeSec=, MaxAgeSec= and
1929 ForwardDelaySec= bridge parameters in .netdev files.
1931 - Configuring PreferredSource= for static routes in
1934 * The "ask-password" framework used to query for LUKS harddisk
1935 passwords or SSL passwords during boot gained support for
1936 caching passwords in the kernel keyring, if it is
1937 available. This makes sure that the user only has to type in
1938 a passphrase once if there are multiple objects to unlock
1939 with the same one. Previously, such password caching was
1940 available only when Plymouth was used; this moves the
1941 caching logic into the systemd codebase itself. The
1942 "systemd-ask-password" utility gained a new --keyname=
1943 switch to control which kernel keyring key to use for
1944 caching a password in. This functionality is also useful for
1945 enabling display managers such as gdm to automatically
1946 unlock the user's GNOME keyring if its passphrase, the
1947 user's password and the harddisk password are the same, if
1948 gdm-autologin is used.
1950 * When downloading tar or raw images using "machinectl
1951 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
1952 file is now also downloaded, if it is available and stored
1953 next to the image file.
1955 * Units of type ".socket" gained a new boolean setting
1956 Writable= which is only useful in conjunction with
1957 ListenSpecial=. If true, enables opening the specified
1958 special file in O_RDWR mode rather than O_RDONLY mode.
1960 * systemd-rfkill has been reworked to become a singleton
1961 service that is activated through /dev/rfkill on each rfkill
1962 state change and saves the settings to disk. This way,
1963 systemd-rfkill is now compatible with devices that exist
1964 only intermittendly, and even restores state if the previous
1965 system shutdown was abrupt rather than clean.
1967 * The journal daemon gained support for vacuuming old journal
1968 files controlled by the number of files that shall remain,
1969 in addition to the already existing control by size and by
1970 date. This is useful as journal interleaving performance
1971 degrades with too many separate journal files, and allows
1972 putting an effective limit on them. The new setting defaults
1973 to 100, but this may be changed by setting SystemMaxFiles=
1974 and RuntimeMaxFiles= in journald.conf. Also, the
1975 "journalctl" tool gained the new --vacuum-files= switch to
1976 manually vacuum journal files to leave only the specified
1977 number of files in place.
1979 * udev will now create /dev/disk/by-path links for ATA devices
1980 on kernels where that is supported.
1982 * Galician, Serbian, Turkish and Korean translations were added.
1984 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
1985 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
1986 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
1987 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
1988 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
1989 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
1990 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
1991 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
1992 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
1993 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
1994 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
1995 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
1996 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
1997 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
1998 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
1999 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
2000 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
2001 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
2003 — Berlin, 2015-10-07
2007 * The DHCP implementation of systemd-networkd gained a set of
2010 - The DHCP server now supports emitting DNS and NTP
2011 information. It may be enabled and configured via
2012 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
2013 and NTP information is enabled, but no servers are
2014 configured, the corresponding uplink information (if there
2015 is any) is propagated.
2017 - Server and client now support transmission and reception
2018 of timezone information. It can be configured via the
2019 newly introduced network options UseTimezone=,
2020 EmitTimezone=, and Timezone=. Transmission of timezone
2021 information is enabled between host and containers by
2022 default now: the container will change its local timezone
2023 to what the host has set.
2025 - Lease timeouts can now be configured via
2026 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
2028 - The DHCP server improved on the stability of
2029 leases. Clients are more likely to get the same lease
2030 information back, even if the server loses state.
2032 - The DHCP server supports two new configuration options to
2033 control the lease address pool metrics, PoolOffset= and
2036 * The encapsulation limit of tunnels in systemd-networkd may
2037 now be configured via 'EncapsulationLimit='. It allows
2038 modifying the maximum additional levels of encapsulation
2039 that are permitted to be prepended to a packet.
2041 * systemd now supports the concept of user buses replacing
2042 session buses, if used with dbus-1.10 (and enabled via dbus
2043 --enable-user-session). It previously only supported this on
2044 kdbus-enabled systems, and this release expands this to
2045 'dbus-daemon' systems.
2047 * systemd-networkd now supports predictable interface names
2050 * systemd now optionally supports the new Linux kernel
2051 "unified" control group hierarchy. If enabled via the kernel
2052 command-line option 'systemd.unified_cgroup_hierarchy=1',
2053 systemd will try to mount the unified cgroup hierarchy
2054 directly on /sys/fs/cgroup. If not enabled, or not
2055 available, systemd will fall back to the legacy cgroup
2056 hierarchy setup, as before. Host system and containers can
2057 mix and match legacy and unified hierarchies as they
2058 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
2059 environment variable to individually select the hierarchy to
2060 use for executed containers. By default, nspawn will use the
2061 unified hierarchy for the containers if the host uses the
2062 unified hierarchy, and the legacy hierarchy otherwise.
2063 Please note that at this point the unified hierarchy is an
2064 experimental kernel feature and is likely to change in one
2065 of the next kernel releases. Therefore, it should not be
2066 enabled by default in downstream distributions yet. The
2067 minimum required kernel version for the unified hierarchy to
2068 work is 4.2. Note that when the unified hierarchy is used
2069 for the first time delegated access to controllers is
2070 safe. Because of this systemd-nspawn containers will get
2071 access to controllers now, as will systemd user
2072 sessions. This means containers and user sessions may now
2073 manage their own resources, partitioning up what the system
2076 * A new special scope unit "init.scope" has been introduced
2077 that encapsulates PID 1 of the system. It may be used to
2078 determine resource usage and enforce resource limits on PID
2079 1 itself. PID 1 hence moved out of the root of the control
2082 * The cgtop tool gained support for filtering out kernel
2083 threads when counting tasks in a control group. Also, the
2084 count of processes is now recursively summed up by
2085 default. Two options -k and --recursive= have been added to
2086 revert to old behaviour. The tool has also been updated to
2087 work correctly in containers now.
2089 * systemd-nspawn's --bind= and --bind-ro= options have been
2090 extended to allow creation of non-recursive bind mounts.
2092 * libsystemd gained two new calls sd_pid_get_cgroup() and
2093 sd_peer_get_cgroup() which return the control group path of
2094 a process or peer of a connected AF_UNIX socket. This
2095 function call is particularly useful when implementing
2096 delegated subtrees support in the control group hierarchy.
2098 * The "sd-event" event loop API of libsystemd now supports
2099 correct dequeuing of real-time signals, without losing
2102 * When systemd requests a PolicyKit decision when managing
2103 units it will now add additional fields to the request,
2104 including unit name and desired operation. This enables more
2105 powerful PolicyKit policies, that make decisions depending
2106 on these parameters.
2108 * nspawn learnt support for .nspawn settings files, that may
2109 accompany the image files or directories of containers, and
2110 may contain additional settings for the container. This is
2111 an alternative to configuring container parameters via the
2112 nspawn command line.
2114 Contributions from: Cristian Rodríguez, Daniel Mack, David
2115 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
2116 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
2117 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
2118 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
2119 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
2120 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
2121 Andersen, Tom Gundersen, Torstein Husebø
2123 — Berlin, 2015-09-08
2127 * machinectl gained a new verb 'shell' which opens a fresh
2128 shell on the target container or the host. It is similar to
2129 the existing 'login' command of machinectl, but spawns the
2130 shell directly without prompting for username or
2131 password. The pseudo machine '.host' now refers to the local
2132 host and is used by default. Hence, 'machinectl shell' can
2133 be used as replacement for 'su -' which spawns a session as
2134 a fresh systemd unit in a way that is fully isolated from
2135 the originating session.
2137 * systemd-networkd learned to cope with private-zone DHCP
2138 options and allows other programs to query the values.
2140 * SELinux access control when enabling/disabling units is no
2141 longer enforced with this release. The previous
2142 implementation was incorrect, and a new corrected
2143 implementation is not yet available. As unit file operations
2144 are still protected via PolicyKit and D-Bus policy this is
2145 not a security problem. Yet, distributions which care about
2146 optimal SELinux support should probably not stabilize on
2149 * sd-bus gained support for matches of type "arg0has=", that
2150 test for membership of strings in string arrays sent in bus
2153 * systemd-resolved now dumps the contents of its DNS and LLMNR
2154 caches to the logs on reception of the SIGUSR1 signal. This
2155 is useful to debug DNS behaviour.
2157 * The coredumpctl tool gained a new --directory= option to
2158 operate on journal files in a specific directory.
2160 * "systemctl reboot" and related commands gained a new
2161 "--message=" option which may be used to set a free-text
2162 wall message when shutting down or rebooting the
2163 system. This message is also logged, which is useful for
2164 figuring out the reason for a reboot or shutdown a
2167 * The "systemd-resolve-host" tool's -i switch now takes
2168 network interface numbers as alternative to interface names.
2170 * A new unit file setting for services has been introduced:
2171 UtmpMode= allows configuration of how precisely systemd
2172 handles utmp and wtmp entries for the service if this is
2173 enabled. This allows writing services that appear similar to
2174 user sessions in the output of the "w", "who", "last" and
2177 * systemd-resolved will now locally synthesize DNS resource
2178 records for the "localhost" and "gateway" domains as well as
2179 the local hostname. This should ensure that clients querying
2180 RRs via resolved will get similar results as those going via
2181 NSS, if nss-myhostname is enabled.
2183 Contributions from: Alastair Hughes, Alex Crawford, Daniel
2184 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
2185 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
2186 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
2187 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
2188 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
2189 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
2190 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
2191 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
2192 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
2193 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
2194 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2196 — Berlin, 2015-08-27
2200 * The systemd-efi-boot-generator functionality was merged into
2201 systemd-gpt-auto-generator.
2203 * systemd-networkd now supports Group Policy for vxlan
2204 devices. It can be enabled via the new boolean configuration
2205 option called 'GroupPolicyExtension='.
2207 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
2208 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
2209 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
2211 — Berlin, 2015-07-31
2215 * The python-systemd code has been removed from the systemd repository.
2216 A new repository has been created which accommodates the code from
2217 now on, and we kindly ask distributions to create a separate package
2218 for this: https://github.com/systemd/python-systemd
2220 * The systemd daemon will now reload its main configuration
2221 (/etc/systemd/system.conf) on daemon-reload.
2223 * sd-dhcp now exposes vendor specific extensions via
2224 sd_dhcp_lease_get_vendor_specific().
2226 * systemd-networkd gained a number of new configuration options.
2228 - A new boolean configuration option for TAP devices called
2229 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
2230 device, thus allowing to send and receive GSO packets.
2232 - A new tunnel configuration option called 'CopyDSCP='.
2233 If enabled, the DSCP field of ip6 tunnels is copied into the
2234 decapsulated packet.
2236 - A set of boolean bridge configuration options were added.
2237 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
2238 and 'UnicastFlood=' are now parsed by networkd and applied to the
2239 respective bridge link device via the respective IFLA_BRPORT_*
2242 - A new string configuration option to override the hostname sent
2243 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
2244 is true, networkd will use the configured hostname instead of the
2245 system hostname when sending DHCP requests.
2247 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
2248 networkd will configure the IPv6 flow-label of the tunnel device
2249 according to RFC2460.
2251 - The 'macvtap' virtual network devices are now supported, similar to
2252 the already supported 'macvlan' devices.
2254 * systemd-resolved now implements RFC5452 to improve resilience against
2255 cache poisoning. Additionally, source port randomization is enabled
2256 by default to further protect against DNS spoofing attacks.
2258 * nss-mymachines now supports translating UIDs and GIDs of running
2259 containers with user-namespaces enabled. If a container 'foo'
2260 translates a host uid 'UID' to the container uid 'TUID', then
2261 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
2262 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
2263 mapped as 'vg-foo-TGID'.
2265 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
2266 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
2267 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
2268 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
2269 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
2270 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
2271 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
2272 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
2273 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
2274 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
2276 — Berlin, 2015-07-29
2280 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
2281 There are no known issues with current sysfs, and udev does not need
2282 or should be used to work around such bugs.
2284 * udev does no longer enable USB HID power management. Several reports
2285 indicate, that some devices cannot handle that setting.
2287 * The udev accelerometer helper was removed. The functionality
2288 is now fully included in iio-sensor-proxy. But this means,
2289 older iio-sensor-proxy versions will no longer provide
2290 accelerometer/orientation data with this systemd version.
2291 Please upgrade iio-sensor-proxy to version 1.0.
2293 * networkd gained a new configuration option IPv6PrivacyExtensions=
2294 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
2295 for Stateless Address") on selected networks.
2297 * For the sake of fewer build-time dependencies and less code in the
2298 main repository, the python bindings are about to be removed in the
2299 next release. A new repository has been created which accommodates
2300 the code from now on, and we kindly ask distributions to create a
2301 separate package for this. The removal will take place in v223.
2303 https://github.com/systemd/python-systemd
2305 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
2306 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
2307 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
2308 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
2309 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
2310 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
2311 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
2312 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
2313 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
2314 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2316 — Berlin, 2015-07-07
2320 * The sd-bus.h and sd-event.h APIs have now been declared
2321 stable and have been added to the official interface of
2322 libsystemd.so. sd-bus implements an alternative D-Bus client
2323 library, that is relatively easy to use, very efficient and
2324 supports both classic D-Bus as well as kdbus as transport
2325 backend. sd-event is a generic event loop abstraction that
2326 is built around Linux epoll, but adds features such as event
2327 prioritization or efficient timer handling. Both APIs are good
2328 choices for C programs looking for a bus and/or event loop
2329 implementation that is minimal and does not have to be
2330 portable to other kernels.
2332 * kdbus support is no longer compile-time optional. It is now
2333 always built-in. However, it can still be disabled at
2334 runtime using the kdbus=0 kernel command line setting, and
2335 that setting may be changed to default to off, by specifying
2336 --disable-kdbus at build-time. Note though that the kernel
2337 command line setting has no effect if the kdbus.ko kernel
2338 module is not installed, in which case kdbus is (obviously)
2339 also disabled. We encourage all downstream distributions to
2340 begin testing kdbus by adding it to the kernel images in the
2341 development distributions, and leaving kdbus support in
2344 * The minimal required util-linux version has been bumped to
2347 * Support for chkconfig (--enable-chkconfig) was removed in
2348 favor of calling an abstraction tool
2349 /lib/systemd/systemd-sysv-install. This needs to be
2350 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
2351 in README for details.
2353 * If there's a systemd unit and a SysV init script for the
2354 same service name, and the user executes "systemctl enable"
2355 for it (or a related call), then this will now enable both
2356 (or execute the related operation on both), not just the
2359 * The libudev API documentation has been converted from gtkdoc
2362 * gudev has been removed from the systemd tree, it is now an
2365 * The systemd-cgtop tool learnt a new --raw switch to generate
2366 "raw" (machine parsable) output.
2368 * networkd's IPForwarding= .network file setting learnt the
2369 new setting "kernel", which ensures that networkd does not
2370 change the IP forwarding sysctl from the default kernel
2373 * The systemd-logind bus API now exposes a new boolean
2374 property "Docked" that reports whether logind considers the
2375 system "docked", i.e. connected to a docking station or not.
2377 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
2378 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
2379 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
2380 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
2381 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
2382 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
2383 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
2384 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
2385 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
2386 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
2387 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
2388 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
2389 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
2390 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
2391 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
2392 Fink, Zbigniew Jędrzejewski-Szmek
2394 — Berlin, 2015-06-19
2398 * The gudev library has been extracted into a separate repository
2399 available at: https://git.gnome.org/browse/libgudev/
2400 It is now managed as part of the Gnome project. Distributions
2401 are recommended to pass --disable-gudev to systemd and use
2402 gudev from the Gnome project instead. gudev is still included
2403 in systemd, for now. It will be removed soon, though. Please
2404 also see the announcement-thread on systemd-devel:
2405 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
2407 * systemd now exposes a CPUUsageNSec= property for each
2408 service unit on the bus, that contains the overall consumed
2409 CPU time of a service (the sum of what each process of the
2410 service consumed). This value is only available if
2411 CPUAccounting= is turned on for a service, and is then shown
2412 in the "systemctl status" output.
2414 * Support for configuring alternative mappings of the old SysV
2415 runlevels to systemd targets has been removed. They are now
2416 hardcoded in a way that runlevels 2, 3, 4 all map to
2417 multi-user.target and 5 to graphical.target (which
2418 previously was already the default behaviour).
2420 * The auto-mounter logic gained support for mount point
2421 expiry, using a new TimeoutIdleSec= setting in .automount
2422 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
2424 * The EFI System Partition (ESP) as mounted to /boot by
2425 systemd-efi-boot-generator will now be unmounted
2426 automatically after 2 minutes of not being used. This should
2427 minimize the risk of ESP corruptions.
2429 * New /etc/fstab options x-systemd.requires= and
2430 x-systemd.requires-mounts-for= are now supported to express
2431 additional dependencies for mounts. This is useful for
2432 journalling file systems that support external journal
2433 devices or overlay file systems that require underlying file
2434 systems to be mounted.
2436 * systemd does not support direct live-upgrades (via systemctl
2437 daemon-reexec) from versions older than v44 anymore. As no
2438 distribution we are aware of shipped such old versions in a
2439 stable release this should not be problematic.
2441 * When systemd forks off a new per-connection service instance
2442 it will now set the $REMOTE_ADDR environment variable to the
2443 remote IP address, and $REMOTE_PORT environment variable to
2444 the remote IP port. This behaviour is similar to the
2445 corresponding environment variables defined by CGI.
2447 * systemd-networkd gained support for uplink failure
2448 detection. The BindCarrier= option allows binding interface
2449 configuration dynamically to the link sense of other
2450 interfaces. This is useful to achieve behaviour like in
2453 * systemd-networkd gained support for configuring the DHCP
2454 client identifier to use when requesting leases.
2456 * systemd-networkd now has a per-network UseNTP= option to
2457 configure whether NTP server information acquired via DHCP
2458 is passed on to services like systemd-timesyncd.
2460 * systemd-networkd gained support for vti6 tunnels.
2462 * Note that systemd-networkd manages the sysctl variable
2463 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
2464 it is configured for since v219. The variable controls IP
2465 forwarding, and is a per-interface alternative to the global
2466 /proc/sys/net/ipv[46]/ip_forward. This setting is
2467 configurable in the IPForward= option, which defaults to
2468 "no". This means if networkd is used for an interface it is
2469 no longer sufficient to set the global sysctl option to turn
2470 on IP forwarding! Instead, the .network file option
2471 IPForward= needs to be turned on! Note that the
2472 implementation of this behaviour was broken in v219 and has
2475 * Many bonding and vxlan options are now configurable in
2478 * systemd-nspawn gained a new --property= setting to set unit
2479 properties for the container scope. This is useful for
2480 setting resource parameters (e.g. "CPUShares=500") on
2481 containers started from the command line.
2483 * systemd-nspawn gained a new --private-users= switch to make
2484 use of user namespacing available on recent Linux kernels.
2486 * systemd-nspawn may now be called as part of a shell pipeline
2487 in which case the pipes used for stdin and stdout are passed
2488 directly to the process invoked in the container, without
2489 indirection via a pseudo tty.
2491 * systemd-nspawn gained a new switch to control the UNIX
2492 signal to use when killing the init process of the container
2495 * systemd-nspawn gained a new --overlay= switch for mounting
2496 overlay file systems into the container using the new kernel
2499 * When a container image is imported via systemd-importd and
2500 the host file system is not btrfs, a loopback block device
2501 file is created in /var/lib/machines.raw with a btrfs file
2502 system inside. It is then mounted to /var/lib/machines to
2503 enable btrfs features for container management. The loopback
2504 file and btrfs file system is grown as needed when container
2505 images are imported via systemd-importd.
2507 * systemd-machined/systemd-importd gained support for btrfs
2508 quota, to enforce container disk space limits on disk. This
2509 is exposed in "machinectl set-limit".
2511 * systemd-importd now can import containers from local .tar,
2512 .raw and .qcow2 images, and export them to .tar and .raw. It
2513 can also import dkr v2 images now from the network (on top
2516 * systemd-importd gained support for verifying downloaded
2517 images with gpg2 (previously only gpg1 was supported).
2519 * systemd-machined, systemd-logind, systemd: most bus calls
2520 are now accessible to unprivileged processes via
2521 PolicyKit. Also, systemd-logind will now allow users to kill
2522 their own sessions without further privileges or
2525 * systemd-shutdownd has been removed. This service was
2526 previously responsible for implementing scheduled shutdowns
2527 as exposed in /usr/bin/shutdown's time parameter. This
2528 functionality has now been moved into systemd-logind and is
2529 accessible via a bus interface.
2531 * "systemctl reboot" gained a new switch --firmware-setup that
2532 can be used to reboot into the EFI firmware setup, if that
2533 is available. systemd-logind now exposes an API on the bus
2534 to trigger such reboots, in case graphical desktop UIs want
2535 to cover this functionality.
2537 * "systemctl enable", "systemctl disable" and "systemctl mask"
2538 now support a new "--now" switch. If specified the units
2539 that are enabled will also be started, and the ones
2540 disabled/masked also stopped.
2542 * The Gummiboot EFI boot loader tool has been merged into
2543 systemd, and renamed to "systemd-boot". The bootctl tool has been
2544 updated to support systemd-boot.
2546 * An EFI kernel stub has been added that may be used to create
2547 kernel EFI binaries that contain not only the actual kernel,
2548 but also an initrd, boot splash, command line and OS release
2549 information. This combined binary can then be signed as a
2550 single image, so that the firmware can verify it all in one
2551 step. systemd-boot has special support for EFI binaries created
2552 like this and can extract OS release information from them
2553 and show them in the boot menu. This functionality is useful
2554 to implement cryptographically verified boot schemes.
2556 * Optional support has been added to systemd-fsck to pass
2557 fsck's progress report to an AF_UNIX socket in the file
2560 * udev will no longer create device symlinks for all block
2561 devices by default. A blacklist for excluding special block
2562 devices from this logic has been turned into a whitelist
2563 that requires picking block devices explicitly that require
2566 * A new (currently still internal) API sd-device.h has been
2567 added to libsystemd. This modernized API is supposed to
2568 replace libudev eventually. In fact, already much of libudev
2569 is now just a wrapper around sd-device.h.
2571 * A new hwdb database for storing metadata about pointing
2572 stick devices has been added.
2574 * systemd-tmpfiles gained support for setting file attributes
2575 similar to the "chattr" tool with new 'h' and 'H' lines.
2577 * systemd-journald will no longer unconditionally set the
2578 btrfs NOCOW flag on new journal files. This is instead done
2579 with tmpfiles snippet using the new 'h' line type. This
2580 allows easy disabling of this logic, by masking the
2581 journal-nocow.conf tmpfiles file.
2583 * systemd-journald will now translate audit message types to
2584 human readable identifiers when writing them to the
2585 journal. This should improve readability of audit messages.
2587 * The LUKS logic gained support for the offset= and skip=
2588 options in /etc/crypttab, as previously implemented by
2591 * /usr/lib/os-release gained a new optional field VARIANT= for
2592 distributions that support multiple variants (such as a
2593 desktop edition, a server edition, ...)
2595 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
2596 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
2597 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
2598 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
2599 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
2600 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
2601 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
2602 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
2603 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
2604 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
2605 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
2606 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
2607 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
2608 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
2609 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
2610 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
2611 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
2612 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
2613 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
2614 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
2615 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
2616 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
2617 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
2618 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
2619 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
2620 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
2621 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
2623 — Berlin, 2015-05-22
2627 * Introduce a new API "sd-hwdb.h" for querying the hardware
2628 metadata database. With this minimal interface one can query
2629 and enumerate the udev hwdb, decoupled from the old libudev
2630 library. libudev's interface for this is now only a wrapper
2631 around sd-hwdb. A new tool systemd-hwdb has been added to
2632 interface with and update the database.
2634 * When any of systemd's tools copies files (for example due to
2635 tmpfiles' C lines) a btrfs reflink will attempted first,
2636 before bytewise copying is done.
2638 * systemd-nspawn gained a new --ephemeral switch. When
2639 specified a btrfs snapshot is taken of the container's root
2640 directory, and immediately removed when the container
2641 terminates again. Thus, a container can be started whose
2642 changes never alter the container's root directory, and are
2643 lost on container termination. This switch can also be used
2644 for starting a container off the root file system of the
2645 host without affecting the host OS. This switch is only
2646 available on btrfs file systems.
2648 * systemd-nspawn gained a new --template= switch. It takes the
2649 path to a container tree to use as template for the tree
2650 specified via --directory=, should that directory be
2651 missing. This allows instantiating containers dynamically,
2652 on first run. This switch is only available on btrfs file
2655 * When a .mount unit refers to a mount point on which multiple
2656 mounts are stacked, and the .mount unit is stopped all of
2657 the stacked mount points will now be unmounted until no
2658 mount point remains.
2660 * systemd now has an explicit notion of supported and
2661 unsupported unit types. Jobs enqueued for unsupported unit
2662 types will now fail with an "unsupported" error code. More
2663 specifically .swap, .automount and .device units are not
2664 supported in containers, .busname units are not supported on
2665 non-kdbus systems. .swap and .automount are also not
2666 supported if their respective kernel compile time options
2669 * machinectl gained support for two new "copy-from" and
2670 "copy-to" commands for copying files from a running
2671 container to the host or vice versa.
2673 * machinectl gained support for a new "bind" command to bind
2674 mount host directories into local containers. This is
2675 currently only supported for nspawn containers.
2677 * networkd gained support for configuring bridge forwarding
2678 database entries (fdb) from .network files.
2680 * A new tiny daemon "systemd-importd" has been added that can
2681 download container images in tar, raw, qcow2 or dkr formats,
2682 and make them available locally in /var/lib/machines, so
2683 that they can run as nspawn containers. The daemon can GPG
2684 verify the downloads (not supported for dkr, since it has no
2685 provisions for verifying downloads). It will transparently
2686 decompress bz2, xz, gzip compressed downloads if necessary,
2687 and restore sparse files on disk. The daemon uses privilege
2688 separation to ensure the actual download logic runs with
2689 fewer privileges than the daemon itself. machinectl has
2690 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
2691 make the functionality of importd available to the
2692 user. With this in place the Fedora and Ubuntu "Cloud"
2693 images can be downloaded and booted as containers unmodified
2694 (the Fedora images lack the appropriate GPG signature files
2695 currently, so they cannot be verified, but this will change
2696 soon, hopefully). Note that downloading images is currently
2697 only fully supported on btrfs.
2699 * machinectl is now able to list container images found in
2700 /var/lib/machines, along with some metadata about sizes of
2701 disk and similar. If the directory is located on btrfs and
2702 quota is enabled, this includes quota display. A new command
2703 "image-status" has been added that shows additional
2704 information about images.
2706 * machinectl is now able to clone container images
2707 efficiently, if the underlying file system (btrfs) supports
2708 it, with the new "machinectl clone" command. It also
2709 gained commands for renaming and removing images, as well as
2710 marking them read-only or read-write (supported also on
2711 legacy file systems).
2713 * networkd gained support for collecting LLDP network
2714 announcements, from hardware that supports this. This is
2715 shown in networkctl output.
2717 * systemd-run gained support for a new -t (--pty) switch for
2718 invoking a binary on a pty whose input and output is
2719 connected to the invoking terminal. This allows executing
2720 processes as system services while interactively
2721 communicating with them via the terminal. Most interestingly
2722 this is supported across container boundaries. Invoking
2723 "systemd-run -t /bin/bash" is an alternative to running a
2724 full login session, the difference being that the former
2725 will not register a session, nor go through the PAM session
2728 * tmpfiles gained support for a new "v" line type for creating
2729 btrfs subvolumes. If the underlying file system is a legacy
2730 file system, this automatically degrades to creating a
2731 normal directory. Among others /var/lib/machines is now
2732 created like this at boot, should it be missing.
2734 * The directory /var/lib/containers/ has been deprecated and
2735 been replaced by /var/lib/machines. The term "machines" has
2736 been used in the systemd context as generic term for both
2737 VMs and containers, and hence appears more appropriate for
2738 this, as the directory can also contain raw images bootable
2741 * systemd-nspawn when invoked with -M but without --directory=
2742 or --image= is now capable of searching for the container
2743 root directory, subvolume or disk image automatically, in
2744 /var/lib/machines. systemd-nspawn@.service has been updated
2745 to make use of this, thus allowing it to be used for raw
2748 * A new machines.target unit has been introduced that is
2749 supposed to group all containers/VMs invoked as services on
2750 the system. systemd-nspawn@.service has been updated to
2751 integrate with that.
2753 * machinectl gained a new "start" command, for invoking a
2754 container as a service. "machinectl start foo" is mostly
2755 equivalent to "systemctl start systemd-nspawn@foo.service",
2756 but handles escaping in a nicer way.
2758 * systemd-nspawn will now mount most of the cgroupfs tree
2759 read-only into each container, with the exception of the
2760 container's own subtree in the name=systemd hierarchy.
2762 * journald now sets the special FS_NOCOW file flag for its
2763 journal files. This should improve performance on btrfs, by
2764 avoiding heavy fragmentation when journald's write-pattern
2765 is used on COW file systems. It degrades btrfs' data
2766 integrity guarantees for the files to the same levels as for
2767 ext3/ext4 however. This should be OK though as journald does
2768 its own data integrity checks and all its objects are
2769 checksummed on disk. Also, journald should handle btrfs disk
2770 full events a lot more gracefully now, by processing SIGBUS
2771 errors, and not relying on fallocate() anymore.
2773 * When journald detects that journal files it is writing to
2774 have been deleted it will immediately start new journal
2777 * systemd now provides a way to store file descriptors
2778 per-service in PID 1. This is useful for daemons to ensure
2779 that fds they require are not lost during a daemon
2780 restart. The fds are passed to the daemon on the next
2781 invocation in the same way socket activation fds are
2782 passed. This is now used by journald to ensure that the
2783 various sockets connected to all the system's stdout/stderr
2784 are not lost when journald is restarted. File descriptors
2785 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
2786 an extension to sd_notify(). Note that a limit is enforced
2787 on the number of fds a service can store in PID 1, and it
2788 defaults to 0, so that no fds may be stored, unless this is
2789 explicitly turned on.
2791 * The default TERM variable to use for units connected to a
2792 terminal, when no other value is explicitly is set is now
2793 vt220 rather than vt102. This should be fairly safe still,
2794 but allows PgUp/PgDn work.
2796 * The /etc/crypttab option header= as known from Debian is now
2799 * "loginctl user-status" and "loginctl session-status" will
2800 now show the last 10 lines of log messages of the
2801 user/session following the status output. Similar,
2802 "machinectl status" will show the last 10 log lines
2803 associated with a virtual machine or container
2804 service. (Note that this is usually not the log messages
2805 done in the VM/container itself, but simply what the
2806 container manager logs. For nspawn this includes all console
2809 * "loginctl session-status" without further argument will now
2810 show the status of the session of the caller. Similar,
2811 "lock-session", "unlock-session", "activate",
2812 "enable-linger", "disable-linger" may now be called without
2813 session/user parameter in which case they apply to the
2814 caller's session/user.
2816 * An X11 session scriptlet is now shipped that uploads
2817 $DISPLAY and $XAUTHORITY into the environment of the systemd
2818 --user daemon if a session begins. This should improve
2819 compatibility with X11 enabled applications run as systemd
2822 * Generators are now subject to masking via /etc and /run, the
2823 same way as unit files.
2825 * networkd .network files gained support for configuring
2826 per-link IPv4/IPv6 packet forwarding as well as IPv4
2827 masquerading. This is by default turned on for veth links to
2828 containers, as registered by systemd-nspawn. This means that
2829 nspawn containers run with --network-veth will now get
2830 automatic routed access to the host's networks without any
2831 further configuration or setup, as long as networkd runs on
2834 * systemd-nspawn gained the --port= (-p) switch to expose TCP
2835 or UDP posts of a container on the host. With this in place
2836 it is possible to run containers with private veth links
2837 (--network-veth), and have their functionality exposed on
2838 the host as if their services were running directly on the
2841 * systemd-nspawn's --network-veth switch now gained a short
2842 version "-n", since with the changes above it is now truly
2843 useful out-of-the-box. The systemd-nspawn@.service has been
2844 updated to make use of it too by default.
2846 * systemd-nspawn will now maintain a per-image R/W lock, to
2847 ensure that the same image is not started more than once
2848 writable. (It's OK to run an image multiple times
2849 simultaneously in read-only mode.)
2851 * systemd-nspawn's --image= option is now capable of
2852 dissecting and booting MBR and GPT disk images that contain
2853 only a single active Linux partition. Previously it
2854 supported only GPT disk images with proper GPT type
2855 IDs. This allows running cloud images from major
2856 distributions directly with systemd-nspawn, without
2859 * In addition to collecting mouse dpi data in the udev
2860 hardware database, there's now support for collecting angle
2861 information for mouse scroll wheels. The database is
2862 supposed to guarantee similar scrolling behavior on mice
2863 that it knows about. There's also support for collecting
2864 information about Touchpad types.
2866 * udev's input_id built-in will now also collect touch screen
2867 dimension data and attach it to probed devices.
2869 * /etc/os-release gained support for a Distribution Privacy
2872 * networkd gained support for creating "ipvlan", "gretap",
2873 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
2875 * systemd-tmpfiles gained support for "a" lines for setting
2878 * systemd-nspawn will now mount /tmp in the container to
2879 tmpfs, automatically.
2881 * systemd now exposes the memory.usage_in_bytes cgroup
2882 attribute and shows it for each service in the "systemctl
2883 status" output, if available.
2885 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
2886 immediate reboot is triggered. This useful if shutdown is
2887 hung and is unable to complete, to expedite the
2888 operation. Note that this kind of reboot will still unmount
2889 all file systems, and hence should not result in fsck being
2892 * A .device unit for an optical block device will now be
2893 considered active only when a medium is in the drive. Also,
2894 mount units are now bound to their backing devices thus
2895 triggering automatic unmounting when devices become
2896 unavailable. With this in place systemd will now
2897 automatically unmount left-over mounts when a CD-ROM is
2898 ejected or an USB stick is yanked from the system.
2900 * networkd-wait-online now has support for waiting for
2901 specific interfaces only (with globbing), and for giving up
2902 after a configurable timeout.
2904 * networkd now exits when idle. It will be automatically
2905 restarted as soon as interfaces show up, are removed or
2906 change state. networkd will stay around as long as there is
2907 at least one DHCP state machine or similar around, that keep
2910 * networkd may now configure IPv6 link-local addressing in
2911 addition to IPv4 link-local addressing.
2913 * The IPv6 "token" for use in SLAAC may now be configured for
2914 each .network interface in networkd.
2916 * Routes configured with networkd may now be assigned a scope
2919 * networkd's [Match] sections now support globbing and lists
2920 of multiple space-separated matches per item.
2922 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
2923 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
2924 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
2925 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
2926 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
2927 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
2928 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
2929 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
2930 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
2931 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
2932 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
2933 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
2934 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
2935 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
2936 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
2937 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
2938 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
2939 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
2940 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
2941 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
2942 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
2943 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
2944 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
2945 Hoffmann, Zbigniew Jędrzejewski-Szmek
2947 — Berlin, 2015-02-16
2951 * When querying unit file enablement status (for example via
2952 "systemctl is-enabled"), a new state "indirect" is now known
2953 which indicates that a unit might not be enabled itself, but
2954 another unit listed in its Also= setting might be.
2956 * Similar to the various existing ConditionXYZ= settings for
2957 units, there are now matching AssertXYZ= settings. While
2958 failing conditions cause a unit to be skipped, but its job
2959 to succeed, failing assertions declared like this will cause
2960 a unit start operation and its job to fail.
2962 * hostnamed now knows a new chassis type "embedded".
2964 * systemctl gained a new "edit" command. When used on a unit
2965 file, this allows extending unit files with .d/ drop-in
2966 configuration snippets or editing the full file (after
2967 copying it from /usr/lib to /etc). This will invoke the
2968 user's editor (as configured with $EDITOR), and reload the
2969 modified configuration after editing.
2971 * "systemctl status" now shows the suggested enablement state
2972 for a unit, as declared in the (usually vendor-supplied)
2973 system preset files.
2975 * nss-myhostname will now resolve the single-label host name
2976 "gateway" to the locally configured default IP routing
2977 gateways, ordered by their metrics. This assigns a stable
2978 name to the used gateways, regardless which ones are
2979 currently configured. Note that the name will only be
2980 resolved after all other name sources (if nss-myhostname is
2981 configured properly) and should hence not negatively impact
2982 systems that use the single-label host name "gateway" in
2985 * systemd-inhibit now allows filtering by mode when listing
2988 * Scope and service units gained a new "Delegate" boolean
2989 property, which, when set, allows processes running inside the
2990 unit to further partition resources. This is primarily
2991 useful for systemd user instances as well as container
2994 * journald will now pick up audit messages directly from
2995 the kernel, and log them like any other log message. The
2996 audit fields are split up and fully indexed. This means that
2997 journalctl in many ways is now a (nicer!) alternative to
2998 ausearch, the traditional audit client. Note that this
2999 implements only a minimal audit client. If you want the
3000 special audit modes like reboot-on-log-overflow, please use
3001 the traditional auditd instead, which can be used in
3002 parallel to journald.
3004 * The ConditionSecurity= unit file option now understands the
3005 special string "audit" to check whether auditing is
3008 * journalctl gained two new commands --vacuum-size= and
3009 --vacuum-time= to delete old journal files until the
3010 remaining ones take up no more than the specified size on disk,
3011 or are not older than the specified time.
3013 * A new, native PPPoE library has been added to sd-network,
3014 systemd's library of light-weight networking protocols. This
3015 library will be used in a future version of networkd to
3016 enable PPPoE communication without an external pppd daemon.
3018 * The busctl tool now understands a new "capture" verb that
3019 works similar to "monitor", but writes a packet capture
3020 trace to STDOUT that can be redirected to a file which is
3021 compatible with libcap's capture file format. This can then
3022 be loaded in Wireshark and similar tools to inspect bus
3025 * The busctl tool now understands a new "tree" verb that shows
3026 the object trees of a specific service on the bus, or of all
3029 * The busctl tool now understands a new "introspect" verb that
3030 shows all interfaces and members of objects on the bus,
3031 including their signature and values. This is particularly
3032 useful to get more information about bus objects shown by
3033 the new "busctl tree" command.
3035 * The busctl tool now understands new verbs "call",
3036 "set-property" and "get-property" for invoking bus method
3037 calls, setting and getting bus object properties in a
3040 * busctl gained a new --augment-creds= argument that controls
3041 whether the tool shall augment credential information it
3042 gets from the bus with data from /proc, in a possibly
3045 * nspawn's --link-journal= switch gained two new values
3046 "try-guest" and "try-host" that work like "guest" and
3047 "host", but do not fail if the host has no persistent
3048 journalling enabled. -j is now equivalent to
3049 --link-journal=try-guest.
3051 * macvlan network devices created by nspawn will now have
3052 stable MAC addresses.
3054 * A new SmackProcessLabel= unit setting has been added, which
3055 controls the SMACK security label processes forked off by
3056 the respective unit shall use.
3058 * If compiled with --enable-xkbcommon, systemd-localed will
3059 verify x11 keymap settings by compiling the given keymap. It
3060 will spew out warnings if the compilation fails. This
3061 requires libxkbcommon to be installed.
3063 * When a coredump is collected, a larger number of metadata
3064 fields is now collected and included in the journal records
3065 created for it. More specifically, control group membership,
3066 environment variables, memory maps, working directory,
3067 chroot directory, /proc/$PID/status, and a list of open file
3068 descriptors is now stored in the log entry.
3070 * The udev hwdb now contains DPI information for mice. For
3073 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
3075 * All systemd programs that read standalone configuration
3076 files in /etc now also support a corresponding series of
3077 .conf.d configuration directories in /etc/, /run/,
3078 /usr/local/lib/, /usr/lib/, and (if configured with
3079 --enable-split-usr) /lib/. In particular, the following
3080 configuration files now have corresponding configuration
3081 directories: system.conf user.conf, logind.conf,
3082 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
3083 resolved.conf, timesyncd.conf, journal-remote.conf, and
3084 journal-upload.conf. Note that distributions should use the
3085 configuration directories in /usr/lib/; the directories in
3086 /etc/ are reserved for the system administrator.
3088 * systemd-rfkill will no longer take the rfkill device name
3089 into account when storing rfkill state on disk, as the name
3090 might be dynamically assigned and not stable. Instead, the
3091 ID_PATH udev variable combined with the rfkill type (wlan,
3092 bluetooth, ...) is used.
3094 * A new service systemd-machine-id-commit.service has been
3095 added. When used on systems where /etc is read-only during
3096 boot, and /etc/machine-id is not initialized (but an empty
3097 file), this service will copy the temporary machine ID
3098 created as replacement into /etc after the system is fully
3099 booted up. This is useful for systems that are freshly
3100 installed with a non-initialized machine ID, but should get
3101 a fixed machine ID for subsequent boots.
3103 * networkd's .netdev files now provide a large set of
3104 configuration parameters for VXLAN devices. Similarly, the
3105 bridge port cost parameter is now configurable in .network
3106 files. There's also new support for configuring IP source
3107 routing. networkd .link files gained support for a new
3108 OriginalName= match that is useful to match against the
3109 original interface name the kernel assigned. .network files
3110 may include MTU= and MACAddress= fields for altering the MTU
3111 and MAC address while being connected to a specific network
3114 * The LUKS logic gained supported for configuring
3115 UUID-specific key files. There's also new support for naming
3116 LUKS device from the kernel command line, using the new
3117 luks.name= argument.
3119 * Timer units may now be transiently created via the bus API
3120 (this was previously already available for scope and service
3121 units). In addition it is now possible to create multiple
3122 transient units at the same time with a single bus call. The
3123 "systemd-run" tool has been updated to make use of this for
3124 running commands on a specified time, in at(1)-style.
3126 * tmpfiles gained support for "t" lines, for assigning
3127 extended attributes to files. Among other uses this may be
3128 used to assign SMACK labels to files.
3130 Contributions from: Alin Rauta, Alison Chaiken, Andrej
3131 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
3132 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
3133 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
3134 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
3135 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
3136 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
3137 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
3138 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
3139 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
3140 Michael Biebl, Michael Chapman, Michael Marineau, Michal
3141 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
3142 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
3143 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
3144 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
3145 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
3146 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
3147 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
3149 — Berlin, 2014-12-10
3153 * journalctl gained the new options -t/--identifier= to match
3154 on the syslog identifier (aka "tag"), as well as --utc to
3155 show log timestamps in the UTC timezone. journalctl now also
3156 accepts -n/--lines=all to disable line capping in a pager.
3158 * journalctl gained a new switch, --flush, that synchronously
3159 flushes logs from /run/log/journal to /var/log/journal if
3160 persistent storage is enabled. systemd-journal-flush.service
3161 now waits until the operation is complete.
3163 * Services can notify the manager before they start a reload
3164 (by sending RELOADING=1) or shutdown (by sending
3165 STOPPING=1). This allows the manager to track and show the
3166 internal state of daemons and closes a race condition when
3167 the process is still running but has closed its D-Bus
3170 * Services with Type=oneshot do not have to have any ExecStart
3173 * User units are now loaded also from
3174 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
3175 /run/systemd/user directory that was already previously
3176 supported, but is under the control of the user.
3178 * Job timeouts (i.e. time-outs on the time a job that is
3179 queued stays in the run queue) can now optionally result in
3180 immediate reboot or power-off actions (JobTimeoutAction= and
3181 JobTimeoutRebootArgument=). This is useful on ".target"
3182 units, to limit the maximum time a target remains
3183 undispatched in the run queue, and to trigger an emergency
3184 operation in such a case. This is now used by default to
3185 turn off the system if boot-up (as defined by everything in
3186 basic.target) hangs and does not complete for at least
3187 15min. Also, if power-off or reboot hang for at least 30min
3188 an immediate power-off/reboot operation is triggered. This
3189 functionality is particularly useful to increase reliability
3190 on embedded devices, but also on laptops which might
3191 accidentally get powered on when carried in a backpack and
3192 whose boot stays stuck in a hard disk encryption passphrase
3195 * systemd-logind can be configured to also handle lid switch
3196 events even when the machine is docked or multiple displays
3197 are attached (HandleLidSwitchDocked= option).
3199 * A helper binary and a service have been added which can be
3200 used to resume from hibernation in the initramfs. A
3201 generator will parse the resume= option on the kernel
3202 command line to trigger resume.
3204 * A user console daemon systemd-consoled has been
3205 added. Currently, it is a preview, and will so far open a
3206 single terminal on each session of the user marked as
3207 Desktop=systemd-console.
3209 * Route metrics can be specified for DHCP routes added by
3212 * The SELinux context of socket-activated services can be set
3213 from the information provided by the networking stack
3214 (SELinuxContextFromNet= option).
3216 * Userspace firmware loading support has been removed and
3217 the minimum supported kernel version is thus bumped to 3.7.
3219 * Timeout for udev workers has been increased from 1 to 3
3220 minutes, but a warning will be printed after 1 minute to
3221 help diagnose kernel modules that take a long time to load.
3223 * Udev rules can now remove tags on devices with TAG-="foobar".
3225 * systemd's readahead implementation has been removed. In many
3226 circumstances it didn't give expected benefits even for
3227 rotational disk drives and was becoming less relevant in the
3228 age of SSDs. As none of the developers has been using
3229 rotating media anymore, and nobody stepped up to actively
3230 maintain this component of systemd it has now been removed.
3232 * Swap units can use Options= to specify discard options.
3233 Discard options specified for swaps in /etc/fstab are now
3236 * Docker containers are now detected as a separate type of
3239 * The Password Agent protocol gained support for queries where
3240 the user input is shown, useful e.g. for user names.
3241 systemd-ask-password gained a new --echo option to turn that
3244 * The default sysctl.d/ snippets will now set:
3246 net.core.default_qdisc = fq_codel
3248 This selects Fair Queuing Controlled Delay as the default
3249 queuing discipline for network interfaces. fq_codel helps
3250 fight the network bufferbloat problem. It is believed to be
3251 a good default with no tuning required for most workloads.
3252 Downstream distributions may override this choice. On 10Gbit
3253 servers that do not do forwarding, "fq" may perform better.
3254 Systems without a good clocksource should use "pfifo_fast".
3256 * If kdbus is enabled during build a new option BusPolicy= is
3257 available for service units, that allows locking all service
3258 processes into a stricter bus policy, in order to limit
3259 access to various bus services, or even hide most of them
3260 from the service's view entirely.
3262 * networkctl will now show the .network and .link file
3263 networkd has applied to a specific interface.
3265 * sd-login gained a new API call sd_session_get_desktop() to
3266 query which desktop environment has been selected for a
3269 * UNIX utmp support is now compile-time optional to support
3270 legacy-free systems.
3272 * systemctl gained two new commands "add-wants" and
3273 "add-requires" for pulling in units from specific targets
3276 * If the word "rescue" is specified on the kernel command line
3277 the system will now boot into rescue mode (aka
3278 rescue.target), which was previously available only by
3279 specifying "1" or "systemd.unit=rescue.target" on the kernel
3280 command line. This new kernel command line option nicely
3281 mirrors the already existing "emergency" kernel command line
3284 * New kernel command line options mount.usr=, mount.usrflags=,
3285 mount.usrfstype= have been added that match root=, rootflags=,
3286 rootfstype= but allow mounting a specific file system to
3289 * The $NOTIFY_SOCKET is now also passed to control processes of
3290 services, not only the main process.
3292 * This version reenables support for fsck's -l switch. This
3293 means at least version v2.25 of util-linux is required for
3294 operation, otherwise dead-locks on device nodes may
3295 occur. Again: you need to update util-linux to at least
3296 v2.25 when updating systemd to v217.
3298 * The "multi-seat-x" tool has been removed from systemd, as
3299 its functionality has been integrated into X servers 1.16,
3300 and the tool is hence redundant. It is recommended to update
3301 display managers invoking this tool to simply invoke X
3302 directly from now on, again.
3304 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
3305 message flag has been added for all of systemd's PolicyKit
3306 authenticated method calls has been added. In particular
3307 this now allows optional interactive authorization via
3308 PolicyKit for many of PID1's privileged operations such as
3309 unit file enabling and disabling.
3311 * "udevadm hwdb --update" learnt a new switch "--usr" for
3312 placing the rebuilt hardware database in /usr instead of
3313 /etc. When used only hardware database entries stored in
3314 /usr will be used, and any user database entries in /etc are
3315 ignored. This functionality is useful for vendors to ship a
3316 pre-built database on systems where local configuration is
3317 unnecessary or unlikely.
3319 * Calendar time specifications in .timer units now also
3320 understand the strings "semi-annually", "quarterly" and
3321 "minutely" as shortcuts (in addition to the preexisting
3322 "anually", "hourly", ...).
3324 * systemd-tmpfiles will now correctly create files in /dev
3325 at boot which are marked for creation only at boot. It is
3326 recommended to always create static device nodes with 'c!'
3327 and 'b!', so that they are created only at boot and not
3328 overwritten at runtime.
3330 * When the watchdog logic is used for a service (WatchdogSec=)
3331 and the watchdog timeout is hit the service will now be
3332 terminated with SIGABRT (instead of just SIGTERM), in order
3333 to make sure a proper coredump and backtrace is
3334 generated. This ensures that hanging services will result in
3335 similar coredump/backtrace behaviour as services that hit a
3338 Contributions from: Andreas Henriksson, Andrei Borzenkov,
3339 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
3340 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
3341 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
3342 Herrmann, David Sommerseth, David Strauss, Emil Renner
3343 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
3344 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
3345 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
3346 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
3347 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
3348 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
3349 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
3350 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
3351 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
3352 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
3353 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
3354 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
3355 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
3356 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
3357 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
3358 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
3361 — Berlin, 2014-10-28
3365 * timedated no longer reads NTP implementation unit names from
3366 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
3367 implementations should add a
3369 Conflicts=systemd-timesyncd.service
3371 to their unit files to take over and replace systemd's NTP
3372 default functionality.
3374 * systemd-sysusers gained a new line type "r" for configuring
3375 which UID/GID ranges to allocate system users/groups
3376 from. Lines of type "u" may now add an additional column
3377 that specifies the home directory for the system user to be
3378 created. Also, systemd-sysusers may now optionally read user
3379 information from STDIN instead of a file. This is useful for
3380 invoking it from RPM preinst scriptlets that need to create
3381 users before the first RPM file is installed since these
3382 files might need to be owned by them. A new
3383 %sysusers_create_inline RPM macro has been introduced to do
3384 just that. systemd-sysusers now updates the shadow files as
3385 well as the user/group databases, which should enhance
3386 compatibility with certain tools like grpck.
3388 * A number of bus APIs of PID 1 now optionally consult
3389 PolicyKit to permit access for otherwise unprivileged
3390 clients under certain conditions. Note that this currently
3391 doesn't support interactive authentication yet, but this is
3392 expected to be added eventually, too.
3394 * /etc/machine-info now has new fields for configuring the
3395 deployment environment of the machine, as well as the
3396 location of the machine. hostnamectl has been updated with
3397 new command to update these fields.
3399 * systemd-timesyncd has been updated to automatically acquire
3400 NTP server information from systemd-networkd, which might
3401 have been discovered via DHCP.
3403 * systemd-resolved now includes a caching DNS stub resolver
3404 and a complete LLMNR name resolution implementation. A new
3405 NSS module "nss-resolve" has been added which can be used
3406 instead of glibc's own "nss-dns" to resolve hostnames via
3407 systemd-resolved. Hostnames, addresses and arbitrary RRs may
3408 be resolved via systemd-resolved D-Bus APIs. In contrast to
3409 the glibc internal resolver systemd-resolved is aware of
3410 multi-homed system, and keeps DNS server and caches separate
3411 and per-interface. Queries are sent simultaneously on all
3412 interfaces that have DNS servers configured, in order to
3413 properly handle VPNs and local LANs which might resolve
3414 separate sets of domain names. systemd-resolved may acquire
3415 DNS server information from systemd-networkd automatically,
3416 which in turn might have discovered them via DHCP. A tool
3417 "systemd-resolve-host" has been added that may be used to
3418 query the DNS logic in resolved. systemd-resolved implements
3419 IDNA and automatically uses IDNA or UTF-8 encoding depending
3420 on whether classic DNS or LLMNR is used as transport. In the
3421 next releases we intend to add a DNSSEC and mDNS/DNS-SD
3422 implementation to systemd-resolved.
3424 * A new NSS module nss-mymachines has been added, that
3425 automatically resolves the names of all local registered
3426 containers to their respective IP addresses.
3428 * A new client tool "networkctl" for systemd-networkd has been
3429 added. It currently is entirely passive and will query
3430 networking configuration from udev, rtnetlink and networkd,
3431 and present it to the user in a very friendly
3432 way. Eventually, we hope to extend it to become a full
3433 control utility for networkd.
3435 * .socket units gained a new DeferAcceptSec= setting that
3436 controls the kernels' TCP_DEFER_ACCEPT sockopt for
3437 TCP. Similarly, support for controlling TCP keep-alive
3438 settings has been added (KeepAliveTimeSec=,
3439 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
3440 turning off Nagle's algorithm on TCP has been added
3443 * logind learned a new session type "web", for use in projects
3444 like Cockpit which register web clients as PAM sessions.
3446 * timer units with at least one OnCalendar= setting will now
3447 be started only after time-sync.target has been
3448 reached. This way they will not elapse before the system
3449 clock has been corrected by a local NTP client or
3450 similar. This is particular useful on RTC-less embedded
3451 machines, that come up with an invalid system clock.
3453 * systemd-nspawn's --network-veth= switch should now result in
3454 stable MAC addresses for both the outer and the inner side
3457 * systemd-nspawn gained a new --volatile= switch for running
3458 container instances with /etc or /var unpopulated.
3460 * The kdbus client code has been updated to use the new Linux
3461 3.17 memfd subsystem instead of the old kdbus-specific one.
3463 * systemd-networkd's DHCP client and server now support
3464 FORCERENEW. There are also new configuration options to
3465 configure the vendor client identifier and broadcast mode
3468 * systemd will no longer inform the kernel about the current
3469 timezone, as this is necessarily incorrect and racy as the
3470 kernel has no understanding of DST and similar
3471 concepts. This hence means FAT timestamps will be always
3472 considered UTC, similar to what Android is already
3473 doing. Also, when the RTC is configured to the local time
3474 (rather than UTC) systemd will never synchronize back to it,
3475 as this might confuse Windows at a later boot.
3477 * systemd-analyze gained a new command "verify" for offline
3478 validation of unit files.
3480 * systemd-networkd gained support for a couple of additional
3481 settings for bonding networking setups. Also, the metric for
3482 statically configured routes may now be configured. For
3483 network interfaces where this is appropriate the peer IP
3484 address may now be configured.
3486 * systemd-networkd's DHCP client will no longer request
3487 broadcasting by default, as this tripped up some networks.
3488 For hardware where broadcast is required the feature should
3489 be switched back on using RequestBroadcast=yes.
3491 * systemd-networkd will now set up IPv4LL addresses (when
3492 enabled) even if DHCP is configured successfully.
3494 * udev will now default to respect network device names given
3495 by the kernel when the kernel indicates that these are
3496 predictable. This behavior can be tweaked by changing
3497 NamePolicy= in the relevant .link file.
3499 * A new library systemd-terminal has been added that
3500 implements full TTY stream parsing and rendering. This
3501 library is supposed to be used later on for implementing a
3502 full userspace VT subsystem, replacing the current kernel
3505 * A new tool systemd-journal-upload has been added to push
3506 journal data to a remote system running
3507 systemd-journal-remote.
3509 * journald will no longer forward all local data to another
3510 running syslog daemon. This change has been made because
3511 rsyslog (which appears to be the most commonly used syslog
3512 implementation these days) no longer makes use of this, and
3513 instead pulls the data out of the journal on its own. Since
3514 forwarding the messages to a non-existent syslog server is
3515 more expensive than we assumed we have now turned this
3516 off. If you run a syslog server that is not a recent rsyslog
3517 version, you have to turn this option on again
3518 (ForwardToSyslog= in journald.conf).
3520 * journald now optionally supports the LZ4 compressor for
3521 larger journal fields. This compressor should perform much
3522 better than XZ which was the previous default.
3524 * machinectl now shows the IP addresses of local containers,
3525 if it knows them, plus the interface name of the container.
3527 * A new tool "systemd-escape" has been added that makes it
3528 easy to escape strings to build unit names and similar.
3530 * sd_notify() messages may now include a new ERRNO= field
3531 which is parsed and collected by systemd and shown among the
3532 "systemctl status" output for a service.
3534 * A new component "systemd-firstboot" has been added that
3535 queries the most basic systemd information (timezone,
3536 hostname, root password) interactively on first
3537 boot. Alternatively it may also be used to provision these
3538 things offline on OS images installed into directories.
3540 * The default sysctl.d/ snippets will now set
3542 net.ipv4.conf.default.promote_secondaries=1
3544 This has the benefit of no flushing secondary IP addresses
3545 when primary addresses are removed.
3547 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
3548 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
3549 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
3550 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
3551 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
3552 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
3553 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
3554 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
3555 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
3556 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
3557 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
3558 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
3559 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
3560 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
3561 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
3563 — Berlin, 2014-08-19
3567 * A new tool systemd-sysusers has been added. This tool
3568 creates system users and groups in /etc/passwd and
3569 /etc/group, based on static declarative system user/group
3570 definitions in /usr/lib/sysusers.d/. This is useful to
3571 enable factory resets and volatile systems that boot up with
3572 an empty /etc directory, and thus need system users and
3573 groups created during early boot. systemd now also ships
3574 with two default sysusers.d/ files for the most basic
3575 users and groups systemd and the core operating system
3578 * A new tmpfiles snippet has been added that rebuilds the
3579 essential files in /etc on boot, should they be missing.
3581 * A directive for ensuring automatic clean-up of
3582 /var/cache/man/ has been removed from the default
3583 configuration. This line should now be shipped by the man
3584 implementation. The necessary change has been made to the
3585 man-db implementation. Note that you need to update your man
3586 implementation to one that ships this line, otherwise no
3587 automatic clean-up of /var/cache/man will take place.
3589 * A new condition ConditionNeedsUpdate= has been added that
3590 may conditionalize services to only run when /etc or /var
3591 are "older" than the vendor operating system resources in
3592 /usr. This is useful for reconstructing or updating /etc
3593 after an offline update of /usr or a factory reset, on the
3594 next reboot. Services that want to run once after such an
3595 update or reset should use this condition and order
3596 themselves before the new systemd-update-done.service, which
3597 will mark the two directories as fully updated. A number of
3598 service files have been added making use of this, to rebuild
3599 the udev hardware database, the journald message catalog and
3600 dynamic loader cache (ldconfig). The systemd-sysusers tool
3601 described above also makes use of this now. With this in
3602 place it is now possible to start up a minimal operating
3603 system with /etc empty cleanly. For more information on the
3604 concepts involved see this recent blog story:
3606 http://0pointer.de/blog/projects/stateless.html
3608 * A new system group "input" has been introduced, and all
3609 input device nodes get this group assigned. This is useful
3610 for system-level software to get access to input devices. It
3611 complements what is already done for "audio" and "video".
3613 * systemd-networkd learnt minimal DHCPv4 server support in
3614 addition to the existing DHCPv4 client support. It also
3615 learnt DHCPv6 client and IPv6 Router Solicitation client
3616 support. The DHCPv4 client gained support for static routes
3617 passed in from the server. Note that the [DHCPv4] section
3618 known in older systemd-networkd versions has been renamed to
3619 [DHCP] and is now also used by the DHCPv6 client. Existing
3620 .network files using settings of this section should be
3621 updated, though compatibility is maintained. Optionally, the
3622 client hostname may now be sent to the DHCP server.
3624 * networkd gained support for vxlan virtual networks as well
3625 as tun/tap and dummy devices.
3627 * networkd gained support for automatic allocation of address
3628 ranges for interfaces from a system-wide pool of
3629 addresses. This is useful for dynamically managing a large
3630 number of interfaces with a single network configuration
3631 file. In particular this is useful to easily assign
3632 appropriate IP addresses to the veth links of a large number
3633 of nspawn instances.
3635 * RPM macros for processing sysusers, sysctl and binfmt
3636 drop-in snippets at package installation time have been
3639 * The /etc/os-release file should now be placed in
3640 /usr/lib/os-release. The old location is automatically
3641 created as symlink. /usr/lib is the more appropriate
3642 location of this file, since it shall actually describe the
3643 vendor operating system shipped in /usr, and not the
3644 configuration stored in /etc.
3646 * .mount units gained a new boolean SloppyOptions= setting
3647 that maps to mount(8)'s -s option which enables permissive
3648 parsing of unknown mount options.
3650 * tmpfiles learnt a new "L+" directive which creates a symlink
3651 but (unlike "L") deletes a pre-existing file first, should
3652 it already exist and not already be the correct
3653 symlink. Similarly, "b+", "c+" and "p+" directives have been
3654 added as well, which create block and character devices, as
3655 well as fifos in the filesystem, possibly removing any
3656 pre-existing files of different types.
3658 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
3659 'argument' field (which so far specified the source to
3660 symlink/copy the files from) is now optional. If omitted the
3661 same file os copied from /usr/share/factory/ suffixed by the
3662 full destination path. This is useful for populating /etc
3663 with essential files, by copying them from vendor defaults
3664 shipped in /usr/share/factory/etc.
3666 * A new command "systemctl preset-all" has been added that
3667 applies the service preset settings to all installed unit
3668 files. A new switch --preset-mode= has been added that
3669 controls whether only enable or only disable operations
3672 * A new command "systemctl is-system-running" has been added
3673 that allows checking the overall state of the system, for
3674 example whether it is fully up and running.
3676 * When the system boots up with an empty /etc, the equivalent
3677 to "systemctl preset-all" is executed during early boot, to
3678 make sure all default services are enabled after a factory
3681 * systemd now contains a minimal preset file that enables the
3682 most basic services systemd ships by default.
3684 * Unit files' [Install] section gained a new DefaultInstance=
3685 field for defining the default instance to create if a
3686 template unit is enabled with no instance specified.
3688 * A new passive target cryptsetup-pre.target has been added
3689 that may be used by services that need to make they run and
3690 finish before the first LUKS cryptographic device is set up.
3692 * The /dev/loop-control and /dev/btrfs-control device nodes
3693 are now owned by the "disk" group by default, opening up
3694 access to this group.
3696 * systemd-coredump will now automatically generate a
3697 stack trace of all core dumps taking place on the system,
3698 based on elfutils' libdw library. This stack trace is logged
3701 * systemd-coredump may now optionally store coredumps directly
3702 on disk (in /var/lib/systemd/coredump, possibly compressed),
3703 instead of storing them unconditionally in the journal. This
3704 mode is the new default. A new configuration file
3705 /etc/systemd/coredump.conf has been added to configure this
3706 and other parameters of systemd-coredump.
3708 * coredumpctl gained a new "info" verb to show details about a
3709 specific coredump. A new switch "-1" has also been added
3710 that makes sure to only show information about the most
3711 recent entry instead of all entries. Also, as the tool is
3712 generally useful now the "systemd-" prefix of the binary
3713 name has been removed. Distributions that want to maintain
3714 compatibility with the old name should add a symlink from
3715 the old name to the new name.
3717 * journald's SplitMode= now defaults to "uid". This makes sure
3718 that unprivileged users can access their own coredumps with
3719 coredumpctl without restrictions.
3721 * New kernel command line options "systemd.wants=" (for
3722 pulling an additional unit during boot), "systemd.mask="
3723 (for masking a specific unit for the boot), and
3724 "systemd.debug-shell" (for enabling the debug shell on tty9)
3725 have been added. This is implemented in the new generator
3726 "systemd-debug-generator".
3728 * systemd-nspawn will now by default filter a couple of
3729 syscalls for containers, among them those required for
3730 kernel module loading, direct x86 IO port access, swap
3731 management, and kexec. Most importantly though
3732 open_by_handle_at() is now prohibited for containers,
3733 closing a hole similar to a recently discussed vulnerability
3734 in docker regarding access to files on file hierarchies the
3735 container should normally not have access to. Note that, for
3736 nspawn, we generally make no security claims anyway (and
3737 this is explicitly documented in the man page), so this is
3738 just a fix for one of the most obvious problems.
3740 * A new man page file-hierarchy(7) has been added that
3741 contains a minimized, modernized version of the file system
3742 layout systemd expects, similar in style to the FHS
3743 specification or hier(5). A new tool systemd-path(1) has
3744 been added to query many of these paths for the local
3747 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
3748 longer done. Since the directory now has a per-user size
3749 limit, and is cleaned on logout this appears unnecessary,
3750 in particular since this now brings the lifecycle of this
3751 directory closer in line with how IPC objects are handled.
3753 * systemd.pc now exports a number of additional directories,
3754 including $libdir (which is useful to identify the library
3755 path for the primary architecture of the system), and a
3756 couple of drop-in directories.
3758 * udev's predictable network interface names now use the dev_port
3759 sysfs attribute, introduced in linux 3.15 instead of dev_id to
3760 distinguish between ports of the same PCI function. dev_id should
3761 only be used for ports using the same HW address, hence the need
3764 * machined has been updated to export the OS version of a
3765 container (read from /etc/os-release and
3766 /usr/lib/os-release) on the bus. This is now shown in
3767 "machinectl status" for a machine.
3769 * A new service setting RestartForceExitStatus= has been
3770 added. If configured to a set of exit signals or process
3771 return values, the service will be restarted when the main
3772 daemon process exits with any of them, regardless of the
3775 * systemctl's -H switch for connecting to remote systemd
3776 machines has been extended so that it may be used to
3777 directly connect to a specific container on the
3778 host. "systemctl -H root@foobar:waldi" will now connect as
3779 user "root" to host "foobar", and then proceed directly to
3780 the container named "waldi". Note that currently you have to
3781 authenticate as user "root" for this to work, as entering
3782 containers is a privileged operation.
3784 Contributions from: Andreas Henriksson, Benjamin Steinwender,
3785 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
3786 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
3787 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
3788 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
3789 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
3790 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
3791 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
3792 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
3793 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
3794 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
3795 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
3797 — Berlin, 2014-07-03
3801 * As an experimental feature, udev now tries to lock the
3802 disk device node (flock(LOCK_SH|LOCK_NB)) while it
3803 executes events for the disk or any of its partitions.
3804 Applications like partitioning programs can lock the
3805 disk device node (flock(LOCK_EX)) and claim temporary
3806 device ownership that way; udev will entirely skip all event
3807 handling for this disk and its partitions. If the disk
3808 was opened for writing, the close will trigger a partition
3809 table rescan in udev's "watch" facility, and if needed
3810 synthesize "change" events for the disk and all its partitions.
3811 This is now unconditionally enabled, and if it turns out to
3812 cause major problems, we might turn it on only for specific
3813 devices, or might need to disable it entirely. Device Mapper
3814 devices are excluded from this logic.
3816 * We temporarily dropped the "-l" switch for fsck invocations,
3817 since they collide with the flock() logic above. util-linux
3818 upstream has been changed already to avoid this conflict,
3819 and we will readd "-l" as soon as util-linux with this
3820 change has been released.
3822 * The dependency on libattr has been removed. Since a long
3823 time, the extended attribute calls have moved to glibc, and
3824 libattr is thus unnecessary.
3826 * Virtualization detection works without privileges now. This
3827 means the systemd-detect-virt binary no longer requires
3828 CAP_SYS_PTRACE file capabilities, and our daemons can run
3829 with fewer privileges.
3831 * systemd-networkd now runs under its own "systemd-network"
3832 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
3833 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
3834 loses the ability to write to files owned by root this way.
3836 * Similarly, systemd-resolved now runs under its own
3837 "systemd-resolve" user with no capabilities remaining.
3839 * Similarly, systemd-bus-proxyd now runs under its own
3840 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
3842 * systemd-networkd gained support for setting up "veth"
3843 virtual Ethernet devices for container connectivity, as well
3844 as GRE and VTI tunnels.
3846 * systemd-networkd will no longer automatically attempt to
3847 manually load kernel modules necessary for certain tunnel
3848 transports. Instead, it is assumed the kernel loads them
3849 automatically when required. This only works correctly on
3850 very new kernels. On older kernels, please consider adding
3851 the kernel modules to /etc/modules-load.d/ as a work-around.
3853 * The resolv.conf file systemd-resolved generates has been
3854 moved to /run/systemd/resolve/. If you have a symlink from
3855 /etc/resolv.conf, it might be necessary to correct it.
3857 * Two new service settings, ProtectHome= and ProtectSystem=,
3858 have been added. When enabled, they will make the user data
3859 (such as /home) inaccessible or read-only and the system
3860 (such as /usr) read-only, for specific services. This allows
3861 very light-weight per-service sandboxing to avoid
3862 modifications of user data or system files from
3863 services. These two new switches have been enabled for all
3864 of systemd's long-running services, where appropriate.
3866 * Socket units gained new SocketUser= and SocketGroup=
3867 settings to set the owner user and group of AF_UNIX sockets
3868 and FIFOs in the file system.
3870 * Socket units gained a new RemoveOnStop= setting. If enabled,
3871 all FIFOS and sockets in the file system will be removed
3872 when the specific socket unit is stopped.
3874 * Socket units gained a new Symlinks= setting. It takes a list
3875 of symlinks to create to file system sockets or FIFOs
3876 created by the specific Unix sockets. This is useful to
3877 manage symlinks to socket nodes with the same life-cycle as
3880 * The /dev/log socket and /dev/initctl FIFO have been moved to
3881 /run, and have been replaced by symlinks. This allows
3882 connecting to these facilities even if PrivateDevices=yes is
3883 used for a service (which makes /dev/log itself unavailable,
3884 but /run is left). This also has the benefit of ensuring
3885 that /dev only contains device nodes, directories and
3886 symlinks, and nothing else.
3888 * sd-daemon gained two new calls sd_pid_notify() and
3889 sd_pid_notifyf(). They are similar to sd_notify() and
3890 sd_notifyf(), but allow overriding of the source PID of
3891 notification messages if permissions permit this. This is
3892 useful to send notify messages on behalf of a different
3893 process (for example, the parent process). The
3894 systemd-notify tool has been updated to make use of this
3895 when sending messages (so that notification messages now
3896 originate from the shell script invoking systemd-notify and
3897 not the systemd-notify process itself. This should minimize
3898 a race where systemd fails to associate notification
3899 messages to services when the originating process already
3902 * A new "on-abnormal" setting for Restart= has been added. If
3903 set, it will result in automatic restarts on all "abnormal"
3904 reasons for a process to exit, which includes unclean
3905 signals, core dumps, timeouts and watchdog timeouts, but
3906 does not include clean and unclean exit codes or clean
3907 signals. Restart=on-abnormal is an alternative for
3908 Restart=on-failure for services that shall be able to
3909 terminate and avoid restarts on certain errors, by
3910 indicating so with an unclean exit code. Restart=on-failure
3911 or Restart=on-abnormal is now the recommended setting for
3912 all long-running services.
3914 * If the InaccessibleDirectories= service setting points to a
3915 mount point (or if there are any submounts contained within
3916 it), it is now attempted to completely unmount it, to make
3917 the file systems truly unavailable for the respective
3920 * The ReadOnlyDirectories= service setting and
3921 systemd-nspawn's --read-only parameter are now recursively
3922 applied to all submounts, too.
3924 * Mount units may now be created transiently via the bus APIs.
3926 * The support for SysV and LSB init scripts has been removed
3927 from the systemd daemon itself. Instead, it is now
3928 implemented as a generator that creates native systemd units
3929 from these scripts when needed. This enables us to remove a
3930 substantial amount of legacy code from PID 1, following the
3931 fact that many distributions only ship a very small number
3932 of LSB/SysV init scripts nowadays.
3934 * Privileged Xen (dom0) domains are not considered
3935 virtualization anymore by the virtualization detection
3936 logic. After all, they generally have unrestricted access to
3937 the hardware and usually are used to manage the unprivileged
3940 * systemd-tmpfiles gained a new "C" line type, for copying
3941 files or entire directories.
3943 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
3944 lines. So far, they have been non-globbing versions of the
3945 latter, and have thus been redundant. In future, it is
3946 recommended to only use "z". "m" has hence been removed
3947 from the documentation, even though it stays supported.
3949 * A tmpfiles snippet to recreate the most basic structure in
3950 /var has been added. This is enough to create the /var/run →
3951 /run symlink and create a couple of structural
3952 directories. This allows systems to boot up with an empty or
3953 volatile /var. Of course, while with this change, the core OS
3954 now is capable with dealing with a volatile /var, not all
3955 user services are ready for it. However, we hope that sooner
3956 or later, many service daemons will be changed upstream so
3957 that they are able to automatically create their necessary
3958 directories in /var at boot, should they be missing. This is
3959 the first step to allow state-less systems that only require
3960 the vendor image for /usr to boot.
3962 * systemd-nspawn has gained a new --tmpfs= switch to mount an
3963 empty tmpfs instance to a specific directory. This is
3964 particularly useful for making use of the automatic
3965 reconstruction of /var (see above), by passing --tmpfs=/var.
3967 * Access modes specified in tmpfiles snippets may now be
3968 prefixed with "~", which indicates that they shall be masked
3969 by whether the existing file or directory is currently
3970 writable, readable or executable at all. Also, if specified,
3971 the sgid/suid/sticky bits will be masked for all
3974 * A new passive target unit "network-pre.target" has been
3975 added which is useful for services that shall run before any
3976 network is configured, for example firewall scripts.
3978 * The "floppy" group that previously owned the /dev/fd*
3979 devices is no longer used. The "disk" group is now used
3980 instead. Distributions should probably deprecate usage of
3983 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
3984 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
3985 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
3986 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
3987 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
3988 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
3991 — Berlin, 2014-06-11
3995 * A new "systemd-timesyncd" daemon has been added for
3996 synchronizing the system clock across the network. It
3997 implements an SNTP client. In contrast to NTP
3998 implementations such as chrony or the NTP reference server,
3999 this only implements a client side, and does not bother with
4000 the full NTP complexity, focusing only on querying time from
4001 one remote server and synchronizing the local clock to
4002 it. Unless you intend to serve NTP to networked clients or
4003 want to connect to local hardware clocks, this simple NTP
4004 client should be more than appropriate for most
4005 installations. The daemon runs with minimal privileges, and
4006 has been hooked up with networkd to only operate when
4007 network connectivity is available. The daemon saves the
4008 current clock to disk every time a new NTP sync has been
4009 acquired, and uses this to possibly correct the system clock
4010 early at bootup, in order to accommodate for systems that
4011 lack an RTC such as the Raspberry Pi and embedded devices,
4012 and to make sure that time monotonically progresses on these
4013 systems, even if it is not always correct. To make use of
4014 this daemon, a new system user and group "systemd-timesync"
4015 needs to be created on installation of systemd.
4017 * The queue "seqnum" interface of libudev has been disabled, as
4018 it was generally incompatible with device namespacing as
4019 sequence numbers of devices go "missing" if the devices are
4020 part of a different namespace.
4022 * "systemctl list-timers" and "systemctl list-sockets" gained
4023 a --recursive switch for showing units of these types also
4024 for all local containers, similar in style to the already
4025 supported --recursive switch for "systemctl list-units".
4027 * A new RebootArgument= setting has been added for service
4028 units, which may be used to specify a kernel reboot argument
4029 to use when triggering reboots with StartLimitAction=.
4031 * A new FailureAction= setting has been added for service
4032 units which may be used to specify an operation to trigger
4033 when a service fails. This works similarly to
4034 StartLimitAction=, but unlike it, controls what is done
4035 immediately rather than only after several attempts to
4036 restart the service in question.
4038 * hostnamed got updated to also expose the kernel name,
4039 release, and version on the bus. This is useful for
4040 executing commands like hostnamectl with the -H switch.
4041 systemd-analyze makes use of this to properly display
4042 details when running non-locally.
4044 * The bootchart tool can now show cgroup information in the
4045 graphs it generates.
4047 * The CFS CPU quota cgroup attribute is now exposed for
4048 services. The new CPUQuota= switch has been added for this
4049 which takes a percentage value. Setting this will have the
4050 result that a service may never get more CPU time than the
4051 specified percentage, even if the machine is otherwise idle.
4053 * systemd-networkd learned IPIP and SIT tunnel support.
4055 * LSB init scripts exposing a dependency on $network will now
4056 get a dependency on network-online.target rather than simply
4057 network.target. This should bring LSB handling closer to
4058 what it was on SysV systems.
4060 * A new fsck.repair= kernel option has been added to control
4061 how fsck shall deal with unclean file systems at boot.
4063 * The (.ini) configuration file parser will now silently
4064 ignore sections whose name begins with "X-". This may be
4065 used to maintain application-specific extension sections in unit
4068 * machined gained a new API to query the IP addresses of
4069 registered containers. "machinectl status" has been updated
4070 to show these addresses in its output.
4072 * A new call sd_uid_get_display() has been added to the
4073 sd-login APIs for querying the "primary" session of a
4074 user. The "primary" session of the user is elected from the
4075 user's sessions and generally a graphical session is
4076 preferred over a text one.
4078 * A minimal systemd-resolved daemon has been added. It
4079 currently simply acts as a companion to systemd-networkd and
4080 manages resolv.conf based on per-interface DNS
4081 configuration, possibly supplied via DHCP. In the long run
4082 we hope to extend this into a local DNSSEC enabled DNS and
4085 * The systemd-networkd-wait-online tool is now enabled by
4086 default. It will delay network-online.target until a network
4087 connection has been configured. The tool primarily integrates
4088 with networkd, but will also make a best effort to make sense
4089 of network configuration performed in some other way.
4091 * Two new service options StartupCPUShares= and
4092 StartupBlockIOWeight= have been added that work similarly to
4093 CPUShares= and BlockIOWeight= however only apply during
4094 system startup. This is useful to prioritize certain services
4095 differently during bootup than during normal runtime.
4097 * hostnamed has been changed to prefer the statically
4098 configured hostname in /etc/hostname (unless set to
4099 'localhost' or empty) over any dynamic one supplied by
4100 dhcp. With this change, the rules for picking the hostname
4101 match more closely the rules of other configuration settings
4102 where the local administrator's configuration in /etc always
4103 overrides any other settings.
4105 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
4106 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
4107 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
4108 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
4109 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
4110 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
4111 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
4112 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
4113 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
4114 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
4115 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
4116 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
4117 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
4118 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
4119 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
4120 Lindskog, WaLyong Cho, Will Woods, Zbigniew
4123 — Beijing, 2014-05-28
4127 * When restoring the screen brightness at boot, stay away from
4128 the darkest setting or from the lowest 5% of the available
4129 range, depending on which is the larger value of both. This
4130 should effectively protect the user from rebooting into a
4131 black screen, should the brightness have been set to minimum
4134 * sd-login gained a new sd_machine_get_class() call to
4135 determine the class ("vm" or "container") of a machine
4136 registered with machined.
4138 * sd-login gained new calls
4139 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
4140 to query the identity of the peer of a local AF_UNIX
4141 connection. They operate similarly to their sd_pid_get_xyz()
4144 * PID 1 will now maintain a system-wide system state engine
4145 with the states "starting", "running", "degraded",
4146 "maintenance", "stopping". These states are bound to system
4147 startup, normal runtime, runtime with at least one failed
4148 service, rescue/emergency mode and system shutdown. This
4149 state is shown in the "systemctl status" output when no unit
4150 name is passed. It is useful to determine system state, in
4151 particularly when doing so for many systems or containers at
4154 * A new command "list-machines" has been added to "systemctl"
4155 that lists all local OS containers and shows their system
4156 state (see above), if systemd runs inside of them.
4158 * systemctl gained a new "-r" switch to recursively enumerate
4159 units on all local containers, when used with the
4160 "list-unit" command (which is the default one that is
4161 executed when no parameters are specified).
4163 * The GPT automatic partition discovery logic will now honour
4164 two GPT partition flags: one may be set on a partition to
4165 cause it to be mounted read-only, and the other may be set
4166 on a partition to ignore it during automatic discovery.
4168 * Two new GPT type UUIDs have been added for automatic root
4169 partition discovery, for 32-bit and 64-bit ARM. This is not
4170 particularly useful for discovering the root directory on
4171 these architectures during bare-metal boots (since UEFI is
4172 not common there), but still very useful to allow booting of
4173 ARM disk images in nspawn with the -i option.
4175 * MAC addresses of interfaces created with nspawn's
4176 --network-interface= switch will now be generated from the
4177 machine name, and thus be stable between multiple invocations
4180 * logind will now automatically remove all IPC objects owned
4181 by a user if she or he fully logs out. This makes sure that
4182 users who are logged out cannot continue to consume IPC
4183 resources. This covers SysV memory, semaphores and message
4184 queues as well as POSIX shared memory and message
4185 queues. Traditionally, SysV and POSIX IPC had no life-cycle
4186 limits. With this functionality, that is corrected. This may
4187 be turned off by using the RemoveIPC= switch of logind.conf.
4189 * The systemd-machine-id-setup and tmpfiles tools gained a
4190 --root= switch to operate on a specific root directory,
4193 * journald can now forward logged messages to the TTYs of all
4194 logged in users ("wall"). This is the default for all
4195 emergency messages now.
4197 * A new tool systemd-journal-remote has been added to stream
4198 journal log messages across the network.
4200 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
4201 controller trees are mounted into it. Note that the
4202 directories mounted beneath it are not read-only. This is a
4203 security measure and is particularly useful because glibc
4204 actually includes a search logic to pick any tmpfs it can
4205 find to implement shm_open() if /dev/shm is not available
4206 (which it might very well be in namespaced setups).
4208 * machinectl gained a new "poweroff" command to cleanly power
4209 down a local OS container.
4211 * The PrivateDevices= unit file setting will now also drop the
4212 CAP_MKNOD capability from the capability bound set, and
4213 imply DevicePolicy=closed.
4215 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
4216 comprehensively on all long-running systemd services where
4217 this is appropriate.
4219 * systemd-udevd will now run in a disassociated mount
4220 namespace. To mount directories from udev rules, make sure to
4221 pull in mount units via SYSTEMD_WANTS properties.
4223 * The kdbus support gained support for uploading policy into
4224 the kernel. sd-bus gained support for creating "monitoring"
4225 connections that can eavesdrop into all bus communication
4226 for debugging purposes.
4228 * Timestamps may now be specified in seconds since the UNIX
4229 epoch Jan 1st, 1970 by specifying "@" followed by the value
4232 * Native tcpwrap support in systemd has been removed. tcpwrap
4233 is old code, not really maintained anymore and has serious
4234 shortcomings, and better options such as firewalls
4235 exist. For setups that require tcpwrap usage, please
4236 consider invoking your socket-activated service via tcpd,
4237 like on traditional inetd.
4239 * A new system.conf configuration option
4240 DefaultTimerAccuracySec= has been added that controls the
4241 default AccuracySec= setting of .timer units.
4243 * Timer units gained a new WakeSystem= switch. If enabled,
4244 timers configured this way will cause the system to resume
4245 from system suspend (if the system supports that, which most
4248 * Timer units gained a new Persistent= switch. If enabled,
4249 timers configured this way will save to disk when they have
4250 been last triggered. This information is then used on next
4251 reboot to possible execute overdue timer events, that
4252 could not take place because the system was powered off.
4253 This enables simple anacron-like behaviour for timer units.
4255 * systemctl's "list-timers" will now also list the time a
4256 timer unit was last triggered in addition to the next time
4257 it will be triggered.
4259 * systemd-networkd will now assign predictable IPv4LL
4260 addresses to its local interfaces.
4262 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
4263 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
4264 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
4265 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
4266 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
4267 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
4268 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
4269 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
4272 — Berlin, 2014-03-25
4276 * A new unit file setting RestrictAddressFamilies= has been
4277 added to restrict which socket address families unit
4278 processes gain access to. This takes address family names
4279 like "AF_INET" or "AF_UNIX", and is useful to minimize the
4280 attack surface of services via exotic protocol stacks. This
4281 is built on seccomp system call filters.
4283 * Two new unit file settings RuntimeDirectory= and
4284 RuntimeDirectoryMode= have been added that may be used to
4285 manage a per-daemon runtime directories below /run. This is
4286 an alternative for setting up directory permissions with
4287 tmpfiles snippets, and has the advantage that the runtime
4288 directory's lifetime is bound to the daemon runtime and that
4289 the daemon starts up with an empty directory each time. This
4290 is particularly useful when writing services that drop
4291 privileges using the User= or Group= setting.
4293 * The DeviceAllow= unit setting now supports globbing for
4294 matching against device group names.
4296 * The systemd configuration file system.conf gained new
4297 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
4298 DefaultMemoryAccounting= to globally turn on/off accounting
4299 for specific resources (cgroups) for all units. These
4300 settings may still be overridden individually in each unit
4303 * systemd-gpt-auto-generator is now able to discover /srv and
4304 root partitions in addition to /home and swap partitions. It
4305 also supports LUKS-encrypted partitions now. With this in
4306 place, automatic discovery of partitions to mount following
4307 the Discoverable Partitions Specification
4308 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
4309 is now a lot more complete. This allows booting without
4310 /etc/fstab and without root= on the kernel command line on
4311 systems prepared appropriately.
4313 * systemd-nspawn gained a new --image= switch which allows
4314 booting up disk images and Linux installations on any block
4315 device that follow the Discoverable Partitions Specification
4316 (see above). This means that installations made with
4317 appropriately updated installers may now be started and
4318 deployed using container managers, completely
4319 unmodified. (We hope that libvirt-lxc will add support for
4320 this feature soon, too.)
4322 * systemd-nspawn gained a new --network-macvlan= setting to
4323 set up a private macvlan interface for the
4324 container. Similarly, systemd-networkd gained a new
4325 Kind=macvlan setting in .netdev files.
4327 * systemd-networkd now supports configuring local addresses
4330 * A new tool systemd-network-wait-online has been added to
4331 synchronously wait for network connectivity using
4334 * The sd-bus.h bus API gained a new sd_bus_track object for
4335 tracking the life-cycle of bus peers. Note that sd-bus.h is
4336 still not a public API though (unless you specify
4337 --enable-kdbus on the configure command line, which however
4338 voids your warranty and you get no API stability guarantee).
4340 * The $XDG_RUNTIME_DIR runtime directories for each user are
4341 now individual tmpfs instances, which has the benefit of
4342 introducing separate pools for each user, with individual
4343 size limits, and thus making sure that unprivileged clients
4344 can no longer negatively impact the system or other users by
4345 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
4346 RuntimeDirectorySize= has been introduced that allows
4347 controlling the default size limit for all users. It
4348 defaults to 10% of the available physical memory. This is no
4349 replacement for quotas on tmpfs though (which the kernel
4350 still does not support), as /dev/shm and /tmp are still
4351 shared resources used by both the system and unprivileged
4354 * logind will now automatically turn off automatic suspending
4355 on laptop lid close when more than one display is
4356 connected. This was previously expected to be implemented
4357 individually in desktop environments (such as GNOME),
4358 however has been added to logind now, in order to fix a
4359 boot-time race where a desktop environment might not have
4360 been started yet and thus not been able to take an inhibitor
4361 lock at the time where logind already suspends the system
4362 due to a closed lid.
4364 * logind will now wait at least 30s after each system
4365 suspend/resume cycle, and 3min after system boot before
4366 suspending the system due to a closed laptop lid. This
4367 should give USB docking stations and similar enough time to
4368 be probed and configured after system resume and boot in
4369 order to then act as suspend blocker.
4371 * systemd-run gained a new --property= setting which allows
4372 initialization of resource control properties (and others)
4373 for the created scope or service unit. Example: "systemd-run
4374 --property=BlockIOWeight=10 updatedb" may be used to run
4375 updatedb at a low block IO scheduling weight.
4377 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
4378 now also work in --scope mode.
4380 * When systemd is compiled with kdbus support, basic support
4381 for enforced policies is now in place. (Note that enabling
4382 kdbus still voids your warranty and no API compatibility
4385 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
4386 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
4387 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
4388 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
4389 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
4390 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
4391 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
4392 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
4393 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
4394 Zbigniew Jędrzejewski-Szmek
4396 — Berlin, 2014-03-12
4400 * systemd will now relabel /dev after loading the SMACK policy
4401 according to SMACK rules.
4403 * A new unit file option AppArmorProfile= has been added to
4404 set the AppArmor profile for the processes of a unit.
4406 * A new condition check ConditionArchitecture= has been added
4407 to conditionalize units based on the system architecture, as
4408 reported by uname()'s "machine" field.
4410 * systemd-networkd now supports matching on the system
4411 virtualization, architecture, kernel command line, host name
4414 * logind is now a lot more aggressive when suspending the
4415 machine due to a closed laptop lid. Instead of acting only
4416 on the lid close action, it will continuously watch the lid
4417 status and act on it. This is useful for laptops where the
4418 power button is on the outside of the chassis so that it can
4419 be reached without opening the lid (such as the Lenovo
4420 Yoga). On those machines, logind will now immediately
4421 re-suspend the machine if the power button has been
4422 accidentally pressed while the laptop was suspended and in a
4423 backpack or similar.
4425 * logind will now watch SW_DOCK switches and inhibit reaction
4426 to the lid switch if it is pressed. This means that logind
4427 will not suspend the machine anymore if the lid is closed
4428 and the system is docked, if the laptop supports SW_DOCK
4429 notifications via the input layer. Note that ACPI docking
4430 stations do not generate this currently. Also note that this
4431 logic is usually not fully sufficient and Desktop
4432 Environments should take a lid switch inhibitor lock when an
4433 external display is connected, as systemd will not watch
4436 * nspawn will now make use of the devices cgroup controller by
4437 default, and only permit creation of and access to the usual
4438 API device nodes like /dev/null or /dev/random, as well as
4439 access to (but not creation of) the pty devices.
4441 * We will now ship a default .network file for
4442 systemd-networkd that automatically configures DHCP for
4443 network interfaces created by nspawn's --network-veth or
4444 --network-bridge= switches.
4446 * systemd will now understand the usual M, K, G, T suffixes
4447 according to SI conventions (i.e. to the base 1000) when
4448 referring to throughput and hardware metrics. It will stay
4449 with IEC conventions (i.e. to the base 1024) for software
4450 metrics, according to what is customary according to
4451 Wikipedia. We explicitly document which base applies for
4452 each configuration option.
4454 * The DeviceAllow= setting in unit files now supports a syntax
4455 to whitelist an entire group of devices node majors at once,
4456 based on the /proc/devices listing. For example, with the
4457 string "char-pts", it is now possible to whitelist all
4458 current and future pseudo-TTYs at once.
4460 * sd-event learned a new "post" event source. Event sources of
4461 this type are triggered by the dispatching of any event
4462 source of a type that is not "post". This is useful for
4463 implementing clean-up and check event sources that are
4464 triggered by other work being done in the program.
4466 * systemd-networkd is no longer statically enabled, but uses
4467 the usual [Install] sections so that it can be
4468 enabled/disabled using systemctl. It still is enabled by
4471 * When creating a veth interface pair with systemd-nspawn, the
4472 host side will now be prefixed with "vb-" if
4473 --network-bridge= is used, and with "ve-" if --network-veth
4474 is used. This way, it is easy to distinguish these cases on
4475 the host, for example to apply different configuration to
4476 them with systemd-networkd.
4478 * The compatibility libraries for libsystemd-journal.so,
4479 libsystem-id128.so, libsystemd-login.so and
4480 libsystemd-daemon.so do not make use of IFUNC
4481 anymore. Instead, we now build libsystemd.so multiple times
4482 under these alternative names. This means that the footprint
4483 is drastically increased, but given that these are
4484 transitional compatibility libraries, this should not matter
4485 much. This change has been made necessary to support the ARM
4486 platform for these compatibility libraries, as the ARM
4487 toolchain is not really at the same level as the toolchain
4488 for other architectures like x86 and does not support
4489 IFUNC. Please make sure to use --enable-compat-libs only
4490 during a transitional period!
4492 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
4493 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
4494 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
4495 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
4496 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
4497 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
4498 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
4499 Zbigniew Jędrzejewski-Szmek
4501 — Berlin, 2014-02-24
4505 * A new component "systemd-networkd" has been added that can
4506 be used to configure local network interfaces statically or
4507 via DHCP. It is capable of bringing up bridges, VLANs, and
4508 bonding. Currently, no hook-ups for interactive network
4509 configuration are provided. Use this for your initrd,
4510 container, embedded, or server setup if you need a simple,
4511 yet powerful, network configuration solution. This
4512 configuration subsystem is quite nifty, as it allows wildcard
4513 hotplug matching in interfaces. For example, with a single
4514 configuration snippet, you can configure that all Ethernet
4515 interfaces showing up are automatically added to a bridge,
4516 or similar. It supports link-sensing and more.
4518 * A new tool "systemd-socket-proxyd" has been added which can
4519 act as a bidirectional proxy for TCP sockets. This is
4520 useful for adding socket activation support to services that
4521 do not actually support socket activation, including virtual
4522 machines and the like.
4524 * Add a new tool to save/restore rfkill state on
4527 * Save/restore state of keyboard backlights in addition to
4528 display backlights on shutdown/boot.
4530 * udev learned a new SECLABEL{} construct to label device
4531 nodes with a specific security label when they appear. For
4532 now, only SECLABEL{selinux} is supported, but the syntax is
4533 prepared for additional security frameworks.
4535 * udev gained a new scheme to configure link-level attributes
4536 from files in /etc/systemd/network/*.link. These files can
4537 match against MAC address, device path, driver name and type,
4538 and will apply attributes like the naming policy, link speed,
4539 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
4540 address assignment policy (randomized, ...).
4542 * The configuration of network interface naming rules for
4543 "permanent interface names" has changed: a new NamePolicy=
4544 setting in the [Link] section of .link files determines the
4545 priority of possible naming schemes (onboard, slot, MAC,
4546 path). The default value of this setting is determined by
4547 /usr/lib/net/links/99-default.link. Old
4548 80-net-name-slot.rules udev configuration file has been
4549 removed, so local configuration overriding this file should
4550 be adapted to override 99-default.link instead.
4552 * When the User= switch is used in a unit file, also
4553 initialize $SHELL= based on the user database entry.
4555 * systemd no longer depends on libdbus. All communication is
4556 now done with sd-bus, systemd's low-level bus library
4559 * kdbus support has been added to PID 1 itself. When kdbus is
4560 enabled, this causes PID 1 to set up the system bus and
4561 enable support for a new ".busname" unit type that
4562 encapsulates bus name activation on kdbus. It works a little
4563 bit like ".socket" units, except for bus names. A new
4564 generator has been added that converts classic dbus1 service
4565 activation files automatically into native systemd .busname
4568 * sd-bus: add a light-weight vtable implementation that allows
4569 defining objects on the bus with a simple static const
4570 vtable array of its methods, signals and properties.
4572 * systemd will not generate or install static dbus
4573 introspection data anymore to /usr/share/dbus-1/interfaces,
4574 as the precise format of these files is unclear, and
4575 nothing makes use of it.
4577 * A proxy daemon is now provided to proxy clients connecting
4578 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
4579 compatibility with classic D-Bus.
4581 * A bus driver implementation has been added that supports the
4582 classic D-Bus bus driver calls on kdbus, also for
4583 compatibility purposes.
4585 * A new API "sd-event.h" has been added that implements a
4586 minimal event loop API built around epoll. It provides a
4587 couple of features that direct epoll usage is lacking:
4588 prioritization of events, scales to large numbers of timer
4589 events, per-event timer slack (accuracy), system-wide
4590 coalescing of timer events, exit handlers, watchdog
4591 supervision support using systemd's sd_notify() API, child
4594 * A new API "sd-rntl.h" has been added that provides an API
4595 around the route netlink interface of the kernel, similar in
4596 style to "sd-bus.h".
4598 * A new API "sd-dhcp-client.h" has been added that provides a
4599 small DHCPv4 client-side implementation. This is used by
4602 * There is a new kernel command line option
4603 "systemd.restore_state=0|1". When set to "0", none of the
4604 systemd tools will restore saved runtime state to hardware
4605 devices. More specifically, the rfkill and backlight states
4608 * The FsckPassNo= compatibility option in mount/service units
4609 has been removed. The fstab generator will now add the
4610 necessary dependencies automatically, and does not require
4611 PID1's support for that anymore.
4613 * journalctl gained a new switch, --list-boots, that lists
4614 recent boots with their times and boot IDs.
4616 * The various tools like systemctl, loginctl, timedatectl,
4617 busctl, systemd-run, ... have gained a new switch "-M" to
4618 connect to a specific, local OS container (as direct
4619 connection, without requiring SSH). This works on any
4620 container that is registered with machined, such as those
4621 created by libvirt-lxc or nspawn.
4623 * systemd-run and systemd-analyze also gained support for "-H"
4624 to connect to remote hosts via SSH. This is particularly
4625 useful for systemd-run because it enables queuing of jobs
4626 onto remote systems.
4628 * machinectl gained a new command "login" to open a getty
4629 login in any local container. This works with any container
4630 that is registered with machined (such as those created by
4631 libvirt-lxc or nspawn), and which runs systemd inside.
4633 * machinectl gained a new "reboot" command that may be used to
4634 trigger a reboot on a specific container that is registered
4635 with machined. This works on any container that runs an init
4636 system of some kind.
4638 * systemctl gained a new "list-timers" command to print a nice
4639 listing of installed timer units with the times they elapse
4642 * Alternative reboot() parameters may now be specified on the
4643 "systemctl reboot" command line and are passed to the
4644 reboot() system call.
4646 * systemctl gained a new --job-mode= switch to configure the
4647 mode to queue a job with. This is a more generic version of
4648 --fail, --irreversible, and --ignore-dependencies, which are
4649 still available but not advertised anymore.
4651 * /etc/systemd/system.conf gained new settings to configure
4652 various default timeouts of units, as well as the default
4653 start limit interval and burst. These may still be overridden
4656 * PID1 will now export on the bus profile data of the security
4657 policy upload process (such as the SELinux policy upload to
4660 * journald: when forwarding logs to the console, include
4661 timestamps (following the setting in
4662 /sys/module/printk/parameters/time).
4664 * OnCalendar= in timer units now understands the special
4665 strings "yearly" and "annually". (Both are equivalent)
4667 * The accuracy of timer units is now configurable with the new
4668 AccuracySec= setting. It defaults to 1min.
4670 * A new dependency type JoinsNamespaceOf= has been added that
4671 allows running two services within the same /tmp and network
4672 namespace, if PrivateNetwork= or PrivateTmp= are used.
4674 * A new command "cat" has been added to systemctl. It outputs
4675 the original unit file of a unit, and concatenates the
4676 contents of additional "drop-in" unit file snippets, so that
4677 the full configuration is shown.
4679 * systemctl now supports globbing on the various "list-xyz"
4680 commands, like "list-units" or "list-sockets", as well as on
4681 those commands which take multiple unit names.
4683 * journalctl's --unit= switch gained support for globbing.
4685 * All systemd daemons now make use of the watchdog logic so
4686 that systemd automatically notices when they hang.
4688 * If the $container_ttys environment variable is set,
4689 getty-generator will automatically spawn a getty for each
4690 listed tty. This is useful for container managers to request
4691 login gettys to be spawned on as many ttys as needed.
4693 * %h, %s, %U specifier support is not available anymore when
4694 used in unit files for PID 1. This is because NSS calls are
4695 not safe from PID 1. They stay available for --user
4696 instances of systemd, and as special case for the root user.
4698 * loginctl gained a new "--no-legend" switch to turn off output
4701 * The "sd-login.h" API gained three new calls:
4702 sd_session_is_remote(), sd_session_get_remote_user(),
4703 sd_session_get_remote_host() to query information about
4706 * The udev hardware database now also carries vendor/product
4707 information of SDIO devices.
4709 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
4710 determine whether watchdog notifications are requested by
4713 * Socket-activated per-connection services now include a
4714 short description of the connection parameters in the
4717 * tmpfiles gained a new "--boot" option. When this is not used,
4718 only lines where the command character is not suffixed with
4719 "!" are executed. When this option is specified, those
4720 options are executed too. This partitions tmpfiles
4721 directives into those that can be safely executed at any
4722 time, and those which should be run only at boot (for
4723 example, a line that creates /run/nologin).
4725 * A new API "sd-resolve.h" has been added which provides a simple
4726 asynchronous wrapper around glibc NSS host name resolution
4727 calls, such as getaddrinfo(). In contrast to glibc's
4728 getaddrinfo_a(), it does not use signals. In contrast to most
4729 other asynchronous name resolution libraries, this one does
4730 not reimplement DNS, but reuses NSS, so that alternate
4731 host name resolution systems continue to work, such as mDNS,
4732 LDAP, etc. This API is based on libasyncns, but it has been
4733 cleaned up for inclusion in systemd.
4735 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
4736 "sd-daemon.h" are no longer found in individual libraries
4737 libsystemd-journal.so, libsystemd-login.so,
4738 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
4739 merged them into a single library, libsystemd.so, which
4740 provides all symbols. The reason for this is cyclic
4741 dependencies, as these libraries tend to use each other's
4742 symbols. So far, we have managed to workaround that by linking
4743 a copy of a good part of our code into each of these
4744 libraries again and again, which, however, makes certain
4745 things hard to do, like sharing static variables. Also, it
4746 substantially increases footprint. With this change, there
4747 is only one library for the basic APIs systemd
4748 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
4749 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
4750 library as well, however are subject to the --enable-kdbus
4751 switch (see below). Note that "sd-dhcp-client.h" is not part
4752 of this library (this is because it only consumes, never
4753 provides, services of/to other APIs). To make the transition
4754 easy from the separate libraries to the unified one, we
4755 provide the --enable-compat-libs compile-time switch which
4756 will generate stub libraries that are compatible with the
4757 old ones but redirect all calls to the new one.
4759 * All of the kdbus logic and the new APIs "sd-bus.h",
4760 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
4761 and "sd-utf8.h" are compile-time optional via the
4762 "--enable-kdbus" switch, and they are not compiled in by
4763 default. To make use of kdbus, you have to explicitly enable
4764 the switch. Note however, that neither the kernel nor the
4765 userspace API for all of this is considered stable yet. We
4766 want to maintain the freedom to still change the APIs for
4767 now. By specifying this build-time switch, you acknowledge
4768 that you are aware of the instability of the current
4771 * Also, note that while kdbus is pretty much complete,
4772 it lacks one thing: proper policy support. This means you
4773 can build a fully working system with all features; however,
4774 it will be highly insecure. Policy support will be added in
4775 one of the next releases, at the same time that we will
4776 declare the APIs stable.
4778 * When the kernel command line argument "kdbus" is specified,
4779 systemd will automatically load the kdbus.ko kernel module. At
4780 this stage of development, it is only useful for testing kdbus
4781 and should not be used in production. Note: if "--enable-kdbus"
4782 is specified, and the kdbus.ko kernel module is available, and
4783 "kdbus" is added to the kernel command line, the entire system
4784 runs with kdbus instead of dbus-daemon, with the above mentioned
4785 problem of missing the system policy enforcement. Also a future
4786 version of kdbus.ko or a newer systemd will not be compatible with
4787 each other, and will unlikely be able to boot the machine if only
4788 one of them is updated.
4790 * systemctl gained a new "import-environment" command which
4791 uploads the caller's environment (or parts thereof) into the
4792 service manager so that it is inherited by services started
4793 by the manager. This is useful to upload variables like
4794 $DISPLAY into the user service manager.
4796 * A new PrivateDevices= switch has been added to service units
4797 which allows running a service with a namespaced /dev
4798 directory that does not contain any device nodes for
4799 physical devices. More specifically, it only includes devices
4800 such as /dev/null, /dev/urandom, and /dev/zero which are API
4803 * logind has been extended to support behaviour like VT
4804 switching on seats that do not support a VT. This makes
4805 multi-session available on seats that are not the first seat
4806 (seat0), and on systems where kernel support for VTs has
4807 been disabled at compile-time.
4809 * If a process holds a delay lock for system sleep or shutdown
4810 and fails to release it in time, we will now log its
4811 identity. This makes it easier to identify processes that
4812 cause slow suspends or power-offs.
4814 * When parsing /etc/crypttab, support for a new key-slot=
4815 option as supported by Debian is added. It allows indicating
4816 which LUKS slot to use on disk, speeding up key loading.
4818 * The sd_journald_sendv() API call has been checked and
4819 officially declared to be async-signal-safe so that it may
4820 be invoked from signal handlers for logging purposes.
4822 * Boot-time status output is now enabled automatically after a
4823 short timeout if boot does not progress, in order to give
4824 the user an indication what she or he is waiting for.
4826 * The boot-time output has been improved to show how much time
4827 remains until jobs expire.
4829 * The KillMode= switch in service units gained a new possible
4830 value "mixed". If set, and the unit is shut down, then the
4831 initial SIGTERM signal is sent only to the main daemon
4832 process, while the following SIGKILL signal is sent to
4833 all remaining processes of the service.
4835 * When a scope unit is registered, a new property "Controller"
4836 may be set. If set to a valid bus name, systemd will send a
4837 RequestStop() signal to this name when it would like to shut
4838 down the scope. This may be used to hook manager logic into
4839 the shutdown logic of scope units. Also, scope units may now
4840 be put in a special "abandoned" state, in which case the
4841 manager process which created them takes no further
4842 responsibilities for it.
4844 * When reading unit files, systemd will now verify
4845 the access mode of these files, and warn about certain
4846 suspicious combinations. This has been added to make it
4847 easier to track down packaging bugs where unit files are
4848 marked executable or world-writable.
4850 * systemd-nspawn gained a new "--setenv=" switch to set
4851 container-wide environment variables. The similar option in
4852 systemd-activate was renamed from "--environment=" to
4853 "--setenv=" for consistency.
4855 * systemd-nspawn has been updated to create a new kdbus domain
4856 for each container that is invoked, thus allowing each
4857 container to have its own set of system and user buses,
4858 independent of the host.
4860 * systemd-nspawn gained a new --drop-capability= switch to run
4861 the container with less capabilities than the default. Both
4862 --drop-capability= and --capability= now take the special
4863 string "all" for dropping or keeping all capabilities.
4865 * systemd-nspawn gained new switches for executing containers
4866 with specific SELinux labels set.
4868 * systemd-nspawn gained a new --quiet switch to not generate
4869 any additional output but the container's own console
4872 * systemd-nspawn gained a new --share-system switch to run a
4873 container without PID namespacing enabled.
4875 * systemd-nspawn gained a new --register= switch to control
4876 whether the container is registered with systemd-machined or
4877 not. This is useful for containers that do not run full
4878 OS images, but only specific apps.
4880 * systemd-nspawn gained a new --keep-unit which may be used
4881 when invoked as the only program from a service unit, and
4882 results in registration of the unit service itself in
4883 systemd-machined, instead of a newly opened scope unit.
4885 * systemd-nspawn gained a new --network-interface= switch for
4886 moving arbitrary interfaces to the container. The new
4887 --network-veth switch creates a virtual Ethernet connection
4888 between host and container. The new --network-bridge=
4889 switch then allows assigning the host side of this virtual
4890 Ethernet connection to a bridge device.
4892 * systemd-nspawn gained a new --personality= switch for
4893 setting the kernel personality for the container. This is
4894 useful when running a 32-bit container on a 64-bit host. A
4895 similar option Personality= is now also available for service
4898 * logind will now also track a "Desktop" identifier for each
4899 session which encodes the desktop environment of it. This is
4900 useful for desktop environments that want to identify
4901 multiple running sessions of itself easily.
4903 * A new SELinuxContext= setting for service units has been
4904 added that allows setting a specific SELinux execution
4905 context for a service.
4907 * Most systemd client tools will now honour $SYSTEMD_LESS for
4908 settings of the "less" pager. By default, these tools will
4909 override $LESS to allow certain operations to work, such as
4910 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
4911 influence this logic.
4913 * systemd's "seccomp" hook-up has been changed to make use of
4914 the libseccomp library instead of using its own
4915 implementation. This has benefits for portability among
4918 * For usage together with SystemCallFilter=, a new
4919 SystemCallErrorNumber= setting has been introduced that
4920 allows configuration of a system error number to be returned
4921 on filtered system calls, instead of immediately killing the
4922 process. Also, SystemCallArchitectures= has been added to
4923 limit access to system calls of a particular architecture
4924 (in order to turn off support for unused secondary
4925 architectures). There is also a global
4926 SystemCallArchitectures= setting in system.conf now to turn
4927 off support for non-native system calls system-wide.
4929 * systemd requires a kernel with a working name_to_handle_at(),
4930 please see the kernel config requirements in the README file.
4932 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
4933 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
4934 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
4935 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
4936 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
4937 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
4938 Elia Pinto, Florian Weimer, George McCollister, Goffredo
4939 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
4940 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
4941 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
4942 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
4943 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
4944 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
4945 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
4946 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
4947 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
4948 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
4949 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
4950 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
4951 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
4952 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
4953 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
4954 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
4955 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
4957 — Berlin, 2014-02-20
4961 * logind has gained support for facilitating privileged input
4962 and drm device access for unprivileged clients. This work is
4963 useful to allow Wayland display servers (and similar
4964 programs, such as kmscon) to run under the user's ID and
4965 access input and drm devices which are normally
4966 protected. When this is used (and the kernel is new enough)
4967 logind will "mute" IO on the file descriptors passed to
4968 Wayland as long as it is in the background and "unmute" it
4969 if it returns into the foreground. This allows secure
4970 session switching without allowing background sessions to
4971 eavesdrop on input and display data. This also introduces
4972 session switching support if VT support is turned off in the
4973 kernel, and on seats that are not seat0.
4975 * A new kernel command line option luks.options= is understood
4976 now which allows specifying LUKS options for usage for LUKS
4977 encrypted partitions specified with luks.uuid=.
4979 * tmpfiles.d(5) snippets may now use specifier expansion in
4980 path names. More specifically %m, %b, %H, %v, are now
4981 replaced by the local machine id, boot id, hostname, and
4982 kernel version number.
4984 * A new tmpfiles.d(5) command "m" has been introduced which
4985 may be used to change the owner/group/access mode of a file
4986 or directory if it exists, but do nothing if it does not.
4988 * This release removes high-level support for the
4989 MemorySoftLimit= cgroup setting. The underlying kernel
4990 cgroup attribute memory.soft_limit= is currently badly
4991 designed and likely to be removed from the kernel API in its
4992 current form, hence we should not expose it for now.
4994 * The memory.use_hierarchy cgroup attribute is now enabled for
4995 all cgroups systemd creates in the memory cgroup
4996 hierarchy. This option is likely to be come the built-in
4997 default in the kernel anyway, and the non-hierarchical mode
4998 never made much sense in the intrinsically hierarchical
5001 * A new field _SYSTEMD_SLICE= is logged along with all journal
5002 messages containing the slice a message was generated
5003 from. This is useful to allow easy per-customer filtering of
5004 logs among other things.
5006 * systemd-journald will no longer adjust the group of journal
5007 files it creates to the "systemd-journal" group. Instead we
5008 rely on the journal directory to be owned by the
5009 "systemd-journal" group, and its setgid bit set, so that the
5010 kernel file system layer will automatically enforce that
5011 journal files inherit this group assignment. The reason for
5012 this change is that we cannot allow NSS look-ups from
5013 journald which would be necessary to resolve
5014 "systemd-journal" to a numeric GID, because this might
5015 create deadlocks if NSS involves synchronous queries to
5016 other daemons (such as nscd, or sssd) which in turn are
5017 logging clients of journald and might block on it, which
5018 would then dead lock. A tmpfiles.d(5) snippet included in
5019 systemd will make sure the setgid bit and group are
5020 properly set on the journal directory if it exists on every
5021 boot. However, we recommend adjusting it manually after
5022 upgrades too (or from RPM scriptlets), so that the change is
5023 not delayed until next reboot.
5025 * Backlight and random seed files in /var/lib/ have moved into
5026 the /var/lib/systemd/ directory, in order to centralize all
5027 systemd generated files in one directory.
5029 * Boot time performance measurements (as displayed by
5030 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
5031 performance information if that's available to determine how
5032 much time BIOS and boot loader initialization required. With
5033 a sufficiently new BIOS you hence no longer need to boot
5034 with Gummiboot to get access to such information.
5036 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
5037 Cristian Rodríguez, Dave Reisner, David Herrmann, David
5038 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
5039 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
5040 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
5041 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
5042 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5044 — Berlin, 2013-10-02
5048 * The Restart= option for services now understands a new
5049 on-watchdog setting, which will restart the service
5050 automatically if the service stops sending out watchdog keep
5051 alive messages (as configured with WatchdogSec=).
5053 * The getty generator (which is responsible for bringing up a
5054 getty on configured serial consoles) will no longer only
5055 start a getty on the primary kernel console but on all
5056 others, too. This makes the order in which console= is
5057 specified on the kernel command line less important.
5059 * libsystemd-logind gained a new sd_session_get_vt() call to
5060 retrieve the VT number of a session.
5062 * If the option "tries=0" is set for an entry of /etc/crypttab
5063 its passphrase is queried indefinitely instead of any
5064 maximum number of tries.
5066 * If a service with a configure PID file terminates its PID
5067 file will now be removed automatically if it still exists
5068 afterwards. This should put an end to stale PID files.
5070 * systemd-run will now also take relative binary path names
5071 for execution and no longer insists on absolute paths.
5073 * InaccessibleDirectories= and ReadOnlyDirectories= now take
5074 paths that are optionally prefixed with "-" to indicate that
5075 it should not be considered a failure if they do not exist.
5077 * journalctl -o (and similar commands) now understands a new
5078 output mode "short-precise", it is similar to "short" but
5079 shows timestamps with usec accuracy.
5081 * The option "discard" (as known from Debian) is now
5082 synonymous to "allow-discards" in /etc/crypttab. In fact,
5083 "discard" is preferred now (since it is easier to remember
5086 * Some licensing clean-ups were made, so that more code is now
5087 LGPL-2.1 licensed than before.
5089 * A minimal tool to save/restore the display backlight
5090 brightness across reboots has been added. It will store the
5091 backlight setting as late as possible at shutdown, and
5092 restore it as early as possible during reboot.
5094 * A logic to automatically discover and enable home and swap
5095 partitions on GPT disks has been added. With this in place
5096 /etc/fstab becomes optional for many setups as systemd can
5097 discover certain partitions located on the root disk
5098 automatically. Home partitions are recognized under their
5099 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
5100 partitions are recognized under their GPT type ID
5101 0657fd6da4ab43c484e50933c84b4f4f.
5103 * systemd will no longer pass any environment from the kernel
5104 or initrd to system services. If you want to set an
5105 environment for all services, do so via the kernel command
5106 line systemd.setenv= assignment.
5108 * The systemd-sysctl tool no longer natively reads the file
5109 /etc/sysctl.conf. If desired, the file should be symlinked
5110 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
5111 legacy support by a symlink rather than built-in code, it
5112 also makes the otherwise hidden order of application of the
5113 different files visible. (Note that this partly reverts to a
5114 pre-198 application order of sysctl knobs!)
5116 * The "systemctl set-log-level" and "systemctl dump" commands
5117 have been moved to systemd-analyze.
5119 * systemd-run learned the new --remain-after-exit switch,
5120 which causes the scope unit not to be cleaned up
5121 automatically after the process terminated.
5123 * tmpfiles learned a new --exclude-prefix= switch to exclude
5124 certain paths from operation.
5126 * journald will now automatically flush all messages to disk
5127 as soon as a message at the log level CRIT, ALERT or EMERG
5130 Contributions from: Andrew Cook, Brandon Philips, Christian
5131 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
5132 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
5133 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
5134 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
5135 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
5136 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
5137 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
5138 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
5139 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
5140 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
5141 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
5142 William Giokas, Zbigniew Jędrzejewski-Szmek
5144 — Berlin, 2013-09-13
5148 * The documentation has been updated to cover the various new
5149 concepts introduced with 205.
5151 * Unit files now understand the new %v specifier which
5152 resolves to the kernel version string as returned by "uname
5155 * systemctl now supports filtering the unit list output by
5156 load state, active state and sub state, using the new
5159 * "systemctl status" will now show the results of the
5160 condition checks (like ConditionPathExists= and similar) of
5161 the last start attempts of the unit. They are also logged to
5164 * "journalctl -b" may now be used to look for boot output of a
5165 specific boot. Try "journalctl -b -1" for the previous boot,
5166 but the syntax is substantially more powerful.
5168 * "journalctl --show-cursor" has been added which prints the
5169 cursor string the last shown log line. This may then be used
5170 with the new "journalctl --after-cursor=" switch to continue
5171 browsing logs from that point on.
5173 * "journalctl --force" may now be used to force regeneration
5176 * Creation of "dead" device nodes has been moved from udev
5177 into kmod and tmpfiles. Previously, udev would read the kmod
5178 databases to pre-generate dead device nodes based on meta
5179 information contained in kernel modules, so that these would
5180 be auto-loaded on access rather then at boot. As this
5181 does not really have much to do with the exposing actual
5182 kernel devices to userspace this has always been slightly
5183 alien in the udev codebase. Following the new scheme kmod
5184 will now generate a runtime snippet for tmpfiles from the
5185 module meta information and it now is tmpfiles' job to the
5186 create the nodes. This also allows overriding access and
5187 other parameters for the nodes using the usual tmpfiles
5188 facilities. As side effect this allows us to remove the
5189 CAP_SYS_MKNOD capability bit from udevd entirely.
5191 * logind's device ACLs may now be applied to these "dead"
5192 devices nodes too, thus finally allowing managed access to
5193 devices such as /dev/snd/sequencer without loading the
5194 backing module right-away.
5196 * A new RPM macro has been added that may be used to apply
5197 tmpfiles configuration during package installation.
5199 * systemd-detect-virt and ConditionVirtualization= now can
5200 detect User-Mode-Linux machines (UML).
5202 * journald will now implicitly log the effective capabilities
5203 set of processes in the message metadata.
5205 * systemd-cryptsetup has gained support for TrueCrypt volumes.
5207 * The initrd interface has been simplified (more specifically,
5208 support for passing performance data via environment
5209 variables and fsck results via files in /run has been
5210 removed). These features were non-essential, and are
5211 nowadays available in a much nicer way by having systemd in
5212 the initrd serialize its state and have the hosts systemd
5213 deserialize it again.
5215 * The udev "keymap" data files and tools to apply keyboard
5216 specific mappings of scan to key codes, and force-release
5217 scan code lists have been entirely replaced by a udev
5218 "keyboard" builtin and a hwdb data file.
5220 * systemd will now honour the kernel's "quiet" command line
5221 argument also during late shutdown, resulting in a
5222 completely silent shutdown when used.
5224 * There's now an option to control the SO_REUSEPORT socket
5225 option in .socket units.
5227 * Instance units will now automatically get a per-template
5228 subslice of system.slice unless something else is explicitly
5229 configured. For example, instances of sshd@.service will now
5230 implicitly be placed in system-sshd.slice rather than
5231 system.slice as before.
5233 * Test coverage support may now be enabled at build time.
5235 Contributions from: Dave Reisner, Frederic Crozat, Harald
5236 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
5237 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
5238 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
5239 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
5240 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
5241 Giokas, Zbigniew Jędrzejewski-Szmek
5243 — Berlin, 2013-07-23
5247 * Two new unit types have been introduced:
5249 Scope units are very similar to service units, however, are
5250 created out of pre-existing processes — instead of PID 1
5251 forking off the processes. By using scope units it is
5252 possible for system services and applications to group their
5253 own child processes (worker processes) in a powerful way
5254 which then maybe used to organize them, or kill them
5255 together, or apply resource limits on them.
5257 Slice units may be used to partition system resources in an
5258 hierarchical fashion and then assign other units to them. By
5259 default there are now three slices: system.slice (for all
5260 system services), user.slice (for all user sessions),
5261 machine.slice (for VMs and containers).
5263 Slices and scopes have been introduced primarily in
5264 context of the work to move cgroup handling to a
5265 single-writer scheme, where only PID 1
5266 creates/removes/manages cgroups.
5268 * There's a new concept of "transient" units. In contrast to
5269 normal units these units are created via an API at runtime,
5270 not from configuration from disk. More specifically this
5271 means it is now possible to run arbitrary programs as
5272 independent services, with all execution parameters passed
5273 in via bus APIs rather than read from disk. Transient units
5274 make systemd substantially more dynamic then it ever was,
5275 and useful as a general batch manager.
5277 * logind has been updated to make use of scope and slice units
5278 for managing user sessions. As a user logs in he will get
5279 his own private slice unit, to which all sessions are added
5280 as scope units. We also added support for automatically
5281 adding an instance of user@.service for the user into the
5282 slice. Effectively logind will no longer create cgroup
5283 hierarchies on its own now, it will defer entirely to PID 1
5284 for this by means of scope, service and slice units. Since
5285 user sessions this way become entities managed by PID 1
5286 the output of "systemctl" is now a lot more comprehensive.
5288 * A new mini-daemon "systemd-machined" has been added which
5289 may be used by virtualization managers to register local
5290 VMs/containers. nspawn has been updated accordingly, and
5291 libvirt will be updated shortly. machined will collect a bit
5292 of meta information about the VMs/containers, and assign
5293 them their own scope unit (see above). The collected
5294 meta-data is then made available via the "machinectl" tool,
5295 and exposed in "ps" and similar tools. machined/machinectl
5296 is compile-time optional.
5298 * As discussed earlier, the low-level cgroup configuration
5299 options ControlGroup=, ControlGroupModify=,
5300 ControlGroupPersistent=, ControlGroupAttribute= have been
5301 removed. Please use high-level attribute settings instead as
5302 well as slice units.
5304 * A new bus call SetUnitProperties() has been added to alter
5305 various runtime parameters of a unit. This is primarily
5306 useful to alter cgroup parameters dynamically in a nice way,
5307 but will be extended later on to make more properties
5308 modifiable at runtime. systemctl gained a new set-properties
5309 command that wraps this call.
5311 * A new tool "systemd-run" has been added which can be used to
5312 run arbitrary command lines as transient services or scopes,
5313 while configuring a number of settings via the command
5314 line. This tool is currently very basic, however already
5315 very useful. We plan to extend this tool to even allow
5316 queuing of execution jobs with time triggers from the
5317 command line, similar in fashion to "at".
5319 * nspawn will now inform the user explicitly that kernels with
5320 audit enabled break containers, and suggest the user to turn
5323 * Support for detecting the IMA and AppArmor security
5324 frameworks with ConditionSecurity= has been added.
5326 * journalctl gained a new "-k" switch for showing only kernel
5327 messages, mimicking dmesg output; in addition to "--user"
5328 and "--system" switches for showing only user's own logs
5331 * systemd-delta can now show information about drop-in
5332 snippets extending unit files.
5334 * libsystemd-bus has been substantially updated but is still
5335 not available as public API.
5337 * systemd will now look for the "debug" argument on the kernel
5338 command line and enable debug logging, similar to what
5339 "systemd.log_level=debug" already did before.
5341 * "systemctl set-default", "systemctl get-default" has been
5342 added to configure the default.target symlink, which
5343 controls what to boot into by default.
5345 * "systemctl set-log-level" has been added as a convenient
5346 way to raise and lower systemd logging threshold.
5348 * "systemd-analyze plot" will now show the time the various
5349 generators needed for execution, as well as information
5350 about the unit file loading.
5352 * libsystemd-journal gained a new sd_journal_open_files() call
5353 for opening specific journal files. journactl also gained a
5354 new switch to expose this new functionality. Previously we
5355 only supported opening all files from a directory, or all
5356 files from the system, as opening individual files only is
5357 racy due to journal file rotation.
5359 * systemd gained the new DefaultEnvironment= setting in
5360 /etc/systemd/system.conf to set environment variables for
5363 * If a privileged process logs a journal message with the
5364 OBJECT_PID= field set, then journald will automatically
5365 augment this with additional OBJECT_UID=, OBJECT_GID=,
5366 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
5367 system services want to log events about specific client
5368 processes. journactl/systemctl has been updated to make use
5369 of this information if all log messages regarding a specific
5372 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
5373 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
5374 Reisner, David Coppa, David King, David Strauss, Eelco
5375 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
5376 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
5377 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
5378 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
5379 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
5380 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
5381 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
5382 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
5383 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
5384 Łukasz Stelmach, 장동준
5388 * The Python bindings gained some minimal support for the APIs
5389 exposed by libsystemd-logind.
5391 * ConditionSecurity= gained support for detecting SMACK. Since
5392 this condition already supports SELinux and AppArmor we only
5393 miss IMA for this. Patches welcome!
5395 Contributions from: Karol Lewandowski, Lennart Poettering,
5396 Zbigniew Jędrzejewski-Szmek
5400 * systemd-nspawn will now create /etc/resolv.conf if
5401 necessary, before bind-mounting the host's file onto it.
5403 * systemd-nspawn will now store meta information about a
5404 container on the container's cgroup as extended attribute
5405 fields, including the root directory.
5407 * The cgroup hierarchy has been reworked in many ways. All
5408 objects any of the components systemd creates in the cgroup
5409 tree are now suffixed. More specifically, user sessions are
5410 now placed in cgroups suffixed with ".session", users in
5411 cgroups suffixed with ".user", and nspawn containers in
5412 cgroups suffixed with ".nspawn". Furthermore, all cgroup
5413 names are now escaped in a simple scheme to avoid collision
5414 of userspace object names with kernel filenames. This work
5415 is preparation for making these objects relocatable in the
5416 cgroup tree, in order to allow easy resource partitioning of
5417 these objects without causing naming conflicts.
5419 * systemctl list-dependencies gained the new switches
5420 --plain, --reverse, --after and --before.
5422 * systemd-inhibit now shows the process name of processes that
5423 have taken an inhibitor lock.
5425 * nss-myhostname will now also resolve "localhost"
5426 implicitly. This makes /etc/hosts an optional file and
5427 nicely handles that on IPv6 ::1 maps to both "localhost" and
5430 * libsystemd-logind.so gained a new call
5431 sd_get_machine_names() to enumerate running containers and
5432 VMs (currently only supported by very new libvirt and
5433 nspawn). sd_login_monitor can now be used to watch
5434 VMs/containers coming and going.
5436 * .include is not allowed recursively anymore, and only in
5437 unit files. Usually it is better to use drop-in snippets in
5438 .d/*.conf anyway, as introduced with systemd 198.
5440 * systemd-analyze gained a new "critical-chain" command that
5441 determines the slowest chain of units run during system
5442 boot-up. It is very useful for tracking down where
5443 optimizing boot time is the most beneficial.
5445 * systemd will no longer allow manipulating service paths in
5446 the name=systemd:/system cgroup tree using ControlGroup= in
5447 units. (But is still fine with it in all other dirs.)
5449 * There's a new systemd-nspawn@.service service file that may
5450 be used to easily run nspawn containers as system
5451 services. With the container's root directory in
5452 /var/lib/container/foobar it is now sufficient to run
5453 "systemctl start systemd-nspawn@foobar.service" to boot it.
5455 * systemd-cgls gained a new parameter "--machine" to list only
5456 the processes within a certain container.
5458 * ConditionSecurity= now can check for "apparmor". We still
5459 are lacking checks for SMACK and IMA for this condition
5460 check though. Patches welcome!
5462 * A new configuration file /etc/systemd/sleep.conf has been
5463 added that may be used to configure which kernel operation
5464 systemd is supposed to execute when "suspend", "hibernate"
5465 or "hybrid-sleep" is requested. This makes the new kernel
5466 "freeze" state accessible to the user.
5468 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
5469 the passed argument if applicable.
5471 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
5472 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
5473 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
5474 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
5475 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
5476 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
5477 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
5482 * The output of 'systemctl list-jobs' got some polishing. The
5483 '--type=' argument may now be passed more than once. A new
5484 command 'systemctl list-sockets' has been added which shows
5485 a list of kernel sockets systemd is listening on with the
5486 socket units they belong to, plus the units these socket
5489 * The experimental libsystemd-bus library got substantial
5490 updates to work in conjunction with the (also experimental)
5491 kdbus kernel project. It works well enough to exchange
5492 messages with some sophistication. Note that kdbus is not
5493 ready yet, and the library is mostly an elaborate test case
5494 for now, and not installable.
5496 * systemd gained a new unit 'systemd-static-nodes.service'
5497 that generates static device nodes earlier during boot, and
5498 can run in conjunction with udev.
5500 * libsystemd-login gained a new call sd_pid_get_user_unit()
5501 to retrieve the user systemd unit a process is running
5502 in. This is useful for systems where systemd is used as
5505 * systemd-nspawn now places all containers in the new /machine
5506 top-level cgroup directory in the name=systemd
5507 hierarchy. libvirt will soon do the same, so that we get a
5508 uniform separation of /system, /user and /machine for system
5509 services, user processes and containers/virtual
5510 machines. This new cgroup hierarchy is also useful to stick
5511 stable names to specific container instances, which can be
5512 recognized later this way (this name may be controlled
5513 via systemd-nspawn's new -M switch). libsystemd-login also
5514 gained a new call sd_pid_get_machine_name() to retrieve the
5515 name of the container/VM a specific process belongs to.
5517 * bootchart can now store its data in the journal.
5519 * libsystemd-journal gained a new call
5520 sd_journal_add_conjunction() for AND expressions to the
5521 matching logic. This can be used to express more complex
5522 logical expressions.
5524 * journactl can now take multiple --unit= and --user-unit=
5527 * The cryptsetup logic now understands the "luks.key=" kernel
5528 command line switch for specifying a file to read the
5529 decryption key from. Also, if a configured key file is not
5530 found the tool will now automatically fall back to prompting
5533 * Python systemd.journal module was updated to wrap recently
5534 added functions from libsystemd-journal. The interface was
5535 changed to bring the low level interface in s.j._Reader
5536 closer to the C API, and the high level interface in
5537 s.j.Reader was updated to wrap and convert all data about
5540 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
5541 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
5542 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
5543 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
5544 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
5545 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5549 * journalctl --update-catalog now understands a new --root=
5550 option to operate on catalogs found in a different root
5553 * During shutdown after systemd has terminated all running
5554 services a final killing loop kills all remaining left-over
5555 processes. We will now print the name of these processes
5556 when we send SIGKILL to them, since this usually indicates a
5559 * If /etc/crypttab refers to password files stored on
5560 configured mount points automatic dependencies will now be
5561 generated to ensure the specific mount is established first
5562 before the key file is attempted to be read.
5564 * 'systemctl status' will now show information about the
5565 network sockets a socket unit is listening on.
5567 * 'systemctl status' will also shown information about any
5568 drop-in configuration file for units. (Drop-In configuration
5569 files in this context are files such as
5570 /etc/systemd/systemd/foobar.service.d/*.conf)
5572 * systemd-cgtop now optionally shows summed up CPU times of
5573 cgroups. Press '%' while running cgtop to switch between
5574 percentage and absolute mode. This is useful to determine
5575 which cgroups use up the most CPU time over the entire
5576 runtime of the system. systemd-cgtop has also been updated
5577 to be 'pipeable' for processing with further shell tools.
5579 * 'hostnamectl set-hostname' will now allow setting of FQDN
5582 * The formatting and parsing of time span values has been
5583 changed. The parser now understands fractional expressions
5584 such as "5.5h". The formatter will now output fractional
5585 expressions for all time spans under 1min, i.e. "5.123456s"
5586 rather than "5s 123ms 456us". For time spans under 1s
5587 millisecond values are shown, for those under 1ms
5588 microsecond values are shown. This should greatly improve
5589 all time-related output of systemd.
5591 * libsystemd-login and libsystemd-journal gained new
5592 functions for querying the poll() events mask and poll()
5593 timeout value for integration into arbitrary event
5596 * localectl gained the ability to list available X11 keymaps
5597 (models, layouts, variants, options).
5599 * 'systemd-analyze dot' gained the ability to filter for
5600 specific units via shell-style globs, to create smaller,
5601 more useful graphs. I.e. it is now possible to create simple
5602 graphs of all the dependencies between only target units, or
5603 of all units that Avahi has dependencies with.
5605 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
5606 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
5607 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
5608 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
5609 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
5610 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
5611 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
5615 * The boot-time readahead implementation for rotating media
5616 will now read the read-ahead data in multiple passes which
5617 consist of all read requests made in equidistant time
5618 intervals. This means instead of strictly reading read-ahead
5619 data in its physical order on disk we now try to find a
5620 middle ground between physical and access time order.
5622 * /etc/os-release files gained a new BUILD_ID= field for usage
5623 on operating systems that provide continuous builds of OS
5626 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
5627 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
5628 William Douglas, Zbigniew Jędrzejewski-Szmek
5632 * systemd-python gained an API exposing libsystemd-daemon.
5634 * The SMACK setup logic gained support for uploading CIPSO
5637 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
5638 ReadOnlyDirectories= and InaccessibleDirectories= has
5639 changed. The private /tmp and /var/tmp directories are now
5640 shared by all processes of a service (which means
5641 ExecStartPre= may now leave data in /tmp that ExecStart= of
5642 the same service can still access). When a service is
5643 stopped its temporary directories are immediately deleted
5644 (normal clean-up with tmpfiles is still done in addition to
5647 * By default, systemd will now set a couple of sysctl
5648 variables in the kernel: the safe sysrq options are turned
5649 on, IP route verification is turned on, and source routing
5650 disabled. The recently added hardlink and softlink
5651 protection of the kernel is turned on. These settings should
5652 be reasonably safe, and good defaults for all new systems.
5654 * The predictable network naming logic may now be turned off
5655 with a new kernel command line switch: net.ifnames=0.
5657 * A new libsystemd-bus module has been added that implements a
5658 pretty complete D-Bus client library. For details see:
5660 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
5662 * journald will now explicitly flush the journal files to disk
5663 at the latest 5min after each write. The file will then also
5664 be marked offline until the next write. This should increase
5665 reliability in case of a crash. The synchronization delay
5666 can be configured via SyncIntervalSec= in journald.conf.
5668 * There's a new remote-fs-setup.target unit that can be used
5669 to pull in specific services when at least one remote file
5670 system is to be mounted.
5672 * There are new targets timers.target and paths.target as
5673 canonical targets to pull user timer and path units in
5674 from. This complements sockets.target with a similar
5675 purpose for socket units.
5677 * libudev gained a new call udev_device_set_attribute_value()
5678 to set sysfs attributes of a device.
5680 * The udev daemon now sets the default number of worker
5681 processes executed in parallel based on the number of available
5682 CPUs instead of the amount of available RAM. This is supposed
5683 to provide a more reliable default and limit a too aggressive
5684 parallelism for setups with 1000s of devices connected.
5686 Contributions from: Auke Kok, Colin Walters, Cristian
5687 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
5688 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
5689 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
5690 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
5691 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
5692 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
5693 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
5694 Zbigniew Jędrzejewski-Szmek
5698 * Configuration of unit files may now be extended via drop-in
5699 files without having to edit/override the unit files
5700 themselves. More specifically, if the administrator wants to
5701 change one value for a service file foobar.service he can
5702 now do so by dropping in a configuration snippet into
5703 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
5704 will load all these snippets and apply them on top of the
5705 main unit configuration file, possibly extending or
5706 overriding its settings. Using these drop-in snippets is
5707 generally nicer than the two earlier options for changing
5708 unit files locally: copying the files from
5709 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
5710 them there; or creating a new file in /etc/systemd/system/
5711 that incorporates the original one via ".include". Drop-in
5712 snippets into these .d/ directories can be placed in any
5713 directory systemd looks for units in, and the usual
5714 overriding semantics between /usr/lib, /etc and /run apply
5717 * Most unit file settings which take lists of items can now be
5718 reset by assigning the empty string to them. For example,
5719 normally, settings such as Environment=FOO=BAR append a new
5720 environment variable assignment to the environment block,
5721 each time they are used. By assigning Environment= the empty
5722 string the environment block can be reset to empty. This is
5723 particularly useful with the .d/*.conf drop-in snippets
5724 mentioned above, since this adds the ability to reset list
5725 settings from vendor unit files via these drop-ins.
5727 * systemctl gained a new "list-dependencies" command for
5728 listing the dependencies of a unit recursively.
5730 * Inhibitors are now honored and listed by "systemctl
5731 suspend", "systemctl poweroff" (and similar) too, not only
5732 GNOME. These commands will also list active sessions by
5735 * Resource limits (as exposed by the various control group
5736 controllers) can now be controlled dynamically at runtime
5737 for all units. More specifically, you can now use a command
5738 like "systemctl set-cgroup-attr foobar.service cpu.shares
5739 2000" to alter the CPU shares a specific service gets. These
5740 settings are stored persistently on disk, and thus allow the
5741 administrator to easily adjust the resource usage of
5742 services with a few simple commands. This dynamic resource
5743 management logic is also available to other programs via the
5744 bus. Almost any kernel cgroup attribute and controller is
5747 * systemd-vconsole-setup will now copy all font settings to
5748 all allocated VTs, where it previously applied them only to
5751 * libsystemd-login gained the new sd_session_get_tty() API
5754 * This release drops support for a few legacy or
5755 distribution-specific LSB facility names when parsing init
5756 scripts: $x-display-manager, $mail-transfer-agent,
5757 $mail-transport-agent, $mail-transfer-agent, $smtp,
5758 $null. Also, the mail-transfer-agent.target unit backing
5759 this has been removed. Distributions which want to retain
5760 compatibility with this should carry the burden for
5761 supporting this themselves and patch support for these back
5762 in, if they really need to. Also, the facilities $syslog and
5763 $local_fs are now ignored, since systemd does not support
5764 early-boot LSB init scripts anymore, and these facilities
5765 are implied anyway for normal services. syslog.target has
5768 * There are new bus calls on PID1's Manager object for
5769 cancelling jobs, and removing snapshot units. Previously,
5770 both calls were only available on the Job and Snapshot
5773 * systemd-journal-gatewayd gained SSL support.
5775 * The various "environment" files, such as /etc/locale.conf
5776 now support continuation lines with a backslash ("\") as
5777 last character in the line, similarly in style (but different)
5778 to how this is supported in shells.
5780 * For normal user processes the _SYSTEMD_USER_UNIT= field is
5781 now implicitly appended to every log entry logged. systemctl
5782 has been updated to filter by this field when operating on a
5783 user systemd instance.
5785 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
5786 CAP_AUDIT_CONTROL capabilities to the capabilities set for
5787 the container. This makes it easier to boot unmodified
5788 Fedora systems in a container, which however still requires
5789 audit=0 to be passed on the kernel command line. Auditing in
5790 kernel and userspace is unfortunately still too broken in
5791 context of containers, hence we recommend compiling it out
5792 of the kernel or using audit=0. Hopefully this will be fixed
5793 one day for good in the kernel.
5795 * nspawn gained the new --bind= and --bind-ro= parameters to
5796 bind mount specific directories from the host into the
5799 * nspawn will now mount its own devpts file system instance
5800 into the container, in order not to leak pty devices from
5801 the host into the container.
5803 * systemd will now read the firmware boot time performance
5804 information from the EFI variables, if the used boot loader
5805 supports this, and takes it into account for boot performance
5806 analysis via "systemd-analyze". This is currently supported
5807 only in conjunction with Gummiboot, but could be supported
5808 by other boot loaders too. For details see:
5810 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
5812 * A new generator has been added that automatically mounts the
5813 EFI System Partition (ESP) to /boot, if that directory
5814 exists, is empty, and no other file system has been
5815 configured to be mounted there.
5817 * logind will now send out PrepareForSleep(false) out
5818 unconditionally, after coming back from suspend. This may be
5819 used by applications as asynchronous notification for
5820 system resume events.
5822 * "systemctl unlock-sessions" has been added, that allows
5823 unlocking the screens of all user sessions at once, similar
5824 to how "systemctl lock-sessions" already locked all users
5825 sessions. This is backed by a new D-Bus call UnlockSessions().
5827 * "loginctl seat-status" will now show the master device of a
5828 seat. (i.e. the device of a seat that needs to be around for
5829 the seat to be considered available, usually the graphics
5832 * tmpfiles gained a new "X" line type, that allows
5833 configuration of files and directories (with wildcards) that
5834 shall be excluded from automatic cleanup ("aging").
5836 * udev default rules set the device node permissions now only
5837 at "add" events, and do not change them any longer with a
5838 later "change" event.
5840 * The log messages for lid events and power/sleep keypresses
5841 now carry a message ID.
5843 * We now have a substantially larger unit test suite, but this
5844 continues to be work in progress.
5846 * udevadm hwdb gained a new --root= parameter to change the
5847 root directory to operate relative to.
5849 * logind will now issue a background sync() request to the kernel
5850 early at shutdown, so that dirty buffers are flushed to disk early
5851 instead of at the last moment, in order to optimize shutdown
5854 * A new bootctl tool has been added that is an interface for
5855 certain boot loader operations. This is currently a preview
5856 and is likely to be extended into a small mechanism daemon
5857 like timedated, localed, hostnamed, and can be used by
5858 graphical UIs to enumerate available boot options, and
5859 request boot into firmware operations.
5861 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
5862 the rest of the package. It also has been updated to work
5863 correctly in initrds.
5865 * Policykit previously has been runtime optional, and is now
5866 also compile time optional via a configure switch.
5868 * systemd-analyze has been reimplemented in C. Also "systemctl
5869 dot" has moved into systemd-analyze.
5871 * "systemctl status" with no further parameters will now print
5872 the status of all active or failed units.
5874 * Operations such as "systemctl start" can now be executed
5875 with a new mode "--irreversible" which may be used to queue
5876 operations that cannot accidentally be reversed by a later
5877 job queuing. This is by default used to make shutdown
5878 requests more robust.
5880 * The Python API of systemd now gained a new module for
5881 reading journal files.
5883 * A new tool kernel-install has been added that can install
5884 kernel images according to the Boot Loader Specification:
5886 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
5888 * Boot time console output has been improved to provide
5889 animated boot time output for hanging jobs.
5891 * A new tool systemd-activate has been added which can be used
5892 to test socket activation with, directly from the command
5893 line. This should make it much easier to test and debug
5894 socket activation in daemons.
5896 * journalctl gained a new "--reverse" (or -r) option to show
5897 journal output in reverse order (i.e. newest line first).
5899 * journalctl gained a new "--pager-end" (or -e) option to jump
5900 to immediately jump to the end of the journal in the
5901 pager. This is only supported in conjunction with "less".
5903 * journalctl gained a new "--user-unit=" option, that works
5904 similarly to "--unit=" but filters for user units rather than
5907 * A number of unit files to ease adoption of systemd in
5908 initrds has been added. This moves some minimal logic from
5909 the various initrd implementations into systemd proper.
5911 * The journal files are now owned by a new group
5912 "systemd-journal", which exists specifically to allow access
5913 to the journal, and nothing else. Previously, we used the
5914 "adm" group for that, which however possibly covers more
5915 than just journal/log file access. This new group is now
5916 already used by systemd-journal-gatewayd to ensure this
5917 daemon gets access to the journal files and as little else
5918 as possible. Note that "make install" will also set FS ACLs
5919 up for /var/log/journal to give "adm" and "wheel" read
5920 access to it, in addition to "systemd-journal" which owns
5921 the journal files. We recommend that packaging scripts also
5922 add read access to "adm" + "wheel" to /var/log/journal, and
5923 all existing/future journal files. To normal users and
5924 administrators little changes, however packagers need to
5925 ensure to create the "systemd-journal" system group at
5926 package installation time.
5928 * The systemd-journal-gatewayd now runs as unprivileged user
5929 systemd-journal-gateway:systemd-journal-gateway. Packaging
5930 scripts need to create these system user/group at
5933 * timedated now exposes a new boolean property CanNTP that
5934 indicates whether a local NTP service is available or not.
5936 * systemd-detect-virt will now also detect xen PVs
5938 * The pstore file system is now mounted by default, if it is
5941 * In addition to the SELinux and IMA policies we will now also
5942 load SMACK policies at early boot.
5944 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
5945 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
5946 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
5947 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
5948 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
5949 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
5950 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
5951 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
5952 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
5953 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
5954 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
5955 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
5956 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
5957 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
5961 * Timer units now support calendar time events in addition to
5962 monotonic time events. That means you can now trigger a unit
5963 based on a calendar time specification such as "Thu,Fri
5964 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
5965 or fifth day of any month of the year 2013, given that it is
5966 a thursday or friday. This brings timer event support
5967 considerably closer to cron's capabilities. For details on
5968 the supported calendar time specification language see
5971 * udev now supports a number of different naming policies for
5972 network interfaces for predictable names, and a combination
5973 of these policies is now the default. Please see this wiki
5974 document for details:
5976 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
5978 * Auke Kok's bootchart implementation has been added to the
5979 systemd tree. It is an optional component that can graph the
5980 boot in quite some detail. It is one of the best bootchart
5981 implementations around and minimal in its code and
5984 * nss-myhostname has been integrated into the systemd source
5985 tree. nss-myhostname guarantees that the local hostname
5986 always stays resolvable via NSS. It has been a weak
5987 requirement of systemd-hostnamed since a long time, and
5988 since its code is actually trivial we decided to just
5989 include it in systemd's source tree. It can be turned off
5990 with a configure switch.
5992 * The read-ahead logic is now capable of properly detecting
5993 whether a btrfs file system is on SSD or rotating media, in
5994 order to optimize the read-ahead scheme. Previously, it was
5995 only capable of detecting this on traditional file systems
5998 * In udev, additional device properties are now read from the
5999 IAB in addition to the OUI database. Also, Bluetooth company
6000 identities are attached to the devices as well.
6002 * In service files %U may be used as specifier that is
6003 replaced by the configured user name of the service.
6005 * nspawn may now be invoked without a controlling TTY. This
6006 makes it suitable for invocation as its own service. This
6007 may be used to set up a simple containerized server system
6008 using only core OS tools.
6010 * systemd and nspawn can now accept socket file descriptors
6011 when they are started for socket activation. This enables
6012 implementation of socket activated nspawn
6013 containers. i.e. think about autospawning an entire OS image
6014 when the first SSH or HTTP connection is received. We expect
6015 that similar functionality will also be added to libvirt-lxc
6018 * journalctl will now suppress ANSI color codes when
6019 presenting log data.
6021 * systemctl will no longer show control group information for
6022 a unit if the control group is empty anyway.
6024 * logind can now automatically suspend/hibernate/shutdown the
6027 * /etc/machine-info and hostnamed now also expose the chassis
6028 type of the system. This can be used to determine whether
6029 the local system is a laptop, desktop, handset or
6030 tablet. This information may either be configured by the
6031 user/vendor or is automatically determined from ACPI and DMI
6032 information if possible.
6034 * A number of PolicyKit actions are now bound together with
6035 "imply" rules. This should simplify creating UIs because
6036 many actions will now authenticate similar ones as well.
6038 * Unit files learnt a new condition ConditionACPower= which
6039 may be used to conditionalize a unit depending on whether an
6040 AC power source is connected or not, of whether the system
6041 is running on battery power.
6043 * systemctl gained a new "is-failed" verb that may be used in
6044 shell scripts and suchlike to check whether a specific unit
6045 is in the "failed" state.
6047 * The EnvironmentFile= setting in unit files now supports file
6048 globbing, and can hence be used to easily read a number of
6049 environment files at once.
6051 * systemd will no longer detect and recognize specific
6052 distributions. All distribution-specific #ifdeffery has been
6053 removed, systemd is now fully generic and
6054 distribution-agnostic. Effectively, not too much is lost as
6055 a lot of the code is still accessible via explicit configure
6056 switches. However, support for some distribution specific
6057 legacy configuration file formats has been dropped. We
6058 recommend distributions to simply adopt the configuration
6059 files everybody else uses now and convert the old
6060 configuration from packaging scripts. Most distributions
6061 already did that. If that's not possible or desirable,
6062 distributions are welcome to forward port the specific
6063 pieces of code locally from the git history.
6065 * When logging a message about a unit systemd will now always
6066 log the unit name in the message meta data.
6068 * localectl will now also discover system locale data that is
6069 not stored in locale archives, but directly unpacked.
6071 * logind will no longer unconditionally use framebuffer
6072 devices as seat masters, i.e. as devices that are required
6073 to be existing before a seat is considered preset. Instead,
6074 it will now look for all devices that are tagged as
6075 "seat-master" in udev. By default, framebuffer devices will
6076 be marked as such, but depending on local systems, other
6077 devices might be marked as well. This may be used to
6078 integrate graphics cards using closed source drivers (such
6079 as NVidia ones) more nicely into logind. Note however, that
6080 we recommend using the open source NVidia drivers instead,
6081 and no udev rules for the closed-source drivers will be
6082 shipped from us upstream.
6084 Contributions from: Adam Williamson, Alessandro Crismani, Auke
6085 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
6086 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
6087 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
6088 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
6089 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
6090 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
6091 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
6092 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
6093 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
6094 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
6095 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
6100 * udev gained support for loading additional device properties
6101 from an indexed database that is keyed by vendor/product IDs
6102 and similar device identifiers. For the beginning this
6103 "hwdb" is populated with data from the well-known PCI and
6104 USB database, but also includes PNP, ACPI and OID data. In
6105 the longer run this indexed database shall grow into
6106 becoming the one central database for non-essential
6107 userspace device metadata. Previously, data from the PCI/USB
6108 database was only attached to select devices, since the
6109 lookup was a relatively expensive operation due to O(n) time
6110 complexity (with n being the number of entries in the
6111 database). Since this is now O(1), we decided to add in this
6112 data for all devices where this is available, by
6113 default. Note that the indexed database needs to be rebuilt
6114 when new data files are installed. To achieve this you need
6115 to update your packaging scripts to invoke "udevadm hwdb
6116 --update" after installation of hwdb data files. For
6117 RPM-based distributions we introduced the new
6118 %udev_hwdb_update macro for this purpose.
6120 * The Journal gained support for the "Message Catalog", an
6121 indexed database to link up additional information with
6122 journal entries. For further details please check:
6124 https://www.freedesktop.org/wiki/Software/systemd/catalog
6126 The indexed message catalog database also needs to be
6127 rebuilt after installation of message catalog files. Use
6128 "journalctl --update-catalog" for this. For RPM-based
6129 distributions we introduced the %journal_catalog_update
6130 macro for this purpose.
6132 * The Python Journal bindings gained support for the standard
6133 Python logging framework.
6135 * The Journal API gained new functions for checking whether
6136 the underlying file system of a journal file is capable of
6137 properly reporting file change notifications, or whether
6138 applications that want to reflect journal changes "live"
6139 need to recheck journal files continuously in appropriate
6142 * It is now possible to set the "age" field for tmpfiles
6143 entries to 0, indicating that files matching this entry
6144 shall always be removed when the directories are cleaned up.
6146 * coredumpctl gained a new "gdb" verb which invokes gdb
6147 right-away on the selected coredump.
6149 * There's now support for "hybrid sleep" on kernels that
6150 support this, in addition to "suspend" and "hibernate". Use
6151 "systemctl hybrid-sleep" to make use of this.
6153 * logind's HandleSuspendKey= setting (and related settings)
6154 now gained support for a new "lock" setting to simply
6155 request the screen lock on all local sessions, instead of
6156 actually executing a suspend or hibernation.
6158 * systemd will now mount the EFI variables file system by
6161 * Socket units now gained support for configuration of the
6162 SMACK security label.
6164 * timedatectl will now output the time of the last and next
6165 daylight saving change.
6167 * We dropped support for various legacy and distro-specific
6168 concepts, such as insserv, early-boot SysV services
6169 (i.e. those for non-standard runlevels such as 'b' or 'S')
6170 or ArchLinux /etc/rc.conf support. We recommend the
6171 distributions who still need support this to either continue
6172 to maintain the necessary patches downstream, or find a
6173 different solution. (Talk to us if you have questions!)
6175 * Various systemd components will now bypass PolicyKit checks
6176 for root and otherwise handle properly if PolicyKit is not
6177 found to be around. This should fix most issues for
6178 PolicyKit-less systems. Quite frankly this should have been
6179 this way since day one. It is absolutely our intention to
6180 make systemd work fine on PolicyKit-less systems, and we
6181 consider it a bug if something does not work as it should if
6182 PolicyKit is not around.
6184 * For embedded systems it is now possible to build udev and
6185 systemd without blkid and/or kmod support.
6187 * "systemctl switch-root" is now capable of switching root
6188 more than once. I.e. in addition to transitions from the
6189 initrd to the host OS it is now possible to transition to
6190 further OS images from the host. This is useful to implement
6191 offline updating tools.
6193 * Various other additions have been made to the RPM macros
6194 shipped with systemd. Use %udev_rules_update() after
6195 installing new udev rules files. %_udevhwdbdir,
6196 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
6197 %_sysctldir are now available which resolve to the right
6198 directories for packages to place various data files in.
6200 * journalctl gained the new --full switch (in addition to
6201 --all, to disable ellipsation for long messages.
6203 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
6204 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
6205 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
6206 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
6207 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
6208 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
6209 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
6210 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
6211 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
6215 * journalctl gained new --since= and --until= switches to
6216 filter by time. It also now supports nice filtering for
6217 units via --unit=/-u.
6219 * Type=oneshot services may use ExecReload= and do the
6222 * The journal daemon now supports time-based rotation and
6223 vacuuming, in addition to the usual disk-space based
6226 * The journal will now index the available field values for
6227 each field name. This enables clients to show pretty drop
6228 downs of available match values when filtering. The bash
6229 completion of journalctl has been updated
6230 accordingly. journalctl gained a new switch -F to list all
6231 values a certain field takes in the journal database.
6233 * More service events are now written as structured messages
6234 to the journal, and made recognizable via message IDs.
6236 * The timedated, localed and hostnamed mini-services which
6237 previously only provided support for changing time, locale
6238 and hostname settings from graphical DEs such as GNOME now
6239 also have a minimal (but very useful) text-based client
6240 utility each. This is probably the nicest way to changing
6241 these settings from the command line now, especially since
6242 it lists available options and is fully integrated with bash
6245 * There's now a new tool "systemd-coredumpctl" to list and
6246 extract coredumps from the journal.
6248 * We now install a README each in /var/log/ and
6249 /etc/rc.d/init.d explaining where the system logs and init
6250 scripts went. This hopefully should help folks who go to
6251 that dirs and look into the otherwise now empty void and
6252 scratch their heads.
6254 * When user-services are invoked (by systemd --user) the
6255 $MANAGERPID env var is set to the PID of systemd.
6257 * SIGRTMIN+24 when sent to a --user instance will now result
6258 in immediate termination of systemd.
6260 * gatewayd received numerous feature additions such as a
6261 "follow" mode, for live syncing and filtering.
6263 * browse.html now allows filtering and showing detailed
6264 information on specific entries. Keyboard navigation and
6265 mouse screen support has been added.
6267 * gatewayd/journalctl now supports HTML5/JSON
6268 Server-Sent-Events as output.
6270 * The SysV init script compatibility logic will now
6271 heuristically determine whether a script supports the
6272 "reload" verb, and only then make this available as
6275 * "systemctl status --follow" has been removed, use "journalctl
6278 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
6279 have been removed since they are hardly useful to be
6282 * And I'd like to take the opportunity to specifically mention
6283 Zbigniew for his great contributions. Zbigniew, you rock!
6285 Contributions from: Andrew Eikum, Christian Hesse, Colin
6286 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
6287 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
6288 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
6289 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
6290 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
6291 Jędrzejewski-Szmek, Сковорода Никита Андреевич
6295 * If /etc/vconsole.conf is non-existent or empty we will no
6296 longer load any console font or key map at boot by
6297 default. Instead the kernel defaults will be left
6298 intact. This is definitely the right thing to do, as no
6299 configuration should mean no configuration, and hard-coding
6300 font names that are different on all archs is probably a bad
6301 idea. Also, the kernel default key map and font should be
6302 good enough for most cases anyway, and mostly identical to
6303 the userspace fonts/key maps we previously overloaded them
6304 with. If distributions want to continue to default to a
6305 non-kernel font or key map they should ship a default
6306 /etc/vconsole.conf with the appropriate contents.
6308 Contributions from: Colin Walters, Daniel J Walsh, Dave
6309 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
6310 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6314 * journalctl gained a new --cursor= switch to show entries
6315 starting from the specified location in the journal.
6317 * We now enforce a size limit on journal entry fields exported
6318 with "-o json" in journalctl. Fields larger than 4K will be
6319 assigned null. This can be turned off with --all.
6321 * An (optional) journal gateway daemon is now available as
6322 "systemd-journal-gatewayd.service". This service provides
6323 access to the journal via HTTP and JSON. This functionality
6324 will be used to implement live log synchronization in both
6325 pull and push modes, but has various other users too, such
6326 as easy log access for debugging of embedded devices. Right
6327 now it is already useful to retrieve the journal via HTTP:
6329 # systemctl start systemd-journal-gatewayd.service
6330 # wget http://localhost:19531/entries
6332 This will download the journal contents in a
6333 /var/log/messages compatible format. The same as JSON:
6335 # curl -H"Accept: application/json" http://localhost:19531/entries
6337 This service is also accessible via a web browser where a
6338 single static HTML5 app is served that uses the JSON logic
6339 to enable the user to do some basic browsing of the
6340 journal. This will be extended later on. Here's an example
6341 screenshot of this app in its current state:
6343 http://0pointer.de/public/journal-gatewayd
6345 Contributions from: Kay Sievers, Lennart Poettering, Robert
6346 Milasan, Tom Gundersen
6350 * The bash completion logic is now available for journalctl
6353 * We do not mount the "cpuset" controller anymore together with
6354 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
6355 started if no parameters are assigned to it. "cpuset" hence
6356 broke code that assumed it could create "cpu" groups and
6359 * journalctl -f will now subscribe to terminal size changes,
6360 and line break accordingly.
6362 Contributions from: Dave Reisner, Kay Sievers, Lennart
6363 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
6367 * nspawn will now create a symlink /etc/localtime in the
6368 container environment, copying the host's timezone
6369 setting. Previously this has been done via a bind mount, but
6370 since symlinks cannot be bind mounted this has now been
6371 changed to create/update the appropriate symlink.
6373 * journalctl -n's line number argument is now optional, and
6374 will default to 10 if omitted.
6376 * journald will now log the maximum size the journal files may
6377 take up on disk. This is particularly useful if the default
6378 built-in logic of determining this parameter from the file
6379 system size is used. Use "systemctl status
6380 systemd-journald.service" to see this information.
6382 * The multi-seat X wrapper tool has been stripped down. As X
6383 is now capable of enumerating graphics devices via udev in a
6384 seat-aware way the wrapper is not strictly necessary
6385 anymore. A stripped down temporary stop-gap is still shipped
6386 until the upstream display managers have been updated to
6387 fully support the new X logic. Expect this wrapper to be
6388 removed entirely in one of the next releases.
6390 * HandleSleepKey= in logind.conf has been split up into
6391 HandleSuspendKey= and HandleHibernateKey=. The old setting
6392 is not available anymore. X11 and the kernel are
6393 distinguishing between these keys and we should too. This
6394 also means the inhibition lock for these keys has been split
6397 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
6398 Poettering, Lukas Nykryn, Václav Pavlín
6402 * Whenever a unit changes state we will now log this to the
6403 journal and show along the unit's own log output in
6406 * ConditionPathIsMountPoint= can now properly detect bind
6407 mount points too. (Previously, a bind mount of one file
6408 system to another place in the same file system could not be
6409 detected as mount, since they shared struct stat's st_dev
6412 * We will now mount the cgroup controllers cpu, cpuacct,
6413 cpuset and the controllers net_cls, net_prio together by
6416 * nspawn containers will now have a virtualized boot
6417 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
6418 over with a randomized ID at container initialization). This
6419 has the effect of making "journalctl -b" do the right thing
6422 * The JSON output journal serialization has been updated not
6423 to generate "endless" list objects anymore, but rather one
6424 JSON object per line. This is more in line how most JSON
6425 parsers expect JSON objects. The new output mode
6426 "json-pretty" has been added to provide similar output, but
6427 neatly aligned for readability by humans.
6429 * We dropped all explicit sync() invocations in the shutdown
6430 code. The kernel does this implicitly anyway in the kernel
6431 reboot() syscall. halt(8)'s -n option is now a compatibility
6434 * We now support virtualized reboot() in containers, as
6435 supported by newer kernels. We will fall back to exit() if
6436 CAP_SYS_REBOOT is not available to the container. Also,
6437 nspawn makes use of this now and will actually reboot the
6438 container if the containerized OS asks for that.
6440 * journalctl will only show local log output by default
6441 now. Use --merge (-m) to show remote log output, too.
6443 * libsystemd-journal gained the new sd_journal_get_usage()
6444 call to determine the current disk usage of all journal
6445 files. This is exposed in the new "journalctl --disk-usage"
6448 * journald gained a new configuration setting SplitMode= in
6449 journald.conf which may be used to control how user journals
6450 are split off. See journald.conf(5) for details.
6452 * A new condition type ConditionFileNotEmpty= has been added.
6454 * tmpfiles' "w" lines now support file globbing, to write
6455 multiple files at once.
6457 * We added Python bindings for the journal submission
6458 APIs. More Python APIs for a number of selected APIs will
6459 likely follow. Note that we intend to add native bindings
6460 only for the Python language, as we consider it common
6461 enough to deserve bindings shipped within systemd. There are
6462 various projects outside of systemd that provide bindings
6463 for languages such as PHP or Lua.
6465 * Many conditions will now resolve specifiers such as %i. In
6466 addition, PathChanged= and related directives of .path units
6467 now support specifiers as well.
6469 * There's now a new RPM macro definition for the system preset
6472 * journald will now warn if it ca not forward a message to the
6473 syslog daemon because its socket is full.
6475 * timedated will no longer write or process /etc/timezone,
6476 except on Debian. As we do not support late mounted /usr
6477 anymore /etc/localtime always being a symlink is now safe,
6478 and hence the information in /etc/timezone is not necessary
6481 * logind will now always reserve one VT for a text getty (VT6
6482 by default). Previously if more than 6 X sessions where
6483 started they took up all the VTs with auto-spawned gettys,
6484 so that no text gettys were available anymore.
6486 * udev will now automatically inform the btrfs kernel logic
6487 about btrfs RAID components showing up. This should make
6488 simple hotplug based btrfs RAID assembly work.
6490 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
6491 (but not for its children which will stay at the kernel
6492 default). This should allow setups with a lot more listening
6495 * systemd will now always pass the configured timezone to the
6496 kernel at boot. timedated will do the same when the timezone
6499 * logind's inhibition logic has been updated. By default,
6500 logind will now handle the lid switch, the power and sleep
6501 keys all the time, even in graphical sessions. If DEs want
6502 to handle these events on their own they should take the new
6503 handle-power-key, handle-sleep-key and handle-lid-switch
6504 inhibitors during their runtime. A simple way to achieve
6505 that is to invoke the DE wrapped in an invocation of:
6507 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
6509 * Access to unit operations is now checked via SELinux taking
6510 the unit file label and client process label into account.
6512 * systemd will now notify the administrator in the journal
6513 when he over-mounts a non-empty directory.
6515 * There are new specifiers that are resolved in unit files,
6516 for the host name (%H), the machine ID (%m) and the boot ID
6519 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
6520 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
6521 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
6522 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
6523 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
6524 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
6525 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
6529 * Support for reading structured kernel messages from
6530 /dev/kmsg has now been added and is enabled by default.
6532 * Support for reading kernel messages from /proc/kmsg has now
6533 been removed. If you want kernel messages in the journal
6534 make sure to run a recent kernel (>= 3.5) that supports
6535 reading structured messages from /dev/kmsg (see
6536 above). /proc/kmsg is now exclusive property of classic
6537 syslog daemons again.
6539 * The libudev API gained the new
6540 udev_device_new_from_device_id() call.
6542 * The logic for file system namespace (ReadOnlyDirectory=,
6543 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
6544 require pivot_root() anymore. This means fewer temporary
6545 directories are created below /tmp for this feature.
6547 * nspawn containers will now see and receive all submounts
6548 made on the host OS below the root file system of the
6551 * Forward Secure Sealing is now supported for Journal files,
6552 which provide cryptographical sealing of journal files so
6553 that attackers cannot alter log history anymore without this
6554 being detectable. Lennart will soon post a blog story about
6555 this explaining it in more detail.
6557 * There are two new service settings RestartPreventExitStatus=
6558 and SuccessExitStatus= which allow configuration of exit
6559 status (exit code or signal) which will be excepted from the
6560 restart logic, resp. consider successful.
6562 * journalctl gained the new --verify switch that can be used
6563 to check the integrity of the structure of journal files and
6564 (if Forward Secure Sealing is enabled) the contents of
6567 * nspawn containers will now be run with /dev/stdin, /dev/fd/
6568 and similar symlinks pre-created. This makes running shells
6569 as container init process a lot more fun.
6571 * The fstab support can now handle PARTUUID= and PARTLABEL=
6574 * A new ConditionHost= condition has been added to match
6575 against the hostname (with globs) and machine ID. This is
6576 useful for clusters where a single OS image is used to
6577 provision a large number of hosts which shall run slightly
6578 different sets of services.
6580 * Services which hit the restart limit will now be placed in a
6583 Contributions from: Bertram Poettering, Dave Reisner, Huang
6584 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
6585 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
6589 * When running in --user mode systemd will now become a
6590 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
6591 tree a lot more organized.
6593 * A new PartOf= unit dependency type has been introduced that
6594 may be used to group services in a natural way.
6596 * "systemctl enable" may now be used to enable instances of
6599 * journalctl now prints error log levels in red, and
6600 warning/notice log levels in bright white. It also supports
6601 filtering by log level now.
6603 * cgtop gained a new -n switch (similar to top), to configure
6604 the maximum number of iterations to run for. It also gained
6605 -b, to run in batch mode (accepting no input).
6607 * The suffix ".service" may now be omitted on most systemctl
6608 command lines involving service unit names.
6610 * There's a new bus call in logind to lock all sessions, as
6611 well as a loginctl verb for it "lock-sessions".
6613 * libsystemd-logind.so gained a new call sd_journal_perror()
6614 that works similar to libc perror() but logs to the journal
6615 and encodes structured information about the error number.
6617 * /etc/crypttab entries now understand the new keyfile-size=
6620 * shutdown(8) now can send a (configurable) wall message when
6621 a shutdown is cancelled.
6623 * The mount propagation mode for the root file system will now
6624 default to "shared", which is useful to make containers work
6625 nicely out-of-the-box so that they receive new mounts from
6626 the host. This can be undone locally by running "mount
6627 --make-rprivate /" if needed.
6629 * The prefdm.service file has been removed. Distributions
6630 should maintain this unit downstream if they intend to keep
6631 it around. However, we recommend writing normal unit files
6632 for display managers instead.
6634 * Since systemd is a crucial part of the OS we will now
6635 default to a number of compiler switches that improve
6636 security (hardening) such as read-only relocations, stack
6637 protection, and suchlike.
6639 * The TimeoutSec= setting for services is now split into
6640 TimeoutStartSec= and TimeoutStopSec= to allow configuration
6641 of individual time outs for the start and the stop phase of
6644 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
6645 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
6646 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
6647 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
6648 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
6649 Gundersen, Zbigniew Jędrzejewski-Szmek
6653 * The journal and id128 C APIs are now fully documented as man
6656 * Extra safety checks have been added when transitioning from
6657 the initial RAM disk to the main system to avoid accidental
6660 * /etc/crypttab entries now understand the new keyfile-offset=
6663 * systemctl -t can now be used to filter by unit load state.
6665 * The journal C API gained the new sd_journal_wait() call to
6666 make writing synchronous journal clients easier.
6668 * journalctl gained the new -D switch to show journals from a
6671 * journalctl now displays a special marker between log
6672 messages of two different boots.
6674 * The journal is now explicitly flushed to /var via a service
6675 systemd-journal-flush.service, rather than implicitly simply
6676 by seeing /var/log/journal to be writable.
6678 * journalctl (and the journal C APIs) can now match for much
6679 more complex expressions, with alternatives and
6682 * When transitioning from the initial RAM disk to the main
6683 system we will now kill all processes in a killing spree to
6684 ensure no processes stay around by accident.
6686 * Three new specifiers may be used in unit files: %u, %h, %s
6687 resolve to the user name, user home directory resp. user
6688 shell. This is useful for running systemd user instances.
6690 * We now automatically rotate journal files if their data
6691 object hash table gets a fill level > 75%. We also size the
6692 hash table based on the configured maximum file size. This
6693 together should lower hash collisions drastically and thus
6694 speed things up a bit.
6696 * journalctl gained the new "--header" switch to introspect
6697 header data of journal files.
6699 * A new setting SystemCallFilters= has been added to services
6700 which may be used to apply blacklists or whitelists to
6701 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
6703 * nspawn gained a new --link-journal= switch (and quicker: -j)
6704 to link the container journal with the host. This makes it
6705 very easy to centralize log viewing on the host for all
6706 guests while still keeping the journal files separated.
6708 * Many bugfixes and optimizations
6710 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
6711 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
6712 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
6717 * Several tools now understand kernel command line arguments,
6718 which are only read when run in an initial RAM disk. They
6719 usually follow closely their normal counterparts, but are
6722 * There's a new tool to analyze the readahead files that are
6723 automatically generated at boot. Use:
6725 /usr/lib/systemd/systemd-readahead analyze /.readahead
6727 * We now provide an early debug shell on tty9 if this enabled. Use:
6729 systemctl enable debug-shell.service
6731 * All plymouth related units have been moved into the Plymouth
6732 package. Please make sure to upgrade your Plymouth version
6735 * systemd-tmpfiles now supports getting passed the basename of
6736 a configuration file only, in which case it will look for it
6737 in all appropriate directories automatically.
6739 * udevadm info now takes a /dev or /sys path as argument, and
6740 does the right thing. Example:
6742 udevadm info /dev/sda
6743 udevadm info /sys/class/block/sda
6745 * systemctl now prints a warning if a unit is stopped but a
6746 unit that might trigger it continues to run. Example: a
6747 service is stopped but the socket that activates it is left
6750 * "systemctl status" will now mention if the log output was
6751 shortened due to rotation since a service has been started.
6753 * The journal API now exposes functions to determine the
6754 "cutoff" times due to rotation.
6756 * journald now understands SIGUSR1 and SIGUSR2 for triggering
6757 immediately flushing of runtime logs to /var if possible,
6758 resp. for triggering immediate rotation of the journal
6761 * It is now considered an error if a service is attempted to
6762 be stopped that is not loaded.
6764 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
6766 * systemd-analyze now supports Python 3
6768 * tmpfiles now supports cleaning up directories via aging
6769 where the first level dirs are always kept around but
6770 directories beneath it automatically aged. This is enabled
6771 by prefixing the age field with '~'.
6773 * Seat objects now expose CanGraphical, CanTTY properties
6774 which is required to deal with very fast bootups where the
6775 display manager might be running before the graphics drivers
6776 completed initialization.
6778 * Seat objects now expose a State property.
6780 * We now include RPM macros for service enabling/disabling
6781 based on the preset logic. We recommend RPM based
6782 distributions to make use of these macros if possible. This
6783 makes it simpler to reuse RPM spec files across
6786 * We now make sure that the collected systemd unit name is
6787 always valid when services log to the journal via
6790 * There's a new man page kernel-command-line(7) detailing all
6791 command line options we understand.
6793 * The fstab generator may now be disabled at boot by passing
6794 fstab=0 on the kernel command line.
6796 * A new kernel command line option modules-load= is now understood
6797 to load a specific kernel module statically, early at boot.
6799 * Unit names specified on the systemctl command line are now
6800 automatically escaped as needed. Also, if file system or
6801 device paths are specified they are automatically turned
6802 into the appropriate mount or device unit names. Example:
6804 systemctl status /home
6805 systemctl status /dev/sda
6807 * The SysVConsole= configuration option has been removed from
6808 system.conf parsing.
6810 * The SysV search path is no longer exported on the D-Bus
6813 * The Names= option has been removed from unit file parsing.
6815 * There's a new man page bootup(7) detailing the boot process.
6817 * Every unit and every generator we ship with systemd now
6818 comes with full documentation. The self-explanatory boot is
6821 * A couple of services gained "systemd-" prefixes in their
6822 name if they wrap systemd code, rather than only external
6823 code. Among them fsck@.service which is now
6824 systemd-fsck@.service.
6826 * The HaveWatchdog property has been removed from the D-Bus
6829 * systemd.confirm_spawn= on the kernel command line should now
6832 * There's a new man page crypttab(5) which details all options
6833 we actually understand.
6835 * systemd-nspawn gained a new --capability= switch to pass
6836 additional capabilities to the container.
6838 * timedated will now read known NTP implementation unit names
6839 from /usr/lib/systemd/ntp-units.d/*.list,
6840 systemd-timedated-ntp.target has been removed.
6842 * journalctl gained a new switch "-b" that lists log data of
6843 the current boot only.
6845 * The notify socket is in the abstract namespace again, in
6846 order to support daemons which chroot() at start-up.
6848 * There is a new Storage= configuration option for journald
6849 which allows configuration of where log data should go. This
6850 also provides a way to disable journal logging entirely, so
6851 that data collected is only forwarded to the console, the
6852 kernel log buffer or another syslog implementation.
6854 * Many bugfixes and optimizations
6856 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
6857 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
6858 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
6859 Shawn Landden, Tom Gundersen
6863 * "systemctl help <unit>" now shows the man page if one is
6866 * Several new man pages have been added.
6868 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
6869 MaxLevelConsole= can now be specified in
6870 journald.conf. These options allow reducing the amount of
6871 data stored on disk or forwarded by the log level.
6873 * TimerSlackNSec= can now be specified in system.conf for
6874 PID1. This allows system-wide power savings.
6876 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
6877 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
6882 * logind is now capable of (optionally) handling power and
6883 sleep keys as well as the lid switch.
6885 * journalctl now understands the syntax "journalctl
6886 /usr/bin/avahi-daemon" to get all log output of a specific
6889 * CapabilityBoundingSet= in system.conf now also influences
6890 the capability bound set of usermode helpers of the kernel.
6892 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
6893 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
6894 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
6895 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
6899 * Note that we skipped 139 releases here in order to set the
6900 new version to something that is greater than both udev's
6901 and systemd's most recent version number.
6903 * udev: all udev sources are merged into the systemd source tree now.
6904 All future udev development will happen in the systemd tree. It
6905 is still fully supported to use the udev daemon and tools without
6906 systemd running, like in initramfs or other init systems. Building
6907 udev though, will require the *build* of the systemd tree, but
6908 udev can be properly *run* without systemd.
6910 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
6911 should be used to create dead device nodes as workarounds for broken
6914 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
6915 no longer supported. udev_monitor_new_from_netlink() needs to be
6916 used to subscribe to events.
6918 * udev: when udevd is started by systemd, processes which are left
6919 behind by forking them off of udev rules, are unconditionally cleaned
6920 up and killed now after the event handling has finished. Services or
6921 daemons must be started as systemd services. Services can be
6922 pulled-in by udev to get started, but they can no longer be directly
6923 forked by udev rules.
6925 * udev: the daemon binary is called systemd-udevd now and installed
6926 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
6927 to adapt to that, create symlink, or rename the binary after building
6930 * libudev no longer provides these symbols:
6931 udev_monitor_from_socket()
6932 udev_queue_get_failed_list_entry()
6933 udev_get_{dev,sys,run}_path()
6934 The versions number was bumped and symbol versioning introduced.
6936 * systemd-loginctl and systemd-journalctl have been renamed
6937 to loginctl and journalctl to match systemctl.
6939 * The config files: /etc/systemd/systemd-logind.conf and
6940 /etc/systemd/systemd-journald.conf have been renamed to
6941 logind.conf and journald.conf. Package updates should rename
6942 the files to the new names on upgrade.
6944 * For almost all files the license is now LGPL2.1+, changed
6945 from the previous GPL2.0+. Exceptions are some minor stuff
6946 of udev (which will be changed to LGPL2.1 eventually, too),
6947 and the MIT licensed sd-daemon.[ch] library that is suitable
6948 to be used as drop-in files.
6950 * systemd and logind now handle system sleep states, in
6951 particular suspending and hibernating.
6953 * logind now implements a sleep/shutdown/idle inhibiting logic
6954 suitable for a variety of uses. Soonishly Lennart will blog
6955 about this in more detail.
6957 * var-run.mount and var-lock.mount are no longer provided
6958 (which previously bind mounted these directories to their new
6959 places). Distributions which have not converted these
6960 directories to symlinks should consider stealing these files
6961 from git history and add them downstream.
6963 * We introduced the Documentation= field for units and added
6964 this to all our shipped units. This is useful to make it
6965 easier to explore the boot and the purpose of the various
6968 * All smaller setup units (such as
6969 systemd-vconsole-setup.service) now detect properly if they
6970 are run in a container and are skipped when
6971 appropriate. This guarantees an entirely noise-free boot in
6972 Linux container environments such as systemd-nspawn.
6974 * A framework for implementing offline system updates is now
6975 integrated, for details see:
6976 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
6978 * A new service type Type=idle is available now which helps us
6979 avoiding ugly interleaving of getty output and boot status
6982 * There's now a system-wide CapabilityBoundingSet= option to
6983 globally reduce the set of capabilities for the
6984 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
6985 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
6986 even CAP_NET_ADMIN system-wide for secure systems.
6988 * There are now system-wide DefaultLimitXXX= options to
6989 globally change the defaults of the various resource limits
6990 for all units started by PID 1.
6992 * Harald Hoyer's systemd test suite has been integrated into
6993 systemd which allows easy testing of systemd builds in qemu
6994 and nspawn. (This is really awesome! Ask us for details!)
6996 * The fstab parser is now implemented as generator, not inside
6999 * systemctl will now warn you if .mount units generated from
7000 /etc/fstab are out of date due to changes in fstab that
7001 have not been read by systemd yet.
7003 * systemd is now suitable for usage in initrds. Dracut has
7004 already been updated to make use of this. With this in place
7005 initrds get a slight bit faster but primarily are much
7006 easier to introspect and debug since "systemctl status" in
7007 the host system can be used to introspect initrd services,
7008 and the journal from the initrd is kept around too.
7010 * systemd-delta has been added, a tool to explore differences
7011 between user/admin configuration and vendor defaults.
7013 * PrivateTmp= now affects both /tmp and /var/tmp.
7015 * Boot time status messages are now much prettier and feature
7016 proper english language. Booting up systemd has never been
7019 * Read-ahead pack files now include the inode number of all
7020 files to pre-cache. When the inode changes the pre-caching
7021 is not attempted. This should be nicer to deal with updated
7022 packages which might result in changes of read-ahead
7025 * We now temporaritly lower the kernel's read_ahead_kb variable
7026 when collecting read-ahead data to ensure the kernel's
7027 built-in read-ahead does not add noise to our measurements
7028 of necessary blocks to pre-cache.
7030 * There's now RequiresMountsFor= to add automatic dependencies
7031 for all mounts necessary for a specific file system path.
7033 * MountAuto= and SwapAuto= have been removed from
7034 system.conf. Mounting file systems at boot has to take place
7037 * nspawn now learned a new switch --uuid= to set the machine
7038 ID on the command line.
7040 * nspawn now learned the -b switch to automatically search
7043 * vt102 is now the default TERM for serial TTYs, upgraded from
7046 * systemd-logind now works on VT-less systems.
7048 * The build tree has been reorganized. The individual
7049 components now have directories of their own.
7051 * A new condition type ConditionPathIsReadWrite= is now available.
7053 * nspawn learned the new -C switch to create cgroups for the
7054 container in other hierarchies.
7056 * We now have support for hardware watchdogs, configurable in
7059 * The scheduled shutdown logic now has a public API.
7061 * We now mount /tmp as tmpfs by default, but this can be
7062 masked and /etc/fstab can override it.
7064 * Since udisks does not make use of /media anymore we are not
7065 mounting a tmpfs on it anymore.
7067 * journalctl gained a new --local switch to only interleave
7068 locally generated journal files.
7070 * We can now load the IMA policy at boot automatically.
7072 * The GTK tools have been split off into a systemd-ui.
7074 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
7075 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
7076 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
7077 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
7078 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
7079 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
7080 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
7081 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
7082 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
7087 * This is mostly a bugfix release
7089 * Support optional initialization of the machine ID from the
7090 KVM or container configured UUID.
7092 * Support immediate reboots with "systemctl reboot -ff"
7094 * Show /etc/os-release data in systemd-analyze output
7096 * Many bugfixes for the journal, including endianness fixes and
7097 ensuring that disk space enforcement works
7099 * sd-login.h is C++ compatible again
7101 * Extend the /etc/os-release format on request of the Debian
7104 * We now refuse non-UTF8 strings used in various configuration
7105 and unit files. This is done to ensure we do not pass invalid
7106 data over D-Bus or expose it elsewhere.
7108 * Register Mimo USB Screens as suitable for automatic seat
7111 * Read SELinux client context from journal clients in a race
7114 * Reorder configuration file lookup order. /etc now always
7115 overrides /run in order to allow the administrator to always
7116 and unconditionally override vendor-supplied or
7117 automatically generated data.
7119 * The various user visible bits of the journal now have man
7120 pages. We still lack man pages for the journal API calls
7123 * We now ship all man pages in HTML format again in the
7126 Contributions from: Dave Reisner, Dirk Eibach, Frederic
7127 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
7128 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
7133 * This is mostly a bugfix release
7135 * systems lacking /etc/os-release are no longer supported.
7137 * Various functionality updates to libsystemd-login.so
7139 * Track class of PAM logins to distinguish greeters from
7142 Contributions from: Kay Sievers, Lennart Poettering, Michael
7147 * This is an important bugfix release for v41.
7149 * Building man pages is now optional which should be useful
7150 for those building systemd from git but unwilling to install
7153 * Watchdog support for supervising services is now usable. In
7154 a future release support for hardware watchdogs
7155 (i.e. /dev/watchdog) will be added building on this.
7157 * Service start rate limiting is now configurable and can be
7158 turned off per service. When a start rate limit is hit a
7159 reboot can automatically be triggered.
7161 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
7163 Contributions from: Benjamin Franzke, Bill Nottingham,
7164 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
7165 Schmidt, Michał Górny, Piotr Drąg
7169 * The systemd binary is installed /usr/lib/systemd/systemd now;
7170 An existing /sbin/init symlink needs to be adapted with the
7173 * The code that loads kernel modules has been ported to invoke
7174 libkmod directly, instead of modprobe. This means we do not
7175 support systems with module-init-tools anymore.
7177 * Watchdog support is now already useful, but still not
7180 * A new kernel command line option systemd.setenv= is
7181 understood to set system wide environment variables
7182 dynamically at boot.
7184 * We now limit the set of capabilities of systemd-journald.
7186 * We now set SIGPIPE to ignore by default, since it only is
7187 useful in shell pipelines, and has little use in general
7188 code. This can be disabled with IgnoreSIPIPE=no in unit
7191 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
7192 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
7197 * This is mostly a bugfix release
7199 * We now expose the reason why a service failed in the
7200 "Result" D-Bus property.
7202 * Rudimentary service watchdog support (will be completed over
7203 the next few releases.)
7205 * When systemd forks off in order execute some service we will
7206 now immediately changes its argv[0] to reflect which process
7207 it will execute. This is useful to minimize the time window
7208 with a generic argv[0], which makes bootcharts more useful
7210 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
7211 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
7212 Mike Kazantsev, Ray Strode
7216 * This is mostly a test release, but incorporates many
7219 * New systemd-cgtop tool to show control groups by their
7222 * Linking against libacl for ACLs is optional again. If
7223 disabled, support tracking device access for active logins
7224 goes becomes unavailable, and so does access to the user
7225 journals by the respective users.
7227 * If a group "adm" exists, journal files are automatically
7228 owned by them, thus allow members of this group full access
7229 to the system journal as well as all user journals.
7231 * The journal now stores the SELinux context of the logging
7232 client for all entries.
7234 * Add C++ inclusion guards to all public headers
7236 * New output mode "cat" in the journal to print only text
7237 messages, without any meta data like date or time.
7239 * Include tiny X server wrapper as a temporary stop-gap to
7240 teach XOrg udev display enumeration. This is used by display
7241 managers such as gdm, and will go away as soon as XOrg
7242 learned native udev hotplugging for display devices.
7244 * Add new systemd-cat tool for executing arbitrary programs
7245 with STDERR/STDOUT connected to the journal. Can also act as
7246 BSD logger replacement, and does so by default.
7248 * Optionally store all locally generated coredumps in the
7249 journal along with meta data.
7251 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
7252 writing short strings to files (for usage for /sys), and for
7253 creating symlinks, character and block device nodes.
7255 * New unit file option ControlGroupPersistent= to make cgroups
7256 persistent, following the mechanisms outlined in
7257 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
7259 * Support multiple local RTCs in a sane way
7261 * No longer monopolize IO when replaying readahead data on
7262 rotating disks, since we might starve non-file-system IO to
7263 death, since fanotify() will not see accesses done by blkid,
7266 * Do not show kernel threads in systemd-cgls anymore, unless
7267 requested with new -k switch.
7269 Contributions from: Dan Horák, Kay Sievers, Lennart
7270 Poettering, Michal Schmidt
7274 * This is mostly a test release, but incorporates many
7277 * The git repository moved to:
7278 git://anongit.freedesktop.org/systemd/systemd
7279 ssh://git.freedesktop.org/git/systemd/systemd
7281 * First release with the journal
7282 http://0pointer.de/blog/projects/the-journal.html
7284 * The journal replaces both systemd-kmsg-syslogd and
7285 systemd-stdout-bridge.
7287 * New sd_pid_get_unit() API call in libsystemd-logind
7289 * Many systemadm clean-ups
7291 * Introduce remote-fs-pre.target which is ordered before all
7292 remote mounts and may be used to start services before all
7295 * Added Mageia support
7297 * Add bash completion for systemd-loginctl
7299 * Actively monitor PID file creation for daemons which exit in
7300 the parent process before having finished writing the PID
7301 file in the daemon process. Daemons which do this need to be
7302 fixed (i.e. PID file creation must have finished before the
7303 parent exits), but we now react a bit more gracefully to them.
7305 * Add colourful boot output, mimicking the well-known output
7306 of existing distributions.
7308 * New option PassCredentials= for socket units, for
7309 compatibility with a recent kernel ABI breakage.
7311 * /etc/rc.local is now hooked in via a generator binary, and
7312 thus will no longer act as synchronization point during
7315 * systemctl list-unit-files now supports --root=.
7317 * systemd-tmpfiles now understands two new commands: z, Z for
7318 relabelling files according to the SELinux database. This is
7319 useful to apply SELinux labels to specific files in /sys,
7322 * Output of SysV services is now forwarded to both the console
7323 and the journal by default, not only just the console.
7325 * New man pages for all APIs from libsystemd-login.
7327 * The build tree got reorganized and the build system is a
7328 lot more modular allowing embedded setups to specifically
7329 select the components of systemd they are interested in.
7331 * Support for Linux systems lacking the kernel VT subsystem is
7334 * configure's --with-rootdir= got renamed to
7335 --with-rootprefix= to follow the naming used by udev and
7338 * Unless specified otherwise we will now install to /usr instead
7339 of /usr/local by default.
7341 * Processes with '@' in argv[0][0] are now excluded from the
7342 final shut-down killing spree, following the logic explained
7344 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
7346 * All processes remaining in a service cgroup when we enter
7347 the START or START_PRE states are now killed with
7348 SIGKILL. That means it is no longer possible to spawn
7349 background processes from ExecStart= lines (which was never
7350 supported anyway, and bad style).
7352 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
7353 reloading of units together.
7355 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
7356 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
7357 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
7358 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
7359 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek