1 systemd System and Service Manager
5 * systemd now depends on util-linux v2.27. More specifically,
6 the newly added mount monitor feature in libmount now
7 replaces systemd's former own implementation.
9 * libmount mandates /etc/mtab not to be regular file, and
10 systemd now enforces this condition at early boot.
11 /etc/mtab has been deprecated and warned about for a very
12 long time, so systems running systemd should already have
13 stopped having this file around as anything else than a
14 symlink to /proc/self/mounts.
16 * Support for the "pids" cgroup controller has been added. It
17 allows accounting the number of tasks in a cgroup and
18 enforcing limits on it. This adds two new setting
19 TasksAccounting= and TasksMax= to each unit, as well as a
20 global option DefaultTasksAccounting=.
22 * Support for the "net_cls" cgroup controller has been added.
23 It allows assigning a net class ID to each task in the
24 cgroup, which can then be used in firewall rules and traffic
25 shaping configurations. Note that the kernel netfilter net
26 class code does not currently work reliably for ingress
27 packets on unestablished sockets.
29 This adds a new config directive called NetClass= to CGroup
30 enabled units. Allowed values are positive numbers for fixed
31 assignments and "auto" for picking a free value
34 * 'systemctl is-system-running' now returns 'offline' if the
35 system is not booted with systemd. This command can now be
36 used as a substitute for 'systemd-notify --booted'.
38 * Watchdog timeouts have been increased to 3 minutes for all
39 in-tree service files. Apparently, disk IO issues are more
40 frequent than we hoped, and user reported >1 minute waiting
43 * 'machine-id-commit' functionality has been merged into
44 'machine-id-setup --commit'. The separate binary has been
47 * The WorkingDirectory= directive in unit files may now be set
48 to the special value '~'. In this case, the working
49 directory is set to the home directory of the user
52 * "machinectl shell" will now open the shell in the home
53 directory of the selected user by default.
55 * The CrashChVT= configuration file setting is renamed to
56 CrashChangeVT=, following our usual logic of not
57 abbreviating unnecessarily. The old directive is still
58 supported for compat reasons. Also, this directive now takes
59 an integer value between 1 and 63, or a boolean value. The
60 formerly supported '-1' value for disabling stays around for
63 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
64 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
65 RootDirectory= properties can now be set for transient
68 * The systemd-analyze tool gained a new "set-log-target" verb
69 to change the logging target the system manager logs to
70 dynamically during runtime. This is similar to how
71 "systemd-analyze set-log-level" already changes the log
74 * In nspawn /sys is now mounted as tmpfs, with only a selected
75 set of subdirectories mounted in from the real sysfs. This
76 enhances security slightly, and is useful for ensuring user
77 namespaces work correctly.
79 * Support for USB FunctionFS activation has been added. This
80 allows implementation of USB gadget services that are
81 activated as soon as they are requested, so that they don't
82 have to run continously, similar to classic socket
85 * The "systemctl exit" command now optionally takes an
86 additional parameter that sets the exit code to return from
87 the systemd manager when exiting. This is only relevant when
88 running the systemd user instance, or when running the
89 system instance in a container.
91 * sd-bus gained the new API calls sd_bus_path_encode_many()
92 and sd_bus_path_decode_many() that allow easy encoding and
93 decoding of multiple identifier strings inside a D-Bus
94 object path. Another new call sd_bus_default_flush_close()
95 has been added to flush and close per-thread default
98 * systemd-cgtop gained support for a -M/--machine= switch to
99 show the control groups within a certain container only.
101 * "systemctl kill" gained support for an optional --fail
102 switch. If specified the requested operation will fail of no
103 processes have been killed, because the unit had no
104 processes attached, or similar.
106 * A new systemd.crash_reboot=1 kernel command line option has
107 been added that triggers a reboot after crashing. This can
108 also be set through CrashReboot= in systemd.conf.
110 * The RuntimeDirectory= setting now understands unit
111 specifiers like %i or %f.
113 * A new (still internal) libary API sd-ipv4acd has been added,
114 that implements address conflict detection for IPv4. It's
115 based on code from sd-ipv4ll, and will be useful for
116 detecting DHCP address conflicts.
118 * File descriptors passed during socket activation may now be
119 named. A new API sd_listen_fds_with_names() is added to
120 access the names. The default names may be overriden,
121 either in the .socket file using the FileDescriptorName=
122 parameter, or by passing FDNAME= when storing the file
123 descriptors using sd_notify().
125 * systemd-networkd gained support for:
127 - Setting the IPv6 Router Advertisment settings via
128 IPv6AcceptRouterAdvertisements= in .network files.
130 - Configuring the HelloTimeSec=, MaxAgeSec= and
131 ForwardDelaySec= bridge parameters in .netdev files.
133 - Configuring PreferredSource= for static routes in
136 * The "ask-password" framework used to query for LUKS harddisk
137 passwords or SSL passwords during boot gained support for
138 caching passwords in the kernel keyring, if it is
139 available. This makes sure that the user only has to type in
140 a passphrase once if there are multiple objects to unlock
141 with the same one. Previously, such password caching was
142 available only when Plymouth was used; this moves the
143 caching logic into the systemd codebase itself. The
144 "systemd-ask-password" utility gained a new --keyname=
145 switch to control which kernel keyring key to use for
146 caching a password in. This functionality is also useful for
147 enabling display managers such as gdm to automatically
148 unlock the user's GNOME keyring if its passphrase, the
149 user's password and the harddisk password are the same, if
150 gdm-autologin is used.
152 * When downloading tar or raw images using "machinectl
153 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
154 file is now also downloaded, if it is available and stored
155 next to the image file.
157 * Units of type ".socket" gained a new boolean setting
158 Writable= which is only useful in conjunction with
159 ListenSpecial=. If true, enables opening the specified
160 special file in O_RDWR mode rather than O_RDONLY mode.
162 * systemd-rfkill has been reworked to become a singleton
163 service that is activated through /dev/rfkill on each rfkill
164 state change and saves the settings to disk. This way,
165 systemd-rfkill is now compatible with devices that exist
166 only intermittendly, and even restores state if the previous
167 system shutdown was abrupt rather than clean.
169 * The journal daemon gained support for vacuuming old journal
170 files controlled by the number of files that shall remain,
171 in addition to the already existing control by size and by
172 date. This is useful as journal interleaving performance
173 degrades with too many seperate journal files, and allows
174 putting an effective limit on them. The new setting defaults
175 to 100, but this may be changed by setting SystemMaxFiles=
176 and RuntimeMaxFiles= in journald.conf. Also, the
177 "journalctl" tool gained the new --vacuum-files= switch to
178 manually vacuum journal files to leave only the specified
179 number of files in place.
181 * udev will now create /dev/disk/by-path links for ATA devices
182 on kernels where that is supported.
184 * Galician, Serbian, Turkish and Korean translations were added.
186 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
187 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
188 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
189 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
190 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
191 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
192 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
193 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
194 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
195 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
196 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
197 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
198 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
199 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
200 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
201 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
202 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
203 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
205 -- Berlin, 2015-10-07
209 * The DHCP implementation of systemd-networkd gained a set of
212 - The DHCP server now supports emitting DNS and NTP
213 information. It may be enabled and configured via
214 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
215 and NTP information is enabled, but no servers are
216 configured, the corresponding uplink information (if there
217 is any) is propagated.
219 - Server and client now support transmission and reception
220 of timezone information. It can be configured via the
221 newly introduced network options UseTimezone=,
222 EmitTimezone=, and Timezone=. Transmission of timezone
223 information is enabled between host and containers by
224 default now: the container will change its local timezone
225 to what the host has set.
227 - Lease timeouts can now be configured via
228 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
230 - The DHCP server improved on the stability of
231 leases. Clients are more likely to get the same lease
232 information back, even if the server loses state.
234 - The DHCP server supports two new configuration options to
235 control the lease address pool metrics, PoolOffset= and
238 * The encapsulation limit of tunnels in systemd-networkd may
239 now be configured via 'EncapsulationLimit='. It allows
240 modifying the maximum additional levels of encapsulation
241 that are permitted to be prepended to a packet.
243 * systemd now supports the concept of user buses replacing
244 session buses, if used with dbus-1.10 (and enabled via dbus
245 --enable-user-session). It previously only supported this on
246 kdbus-enabled systems, and this release expands this to
247 'dbus-daemon' systems.
249 * systemd-networkd now supports predictable interface names
252 * systemd now optionally supports the new Linux kernel
253 "unified" control group hierarchy. If enabled via the kernel
254 command-line option 'systemd.unified_cgroup_hierarchy=1',
255 systemd will try to mount the unified cgroup hierarchy
256 directly on /sys/fs/cgroup. If not enabled, or not
257 available, systemd will fall back to the legacy cgroup
258 hierarchy setup, as before. Host system and containers can
259 mix and match legacy and unified hierarchies as they
260 wish. nspawn understands the $UNIFIED_CROUP_HIERARCHY
261 environment variable to individually select the hierarchy to
262 use for executed containers. By default, nspawn will use the
263 unified hierarchy for the containers if the host uses the
264 unified hierarchy, and the legacy hierarchy otherwise.
265 Please note that at this point the unified hierarchy is an
266 experimental kernel feature and is likely to change in one
267 of the next kernel releases. Therefore, it should not be
268 enabled by default in downstream distributions yet. The
269 minimum required kernel version for the unified hierarchy to
270 work is 4.2. Note that when the unified hierarchy is used
271 for the first time delegated access to controllers is
272 safe. Because of this systemd-nspawn containers will get
273 access to controllers now, as will systemd user
274 sessions. This means containers and user sessions may now
275 manage their own resources, partitioning up what the system
278 * A new special scope unit "init.scope" has been introduced
279 that encapsulates PID 1 of the system. It may be used to
280 determine resource usage and enforce resource limits on PID
281 1 itself. PID 1 hence moved out of the root of the control
284 * The cgtop tool gained support for filtering out kernel
285 threads when counting tasks in a control group. Also, the
286 count of processes is now recursively summed up by
287 default. Two options -k and --recursive= have been added to
288 revert to old behaviour. The tool has also been updated to
289 work correctly in containers now.
291 * systemd-nspawn's --bind= and --bind-ro= options have been
292 extended to allow creation of non-recursive bind mounts.
294 * libsystemd gained two new calls sd_pid_get_cgroup() and
295 sd_peer_get_cgroup() which return the control group path of
296 a process or peer of a connected AF_UNIX socket. This
297 function call is particularly useful when implementing
298 delegated subtrees support in the control group hierarchy.
300 * The "sd-event" event loop API of libsystemd now supports
301 correct dequeuing of real-time signals, without losing
304 * When systemd requests a PolicyKit decision when managing
305 units it will now add additional fields to the request,
306 including unit name and desired operation. This enables more
307 powerful PolicyKit policies, that make decisions depending
310 * nspawn learnt support for .nspawn settings files, that may
311 accompany the image files or directories of containers, and
312 may contain additional settings for the container. This is
313 an alternative to configuring container parameters via the
316 Contributions from: Cristian Rodríguez, Daniel Mack, David
317 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
318 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
319 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
320 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
321 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
322 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
323 Andersen, Tom Gundersen, Torstein Husebø
325 -- Berlin, 2015-09-08
329 * machinectl gained a new verb 'shell' which opens a fresh
330 shell on the target container or the host. It is similar to
331 the existing 'login' command of machinectl, but spawns the
332 shell directly without prompting for username or
333 password. The pseudo machine '.host' now refers to the local
334 host and is used by default. Hence, 'machinectl shell' can
335 be used as replacement for 'su -' which spawns a session as
336 a fresh systemd unit in a way that is fully isolated from
337 the originating session.
339 * systemd-networkd learned to cope with private-zone DHCP
340 options and allows other programs to query the values.
342 * SELinux access control when enabling/disabling units is no
343 longer enforced with this release. The previous
344 implementation was incorrect, and a new corrected
345 implementation is not yet available. As unit file operations
346 are still protected via PolicyKit and D-Bus policy this is
347 not a security problem. Yet, distributions which care about
348 optimal SELinux support should probably not stabilize on
351 * sd-bus gained support for matches of type "arg0has=", that
352 test for membership of strings in string arrays sent in bus
355 * systemd-resolved now dumps the contents of its DNS and LLMNR
356 caches to the logs on reception of the SIGUSR1 signal. This
357 is useful to debug DNS behaviour.
359 * The coredumpctl tool gained a new --directory= option to
360 operate on journal files in a specific directory.
362 * "systemctl reboot" and related commands gained a new
363 "--message=" option which may be used to set a free-text
364 wall message when shutting down or rebooting the
365 system. This message is also logged, which is useful for
366 figuring out the reason for a reboot or shutdown a
369 * The "systemd-resolve-host" tool's -i switch now takes
370 network interface numbers as alternative to interface names.
372 * A new unit file setting for services has been introduced:
373 UtmpMode= allows configuration of how precisely systemd
374 handles utmp and wtmp entries for the service if this is
375 enabled. This allows writing services that appear similar to
376 user sessions in the output of the "w", "who", "last" and
379 * systemd-resolved will now locally synthesize DNS resource
380 records for the "localhost" and "gateway" domains as well as
381 the local hostname. This should ensure that clients querying
382 RRs via resolved will get similar results as those going via
383 NSS, if nss-myhostname is enabled.
385 Contributions from: Alastair Hughes, Alex Crawford, Daniel
386 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
387 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
388 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
389 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
390 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
391 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
392 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
393 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
394 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
395 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
396 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
398 -- Berlin, 2015-08-27
402 * The systemd-efi-boot-generator functionality was merged into
403 systemd-gpt-auto-generator.
405 * systemd-networkd now supports Group Policy for vxlan
406 devices. It can be enabled via the new boolean configuration
407 option called 'GroupPolicyExtension='.
409 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
410 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
411 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
413 -- Berlin, 2015-07-31
417 * The python-systemd code has been removed from the systemd repository.
418 A new repository has been created which accommodates the code from
419 now on, and we kindly ask distributions to create a separate package
420 for this: https://github.com/systemd/python-systemd
422 * The systemd daemon will now reload its main configuration
423 (/etc/systemd/system.conf) on daemon-reload.
425 * sd-dhcp now exposes vendor specific extensions via
426 sd_dhcp_lease_get_vendor_specific().
428 * systemd-networkd gained a number of new configuration options.
430 - A new boolean configuration option for TAP devices called
431 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
432 device, thus allowing to send and receive GSO packets.
434 - A new tunnel configuration option called 'CopyDSCP='.
435 If enabled, the DSCP field of ip6 tunnels is copied into the
438 - A set of boolean bridge configuration options were added.
439 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
440 and 'UnicastFlood=' are now parsed by networkd and applied to the
441 respective bridge link device via the respective IFLA_BRPORT_*
444 - A new string configuration option to override the hostname sent
445 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
446 is true, networkd will use the configured hostname instead of the
447 system hostname when sending DHCP requests.
449 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
450 networkd will configure the IPv6 flow-label of the tunnel device
451 according to RFC2460.
453 - The 'macvtap' virtual network devices are now supported, similar to
454 the already supported 'macvlan' devices.
456 * systemd-resolved now implements RFC5452 to improve resilience against
457 cache poisoning. Additionally, source port randomization is enabled
458 by default to further protect against DNS spoofing attacks.
460 * nss-mymachines now supports translating UIDs and GIDs of running
461 containers with user-namespaces enabled. If a container 'foo'
462 translates a host uid 'UID' to the container uid 'TUID', then
463 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
464 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
465 mapped as 'vg-foo-TGID'.
467 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
468 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
469 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
470 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
471 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
472 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
473 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
474 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
475 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
476 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
478 -- Berlin, 2015-07-29
482 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
483 There are no known issues with current sysfs, and udev does not need
484 or should be used to work around such bugs.
486 * udev does no longer enable USB HID power management. Several reports
487 indicate, that some devices cannot handle that setting.
489 * The udev accelerometer helper was removed. The functionality
490 is now fully included in iio-sensor-proxy. But this means,
491 older iio-sensor-proxy versions will no longer provide
492 accelerometer/orientation data with this systemd version.
493 Please upgrade iio-sensor-proxy to version 1.0.
495 * networkd gained a new configuration option IPv6PrivacyExtensions=
496 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
497 for Stateless Address") on selected networks.
499 * For the sake of fewer build-time dependencies and less code in the
500 main repository, the python bindings are about to be removed in the
501 next release. A new repository has been created which accommodates
502 the code from now on, and we kindly ask distributions to create a
503 separate package for this. The removal will take place in v223.
505 https://github.com/systemd/python-systemd
507 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
508 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
509 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
510 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
511 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
512 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
513 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
514 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
515 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
516 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
518 -- Berlin, 2015-07-07
522 * The sd-bus.h and sd-event.h APIs have now been declared
523 stable and have been added to the official interface of
524 libsystemd.so. sd-bus implements an alternative D-Bus client
525 library, that is relatively easy to use, very efficient and
526 supports both classic D-Bus as well as kdbus as transport
527 backend. sd-event is a generic event loop abstraction that
528 is built around Linux epoll, but adds features such as event
529 prioritization or efficient timer handling. Both APIs are good
530 choices for C programs looking for a bus and/or event loop
531 implementation that is minimal and does not have to be
532 portable to other kernels.
534 * kdbus support is no longer compile-time optional. It is now
535 always built-in. However, it can still be disabled at
536 runtime using the kdbus=0 kernel command line setting, and
537 that setting may be changed to default to off, by specifying
538 --disable-kdbus at build-time. Note though that the kernel
539 command line setting has no effect if the kdbus.ko kernel
540 module is not installed, in which case kdbus is (obviously)
541 also disabled. We encourage all downstream distributions to
542 begin testing kdbus by adding it to the kernel images in the
543 development distributions, and leaving kdbus support in
546 * The minimal required util-linux version has been bumped to
549 * Support for chkconfig (--enable-chkconfig) was removed in
550 favor of calling an abstraction tool
551 /lib/systemd/systemd-sysv-install. This needs to be
552 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
553 in README for details.
555 * If there's a systemd unit and a SysV init script for the
556 same service name, and the user executes "systemctl enable"
557 for it (or a related call), then this will now enable both
558 (or execute the related operation on both), not just the
561 * The libudev API documentation has been converted from gtkdoc
564 * gudev has been removed from the systemd tree, it is now an
567 * The systemd-cgtop tool learnt a new --raw switch to generate
568 "raw" (machine parsable) output.
570 * networkd's IPForwarding= .network file setting learnt the
571 new setting "kernel", which ensures that networkd does not
572 change the IP forwarding sysctl from the default kernel
575 * The systemd-logind bus API now exposes a new boolean
576 property "Docked" that reports whether logind considers the
577 system "docked", i.e. connected to a docking station or not.
579 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
580 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
581 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
582 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
583 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
584 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
585 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
586 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
587 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
588 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
589 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
590 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
591 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
592 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
593 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
594 Fink, Zbigniew Jędrzejewski-Szmek
596 -- Berlin, 2015-06-19
600 * The gudev library has been extracted into a separate repository
601 available at: https://git.gnome.org/browse/libgudev/
602 It is now managed as part of the Gnome project. Distributions
603 are recommended to pass --disable-gudev to systemd and use
604 gudev from the Gnome project instead. gudev is still included
605 in systemd, for now. It will be removed soon, though. Please
606 also see the announcement-thread on systemd-devel:
607 http://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
609 * systemd now exposes a CPUUsageNSec= property for each
610 service unit on the bus, that contains the overall consumed
611 CPU time of a service (the sum of what each process of the
612 service consumed). This value is only available if
613 CPUAccounting= is turned on for a service, and is then shown
614 in the "systemctl status" output.
616 * Support for configuring alternative mappings of the old SysV
617 runlevels to systemd targets has been removed. They are now
618 hardcoded in a way that runlevels 2, 3, 4 all map to
619 multi-user.target and 5 to graphical.target (which
620 previously was already the default behaviour).
622 * The auto-mounter logic gained support for mount point
623 expiry, using a new TimeoutIdleSec= setting in .automount
624 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
626 * The EFI System Partition (ESP) as mounted to /boot by
627 systemd-efi-boot-generator will now be unmounted
628 automatically after 2 minutes of not being used. This should
629 minimize the risk of ESP corruptions.
631 * New /etc/fstab options x-systemd.requires= and
632 x-systemd.requires-mounts-for= are now supported to express
633 additional dependencies for mounts. This is useful for
634 journalling file systems that support external journal
635 devices or overlay file systems that require underlying file
636 systems to be mounted.
638 * systemd does not support direct live-upgrades (via systemctl
639 daemon-reexec) from versions older than v44 anymore. As no
640 distribution we are aware of shipped such old versions in a
641 stable release this should not be problematic.
643 * When systemd forks off a new per-connection service instance
644 it will now set the $REMOTE_ADDR environment variable to the
645 remote IP address, and $REMOTE_PORT environment variable to
646 the remote IP port. This behaviour is similar to the
647 corresponding environment variables defined by CGI.
649 * systemd-networkd gained support for uplink failure
650 detection. The BindCarrier= option allows binding interface
651 configuration dynamically to the link sense of other
652 interfaces. This is useful to achieve behaviour like in
655 * systemd-networkd gained support for configuring the DHCP
656 client identifier to use when requesting leases.
658 * systemd-networkd now has a per-network UseNTP= option to
659 configure whether NTP server information acquired via DHCP
660 is passed on to services like systemd-timesyncd.
662 * systemd-networkd gained support for vti6 tunnels.
664 * Note that systemd-networkd manages the sysctl variable
665 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
666 it is configured for since v219. The variable controls IP
667 forwarding, and is a per-interface alternative to the global
668 /proc/sys/net/ipv[46]/ip_forward. This setting is
669 configurable in the IPForward= option, which defaults to
670 "no". This means if networkd is used for an interface it is
671 no longer sufficient to set the global sysctl option to turn
672 on IP forwarding! Instead, the .network file option
673 IPForward= needs to be turned on! Note that the
674 implementation of this behaviour was broken in v219 and has
677 * Many bonding and vxlan options are now configurable in
680 * systemd-nspawn gained a new --property= setting to set unit
681 properties for the container scope. This is useful for
682 setting resource parameters (e.g "CPUShares=500") on
683 containers started from the command line.
685 * systemd-nspawn gained a new --private-users= switch to make
686 use of user namespacing available on recent Linux kernels.
688 * systemd-nspawn may now be called as part of a shell pipeline
689 in which case the pipes used for stdin and stdout are passed
690 directly to the process invoked in the container, without
691 indirection via a pseudo tty.
693 * systemd-nspawn gained a new switch to control the UNIX
694 signal to use when killing the init process of the container
697 * systemd-nspawn gained a new --overlay= switch for mounting
698 overlay file systems into the container using the new kernel
701 * When a container image is imported via systemd-importd and
702 the host file system is not btrfs, a loopback block device
703 file is created in /var/lib/machines.raw with a btrfs file
704 system inside. It is then mounted to /var/lib/machines to
705 enable btrfs features for container management. The loopback
706 file and btrfs file system is grown as needed when container
707 images are imported via systemd-importd.
709 * systemd-machined/systemd-importd gained support for btrfs
710 quota, to enforce container disk space limits on disk. This
711 is exposed in "machinectl set-limit".
713 * systemd-importd now can import containers from local .tar,
714 .raw and .qcow2 images, and export them to .tar and .raw. It
715 can also import dkr v2 images now from the network (on top
718 * systemd-importd gained support for verifying downloaded
719 images with gpg2 (previously only gpg1 was supported).
721 * systemd-machined, systemd-logind, systemd: most bus calls
722 are now accessible to unprivileged processes via
723 PolicyKit. Also, systemd-logind will now allow users to kill
724 their own sessions without further privileges or
727 * systemd-shutdownd has been removed. This service was
728 previously responsible for implementing scheduled shutdowns
729 as exposed in /usr/bin/shutdown's time parameter. This
730 functionality has now been moved into systemd-logind and is
731 accessible via a bus interface.
733 * "systemctl reboot" gained a new switch --firmware-setup that
734 can be used to reboot into the EFI firmware setup, if that
735 is available. systemd-logind now exposes an API on the bus
736 to trigger such reboots, in case graphical desktop UIs want
737 to cover this functionality.
739 * "systemctl enable", "systemctl disable" and "systemctl mask"
740 now support a new "--now" switch. If specified the units
741 that are enabled will also be started, and the ones
742 disabled/masked also stopped.
744 * The Gummiboot EFI boot loader tool has been merged into
745 systemd, and renamed to "systemd-boot". The bootctl tool has been
746 updated to support systemd-boot.
748 * An EFI kernel stub has been added that may be used to create
749 kernel EFI binaries that contain not only the actual kernel,
750 but also an initrd, boot splash, command line and OS release
751 information. This combined binary can then be signed as a
752 single image, so that the firmware can verify it all in one
753 step. systemd-boot has special support for EFI binaries created
754 like this and can extract OS release information from them
755 and show them in the boot menu. This functionality is useful
756 to implement cryptographically verified boot schemes.
758 * Optional support has been added to systemd-fsck to pass
759 fsck's progress report to an AF_UNIX socket in the file
762 * udev will no longer create device symlinks for all block
763 devices by default. A blacklist for excluding special block
764 devices from this logic has been turned into a whitelist
765 that requires picking block devices explicitly that require
768 * A new (currently still internal) API sd-device.h has been
769 added to libsystemd. This modernized API is supposed to
770 replace libudev eventually. In fact, already much of libudev
771 is now just a wrapper around sd-device.h.
773 * A new hwdb database for storing metadata about pointing
774 stick devices has been added.
776 * systemd-tmpfiles gained support for setting file attributes
777 similar to the "chattr" tool with new 'h' and 'H' lines.
779 * systemd-journald will no longer unconditionally set the
780 btrfs NOCOW flag on new journal files. This is instead done
781 with tmpfiles snippet using the new 'h' line type. This
782 allows easy disabling of this logic, by masking the
783 journal-nocow.conf tmpfiles file.
785 * systemd-journald will now translate audit message types to
786 human readable identifiers when writing them to the
787 journal. This should improve readability of audit messages.
789 * The LUKS logic gained support for the offset= and skip=
790 options in /etc/crypttab, as previously implemented by
793 * /usr/lib/os-release gained a new optional field VARIANT= for
794 distributions that support multiple variants (such as a
795 desktop edition, a server edition, ...)
797 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
798 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
799 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
800 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
801 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
802 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
803 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
804 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
805 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
806 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
807 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
808 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
809 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
810 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
811 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
812 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
813 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
814 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
815 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
816 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
817 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
818 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
819 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
820 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
821 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
822 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
823 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
825 -- Berlin, 2015-05-22
829 * Introduce a new API "sd-hwdb.h" for querying the hardware
830 metadata database. With this minimal interface one can query
831 and enumerate the udev hwdb, decoupled from the old libudev
832 library. libudev's interface for this is now only a wrapper
833 around sd-hwdb. A new tool systemd-hwdb has been added to
834 interface with and update the database.
836 * When any of systemd's tools copies files (for example due to
837 tmpfiles' C lines) a btrfs reflink will attempted first,
838 before bytewise copying is done.
840 * systemd-nspawn gained a new --ephemeral switch. When
841 specified a btrfs snapshot is taken of the container's root
842 directory, and immediately removed when the container
843 terminates again. Thus, a container can be started whose
844 changes never alter the container's root directory, and are
845 lost on container termination. This switch can also be used
846 for starting a container off the root file system of the
847 host without affecting the host OS. This switch is only
848 available on btrfs file systems.
850 * systemd-nspawn gained a new --template= switch. It takes the
851 path to a container tree to use as template for the tree
852 specified via --directory=, should that directory be
853 missing. This allows instantiating containers dynamically,
854 on first run. This switch is only available on btrfs file
857 * When a .mount unit refers to a mount point on which multiple
858 mounts are stacked, and the .mount unit is stopped all of
859 the stacked mount points will now be unmounted until no
862 * systemd now has an explicit notion of supported and
863 unsupported unit types. Jobs enqueued for unsupported unit
864 types will now fail with an "unsupported" error code. More
865 specifically .swap, .automount and .device units are not
866 supported in containers, .busname units are not supported on
867 non-kdbus systems. .swap and .automount are also not
868 supported if their respective kernel compile time options
871 * machinectl gained support for two new "copy-from" and
872 "copy-to" commands for copying files from a running
873 container to the host or vice versa.
875 * machinectl gained support for a new "bind" command to bind
876 mount host directories into local containers. This is
877 currently only supported for nspawn containers.
879 * networkd gained support for configuring bridge forwarding
880 database entries (fdb) from .network files.
882 * A new tiny daemon "systemd-importd" has been added that can
883 download container images in tar, raw, qcow2 or dkr formats,
884 and make them available locally in /var/lib/machines, so
885 that they can run as nspawn containers. The daemon can GPG
886 verify the downloads (not supported for dkr, since it has no
887 provisions for verifying downloads). It will transparently
888 decompress bz2, xz, gzip compressed downloads if necessary,
889 and restore sparse files on disk. The daemon uses privilege
890 separation to ensure the actual download logic runs with
891 fewer privileges than the daemon itself. machinectl has
892 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
893 make the functionality of importd available to the
894 user. With this in place the Fedora and Ubuntu "Cloud"
895 images can be downloaded and booted as containers unmodified
896 (the Fedora images lack the appropriate GPG signature files
897 currently, so they cannot be verified, but this will change
898 soon, hopefully). Note that downloading images is currently
899 only fully supported on btrfs.
901 * machinectl is now able to list container images found in
902 /var/lib/machines, along with some metadata about sizes of
903 disk and similar. If the directory is located on btrfs and
904 quota is enabled, this includes quota display. A new command
905 "image-status" has been added that shows additional
906 information about images.
908 * machinectl is now able to clone container images
909 efficiently, if the underlying file system (btrfs) supports
910 it, with the new "machinectl list-images" command. It also
911 gained commands for renaming and removing images, as well as
912 marking them read-only or read-write (supported also on
913 legacy file systems).
915 * networkd gained support for collecting LLDP network
916 announcements, from hardware that supports this. This is
917 shown in networkctl output.
919 * systemd-run gained support for a new -t (--pty) switch for
920 invoking a binary on a pty whose input and output is
921 connected to the invoking terminal. This allows executing
922 processes as system services while interactively
923 communicating with them via the terminal. Most interestingly
924 this is supported across container boundaries. Invoking
925 "systemd-run -t /bin/bash" is an alternative to running a
926 full login session, the difference being that the former
927 will not register a session, nor go through the PAM session
930 * tmpfiles gained support for a new "v" line type for creating
931 btrfs subvolumes. If the underlying file system is a legacy
932 file system, this automatically degrades to creating a
933 normal directory. Among others /var/lib/machines is now
934 created like this at boot, should it be missing.
936 * The directory /var/lib/containers/ has been deprecated and
937 been replaced by /var/lib/machines. The term "machines" has
938 been used in the systemd context as generic term for both
939 VMs and containers, and hence appears more appropriate for
940 this, as the directory can also contain raw images bootable
943 * systemd-nspawn when invoked with -M but without --directory=
944 or --image= is now capable of searching for the container
945 root directory, subvolume or disk image automatically, in
946 /var/lib/machines. systemd-nspawn@.service has been updated
947 to make use of this, thus allowing it to be used for raw
950 * A new machines.target unit has been introduced that is
951 supposed to group all containers/VMs invoked as services on
952 the system. systemd-nspawn@.service has been updated to
955 * machinectl gained a new "start" command, for invoking a
956 container as a service. "machinectl start foo" is mostly
957 equivalent to "systemctl start systemd-nspawn@foo.service",
958 but handles escaping in a nicer way.
960 * systemd-nspawn will now mount most of the cgroupfs tree
961 read-only into each container, with the exception of the
962 container's own subtree in the name=systemd hierarchy.
964 * journald now sets the special FS_NOCOW file flag for its
965 journal files. This should improve performance on btrfs, by
966 avoiding heavy fragmentation when journald's write-pattern
967 is used on COW file systems. It degrades btrfs' data
968 integrity guarantees for the files to the same levels as for
969 ext3/ext4 however. This should be OK though as journald does
970 its own data integrity checks and all its objects are
971 checksummed on disk. Also, journald should handle btrfs disk
972 full events a lot more gracefully now, by processing SIGBUS
973 errors, and not relying on fallocate() anymore.
975 * When journald detects that journal files it is writing to
976 have been deleted it will immediately start new journal
979 * systemd now provides a way to store file descriptors
980 per-service in PID 1.This is useful for daemons to ensure
981 that fds they require are not lost during a daemon
982 restart. The fds are passed to the daemon on the next
983 invocation in the same way socket activation fds are
984 passed. This is now used by journald to ensure that the
985 various sockets connected to all the system's stdout/stderr
986 are not lost when journald is restarted. File descriptors
987 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
988 an extension to sd_notify(). Note that a limit is enforced
989 on the number of fds a service can store in PID 1, and it
990 defaults to 0, so that no fds may be stored, unless this is
991 explicitly turned on.
993 * The default TERM variable to use for units connected to a
994 terminal, when no other value is explicitly is set is now
995 vt220 rather than vt102. This should be fairly safe still,
996 but allows PgUp/PgDn work.
998 * The /etc/crypttab option header= as known from Debian is now
1001 * "loginctl user-status" and "loginctl session-status" will
1002 now show the last 10 lines of log messages of the
1003 user/session following the status output. Similar,
1004 "machinectl status" will show the last 10 log lines
1005 associated with a virtual machine or container
1006 service. (Note that this is usually not the log messages
1007 done in the VM/container itself, but simply what the
1008 container manager logs. For nspawn this includes all console
1011 * "loginctl session-status" without further argument will now
1012 show the status of the session of the caller. Similar,
1013 "lock-session", "unlock-session", "activate",
1014 "enable-linger", "disable-linger" may now be called without
1015 session/user parameter in which case they apply to the
1016 caller's session/user.
1018 * An X11 session scriptlet is now shipped that uploads
1019 $DISPLAY and $XAUTHORITY into the environment of the systemd
1020 --user daemon if a session begins. This should improve
1021 compatibility with X11 enabled applications run as systemd
1024 * Generators are now subject to masking via /etc and /run, the
1025 same way as unit files.
1027 * networkd .network files gained support for configuring
1028 per-link IPv4/IPv6 packet forwarding as well as IPv4
1029 masquerading. This is by default turned on for veth links to
1030 containers, as registered by systemd-nspawn. This means that
1031 nspawn containers run with --network-veth will now get
1032 automatic routed access to the host's networks without any
1033 further configuration or setup, as long as networkd runs on
1036 * systemd-nspawn gained the --port= (-p) switch to expose TCP
1037 or UDP posts of a container on the host. With this in place
1038 it is possible to run containers with private veth links
1039 (--network-veth), and have their functionality exposed on
1040 the host as if their services were running directly on the
1043 * systemd-nspawn's --network-veth switch now gained a short
1044 version "-n", since with the changes above it is now truly
1045 useful out-of-the-box. The systemd-nspawn@.service has been
1046 updated to make use of it too by default.
1048 * systemd-nspawn will now maintain a per-image R/W lock, to
1049 ensure that the same image is not started more than once
1050 writable. (It's OK to run an image multiple times
1051 simultaneously in read-only mode.)
1053 * systemd-nspawn's --image= option is now capable of
1054 dissecting and booting MBR and GPT disk images that contain
1055 only a single active Linux partition. Previously it
1056 supported only GPT disk images with proper GPT type
1057 IDs. This allows running cloud images from major
1058 distributions directly with systemd-nspawn, without
1061 * In addition to collecting mouse dpi data in the udev
1062 hardware database, there's now support for collecting angle
1063 information for mouse scroll wheels. The database is
1064 supposed to guarantee similar scrolling behavior on mice
1065 that it knows about. There's also support for collecting
1066 information about Touchpad types.
1068 * udev's input_id built-in will now also collect touch screen
1069 dimension data and attach it to probed devices.
1071 * /etc/os-release gained support for a Distribution Privacy
1074 * networkd gained support for creating "ipvlan", "gretap",
1075 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
1077 * systemd-tmpfiles gained support for "a" lines for setting
1080 * systemd-nspawn will now mount /tmp in the container to
1081 tmpfs, automatically.
1083 * systemd now exposes the memory.usage_in_bytes cgroup
1084 attribute and shows it for each service in the "systemctl
1085 status" output, if available.
1087 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
1088 immediate reboot is triggered. This useful if shutdown is
1089 hung and is unable to complete, to expedite the
1090 operation. Note that this kind of reboot will still unmount
1091 all file systems, and hence should not result in fsck being
1094 * A .device unit for an optical block device will now be
1095 considered active only when a medium is in the drive. Also,
1096 mount units are now bound to their backing devices thus
1097 triggering automatic unmounting when devices become
1098 unavailable. With this in place systemd will now
1099 automatically unmount left-over mounts when a CD-ROM is
1100 ejected or an USB stick is yanked from the system.
1102 * networkd-wait-online now has support for waiting for
1103 specific interfaces only (with globbing), and for giving up
1104 after a configurable timeout.
1106 * networkd now exits when idle. It will be automatically
1107 restarted as soon as interfaces show up, are removed or
1108 change state. networkd will stay around as long as there is
1109 at least one DHCP state machine or similar around, that keep
1112 * networkd may now configure IPv6 link-local addressing in
1113 addition to IPv4 link-local addressing.
1115 * The IPv6 "token" for use in SLAAC may now be configured for
1116 each .network interface in networkd.
1118 * Routes configured with networkd may now be assigned a scope
1121 * networkd's [Match] sections now support globbing and lists
1122 of multiple space-separated matches per item.
1124 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
1125 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
1126 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
1127 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
1128 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
1129 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
1130 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
1131 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
1132 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
1133 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
1134 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
1135 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
1136 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
1137 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
1138 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
1139 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
1140 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
1141 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
1142 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
1143 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
1144 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
1145 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
1146 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
1147 Hoffmann, Zbigniew Jędrzejewski-Szmek
1149 -- Berlin, 2015-02-16
1153 * When querying unit file enablement status (for example via
1154 "systemctl is-enabled"), a new state "indirect" is now known
1155 which indicates that a unit might not be enabled itself, but
1156 another unit listed in its Also= setting might be.
1158 * Similar to the various existing ConditionXYZ= settings for
1159 units there are now matching AssertXYZ= settings. While
1160 failing conditions cause a unit to be skipped, but its job
1161 to succeed, failing assertions declared like this will cause
1162 a unit start operation and its job to fail.
1164 * hostnamed now knows a new chassis type "embedded".
1166 * systemctl gained a new "edit" command. When used on a unit
1167 file this allows extending unit files with .d/ drop-in
1168 configuration snippets or editing the full file (after
1169 copying it from /usr/lib to /etc). This will invoke the
1170 user's editor (as configured with $EDITOR), and reload the
1171 modified configuration after editing.
1173 * "systemctl status" now shows the suggested enablement state
1174 for a unit, as declared in the (usually vendor-supplied)
1175 system preset files.
1177 * nss-myhostname will now resolve the single-label host name
1178 "gateway" to the locally configured default IP routing
1179 gateways, ordered by their metrics. This assigns a stable
1180 name to the used gateways, regardless which ones are
1181 currently configured. Note that the name will only be
1182 resolved after all other name sources (if nss-myhostname is
1183 configured properly) and should hence not negatively impact
1184 systems that use the single-label host name "gateway" in
1187 * systemd-inhibit now allows filtering by mode when listing
1190 * Scope and service units gained a new "Delegate" boolean
1191 property, which when set allows processes running inside the
1192 unit to further partition resources. This is primarily
1193 useful for systemd user instances as well as container
1196 * journald will now pick up audit messages directly from
1197 the kernel, and log them like any other log message. The
1198 audit fields are split up and fully indexed. This means that
1199 journalctl in many ways is now a (nicer!) alternative to
1200 ausearch, the traditional audit client. Note that this
1201 implements only a minimal audit client, if you want the
1202 special audit modes like reboot-on-log-overflow, please use
1203 the traditional auditd instead, which can be used in
1204 parallel to journald.
1206 * The ConditionSecurity= unit file option now understands the
1207 special string "audit" to check whether auditing is
1210 * journalctl gained two new commands --vacuum-size= and
1211 --vacuum-time= to delete old journal files until the
1212 remaining ones take up no more the specified size on disk,
1213 or are not older than the specified time.
1215 * A new, native PPPoE library has been added to sd-network,
1216 systemd's library of light-weight networking protocols. This
1217 library will be used in a future version of networkd to
1218 enable PPPoE communication without an external pppd daemon.
1220 * The busctl tool now understands a new "capture" verb that
1221 works similar to "monitor", but writes a packet capture
1222 trace to STDOUT that can be redirected to a file which is
1223 compatible with libcap's capture file format. This can then
1224 be loaded in Wireshark and similar tools to inspect bus
1227 * The busctl tool now understands a new "tree" verb that shows
1228 the object trees of a specific service on the bus, or of all
1231 * The busctl tool now understands a new "introspect" verb that
1232 shows all interfaces and members of objects on the bus,
1233 including their signature and values. This is particularly
1234 useful to get more information about bus objects shown by
1235 the new "busctl tree" command.
1237 * The busctl tool now understands new verbs "call",
1238 "set-property" and "get-property" for invoking bus method
1239 calls, setting and getting bus object properties in a
1242 * busctl gained a new --augment-creds= argument that controls
1243 whether the tool shall augment credential information it
1244 gets from the bus with data from /proc, in a possibly
1247 * nspawn's --link-journal= switch gained two new values
1248 "try-guest" and "try-host" that work like "guest" and
1249 "host", but do not fail if the host has no persistent
1250 journalling enabled. -j is now equivalent to
1251 --link-journal=try-guest.
1253 * macvlan network devices created by nspawn will now have
1254 stable MAC addresses.
1256 * A new SmackProcessLabel= unit setting has been added, which
1257 controls the SMACK security label processes forked off by
1258 the respective unit shall use.
1260 * If compiled with --enable-xkbcommon, systemd-localed will
1261 verify x11 keymap settings by compiling the given keymap. It
1262 will spew out warnings if the compilation fails. This
1263 requires libxkbcommon to be installed.
1265 * When a coredump is collected a larger number of metadata
1266 fields is now collected and included in the journal records
1267 created for it. More specifically control group membership,
1268 environment variables, memory maps, working directory,
1269 chroot directory, /proc/$PID/status, and a list of open file
1270 descriptors is now stored in the log entry.
1272 * The udev hwdb now contains DPI information for mice. For
1275 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
1277 * All systemd programs that read standalone configuration
1278 files in /etc now also support a corresponding series of
1279 .conf.d configuration directories in /etc/, /run/,
1280 /usr/local/lib/, /usr/lib/, and (if configured with
1281 --enable-split-usr) /lib/. In particular, the following
1282 configuration files now have corresponding configuration
1283 directories: system.conf user.conf, logind.conf,
1284 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
1285 resolved.conf, timesyncd.conf, journal-remote.conf, and
1286 journal-upload.conf. Note that distributions should use the
1287 configuration directories in /usr/lib/; the directories in
1288 /etc/ are reserved for the system administrator.
1290 * systemd-rfkill will no longer take the rfkill device name
1291 into account when storing rfkill state on disk, as the name
1292 might be dynamically assigned and not stable. Instead, the
1293 ID_PATH udev variable combined with the rfkill type (wlan,
1294 bluetooth, ...) is used.
1296 * A new service systemd-machine-id-commit.service has been
1297 added. When used on systems where /etc is read-only during
1298 boot, and /etc/machine-id is not initialized (but an empty
1299 file), this service will copy the temporary machine ID
1300 created as replacement into /etc after the system is fully
1301 booted up. This is useful for systems that are freshly
1302 installed with a non-initialized machine ID, but should get
1303 a fixed machine ID for subsequent boots.
1305 * networkd's .netdev files now provide a large set of
1306 configuration parameters for VXLAN devices. Similar, the
1307 bridge port cost parameter is now configurable in .network
1308 files. There's also new support for configuring IP source
1309 routing. networkd .link files gained support for a new
1310 OriginalName= match that is useful to match against the
1311 original interface name the kernel assigned. .network files
1312 may include MTU= and MACAddress= fields for altering the MTU
1313 and MAC address while being connected to a specific network
1316 * The LUKS logic gained supported for configuring
1317 UUID-specific key files. There's also new support for naming
1318 LUKS device from the kernel command line, using the new
1319 luks.name= argument.
1321 * Timer units may now be transiently created via the bus API
1322 (this was previously already available for scope and service
1323 units). In addition it is now possible to create multiple
1324 transient units at the same time with a single bus call. The
1325 "systemd-run" tool has been updated to make use of this for
1326 running commands on a specified time, in at(1)-style.
1328 * tmpfiles gained support for "t" lines, for assigning
1329 extended attributes to files. Among other uses this may be
1330 used to assign SMACK labels to files.
1332 Contributions from: Alin Rauta, Alison Chaiken, Andrej
1333 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
1334 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
1335 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
1336 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
1337 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
1338 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
1339 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
1340 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
1341 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
1342 Michael Biebl, Michael Chapman, Michael Marineau, Michal
1343 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
1344 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
1345 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
1346 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
1347 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
1348 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
1349 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
1351 -- Berlin, 2014-12-10
1355 * journalctl gained the new options -t/--identifier= to match
1356 on the syslog identifier (aka "tag"), as well as --utc to
1357 show log timestamps in the UTC timezone. journalctl now also
1358 accepts -n/--lines=all to disable line capping in a pager.
1360 * journalctl gained a new switch, --flush, that synchronously
1361 flushes logs from /run/log/journal to /var/log/journal if
1362 persistent storage is enabled. systemd-journal-flush.service
1363 now waits until the operation is complete.
1365 * Services can notify the manager before they start a reload
1366 (by sending RELOADING=1) or shutdown (by sending
1367 STOPPING=1). This allows the manager to track and show the
1368 internal state of daemons and closes a race condition when
1369 the process is still running but has closed its D-Bus
1372 * Services with Type=oneshot do not have to have any ExecStart
1375 * User units are now loaded also from
1376 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
1377 /run/systemd/user directory that was already previously
1378 supported, but is under the control of the user.
1380 * Job timeouts (i.e. time-outs on the time a job that is
1381 queued stays in the run queue) can now optionally result in
1382 immediate reboot or power-off actions (JobTimeoutAction= and
1383 JobTimeoutRebootArgument=). This is useful on ".target"
1384 units, to limit the maximum time a target remains
1385 undispatched in the run queue, and to trigger an emergency
1386 operation in such a case. This is now used by default to
1387 turn off the system if boot-up (as defined by everything in
1388 basic.target) hangs and does not complete for at least
1389 15min. Also, if power-off or reboot hang for at least 30min
1390 an immediate power-off/reboot operation is triggered. This
1391 functionality is particularly useful to increase reliability
1392 on embedded devices, but also on laptops which might
1393 accidentally get powered on when carried in a backpack and
1394 whose boot stays stuck in a hard disk encryption passphrase
1397 * systemd-logind can be configured to also handle lid switch
1398 events even when the machine is docked or multiple displays
1399 are attached (HandleLidSwitchDocked= option).
1401 * A helper binary and a service have been added which can be
1402 used to resume from hibernation in the initramfs. A
1403 generator will parse the resume= option on the kernel
1404 command line to trigger resume.
1406 * A user console daemon systemd-consoled has been
1407 added. Currently, it is a preview, and will so far open a
1408 single terminal on each session of the user marked as
1409 Desktop=systemd-console.
1411 * Route metrics can be specified for DHCP routes added by
1414 * The SELinux context of socket-activated services can be set
1415 from the information provided by the networking stack
1416 (SELinuxContextFromNet= option).
1418 * Userspace firmware loading support has been removed and
1419 the minimum supported kernel version is thus bumped to 3.7.
1421 * Timeout for udev workers has been increased from 1 to 3
1422 minutes, but a warning will be printed after 1 minute to
1423 help diagnose kernel modules that take a long time to load.
1425 * Udev rules can now remove tags on devices with TAG-="foobar".
1427 * systemd's readahead implementation has been removed. In many
1428 circumstances it didn't give expected benefits even for
1429 rotational disk drives and was becoming less relevant in the
1430 age of SSDs. As none of the developers has been using
1431 rotating media anymore, and nobody stepped up to actively
1432 maintain this component of systemd it has now been removed.
1434 * Swap units can use Options= to specify discard options.
1435 Discard options specified for swaps in /etc/fstab are now
1438 * Docker containers are now detected as a separate type of
1441 * The Password Agent protocol gained support for queries where
1442 the user input is shown, useful e.g. for user names.
1443 systemd-ask-password gained a new --echo option to turn that
1446 * The default sysctl.d/ snippets will now set:
1448 net.core.default_qdisc = fq_codel
1450 This selects Fair Queuing Controlled Delay as the default
1451 queuing discipline for network interfaces. fq_codel helps
1452 fight the network bufferbloat problem. It is believed to be
1453 a good default with no tuning required for most workloads.
1454 Downstream distributions may override this choice. On 10Gbit
1455 servers that do not do forwarding, "fq" may perform better.
1456 Systems without a good clocksource should use "pfifo_fast".
1458 * If kdbus is enabled during build a new option BusPolicy= is
1459 available for service units, that allows locking all service
1460 processes into a stricter bus policy, in order to limit
1461 access to various bus services, or even hide most of them
1462 from the service's view entirely.
1464 * networkctl will now show the .network and .link file
1465 networkd has applied to a specific interface.
1467 * sd-login gained a new API call sd_session_get_desktop() to
1468 query which desktop environment has been selected for a
1471 * UNIX utmp support is now compile-time optional to support
1472 legacy-free systems.
1474 * systemctl gained two new commands "add-wants" and
1475 "add-requires" for pulling in units from specific targets
1478 * If the word "rescue" is specified on the kernel command line
1479 the system will now boot into rescue mode (aka
1480 rescue.target), which was previously available only by
1481 specifying "1" or "systemd.unit=rescue.target" on the kernel
1482 command line. This new kernel command line option nicely
1483 mirrors the already existing "emergency" kernel command line
1486 * New kernel command line options mount.usr=, mount.usrflags=,
1487 mount.usrfstype= have been added that match root=, rootflags=,
1488 rootfstype= but allow mounting a specific file system to
1491 * The $NOTIFY_SOCKET is now also passed to control processes of
1492 services, not only the main process.
1494 * This version reenables support for fsck's -l switch. This
1495 means at least version v2.25 of util-linux is required for
1496 operation, otherwise dead-locks on device nodes may
1497 occur. Again: you need to update util-linux to at least
1498 v2.25 when updating systemd to v217.
1500 * The "multi-seat-x" tool has been removed from systemd, as
1501 its functionality has been integrated into X servers 1.16,
1502 and the tool is hence redundant. It is recommended to update
1503 display managers invoking this tool to simply invoke X
1504 directly from now on, again.
1506 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
1507 message flag has been added for all of systemd's PolicyKit
1508 authenticated method calls has been added. In particular
1509 this now allows optional interactive authorization via
1510 PolicyKit for many of PID1's privileged operations such as
1511 unit file enabling and disabling.
1513 * "udevadm hwdb --update" learnt a new switch "--usr" for
1514 placing the rebuilt hardware database in /usr instead of
1515 /etc. When used only hardware database entries stored in
1516 /usr will be used, and any user database entries in /etc are
1517 ignored. This functionality is useful for vendors to ship a
1518 pre-built database on systems where local configuration is
1519 unnecessary or unlikely.
1521 * Calendar time specifications in .timer units now also
1522 understand the strings "semi-annually", "quarterly" and
1523 "minutely" as shortcuts (in addition to the preexisting
1524 "anually", "hourly", ...).
1526 * systemd-tmpfiles will now correctly create files in /dev
1527 at boot which are marked for creation only at boot. It is
1528 recommended to always create static device nodes with 'c!'
1529 and 'b!', so that they are created only at boot and not
1530 overwritten at runtime.
1532 * When the watchdog logic is used for a service (WatchdogSec=)
1533 and the watchdog timeout is hit the service will now be
1534 terminated with SIGABRT (instead of just SIGTERM), in order
1535 to make sure a proper coredump and backtrace is
1536 generated. This ensures that hanging services will result in
1537 similar coredump/backtrace behaviour as services that hit a
1540 Contributions from: Andreas Henriksson, Andrei Borzenkov,
1541 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
1542 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
1543 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
1544 Herrmann, David Sommerseth, David Strauss, Emil Renner
1545 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
1546 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
1547 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
1548 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
1549 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
1550 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
1551 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
1552 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
1553 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
1554 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
1555 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
1556 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
1557 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
1558 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
1559 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
1560 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
1563 -- Berlin, 2014-10-28
1567 * timedated no longer reads NTP implementation unit names from
1568 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
1569 implementations should add a
1571 Conflicts=systemd-timesyncd.service
1573 to their unit files to take over and replace systemd's NTP
1574 default functionality.
1576 * systemd-sysusers gained a new line type "r" for configuring
1577 which UID/GID ranges to allocate system users/groups
1578 from. Lines of type "u" may now add an additional column
1579 that specifies the home directory for the system user to be
1580 created. Also, systemd-sysusers may now optionally read user
1581 information from STDIN instead of a file. This is useful for
1582 invoking it from RPM preinst scriptlets that need to create
1583 users before the first RPM file is installed since these
1584 files might need to be owned by them. A new
1585 %sysusers_create_inline RPM macro has been introduced to do
1586 just that. systemd-sysusers now updates the shadow files as
1587 well as the user/group databases, which should enhance
1588 compatibility with certain tools like grpck.
1590 * A number of bus APIs of PID 1 now optionally consult
1591 PolicyKit to permit access for otherwise unprivileged
1592 clients under certain conditions. Note that this currently
1593 doesn't support interactive authentication yet, but this is
1594 expected to be added eventually, too.
1596 * /etc/machine-info now has new fields for configuring the
1597 deployment environment of the machine, as well as the
1598 location of the machine. hostnamectl has been updated with
1599 new command to update these fields.
1601 * systemd-timesyncd has been updated to automatically acquire
1602 NTP server information from systemd-networkd, which might
1603 have been discovered via DHCP.
1605 * systemd-resolved now includes a caching DNS stub resolver
1606 and a complete LLMNR name resolution implementation. A new
1607 NSS module "nss-resolve" has been added which can be used
1608 instead of glibc's own "nss-dns" to resolve hostnames via
1609 systemd-resolved. Hostnames, addresses and arbitrary RRs may
1610 be resolved via systemd-resolved D-Bus APIs. In contrast to
1611 the glibc internal resolver systemd-resolved is aware of
1612 multi-homed system, and keeps DNS server and caches separate
1613 and per-interface. Queries are sent simultaneously on all
1614 interfaces that have DNS servers configured, in order to
1615 properly handle VPNs and local LANs which might resolve
1616 separate sets of domain names. systemd-resolved may acquire
1617 DNS server information from systemd-networkd automatically,
1618 which in turn might have discovered them via DHCP. A tool
1619 "systemd-resolve-host" has been added that may be used to
1620 query the DNS logic in resolved. systemd-resolved implements
1621 IDNA and automatically uses IDNA or UTF-8 encoding depending
1622 on whether classic DNS or LLMNR is used as transport. In the
1623 next releases we intend to add a DNSSEC and mDNS/DNS-SD
1624 implementation to systemd-resolved.
1626 * A new NSS module nss-mymachines has been added, that
1627 automatically resolves the names of all local registered
1628 containers to their respective IP addresses.
1630 * A new client tool "networkctl" for systemd-networkd has been
1631 added. It currently is entirely passive and will query
1632 networking configuration from udev, rtnetlink and networkd,
1633 and present it to the user in a very friendly
1634 way. Eventually, we hope to extend it to become a full
1635 control utility for networkd.
1637 * .socket units gained a new DeferAcceptSec= setting that
1638 controls the kernels' TCP_DEFER_ACCEPT sockopt for
1639 TCP. Similar, support for controlling TCP keep-alive
1640 settings has been added (KeepAliveTimeSec=,
1641 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
1642 turning off Nagle's algorithm on TCP has been added
1645 * logind learned a new session type "web", for use in projects
1646 like Cockpit which register web clients as PAM sessions.
1648 * timer units with at least one OnCalendar= setting will now
1649 be started only after timer-sync.target has been
1650 reached. This way they will not elapse before the system
1651 clock has been corrected by a local NTP client or
1652 similar. This is particular useful on RTC-less embedded
1653 machines, that come up with an invalid system clock.
1655 * systemd-nspawn's --network-veth= switch should now result in
1656 stable MAC addresses for both the outer and the inner side
1659 * systemd-nspawn gained a new --volatile= switch for running
1660 container instances with /etc or /var unpopulated.
1662 * The kdbus client code has been updated to use the new Linux
1663 3.17 memfd subsystem instead of the old kdbus-specific one.
1665 * systemd-networkd's DHCP client and server now support
1666 FORCERENEW. There are also new configuration options to
1667 configure the vendor client identifier and broadcast mode
1670 * systemd will no longer inform the kernel about the current
1671 timezone, as this is necessarily incorrect and racy as the
1672 kernel has no understanding of DST and similar
1673 concepts. This hence means FAT timestamps will be always
1674 considered UTC, similar to what Android is already
1675 doing. Also, when the RTC is configured to the local time
1676 (rather than UTC) systemd will never synchronize back to it,
1677 as this might confuse Windows at a later boot.
1679 * systemd-analyze gained a new command "verify" for offline
1680 validation of unit files.
1682 * systemd-networkd gained support for a couple of additional
1683 settings for bonding networking setups. Also, the metric for
1684 statically configured routes may now be configured. For
1685 network interfaces where this is appropriate the peer IP
1686 address may now be configured.
1688 * systemd-networkd's DHCP client will no longer request
1689 broadcasting by default, as this tripped up some networks.
1690 For hardware where broadcast is required the feature should
1691 be switched back on using RequestBroadcast=yes.
1693 * systemd-networkd will now set up IPv4LL addresses (when
1694 enabled) even if DHCP is configured successfully.
1696 * udev will now default to respect network device names given
1697 by the kernel when the kernel indicates that these are
1698 predictable. This behavior can be tweaked by changing
1699 NamePolicy= in the relevant .link file.
1701 * A new library systemd-terminal has been added that
1702 implements full TTY stream parsing and rendering. This
1703 library is supposed to be used later on for implementing a
1704 full userspace VT subsystem, replacing the current kernel
1707 * A new tool systemd-journal-upload has been added to push
1708 journal data to a remote system running
1709 systemd-journal-remote.
1711 * journald will no longer forward all local data to another
1712 running syslog daemon. This change has been made because
1713 rsyslog (which appears to be the most commonly used syslog
1714 implementation these days) no longer makes use of this, and
1715 instead pulls the data out of the journal on its own. Since
1716 forwarding the messages to a non-existent syslog server is
1717 more expensive than we assumed we have now turned this
1718 off. If you run a syslog server that is not a recent rsyslog
1719 version, you have to turn this option on again
1720 (ForwardToSyslog= in journald.conf).
1722 * journald now optionally supports the LZ4 compressor for
1723 larger journal fields. This compressor should perform much
1724 better than XZ which was the previous default.
1726 * machinectl now shows the IP addresses of local containers,
1727 if it knows them, plus the interface name of the container.
1729 * A new tool "systemd-escape" has been added that makes it
1730 easy to escape strings to build unit names and similar.
1732 * sd_notify() messages may now include a new ERRNO= field
1733 which is parsed and collected by systemd and shown among the
1734 "systemctl status" output for a service.
1736 * A new component "systemd-firstboot" has been added that
1737 queries the most basic systemd information (timezone,
1738 hostname, root password) interactively on first
1739 boot. Alternatively it may also be used to provision these
1740 things offline on OS images installed into directories.
1742 * The default sysctl.d/ snippets will now set
1744 net.ipv4.conf.default.promote_secondaries=1
1746 This has the benefit of no flushing secondary IP addresses
1747 when primary addresses are removed.
1749 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
1750 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
1751 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
1752 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
1753 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
1754 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
1755 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
1756 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
1757 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
1758 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
1759 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
1760 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
1761 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
1762 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
1763 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
1765 -- Berlin, 2014-08-19
1769 * A new tool systemd-sysusers has been added. This tool
1770 creates system users and groups in /etc/passwd and
1771 /etc/group, based on static declarative system user/group
1772 definitions in /usr/lib/sysusers.d/. This is useful to
1773 enable factory resets and volatile systems that boot up with
1774 an empty /etc directory, and thus need system users and
1775 groups created during early boot. systemd now also ships
1776 with two default sysusers.d/ files for the most basic
1777 users and groups systemd and the core operating system
1780 * A new tmpfiles snippet has been added that rebuilds the
1781 essential files in /etc on boot, should they be missing.
1783 * A directive for ensuring automatic clean-up of
1784 /var/cache/man/ has been removed from the default
1785 configuration. This line should now be shipped by the man
1786 implementation. The necessary change has been made to the
1787 man-db implementation. Note that you need to update your man
1788 implementation to one that ships this line, otherwise no
1789 automatic clean-up of /var/cache/man will take place.
1791 * A new condition ConditionNeedsUpdate= has been added that
1792 may conditionalize services to only run when /etc or /var
1793 are "older" than the vendor operating system resources in
1794 /usr. This is useful for reconstructing or updating /etc
1795 after an offline update of /usr or a factory reset, on the
1796 next reboot. Services that want to run once after such an
1797 update or reset should use this condition and order
1798 themselves before the new systemd-update-done.service, which
1799 will mark the two directories as fully updated. A number of
1800 service files have been added making use of this, to rebuild
1801 the udev hardware database, the journald message catalog and
1802 dynamic loader cache (ldconfig). The systemd-sysusers tool
1803 described above also makes use of this now. With this in
1804 place it is now possible to start up a minimal operating
1805 system with /etc empty cleanly. For more information on the
1806 concepts involved see this recent blog story:
1808 http://0pointer.de/blog/projects/stateless.html
1810 * A new system group "input" has been introduced, and all
1811 input device nodes get this group assigned. This is useful
1812 for system-level software to get access to input devices. It
1813 complements what is already done for "audio" and "video".
1815 * systemd-networkd learnt minimal DHCPv4 server support in
1816 addition to the existing DHCPv4 client support. It also
1817 learnt DHCPv6 client and IPv6 Router Solicitation client
1818 support. The DHCPv4 client gained support for static routes
1819 passed in from the server. Note that the [DHCPv4] section
1820 known in older systemd-networkd versions has been renamed to
1821 [DHCP] and is now also used by the DHCPv6 client. Existing
1822 .network files using settings of this section should be
1823 updated, though compatibility is maintained. Optionally, the
1824 client hostname may now be sent to the DHCP server.
1826 * networkd gained support for vxlan virtual networks as well
1827 as tun/tap and dummy devices.
1829 * networkd gained support for automatic allocation of address
1830 ranges for interfaces from a system-wide pool of
1831 addresses. This is useful for dynamically managing a large
1832 number of interfaces with a single network configuration
1833 file. In particular this is useful to easily assign
1834 appropriate IP addresses to the veth links of a large number
1835 of nspawn instances.
1837 * RPM macros for processing sysusers, sysctl and binfmt
1838 drop-in snippets at package installation time have been
1841 * The /etc/os-release file should now be placed in
1842 /usr/lib/os-release. The old location is automatically
1843 created as symlink. /usr/lib is the more appropriate
1844 location of this file, since it shall actually describe the
1845 vendor operating system shipped in /usr, and not the
1846 configuration stored in /etc.
1848 * .mount units gained a new boolean SloppyOptions= setting
1849 that maps to mount(8)'s -s option which enables permissive
1850 parsing of unknown mount options.
1852 * tmpfiles learnt a new "L+" directive which creates a symlink
1853 but (unlike "L") deletes a pre-existing file first, should
1854 it already exist and not already be the correct
1855 symlink. Similar, "b+", "c+" and "p+" directives have been
1856 added as well, which create block and character devices, as
1857 well as fifos in the filesystem, possibly removing any
1858 pre-existing files of different types.
1860 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
1861 'argument' field (which so far specified the source to
1862 symlink/copy the files from) is now optional. If omitted the
1863 same file os copied from /usr/share/factory/ suffixed by the
1864 full destination path. This is useful for populating /etc
1865 with essential files, by copying them from vendor defaults
1866 shipped in /usr/share/factory/etc.
1868 * A new command "systemctl preset-all" has been added that
1869 applies the service preset settings to all installed unit
1870 files. A new switch --preset-mode= has been added that
1871 controls whether only enable or only disable operations
1874 * A new command "systemctl is-system-running" has been added
1875 that allows checking the overall state of the system, for
1876 example whether it is fully up and running.
1878 * When the system boots up with an empty /etc, the equivalent
1879 to "systemctl preset-all" is executed during early boot, to
1880 make sure all default services are enabled after a factory
1883 * systemd now contains a minimal preset file that enables the
1884 most basic services systemd ships by default.
1886 * Unit files' [Install] section gained a new DefaultInstance=
1887 field for defining the default instance to create if a
1888 template unit is enabled with no instance specified.
1890 * A new passive target cryptsetup-pre.target has been added
1891 that may be used by services that need to make they run and
1892 finish before the first LUKS cryptographic device is set up.
1894 * The /dev/loop-control and /dev/btrfs-control device nodes
1895 are now owned by the "disk" group by default, opening up
1896 access to this group.
1898 * systemd-coredump will now automatically generate a
1899 stack trace of all core dumps taking place on the system,
1900 based on elfutils' libdw library. This stack trace is logged
1903 * systemd-coredump may now optionally store coredumps directly
1904 on disk (in /var/lib/systemd/coredump, possibly compressed),
1905 instead of storing them unconditionally in the journal. This
1906 mode is the new default. A new configuration file
1907 /etc/systemd/coredump.conf has been added to configure this
1908 and other parameters of systemd-coredump.
1910 * coredumpctl gained a new "info" verb to show details about a
1911 specific coredump. A new switch "-1" has also been added
1912 that makes sure to only show information about the most
1913 recent entry instead of all entries. Also, as the tool is
1914 generally useful now the "systemd-" prefix of the binary
1915 name has been removed. Distributions that want to maintain
1916 compatibility with the old name should add a symlink from
1917 the old name to the new name.
1919 * journald's SplitMode= now defaults to "uid". This makes sure
1920 that unprivileged users can access their own coredumps with
1921 coredumpctl without restrictions.
1923 * New kernel command line options "systemd.wants=" (for
1924 pulling an additional unit during boot), "systemd.mask="
1925 (for masking a specific unit for the boot), and
1926 "systemd.debug-shell" (for enabling the debug shell on tty9)
1927 have been added. This is implemented in the new generator
1928 "systemd-debug-generator".
1930 * systemd-nspawn will now by default filter a couple of
1931 syscalls for containers, among them those required for
1932 kernel module loading, direct x86 IO port access, swap
1933 management, and kexec. Most importantly though
1934 open_by_handle_at() is now prohibited for containers,
1935 closing a hole similar to a recently discussed vulnerability
1936 in docker regarding access to files on file hierarchies the
1937 container should normally not have access to. Note that for
1938 nspawn we generally make no security claims anyway (and
1939 this is explicitly documented in the man page), so this is
1940 just a fix for one of the most obvious problems.
1942 * A new man page file-hierarchy(7) has been added that
1943 contains a minimized, modernized version of the file system
1944 layout systemd expects, similar in style to the FHS
1945 specification or hier(5). A new tool systemd-path(1) has
1946 been added to query many of these paths for the local
1949 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
1950 longer done. Since the directory now has a per-user size
1951 limit, and is cleaned on logout this appears unnecessary,
1952 in particular since this now brings the lifecycle of this
1953 directory closer in line with how IPC objects are handled.
1955 * systemd.pc now exports a number of additional directories,
1956 including $libdir (which is useful to identify the library
1957 path for the primary architecture of the system), and a
1958 couple of drop-in directories.
1960 * udev's predictable network interface names now use the dev_port
1961 sysfs attribute, introduced in linux 3.15 instead of dev_id to
1962 distinguish between ports of the same PCI function. dev_id should
1963 only be used for ports using the same HW address, hence the need
1966 * machined has been updated to export the OS version of a
1967 container (read from /etc/os-release and
1968 /usr/lib/os-release) on the bus. This is now shown in
1969 "machinectl status" for a machine.
1971 * A new service setting RestartForceExitStatus= has been
1972 added. If configured to a set of exit signals or process
1973 return values, the service will be restarted when the main
1974 daemon process exits with any of them, regardless of the
1977 * systemctl's -H switch for connecting to remote systemd
1978 machines has been extended so that it may be used to
1979 directly connect to a specific container on the
1980 host. "systemctl -H root@foobar:waldi" will now connect as
1981 user "root" to host "foobar", and then proceed directly to
1982 the container named "waldi". Note that currently you have to
1983 authenticate as user "root" for this to work, as entering
1984 containers is a privileged operation.
1986 Contributions from: Andreas Henriksson, Benjamin Steinwender,
1987 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
1988 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
1989 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
1990 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
1991 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
1992 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
1993 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
1994 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
1995 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
1996 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
1997 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
1999 -- Berlin, 2014-07-03
2003 * As an experimental feature, udev now tries to lock the
2004 disk device node (flock(LOCK_SH|LOCK_NB)) while it
2005 executes events for the disk or any of its partitions.
2006 Applications like partitioning programs can lock the
2007 disk device node (flock(LOCK_EX)) and claim temporary
2008 device ownership that way; udev will entirely skip all event
2009 handling for this disk and its partitions. If the disk
2010 was opened for writing, the close will trigger a partition
2011 table rescan in udev's "watch" facility, and if needed
2012 synthesize "change" events for the disk and all its partitions.
2013 This is now unconditionally enabled, and if it turns out to
2014 cause major problems, we might turn it on only for specific
2015 devices, or might need to disable it entirely. Device Mapper
2016 devices are excluded from this logic.
2018 * We temporarily dropped the "-l" switch for fsck invocations,
2019 since they collide with the flock() logic above. util-linux
2020 upstream has been changed already to avoid this conflict,
2021 and we will readd "-l" as soon as util-linux with this
2022 change has been released.
2024 * The dependency on libattr has been removed. Since a long
2025 time, the extended attribute calls have moved to glibc, and
2026 libattr is thus unnecessary.
2028 * Virtualization detection works without priviliges now. This
2029 means the systemd-detect-virt binary no longer requires
2030 CAP_SYS_PTRACE file capabilities, and our daemons can run
2031 with fewer privileges.
2033 * systemd-networkd now runs under its own "systemd-network"
2034 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
2035 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
2036 loses the ability to write to files owned by root this way.
2038 * Similar, systemd-resolved now runs under its own
2039 "systemd-resolve" user with no capabilities remaining.
2041 * Similar, systemd-bus-proxyd now runs under its own
2042 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
2044 * systemd-networkd gained support for setting up "veth"
2045 virtual ethernet devices for container connectivity, as well
2046 as GRE and VTI tunnels.
2048 * systemd-networkd will no longer automatically attempt to
2049 manually load kernel modules necessary for certain tunnel
2050 transports. Instead, it is assumed the kernel loads them
2051 automatically when required. This only works correctly on
2052 very new kernels. On older kernels, please consider adding
2053 the kernel modules to /etc/modules-load.d/ as a work-around.
2055 * The resolv.conf file systemd-resolved generates has been
2056 moved to /run/systemd/resolve/. If you have a symlink from
2057 /etc/resolv.conf, it might be necessary to correct it.
2059 * Two new service settings, ProtectHome= and ProtectSystem=,
2060 have been added. When enabled, they will make the user data
2061 (such as /home) inaccessible or read-only and the system
2062 (such as /usr) read-only, for specific services. This allows
2063 very light-weight per-service sandboxing to avoid
2064 modifications of user data or system files from
2065 services. These two new switches have been enabled for all
2066 of systemd's long-running services, where appropriate.
2068 * Socket units gained new SocketUser= and SocketGroup=
2069 settings to set the owner user and group of AF_UNIX sockets
2070 and FIFOs in the file system.
2072 * Socket units gained a new RemoveOnStop= setting. If enabled,
2073 all FIFOS and sockets in the file system will be removed
2074 when the specific socket unit is stopped.
2076 * Socket units gained a new Symlinks= setting. It takes a list
2077 of symlinks to create to file system sockets or FIFOs
2078 created by the specific Unix sockets. This is useful to
2079 manage symlinks to socket nodes with the same life-cycle as
2082 * The /dev/log socket and /dev/initctl FIFO have been moved to
2083 /run, and have been replaced by symlinks. This allows
2084 connecting to these facilities even if PrivateDevices=yes is
2085 used for a service (which makes /dev/log itself unavailable,
2086 but /run is left). This also has the benefit of ensuring
2087 that /dev only contains device nodes, directories and
2088 symlinks, and nothing else.
2090 * sd-daemon gained two new calls sd_pid_notify() and
2091 sd_pid_notifyf(). They are similar to sd_notify() and
2092 sd_notifyf(), but allow overriding of the source PID of
2093 notification messages if permissions permit this. This is
2094 useful to send notify messages on behalf of a different
2095 process (for example, the parent process). The
2096 systemd-notify tool has been updated to make use of this
2097 when sending messages (so that notification messages now
2098 originate from the shell script invoking systemd-notify and
2099 not the systemd-notify process itself. This should minimize
2100 a race where systemd fails to associate notification
2101 messages to services when the originating process already
2104 * A new "on-abnormal" setting for Restart= has been added. If
2105 set, it will result in automatic restarts on all "abnormal"
2106 reasons for a process to exit, which includes unclean
2107 signals, core dumps, timeouts and watchdog timeouts, but
2108 does not include clean and unclean exit codes or clean
2109 signals. Restart=on-abnormal is an alternative for
2110 Restart=on-failure for services that shall be able to
2111 terminate and avoid restarts on certain errors, by
2112 indicating so with an unclean exit code. Restart=on-failure
2113 or Restart=on-abnormal is now the recommended setting for
2114 all long-running services.
2116 * If the InaccessibleDirectories= service setting points to a
2117 mount point (or if there are any submounts contained within
2118 it), it is now attempted to completely unmount it, to make
2119 the file systems truly unavailable for the respective
2122 * The ReadOnlyDirectories= service setting and
2123 systemd-nspawn's --read-only parameter are now recursively
2124 applied to all submounts, too.
2126 * Mount units may now be created transiently via the bus APIs.
2128 * The support for SysV and LSB init scripts has been removed
2129 from the systemd daemon itself. Instead, it is now
2130 implemented as a generator that creates native systemd units
2131 from these scripts when needed. This enables us to remove a
2132 substantial amount of legacy code from PID 1, following the
2133 fact that many distributions only ship a very small number
2134 of LSB/SysV init scripts nowadays.
2136 * Privileged Xen (dom0) domains are not considered
2137 virtualization anymore by the virtualization detection
2138 logic. After all, they generally have unrestricted access to
2139 the hardware and usually are used to manage the unprivileged
2142 * systemd-tmpfiles gained a new "C" line type, for copying
2143 files or entire directories.
2145 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
2146 lines. So far, they have been non-globbing versions of the
2147 latter, and have thus been redundant. In future, it is
2148 recommended to only use "z". "m" has hence been removed
2149 from the documentation, even though it stays supported.
2151 * A tmpfiles snippet to recreate the most basic structure in
2152 /var has been added. This is enough to create the /var/run →
2153 /run symlink and create a couple of structural
2154 directories. This allows systems to boot up with an empty or
2155 volatile /var. Of course, while with this change, the core OS
2156 now is capable with dealing with a volatile /var, not all
2157 user services are ready for it. However, we hope that sooner
2158 or later, many service daemons will be changed upstream so
2159 that they are able to automatically create their necessary
2160 directories in /var at boot, should they be missing. This is
2161 the first step to allow state-less systems that only require
2162 the vendor image for /usr to boot.
2164 * systemd-nspawn has gained a new --tmpfs= switch to mount an
2165 empty tmpfs instance to a specific directory. This is
2166 particularly useful for making use of the automatic
2167 reconstruction of /var (see above), by passing --tmpfs=/var.
2169 * Access modes specified in tmpfiles snippets may now be
2170 prefixed with "~", which indicates that they shall be masked
2171 by whether the existing file or directory is currently
2172 writable, readable or executable at all. Also, if specified,
2173 the sgid/suid/sticky bits will be masked for all
2176 * A new passive target unit "network-pre.target" has been
2177 added which is useful for services that shall run before any
2178 network is configured, for example firewall scripts.
2180 * The "floppy" group that previously owned the /dev/fd*
2181 devices is no longer used. The "disk" group is now used
2182 instead. Distributions should probably deprecate usage of
2185 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
2186 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
2187 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
2188 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
2189 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
2190 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
2193 -- Berlin, 2014-06-11
2197 * A new "systemd-timesyncd" daemon has been added for
2198 synchronizing the system clock across the network. It
2199 implements an SNTP client. In contrast to NTP
2200 implementations such as chrony or the NTP reference server,
2201 this only implements a client side, and does not bother with
2202 the full NTP complexity, focusing only on querying time from
2203 one remote server and synchronizing the local clock to
2204 it. Unless you intend to serve NTP to networked clients or
2205 want to connect to local hardware clocks, this simple NTP
2206 client should be more than appropriate for most
2207 installations. The daemon runs with minimal privileges, and
2208 has been hooked up with networkd to only operate when
2209 network connectivity is available. The daemon saves the
2210 current clock to disk every time a new NTP sync has been
2211 acquired, and uses this to possibly correct the system clock
2212 early at bootup, in order to accommodate for systems that
2213 lack an RTC such as the Raspberry Pi and embedded devices,
2214 and to make sure that time monotonically progresses on these
2215 systems, even if it is not always correct. To make use of
2216 this daemon, a new system user and group "systemd-timesync"
2217 needs to be created on installation of systemd.
2219 * The queue "seqnum" interface of libudev has been disabled, as
2220 it was generally incompatible with device namespacing as
2221 sequence numbers of devices go "missing" if the devices are
2222 part of a different namespace.
2224 * "systemctl list-timers" and "systemctl list-sockets" gained
2225 a --recursive switch for showing units of these types also
2226 for all local containers, similar in style to the already
2227 supported --recursive switch for "systemctl list-units".
2229 * A new RebootArgument= setting has been added for service
2230 units, which may be used to specify a kernel reboot argument
2231 to use when triggering reboots with StartLimitAction=.
2233 * A new FailureAction= setting has been added for service
2234 units which may be used to specify an operation to trigger
2235 when a service fails. This works similarly to
2236 StartLimitAction=, but unlike it, controls what is done
2237 immediately rather than only after several attempts to
2238 restart the service in question.
2240 * hostnamed got updated to also expose the kernel name,
2241 release, and version on the bus. This is useful for
2242 executing commands like hostnamectl with the -H switch.
2243 systemd-analyze makes use of this to properly display
2244 details when running non-locally.
2246 * The bootchart tool can now show cgroup information in the
2247 graphs it generates.
2249 * The CFS CPU quota cgroup attribute is now exposed for
2250 services. The new CPUQuota= switch has been added for this
2251 which takes a percentage value. Setting this will have the
2252 result that a service may never get more CPU time than the
2253 specified percentage, even if the machine is otherwise idle.
2255 * systemd-networkd learned IPIP and SIT tunnel support.
2257 * LSB init scripts exposing a dependency on $network will now
2258 get a dependency on network-online.target rather than simply
2259 network.target. This should bring LSB handling closer to
2260 what it was on SysV systems.
2262 * A new fsck.repair= kernel option has been added to control
2263 how fsck shall deal with unclean file systems at boot.
2265 * The (.ini) configuration file parser will now silently
2266 ignore sections whose name begins with "X-". This may be
2267 used to maintain application-specific extension sections in unit
2270 * machined gained a new API to query the IP addresses of
2271 registered containers. "machinectl status" has been updated
2272 to show these addresses in its output.
2274 * A new call sd_uid_get_display() has been added to the
2275 sd-login APIs for querying the "primary" session of a
2276 user. The "primary" session of the user is elected from the
2277 user's sessions and generally a graphical session is
2278 preferred over a text one.
2280 * A minimal systemd-resolved daemon has been added. It
2281 currently simply acts as a companion to systemd-networkd and
2282 manages resolv.conf based on per-interface DNS
2283 configuration, possibly supplied via DHCP. In the long run
2284 we hope to extend this into a local DNSSEC enabled DNS and
2287 * The systemd-networkd-wait-online tool is now enabled by
2288 default. It will delay network-online.target until a network
2289 connection has been configured. The tool primarily integrates
2290 with networkd, but will also make a best effort to make sense
2291 of network configuration performed in some other way.
2293 * Two new service options StartupCPUShares= and
2294 StartupBlockIOWeight= have been added that work similarly to
2295 CPUShares= and BlockIOWeight= however only apply during
2296 system startup. This is useful to prioritize certain services
2297 differently during bootup than during normal runtime.
2299 * hostnamed has been changed to prefer the statically
2300 configured hostname in /etc/hostname (unless set to
2301 'localhost' or empty) over any dynamic one supplied by
2302 dhcp. With this change, the rules for picking the hostname
2303 match more closely the rules of other configuration settings
2304 where the local administrator's configuration in /etc always
2305 overrides any other settings.
2307 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
2308 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
2309 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
2310 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
2311 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
2312 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
2313 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
2314 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
2315 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
2316 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
2317 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
2318 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
2319 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
2320 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
2321 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
2322 Lindskog, WaLyong Cho, Will Woods, Zbigniew
2325 -- Beijing, 2014-05-28
2329 * When restoring the screen brightness at boot, stay away from
2330 the darkest setting or from the lowest 5% of the available
2331 range, depending on which is the larger value of both. This
2332 should effectively protect the user from rebooting into a
2333 black screen, should the brightness have been set to minimum
2336 * sd-login gained a new sd_machine_get_class() call to
2337 determine the class ("vm" or "container") of a machine
2338 registered with machined.
2340 * sd-login gained new calls
2341 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
2342 to query the identity of the peer of a local AF_UNIX
2343 connection. They operate similarly to their sd_pid_get_xyz()
2346 * PID 1 will now maintain a system-wide system state engine
2347 with the states "starting", "running", "degraded",
2348 "maintenance", "stopping". These states are bound to system
2349 startup, normal runtime, runtime with at least one failed
2350 service, rescue/emergency mode and system shutdown. This
2351 state is shown in the "systemctl status" output when no unit
2352 name is passed. It is useful to determine system state, in
2353 particularly when doing so for many systems or containers at
2356 * A new command "list-machines" has been added to "systemctl"
2357 that lists all local OS containers and shows their system
2358 state (see above), if systemd runs inside of them.
2360 * systemctl gained a new "-r" switch to recursively enumerate
2361 units on all local containers, when used with the
2362 "list-unit" command (which is the default one that is
2363 executed when no parameters are specified).
2365 * The GPT automatic partition discovery logic will now honour
2366 two GPT partition flags: one may be set on a partition to
2367 cause it to be mounted read-only, and the other may be set
2368 on a partition to ignore it during automatic discovery.
2370 * Two new GPT type UUIDs have been added for automatic root
2371 partition discovery, for 32-bit and 64-bit ARM. This is not
2372 particularly useful for discovering the root directory on
2373 these architectures during bare-metal boots (since UEFI is
2374 not common there), but still very useful to allow booting of
2375 ARM disk images in nspawn with the -i option.
2377 * MAC addresses of interfaces created with nspawn's
2378 --network-interface= switch will now be generated from the
2379 machine name, and thus be stable between multiple invocations
2382 * logind will now automatically remove all IPC objects owned
2383 by a user if she or he fully logs out. This makes sure that
2384 users who are logged out cannot continue to consume IPC
2385 resources. This covers SysV memory, semaphores and message
2386 queues as well as POSIX shared memory and message
2387 queues. Traditionally, SysV and POSIX IPC had no life-cycle
2388 limits. With this functionality, that is corrected. This may
2389 be turned off by using the RemoveIPC= switch of logind.conf.
2391 * The systemd-machine-id-setup and tmpfiles tools gained a
2392 --root= switch to operate on a specific root directory,
2395 * journald can now forward logged messages to the TTYs of all
2396 logged in users ("wall"). This is the default for all
2397 emergency messages now.
2399 * A new tool systemd-journal-remote has been added to stream
2400 journal log messages across the network.
2402 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
2403 controller trees are mounted into it. Note that the
2404 directories mounted beneath it are not read-only. This is a
2405 security measure and is particularly useful because glibc
2406 actually includes a search logic to pick any tmpfs it can
2407 find to implement shm_open() if /dev/shm is not available
2408 (which it might very well be in namespaced setups).
2410 * machinectl gained a new "poweroff" command to cleanly power
2411 down a local OS container.
2413 * The PrivateDevices= unit file setting will now also drop the
2414 CAP_MKNOD capability from the capability bound set, and
2415 imply DevicePolicy=closed.
2417 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
2418 comprehensively on all long-running systemd services where
2419 this is appropriate.
2421 * systemd-udevd will now run in a disassociated mount
2422 namespace. To mount directories from udev rules, make sure to
2423 pull in mount units via SYSTEMD_WANTS properties.
2425 * The kdbus support gained support for uploading policy into
2426 the kernel. sd-bus gained support for creating "monitoring"
2427 connections that can eavesdrop into all bus communication
2428 for debugging purposes.
2430 * Timestamps may now be specified in seconds since the UNIX
2431 epoch Jan 1st, 1970 by specifying "@" followed by the value
2434 * Native tcpwrap support in systemd has been removed. tcpwrap
2435 is old code, not really maintained anymore and has serious
2436 shortcomings, and better options such as firewalls
2437 exist. For setups that require tcpwrap usage, please
2438 consider invoking your socket-activated service via tcpd,
2439 like on traditional inetd.
2441 * A new system.conf configuration option
2442 DefaultTimerAccuracySec= has been added that controls the
2443 default AccuracySec= setting of .timer units.
2445 * Timer units gained a new WakeSystem= switch. If enabled,
2446 timers configured this way will cause the system to resume
2447 from system suspend (if the system supports that, which most
2450 * Timer units gained a new Persistent= switch. If enabled,
2451 timers configured this way will save to disk when they have
2452 been last triggered. This information is then used on next
2453 reboot to possible execute overdue timer events, that
2454 could not take place because the system was powered off.
2455 This enables simple anacron-like behaviour for timer units.
2457 * systemctl's "list-timers" will now also list the time a
2458 timer unit was last triggered in addition to the next time
2459 it will be triggered.
2461 * systemd-networkd will now assign predictable IPv4LL
2462 addresses to its local interfaces.
2464 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
2465 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
2466 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
2467 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
2468 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
2469 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
2470 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
2471 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
2474 -- Berlin, 2014-03-25
2478 * A new unit file setting RestrictAddressFamilies= has been
2479 added to restrict which socket address families unit
2480 processes gain access to. This takes address family names
2481 like "AF_INET" or "AF_UNIX", and is useful to minimize the
2482 attack surface of services via exotic protocol stacks. This
2483 is built on seccomp system call filters.
2485 * Two new unit file settings RuntimeDirectory= and
2486 RuntimeDirectoryMode= have been added that may be used to
2487 manage a per-daemon runtime directories below /run. This is
2488 an alternative for setting up directory permissions with
2489 tmpfiles snippets, and has the advantage that the runtime
2490 directory's lifetime is bound to the daemon runtime and that
2491 the daemon starts up with an empty directory each time. This
2492 is particularly useful when writing services that drop
2493 privileges using the User= or Group= setting.
2495 * The DeviceAllow= unit setting now supports globbing for
2496 matching against device group names.
2498 * The systemd configuration file system.conf gained new
2499 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
2500 DefaultMemoryAccounting= to globally turn on/off accounting
2501 for specific resources (cgroups) for all units. These
2502 settings may still be overridden individually in each unit
2505 * systemd-gpt-auto-generator is now able to discover /srv and
2506 root partitions in addition to /home and swap partitions. It
2507 also supports LUKS-encrypted partitions now. With this in
2508 place, automatic discovery of partitions to mount following
2509 the Discoverable Partitions Specification
2510 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
2511 is now a lot more complete. This allows booting without
2512 /etc/fstab and without root= on the kernel command line on
2513 systems prepared appropriately.
2515 * systemd-nspawn gained a new --image= switch which allows
2516 booting up disk images and Linux installations on any block
2517 device that follow the Discoverable Partitions Specification
2518 (see above). This means that installations made with
2519 appropriately updated installers may now be started and
2520 deployed using container managers, completely
2521 unmodified. (We hope that libvirt-lxc will add support for
2522 this feature soon, too.)
2524 * systemd-nspawn gained a new --network-macvlan= setting to
2525 set up a private macvlan interface for the
2526 container. Similarly, systemd-networkd gained a new
2527 Kind=macvlan setting in .netdev files.
2529 * systemd-networkd now supports configuring local addresses
2532 * A new tool systemd-network-wait-online has been added to
2533 synchronously wait for network connectivity using
2536 * The sd-bus.h bus API gained a new sd_bus_track object for
2537 tracking the life-cycle of bus peers. Note that sd-bus.h is
2538 still not a public API though (unless you specify
2539 --enable-kdbus on the configure command line, which however
2540 voids your warranty and you get no API stability guarantee).
2542 * The $XDG_RUNTIME_DIR runtime directories for each user are
2543 now individual tmpfs instances, which has the benefit of
2544 introducing separate pools for each user, with individual
2545 size limits, and thus making sure that unprivileged clients
2546 can no longer negatively impact the system or other users by
2547 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
2548 RuntimeDirectorySize= has been introduced that allows
2549 controlling the default size limit for all users. It
2550 defaults to 10% of the available physical memory. This is no
2551 replacement for quotas on tmpfs though (which the kernel
2552 still does not support), as /dev/shm and /tmp are still
2553 shared resources used by both the system and unprivileged
2556 * logind will now automatically turn off automatic suspending
2557 on laptop lid close when more than one display is
2558 connected. This was previously expected to be implemented
2559 individually in desktop environments (such as GNOME),
2560 however has been added to logind now, in order to fix a
2561 boot-time race where a desktop environment might not have
2562 been started yet and thus not been able to take an inhibitor
2563 lock at the time where logind already suspends the system
2564 due to a closed lid.
2566 * logind will now wait at least 30s after each system
2567 suspend/resume cycle, and 3min after system boot before
2568 suspending the system due to a closed laptop lid. This
2569 should give USB docking stations and similar enough time to
2570 be probed and configured after system resume and boot in
2571 order to then act as suspend blocker.
2573 * systemd-run gained a new --property= setting which allows
2574 initialization of resource control properties (and others)
2575 for the created scope or service unit. Example: "systemd-run
2576 --property=BlockIOWeight=10 updatedb" may be used to run
2577 updatedb at a low block IO scheduling weight.
2579 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
2580 now also work in --scope mode.
2582 * When systemd is compiled with kdbus support, basic support
2583 for enforced policies is now in place. (Note that enabling
2584 kdbus still voids your warranty and no API compatibility
2587 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
2588 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
2589 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
2590 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
2591 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
2592 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
2593 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
2594 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
2595 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
2596 Zbigniew Jędrzejewski-Szmek
2598 -- Berlin, 2014-03-12
2602 * systemd will now relabel /dev after loading the SMACK policy
2603 according to SMACK rules.
2605 * A new unit file option AppArmorProfile= has been added to
2606 set the AppArmor profile for the processes of a unit.
2608 * A new condition check ConditionArchitecture= has been added
2609 to conditionalize units based on the system architecture, as
2610 reported by uname()'s "machine" field.
2612 * systemd-networkd now supports matching on the system
2613 virtualization, architecture, kernel command line, host name
2616 * logind is now a lot more aggressive when suspending the
2617 machine due to a closed laptop lid. Instead of acting only
2618 on the lid close action, it will continuously watch the lid
2619 status and act on it. This is useful for laptops where the
2620 power button is on the outside of the chassis so that it can
2621 be reached without opening the lid (such as the Lenovo
2622 Yoga). On those machines, logind will now immediately
2623 re-suspend the machine if the power button has been
2624 accidentally pressed while the laptop was suspended and in a
2625 backpack or similar.
2627 * logind will now watch SW_DOCK switches and inhibit reaction
2628 to the lid switch if it is pressed. This means that logind
2629 will not suspend the machine anymore if the lid is closed
2630 and the system is docked, if the laptop supports SW_DOCK
2631 notifications via the input layer. Note that ACPI docking
2632 stations do not generate this currently. Also note that this
2633 logic is usually not fully sufficient and Desktop
2634 Environments should take a lid switch inhibitor lock when an
2635 external display is connected, as systemd will not watch
2638 * nspawn will now make use of the devices cgroup controller by
2639 default, and only permit creation of and access to the usual
2640 API device nodes like /dev/null or /dev/random, as well as
2641 access to (but not creation of) the pty devices.
2643 * We will now ship a default .network file for
2644 systemd-networkd that automatically configures DHCP for
2645 network interfaces created by nspawn's --network-veth or
2646 --network-bridge= switches.
2648 * systemd will now understand the usual M, K, G, T suffixes
2649 according to SI conventions (i.e. to the base 1000) when
2650 referring to throughput and hardware metrics. It will stay
2651 with IEC conventions (i.e. to the base 1024) for software
2652 metrics, according to what is customary according to
2653 Wikipedia. We explicitly document which base applies for
2654 each configuration option.
2656 * The DeviceAllow= setting in unit files now supports a syntax
2657 to whitelist an entire group of devices node majors at once,
2658 based on the /proc/devices listing. For example, with the
2659 string "char-pts", it is now possible to whitelist all
2660 current and future pseudo-TTYs at once.
2662 * sd-event learned a new "post" event source. Event sources of
2663 this type are triggered by the dispatching of any event
2664 source of a type that is not "post". This is useful for
2665 implementing clean-up and check event sources that are
2666 triggered by other work being done in the program.
2668 * systemd-networkd is no longer statically enabled, but uses
2669 the usual [Install] sections so that it can be
2670 enabled/disabled using systemctl. It still is enabled by
2673 * When creating a veth interface pair with systemd-nspawn, the
2674 host side will now be prefixed with "vb-" if
2675 --network-bridge= is used, and with "ve-" if --network-veth
2676 is used. This way, it is easy to distinguish these cases on
2677 the host, for example to apply different configuration to
2678 them with systemd-networkd.
2680 * The compatibility libraries for libsystemd-journal.so,
2681 libsystem-id128.so, libsystemd-login.so and
2682 libsystemd-daemon.so do not make use of IFUNC
2683 anymore. Instead, we now build libsystemd.so multiple times
2684 under these alternative names. This means that the footprint
2685 is drastically increased, but given that these are
2686 transitional compatibility libraries, this should not matter
2687 much. This change has been made necessary to support the ARM
2688 platform for these compatibility libraries, as the ARM
2689 toolchain is not really at the same level as the toolchain
2690 for other architectures like x86 and does not support
2691 IFUNC. Please make sure to use --enable-compat-libs only
2692 during a transitional period!
2694 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
2695 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
2696 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
2697 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
2698 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
2699 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
2700 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
2701 Zbigniew Jędrzejewski-Szmek
2703 -- Berlin, 2014-02-24
2707 * A new component "systemd-networkd" has been added that can
2708 be used to configure local network interfaces statically or
2709 via DHCP. It is capable of bringing up bridges, VLANs, and
2710 bonding. Currently, no hook-ups for interactive network
2711 configuration are provided. Use this for your initrd,
2712 container, embedded, or server setup if you need a simple,
2713 yet powerful, network configuration solution. This
2714 configuration subsystem is quite nifty, as it allows wildcard
2715 hotplug matching in interfaces. For example, with a single
2716 configuration snippet, you can configure that all Ethernet
2717 interfaces showing up are automatically added to a bridge,
2718 or similar. It supports link-sensing and more.
2720 * A new tool "systemd-socket-proxyd" has been added which can
2721 act as a bidirectional proxy for TCP sockets. This is
2722 useful for adding socket activation support to services that
2723 do not actually support socket activation, including virtual
2724 machines and the like.
2726 * Add a new tool to save/restore rfkill state on
2729 * Save/restore state of keyboard backlights in addition to
2730 display backlights on shutdown/boot.
2732 * udev learned a new SECLABEL{} construct to label device
2733 nodes with a specific security label when they appear. For
2734 now, only SECLABEL{selinux} is supported, but the syntax is
2735 prepared for additional security frameworks.
2737 * udev gained a new scheme to configure link-level attributes
2738 from files in /etc/systemd/network/*.link. These files can
2739 match against MAC address, device path, driver name and type,
2740 and will apply attributes like the naming policy, link speed,
2741 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
2742 address assignment policy (randomized, ...).
2744 * The configuration of network interface naming rules for
2745 "permanent interface names" has changed: a new NamePolicy=
2746 setting in the [Link] section of .link files determines the
2747 priority of possible naming schemes (onboard, slot, mac,
2748 path). The default value of this setting is determined by
2749 /usr/lib/net/links/99-default.link. Old
2750 80-net-name-slot.rules udev configuration file has been
2751 removed, so local configuration overriding this file should
2752 be adapated to override 99-default.link instead.
2754 * When the User= switch is used in a unit file, also
2755 initialize $SHELL= based on the user database entry.
2757 * systemd no longer depends on libdbus. All communication is
2758 now done with sd-bus, systemd's low-level bus library
2761 * kdbus support has been added to PID 1 itself. When kdbus is
2762 enabled, this causes PID 1 to set up the system bus and
2763 enable support for a new ".busname" unit type that
2764 encapsulates bus name activation on kdbus. It works a little
2765 bit like ".socket" units, except for bus names. A new
2766 generator has been added that converts classic dbus1 service
2767 activation files automatically into native systemd .busname
2770 * sd-bus: add a light-weight vtable implementation that allows
2771 defining objects on the bus with a simple static const
2772 vtable array of its methods, signals and properties.
2774 * systemd will not generate or install static dbus
2775 introspection data anymore to /usr/share/dbus-1/interfaces,
2776 as the precise format of these files is unclear, and
2777 nothing makes use of it.
2779 * A proxy daemon is now provided to proxy clients connecting
2780 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
2781 compatibility with classic D-Bus.
2783 * A bus driver implementation has been added that supports the
2784 classic D-Bus bus driver calls on kdbus, also for
2785 compatibility purposes.
2787 * A new API "sd-event.h" has been added that implements a
2788 minimal event loop API built around epoll. It provides a
2789 couple of features that direct epoll usage is lacking:
2790 prioritization of events, scales to large numbers of timer
2791 events, per-event timer slack (accuracy), system-wide
2792 coalescing of timer events, exit handlers, watchdog
2793 supervision support using systemd's sd_notify() API, child
2796 * A new API "sd-rntl.h" has been added that provides an API
2797 around the route netlink interface of the kernel, similar in
2798 style to "sd-bus.h".
2800 * A new API "sd-dhcp-client.h" has been added that provides a
2801 small DHCPv4 client-side implementation. This is used by
2804 * There is a new kernel command line option
2805 "systemd.restore_state=0|1". When set to "0", none of the
2806 systemd tools will restore saved runtime state to hardware
2807 devices. More specifically, the rfkill and backlight states
2810 * The FsckPassNo= compatibility option in mount/service units
2811 has been removed. The fstab generator will now add the
2812 necessary dependencies automatically, and does not require
2813 PID1's support for that anymore.
2815 * journalctl gained a new switch, --list-boots, that lists
2816 recent boots with their times and boot IDs.
2818 * The various tools like systemctl, loginctl, timedatectl,
2819 busctl, systemd-run, ... have gained a new switch "-M" to
2820 connect to a specific, local OS container (as direct
2821 connection, without requiring SSH). This works on any
2822 container that is registered with machined, such as those
2823 created by libvirt-lxc or nspawn.
2825 * systemd-run and systemd-analyze also gained support for "-H"
2826 to connect to remote hosts via SSH. This is particularly
2827 useful for systemd-run because it enables queuing of jobs
2828 onto remote systems.
2830 * machinectl gained a new command "login" to open a getty
2831 login in any local container. This works with any container
2832 that is registered with machined (such as those created by
2833 libvirt-lxc or nspawn), and which runs systemd inside.
2835 * machinectl gained a new "reboot" command that may be used to
2836 trigger a reboot on a specific container that is registered
2837 with machined. This works on any container that runs an init
2838 system of some kind.
2840 * systemctl gained a new "list-timers" command to print a nice
2841 listing of installed timer units with the times they elapse
2844 * Alternative reboot() parameters may now be specified on the
2845 "systemctl reboot" command line and are passed to the
2846 reboot() system call.
2848 * systemctl gained a new --job-mode= switch to configure the
2849 mode to queue a job with. This is a more generic version of
2850 --fail, --irreversible, and --ignore-dependencies, which are
2851 still available but not advertised anymore.
2853 * /etc/systemd/system.conf gained new settings to configure
2854 various default timeouts of units, as well as the default
2855 start limit interval and burst. These may still be overridden
2858 * PID1 will now export on the bus profile data of the security
2859 policy upload process (such as the SELinux policy upload to
2862 * journald: when forwarding logs to the console, include
2863 timestamps (following the setting in
2864 /sys/module/printk/parameters/time).
2866 * OnCalendar= in timer units now understands the special
2867 strings "yearly" and "annually". (Both are equivalent)
2869 * The accuracy of timer units is now configurable with the new
2870 AccuracySec= setting. It defaults to 1min.
2872 * A new dependency type JoinsNamespaceOf= has been added that
2873 allows running two services within the same /tmp and network
2874 namespace, if PrivateNetwork= or PrivateTmp= are used.
2876 * A new command "cat" has been added to systemctl. It outputs
2877 the original unit file of a unit, and concatenates the
2878 contents of additional "drop-in" unit file snippets, so that
2879 the full configuration is shown.
2881 * systemctl now supports globbing on the various "list-xyz"
2882 commands, like "list-units" or "list-sockets", as well as on
2883 those commands which take multiple unit names.
2885 * journalctl's --unit= switch gained support for globbing.
2887 * All systemd daemons now make use of the watchdog logic so
2888 that systemd automatically notices when they hang.
2890 * If the $container_ttys environment variable is set,
2891 getty-generator will automatically spawn a getty for each
2892 listed tty. This is useful for container managers to request
2893 login gettys to be spawned on as many ttys as needed.
2895 * %h, %s, %U specifier support is not available anymore when
2896 used in unit files for PID 1. This is because NSS calls are
2897 not safe from PID 1. They stay available for --user
2898 instances of systemd, and as special case for the root user.
2900 * loginctl gained a new "--no-legend" switch to turn off output
2903 * The "sd-login.h" API gained three new calls:
2904 sd_session_is_remote(), sd_session_get_remote_user(),
2905 sd_session_get_remote_host() to query information about
2908 * The udev hardware database now also carries vendor/product
2909 information of SDIO devices.
2911 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
2912 determine whether watchdog notifications are requested by
2915 * Socket-activated per-connection services now include a
2916 short description of the connection parameters in the
2919 * tmpfiles gained a new "--boot" option. When this is not used,
2920 only lines where the command character is not suffixed with
2921 "!" are executed. When this option is specified, those
2922 options are executed too. This partitions tmpfiles
2923 directives into those that can be safely executed at any
2924 time, and those which should be run only at boot (for
2925 example, a line that creates /run/nologin).
2927 * A new API "sd-resolve.h" has been added which provides a simple
2928 asynchronous wrapper around glibc NSS host name resolution
2929 calls, such as getaddrinfo(). In contrast to glibc's
2930 getaddrinfo_a(), it does not use signals. In contrast to most
2931 other asynchronous name resolution libraries, this one does
2932 not reimplement DNS, but reuses NSS, so that alternate
2933 host name resolution systems continue to work, such as mDNS,
2934 LDAP, etc. This API is based on libasyncns, but it has been
2935 cleaned up for inclusion in systemd.
2937 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
2938 "sd-daemon.h" are no longer found in individual libraries
2939 libsystemd-journal.so, libsystemd-login.so,
2940 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
2941 merged them into a single library, libsystemd.so, which
2942 provides all symbols. The reason for this is cyclic
2943 dependencies, as these libraries tend to use each other's
2944 symbols. So far, we have managed to workaround that by linking
2945 a copy of a good part of our code into each of these
2946 libraries again and again, which, however, makes certain
2947 things hard to do, like sharing static variables. Also, it
2948 substantially increases footprint. With this change, there
2949 is only one library for the basic APIs systemd
2950 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
2951 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
2952 library as well, however are subject to the --enable-kdbus
2953 switch (see below). Note that "sd-dhcp-client.h" is not part
2954 of this library (this is because it only consumes, never
2955 provides, services of/to other APIs). To make the transition
2956 easy from the separate libraries to the unified one, we
2957 provide the --enable-compat-libs compile-time switch which
2958 will generate stub libraries that are compatible with the
2959 old ones but redirect all calls to the new one.
2961 * All of the kdbus logic and the new APIs "sd-bus.h",
2962 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
2963 and "sd-utf8.h" are compile-time optional via the
2964 "--enable-kdbus" switch, and they are not compiled in by
2965 default. To make use of kdbus, you have to explicitly enable
2966 the switch. Note however, that neither the kernel nor the
2967 userspace API for all of this is considered stable yet. We
2968 want to maintain the freedom to still change the APIs for
2969 now. By specifying this build-time switch, you acknowledge
2970 that you are aware of the instability of the current
2973 * Also, note that while kdbus is pretty much complete,
2974 it lacks one thing: proper policy support. This means you
2975 can build a fully working system with all features; however,
2976 it will be highly insecure. Policy support will be added in
2977 one of the next releases, at the same time that we will
2978 declare the APIs stable.
2980 * When the kernel command line argument "kdbus" is specified,
2981 systemd will automatically load the kdbus.ko kernel module. At
2982 this stage of development, it is only useful for testing kdbus
2983 and should not be used in production. Note: if "--enable-kdbus"
2984 is specified, and the kdbus.ko kernel module is available, and
2985 "kdbus" is added to the kernel command line, the entire system
2986 runs with kdbus instead of dbus-daemon, with the above mentioned
2987 problem of missing the system policy enforcement. Also a future
2988 version of kdbus.ko or a newer systemd will not be compatible with
2989 each other, and will unlikely be able to boot the machine if only
2990 one of them is updated.
2992 * systemctl gained a new "import-environment" command which
2993 uploads the caller's environment (or parts thereof) into the
2994 service manager so that it is inherited by services started
2995 by the manager. This is useful to upload variables like
2996 $DISPLAY into the user service manager.
2998 * A new PrivateDevices= switch has been added to service units
2999 which allows running a service with a namespaced /dev
3000 directory that does not contain any device nodes for
3001 physical devices. More specifically, it only includes devices
3002 such as /dev/null, /dev/urandom, and /dev/zero which are API
3005 * logind has been extended to support behaviour like VT
3006 switching on seats that do not support a VT. This makes
3007 multi-session available on seats that are not the first seat
3008 (seat0), and on systems where kernel support for VTs has
3009 been disabled at compile-time.
3011 * If a process holds a delay lock for system sleep or shutdown
3012 and fails to release it in time, we will now log its
3013 identity. This makes it easier to identify processes that
3014 cause slow suspends or power-offs.
3016 * When parsing /etc/crypttab, support for a new key-slot=
3017 option as supported by Debian is added. It allows indicating
3018 which LUKS slot to use on disk, speeding up key loading.
3020 * The sd_journald_sendv() API call has been checked and
3021 officially declared to be async-signal-safe so that it may
3022 be invoked from signal handlers for logging purposes.
3024 * Boot-time status output is now enabled automatically after a
3025 short timeout if boot does not progress, in order to give
3026 the user an indication what she or he is waiting for.
3028 * The boot-time output has been improved to show how much time
3029 remains until jobs expire.
3031 * The KillMode= switch in service units gained a new possible
3032 value "mixed". If set, and the unit is shut down, then the
3033 initial SIGTERM signal is sent only to the main daemon
3034 process, while the following SIGKILL signal is sent to
3035 all remaining processes of the service.
3037 * When a scope unit is registered, a new property "Controller"
3038 may be set. If set to a valid bus name, systemd will send a
3039 RequestStop() signal to this name when it would like to shut
3040 down the scope. This may be used to hook manager logic into
3041 the shutdown logic of scope units. Also, scope units may now
3042 be put in a special "abandoned" state, in which case the
3043 manager process which created them takes no further
3044 responsibilities for it.
3046 * When reading unit files, systemd will now verify
3047 the access mode of these files, and warn about certain
3048 suspicious combinations. This has been added to make it
3049 easier to track down packaging bugs where unit files are
3050 marked executable or world-writable.
3052 * systemd-nspawn gained a new "--setenv=" switch to set
3053 container-wide environment variables. The similar option in
3054 systemd-activate was renamed from "--environment=" to
3055 "--setenv=" for consistency.
3057 * systemd-nspawn has been updated to create a new kdbus domain
3058 for each container that is invoked, thus allowing each
3059 container to have its own set of system and user buses,
3060 independent of the host.
3062 * systemd-nspawn gained a new --drop-capability= switch to run
3063 the container with less capabilities than the default. Both
3064 --drop-capability= and --capability= now take the special
3065 string "all" for dropping or keeping all capabilities.
3067 * systemd-nspawn gained new switches for executing containers
3068 with specific SELinux labels set.
3070 * systemd-nspawn gained a new --quiet switch to not generate
3071 any additional output but the container's own console
3074 * systemd-nspawn gained a new --share-system switch to run a
3075 container without PID namespacing enabled.
3077 * systemd-nspawn gained a new --register= switch to control
3078 whether the container is registered with systemd-machined or
3079 not. This is useful for containers that do not run full
3080 OS images, but only specific apps.
3082 * systemd-nspawn gained a new --keep-unit which may be used
3083 when invoked as the only program from a service unit, and
3084 results in registration of the unit service itself in
3085 systemd-machined, instead of a newly opened scope unit.
3087 * systemd-nspawn gained a new --network-interface= switch for
3088 moving arbitrary interfaces to the container. The new
3089 --network-veth switch creates a virtual Ethernet connection
3090 between host and container. The new --network-bridge=
3091 switch then allows assigning the host side of this virtual
3092 Ethernet connection to a bridge device.
3094 * systemd-nspawn gained a new --personality= switch for
3095 setting the kernel personality for the container. This is
3096 useful when running a 32-bit container on a 64-bit host. A
3097 similar option Personality= is now also available for service
3100 * logind will now also track a "Desktop" identifier for each
3101 session which encodes the desktop environment of it. This is
3102 useful for desktop environments that want to identify
3103 multiple running sessions of itself easily.
3105 * A new SELinuxContext= setting for service units has been
3106 added that allows setting a specific SELinux execution
3107 context for a service.
3109 * Most systemd client tools will now honour $SYSTEMD_LESS for
3110 settings of the "less" pager. By default, these tools will
3111 override $LESS to allow certain operations to work, such as
3112 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
3113 influence this logic.
3115 * systemd's "seccomp" hook-up has been changed to make use of
3116 the libseccomp library instead of using its own
3117 implementation. This has benefits for portability among
3120 * For usage together with SystemCallFilter=, a new
3121 SystemCallErrorNumber= setting has been introduced that
3122 allows configuration of a system error number to be returned
3123 on filtered system calls, instead of immediately killing the
3124 process. Also, SystemCallArchitectures= has been added to
3125 limit access to system calls of a particular architecture
3126 (in order to turn off support for unused secondary
3127 architectures). There is also a global
3128 SystemCallArchitectures= setting in system.conf now to turn
3129 off support for non-native system calls system-wide.
3131 * systemd requires a kernel with a working name_to_handle_at(),
3132 please see the kernel config requirements in the README file.
3134 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
3135 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
3136 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
3137 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
3138 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
3139 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
3140 Elia Pinto, Florian Weimer, George McCollister, Goffredo
3141 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
3142 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
3143 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
3144 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
3145 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
3146 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
3147 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
3148 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
3149 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
3150 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
3151 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
3152 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
3153 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
3154 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
3155 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
3156 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
3157 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
3159 -- Berlin, 2014-02-20
3163 * logind has gained support for facilitating privileged input
3164 and drm device access for unprivileged clients. This work is
3165 useful to allow Wayland display servers (and similar
3166 programs, such as kmscon) to run under the user's ID and
3167 access input and drm devices which are normally
3168 protected. When this is used (and the kernel is new enough)
3169 logind will "mute" IO on the file descriptors passed to
3170 Wayland as long as it is in the background and "unmute" it
3171 if it returns into the foreground. This allows secure
3172 session switching without allowing background sessions to
3173 eavesdrop on input and display data. This also introduces
3174 session switching support if VT support is turned off in the
3175 kernel, and on seats that are not seat0.
3177 * A new kernel command line option luks.options= is understood
3178 now which allows specifying LUKS options for usage for LUKS
3179 encrypted partitions specified with luks.uuid=.
3181 * tmpfiles.d(5) snippets may now use specifier expansion in
3182 path names. More specifically %m, %b, %H, %v, are now
3183 replaced by the local machine id, boot id, hostname, and
3184 kernel version number.
3186 * A new tmpfiles.d(5) command "m" has been introduced which
3187 may be used to change the owner/group/access mode of a file
3188 or directory if it exists, but do nothing if it does not.
3190 * This release removes high-level support for the
3191 MemorySoftLimit= cgroup setting. The underlying kernel
3192 cgroup attribute memory.soft_limit= is currently badly
3193 designed and likely to be removed from the kernel API in its
3194 current form, hence we should not expose it for now.
3196 * The memory.use_hierarchy cgroup attribute is now enabled for
3197 all cgroups systemd creates in the memory cgroup
3198 hierarchy. This option is likely to be come the built-in
3199 default in the kernel anyway, and the non-hierarchical mode
3200 never made much sense in the intrinsically hierarchical
3203 * A new field _SYSTEMD_SLICE= is logged along with all journal
3204 messages containing the slice a message was generated
3205 from. This is useful to allow easy per-customer filtering of
3206 logs among other things.
3208 * systemd-journald will no longer adjust the group of journal
3209 files it creates to the "systemd-journal" group. Instead we
3210 rely on the journal directory to be owned by the
3211 "systemd-journal" group, and its setgid bit set, so that the
3212 kernel file system layer will automatically enforce that
3213 journal files inherit this group assignment. The reason for
3214 this change is that we cannot allow NSS look-ups from
3215 journald which would be necessary to resolve
3216 "systemd-journal" to a numeric GID, because this might
3217 create deadlocks if NSS involves synchronous queries to
3218 other daemons (such as nscd, or sssd) which in turn are
3219 logging clients of journald and might block on it, which
3220 would then dead lock. A tmpfiles.d(5) snippet included in
3221 systemd will make sure the setgid bit and group are
3222 properly set on the journal directory if it exists on every
3223 boot. However, we recommend adjusting it manually after
3224 upgrades too (or from RPM scriptlets), so that the change is
3225 not delayed until next reboot.
3227 * Backlight and random seed files in /var/lib/ have moved into
3228 the /var/lib/systemd/ directory, in order to centralize all
3229 systemd generated files in one directory.
3231 * Boot time performance measurements (as displayed by
3232 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
3233 performance information if that's available to determine how
3234 much time BIOS and boot loader initialization required. With
3235 a sufficiently new BIOS you hence no longer need to boot
3236 with Gummiboot to get access to such information.
3238 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
3239 Cristian Rodríguez, Dave Reisner, David Herrmann, David
3240 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
3241 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
3242 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
3243 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
3244 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3246 -- Berlin, 2013-10-02
3250 * The Restart= option for services now understands a new
3251 on-watchdog setting, which will restart the service
3252 automatically if the service stops sending out watchdog keep
3253 alive messages (as configured with WatchdogSec=).
3255 * The getty generator (which is responsible for bringing up a
3256 getty on configured serial consoles) will no longer only
3257 start a getty on the primary kernel console but on all
3258 others, too. This makes the order in which console= is
3259 specified on the kernel command line less important.
3261 * libsystemd-logind gained a new sd_session_get_vt() call to
3262 retrieve the VT number of a session.
3264 * If the option "tries=0" is set for an entry of /etc/crypttab
3265 its passphrase is queried indefinitely instead of any
3266 maximum number of tries.
3268 * If a service with a configure PID file terminates its PID
3269 file will now be removed automatically if it still exists
3270 afterwards. This should put an end to stale PID files.
3272 * systemd-run will now also take relative binary path names
3273 for execution and no longer insists on absolute paths.
3275 * InaccessibleDirectories= and ReadOnlyDirectories= now take
3276 paths that are optionally prefixed with "-" to indicate that
3277 it should not be considered a failure if they do not exist.
3279 * journalctl -o (and similar commands) now understands a new
3280 output mode "short-precise", it is similar to "short" but
3281 shows timestamps with usec accuracy.
3283 * The option "discard" (as known from Debian) is now
3284 synonymous to "allow-discards" in /etc/crypttab. In fact,
3285 "discard" is preferred now (since it is easier to remember
3288 * Some licensing clean-ups were made, so that more code is now
3289 LGPL-2.1 licensed than before.
3291 * A minimal tool to save/restore the display backlight
3292 brightness across reboots has been added. It will store the
3293 backlight setting as late as possible at shutdown, and
3294 restore it as early as possible during reboot.
3296 * A logic to automatically discover and enable home and swap
3297 partitions on GPT disks has been added. With this in place
3298 /etc/fstab becomes optional for many setups as systemd can
3299 discover certain partitions located on the root disk
3300 automatically. Home partitions are recognized under their
3301 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
3302 partitions are recognized under their GPT type ID
3303 0657fd6da4ab43c484e50933c84b4f4f.
3305 * systemd will no longer pass any environment from the kernel
3306 or initrd to system services. If you want to set an
3307 environment for all services, do so via the kernel command
3308 line systemd.setenv= assignment.
3310 * The systemd-sysctl tool no longer natively reads the file
3311 /etc/sysctl.conf. If desired, the file should be symlinked
3312 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
3313 legacy support by a symlink rather than built-in code, it
3314 also makes the otherwise hidden order of application of the
3315 different files visible. (Note that this partly reverts to a
3316 pre-198 application order of sysctl knobs!)
3318 * The "systemctl set-log-level" and "systemctl dump" commands
3319 have been moved to systemd-analyze.
3321 * systemd-run learned the new --remain-after-exit switch,
3322 which causes the scope unit not to be cleaned up
3323 automatically after the process terminated.
3325 * tmpfiles learned a new --exclude-prefix= switch to exclude
3326 certain paths from operation.
3328 * journald will now automatically flush all messages to disk
3329 as soon as a message at the log level CRIT, ALERT or EMERG
3332 Contributions from: Andrew Cook, Brandon Philips, Christian
3333 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
3334 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
3335 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
3336 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
3337 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
3338 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
3339 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
3340 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
3341 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
3342 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
3343 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
3344 William Giokas, Zbigniew Jędrzejewski-Szmek
3346 -- Berlin, 2013-09-13
3350 * The documentation has been updated to cover the various new
3351 concepts introduced with 205.
3353 * Unit files now understand the new %v specifier which
3354 resolves to the kernel version string as returned by "uname
3357 * systemctl now supports filtering the unit list output by
3358 load state, active state and sub state, using the new
3361 * "systemctl status" will now show the results of the
3362 condition checks (like ConditionPathExists= and similar) of
3363 the last start attempts of the unit. They are also logged to
3366 * "journalctl -b" may now be used to look for boot output of a
3367 specific boot. Try "journalctl -b -1" for the previous boot,
3368 but the syntax is substantially more powerful.
3370 * "journalctl --show-cursor" has been added which prints the
3371 cursor string the last shown log line. This may then be used
3372 with the new "journalctl --after-cursor=" switch to continue
3373 browsing logs from that point on.
3375 * "journalctl --force" may now be used to force regeneration
3378 * Creation of "dead" device nodes has been moved from udev
3379 into kmod and tmpfiles. Previously, udev would read the kmod
3380 databases to pre-generate dead device nodes based on meta
3381 information contained in kernel modules, so that these would
3382 be auto-loaded on access rather then at boot. As this
3383 does not really have much to do with the exposing actual
3384 kernel devices to userspace this has always been slightly
3385 alien in the udev codebase. Following the new scheme kmod
3386 will now generate a runtime snippet for tmpfiles from the
3387 module meta information and it now is tmpfiles' job to the
3388 create the nodes. This also allows overriding access and
3389 other parameters for the nodes using the usual tmpfiles
3390 facilities. As side effect this allows us to remove the
3391 CAP_SYS_MKNOD capability bit from udevd entirely.
3393 * logind's device ACLs may now be applied to these "dead"
3394 devices nodes too, thus finally allowing managed access to
3395 devices such as /dev/snd/sequencer whithout loading the
3396 backing module right-away.
3398 * A new RPM macro has been added that may be used to apply
3399 tmpfiles configuration during package installation.
3401 * systemd-detect-virt and ConditionVirtualization= now can
3402 detect User-Mode-Linux machines (UML).
3404 * journald will now implicitly log the effective capabilities
3405 set of processes in the message metadata.
3407 * systemd-cryptsetup has gained support for TrueCrypt volumes.
3409 * The initrd interface has been simplified (more specifically,
3410 support for passing performance data via environment
3411 variables and fsck results via files in /run has been
3412 removed). These features were non-essential, and are
3413 nowadays available in a much nicer way by having systemd in
3414 the initrd serialize its state and have the hosts systemd
3415 deserialize it again.
3417 * The udev "keymap" data files and tools to apply keyboard
3418 specific mappings of scan to key codes, and force-release
3419 scan code lists have been entirely replaced by a udev
3420 "keyboard" builtin and a hwdb data file.
3422 * systemd will now honour the kernel's "quiet" command line
3423 argument also during late shutdown, resulting in a
3424 completely silent shutdown when used.
3426 * There's now an option to control the SO_REUSEPORT socket
3427 option in .socket units.
3429 * Instance units will now automatically get a per-template
3430 subslice of system.slice unless something else is explicitly
3431 configured. For example, instances of sshd@.service will now
3432 implicitly be placed in system-sshd.slice rather than
3433 system.slice as before.
3435 * Test coverage support may now be enabled at build time.
3437 Contributions from: Dave Reisner, Frederic Crozat, Harald
3438 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
3439 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
3440 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
3441 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
3442 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
3443 Giokas, Zbigniew Jędrzejewski-Szmek
3445 -- Berlin, 2013-07-23
3449 * Two new unit types have been introduced:
3451 Scope units are very similar to service units, however, are
3452 created out of pre-existing processes -- instead of PID 1
3453 forking off the processes. By using scope units it is
3454 possible for system services and applications to group their
3455 own child processes (worker processes) in a powerful way
3456 which then maybe used to organize them, or kill them
3457 together, or apply resource limits on them.
3459 Slice units may be used to partition system resources in an
3460 hierarchical fashion and then assign other units to them. By
3461 default there are now three slices: system.slice (for all
3462 system services), user.slice (for all user sessions),
3463 machine.slice (for VMs and containers).
3465 Slices and scopes have been introduced primarily in
3466 context of the work to move cgroup handling to a
3467 single-writer scheme, where only PID 1
3468 creates/removes/manages cgroups.
3470 * There's a new concept of "transient" units. In contrast to
3471 normal units these units are created via an API at runtime,
3472 not from configuration from disk. More specifically this
3473 means it is now possible to run arbitrary programs as
3474 independent services, with all execution parameters passed
3475 in via bus APIs rather than read from disk. Transient units
3476 make systemd substantially more dynamic then it ever was,
3477 and useful as a general batch manager.
3479 * logind has been updated to make use of scope and slice units
3480 for managing user sessions. As a user logs in he will get
3481 his own private slice unit, to which all sessions are added
3482 as scope units. We also added support for automatically
3483 adding an instance of user@.service for the user into the
3484 slice. Effectively logind will no longer create cgroup
3485 hierarchies on its own now, it will defer entirely to PID 1
3486 for this by means of scope, service and slice units. Since
3487 user sessions this way become entities managed by PID 1
3488 the output of "systemctl" is now a lot more comprehensive.
3490 * A new mini-daemon "systemd-machined" has been added which
3491 may be used by virtualization managers to register local
3492 VMs/containers. nspawn has been updated accordingly, and
3493 libvirt will be updated shortly. machined will collect a bit
3494 of meta information about the VMs/containers, and assign
3495 them their own scope unit (see above). The collected
3496 meta-data is then made available via the "machinectl" tool,
3497 and exposed in "ps" and similar tools. machined/machinectl
3498 is compile-time optional.
3500 * As discussed earlier, the low-level cgroup configuration
3501 options ControlGroup=, ControlGroupModify=,
3502 ControlGroupPersistent=, ControlGroupAttribute= have been
3503 removed. Please use high-level attribute settings instead as
3504 well as slice units.
3506 * A new bus call SetUnitProperties() has been added to alter
3507 various runtime parameters of a unit. This is primarily
3508 useful to alter cgroup parameters dynamically in a nice way,
3509 but will be extended later on to make more properties
3510 modifiable at runtime. systemctl gained a new set-properties
3511 command that wraps this call.
3513 * A new tool "systemd-run" has been added which can be used to
3514 run arbitrary command lines as transient services or scopes,
3515 while configuring a number of settings via the command
3516 line. This tool is currently very basic, however already
3517 very useful. We plan to extend this tool to even allow
3518 queuing of execution jobs with time triggers from the
3519 command line, similar in fashion to "at".
3521 * nspawn will now inform the user explicitly that kernels with
3522 audit enabled break containers, and suggest the user to turn
3525 * Support for detecting the IMA and AppArmor security
3526 frameworks with ConditionSecurity= has been added.
3528 * journalctl gained a new "-k" switch for showing only kernel
3529 messages, mimicking dmesg output; in addition to "--user"
3530 and "--system" switches for showing only user's own logs
3533 * systemd-delta can now show information about drop-in
3534 snippets extending unit files.
3536 * libsystemd-bus has been substantially updated but is still
3537 not available as public API.
3539 * systemd will now look for the "debug" argument on the kernel
3540 command line and enable debug logging, similar to what
3541 "systemd.log_level=debug" already did before.
3543 * "systemctl set-default", "systemctl get-default" has been
3544 added to configure the default.target symlink, which
3545 controls what to boot into by default.
3547 * "systemctl set-log-level" has been added as a convenient
3548 way to raise and lower systemd logging threshold.
3550 * "systemd-analyze plot" will now show the time the various
3551 generators needed for execution, as well as information
3552 about the unit file loading.
3554 * libsystemd-journal gained a new sd_journal_open_files() call
3555 for opening specific journal files. journactl also gained a
3556 new switch to expose this new functionality. Previously we
3557 only supported opening all files from a directory, or all
3558 files from the system, as opening individual files only is
3559 racy due to journal file rotation.
3561 * systemd gained the new DefaultEnvironment= setting in
3562 /etc/systemd/system.conf to set environment variables for
3565 * If a privileged process logs a journal message with the
3566 OBJECT_PID= field set, then journald will automatically
3567 augment this with additional OBJECT_UID=, OBJECT_GID=,
3568 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
3569 system services want to log events about specific client
3570 processes. journactl/systemctl has been updated to make use
3571 of this information if all log messages regarding a specific
3574 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
3575 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
3576 Reisner, David Coppa, David King, David Strauss, Eelco
3577 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
3578 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
3579 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
3580 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
3581 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
3582 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
3583 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
3584 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
3585 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
3586 Łukasz Stelmach, 장동준
3590 * The Python bindings gained some minimal support for the APIs
3591 exposed by libsystemd-logind.
3593 * ConditionSecurity= gained support for detecting SMACK. Since
3594 this condition already supports SELinux and AppArmor we only
3595 miss IMA for this. Patches welcome!
3597 Contributions from: Karol Lewandowski, Lennart Poettering,
3598 Zbigniew Jędrzejewski-Szmek
3602 * systemd-nspawn will now create /etc/resolv.conf if
3603 necessary, before bind-mounting the host's file onto it.
3605 * systemd-nspawn will now store meta information about a
3606 container on the container's cgroup as extended attribute
3607 fields, including the root directory.
3609 * The cgroup hierarchy has been reworked in many ways. All
3610 objects any of the components systemd creates in the cgroup
3611 tree are now suffixed. More specifically, user sessions are
3612 now placed in cgroups suffixed with ".session", users in
3613 cgroups suffixed with ".user", and nspawn containers in
3614 cgroups suffixed with ".nspawn". Furthermore, all cgroup
3615 names are now escaped in a simple scheme to avoid collision
3616 of userspace object names with kernel filenames. This work
3617 is preparation for making these objects relocatable in the
3618 cgroup tree, in order to allow easy resource partitioning of
3619 these objects without causing naming conflicts.
3621 * systemctl list-dependencies gained the new switches
3622 --plain, --reverse, --after and --before.
3624 * systemd-inhibit now shows the process name of processes that
3625 have taken an inhibitor lock.
3627 * nss-myhostname will now also resolve "localhost"
3628 implicitly. This makes /etc/hosts an optional file and
3629 nicely handles that on IPv6 ::1 maps to both "localhost" and
3632 * libsystemd-logind.so gained a new call
3633 sd_get_machine_names() to enumerate running containers and
3634 VMs (currently only supported by very new libvirt and
3635 nspawn). sd_login_monitor can now be used to watch
3636 VMs/containers coming and going.
3638 * .include is not allowed recursively anymore, and only in
3639 unit files. Usually it is better to use drop-in snippets in
3640 .d/*.conf anyway, as introduced with systemd 198.
3642 * systemd-analyze gained a new "critical-chain" command that
3643 determines the slowest chain of units run during system
3644 boot-up. It is very useful for tracking down where
3645 optimizing boot time is the most beneficial.
3647 * systemd will no longer allow manipulating service paths in
3648 the name=systemd:/system cgroup tree using ControlGroup= in
3649 units. (But is still fine with it in all other dirs.)
3651 * There's a new systemd-nspawn@.service service file that may
3652 be used to easily run nspawn containers as system
3653 services. With the container's root directory in
3654 /var/lib/container/foobar it is now sufficient to run
3655 "systemctl start systemd-nspawn@foobar.service" to boot it.
3657 * systemd-cgls gained a new parameter "--machine" to list only
3658 the processes within a certain container.
3660 * ConditionSecurity= now can check for "apparmor". We still
3661 are lacking checks for SMACK and IMA for this condition
3662 check though. Patches welcome!
3664 * A new configuration file /etc/systemd/sleep.conf has been
3665 added that may be used to configure which kernel operation
3666 systemd is supposed to execute when "suspend", "hibernate"
3667 or "hybrid-sleep" is requested. This makes the new kernel
3668 "freeze" state accessible to the user.
3670 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
3671 the passed argument if applicable.
3673 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
3674 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
3675 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
3676 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
3677 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
3678 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
3679 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
3684 * The output of 'systemctl list-jobs' got some polishing. The
3685 '--type=' argument may now be passed more than once. A new
3686 command 'systemctl list-sockets' has been added which shows
3687 a list of kernel sockets systemd is listening on with the
3688 socket units they belong to, plus the units these socket
3691 * The experimental libsystemd-bus library got substantial
3692 updates to work in conjunction with the (also experimental)
3693 kdbus kernel project. It works well enough to exchange
3694 messages with some sophistication. Note that kdbus is not
3695 ready yet, and the library is mostly an elaborate test case
3696 for now, and not installable.
3698 * systemd gained a new unit 'systemd-static-nodes.service'
3699 that generates static device nodes earlier during boot, and
3700 can run in conjunction with udev.
3702 * libsystemd-login gained a new call sd_pid_get_user_unit()
3703 to retrieve the user systemd unit a process is running
3704 in. This is useful for systems where systemd is used as
3707 * systemd-nspawn now places all containers in the new /machine
3708 top-level cgroup directory in the name=systemd
3709 hierarchy. libvirt will soon do the same, so that we get a
3710 uniform separation of /system, /user and /machine for system
3711 services, user processes and containers/virtual
3712 machines. This new cgroup hierarchy is also useful to stick
3713 stable names to specific container instances, which can be
3714 recognized later this way (this name may be controlled
3715 via systemd-nspawn's new -M switch). libsystemd-login also
3716 gained a new call sd_pid_get_machine_name() to retrieve the
3717 name of the container/VM a specific process belongs to.
3719 * bootchart can now store its data in the journal.
3721 * libsystemd-journal gained a new call
3722 sd_journal_add_conjunction() for AND expressions to the
3723 matching logic. This can be used to express more complex
3724 logical expressions.
3726 * journactl can now take multiple --unit= and --user-unit=
3729 * The cryptsetup logic now understands the "luks.key=" kernel
3730 command line switch for specifying a file to read the
3731 decryption key from. Also, if a configured key file is not
3732 found the tool will now automatically fall back to prompting
3735 * Python systemd.journal module was updated to wrap recently
3736 added functions from libsystemd-journal. The interface was
3737 changed to bring the low level interface in s.j._Reader
3738 closer to the C API, and the high level interface in
3739 s.j.Reader was updated to wrap and convert all data about
3742 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
3743 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
3744 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
3745 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
3746 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
3747 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3751 * journalctl --update-catalog now understands a new --root=
3752 option to operate on catalogs found in a different root
3755 * During shutdown after systemd has terminated all running
3756 services a final killing loop kills all remaining left-over
3757 processes. We will now print the name of these processes
3758 when we send SIGKILL to them, since this usually indicates a
3761 * If /etc/crypttab refers to password files stored on
3762 configured mount points automatic dependencies will now be
3763 generated to ensure the specific mount is established first
3764 before the key file is attempted to be read.
3766 * 'systemctl status' will now show information about the
3767 network sockets a socket unit is listening on.
3769 * 'systemctl status' will also shown information about any
3770 drop-in configuration file for units. (Drop-In configuration
3771 files in this context are files such as
3772 /etc/systemd/systemd/foobar.service.d/*.conf)
3774 * systemd-cgtop now optionally shows summed up CPU times of
3775 cgroups. Press '%' while running cgtop to switch between
3776 percentage and absolute mode. This is useful to determine
3777 which cgroups use up the most CPU time over the entire
3778 runtime of the system. systemd-cgtop has also been updated
3779 to be 'pipeable' for processing with further shell tools.
3781 * 'hostnamectl set-hostname' will now allow setting of FQDN
3784 * The formatting and parsing of time span values has been
3785 changed. The parser now understands fractional expressions
3786 such as "5.5h". The formatter will now output fractional
3787 expressions for all time spans under 1min, i.e. "5.123456s"
3788 rather than "5s 123ms 456us". For time spans under 1s
3789 millisecond values are shown, for those under 1ms
3790 microsecond values are shown. This should greatly improve
3791 all time-related output of systemd.
3793 * libsystemd-login and libsystemd-journal gained new
3794 functions for querying the poll() events mask and poll()
3795 timeout value for integration into arbitrary event
3798 * localectl gained the ability to list available X11 keymaps
3799 (models, layouts, variants, options).
3801 * 'systemd-analyze dot' gained the ability to filter for
3802 specific units via shell-style globs, to create smaller,
3803 more useful graphs. I.e. it is now possible to create simple
3804 graphs of all the dependencies between only target units, or
3805 of all units that Avahi has dependencies with.
3807 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
3808 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
3809 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
3810 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
3811 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
3812 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
3813 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
3817 * The boot-time readahead implementation for rotating media
3818 will now read the read-ahead data in multiple passes which
3819 consist of all read requests made in equidistant time
3820 intervals. This means instead of strictly reading read-ahead
3821 data in its physical order on disk we now try to find a
3822 middle ground between physical and access time order.
3824 * /etc/os-release files gained a new BUILD_ID= field for usage
3825 on operating systems that provide continuous builds of OS
3828 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
3829 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
3830 William Douglas, Zbigniew Jędrzejewski-Szmek
3834 * systemd-python gained an API exposing libsystemd-daemon.
3836 * The SMACK setup logic gained support for uploading CIPSO
3839 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
3840 ReadOnlyDirectories= and InaccessibleDirectories= has
3841 changed. The private /tmp and /var/tmp directories are now
3842 shared by all processes of a service (which means
3843 ExecStartPre= may now leave data in /tmp that ExecStart= of
3844 the same service can still access). When a service is
3845 stopped its temporary directories are immediately deleted
3846 (normal clean-up with tmpfiles is still done in addition to
3849 * By default, systemd will now set a couple of sysctl
3850 variables in the kernel: the safe sysrq options are turned
3851 on, IP route verification is turned on, and source routing
3852 disabled. The recently added hardlink and softlink
3853 protection of the kernel is turned on. These settings should
3854 be reasonably safe, and good defaults for all new systems.
3856 * The predictable network naming logic may now be turned off
3857 with a new kernel command line switch: net.ifnames=0.
3859 * A new libsystemd-bus module has been added that implements a
3860 pretty complete D-Bus client library. For details see:
3862 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
3864 * journald will now explicitly flush the journal files to disk
3865 at the latest 5min after each write. The file will then also
3866 be marked offline until the next write. This should increase
3867 reliability in case of a crash. The synchronization delay
3868 can be configured via SyncIntervalSec= in journald.conf.
3870 * There's a new remote-fs-setup.target unit that can be used
3871 to pull in specific services when at least one remote file
3872 system is to be mounted.
3874 * There are new targets timers.target and paths.target as
3875 canonical targets to pull user timer and path units in
3876 from. This complements sockets.target with a similar
3877 purpose for socket units.
3879 * libudev gained a new call udev_device_set_attribute_value()
3880 to set sysfs attributes of a device.
3882 * The udev daemon now sets the default number of worker
3883 processes executed in parallel based on the number of available
3884 CPUs instead of the amount of available RAM. This is supposed
3885 to provide a more reliable default and limit a too aggressive
3886 paralellism for setups with 1000s of devices connected.
3888 Contributions from: Auke Kok, Colin Walters, Cristian
3889 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
3890 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
3891 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
3892 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
3893 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
3894 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
3895 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
3896 Zbigniew Jędrzejewski-Szmek
3900 * Configuration of unit files may now be extended via drop-in
3901 files without having to edit/override the unit files
3902 themselves. More specifically, if the administrator wants to
3903 change one value for a service file foobar.service he can
3904 now do so by dropping in a configuration snippet into
3905 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
3906 will load all these snippets and apply them on top of the
3907 main unit configuration file, possibly extending or
3908 overriding its settings. Using these drop-in snippets is
3909 generally nicer than the two earlier options for changing
3910 unit files locally: copying the files from
3911 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
3912 them there; or creating a new file in /etc/systemd/system/
3913 that incorporates the original one via ".include". Drop-in
3914 snippets into these .d/ directories can be placed in any
3915 directory systemd looks for units in, and the usual
3916 overriding semantics between /usr/lib, /etc and /run apply
3919 * Most unit file settings which take lists of items can now be
3920 reset by assigning the empty string to them. For example,
3921 normally, settings such as Environment=FOO=BAR append a new
3922 environment variable assignment to the environment block,
3923 each time they are used. By assigning Environment= the empty
3924 string the environment block can be reset to empty. This is
3925 particularly useful with the .d/*.conf drop-in snippets
3926 mentioned above, since this adds the ability to reset list
3927 settings from vendor unit files via these drop-ins.
3929 * systemctl gained a new "list-dependencies" command for
3930 listing the dependencies of a unit recursively.
3932 * Inhibitors are now honored and listed by "systemctl
3933 suspend", "systemctl poweroff" (and similar) too, not only
3934 GNOME. These commands will also list active sessions by
3937 * Resource limits (as exposed by the various control group
3938 controllers) can now be controlled dynamically at runtime
3939 for all units. More specifically, you can now use a command
3940 like "systemctl set-cgroup-attr foobar.service cpu.shares
3941 2000" to alter the CPU shares a specific service gets. These
3942 settings are stored persistently on disk, and thus allow the
3943 administrator to easily adjust the resource usage of
3944 services with a few simple commands. This dynamic resource
3945 management logic is also available to other programs via the
3946 bus. Almost any kernel cgroup attribute and controller is
3949 * systemd-vconsole-setup will now copy all font settings to
3950 all allocated VTs, where it previously applied them only to
3953 * libsystemd-login gained the new sd_session_get_tty() API
3956 * This release drops support for a few legacy or
3957 distribution-specific LSB facility names when parsing init
3958 scripts: $x-display-manager, $mail-transfer-agent,
3959 $mail-transport-agent, $mail-transfer-agent, $smtp,
3960 $null. Also, the mail-transfer-agent.target unit backing
3961 this has been removed. Distributions which want to retain
3962 compatibility with this should carry the burden for
3963 supporting this themselves and patch support for these back
3964 in, if they really need to. Also, the facilities $syslog and
3965 $local_fs are now ignored, since systemd does not support
3966 early-boot LSB init scripts anymore, and these facilities
3967 are implied anyway for normal services. syslog.target has
3970 * There are new bus calls on PID1's Manager object for
3971 cancelling jobs, and removing snapshot units. Previously,
3972 both calls were only available on the Job and Snapshot
3975 * systemd-journal-gatewayd gained SSL support.
3977 * The various "environment" files, such as /etc/locale.conf
3978 now support continuation lines with a backslash ("\") as
3979 last character in the line, similarly in style (but different)
3980 to how this is supported in shells.
3982 * For normal user processes the _SYSTEMD_USER_UNIT= field is
3983 now implicitly appended to every log entry logged. systemctl
3984 has been updated to filter by this field when operating on a
3985 user systemd instance.
3987 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
3988 CAP_AUDIT_CONTROL capabilities to the capabilities set for
3989 the container. This makes it easier to boot unmodified
3990 Fedora systems in a container, which however still requires
3991 audit=0 to be passed on the kernel command line. Auditing in
3992 kernel and userspace is unfortunately still too broken in
3993 context of containers, hence we recommend compiling it out
3994 of the kernel or using audit=0. Hopefully this will be fixed
3995 one day for good in the kernel.
3997 * nspawn gained the new --bind= and --bind-ro= parameters to
3998 bind mount specific directories from the host into the
4001 * nspawn will now mount its own devpts file system instance
4002 into the container, in order not to leak pty devices from
4003 the host into the container.
4005 * systemd will now read the firmware boot time performance
4006 information from the EFI variables, if the used boot loader
4007 supports this, and takes it into account for boot performance
4008 analysis via "systemd-analyze". This is currently supported
4009 only in conjunction with Gummiboot, but could be supported
4010 by other boot loaders too. For details see:
4012 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
4014 * A new generator has been added that automatically mounts the
4015 EFI System Partition (ESP) to /boot, if that directory
4016 exists, is empty, and no other file system has been
4017 configured to be mounted there.
4019 * logind will now send out PrepareForSleep(false) out
4020 unconditionally, after coming back from suspend. This may be
4021 used by applications as asynchronous notification for
4022 system resume events.
4024 * "systemctl unlock-sessions" has been added, that allows
4025 unlocking the screens of all user sessions at once, similar
4026 to how "systemctl lock-sessions" already locked all users
4027 sessions. This is backed by a new D-Bus call UnlockSessions().
4029 * "loginctl seat-status" will now show the master device of a
4030 seat. (i.e. the device of a seat that needs to be around for
4031 the seat to be considered available, usually the graphics
4034 * tmpfiles gained a new "X" line type, that allows
4035 configuration of files and directories (with wildcards) that
4036 shall be excluded from automatic cleanup ("aging").
4038 * udev default rules set the device node permissions now only
4039 at "add" events, and do not change them any longer with a
4040 later "change" event.
4042 * The log messages for lid events and power/sleep keypresses
4043 now carry a message ID.
4045 * We now have a substantially larger unit test suite, but this
4046 continues to be work in progress.
4048 * udevadm hwdb gained a new --root= parameter to change the
4049 root directory to operate relative to.
4051 * logind will now issue a background sync() request to the kernel
4052 early at shutdown, so that dirty buffers are flushed to disk early
4053 instead of at the last moment, in order to optimize shutdown
4056 * A new bootctl tool has been added that is an interface for
4057 certain boot loader operations. This is currently a preview
4058 and is likely to be extended into a small mechanism daemon
4059 like timedated, localed, hostnamed, and can be used by
4060 graphical UIs to enumerate available boot options, and
4061 request boot into firmware operations.
4063 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
4064 the rest of the package. It also has been updated to work
4065 correctly in initrds.
4067 * Policykit previously has been runtime optional, and is now
4068 also compile time optional via a configure switch.
4070 * systemd-analyze has been reimplemented in C. Also "systemctl
4071 dot" has moved into systemd-analyze.
4073 * "systemctl status" with no further parameters will now print
4074 the status of all active or failed units.
4076 * Operations such as "systemctl start" can now be executed
4077 with a new mode "--irreversible" which may be used to queue
4078 operations that cannot accidentally be reversed by a later
4079 job queuing. This is by default used to make shutdown
4080 requests more robust.
4082 * The Python API of systemd now gained a new module for
4083 reading journal files.
4085 * A new tool kernel-install has been added that can install
4086 kernel images according to the Boot Loader Specification:
4088 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
4090 * Boot time console output has been improved to provide
4091 animated boot time output for hanging jobs.
4093 * A new tool systemd-activate has been added which can be used
4094 to test socket activation with, directly from the command
4095 line. This should make it much easier to test and debug
4096 socket activation in daemons.
4098 * journalctl gained a new "--reverse" (or -r) option to show
4099 journal output in reverse order (i.e. newest line first).
4101 * journalctl gained a new "--pager-end" (or -e) option to jump
4102 to immediately jump to the end of the journal in the
4103 pager. This is only supported in conjunction with "less".
4105 * journalctl gained a new "--user-unit=" option, that works
4106 similarly to "--unit=" but filters for user units rather than
4109 * A number of unit files to ease adoption of systemd in
4110 initrds has been added. This moves some minimal logic from
4111 the various initrd implementations into systemd proper.
4113 * The journal files are now owned by a new group
4114 "systemd-journal", which exists specifically to allow access
4115 to the journal, and nothing else. Previously, we used the
4116 "adm" group for that, which however possibly covers more
4117 than just journal/log file access. This new group is now
4118 already used by systemd-journal-gatewayd to ensure this
4119 daemon gets access to the journal files and as little else
4120 as possible. Note that "make install" will also set FS ACLs
4121 up for /var/log/journal to give "adm" and "wheel" read
4122 access to it, in addition to "systemd-journal" which owns
4123 the journal files. We recommend that packaging scripts also
4124 add read access to "adm" + "wheel" to /var/log/journal, and
4125 all existing/future journal files. To normal users and
4126 administrators little changes, however packagers need to
4127 ensure to create the "systemd-journal" system group at
4128 package installation time.
4130 * The systemd-journal-gatewayd now runs as unprivileged user
4131 systemd-journal-gateway:systemd-journal-gateway. Packaging
4132 scripts need to create these system user/group at
4135 * timedated now exposes a new boolean property CanNTP that
4136 indicates whether a local NTP service is available or not.
4138 * systemd-detect-virt will now also detect xen PVs
4140 * The pstore file system is now mounted by default, if it is
4143 * In addition to the SELinux and IMA policies we will now also
4144 load SMACK policies at early boot.
4146 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
4147 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
4148 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
4149 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
4150 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
4151 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
4152 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
4153 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
4154 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
4155 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
4156 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
4157 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
4158 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
4159 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
4163 * Timer units now support calendar time events in addition to
4164 monotonic time events. That means you can now trigger a unit
4165 based on a calendar time specification such as "Thu,Fri
4166 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
4167 or fifth day of any month of the year 2013, given that it is
4168 a thursday or friday. This brings timer event support
4169 considerably closer to cron's capabilities. For details on
4170 the supported calendar time specification language see
4173 * udev now supports a number of different naming policies for
4174 network interfaces for predictable names, and a combination
4175 of these policies is now the default. Please see this wiki
4176 document for details:
4178 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
4180 * Auke Kok's bootchart implementation has been added to the
4181 systemd tree. It is an optional component that can graph the
4182 boot in quite some detail. It is one of the best bootchart
4183 implementations around and minimal in its code and
4186 * nss-myhostname has been integrated into the systemd source
4187 tree. nss-myhostname guarantees that the local hostname
4188 always stays resolvable via NSS. It has been a weak
4189 requirement of systemd-hostnamed since a long time, and
4190 since its code is actually trivial we decided to just
4191 include it in systemd's source tree. It can be turned off
4192 with a configure switch.
4194 * The read-ahead logic is now capable of properly detecting
4195 whether a btrfs file system is on SSD or rotating media, in
4196 order to optimize the read-ahead scheme. Previously, it was
4197 only capable of detecting this on traditional file systems
4200 * In udev, additional device properties are now read from the
4201 IAB in addition to the OUI database. Also, Bluetooth company
4202 identities are attached to the devices as well.
4204 * In service files %U may be used as specifier that is
4205 replaced by the configured user name of the service.
4207 * nspawn may now be invoked without a controlling TTY. This
4208 makes it suitable for invocation as its own service. This
4209 may be used to set up a simple containerized server system
4210 using only core OS tools.
4212 * systemd and nspawn can now accept socket file descriptors
4213 when they are started for socket activation. This enables
4214 implementation of socket activated nspawn
4215 containers. i.e. think about autospawning an entire OS image
4216 when the first SSH or HTTP connection is received. We expect
4217 that similar functionality will also be added to libvirt-lxc
4220 * journalctl will now suppress ANSI color codes when
4221 presenting log data.
4223 * systemctl will no longer show control group information for
4224 a unit if a the control group is empty anyway.
4226 * logind can now automatically suspend/hibernate/shutdown the
4229 * /etc/machine-info and hostnamed now also expose the chassis
4230 type of the system. This can be used to determine whether
4231 the local system is a laptop, desktop, handset or
4232 tablet. This information may either be configured by the
4233 user/vendor or is automatically determined from ACPI and DMI
4234 information if possible.
4236 * A number of PolicyKit actions are now bound together with
4237 "imply" rules. This should simplify creating UIs because
4238 many actions will now authenticate similar ones as well.
4240 * Unit files learnt a new condition ConditionACPower= which
4241 may be used to conditionalize a unit depending on whether an
4242 AC power source is connected or not, of whether the system
4243 is running on battery power.
4245 * systemctl gained a new "is-failed" verb that may be used in
4246 shell scripts and suchlike to check whether a specific unit
4247 is in the "failed" state.
4249 * The EnvironmentFile= setting in unit files now supports file
4250 globbing, and can hence be used to easily read a number of
4251 environment files at once.
4253 * systemd will no longer detect and recognize specific
4254 distributions. All distribution-specific #ifdeffery has been
4255 removed, systemd is now fully generic and
4256 distribution-agnostic. Effectively, not too much is lost as
4257 a lot of the code is still accessible via explicit configure
4258 switches. However, support for some distribution specific
4259 legacy configuration file formats has been dropped. We
4260 recommend distributions to simply adopt the configuration
4261 files everybody else uses now and convert the old
4262 configuration from packaging scripts. Most distributions
4263 already did that. If that's not possible or desirable,
4264 distributions are welcome to forward port the specific
4265 pieces of code locally from the git history.
4267 * When logging a message about a unit systemd will now always
4268 log the unit name in the message meta data.
4270 * localectl will now also discover system locale data that is
4271 not stored in locale archives, but directly unpacked.
4273 * logind will no longer unconditionally use framebuffer
4274 devices as seat masters, i.e. as devices that are required
4275 to be existing before a seat is considered preset. Instead,
4276 it will now look for all devices that are tagged as
4277 "seat-master" in udev. By default framebuffer devices will
4278 be marked as such, but depending on local systems other
4279 devices might be marked as well. This may be used to
4280 integrate graphics cards using closed source drivers (such
4281 as NVidia ones) more nicely into logind. Note however, that
4282 we recommend using the open source NVidia drivers instead,
4283 and no udev rules for the closed-source drivers will be
4284 shipped from us upstream.
4286 Contributions from: Adam Williamson, Alessandro Crismani, Auke
4287 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
4288 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
4289 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
4290 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
4291 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
4292 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
4293 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
4294 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
4295 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
4296 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
4297 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
4302 * udev gained support for loading additional device properties
4303 from an indexed database that is keyed by vendor/product IDs
4304 and similar device identifiers. For the beginning this
4305 "hwdb" is populated with data from the well-known PCI and
4306 USB database, but also includes PNP, ACPI and OID data. In
4307 the longer run this indexed database shall grow into
4308 becoming the one central database for non-essential
4309 userspace device metadata. Previously, data from the PCI/USB
4310 database was only attached to select devices, since the
4311 lookup was a relatively expensive operation due to O(n) time
4312 complexity (with n being the number of entries in the
4313 database). Since this is now O(1), we decided to add in this
4314 data for all devices where this is available, by
4315 default. Note that the indexed database needs to be rebuilt
4316 when new data files are installed. To achieve this you need
4317 to update your packaging scripts to invoke "udevadm hwdb
4318 --update" after installation of hwdb data files. For
4319 RPM-based distributions we introduced the new
4320 %udev_hwdb_update macro for this purpose.
4322 * The Journal gained support for the "Message Catalog", an
4323 indexed database to link up additional information with
4324 journal entries. For further details please check:
4326 http://www.freedesktop.org/wiki/Software/systemd/catalog
4328 The indexed message catalog database also needs to be
4329 rebuilt after installation of message catalog files. Use
4330 "journalctl --update-catalog" for this. For RPM-based
4331 distributions we introduced the %journal_catalog_update
4332 macro for this purpose.
4334 * The Python Journal bindings gained support for the standard
4335 Python logging framework.
4337 * The Journal API gained new functions for checking whether
4338 the underlying file system of a journal file is capable of
4339 properly reporting file change notifications, or whether
4340 applications that want to reflect journal changes "live"
4341 need to recheck journal files continuously in appropriate
4344 * It is now possible to set the "age" field for tmpfiles
4345 entries to 0, indicating that files matching this entry
4346 shall always be removed when the directories are cleaned up.
4348 * coredumpctl gained a new "gdb" verb which invokes gdb
4349 right-away on the selected coredump.
4351 * There's now support for "hybrid sleep" on kernels that
4352 support this, in addition to "suspend" and "hibernate". Use
4353 "systemctl hybrid-sleep" to make use of this.
4355 * logind's HandleSuspendKey= setting (and related settings)
4356 now gained support for a new "lock" setting to simply
4357 request the screen lock on all local sessions, instead of
4358 actually executing a suspend or hibernation.
4360 * systemd will now mount the EFI variables file system by
4363 * Socket units now gained support for configuration of the
4364 SMACK security label.
4366 * timedatectl will now output the time of the last and next
4367 daylight saving change.
4369 * We dropped support for various legacy and distro-specific
4370 concepts, such as insserv, early-boot SysV services
4371 (i.e. those for non-standard runlevels such as 'b' or 'S')
4372 or ArchLinux /etc/rc.conf support. We recommend the
4373 distributions who still need support this to either continue
4374 to maintain the necessary patches downstream, or find a
4375 different solution. (Talk to us if you have questions!)
4377 * Various systemd components will now bypass PolicyKit checks
4378 for root and otherwise handle properly if PolicyKit is not
4379 found to be around. This should fix most issues for
4380 PolicyKit-less systems. Quite frankly this should have been
4381 this way since day one. It is absolutely our intention to
4382 make systemd work fine on PolicyKit-less systems, and we
4383 consider it a bug if something does not work as it should if
4384 PolicyKit is not around.
4386 * For embedded systems it is now possible to build udev and
4387 systemd without blkid and/or kmod support.
4389 * "systemctl switch-root" is now capable of switching root
4390 more than once. I.e. in addition to transitions from the
4391 initrd to the host OS it is now possible to transition to
4392 further OS images from the host. This is useful to implement
4393 offline updating tools.
4395 * Various other additions have been made to the RPM macros
4396 shipped with systemd. Use %udev_rules_update() after
4397 installing new udev rules files. %_udevhwdbdir,
4398 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
4399 %_sysctldir are now available which resolve to the right
4400 directories for packages to place various data files in.
4402 * journalctl gained the new --full switch (in addition to
4403 --all, to disable ellipsation for long messages.
4405 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
4406 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
4407 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
4408 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
4409 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
4410 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
4411 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
4412 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
4413 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
4417 * journalctl gained new --since= and --until= switches to
4418 filter by time. It also now supports nice filtering for
4419 units via --unit=/-u.
4421 * Type=oneshot services may use ExecReload= and do the
4424 * The journal daemon now supports time-based rotation and
4425 vacuuming, in addition to the usual disk-space based
4428 * The journal will now index the available field values for
4429 each field name. This enables clients to show pretty drop
4430 downs of available match values when filtering. The bash
4431 completion of journalctl has been updated
4432 accordingly. journalctl gained a new switch -F to list all
4433 values a certain field takes in the journal database.
4435 * More service events are now written as structured messages
4436 to the journal, and made recognizable via message IDs.
4438 * The timedated, localed and hostnamed mini-services which
4439 previously only provided support for changing time, locale
4440 and hostname settings from graphical DEs such as GNOME now
4441 also have a minimal (but very useful) text-based client
4442 utility each. This is probably the nicest way to changing
4443 these settings from the command line now, especially since
4444 it lists available options and is fully integrated with bash
4447 * There's now a new tool "systemd-coredumpctl" to list and
4448 extract coredumps from the journal.
4450 * We now install a README each in /var/log/ and
4451 /etc/rc.d/init.d explaining where the system logs and init
4452 scripts went. This hopefully should help folks who go to
4453 that dirs and look into the otherwise now empty void and
4454 scratch their heads.
4456 * When user-services are invoked (by systemd --user) the
4457 $MANAGERPID env var is set to the PID of systemd.
4459 * SIGRTMIN+24 when sent to a --user instance will now result
4460 in immediate termination of systemd.
4462 * gatewayd received numerous feature additions such as a
4463 "follow" mode, for live syncing and filtering.
4465 * browse.html now allows filtering and showing detailed
4466 information on specific entries. Keyboard navigation and
4467 mouse screen support has been added.
4469 * gatewayd/journalctl now supports HTML5/JSON
4470 Server-Sent-Events as output.
4472 * The SysV init script compatibility logic will now
4473 heuristically determine whether a script supports the
4474 "reload" verb, and only then make this available as
4477 * "systemctl status --follow" has been removed, use "journalctl
4480 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
4481 have been removed since they are hardly useful to be
4484 * And I'd like to take the opportunity to specifically mention
4485 Zbigniew for his great contributions. Zbigniew, you rock!
4487 Contributions from: Andrew Eikum, Christian Hesse, Colin
4488 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
4489 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
4490 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
4491 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
4492 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
4493 Jędrzejewski-Szmek, Сковорода Никита Андреевич
4497 * If /etc/vconsole.conf is non-existent or empty we will no
4498 longer load any console font or key map at boot by
4499 default. Instead the kernel defaults will be left
4500 intact. This is definitely the right thing to do, as no
4501 configuration should mean no configuration, and hard-coding
4502 font names that are different on all archs is probably a bad
4503 idea. Also, the kernel default key map and font should be
4504 good enough for most cases anyway, and mostly identical to
4505 the userspace fonts/key maps we previously overloaded them
4506 with. If distributions want to continue to default to a
4507 non-kernel font or key map they should ship a default
4508 /etc/vconsole.conf with the appropriate contents.
4510 Contributions from: Colin Walters, Daniel J Walsh, Dave
4511 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
4512 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
4516 * journalctl gained a new --cursor= switch to show entries
4517 starting from the specified location in the journal.
4519 * We now enforce a size limit on journal entry fields exported
4520 with "-o json" in journalctl. Fields larger than 4K will be
4521 assigned null. This can be turned off with --all.
4523 * An (optional) journal gateway daemon is now available as
4524 "systemd-journal-gatewayd.service". This service provides
4525 access to the journal via HTTP and JSON. This functionality
4526 will be used to implement live log synchronization in both
4527 pull and push modes, but has various other users too, such
4528 as easy log access for debugging of embedded devices. Right
4529 now it is already useful to retrieve the journal via HTTP:
4531 # systemctl start systemd-journal-gatewayd.service
4532 # wget http://localhost:19531/entries
4534 This will download the journal contents in a
4535 /var/log/messages compatible format. The same as JSON:
4537 # curl -H"Accept: application/json" http://localhost:19531/entries
4539 This service is also accessible via a web browser where a
4540 single static HTML5 app is served that uses the JSON logic
4541 to enable the user to do some basic browsing of the
4542 journal. This will be extended later on. Here's an example
4543 screenshot of this app in its current state:
4545 http://0pointer.de/public/journal-gatewayd
4547 Contributions from: Kay Sievers, Lennart Poettering, Robert
4548 Milasan, Tom Gundersen
4552 * The bash completion logic is now available for journalctl
4555 * We do not mount the "cpuset" controller anymore together with
4556 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
4557 started if no parameters are assigned to it. "cpuset" hence
4558 broke code that assumed it it could create "cpu" groups and
4561 * journalctl -f will now subscribe to terminal size changes,
4562 and line break accordingly.
4564 Contributions from: Dave Reisner, Kay Sievers, Lennart
4565 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
4569 * nspawn will now create a symlink /etc/localtime in the
4570 container environment, copying the host's timezone
4571 setting. Previously this has been done via a bind mount, but
4572 since symlinks cannot be bind mounted this has now been
4573 changed to create/update the appropriate symlink.
4575 * journalctl -n's line number argument is now optional, and
4576 will default to 10 if omitted.
4578 * journald will now log the maximum size the journal files may
4579 take up on disk. This is particularly useful if the default
4580 built-in logic of determining this parameter from the file
4581 system size is used. Use "systemctl status
4582 systemd-journald.service" to see this information.
4584 * The multi-seat X wrapper tool has been stripped down. As X
4585 is now capable of enumerating graphics devices via udev in a
4586 seat-aware way the wrapper is not strictly necessary
4587 anymore. A stripped down temporary stop-gap is still shipped
4588 until the upstream display managers have been updated to
4589 fully support the new X logic. Expect this wrapper to be
4590 removed entirely in one of the next releases.
4592 * HandleSleepKey= in logind.conf has been split up into
4593 HandleSuspendKey= and HandleHibernateKey=. The old setting
4594 is not available anymore. X11 and the kernel are
4595 distinguishing between these keys and we should too. This
4596 also means the inhibition lock for these keys has been split
4599 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
4600 Poettering, Lukas Nykryn, Václav Pavlín
4604 * Whenever a unit changes state we will now log this to the
4605 journal and show along the unit's own log output in
4608 * ConditionPathIsMountPoint= can now properly detect bind
4609 mount points too. (Previously, a bind mount of one file
4610 system to another place in the same file system could not be
4611 detected as mount, since they shared struct stat's st_dev
4614 * We will now mount the cgroup controllers cpu, cpuacct,
4615 cpuset and the controllers net_cls, net_prio together by
4618 * nspawn containers will now have a virtualized boot
4619 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
4620 over with a randomized ID at container initialization). This
4621 has the effect of making "journalctl -b" do the right thing
4624 * The JSON output journal serialization has been updated not
4625 to generate "endless" list objects anymore, but rather one
4626 JSON object per line. This is more in line how most JSON
4627 parsers expect JSON objects. The new output mode
4628 "json-pretty" has been added to provide similar output, but
4629 neatly aligned for readability by humans.
4631 * We dropped all explicit sync() invocations in the shutdown
4632 code. The kernel does this implicitly anyway in the kernel
4633 reboot() syscall. halt(8)'s -n option is now a compatibility
4636 * We now support virtualized reboot() in containers, as
4637 supported by newer kernels. We will fall back to exit() if
4638 CAP_SYS_REBOOT is not available to the container. Also,
4639 nspawn makes use of this now and will actually reboot the
4640 container if the containerized OS asks for that.
4642 * journalctl will only show local log output by default
4643 now. Use --merge (-m) to show remote log output, too.
4645 * libsystemd-journal gained the new sd_journal_get_usage()
4646 call to determine the current disk usage of all journal
4647 files. This is exposed in the new "journalctl --disk-usage"
4650 * journald gained a new configuration setting SplitMode= in
4651 journald.conf which may be used to control how user journals
4652 are split off. See journald.conf(5) for details.
4654 * A new condition type ConditionFileNotEmpty= has been added.
4656 * tmpfiles' "w" lines now support file globbing, to write
4657 multiple files at once.
4659 * We added Python bindings for the journal submission
4660 APIs. More Python APIs for a number of selected APIs will
4661 likely follow. Note that we intend to add native bindings
4662 only for the Python language, as we consider it common
4663 enough to deserve bindings shipped within systemd. There are
4664 various projects outside of systemd that provide bindings
4665 for languages such as PHP or Lua.
4667 * Many conditions will now resolve specifiers such as %i. In
4668 addition, PathChanged= and related directives of .path units
4669 now support specifiers as well.
4671 * There's now a new RPM macro definition for the system preset
4674 * journald will now warn if it ca not forward a message to the
4675 syslog daemon because its socket is full.
4677 * timedated will no longer write or process /etc/timezone,
4678 except on Debian. As we do not support late mounted /usr
4679 anymore /etc/localtime always being a symlink is now safe,
4680 and hence the information in /etc/timezone is not necessary
4683 * logind will now always reserve one VT for a text getty (VT6
4684 by default). Previously if more than 6 X sessions where
4685 started they took up all the VTs with auto-spawned gettys,
4686 so that no text gettys were available anymore.
4688 * udev will now automatically inform the btrfs kernel logic
4689 about btrfs RAID components showing up. This should make
4690 simple hotplug based btrfs RAID assembly work.
4692 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
4693 (but not for its children which will stay at the kernel
4694 default). This should allow setups with a lot more listening
4697 * systemd will now always pass the configured timezone to the
4698 kernel at boot. timedated will do the same when the timezone
4701 * logind's inhibition logic has been updated. By default,
4702 logind will now handle the lid switch, the power and sleep
4703 keys all the time, even in graphical sessions. If DEs want
4704 to handle these events on their own they should take the new
4705 handle-power-key, handle-sleep-key and handle-lid-switch
4706 inhibitors during their runtime. A simple way to achieve
4707 that is to invoke the DE wrapped in an invocation of:
4709 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
4711 * Access to unit operations is now checked via SELinux taking
4712 the unit file label and client process label into account.
4714 * systemd will now notify the administrator in the journal
4715 when he over-mounts a non-empty directory.
4717 * There are new specifiers that are resolved in unit files,
4718 for the host name (%H), the machine ID (%m) and the boot ID
4721 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
4722 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
4723 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
4724 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
4725 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
4726 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
4727 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
4731 * Support for reading structured kernel messages from
4732 /dev/kmsg has now been added and is enabled by default.
4734 * Support for reading kernel messages from /proc/kmsg has now
4735 been removed. If you want kernel messages in the journal
4736 make sure to run a recent kernel (>= 3.5) that supports
4737 reading structured messages from /dev/kmsg (see
4738 above). /proc/kmsg is now exclusive property of classic
4739 syslog daemons again.
4741 * The libudev API gained the new
4742 udev_device_new_from_device_id() call.
4744 * The logic for file system namespace (ReadOnlyDirectory=,
4745 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
4746 require pivot_root() anymore. This means fewer temporary
4747 directories are created below /tmp for this feature.
4749 * nspawn containers will now see and receive all submounts
4750 made on the host OS below the root file system of the
4753 * Forward Secure Sealing is now supported for Journal files,
4754 which provide cryptographical sealing of journal files so
4755 that attackers cannot alter log history anymore without this
4756 being detectable. Lennart will soon post a blog story about
4757 this explaining it in more detail.
4759 * There are two new service settings RestartPreventExitStatus=
4760 and SuccessExitStatus= which allow configuration of exit
4761 status (exit code or signal) which will be excepted from the
4762 restart logic, resp. consider successful.
4764 * journalctl gained the new --verify switch that can be used
4765 to check the integrity of the structure of journal files and
4766 (if Forward Secure Sealing is enabled) the contents of
4769 * nspawn containers will now be run with /dev/stdin, /dev/fd/
4770 and similar symlinks pre-created. This makes running shells
4771 as container init process a lot more fun.
4773 * The fstab support can now handle PARTUUID= and PARTLABEL=
4776 * A new ConditionHost= condition has been added to match
4777 against the hostname (with globs) and machine ID. This is
4778 useful for clusters where a single OS image is used to
4779 provision a large number of hosts which shall run slightly
4780 different sets of services.
4782 * Services which hit the restart limit will now be placed in a
4785 Contributions from: Bertram Poettering, Dave Reisner, Huang
4786 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
4787 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
4791 * When running in --user mode systemd will now become a
4792 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
4793 tree a lot more organized.
4795 * A new PartOf= unit dependency type has been introduced that
4796 may be used to group services in a natural way.
4798 * "systemctl enable" may now be used to enable instances of
4801 * journalctl now prints error log levels in red, and
4802 warning/notice log levels in bright white. It also supports
4803 filtering by log level now.
4805 * cgtop gained a new -n switch (similar to top), to configure
4806 the maximum number of iterations to run for. It also gained
4807 -b, to run in batch mode (accepting no input).
4809 * The suffix ".service" may now be omitted on most systemctl
4810 command lines involving service unit names.
4812 * There's a new bus call in logind to lock all sessions, as
4813 well as a loginctl verb for it "lock-sessions".
4815 * libsystemd-logind.so gained a new call sd_journal_perror()
4816 that works similar to libc perror() but logs to the journal
4817 and encodes structured information about the error number.
4819 * /etc/crypttab entries now understand the new keyfile-size=
4822 * shutdown(8) now can send a (configurable) wall message when
4823 a shutdown is cancelled.
4825 * The mount propagation mode for the root file system will now
4826 default to "shared", which is useful to make containers work
4827 nicely out-of-the-box so that they receive new mounts from
4828 the host. This can be undone locally by running "mount
4829 --make-rprivate /" if needed.
4831 * The prefdm.service file has been removed. Distributions
4832 should maintain this unit downstream if they intend to keep
4833 it around. However, we recommend writing normal unit files
4834 for display managers instead.
4836 * Since systemd is a crucial part of the OS we will now
4837 default to a number of compiler switches that improve
4838 security (hardening) such as read-only relocations, stack
4839 protection, and suchlike.
4841 * The TimeoutSec= setting for services is now split into
4842 TimeoutStartSec= and TimeoutStopSec= to allow configuration
4843 of individual time outs for the start and the stop phase of
4846 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
4847 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
4848 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
4849 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
4850 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
4851 Gundersen, Zbigniew Jędrzejewski-Szmek
4855 * The journal and id128 C APIs are now fully documented as man
4858 * Extra safety checks have been added when transitioning from
4859 the initial RAM disk to the main system to avoid accidental
4862 * /etc/crypttab entries now understand the new keyfile-offset=
4865 * systemctl -t can now be used to filter by unit load state.
4867 * The journal C API gained the new sd_journal_wait() call to
4868 make writing synchronous journal clients easier.
4870 * journalctl gained the new -D switch to show journals from a
4873 * journalctl now displays a special marker between log
4874 messages of two different boots.
4876 * The journal is now explicitly flushed to /var via a service
4877 systemd-journal-flush.service, rather than implicitly simply
4878 by seeing /var/log/journal to be writable.
4880 * journalctl (and the journal C APIs) can now match for much
4881 more complex expressions, with alternatives and
4884 * When transitioning from the initial RAM disk to the main
4885 system we will now kill all processes in a killing spree to
4886 ensure no processes stay around by accident.
4888 * Three new specifiers may be used in unit files: %u, %h, %s
4889 resolve to the user name, user home directory resp. user
4890 shell. This is useful for running systemd user instances.
4892 * We now automatically rotate journal files if their data
4893 object hash table gets a fill level > 75%. We also size the
4894 hash table based on the configured maximum file size. This
4895 together should lower hash collisions drastically and thus
4896 speed things up a bit.
4898 * journalctl gained the new "--header" switch to introspect
4899 header data of journal files.
4901 * A new setting SystemCallFilters= has been added to services
4902 which may be used to apply blacklists or whitelists to
4903 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
4905 * nspawn gained a new --link-journal= switch (and quicker: -j)
4906 to link the container journal with the host. This makes it
4907 very easy to centralize log viewing on the host for all
4908 guests while still keeping the journal files separated.
4910 * Many bugfixes and optimizations
4912 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
4913 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
4914 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
4919 * Several tools now understand kernel command line arguments,
4920 which are only read when run in an initial RAM disk. They
4921 usually follow closely their normal counterparts, but are
4924 * There's a new tool to analyze the readahead files that are
4925 automatically generated at boot. Use:
4927 /usr/lib/systemd/systemd-readahead analyze /.readahead
4929 * We now provide an early debug shell on tty9 if this enabled. Use:
4931 systemctl enable debug-shell.service
4933 * All plymouth related units have been moved into the Plymouth
4934 package. Please make sure to upgrade your Plymouth version
4937 * systemd-tmpfiles now supports getting passed the basename of
4938 a configuration file only, in which case it will look for it
4939 in all appropriate directories automatically.
4941 * udevadm info now takes a /dev or /sys path as argument, and
4942 does the right thing. Example:
4944 udevadm info /dev/sda
4945 udevadm info /sys/class/block/sda
4947 * systemctl now prints a warning if a unit is stopped but a
4948 unit that might trigger it continues to run. Example: a
4949 service is stopped but the socket that activates it is left
4952 * "systemctl status" will now mention if the log output was
4953 shortened due to rotation since a service has been started.
4955 * The journal API now exposes functions to determine the
4956 "cutoff" times due to rotation.
4958 * journald now understands SIGUSR1 and SIGUSR2 for triggering
4959 immediately flushing of runtime logs to /var if possible,
4960 resp. for triggering immediate rotation of the journal
4963 * It is now considered an error if a service is attempted to
4964 be stopped that is not loaded.
4966 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
4968 * systemd-analyze now supports Python 3
4970 * tmpfiles now supports cleaning up directories via aging
4971 where the first level dirs are always kept around but
4972 directories beneath it automatically aged. This is enabled
4973 by prefixing the age field with '~'.
4975 * Seat objects now expose CanGraphical, CanTTY properties
4976 which is required to deal with very fast bootups where the
4977 display manager might be running before the graphics drivers
4978 completed initialization.
4980 * Seat objects now expose a State property.
4982 * We now include RPM macros for service enabling/disabling
4983 based on the preset logic. We recommend RPM based
4984 distributions to make use of these macros if possible. This
4985 makes it simpler to reuse RPM spec files across
4988 * We now make sure that the collected systemd unit name is
4989 always valid when services log to the journal via
4992 * There's a new man page kernel-command-line(7) detailing all
4993 command line options we understand.
4995 * The fstab generator may now be disabled at boot by passing
4996 fstab=0 on the kernel command line.
4998 * A new kernel command line option modules-load= is now understood
4999 to load a specific kernel module statically, early at boot.
5001 * Unit names specified on the systemctl command line are now
5002 automatically escaped as needed. Also, if file system or
5003 device paths are specified they are automatically turned
5004 into the appropriate mount or device unit names. Example:
5006 systemctl status /home
5007 systemctl status /dev/sda
5009 * The SysVConsole= configuration option has been removed from
5010 system.conf parsing.
5012 * The SysV search path is no longer exported on the D-Bus
5015 * The Names= option is been removed from unit file parsing.
5017 * There's a new man page bootup(7) detailing the boot process.
5019 * Every unit and every generator we ship with systemd now
5020 comes with full documentation. The self-explanatory boot is
5023 * A couple of services gained "systemd-" prefixes in their
5024 name if they wrap systemd code, rather than only external
5025 code. Among them fsck@.service which is now
5026 systemd-fsck@.service.
5028 * The HaveWatchdog property has been removed from the D-Bus
5031 * systemd.confirm_spawn= on the kernel command line should now
5034 * There's a new man page crypttab(5) which details all options
5035 we actually understand.
5037 * systemd-nspawn gained a new --capability= switch to pass
5038 additional capabilities to the container.
5040 * timedated will now read known NTP implementation unit names
5041 from /usr/lib/systemd/ntp-units.d/*.list,
5042 systemd-timedated-ntp.target has been removed.
5044 * journalctl gained a new switch "-b" that lists log data of
5045 the current boot only.
5047 * The notify socket is in the abstract namespace again, in
5048 order to support daemons which chroot() at start-up.
5050 * There is a new Storage= configuration option for journald
5051 which allows configuration of where log data should go. This
5052 also provides a way to disable journal logging entirely, so
5053 that data collected is only forwarded to the console, the
5054 kernel log buffer or another syslog implementation.
5056 * Many bugfixes and optimizations
5058 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
5059 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
5060 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
5061 Shawn Landden, Tom Gundersen
5065 * "systemctl help <unit>" now shows the man page if one is
5068 * Several new man pages have been added.
5070 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
5071 MaxLevelConsole= can now be specified in
5072 journald.conf. These options allow reducing the amount of
5073 data stored on disk or forwarded by the log level.
5075 * TimerSlackNSec= can now be specified in system.conf for
5076 PID1. This allows system-wide power savings.
5078 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
5079 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
5084 * logind is now capable of (optionally) handling power and
5085 sleep keys as well as the lid switch.
5087 * journalctl now understands the syntax "journalctl
5088 /usr/bin/avahi-daemon" to get all log output of a specific
5091 * CapabilityBoundingSet= in system.conf now also influences
5092 the capability bound set of usermode helpers of the kernel.
5094 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
5095 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
5096 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
5097 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
5101 * Note that we skipped 139 releases here in order to set the
5102 new version to something that is greater than both udev's
5103 and systemd's most recent version number.
5105 * udev: all udev sources are merged into the systemd source tree now.
5106 All future udev development will happen in the systemd tree. It
5107 is still fully supported to use the udev daemon and tools without
5108 systemd running, like in initramfs or other init systems. Building
5109 udev though, will require the *build* of the systemd tree, but
5110 udev can be properly *run* without systemd.
5112 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
5113 should be used to create dead device nodes as workarounds for broken
5116 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
5117 no longer supported. udev_monitor_new_from_netlink() needs to be
5118 used to subscribe to events.
5120 * udev: when udevd is started by systemd, processes which are left
5121 behind by forking them off of udev rules, are unconditionally cleaned
5122 up and killed now after the event handling has finished. Services or
5123 daemons must be started as systemd services. Services can be
5124 pulled-in by udev to get started, but they can no longer be directly
5125 forked by udev rules.
5127 * udev: the daemon binary is called systemd-udevd now and installed
5128 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
5129 to adapt to that, create symlink, or rename the binary after building
5132 * libudev no longer provides these symbols:
5133 udev_monitor_from_socket()
5134 udev_queue_get_failed_list_entry()
5135 udev_get_{dev,sys,run}_path()
5136 The versions number was bumped and symbol versioning introduced.
5138 * systemd-loginctl and systemd-journalctl have been renamed
5139 to loginctl and journalctl to match systemctl.
5141 * The config files: /etc/systemd/systemd-logind.conf and
5142 /etc/systemd/systemd-journald.conf have been renamed to
5143 logind.conf and journald.conf. Package updates should rename
5144 the files to the new names on upgrade.
5146 * For almost all files the license is now LGPL2.1+, changed
5147 from the previous GPL2.0+. Exceptions are some minor stuff
5148 of udev (which will be changed to LGPL2.1 eventually, too),
5149 and the MIT licensed sd-daemon.[ch] library that is suitable
5150 to be used as drop-in files.
5152 * systemd and logind now handle system sleep states, in
5153 particular suspending and hibernating.
5155 * logind now implements a sleep/shutdown/idle inhibiting logic
5156 suitable for a variety of uses. Soonishly Lennart will blog
5157 about this in more detail.
5159 * var-run.mount and var-lock.mount are no longer provided
5160 (which prevously bind mounted these directories to their new
5161 places). Distributions which have not converted these
5162 directories to symlinks should consider stealing these files
5163 from git history and add them downstream.
5165 * We introduced the Documentation= field for units and added
5166 this to all our shipped units. This is useful to make it
5167 easier to explore the boot and the purpose of the various
5170 * All smaller setup units (such as
5171 systemd-vconsole-setup.service) now detect properly if they
5172 are run in a container and are skipped when
5173 appropriate. This guarantees an entirely noise-free boot in
5174 Linux container environments such as systemd-nspawn.
5176 * A framework for implementing offline system updates is now
5177 integrated, for details see:
5178 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
5180 * A new service type Type=idle is available now which helps us
5181 avoiding ugly interleaving of getty output and boot status
5184 * There's now a system-wide CapabilityBoundingSet= option to
5185 globally reduce the set of capabilities for the
5186 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
5187 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
5188 even CAP_NET_ADMIN system-wide for secure systems.
5190 * There are now system-wide DefaultLimitXXX= options to
5191 globally change the defaults of the various resource limits
5192 for all units started by PID 1.
5194 * Harald Hoyer's systemd test suite has been integrated into
5195 systemd which allows easy testing of systemd builds in qemu
5196 and nspawn. (This is really awesome! Ask us for details!)
5198 * The fstab parser is now implemented as generator, not inside
5201 * systemctl will now warn you if .mount units generated from
5202 /etc/fstab are out of date due to changes in fstab that
5203 have not been read by systemd yet.
5205 * systemd is now suitable for usage in initrds. Dracut has
5206 already been updated to make use of this. With this in place
5207 initrds get a slight bit faster but primarily are much
5208 easier to introspect and debug since "systemctl status" in
5209 the host system can be used to introspect initrd services,
5210 and the journal from the initrd is kept around too.
5212 * systemd-delta has been added, a tool to explore differences
5213 between user/admin configuration and vendor defaults.
5215 * PrivateTmp= now affects both /tmp and /var/tmp.
5217 * Boot time status messages are now much prettier and feature
5218 proper english language. Booting up systemd has never been
5221 * Read-ahead pack files now include the inode number of all
5222 files to pre-cache. When the inode changes the pre-caching
5223 is not attempted. This should be nicer to deal with updated
5224 packages which might result in changes of read-ahead
5227 * We now temporaritly lower the kernel's read_ahead_kb variable
5228 when collecting read-ahead data to ensure the kernel's
5229 built-in read-ahead does not add noise to our measurements
5230 of necessary blocks to pre-cache.
5232 * There's now RequiresMountsFor= to add automatic dependencies
5233 for all mounts necessary for a specific file system path.
5235 * MountAuto= and SwapAuto= have been removed from
5236 system.conf. Mounting file systems at boot has to take place
5239 * nspawn now learned a new switch --uuid= to set the machine
5240 ID on the command line.
5242 * nspawn now learned the -b switch to automatically search
5245 * vt102 is now the default TERM for serial TTYs, upgraded from
5248 * systemd-logind now works on VT-less systems.
5250 * The build tree has been reorganized. The individual
5251 components now have directories of their own.
5253 * A new condition type ConditionPathIsReadWrite= is now available.
5255 * nspawn learned the new -C switch to create cgroups for the
5256 container in other hierarchies.
5258 * We now have support for hardware watchdogs, configurable in
5261 * The scheduled shutdown logic now has a public API.
5263 * We now mount /tmp as tmpfs by default, but this can be
5264 masked and /etc/fstab can override it.
5266 * Since udisks does not make use of /media anymore we are not
5267 mounting a tmpfs on it anymore.
5269 * journalctl gained a new --local switch to only interleave
5270 locally generated journal files.
5272 * We can now load the IMA policy at boot automatically.
5274 * The GTK tools have been split off into a systemd-ui.
5276 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
5277 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
5278 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
5279 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
5280 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
5281 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
5282 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
5283 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
5284 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
5289 * This is mostly a bugfix release
5291 * Support optional initialization of the machine ID from the
5292 KVM or container configured UUID.
5294 * Support immediate reboots with "systemctl reboot -ff"
5296 * Show /etc/os-release data in systemd-analyze output
5298 * Many bugfixes for the journal, including endianness fixes and
5299 ensuring that disk space enforcement works
5301 * sd-login.h is C++ comptaible again
5303 * Extend the /etc/os-release format on request of the Debian
5306 * We now refuse non-UTF8 strings used in various configuration
5307 and unit files. This is done to ensure we do not pass invalid
5308 data over D-Bus or expose it elsewhere.
5310 * Register Mimo USB Screens as suitable for automatic seat
5313 * Read SELinux client context from journal clients in a race
5316 * Reorder configuration file lookup order. /etc now always
5317 overrides /run in order to allow the administrator to always
5318 and unconditionally override vendor supplied or
5319 automatically generated data.
5321 * The various user visible bits of the journal now have man
5322 pages. We still lack man pages for the journal API calls
5325 * We now ship all man pages in HTML format again in the
5328 Contributions from: Dave Reisner, Dirk Eibach, Frederic
5329 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
5330 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
5335 * This is mostly a bugfix release
5337 * systems lacking /etc/os-release are no longer supported.
5339 * Various functionality updates to libsystemd-login.so
5341 * Track class of PAM logins to distinguish greeters from
5344 Contributions from: Kay Sievers, Lennart Poettering, Michael
5349 * This is an important bugfix release for v41.
5351 * Building man pages is now optional which should be useful
5352 for those building systemd from git but unwilling to install
5355 * Watchdog support for supervising services is now usable. In
5356 a future release support for hardware watchdogs
5357 (i.e. /dev/watchdog) will be added building on this.
5359 * Service start rate limiting is now configurable and can be
5360 turned off per service. When a start rate limit is hit a
5361 reboot can automatically be triggered.
5363 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
5365 Contributions from: Benjamin Franzke, Bill Nottingham,
5366 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
5367 Schmidt, Michał Górny, Piotr Drąg
5371 * The systemd binary is installed /usr/lib/systemd/systemd now;
5372 An existing /sbin/init symlink needs to be adapted with the
5375 * The code that loads kernel modules has been ported to invoke
5376 libkmod directly, instead of modprobe. This means we do not
5377 support systems with module-init-tools anymore.
5379 * Watchdog support is now already useful, but still not
5382 * A new kernel command line option systemd.setenv= is
5383 understood to set system wide environment variables
5384 dynamically at boot.
5386 * We now limit the set of capabilities of systemd-journald.
5388 * We now set SIGPIPE to ignore by default, since it only is
5389 useful in shell pipelines, and has little use in general
5390 code. This can be disabled with IgnoreSIPIPE=no in unit
5393 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
5394 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
5399 * This is mostly a bugfix release
5401 * We now expose the reason why a service failed in the
5402 "Result" D-Bus property.
5404 * Rudimentary service watchdog support (will be completed over
5405 the next few releases.)
5407 * When systemd forks off in order execute some service we will
5408 now immediately changes its argv[0] to reflect which process
5409 it will execute. This is useful to minimize the time window
5410 with a generic argv[0], which makes bootcharts more useful
5412 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
5413 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
5414 Mike Kazantsev, Ray Strode
5418 * This is mostly a test release, but incorporates many
5421 * New systemd-cgtop tool to show control groups by their
5424 * Linking against libacl for ACLs is optional again. If
5425 disabled, support tracking device access for active logins
5426 goes becomes unavailable, and so does access to the user
5427 journals by the respective users.
5429 * If a group "adm" exists, journal files are automatically
5430 owned by them, thus allow members of this group full access
5431 to the system journal as well as all user journals.
5433 * The journal now stores the SELinux context of the logging
5434 client for all entries.
5436 * Add C++ inclusion guards to all public headers
5438 * New output mode "cat" in the journal to print only text
5439 messages, without any meta data like date or time.
5441 * Include tiny X server wrapper as a temporary stop-gap to
5442 teach XOrg udev display enumeration. This is used by display
5443 managers such as gdm, and will go away as soon as XOrg
5444 learned native udev hotplugging for display devices.
5446 * Add new systemd-cat tool for executing arbitrary programs
5447 with STDERR/STDOUT connected to the journal. Can also act as
5448 BSD logger replacement, and does so by default.
5450 * Optionally store all locally generated coredumps in the
5451 journal along with meta data.
5453 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
5454 writing short strings to files (for usage for /sys), and for
5455 creating symlinks, character and block device nodes.
5457 * New unit file option ControlGroupPersistent= to make cgroups
5458 persistent, following the mechanisms outlined in
5459 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
5461 * Support multiple local RTCs in a sane way
5463 * No longer monopolize IO when replaying readahead data on
5464 rotating disks, since we might starve non-file-system IO to
5465 death, since fanotify() will not see accesses done by blkid,
5468 * Do not show kernel threads in systemd-cgls anymore, unless
5469 requested with new -k switch.
5471 Contributions from: Dan Horák, Kay Sievers, Lennart
5472 Poettering, Michal Schmidt
5476 * This is mostly a test release, but incorporates many
5479 * The git repository moved to:
5480 git://anongit.freedesktop.org/systemd/systemd
5481 ssh://git.freedesktop.org/git/systemd/systemd
5483 * First release with the journal
5484 http://0pointer.de/blog/projects/the-journal.html
5486 * The journal replaces both systemd-kmsg-syslogd and
5487 systemd-stdout-bridge.
5489 * New sd_pid_get_unit() API call in libsystemd-logind
5491 * Many systemadm clean-ups
5493 * Introduce remote-fs-pre.target which is ordered before all
5494 remote mounts and may be used to start services before all
5497 * Added Mageia support
5499 * Add bash completion for systemd-loginctl
5501 * Actively monitor PID file creation for daemons which exit in
5502 the parent process before having finished writing the PID
5503 file in the daemon process. Daemons which do this need to be
5504 fixed (i.e. PID file creation must have finished before the
5505 parent exits), but we now react a bit more gracefully to them.
5507 * Add colourful boot output, mimicking the well-known output
5508 of existing distributions.
5510 * New option PassCredentials= for socket units, for
5511 compatibility with a recent kernel ABI breakage.
5513 * /etc/rc.local is now hooked in via a generator binary, and
5514 thus will no longer act as synchronization point during
5517 * systemctl list-unit-files now supports --root=.
5519 * systemd-tmpfiles now understands two new commands: z, Z for
5520 relabelling files according to the SELinux database. This is
5521 useful to apply SELinux labels to specific files in /sys,
5524 * Output of SysV services is now forwarded to both the console
5525 and the journal by default, not only just the console.
5527 * New man pages for all APIs from libsystemd-login.
5529 * The build tree got reorganized and a the build system is a
5530 lot more modular allowing embedded setups to specifically
5531 select the components of systemd they are interested in.
5533 * Support for Linux systems lacking the kernel VT subsystem is
5536 * configure's --with-rootdir= got renamed to
5537 --with-rootprefix= to follow the naming used by udev and
5540 * Unless specified otherwise we will now install to /usr instead
5541 of /usr/local by default.
5543 * Processes with '@' in argv[0][0] are now excluded from the
5544 final shut-down killing spree, following the logic explained
5546 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
5548 * All processes remaining in a service cgroup when we enter
5549 the START or START_PRE states are now killed with
5550 SIGKILL. That means it is no longer possible to spawn
5551 background processes from ExecStart= lines (which was never
5552 supported anyway, and bad style).
5554 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
5555 reloading of units together.
5557 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
5558 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
5559 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
5560 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
5561 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek