1 systemd System and Service Manager
5 * In service units the various ExecXYZ= settings have been extended
6 with an additional special character as first argument of the
7 assigned value: if the character '!' is used the specified command
8 line it will be run with full privileges, regardless of User=,
9 Group=, CapabilityBoundingSet= and similar options. The effect is
10 similar to the existing PermissionsStartOnly= option, but allows
11 configuration of this concept for each executed command line
14 * Services may now alter the service watchdog timeout at runtime by
15 sending a WATCHDOG_USEC= message via sd_notify().
17 * MemoryLimit= and related unit settings now optionally take percentage
18 specifications. The percentage is taken relative to the amount of
19 physical memory in the system (or in case of containers, the assigned
20 amount of memory). This allows scaling service resources neatly with
21 the amount of RAM available on the system. Similarly, systemd-logind's
22 RuntimeDirectorySize= option now also optionally takes percentage
25 * In similar fashion TasksMax= takes percentage values now, too. The
26 value is taken relative to the configured maximum number of processes
27 on the system. The per-service task maximum has been changed to 15%
28 using this functionality. (Effectively this is an increase of 512 →
29 4915 for service units, given the kernel's default pid_max setting.)
31 * Calendar time specifications in .timer units now understand a ".."
32 syntax for time ranges. Example: "4..7:10" may now be used for
33 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
36 * The InaccessableDirectories=, ReadOnlyDirectories= and
37 ReadWriteDirectories= unit file settings have been renamed to
38 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
39 applied to all kinds of file nodes, and not just directories, with
40 the exception of symlinks. Specifically these settings may now be
41 used on block and character device nodes, UNIX sockets and FIFOS as
42 well as regular files. The old names of these settings remain
43 available for compatibility.
45 * systemd will now log about all service processes it kills forcibly
46 (using SIGKILL) because they remained after the clean shutdown phase
47 of the service completed. This should help identifying services that
48 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
49 systemd-logind's configuration a similar log message is generated for
50 processes killed at the end of each session due to this setting.
52 * systemd will now set the $JOURNAL_STREAM environment variable for all
53 services whose stdout/stderr are connected to the Journal (which
54 effectively means by default: all services). The variable contains
55 the device and inode number of the file descriptor used for
56 stdout/stderr. This may be used by invoked programs to detect whether
57 their stdout/stderr is connected to the Journal, in which case they
58 can switch over to direct Journal communication, thus being able to
59 pass extended, structured metadata along with their log messages. As
60 one example, this is now used by glib's logging primitives.
62 * When using systemd's default tmp.mount unit for /tmp, the mount point
63 will now be established with the "nosuid" and "nodev" options. This
64 avoids privilege escalation attacks that put traps and exploits into
65 /tmp. However, this might cause problems if you e. g. put container
66 images or overlays into /tmp; if you need this, override tmp.mount's
67 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
70 * systemd now supports the "memory" cgroup controller also on
73 * The systemd-cgtop tool now optionally takes a control group path as
74 command line argument. If specified, the control group list shown is
75 limited to subgroups of that group.
77 * The SystemCallFilter= unit file setting gained support for
78 pre-defined, named system call filter sets. For example
79 SystemCallFilter=@clock is now an effective way to make all clock
80 changing-related system calls unavailable to a service. A number of
81 similar pre-defined groups are defined. Writing system call filters
82 for system services is simplified substantially with this new
83 concept. Accordingly, all of systemd's own, long-running services now
84 enable system call filtering based on this, by default.
86 * A new service setting MemoryDenyWriteExecute= has been added, taking
87 a boolean value. If turned on, a service may no longer create memory
88 mappings that are writable and executable at the same time. This
89 enhances security for services where this is enabled as it becomes
90 harder to dynamically write and then execute memory in exploited
91 service processes. This option has been enabled for all of systemd's
92 own long-running services.
94 * A new RestrictRealtime= service setting has been added, taking a
95 boolean argument. If set the service's processes may no longer
96 acquire realtime scheduling. This improves security as realtime
97 scheduling may otherwise be used to easily freeze the system.
99 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
100 value. This may be used for requesting that the system manager inside
101 of the container reports start-up completion to nspawn which then
102 propagates this notification further to the service manager
103 supervising nspawn itself. A related option NotifyReady= in .nspawn
104 files has been added too. This functionality allows ordering of the
105 start-up of multiple containers using the usual systemd ordering
108 * machinectl gained a new command "stop" that is an alias for
111 * systemd-resolved gained support for contacting DNS servers on
112 link-local IPv6 addresses.
114 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
115 its caches. A method call for requesting the same operation has been
116 added to the bus API too, and is made available via "systemd-resolve
119 * systemd-resolve gained a new --status switch. If passed a brief
120 summary of the used DNS configuration with per-interface information
123 * resolved.conf gained a new Cache= boolean option, defaulting to
124 on. If turned off local DNS caching is disabled. This comes with a
125 performance penalty in particular when DNSSEC is enabled. Note that
126 resolved disables its internal caching implicitly anyway, when the
127 configured DNS server is on a host-local IP address such as ::1 or
128 127.0.0.1, thus automatically avoiding double local caching.
130 * systemd-resolved now listens on the local IP address 127.0.0.53:53
131 for DNS requests. This improves compatibility with local programs
132 that do not use the libc NSS or systemd-resolved's bus APIs for name
133 resolution. This minimal DNS service is only available to local
134 programs and does not implement the full DNS protocol, but enough to
135 cover local DNS clients. A new, static resolv.conf file, listing just
136 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
137 now recommended to make /etc/resolv.conf a symlink to this file in
138 order to route all DNS lookups to systemd-resolved, regardless if
139 done via NSS, the bus API or raw DNS packets. Note that this local
140 DNS service is not as fully featured as the libc NSS or
141 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
142 used to deliver link-local address information (as this implies
143 sending a local interface index along), LLMNR/mDNS support via this
144 interface is severely restricted. It is thus strongly recommended for
145 all applications to use the libc NSS API or native systemd-resolved
148 * systemd-networkd's bridge support learned a new setting
149 VLANFiltering= for controlling VLAN filtering. Moreover a new section
150 in .network files has been added for configuring VLAN bridging in
151 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
153 * systemd-networkd's IPv6 Router Advertisement code now makes use of
154 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
155 now be acquired without relying on DHCPv6. Two new options
156 UseDomains= and UseDNS= have been added to configure this behaviour.
158 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
159 renamed IPv6AcceptRA=, without altering its behaviour. The old
160 setting name remains available for compatibility reasons.
162 * The systemd-networkd VTI/VTI6 tunneling support gained new options
163 Key=, InputKey= and OutputKey=.
165 * systemd-networkd gained support for VRF ("Virtual Routing Function")
166 interface configuration.
168 * "systemctl edit" may now be used to create new unit files by
169 specifying the --force switch.
171 * sd-event gained a new function sd_event_get_iteration() for
172 requesting the current iteration counter of the event loop. It starts
173 at zero and is increased by one with each event loop iteration.
175 * A new rpm macro %systemd_ordering is provided by the macros.systemd
176 file. It can be used in lieu of %systemd_requires in packages which
177 don't use any systemd functionality and are intended to be installed
178 in minimal containers without systemd present. This macro provides
179 ordering dependecies to ensure that if the package is installed in
180 the same rpm transaction as systemd, systemd will be installed before
181 the scriptlets for the package are executed, allowing unit presets
184 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
185 been added to simplify packaging of generators.
187 * The os-release file gained VERSION_CODENAME field for the
188 distribution nickname (e.g. VERSION_CODENAME=woody).
190 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
191 can be set to disable parsing of metadata and the creation
192 of persistent symlinks for that device.
194 * The change to tag framebuffer devices (/dev/fb*) with "uaccess"
195 to make them available to logged in users has been reverted.
197 * Configuration for "mkosi" is now part of the systemd
198 repository. mkosi is a tool to easily build legacy-free OS images,
199 and is available on github: https://github.com/systemd/mkosi. If
200 "mkosi" is invoked in the build tree a new raw OS image is generated
201 incorporating the systemd sources currently being worked on and a
202 clean, fresh distribution installation. The generated OS image may be
203 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physcial
204 UEFI PC. This functionality is particularly useful to easily test
205 local changes made to systemd in a pristine, defined environment. See
208 Contributions from: Alessandro Puccetti, Alessio Igor Bogani, Alexander
209 Kuleshov, Alexander Kurtz, Alex Gaynor, Andika Triwidada, Andreas
210 Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar Burchardt, Atrotors,
211 Benjamin Drung, Brian Boylston, Christian Hesse, Christian Rebischke,
212 Daniele Medri, Daniel Mack, Dave Reisner, David Herrmann, David
213 Michael, Djalal Harouni, Douglas Christman, Elias Probst, Evgeny
214 Vereshchagin, Federico Mena Quintero, Felipe Sateler, Franck Bui,
215 Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan Janssen,
216 Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke Witteveen, Kai
217 Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart Poettering, Luca
218 Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel Holtmann, Martin
219 Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov, Michael Biebl,
220 Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz, Michal
221 Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran, Otto
222 Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier, Rusty
223 Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
224 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
225 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
226 Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
228 — Somewhere, 2016-XX-XX
232 * DNSSEC is now turned on by default in systemd-resolved (in
233 "allow-downgrade" mode), but may be turned off during compile time by
234 passing "--with-default-dnssec=no" to "configure" (and of course,
235 during runtime with DNSSEC= in resolved.conf). We recommend
236 downstreams to leave this on at least during development cycles and
237 report any issues with the DNSSEC logic upstream. We are very
238 interested in collecting feedback about the DNSSEC validator and its
239 limitations in the wild. Note however, that DNSSEC support is
240 probably nothing downstreams should turn on in stable distros just
241 yet, as it might create incompatibilities with a few DNS servers and
242 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
243 automatically whenever we detect such incompatible setups, but there
244 might be systems we do not cover yet. Hence: please help us testing
245 the DNSSEC code, leave this on where you can, report back, but then
246 again don't consider turning this on in your stable, LTS or
247 production release just yet. (Note that you have to enable
248 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
249 and its DNSSEC mode for host name resolution from local
252 * systemd-resolve conveniently resolves DANE records with the --tlsa
253 option and OPENPGPKEY records with the --openpgp option. It also
254 supports dumping raw DNS record data via the new --raw= switch.
256 * systemd-logind will now by default terminate user processes that are
257 part of the user session scope unit (session-XX.scope) when the user
258 logs out. This behavior is controlled by the KillUserProcesses=
259 setting in logind.conf, and the previous default of "no" is now
260 changed to "yes". This means that user sessions will be properly
261 cleaned up after, but additional steps are necessary to allow
262 intentionally long-running processes to survive logout.
264 While the user is logged in at least once, user@.service is running,
265 and any service that should survive the end of any individual login
266 session can be started at a user service or scope using systemd-run.
267 systemd-run(1) man page has been extended with an example which shows
268 how to run screen in a scope unit underneath user@.service. The same
269 command works for tmux.
271 After the user logs out of all sessions, user@.service will be
272 terminated too, by default, unless the user has "lingering" enabled.
273 To effectively allow users to run long-term tasks even if they are
274 logged out, lingering must be enabled for them. See loginctl(1) for
275 details. The default polkit policy was modified to allow users to
276 set lingering for themselves without authentication.
278 Previous defaults can be restored at compile time by the
279 --without-kill-user-processes option to "configure".
281 * systemd-logind gained new configuration settings SessionsMax= and
282 InhibitorsMax=, both with a default of 8192. It will not register new
283 user sessions or inhibitors above this limit.
285 * systemd-logind will now reload configuration on SIGHUP.
287 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
288 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
289 enable. Also, support for the "io" cgroup controller in the unified
290 hierarchy has been added, so that the "memory", "pids" and "io" are
291 now the controllers that are supported on the unified hierarchy.
293 WARNING: it is not possible to use previous systemd versions with
294 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
295 is necessary to also update systemd in the initramfs if using the
296 unified hierarchy. An updated SELinux policy is also required.
298 * LLDP support has been extended, and both passive (receive-only) and
299 active (sender) modes are supported. Passive mode ("routers-only") is
300 enabled by default in systemd-networkd. Active LLDP mode is enabled
301 by default for containers on the internal network. The "networkctl
302 lldp" command may be used to list information gathered. "networkctl
303 status" will also show basic LLDP information on connected peers now.
305 * The IAID and DUID unique identifier sent in DHCP requests may now be
306 configured for the system and each .network file managed by
307 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
309 * systemd-networkd gained support for configuring proxy ARP support for
310 each interface, via the ProxyArp= setting in .network files. It also
311 gained support for configuring the multicast querier feature of
312 bridge devices, via the new MulticastQuerier= setting in .netdev
313 files. Similarly, snooping on the IGMP traffic can be controlled
314 via the new setting MulticastSnooping=.
316 A new setting PreferredLifetime= has been added for addresses
317 configured in .network file to configure the lifetime intended for an
320 The systemd-networkd DHCP server gained the option EmitRouter=, which
321 defaults to yes, to configure whether the DHCP Option 3 (Router)
324 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
325 systemd-socket-activate and installed into /usr/bin. It is now fully
328 * systemd-journald now uses separate threads to flush changes to disk
329 when closing journal files, thus reducing impact of slow disk I/O on
332 * The sd-journal API gained two new calls
333 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
334 can be used to open journal files using file descriptors instead of
335 file or directory paths. sd_journal_open_container() has been
336 deprecated, sd_journal_open_directory_fd() should be used instead
337 with the flag SD_JOURNAL_OS_ROOT.
339 * journalctl learned a new output mode "-o short-unix" that outputs log
340 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
341 UTC). It also gained support for a new --no-hostname setting to
342 suppress the hostname column in the family of "short" output modes.
344 * systemd-ask-password now optionally skips printing of the password to
345 stdout with --no-output which can be useful in scripts.
347 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
348 (devices tagged with ID_MAKER_TOOL) are now tagged with
349 "uaccess" and are available to logged in users.
351 * The DeviceAllow= unit setting now supports specifiers (with "%").
353 * "systemctl show" gained a new --value switch, which allows print a
354 only the contents of a specific unit property, without also printing
355 the property's name. Similar support was added to "show*" verbs
356 of loginctl and machinectl that output "key=value" lists.
358 * A new unit type "generated" was added for files dynamically generated
359 by generator tools. Similarly, a new unit type "transient" is used
360 for unit files created using the runtime API. "systemctl enable" will
361 refuse to operate on such files.
363 * A new command "systemctl revert" has been added that may be used to
364 revert to the vendor version of a unit file, in case local changes
365 have been made by adding drop-ins or overriding the unit file.
367 * "machinectl clean" gained a new verb to automatically remove all or
368 just hidden container images.
370 * systemd-tmpfiles gained support for a new line type "e" for emptying
371 directories, if they exist, without creating them if they don't.
373 * systemd-nspawn gained support for automatically patching the UID/GIDs
374 of the owners and the ACLs of all files and directories in a
375 container tree to match the UID/GID user namespacing range selected
376 for the container invocation. This mode is enabled via the new
377 --private-user-chown switch. It also gained support for automatically
378 choosing a free, previously unused UID/GID range when starting a
379 container, via the new --private-users=pick setting (which implies
380 --private-user-chown). Together, these options for the first time
381 make user namespacing for nspawn containers fully automatic and thus
382 deployable. The systemd-nspawn@.service template unit file has been
383 changed to use this functionality by default.
385 * systemd-nspawn gained a new --network-zone= switch, that allows
386 creating ad-hoc virtual Ethernet links between multiple containers,
387 that only exist as long as at least one container referencing them is
388 running. This allows easy connecting of multiple containers with a
389 common link that implements an Ethernet broadcast domain. Each of
390 these network "zones" may be named relatively freely by the user, and
391 may be referenced by any number of containers, but each container may
392 only reference one of these "zones". On the lower level, this is
393 implemented by an automatically managed bridge network interface for
394 each zone, that is created when the first container referencing its
395 zone is created and removed when the last one referencing its zone
398 * The default start timeout may now be configured on the kernel command
399 line via systemd.default_timeout_start_sec=. It was already
400 configurable via the DefaultTimeoutStartSec= option in
401 /etc/systemd/system.conf.
403 * Socket units gained a new TriggerLimitIntervalSec= and
404 TriggerLimitBurst= setting to configure a limit on the activation
405 rate of the socket unit.
407 * The LimitNICE= setting now optionally takes normal UNIX nice values
408 in addition to the raw integer limit value. If the specified
409 parameter is prefixed with "+" or "-" and is in the range -20..19 the
410 value is understood as UNIX nice value. If not prefixed like this it
411 is understood as raw RLIMIT_NICE limit.
413 * Note that the effect of the PrivateDevices= unit file setting changed
414 slightly with this release: the per-device /dev file system will be
415 mounted read-only from this version on, and will have "noexec"
416 set. This (minor) change of behavior might cause some (exceptional)
417 legacy software to break, when PrivateDevices=yes is set for its
418 service. Please leave PrivateDevices= off if you run into problems
421 * systemd-bootchart has been split out to a separate repository:
422 https://github.com/systemd/systemd-bootchart
424 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
425 merged into the kernel in its current form.
427 * The compatibility libraries libsystemd-daemon.so,
428 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
429 which have been deprecated since systemd-209 have been removed along
430 with the corresponding pkg-config files. All symbols provided by
431 those libraries are provided by libsystemd.so.
433 * The Capabilities= unit file setting has been removed (it is ignored
434 for backwards compatibility). AmbientCapabilities= and
435 CapabilityBoundingSet= should be used instead.
437 * A new special target has been added, initrd-root-device.target,
438 which creates a synchronization point for dependencies of the root
439 device in early userspace. Initramfs builders must ensure that this
440 target is now included in early userspace.
442 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
443 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
444 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
445 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
446 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
447 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
448 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
449 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
450 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
451 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
452 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
453 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
454 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
455 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
456 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
457 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
458 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
459 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
460 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
461 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
462 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
463 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
464 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
465 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
466 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
469 — Fairfax, 2016-05-21
473 * The systemd-resolved DNS resolver service has gained a substantial
474 set of new features, most prominently it may now act as a DNSSEC
475 validating stub resolver. DNSSEC mode is currently turned off by
476 default, but is expected to be turned on by default in one of the
477 next releases. For now, we invite everybody to test the DNSSEC logic
478 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
479 service also gained a full set of D-Bus interfaces, including calls
480 to configure DNS and DNSSEC settings per link (for use by external
481 network management software). systemd-resolved and systemd-networkd
482 now distinguish between "search" and "routing" domains. The former
483 are used to qualify single-label names, the latter are used purely
484 for routing lookups within certain domains to specific links.
485 resolved now also synthesizes RRs for all entries from /etc/hosts.
487 * The systemd-resolve tool (which is a client utility for
488 systemd-resolved) has been improved considerably and is now fully
489 supported and documented. Hence it has moved from /usr/lib/systemd to
492 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
495 * The coredump collection logic has been reworked: when a coredump is
496 collected it is now written to disk, compressed and processed
497 (including stacktrace extraction) from a new instantiated service
498 systemd-coredump@.service, instead of directly from the
499 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
500 processing large coredumps can take up a substantial amount of
501 resources and time, and this previously happened entirely outside of
502 systemd's service supervision. With the new logic the core_pattern
503 hook only does minimal metadata collection before passing off control
504 to the new instantiated service, which is configured with a time
505 limit, a nice level and other settings to minimize negative impact on
506 the rest of the system. Also note that the new logic will honour the
507 RLIMIT_CORE setting of the crashed process, which now allows users
508 and processes to turn off coredumping for their processes by setting
511 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
512 and all forked processes by default. Previously, PID 1 would leave
513 the setting at "0" for all processes, as set by the kernel. Note that
514 the resource limit traditionally has no effect on the generated
515 coredumps on the system if the /proc/sys/kernel/core_pattern hook
516 logic is used. Since the limit is now honoured (see above) its
517 default has been changed so that the coredumping logic is enabled by
518 default for all processes, while allowing specific opt-out.
520 * When the stacktrace is extracted from processes of system users, this
521 is now done as "systemd-coredump" user, in order to sandbox this
522 potentially security sensitive parsing operation. (Note that when
523 processing coredumps of normal users this is done under the user ID
524 of process that crashed, as before.) Packagers should take notice
525 that it is now necessary to create the "systemd-coredump" system user
526 and group at package installation time.
528 * The systemd-activate socket activation testing tool gained support
529 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
530 and --seqpacket switches. It also has been extended to support both
531 new-style and inetd-style file descriptor passing. Use the new
532 --inetd switch to request inetd-style file descriptor passing.
534 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
535 variable, which takes a boolean value. If set to false, ANSI color
536 output is disabled in the tools even when run on a terminal that
539 * The VXLAN support in networkd now supports two new settings
540 DestinationPort= and PortRange=.
542 * A new systemd.machine_id= kernel command line switch has been added,
543 that may be used to set the machine ID in /etc/machine-id if it is
544 not initialized yet. This command line option has no effect if the
545 file is already initialized.
547 * systemd-nspawn gained a new --as-pid2 switch that invokes any
548 specified command line as PID 2 rather than PID 1 in the
549 container. In this mode PID 1 is a minimal stub init process that
550 implements the special POSIX and Linux semantics of PID 1 regarding
551 signal and child process management. Note that this stub init process
552 is implemented in nspawn itself and requires no support from the
553 container image. This new logic is useful to support running
554 arbitrary commands in the container, as normal processes are
555 generally not prepared to run as PID 1.
557 * systemd-nspawn gained a new --chdir= switch for setting the current
558 working directory for the process started in the container.
560 * "journalctl /dev/sda" will now output all kernel log messages for
561 specified device from the current boot, in addition to all devices
562 that are parents of it. This should make log output about devices
563 pretty useful, as long as kernel drivers attach enough metadata to
564 the log messages. (The usual SATA drivers do.)
566 * The sd-journal API gained two new calls
567 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
568 that report whether log data from /run or /var has been found.
570 * journalctl gained a new switch "--fields" that prints all journal
571 record field names currently in use in the journal. This is backed
572 by two new sd-journal API calls sd_journal_enumerate_fields() and
573 sd_journal_restart_fields().
575 * Most configurable timeouts in systemd now expect an argument of
576 "infinity" to turn them off, instead of "0" as before. The semantics
577 from now on is that a timeout of "0" means "now", and "infinity"
578 means "never". To maintain backwards compatibility, "0" continues to
579 turn off previously existing timeout settings.
581 * "systemctl reload-or-try-restart" has been renamed to "systemctl
582 try-reload-or-restart" to clarify what it actually does: the "try"
583 logic applies to both reloading and restarting, not just restarting.
584 The old name continues to be accepted for compatibility.
586 * On boot-up, when PID 1 detects that the system clock is behind the
587 release date of the systemd version in use, the clock is now set
588 to the latter. Previously, this was already done in timesyncd, in order
589 to avoid running with clocks set to the various clock epochs such as
590 1902, 1938 or 1970. With this change the logic is now done in PID 1
591 in addition to timesyncd during early boot-up, so that it is enforced
592 before the first process is spawned by systemd. Note that the logic
593 in timesyncd remains, as it is more comprehensive and ensures
594 clock monotonicity by maintaining a persistent timestamp file in
595 /var. Since /var is generally not available in earliest boot or the
596 initrd, this part of the logic remains in timesyncd, and is not done
599 * Support for tweaking details in net_cls.class_id through the
600 NetClass= configuration directive has been removed, as the kernel
601 people have decided to deprecate that controller in cgroup v2.
602 Userspace tools such as nftables are moving over to setting rules
603 that are specific to the full cgroup path of a task, which obsoletes
604 these controllers anyway. The NetClass= directive is kept around for
605 legacy compatibility reasons. For a more in-depth description of the
606 kernel change, please refer to the respective upstream commit:
608 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
610 * A new service setting RuntimeMaxSec= has been added that may be used
611 to specify a maximum runtime for a service. If the timeout is hit, the
612 service is terminated and put into a failure state.
614 * A new service setting AmbientCapabilities= has been added. It allows
615 configuration of additional Linux process capabilities that are
616 passed to the activated processes. This is only available on very
619 * The process resource limit settings in service units may now be used
620 to configure hard and soft limits individually.
622 * The various libsystemd APIs such as sd-bus or sd-event now publicly
623 expose support for gcc's __attribute__((cleanup())) C extension.
624 Specifically, for many object destructor functions alternative
625 versions have been added that have names suffixed with "p" and take a
626 pointer to a pointer to the object to destroy, instead of just a
627 pointer to the object itself. This is useful because these destructor
628 functions may be used directly as parameters to the cleanup
629 construct. Internally, systemd has been a heavy user of this GCC
630 extension for a long time, and with this change similar support is
631 now available to consumers of the library outside of systemd. Note
632 that by using this extension in your sources compatibility with old
633 and strictly ANSI compatible C compilers is lost. However, all gcc or
634 LLVM versions of recent years support this extension.
636 * Timer units gained support for a new setting RandomizedDelaySec= that
637 allows configuring some additional randomized delay to the configured
638 time. This is useful to spread out timer events to avoid load peaks in
639 clusters or larger setups.
641 * Calendar time specifications now support sub-second accuracy.
643 * Socket units now support listening on SCTP and UDP-lite protocol
646 * The sd-event API now comes with a full set of man pages.
648 * Older versions of systemd contained experimental support for
649 compressing journal files and coredumps with the LZ4 compressor that
650 was not compatible with the lz4 binary (due to API limitations of the
651 lz4 library). This support has been removed; only support for files
652 compatible with the lz4 binary remains. This LZ4 logic is now
653 officially supported and no longer considered experimental.
655 * The dkr image import logic has been removed again from importd. dkr's
656 micro-services focus doesn't fit into the machine image focus of
657 importd, and quickly got out of date with the upstream dkr API.
659 * Creation of the /run/lock/lockdev/ directory was dropped from
660 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
661 been available for many years. If you still need this, you need to
662 create your own tmpfiles.d config file with:
664 d /run/lock/lockdev 0775 root lock -
666 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
667 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
668 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
669 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
670 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
671 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
672 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
673 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
674 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
675 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
676 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
677 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
678 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
679 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
680 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
681 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
682 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
683 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
684 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
690 * A number of properties previously only settable in unit
691 files are now also available as properties to set when
692 creating transient units programmatically via the bus, as it
693 is exposed with systemd-run's --property=
694 setting. Specifically, these are: SyslogIdentifier=,
695 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
696 EnvironmentFile=, ReadWriteDirectories=,
697 ReadOnlyDirectories=, InaccessibleDirectories=,
698 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
700 * When creating transient services via the bus API it is now
701 possible to pass in a set of file descriptors to use as
702 STDIN/STDOUT/STDERR for the invoked process.
704 * Slice units may now be created transiently via the bus APIs,
705 similar to the way service and scope units may already be
708 * Wherever systemd expects a calendar timestamp specification
709 (like in journalctl's --since= and --until= switches) UTC
710 timestamps are now supported. Timestamps suffixed with "UTC"
711 are now considered to be in Universal Time Coordinated
712 instead of the local timezone. Also, timestamps may now
713 optionally be specified with sub-second accuracy. Both of
714 these additions also apply to recurring calendar event
715 specification, such as OnCalendar= in timer units.
717 * journalctl gained a new "--sync" switch that asks the
718 journal daemon to write all so far unwritten log messages to
719 disk and sync the files, before returning.
721 * systemd-tmpfiles learned two new line types "q" and "Q" that
722 operate like "v", but also set up a basic btrfs quota
723 hierarchy when used on a btrfs file system with quota
726 * tmpfiles' "v", "q" and "Q" will now create a plain directory
727 instead of a subvolume (even on a btrfs file system) if the
728 root directory is a plain directory, and not a
729 subvolume. This should simplify things with certain chroot()
730 environments which are not aware of the concept of btrfs
733 * systemd-detect-virt gained a new --chroot switch to detect
734 whether execution takes place in a chroot() environment.
736 * CPUAffinity= now takes CPU index ranges in addition to
739 * The various memory-related resource limit settings (such as
740 LimitAS=) now understand the usual K, M, G, ... suffixes to
741 the base of 1024 (IEC). Similar, the time-related resource
742 limit settings understand the usual min, h, day, ...
745 * There's a new system.conf setting DefaultTasksMax= to
746 control the default TasksMax= setting for services and
747 scopes running on the system. (TasksMax= is the primary
748 setting that exposes the "pids" cgroup controller on systemd
749 and was introduced in the previous systemd release.) The
750 setting now defaults to 512, which means services that are
751 not explicitly configured otherwise will only be able to
752 create 512 processes or threads at maximum, from this
753 version on. Note that this means that thread- or
754 process-heavy services might need to be reconfigured to set
755 TasksMax= to a higher value. It is sufficient to set
756 TasksMax= in these specific unit files to a higher value, or
757 even "infinity". Similar, there's now a logind.conf setting
758 UserTasksMax= that defaults to 4096 and limits the total
759 number of processes or tasks each user may own
760 concurrently. nspawn containers also have the TasksMax=
761 value set by default now, to 8192. Note that all of this
762 only has an effect if the "pids" cgroup controller is
763 enabled in the kernel. The general benefit of these changes
764 should be a more robust and safer system, that provides a
765 certain amount of per-service fork() bomb protection.
767 * systemd-nspawn gained the new --network-veth-extra= switch
768 to define additional and arbitrarily-named virtual Ethernet
769 links between the host and the container.
771 * A new service execution setting PassEnvironment= has been
772 added that allows importing select environment variables
773 from PID1's environment block into the environment block of
776 * Timer units gained support for a new RemainAfterElapse=
777 setting which takes a boolean argument. It defaults to on,
778 exposing behaviour unchanged to previous releases. If set to
779 off, timer units are unloaded after they elapsed if they
780 cannot elapse again. This is particularly useful for
781 transient timer units, which shall not stay around longer
782 than until they first elapse.
784 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
785 default now (the kernel default is 16). This is beneficial
786 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
787 allows substantially larger numbers of queued
788 datagrams. This should increase the capability of systemd to
789 parallelize boot-up, as logging and sd_notify() are unlikely
790 to stall execution anymore. If you need to change the value
791 from the new defaults, use the usual sysctl.d/ snippets.
793 * The compression framing format used by the journal or
794 coredump processing has changed to be in line with what the
795 official LZ4 tools generate. LZ4 compression support in
796 systemd was considered unsupported previously, as the format
797 was not compatible with the normal tools. With this release
798 this has changed now, and it is hence safe for downstream
799 distributions to turn it on. While not compressing as well
800 as the XZ, LZ4 is substantially faster, which makes
801 it a good default choice for the compression logic in the
802 journal and in coredump handling.
804 * Any reference to /etc/mtab has been dropped from
805 systemd. The file has been obsolete since a while, but
806 systemd refused to work on systems where it was incorrectly
807 set up (it should be a symlink or non-existent). Please make
808 sure to update to util-linux 2.27.1 or newer in conjunction
809 with this systemd release, which also drops any reference to
810 /etc/mtab. If you maintain a distribution make sure that no
811 software you package still references it, as this is a
812 likely source of bugs. There's also a glibc bug pending,
813 asking for removal of any reference to this obsolete file:
815 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
817 Note that only util-linux versions built with
818 --enable-libmount-force-mountinfo are supported.
820 * Support for the ".snapshot" unit type has been removed. This
821 feature turned out to be little useful and little used, and
822 has now been removed from the core and from systemctl.
824 * The dependency types RequiresOverridable= and
825 RequisiteOverridable= have been removed from systemd. They
826 have been used only very sparingly to our knowledge and
827 other options that provide a similar effect (such as
828 systemctl --mode=ignore-dependencies) are much more useful
829 and commonly used. Moreover, they were only half-way
830 implemented as the option to control behaviour regarding
831 these dependencies was never added to systemctl. By removing
832 these dependency types the execution engine becomes a bit
833 simpler. Unit files that use these dependencies should be
834 changed to use the non-Overridable dependency types
835 instead. In fact, when parsing unit files with these
836 options, that's what systemd will automatically convert them
837 too, but it will also warn, asking users to fix the unit
838 files accordingly. Removal of these dependency types should
839 only affect a negligible number of unit files in the wild.
841 * Behaviour of networkd's IPForward= option changed
842 (again). It will no longer maintain a per-interface setting,
843 but propagate one way from interfaces where this is enabled
844 to the global kernel setting. The global setting will be
845 enabled when requested by a network that is set up, but
846 never be disabled again. This change was made to make sure
847 IPv4 and IPv6 behaviour regarding packet forwarding is
848 similar (as the Linux IPv6 stack does not support
849 per-interface control of this setting) and to minimize
852 * In unit files the behaviour of %u, %U, %h, %s has
853 changed. These specifiers will now unconditionally resolve
854 to the various user database fields of the user that the
855 systemd instance is running as, instead of the user
856 configured in the specific unit via User=. Note that this
857 effectively doesn't change much, as resolving of these
858 specifiers was already turned off in the --system instance
859 of systemd, as we cannot do NSS lookups from PID 1. In the
860 --user instance of systemd these specifiers where correctly
861 resolved, but hardly made any sense, since the user instance
862 lacks privileges to do user switches anyway, and User= is
863 hence useless. Morever, even in the --user instance of
864 systemd behaviour was awkward as it would only take settings
865 from User= assignment placed before the specifier into
866 account. In order to unify and simplify the logic around
867 this the specifiers will now always resolve to the
868 credentials of the user invoking the manager (which in case
869 of PID 1 is the root user).
871 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
872 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
873 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
874 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
875 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
876 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
877 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
878 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
879 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
880 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
881 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
882 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
883 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
884 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
891 * systemd now depends on util-linux v2.27. More specifically,
892 the newly added mount monitor feature in libmount now
893 replaces systemd's former own implementation.
895 * libmount mandates /etc/mtab not to be regular file, and
896 systemd now enforces this condition at early boot.
897 /etc/mtab has been deprecated and warned about for a very
898 long time, so systems running systemd should already have
899 stopped having this file around as anything else than a
900 symlink to /proc/self/mounts.
902 * Support for the "pids" cgroup controller has been added. It
903 allows accounting the number of tasks in a cgroup and
904 enforcing limits on it. This adds two new setting
905 TasksAccounting= and TasksMax= to each unit, as well as a
906 global option DefaultTasksAccounting=.
908 * Support for the "net_cls" cgroup controller has been added.
909 It allows assigning a net class ID to each task in the
910 cgroup, which can then be used in firewall rules and traffic
911 shaping configurations. Note that the kernel netfilter net
912 class code does not currently work reliably for ingress
913 packets on unestablished sockets.
915 This adds a new config directive called NetClass= to CGroup
916 enabled units. Allowed values are positive numbers for fixed
917 assignments and "auto" for picking a free value
920 * 'systemctl is-system-running' now returns 'offline' if the
921 system is not booted with systemd. This command can now be
922 used as a substitute for 'systemd-notify --booted'.
924 * Watchdog timeouts have been increased to 3 minutes for all
925 in-tree service files. Apparently, disk IO issues are more
926 frequent than we hoped, and user reported >1 minute waiting
929 * 'machine-id-commit' functionality has been merged into
930 'machine-id-setup --commit'. The separate binary has been
933 * The WorkingDirectory= directive in unit files may now be set
934 to the special value '~'. In this case, the working
935 directory is set to the home directory of the user
938 * "machinectl shell" will now open the shell in the home
939 directory of the selected user by default.
941 * The CrashChVT= configuration file setting is renamed to
942 CrashChangeVT=, following our usual logic of not
943 abbreviating unnecessarily. The old directive is still
944 supported for compat reasons. Also, this directive now takes
945 an integer value between 1 and 63, or a boolean value. The
946 formerly supported '-1' value for disabling stays around for
949 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
950 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
951 RootDirectory= properties can now be set for transient
954 * The systemd-analyze tool gained a new "set-log-target" verb
955 to change the logging target the system manager logs to
956 dynamically during runtime. This is similar to how
957 "systemd-analyze set-log-level" already changes the log
960 * In nspawn /sys is now mounted as tmpfs, with only a selected
961 set of subdirectories mounted in from the real sysfs. This
962 enhances security slightly, and is useful for ensuring user
963 namespaces work correctly.
965 * Support for USB FunctionFS activation has been added. This
966 allows implementation of USB gadget services that are
967 activated as soon as they are requested, so that they don't
968 have to run continuously, similar to classic socket
971 * The "systemctl exit" command now optionally takes an
972 additional parameter that sets the exit code to return from
973 the systemd manager when exiting. This is only relevant when
974 running the systemd user instance, or when running the
975 system instance in a container.
977 * sd-bus gained the new API calls sd_bus_path_encode_many()
978 and sd_bus_path_decode_many() that allow easy encoding and
979 decoding of multiple identifier strings inside a D-Bus
980 object path. Another new call sd_bus_default_flush_close()
981 has been added to flush and close per-thread default
984 * systemd-cgtop gained support for a -M/--machine= switch to
985 show the control groups within a certain container only.
987 * "systemctl kill" gained support for an optional --fail
988 switch. If specified the requested operation will fail of no
989 processes have been killed, because the unit had no
990 processes attached, or similar.
992 * A new systemd.crash_reboot=1 kernel command line option has
993 been added that triggers a reboot after crashing. This can
994 also be set through CrashReboot= in systemd.conf.
996 * The RuntimeDirectory= setting now understands unit
997 specifiers like %i or %f.
999 * A new (still internal) libary API sd-ipv4acd has been added,
1000 that implements address conflict detection for IPv4. It's
1001 based on code from sd-ipv4ll, and will be useful for
1002 detecting DHCP address conflicts.
1004 * File descriptors passed during socket activation may now be
1005 named. A new API sd_listen_fds_with_names() is added to
1006 access the names. The default names may be overridden,
1007 either in the .socket file using the FileDescriptorName=
1008 parameter, or by passing FDNAME= when storing the file
1009 descriptors using sd_notify().
1011 * systemd-networkd gained support for:
1013 - Setting the IPv6 Router Advertisement settings via
1014 IPv6AcceptRouterAdvertisements= in .network files.
1016 - Configuring the HelloTimeSec=, MaxAgeSec= and
1017 ForwardDelaySec= bridge parameters in .netdev files.
1019 - Configuring PreferredSource= for static routes in
1022 * The "ask-password" framework used to query for LUKS harddisk
1023 passwords or SSL passwords during boot gained support for
1024 caching passwords in the kernel keyring, if it is
1025 available. This makes sure that the user only has to type in
1026 a passphrase once if there are multiple objects to unlock
1027 with the same one. Previously, such password caching was
1028 available only when Plymouth was used; this moves the
1029 caching logic into the systemd codebase itself. The
1030 "systemd-ask-password" utility gained a new --keyname=
1031 switch to control which kernel keyring key to use for
1032 caching a password in. This functionality is also useful for
1033 enabling display managers such as gdm to automatically
1034 unlock the user's GNOME keyring if its passphrase, the
1035 user's password and the harddisk password are the same, if
1036 gdm-autologin is used.
1038 * When downloading tar or raw images using "machinectl
1039 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
1040 file is now also downloaded, if it is available and stored
1041 next to the image file.
1043 * Units of type ".socket" gained a new boolean setting
1044 Writable= which is only useful in conjunction with
1045 ListenSpecial=. If true, enables opening the specified
1046 special file in O_RDWR mode rather than O_RDONLY mode.
1048 * systemd-rfkill has been reworked to become a singleton
1049 service that is activated through /dev/rfkill on each rfkill
1050 state change and saves the settings to disk. This way,
1051 systemd-rfkill is now compatible with devices that exist
1052 only intermittendly, and even restores state if the previous
1053 system shutdown was abrupt rather than clean.
1055 * The journal daemon gained support for vacuuming old journal
1056 files controlled by the number of files that shall remain,
1057 in addition to the already existing control by size and by
1058 date. This is useful as journal interleaving performance
1059 degrades with too many separate journal files, and allows
1060 putting an effective limit on them. The new setting defaults
1061 to 100, but this may be changed by setting SystemMaxFiles=
1062 and RuntimeMaxFiles= in journald.conf. Also, the
1063 "journalctl" tool gained the new --vacuum-files= switch to
1064 manually vacuum journal files to leave only the specified
1065 number of files in place.
1067 * udev will now create /dev/disk/by-path links for ATA devices
1068 on kernels where that is supported.
1070 * Galician, Serbian, Turkish and Korean translations were added.
1072 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
1073 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
1074 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
1075 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
1076 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
1077 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
1078 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
1079 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
1080 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
1081 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
1082 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
1083 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
1084 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
1085 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
1086 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
1087 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
1088 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
1089 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
1091 — Berlin, 2015-10-07
1095 * The DHCP implementation of systemd-networkd gained a set of
1098 - The DHCP server now supports emitting DNS and NTP
1099 information. It may be enabled and configured via
1100 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
1101 and NTP information is enabled, but no servers are
1102 configured, the corresponding uplink information (if there
1103 is any) is propagated.
1105 - Server and client now support transmission and reception
1106 of timezone information. It can be configured via the
1107 newly introduced network options UseTimezone=,
1108 EmitTimezone=, and Timezone=. Transmission of timezone
1109 information is enabled between host and containers by
1110 default now: the container will change its local timezone
1111 to what the host has set.
1113 - Lease timeouts can now be configured via
1114 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
1116 - The DHCP server improved on the stability of
1117 leases. Clients are more likely to get the same lease
1118 information back, even if the server loses state.
1120 - The DHCP server supports two new configuration options to
1121 control the lease address pool metrics, PoolOffset= and
1124 * The encapsulation limit of tunnels in systemd-networkd may
1125 now be configured via 'EncapsulationLimit='. It allows
1126 modifying the maximum additional levels of encapsulation
1127 that are permitted to be prepended to a packet.
1129 * systemd now supports the concept of user buses replacing
1130 session buses, if used with dbus-1.10 (and enabled via dbus
1131 --enable-user-session). It previously only supported this on
1132 kdbus-enabled systems, and this release expands this to
1133 'dbus-daemon' systems.
1135 * systemd-networkd now supports predictable interface names
1138 * systemd now optionally supports the new Linux kernel
1139 "unified" control group hierarchy. If enabled via the kernel
1140 command-line option 'systemd.unified_cgroup_hierarchy=1',
1141 systemd will try to mount the unified cgroup hierarchy
1142 directly on /sys/fs/cgroup. If not enabled, or not
1143 available, systemd will fall back to the legacy cgroup
1144 hierarchy setup, as before. Host system and containers can
1145 mix and match legacy and unified hierarchies as they
1146 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
1147 environment variable to individually select the hierarchy to
1148 use for executed containers. By default, nspawn will use the
1149 unified hierarchy for the containers if the host uses the
1150 unified hierarchy, and the legacy hierarchy otherwise.
1151 Please note that at this point the unified hierarchy is an
1152 experimental kernel feature and is likely to change in one
1153 of the next kernel releases. Therefore, it should not be
1154 enabled by default in downstream distributions yet. The
1155 minimum required kernel version for the unified hierarchy to
1156 work is 4.2. Note that when the unified hierarchy is used
1157 for the first time delegated access to controllers is
1158 safe. Because of this systemd-nspawn containers will get
1159 access to controllers now, as will systemd user
1160 sessions. This means containers and user sessions may now
1161 manage their own resources, partitioning up what the system
1164 * A new special scope unit "init.scope" has been introduced
1165 that encapsulates PID 1 of the system. It may be used to
1166 determine resource usage and enforce resource limits on PID
1167 1 itself. PID 1 hence moved out of the root of the control
1170 * The cgtop tool gained support for filtering out kernel
1171 threads when counting tasks in a control group. Also, the
1172 count of processes is now recursively summed up by
1173 default. Two options -k and --recursive= have been added to
1174 revert to old behaviour. The tool has also been updated to
1175 work correctly in containers now.
1177 * systemd-nspawn's --bind= and --bind-ro= options have been
1178 extended to allow creation of non-recursive bind mounts.
1180 * libsystemd gained two new calls sd_pid_get_cgroup() and
1181 sd_peer_get_cgroup() which return the control group path of
1182 a process or peer of a connected AF_UNIX socket. This
1183 function call is particularly useful when implementing
1184 delegated subtrees support in the control group hierarchy.
1186 * The "sd-event" event loop API of libsystemd now supports
1187 correct dequeuing of real-time signals, without losing
1190 * When systemd requests a PolicyKit decision when managing
1191 units it will now add additional fields to the request,
1192 including unit name and desired operation. This enables more
1193 powerful PolicyKit policies, that make decisions depending
1194 on these parameters.
1196 * nspawn learnt support for .nspawn settings files, that may
1197 accompany the image files or directories of containers, and
1198 may contain additional settings for the container. This is
1199 an alternative to configuring container parameters via the
1200 nspawn command line.
1202 Contributions from: Cristian Rodríguez, Daniel Mack, David
1203 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
1204 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
1205 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
1206 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
1207 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
1208 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
1209 Andersen, Tom Gundersen, Torstein Husebø
1211 — Berlin, 2015-09-08
1215 * machinectl gained a new verb 'shell' which opens a fresh
1216 shell on the target container or the host. It is similar to
1217 the existing 'login' command of machinectl, but spawns the
1218 shell directly without prompting for username or
1219 password. The pseudo machine '.host' now refers to the local
1220 host and is used by default. Hence, 'machinectl shell' can
1221 be used as replacement for 'su -' which spawns a session as
1222 a fresh systemd unit in a way that is fully isolated from
1223 the originating session.
1225 * systemd-networkd learned to cope with private-zone DHCP
1226 options and allows other programs to query the values.
1228 * SELinux access control when enabling/disabling units is no
1229 longer enforced with this release. The previous
1230 implementation was incorrect, and a new corrected
1231 implementation is not yet available. As unit file operations
1232 are still protected via PolicyKit and D-Bus policy this is
1233 not a security problem. Yet, distributions which care about
1234 optimal SELinux support should probably not stabilize on
1237 * sd-bus gained support for matches of type "arg0has=", that
1238 test for membership of strings in string arrays sent in bus
1241 * systemd-resolved now dumps the contents of its DNS and LLMNR
1242 caches to the logs on reception of the SIGUSR1 signal. This
1243 is useful to debug DNS behaviour.
1245 * The coredumpctl tool gained a new --directory= option to
1246 operate on journal files in a specific directory.
1248 * "systemctl reboot" and related commands gained a new
1249 "--message=" option which may be used to set a free-text
1250 wall message when shutting down or rebooting the
1251 system. This message is also logged, which is useful for
1252 figuring out the reason for a reboot or shutdown a
1255 * The "systemd-resolve-host" tool's -i switch now takes
1256 network interface numbers as alternative to interface names.
1258 * A new unit file setting for services has been introduced:
1259 UtmpMode= allows configuration of how precisely systemd
1260 handles utmp and wtmp entries for the service if this is
1261 enabled. This allows writing services that appear similar to
1262 user sessions in the output of the "w", "who", "last" and
1265 * systemd-resolved will now locally synthesize DNS resource
1266 records for the "localhost" and "gateway" domains as well as
1267 the local hostname. This should ensure that clients querying
1268 RRs via resolved will get similar results as those going via
1269 NSS, if nss-myhostname is enabled.
1271 Contributions from: Alastair Hughes, Alex Crawford, Daniel
1272 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
1273 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
1274 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
1275 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
1276 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
1277 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
1278 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
1279 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
1280 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
1281 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
1282 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
1284 — Berlin, 2015-08-27
1288 * The systemd-efi-boot-generator functionality was merged into
1289 systemd-gpt-auto-generator.
1291 * systemd-networkd now supports Group Policy for vxlan
1292 devices. It can be enabled via the new boolean configuration
1293 option called 'GroupPolicyExtension='.
1295 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
1296 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
1297 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
1299 — Berlin, 2015-07-31
1303 * The python-systemd code has been removed from the systemd repository.
1304 A new repository has been created which accommodates the code from
1305 now on, and we kindly ask distributions to create a separate package
1306 for this: https://github.com/systemd/python-systemd
1308 * The systemd daemon will now reload its main configuration
1309 (/etc/systemd/system.conf) on daemon-reload.
1311 * sd-dhcp now exposes vendor specific extensions via
1312 sd_dhcp_lease_get_vendor_specific().
1314 * systemd-networkd gained a number of new configuration options.
1316 - A new boolean configuration option for TAP devices called
1317 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
1318 device, thus allowing to send and receive GSO packets.
1320 - A new tunnel configuration option called 'CopyDSCP='.
1321 If enabled, the DSCP field of ip6 tunnels is copied into the
1322 decapsulated packet.
1324 - A set of boolean bridge configuration options were added.
1325 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
1326 and 'UnicastFlood=' are now parsed by networkd and applied to the
1327 respective bridge link device via the respective IFLA_BRPORT_*
1330 - A new string configuration option to override the hostname sent
1331 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
1332 is true, networkd will use the configured hostname instead of the
1333 system hostname when sending DHCP requests.
1335 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
1336 networkd will configure the IPv6 flow-label of the tunnel device
1337 according to RFC2460.
1339 - The 'macvtap' virtual network devices are now supported, similar to
1340 the already supported 'macvlan' devices.
1342 * systemd-resolved now implements RFC5452 to improve resilience against
1343 cache poisoning. Additionally, source port randomization is enabled
1344 by default to further protect against DNS spoofing attacks.
1346 * nss-mymachines now supports translating UIDs and GIDs of running
1347 containers with user-namespaces enabled. If a container 'foo'
1348 translates a host uid 'UID' to the container uid 'TUID', then
1349 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
1350 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
1351 mapped as 'vg-foo-TGID'.
1353 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
1354 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
1355 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
1356 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
1357 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
1358 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
1359 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
1360 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
1361 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
1362 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
1364 — Berlin, 2015-07-29
1368 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
1369 There are no known issues with current sysfs, and udev does not need
1370 or should be used to work around such bugs.
1372 * udev does no longer enable USB HID power management. Several reports
1373 indicate, that some devices cannot handle that setting.
1375 * The udev accelerometer helper was removed. The functionality
1376 is now fully included in iio-sensor-proxy. But this means,
1377 older iio-sensor-proxy versions will no longer provide
1378 accelerometer/orientation data with this systemd version.
1379 Please upgrade iio-sensor-proxy to version 1.0.
1381 * networkd gained a new configuration option IPv6PrivacyExtensions=
1382 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
1383 for Stateless Address") on selected networks.
1385 * For the sake of fewer build-time dependencies and less code in the
1386 main repository, the python bindings are about to be removed in the
1387 next release. A new repository has been created which accommodates
1388 the code from now on, and we kindly ask distributions to create a
1389 separate package for this. The removal will take place in v223.
1391 https://github.com/systemd/python-systemd
1393 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
1394 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
1395 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
1396 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
1397 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
1398 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
1399 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
1400 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
1401 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
1402 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
1404 — Berlin, 2015-07-07
1408 * The sd-bus.h and sd-event.h APIs have now been declared
1409 stable and have been added to the official interface of
1410 libsystemd.so. sd-bus implements an alternative D-Bus client
1411 library, that is relatively easy to use, very efficient and
1412 supports both classic D-Bus as well as kdbus as transport
1413 backend. sd-event is a generic event loop abstraction that
1414 is built around Linux epoll, but adds features such as event
1415 prioritization or efficient timer handling. Both APIs are good
1416 choices for C programs looking for a bus and/or event loop
1417 implementation that is minimal and does not have to be
1418 portable to other kernels.
1420 * kdbus support is no longer compile-time optional. It is now
1421 always built-in. However, it can still be disabled at
1422 runtime using the kdbus=0 kernel command line setting, and
1423 that setting may be changed to default to off, by specifying
1424 --disable-kdbus at build-time. Note though that the kernel
1425 command line setting has no effect if the kdbus.ko kernel
1426 module is not installed, in which case kdbus is (obviously)
1427 also disabled. We encourage all downstream distributions to
1428 begin testing kdbus by adding it to the kernel images in the
1429 development distributions, and leaving kdbus support in
1432 * The minimal required util-linux version has been bumped to
1435 * Support for chkconfig (--enable-chkconfig) was removed in
1436 favor of calling an abstraction tool
1437 /lib/systemd/systemd-sysv-install. This needs to be
1438 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
1439 in README for details.
1441 * If there's a systemd unit and a SysV init script for the
1442 same service name, and the user executes "systemctl enable"
1443 for it (or a related call), then this will now enable both
1444 (or execute the related operation on both), not just the
1447 * The libudev API documentation has been converted from gtkdoc
1450 * gudev has been removed from the systemd tree, it is now an
1453 * The systemd-cgtop tool learnt a new --raw switch to generate
1454 "raw" (machine parsable) output.
1456 * networkd's IPForwarding= .network file setting learnt the
1457 new setting "kernel", which ensures that networkd does not
1458 change the IP forwarding sysctl from the default kernel
1461 * The systemd-logind bus API now exposes a new boolean
1462 property "Docked" that reports whether logind considers the
1463 system "docked", i.e. connected to a docking station or not.
1465 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
1466 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
1467 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
1468 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
1469 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
1470 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
1471 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
1472 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
1473 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
1474 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
1475 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
1476 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
1477 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
1478 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
1479 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
1480 Fink, Zbigniew Jędrzejewski-Szmek
1482 — Berlin, 2015-06-19
1486 * The gudev library has been extracted into a separate repository
1487 available at: https://git.gnome.org/browse/libgudev/
1488 It is now managed as part of the Gnome project. Distributions
1489 are recommended to pass --disable-gudev to systemd and use
1490 gudev from the Gnome project instead. gudev is still included
1491 in systemd, for now. It will be removed soon, though. Please
1492 also see the announcement-thread on systemd-devel:
1493 http://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
1495 * systemd now exposes a CPUUsageNSec= property for each
1496 service unit on the bus, that contains the overall consumed
1497 CPU time of a service (the sum of what each process of the
1498 service consumed). This value is only available if
1499 CPUAccounting= is turned on for a service, and is then shown
1500 in the "systemctl status" output.
1502 * Support for configuring alternative mappings of the old SysV
1503 runlevels to systemd targets has been removed. They are now
1504 hardcoded in a way that runlevels 2, 3, 4 all map to
1505 multi-user.target and 5 to graphical.target (which
1506 previously was already the default behaviour).
1508 * The auto-mounter logic gained support for mount point
1509 expiry, using a new TimeoutIdleSec= setting in .automount
1510 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
1512 * The EFI System Partition (ESP) as mounted to /boot by
1513 systemd-efi-boot-generator will now be unmounted
1514 automatically after 2 minutes of not being used. This should
1515 minimize the risk of ESP corruptions.
1517 * New /etc/fstab options x-systemd.requires= and
1518 x-systemd.requires-mounts-for= are now supported to express
1519 additional dependencies for mounts. This is useful for
1520 journalling file systems that support external journal
1521 devices or overlay file systems that require underlying file
1522 systems to be mounted.
1524 * systemd does not support direct live-upgrades (via systemctl
1525 daemon-reexec) from versions older than v44 anymore. As no
1526 distribution we are aware of shipped such old versions in a
1527 stable release this should not be problematic.
1529 * When systemd forks off a new per-connection service instance
1530 it will now set the $REMOTE_ADDR environment variable to the
1531 remote IP address, and $REMOTE_PORT environment variable to
1532 the remote IP port. This behaviour is similar to the
1533 corresponding environment variables defined by CGI.
1535 * systemd-networkd gained support for uplink failure
1536 detection. The BindCarrier= option allows binding interface
1537 configuration dynamically to the link sense of other
1538 interfaces. This is useful to achieve behaviour like in
1541 * systemd-networkd gained support for configuring the DHCP
1542 client identifier to use when requesting leases.
1544 * systemd-networkd now has a per-network UseNTP= option to
1545 configure whether NTP server information acquired via DHCP
1546 is passed on to services like systemd-timesyncd.
1548 * systemd-networkd gained support for vti6 tunnels.
1550 * Note that systemd-networkd manages the sysctl variable
1551 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
1552 it is configured for since v219. The variable controls IP
1553 forwarding, and is a per-interface alternative to the global
1554 /proc/sys/net/ipv[46]/ip_forward. This setting is
1555 configurable in the IPForward= option, which defaults to
1556 "no". This means if networkd is used for an interface it is
1557 no longer sufficient to set the global sysctl option to turn
1558 on IP forwarding! Instead, the .network file option
1559 IPForward= needs to be turned on! Note that the
1560 implementation of this behaviour was broken in v219 and has
1563 * Many bonding and vxlan options are now configurable in
1566 * systemd-nspawn gained a new --property= setting to set unit
1567 properties for the container scope. This is useful for
1568 setting resource parameters (e.g "CPUShares=500") on
1569 containers started from the command line.
1571 * systemd-nspawn gained a new --private-users= switch to make
1572 use of user namespacing available on recent Linux kernels.
1574 * systemd-nspawn may now be called as part of a shell pipeline
1575 in which case the pipes used for stdin and stdout are passed
1576 directly to the process invoked in the container, without
1577 indirection via a pseudo tty.
1579 * systemd-nspawn gained a new switch to control the UNIX
1580 signal to use when killing the init process of the container
1583 * systemd-nspawn gained a new --overlay= switch for mounting
1584 overlay file systems into the container using the new kernel
1587 * When a container image is imported via systemd-importd and
1588 the host file system is not btrfs, a loopback block device
1589 file is created in /var/lib/machines.raw with a btrfs file
1590 system inside. It is then mounted to /var/lib/machines to
1591 enable btrfs features for container management. The loopback
1592 file and btrfs file system is grown as needed when container
1593 images are imported via systemd-importd.
1595 * systemd-machined/systemd-importd gained support for btrfs
1596 quota, to enforce container disk space limits on disk. This
1597 is exposed in "machinectl set-limit".
1599 * systemd-importd now can import containers from local .tar,
1600 .raw and .qcow2 images, and export them to .tar and .raw. It
1601 can also import dkr v2 images now from the network (on top
1604 * systemd-importd gained support for verifying downloaded
1605 images with gpg2 (previously only gpg1 was supported).
1607 * systemd-machined, systemd-logind, systemd: most bus calls
1608 are now accessible to unprivileged processes via
1609 PolicyKit. Also, systemd-logind will now allow users to kill
1610 their own sessions without further privileges or
1613 * systemd-shutdownd has been removed. This service was
1614 previously responsible for implementing scheduled shutdowns
1615 as exposed in /usr/bin/shutdown's time parameter. This
1616 functionality has now been moved into systemd-logind and is
1617 accessible via a bus interface.
1619 * "systemctl reboot" gained a new switch --firmware-setup that
1620 can be used to reboot into the EFI firmware setup, if that
1621 is available. systemd-logind now exposes an API on the bus
1622 to trigger such reboots, in case graphical desktop UIs want
1623 to cover this functionality.
1625 * "systemctl enable", "systemctl disable" and "systemctl mask"
1626 now support a new "--now" switch. If specified the units
1627 that are enabled will also be started, and the ones
1628 disabled/masked also stopped.
1630 * The Gummiboot EFI boot loader tool has been merged into
1631 systemd, and renamed to "systemd-boot". The bootctl tool has been
1632 updated to support systemd-boot.
1634 * An EFI kernel stub has been added that may be used to create
1635 kernel EFI binaries that contain not only the actual kernel,
1636 but also an initrd, boot splash, command line and OS release
1637 information. This combined binary can then be signed as a
1638 single image, so that the firmware can verify it all in one
1639 step. systemd-boot has special support for EFI binaries created
1640 like this and can extract OS release information from them
1641 and show them in the boot menu. This functionality is useful
1642 to implement cryptographically verified boot schemes.
1644 * Optional support has been added to systemd-fsck to pass
1645 fsck's progress report to an AF_UNIX socket in the file
1648 * udev will no longer create device symlinks for all block
1649 devices by default. A blacklist for excluding special block
1650 devices from this logic has been turned into a whitelist
1651 that requires picking block devices explicitly that require
1654 * A new (currently still internal) API sd-device.h has been
1655 added to libsystemd. This modernized API is supposed to
1656 replace libudev eventually. In fact, already much of libudev
1657 is now just a wrapper around sd-device.h.
1659 * A new hwdb database for storing metadata about pointing
1660 stick devices has been added.
1662 * systemd-tmpfiles gained support for setting file attributes
1663 similar to the "chattr" tool with new 'h' and 'H' lines.
1665 * systemd-journald will no longer unconditionally set the
1666 btrfs NOCOW flag on new journal files. This is instead done
1667 with tmpfiles snippet using the new 'h' line type. This
1668 allows easy disabling of this logic, by masking the
1669 journal-nocow.conf tmpfiles file.
1671 * systemd-journald will now translate audit message types to
1672 human readable identifiers when writing them to the
1673 journal. This should improve readability of audit messages.
1675 * The LUKS logic gained support for the offset= and skip=
1676 options in /etc/crypttab, as previously implemented by
1679 * /usr/lib/os-release gained a new optional field VARIANT= for
1680 distributions that support multiple variants (such as a
1681 desktop edition, a server edition, ...)
1683 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
1684 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
1685 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
1686 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
1687 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
1688 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
1689 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
1690 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
1691 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
1692 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
1693 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
1694 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
1695 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
1696 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
1697 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
1698 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
1699 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
1700 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
1701 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
1702 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
1703 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
1704 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
1705 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
1706 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
1707 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
1708 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
1709 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
1711 — Berlin, 2015-05-22
1715 * Introduce a new API "sd-hwdb.h" for querying the hardware
1716 metadata database. With this minimal interface one can query
1717 and enumerate the udev hwdb, decoupled from the old libudev
1718 library. libudev's interface for this is now only a wrapper
1719 around sd-hwdb. A new tool systemd-hwdb has been added to
1720 interface with and update the database.
1722 * When any of systemd's tools copies files (for example due to
1723 tmpfiles' C lines) a btrfs reflink will attempted first,
1724 before bytewise copying is done.
1726 * systemd-nspawn gained a new --ephemeral switch. When
1727 specified a btrfs snapshot is taken of the container's root
1728 directory, and immediately removed when the container
1729 terminates again. Thus, a container can be started whose
1730 changes never alter the container's root directory, and are
1731 lost on container termination. This switch can also be used
1732 for starting a container off the root file system of the
1733 host without affecting the host OS. This switch is only
1734 available on btrfs file systems.
1736 * systemd-nspawn gained a new --template= switch. It takes the
1737 path to a container tree to use as template for the tree
1738 specified via --directory=, should that directory be
1739 missing. This allows instantiating containers dynamically,
1740 on first run. This switch is only available on btrfs file
1743 * When a .mount unit refers to a mount point on which multiple
1744 mounts are stacked, and the .mount unit is stopped all of
1745 the stacked mount points will now be unmounted until no
1746 mount point remains.
1748 * systemd now has an explicit notion of supported and
1749 unsupported unit types. Jobs enqueued for unsupported unit
1750 types will now fail with an "unsupported" error code. More
1751 specifically .swap, .automount and .device units are not
1752 supported in containers, .busname units are not supported on
1753 non-kdbus systems. .swap and .automount are also not
1754 supported if their respective kernel compile time options
1757 * machinectl gained support for two new "copy-from" and
1758 "copy-to" commands for copying files from a running
1759 container to the host or vice versa.
1761 * machinectl gained support for a new "bind" command to bind
1762 mount host directories into local containers. This is
1763 currently only supported for nspawn containers.
1765 * networkd gained support for configuring bridge forwarding
1766 database entries (fdb) from .network files.
1768 * A new tiny daemon "systemd-importd" has been added that can
1769 download container images in tar, raw, qcow2 or dkr formats,
1770 and make them available locally in /var/lib/machines, so
1771 that they can run as nspawn containers. The daemon can GPG
1772 verify the downloads (not supported for dkr, since it has no
1773 provisions for verifying downloads). It will transparently
1774 decompress bz2, xz, gzip compressed downloads if necessary,
1775 and restore sparse files on disk. The daemon uses privilege
1776 separation to ensure the actual download logic runs with
1777 fewer privileges than the daemon itself. machinectl has
1778 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
1779 make the functionality of importd available to the
1780 user. With this in place the Fedora and Ubuntu "Cloud"
1781 images can be downloaded and booted as containers unmodified
1782 (the Fedora images lack the appropriate GPG signature files
1783 currently, so they cannot be verified, but this will change
1784 soon, hopefully). Note that downloading images is currently
1785 only fully supported on btrfs.
1787 * machinectl is now able to list container images found in
1788 /var/lib/machines, along with some metadata about sizes of
1789 disk and similar. If the directory is located on btrfs and
1790 quota is enabled, this includes quota display. A new command
1791 "image-status" has been added that shows additional
1792 information about images.
1794 * machinectl is now able to clone container images
1795 efficiently, if the underlying file system (btrfs) supports
1796 it, with the new "machinectl clone" command. It also
1797 gained commands for renaming and removing images, as well as
1798 marking them read-only or read-write (supported also on
1799 legacy file systems).
1801 * networkd gained support for collecting LLDP network
1802 announcements, from hardware that supports this. This is
1803 shown in networkctl output.
1805 * systemd-run gained support for a new -t (--pty) switch for
1806 invoking a binary on a pty whose input and output is
1807 connected to the invoking terminal. This allows executing
1808 processes as system services while interactively
1809 communicating with them via the terminal. Most interestingly
1810 this is supported across container boundaries. Invoking
1811 "systemd-run -t /bin/bash" is an alternative to running a
1812 full login session, the difference being that the former
1813 will not register a session, nor go through the PAM session
1816 * tmpfiles gained support for a new "v" line type for creating
1817 btrfs subvolumes. If the underlying file system is a legacy
1818 file system, this automatically degrades to creating a
1819 normal directory. Among others /var/lib/machines is now
1820 created like this at boot, should it be missing.
1822 * The directory /var/lib/containers/ has been deprecated and
1823 been replaced by /var/lib/machines. The term "machines" has
1824 been used in the systemd context as generic term for both
1825 VMs and containers, and hence appears more appropriate for
1826 this, as the directory can also contain raw images bootable
1829 * systemd-nspawn when invoked with -M but without --directory=
1830 or --image= is now capable of searching for the container
1831 root directory, subvolume or disk image automatically, in
1832 /var/lib/machines. systemd-nspawn@.service has been updated
1833 to make use of this, thus allowing it to be used for raw
1836 * A new machines.target unit has been introduced that is
1837 supposed to group all containers/VMs invoked as services on
1838 the system. systemd-nspawn@.service has been updated to
1839 integrate with that.
1841 * machinectl gained a new "start" command, for invoking a
1842 container as a service. "machinectl start foo" is mostly
1843 equivalent to "systemctl start systemd-nspawn@foo.service",
1844 but handles escaping in a nicer way.
1846 * systemd-nspawn will now mount most of the cgroupfs tree
1847 read-only into each container, with the exception of the
1848 container's own subtree in the name=systemd hierarchy.
1850 * journald now sets the special FS_NOCOW file flag for its
1851 journal files. This should improve performance on btrfs, by
1852 avoiding heavy fragmentation when journald's write-pattern
1853 is used on COW file systems. It degrades btrfs' data
1854 integrity guarantees for the files to the same levels as for
1855 ext3/ext4 however. This should be OK though as journald does
1856 its own data integrity checks and all its objects are
1857 checksummed on disk. Also, journald should handle btrfs disk
1858 full events a lot more gracefully now, by processing SIGBUS
1859 errors, and not relying on fallocate() anymore.
1861 * When journald detects that journal files it is writing to
1862 have been deleted it will immediately start new journal
1865 * systemd now provides a way to store file descriptors
1866 per-service in PID 1.This is useful for daemons to ensure
1867 that fds they require are not lost during a daemon
1868 restart. The fds are passed to the daemon on the next
1869 invocation in the same way socket activation fds are
1870 passed. This is now used by journald to ensure that the
1871 various sockets connected to all the system's stdout/stderr
1872 are not lost when journald is restarted. File descriptors
1873 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
1874 an extension to sd_notify(). Note that a limit is enforced
1875 on the number of fds a service can store in PID 1, and it
1876 defaults to 0, so that no fds may be stored, unless this is
1877 explicitly turned on.
1879 * The default TERM variable to use for units connected to a
1880 terminal, when no other value is explicitly is set is now
1881 vt220 rather than vt102. This should be fairly safe still,
1882 but allows PgUp/PgDn work.
1884 * The /etc/crypttab option header= as known from Debian is now
1887 * "loginctl user-status" and "loginctl session-status" will
1888 now show the last 10 lines of log messages of the
1889 user/session following the status output. Similar,
1890 "machinectl status" will show the last 10 log lines
1891 associated with a virtual machine or container
1892 service. (Note that this is usually not the log messages
1893 done in the VM/container itself, but simply what the
1894 container manager logs. For nspawn this includes all console
1897 * "loginctl session-status" without further argument will now
1898 show the status of the session of the caller. Similar,
1899 "lock-session", "unlock-session", "activate",
1900 "enable-linger", "disable-linger" may now be called without
1901 session/user parameter in which case they apply to the
1902 caller's session/user.
1904 * An X11 session scriptlet is now shipped that uploads
1905 $DISPLAY and $XAUTHORITY into the environment of the systemd
1906 --user daemon if a session begins. This should improve
1907 compatibility with X11 enabled applications run as systemd
1910 * Generators are now subject to masking via /etc and /run, the
1911 same way as unit files.
1913 * networkd .network files gained support for configuring
1914 per-link IPv4/IPv6 packet forwarding as well as IPv4
1915 masquerading. This is by default turned on for veth links to
1916 containers, as registered by systemd-nspawn. This means that
1917 nspawn containers run with --network-veth will now get
1918 automatic routed access to the host's networks without any
1919 further configuration or setup, as long as networkd runs on
1922 * systemd-nspawn gained the --port= (-p) switch to expose TCP
1923 or UDP posts of a container on the host. With this in place
1924 it is possible to run containers with private veth links
1925 (--network-veth), and have their functionality exposed on
1926 the host as if their services were running directly on the
1929 * systemd-nspawn's --network-veth switch now gained a short
1930 version "-n", since with the changes above it is now truly
1931 useful out-of-the-box. The systemd-nspawn@.service has been
1932 updated to make use of it too by default.
1934 * systemd-nspawn will now maintain a per-image R/W lock, to
1935 ensure that the same image is not started more than once
1936 writable. (It's OK to run an image multiple times
1937 simultaneously in read-only mode.)
1939 * systemd-nspawn's --image= option is now capable of
1940 dissecting and booting MBR and GPT disk images that contain
1941 only a single active Linux partition. Previously it
1942 supported only GPT disk images with proper GPT type
1943 IDs. This allows running cloud images from major
1944 distributions directly with systemd-nspawn, without
1947 * In addition to collecting mouse dpi data in the udev
1948 hardware database, there's now support for collecting angle
1949 information for mouse scroll wheels. The database is
1950 supposed to guarantee similar scrolling behavior on mice
1951 that it knows about. There's also support for collecting
1952 information about Touchpad types.
1954 * udev's input_id built-in will now also collect touch screen
1955 dimension data and attach it to probed devices.
1957 * /etc/os-release gained support for a Distribution Privacy
1960 * networkd gained support for creating "ipvlan", "gretap",
1961 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
1963 * systemd-tmpfiles gained support for "a" lines for setting
1966 * systemd-nspawn will now mount /tmp in the container to
1967 tmpfs, automatically.
1969 * systemd now exposes the memory.usage_in_bytes cgroup
1970 attribute and shows it for each service in the "systemctl
1971 status" output, if available.
1973 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
1974 immediate reboot is triggered. This useful if shutdown is
1975 hung and is unable to complete, to expedite the
1976 operation. Note that this kind of reboot will still unmount
1977 all file systems, and hence should not result in fsck being
1980 * A .device unit for an optical block device will now be
1981 considered active only when a medium is in the drive. Also,
1982 mount units are now bound to their backing devices thus
1983 triggering automatic unmounting when devices become
1984 unavailable. With this in place systemd will now
1985 automatically unmount left-over mounts when a CD-ROM is
1986 ejected or an USB stick is yanked from the system.
1988 * networkd-wait-online now has support for waiting for
1989 specific interfaces only (with globbing), and for giving up
1990 after a configurable timeout.
1992 * networkd now exits when idle. It will be automatically
1993 restarted as soon as interfaces show up, are removed or
1994 change state. networkd will stay around as long as there is
1995 at least one DHCP state machine or similar around, that keep
1998 * networkd may now configure IPv6 link-local addressing in
1999 addition to IPv4 link-local addressing.
2001 * The IPv6 "token" for use in SLAAC may now be configured for
2002 each .network interface in networkd.
2004 * Routes configured with networkd may now be assigned a scope
2007 * networkd's [Match] sections now support globbing and lists
2008 of multiple space-separated matches per item.
2010 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
2011 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
2012 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
2013 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
2014 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
2015 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
2016 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
2017 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
2018 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
2019 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
2020 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
2021 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
2022 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
2023 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
2024 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
2025 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
2026 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
2027 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
2028 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
2029 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
2030 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
2031 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
2032 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
2033 Hoffmann, Zbigniew Jędrzejewski-Szmek
2035 — Berlin, 2015-02-16
2039 * When querying unit file enablement status (for example via
2040 "systemctl is-enabled"), a new state "indirect" is now known
2041 which indicates that a unit might not be enabled itself, but
2042 another unit listed in its Also= setting might be.
2044 * Similar to the various existing ConditionXYZ= settings for
2045 units, there are now matching AssertXYZ= settings. While
2046 failing conditions cause a unit to be skipped, but its job
2047 to succeed, failing assertions declared like this will cause
2048 a unit start operation and its job to fail.
2050 * hostnamed now knows a new chassis type "embedded".
2052 * systemctl gained a new "edit" command. When used on a unit
2053 file, this allows extending unit files with .d/ drop-in
2054 configuration snippets or editing the full file (after
2055 copying it from /usr/lib to /etc). This will invoke the
2056 user's editor (as configured with $EDITOR), and reload the
2057 modified configuration after editing.
2059 * "systemctl status" now shows the suggested enablement state
2060 for a unit, as declared in the (usually vendor-supplied)
2061 system preset files.
2063 * nss-myhostname will now resolve the single-label host name
2064 "gateway" to the locally configured default IP routing
2065 gateways, ordered by their metrics. This assigns a stable
2066 name to the used gateways, regardless which ones are
2067 currently configured. Note that the name will only be
2068 resolved after all other name sources (if nss-myhostname is
2069 configured properly) and should hence not negatively impact
2070 systems that use the single-label host name "gateway" in
2073 * systemd-inhibit now allows filtering by mode when listing
2076 * Scope and service units gained a new "Delegate" boolean
2077 property, which, when set, allows processes running inside the
2078 unit to further partition resources. This is primarily
2079 useful for systemd user instances as well as container
2082 * journald will now pick up audit messages directly from
2083 the kernel, and log them like any other log message. The
2084 audit fields are split up and fully indexed. This means that
2085 journalctl in many ways is now a (nicer!) alternative to
2086 ausearch, the traditional audit client. Note that this
2087 implements only a minimal audit client. If you want the
2088 special audit modes like reboot-on-log-overflow, please use
2089 the traditional auditd instead, which can be used in
2090 parallel to journald.
2092 * The ConditionSecurity= unit file option now understands the
2093 special string "audit" to check whether auditing is
2096 * journalctl gained two new commands --vacuum-size= and
2097 --vacuum-time= to delete old journal files until the
2098 remaining ones take up no more than the specified size on disk,
2099 or are not older than the specified time.
2101 * A new, native PPPoE library has been added to sd-network,
2102 systemd's library of light-weight networking protocols. This
2103 library will be used in a future version of networkd to
2104 enable PPPoE communication without an external pppd daemon.
2106 * The busctl tool now understands a new "capture" verb that
2107 works similar to "monitor", but writes a packet capture
2108 trace to STDOUT that can be redirected to a file which is
2109 compatible with libcap's capture file format. This can then
2110 be loaded in Wireshark and similar tools to inspect bus
2113 * The busctl tool now understands a new "tree" verb that shows
2114 the object trees of a specific service on the bus, or of all
2117 * The busctl tool now understands a new "introspect" verb that
2118 shows all interfaces and members of objects on the bus,
2119 including their signature and values. This is particularly
2120 useful to get more information about bus objects shown by
2121 the new "busctl tree" command.
2123 * The busctl tool now understands new verbs "call",
2124 "set-property" and "get-property" for invoking bus method
2125 calls, setting and getting bus object properties in a
2128 * busctl gained a new --augment-creds= argument that controls
2129 whether the tool shall augment credential information it
2130 gets from the bus with data from /proc, in a possibly
2133 * nspawn's --link-journal= switch gained two new values
2134 "try-guest" and "try-host" that work like "guest" and
2135 "host", but do not fail if the host has no persistent
2136 journalling enabled. -j is now equivalent to
2137 --link-journal=try-guest.
2139 * macvlan network devices created by nspawn will now have
2140 stable MAC addresses.
2142 * A new SmackProcessLabel= unit setting has been added, which
2143 controls the SMACK security label processes forked off by
2144 the respective unit shall use.
2146 * If compiled with --enable-xkbcommon, systemd-localed will
2147 verify x11 keymap settings by compiling the given keymap. It
2148 will spew out warnings if the compilation fails. This
2149 requires libxkbcommon to be installed.
2151 * When a coredump is collected, a larger number of metadata
2152 fields is now collected and included in the journal records
2153 created for it. More specifically, control group membership,
2154 environment variables, memory maps, working directory,
2155 chroot directory, /proc/$PID/status, and a list of open file
2156 descriptors is now stored in the log entry.
2158 * The udev hwdb now contains DPI information for mice. For
2161 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
2163 * All systemd programs that read standalone configuration
2164 files in /etc now also support a corresponding series of
2165 .conf.d configuration directories in /etc/, /run/,
2166 /usr/local/lib/, /usr/lib/, and (if configured with
2167 --enable-split-usr) /lib/. In particular, the following
2168 configuration files now have corresponding configuration
2169 directories: system.conf user.conf, logind.conf,
2170 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
2171 resolved.conf, timesyncd.conf, journal-remote.conf, and
2172 journal-upload.conf. Note that distributions should use the
2173 configuration directories in /usr/lib/; the directories in
2174 /etc/ are reserved for the system administrator.
2176 * systemd-rfkill will no longer take the rfkill device name
2177 into account when storing rfkill state on disk, as the name
2178 might be dynamically assigned and not stable. Instead, the
2179 ID_PATH udev variable combined with the rfkill type (wlan,
2180 bluetooth, ...) is used.
2182 * A new service systemd-machine-id-commit.service has been
2183 added. When used on systems where /etc is read-only during
2184 boot, and /etc/machine-id is not initialized (but an empty
2185 file), this service will copy the temporary machine ID
2186 created as replacement into /etc after the system is fully
2187 booted up. This is useful for systems that are freshly
2188 installed with a non-initialized machine ID, but should get
2189 a fixed machine ID for subsequent boots.
2191 * networkd's .netdev files now provide a large set of
2192 configuration parameters for VXLAN devices. Similarly, the
2193 bridge port cost parameter is now configurable in .network
2194 files. There's also new support for configuring IP source
2195 routing. networkd .link files gained support for a new
2196 OriginalName= match that is useful to match against the
2197 original interface name the kernel assigned. .network files
2198 may include MTU= and MACAddress= fields for altering the MTU
2199 and MAC address while being connected to a specific network
2202 * The LUKS logic gained supported for configuring
2203 UUID-specific key files. There's also new support for naming
2204 LUKS device from the kernel command line, using the new
2205 luks.name= argument.
2207 * Timer units may now be transiently created via the bus API
2208 (this was previously already available for scope and service
2209 units). In addition it is now possible to create multiple
2210 transient units at the same time with a single bus call. The
2211 "systemd-run" tool has been updated to make use of this for
2212 running commands on a specified time, in at(1)-style.
2214 * tmpfiles gained support for "t" lines, for assigning
2215 extended attributes to files. Among other uses this may be
2216 used to assign SMACK labels to files.
2218 Contributions from: Alin Rauta, Alison Chaiken, Andrej
2219 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
2220 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
2221 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
2222 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
2223 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
2224 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
2225 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
2226 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
2227 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
2228 Michael Biebl, Michael Chapman, Michael Marineau, Michal
2229 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
2230 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
2231 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
2232 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
2233 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
2234 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
2235 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
2237 — Berlin, 2014-12-10
2241 * journalctl gained the new options -t/--identifier= to match
2242 on the syslog identifier (aka "tag"), as well as --utc to
2243 show log timestamps in the UTC timezone. journalctl now also
2244 accepts -n/--lines=all to disable line capping in a pager.
2246 * journalctl gained a new switch, --flush, that synchronously
2247 flushes logs from /run/log/journal to /var/log/journal if
2248 persistent storage is enabled. systemd-journal-flush.service
2249 now waits until the operation is complete.
2251 * Services can notify the manager before they start a reload
2252 (by sending RELOADING=1) or shutdown (by sending
2253 STOPPING=1). This allows the manager to track and show the
2254 internal state of daemons and closes a race condition when
2255 the process is still running but has closed its D-Bus
2258 * Services with Type=oneshot do not have to have any ExecStart
2261 * User units are now loaded also from
2262 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
2263 /run/systemd/user directory that was already previously
2264 supported, but is under the control of the user.
2266 * Job timeouts (i.e. time-outs on the time a job that is
2267 queued stays in the run queue) can now optionally result in
2268 immediate reboot or power-off actions (JobTimeoutAction= and
2269 JobTimeoutRebootArgument=). This is useful on ".target"
2270 units, to limit the maximum time a target remains
2271 undispatched in the run queue, and to trigger an emergency
2272 operation in such a case. This is now used by default to
2273 turn off the system if boot-up (as defined by everything in
2274 basic.target) hangs and does not complete for at least
2275 15min. Also, if power-off or reboot hang for at least 30min
2276 an immediate power-off/reboot operation is triggered. This
2277 functionality is particularly useful to increase reliability
2278 on embedded devices, but also on laptops which might
2279 accidentally get powered on when carried in a backpack and
2280 whose boot stays stuck in a hard disk encryption passphrase
2283 * systemd-logind can be configured to also handle lid switch
2284 events even when the machine is docked or multiple displays
2285 are attached (HandleLidSwitchDocked= option).
2287 * A helper binary and a service have been added which can be
2288 used to resume from hibernation in the initramfs. A
2289 generator will parse the resume= option on the kernel
2290 command line to trigger resume.
2292 * A user console daemon systemd-consoled has been
2293 added. Currently, it is a preview, and will so far open a
2294 single terminal on each session of the user marked as
2295 Desktop=systemd-console.
2297 * Route metrics can be specified for DHCP routes added by
2300 * The SELinux context of socket-activated services can be set
2301 from the information provided by the networking stack
2302 (SELinuxContextFromNet= option).
2304 * Userspace firmware loading support has been removed and
2305 the minimum supported kernel version is thus bumped to 3.7.
2307 * Timeout for udev workers has been increased from 1 to 3
2308 minutes, but a warning will be printed after 1 minute to
2309 help diagnose kernel modules that take a long time to load.
2311 * Udev rules can now remove tags on devices with TAG-="foobar".
2313 * systemd's readahead implementation has been removed. In many
2314 circumstances it didn't give expected benefits even for
2315 rotational disk drives and was becoming less relevant in the
2316 age of SSDs. As none of the developers has been using
2317 rotating media anymore, and nobody stepped up to actively
2318 maintain this component of systemd it has now been removed.
2320 * Swap units can use Options= to specify discard options.
2321 Discard options specified for swaps in /etc/fstab are now
2324 * Docker containers are now detected as a separate type of
2327 * The Password Agent protocol gained support for queries where
2328 the user input is shown, useful e.g. for user names.
2329 systemd-ask-password gained a new --echo option to turn that
2332 * The default sysctl.d/ snippets will now set:
2334 net.core.default_qdisc = fq_codel
2336 This selects Fair Queuing Controlled Delay as the default
2337 queuing discipline for network interfaces. fq_codel helps
2338 fight the network bufferbloat problem. It is believed to be
2339 a good default with no tuning required for most workloads.
2340 Downstream distributions may override this choice. On 10Gbit
2341 servers that do not do forwarding, "fq" may perform better.
2342 Systems without a good clocksource should use "pfifo_fast".
2344 * If kdbus is enabled during build a new option BusPolicy= is
2345 available for service units, that allows locking all service
2346 processes into a stricter bus policy, in order to limit
2347 access to various bus services, or even hide most of them
2348 from the service's view entirely.
2350 * networkctl will now show the .network and .link file
2351 networkd has applied to a specific interface.
2353 * sd-login gained a new API call sd_session_get_desktop() to
2354 query which desktop environment has been selected for a
2357 * UNIX utmp support is now compile-time optional to support
2358 legacy-free systems.
2360 * systemctl gained two new commands "add-wants" and
2361 "add-requires" for pulling in units from specific targets
2364 * If the word "rescue" is specified on the kernel command line
2365 the system will now boot into rescue mode (aka
2366 rescue.target), which was previously available only by
2367 specifying "1" or "systemd.unit=rescue.target" on the kernel
2368 command line. This new kernel command line option nicely
2369 mirrors the already existing "emergency" kernel command line
2372 * New kernel command line options mount.usr=, mount.usrflags=,
2373 mount.usrfstype= have been added that match root=, rootflags=,
2374 rootfstype= but allow mounting a specific file system to
2377 * The $NOTIFY_SOCKET is now also passed to control processes of
2378 services, not only the main process.
2380 * This version reenables support for fsck's -l switch. This
2381 means at least version v2.25 of util-linux is required for
2382 operation, otherwise dead-locks on device nodes may
2383 occur. Again: you need to update util-linux to at least
2384 v2.25 when updating systemd to v217.
2386 * The "multi-seat-x" tool has been removed from systemd, as
2387 its functionality has been integrated into X servers 1.16,
2388 and the tool is hence redundant. It is recommended to update
2389 display managers invoking this tool to simply invoke X
2390 directly from now on, again.
2392 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
2393 message flag has been added for all of systemd's PolicyKit
2394 authenticated method calls has been added. In particular
2395 this now allows optional interactive authorization via
2396 PolicyKit for many of PID1's privileged operations such as
2397 unit file enabling and disabling.
2399 * "udevadm hwdb --update" learnt a new switch "--usr" for
2400 placing the rebuilt hardware database in /usr instead of
2401 /etc. When used only hardware database entries stored in
2402 /usr will be used, and any user database entries in /etc are
2403 ignored. This functionality is useful for vendors to ship a
2404 pre-built database on systems where local configuration is
2405 unnecessary or unlikely.
2407 * Calendar time specifications in .timer units now also
2408 understand the strings "semi-annually", "quarterly" and
2409 "minutely" as shortcuts (in addition to the preexisting
2410 "anually", "hourly", ...).
2412 * systemd-tmpfiles will now correctly create files in /dev
2413 at boot which are marked for creation only at boot. It is
2414 recommended to always create static device nodes with 'c!'
2415 and 'b!', so that they are created only at boot and not
2416 overwritten at runtime.
2418 * When the watchdog logic is used for a service (WatchdogSec=)
2419 and the watchdog timeout is hit the service will now be
2420 terminated with SIGABRT (instead of just SIGTERM), in order
2421 to make sure a proper coredump and backtrace is
2422 generated. This ensures that hanging services will result in
2423 similar coredump/backtrace behaviour as services that hit a
2426 Contributions from: Andreas Henriksson, Andrei Borzenkov,
2427 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
2428 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
2429 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
2430 Herrmann, David Sommerseth, David Strauss, Emil Renner
2431 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
2432 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
2433 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
2434 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
2435 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
2436 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
2437 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
2438 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
2439 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
2440 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
2441 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
2442 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
2443 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
2444 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
2445 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
2446 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
2449 — Berlin, 2014-10-28
2453 * timedated no longer reads NTP implementation unit names from
2454 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
2455 implementations should add a
2457 Conflicts=systemd-timesyncd.service
2459 to their unit files to take over and replace systemd's NTP
2460 default functionality.
2462 * systemd-sysusers gained a new line type "r" for configuring
2463 which UID/GID ranges to allocate system users/groups
2464 from. Lines of type "u" may now add an additional column
2465 that specifies the home directory for the system user to be
2466 created. Also, systemd-sysusers may now optionally read user
2467 information from STDIN instead of a file. This is useful for
2468 invoking it from RPM preinst scriptlets that need to create
2469 users before the first RPM file is installed since these
2470 files might need to be owned by them. A new
2471 %sysusers_create_inline RPM macro has been introduced to do
2472 just that. systemd-sysusers now updates the shadow files as
2473 well as the user/group databases, which should enhance
2474 compatibility with certain tools like grpck.
2476 * A number of bus APIs of PID 1 now optionally consult
2477 PolicyKit to permit access for otherwise unprivileged
2478 clients under certain conditions. Note that this currently
2479 doesn't support interactive authentication yet, but this is
2480 expected to be added eventually, too.
2482 * /etc/machine-info now has new fields for configuring the
2483 deployment environment of the machine, as well as the
2484 location of the machine. hostnamectl has been updated with
2485 new command to update these fields.
2487 * systemd-timesyncd has been updated to automatically acquire
2488 NTP server information from systemd-networkd, which might
2489 have been discovered via DHCP.
2491 * systemd-resolved now includes a caching DNS stub resolver
2492 and a complete LLMNR name resolution implementation. A new
2493 NSS module "nss-resolve" has been added which can be used
2494 instead of glibc's own "nss-dns" to resolve hostnames via
2495 systemd-resolved. Hostnames, addresses and arbitrary RRs may
2496 be resolved via systemd-resolved D-Bus APIs. In contrast to
2497 the glibc internal resolver systemd-resolved is aware of
2498 multi-homed system, and keeps DNS server and caches separate
2499 and per-interface. Queries are sent simultaneously on all
2500 interfaces that have DNS servers configured, in order to
2501 properly handle VPNs and local LANs which might resolve
2502 separate sets of domain names. systemd-resolved may acquire
2503 DNS server information from systemd-networkd automatically,
2504 which in turn might have discovered them via DHCP. A tool
2505 "systemd-resolve-host" has been added that may be used to
2506 query the DNS logic in resolved. systemd-resolved implements
2507 IDNA and automatically uses IDNA or UTF-8 encoding depending
2508 on whether classic DNS or LLMNR is used as transport. In the
2509 next releases we intend to add a DNSSEC and mDNS/DNS-SD
2510 implementation to systemd-resolved.
2512 * A new NSS module nss-mymachines has been added, that
2513 automatically resolves the names of all local registered
2514 containers to their respective IP addresses.
2516 * A new client tool "networkctl" for systemd-networkd has been
2517 added. It currently is entirely passive and will query
2518 networking configuration from udev, rtnetlink and networkd,
2519 and present it to the user in a very friendly
2520 way. Eventually, we hope to extend it to become a full
2521 control utility for networkd.
2523 * .socket units gained a new DeferAcceptSec= setting that
2524 controls the kernels' TCP_DEFER_ACCEPT sockopt for
2525 TCP. Similarly, support for controlling TCP keep-alive
2526 settings has been added (KeepAliveTimeSec=,
2527 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
2528 turning off Nagle's algorithm on TCP has been added
2531 * logind learned a new session type "web", for use in projects
2532 like Cockpit which register web clients as PAM sessions.
2534 * timer units with at least one OnCalendar= setting will now
2535 be started only after timer-sync.target has been
2536 reached. This way they will not elapse before the system
2537 clock has been corrected by a local NTP client or
2538 similar. This is particular useful on RTC-less embedded
2539 machines, that come up with an invalid system clock.
2541 * systemd-nspawn's --network-veth= switch should now result in
2542 stable MAC addresses for both the outer and the inner side
2545 * systemd-nspawn gained a new --volatile= switch for running
2546 container instances with /etc or /var unpopulated.
2548 * The kdbus client code has been updated to use the new Linux
2549 3.17 memfd subsystem instead of the old kdbus-specific one.
2551 * systemd-networkd's DHCP client and server now support
2552 FORCERENEW. There are also new configuration options to
2553 configure the vendor client identifier and broadcast mode
2556 * systemd will no longer inform the kernel about the current
2557 timezone, as this is necessarily incorrect and racy as the
2558 kernel has no understanding of DST and similar
2559 concepts. This hence means FAT timestamps will be always
2560 considered UTC, similar to what Android is already
2561 doing. Also, when the RTC is configured to the local time
2562 (rather than UTC) systemd will never synchronize back to it,
2563 as this might confuse Windows at a later boot.
2565 * systemd-analyze gained a new command "verify" for offline
2566 validation of unit files.
2568 * systemd-networkd gained support for a couple of additional
2569 settings for bonding networking setups. Also, the metric for
2570 statically configured routes may now be configured. For
2571 network interfaces where this is appropriate the peer IP
2572 address may now be configured.
2574 * systemd-networkd's DHCP client will no longer request
2575 broadcasting by default, as this tripped up some networks.
2576 For hardware where broadcast is required the feature should
2577 be switched back on using RequestBroadcast=yes.
2579 * systemd-networkd will now set up IPv4LL addresses (when
2580 enabled) even if DHCP is configured successfully.
2582 * udev will now default to respect network device names given
2583 by the kernel when the kernel indicates that these are
2584 predictable. This behavior can be tweaked by changing
2585 NamePolicy= in the relevant .link file.
2587 * A new library systemd-terminal has been added that
2588 implements full TTY stream parsing and rendering. This
2589 library is supposed to be used later on for implementing a
2590 full userspace VT subsystem, replacing the current kernel
2593 * A new tool systemd-journal-upload has been added to push
2594 journal data to a remote system running
2595 systemd-journal-remote.
2597 * journald will no longer forward all local data to another
2598 running syslog daemon. This change has been made because
2599 rsyslog (which appears to be the most commonly used syslog
2600 implementation these days) no longer makes use of this, and
2601 instead pulls the data out of the journal on its own. Since
2602 forwarding the messages to a non-existent syslog server is
2603 more expensive than we assumed we have now turned this
2604 off. If you run a syslog server that is not a recent rsyslog
2605 version, you have to turn this option on again
2606 (ForwardToSyslog= in journald.conf).
2608 * journald now optionally supports the LZ4 compressor for
2609 larger journal fields. This compressor should perform much
2610 better than XZ which was the previous default.
2612 * machinectl now shows the IP addresses of local containers,
2613 if it knows them, plus the interface name of the container.
2615 * A new tool "systemd-escape" has been added that makes it
2616 easy to escape strings to build unit names and similar.
2618 * sd_notify() messages may now include a new ERRNO= field
2619 which is parsed and collected by systemd and shown among the
2620 "systemctl status" output for a service.
2622 * A new component "systemd-firstboot" has been added that
2623 queries the most basic systemd information (timezone,
2624 hostname, root password) interactively on first
2625 boot. Alternatively it may also be used to provision these
2626 things offline on OS images installed into directories.
2628 * The default sysctl.d/ snippets will now set
2630 net.ipv4.conf.default.promote_secondaries=1
2632 This has the benefit of no flushing secondary IP addresses
2633 when primary addresses are removed.
2635 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
2636 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
2637 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
2638 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
2639 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
2640 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
2641 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
2642 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
2643 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
2644 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
2645 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
2646 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
2647 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
2648 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
2649 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
2651 — Berlin, 2014-08-19
2655 * A new tool systemd-sysusers has been added. This tool
2656 creates system users and groups in /etc/passwd and
2657 /etc/group, based on static declarative system user/group
2658 definitions in /usr/lib/sysusers.d/. This is useful to
2659 enable factory resets and volatile systems that boot up with
2660 an empty /etc directory, and thus need system users and
2661 groups created during early boot. systemd now also ships
2662 with two default sysusers.d/ files for the most basic
2663 users and groups systemd and the core operating system
2666 * A new tmpfiles snippet has been added that rebuilds the
2667 essential files in /etc on boot, should they be missing.
2669 * A directive for ensuring automatic clean-up of
2670 /var/cache/man/ has been removed from the default
2671 configuration. This line should now be shipped by the man
2672 implementation. The necessary change has been made to the
2673 man-db implementation. Note that you need to update your man
2674 implementation to one that ships this line, otherwise no
2675 automatic clean-up of /var/cache/man will take place.
2677 * A new condition ConditionNeedsUpdate= has been added that
2678 may conditionalize services to only run when /etc or /var
2679 are "older" than the vendor operating system resources in
2680 /usr. This is useful for reconstructing or updating /etc
2681 after an offline update of /usr or a factory reset, on the
2682 next reboot. Services that want to run once after such an
2683 update or reset should use this condition and order
2684 themselves before the new systemd-update-done.service, which
2685 will mark the two directories as fully updated. A number of
2686 service files have been added making use of this, to rebuild
2687 the udev hardware database, the journald message catalog and
2688 dynamic loader cache (ldconfig). The systemd-sysusers tool
2689 described above also makes use of this now. With this in
2690 place it is now possible to start up a minimal operating
2691 system with /etc empty cleanly. For more information on the
2692 concepts involved see this recent blog story:
2694 http://0pointer.de/blog/projects/stateless.html
2696 * A new system group "input" has been introduced, and all
2697 input device nodes get this group assigned. This is useful
2698 for system-level software to get access to input devices. It
2699 complements what is already done for "audio" and "video".
2701 * systemd-networkd learnt minimal DHCPv4 server support in
2702 addition to the existing DHCPv4 client support. It also
2703 learnt DHCPv6 client and IPv6 Router Solicitation client
2704 support. The DHCPv4 client gained support for static routes
2705 passed in from the server. Note that the [DHCPv4] section
2706 known in older systemd-networkd versions has been renamed to
2707 [DHCP] and is now also used by the DHCPv6 client. Existing
2708 .network files using settings of this section should be
2709 updated, though compatibility is maintained. Optionally, the
2710 client hostname may now be sent to the DHCP server.
2712 * networkd gained support for vxlan virtual networks as well
2713 as tun/tap and dummy devices.
2715 * networkd gained support for automatic allocation of address
2716 ranges for interfaces from a system-wide pool of
2717 addresses. This is useful for dynamically managing a large
2718 number of interfaces with a single network configuration
2719 file. In particular this is useful to easily assign
2720 appropriate IP addresses to the veth links of a large number
2721 of nspawn instances.
2723 * RPM macros for processing sysusers, sysctl and binfmt
2724 drop-in snippets at package installation time have been
2727 * The /etc/os-release file should now be placed in
2728 /usr/lib/os-release. The old location is automatically
2729 created as symlink. /usr/lib is the more appropriate
2730 location of this file, since it shall actually describe the
2731 vendor operating system shipped in /usr, and not the
2732 configuration stored in /etc.
2734 * .mount units gained a new boolean SloppyOptions= setting
2735 that maps to mount(8)'s -s option which enables permissive
2736 parsing of unknown mount options.
2738 * tmpfiles learnt a new "L+" directive which creates a symlink
2739 but (unlike "L") deletes a pre-existing file first, should
2740 it already exist and not already be the correct
2741 symlink. Similarly, "b+", "c+" and "p+" directives have been
2742 added as well, which create block and character devices, as
2743 well as fifos in the filesystem, possibly removing any
2744 pre-existing files of different types.
2746 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
2747 'argument' field (which so far specified the source to
2748 symlink/copy the files from) is now optional. If omitted the
2749 same file os copied from /usr/share/factory/ suffixed by the
2750 full destination path. This is useful for populating /etc
2751 with essential files, by copying them from vendor defaults
2752 shipped in /usr/share/factory/etc.
2754 * A new command "systemctl preset-all" has been added that
2755 applies the service preset settings to all installed unit
2756 files. A new switch --preset-mode= has been added that
2757 controls whether only enable or only disable operations
2760 * A new command "systemctl is-system-running" has been added
2761 that allows checking the overall state of the system, for
2762 example whether it is fully up and running.
2764 * When the system boots up with an empty /etc, the equivalent
2765 to "systemctl preset-all" is executed during early boot, to
2766 make sure all default services are enabled after a factory
2769 * systemd now contains a minimal preset file that enables the
2770 most basic services systemd ships by default.
2772 * Unit files' [Install] section gained a new DefaultInstance=
2773 field for defining the default instance to create if a
2774 template unit is enabled with no instance specified.
2776 * A new passive target cryptsetup-pre.target has been added
2777 that may be used by services that need to make they run and
2778 finish before the first LUKS cryptographic device is set up.
2780 * The /dev/loop-control and /dev/btrfs-control device nodes
2781 are now owned by the "disk" group by default, opening up
2782 access to this group.
2784 * systemd-coredump will now automatically generate a
2785 stack trace of all core dumps taking place on the system,
2786 based on elfutils' libdw library. This stack trace is logged
2789 * systemd-coredump may now optionally store coredumps directly
2790 on disk (in /var/lib/systemd/coredump, possibly compressed),
2791 instead of storing them unconditionally in the journal. This
2792 mode is the new default. A new configuration file
2793 /etc/systemd/coredump.conf has been added to configure this
2794 and other parameters of systemd-coredump.
2796 * coredumpctl gained a new "info" verb to show details about a
2797 specific coredump. A new switch "-1" has also been added
2798 that makes sure to only show information about the most
2799 recent entry instead of all entries. Also, as the tool is
2800 generally useful now the "systemd-" prefix of the binary
2801 name has been removed. Distributions that want to maintain
2802 compatibility with the old name should add a symlink from
2803 the old name to the new name.
2805 * journald's SplitMode= now defaults to "uid". This makes sure
2806 that unprivileged users can access their own coredumps with
2807 coredumpctl without restrictions.
2809 * New kernel command line options "systemd.wants=" (for
2810 pulling an additional unit during boot), "systemd.mask="
2811 (for masking a specific unit for the boot), and
2812 "systemd.debug-shell" (for enabling the debug shell on tty9)
2813 have been added. This is implemented in the new generator
2814 "systemd-debug-generator".
2816 * systemd-nspawn will now by default filter a couple of
2817 syscalls for containers, among them those required for
2818 kernel module loading, direct x86 IO port access, swap
2819 management, and kexec. Most importantly though
2820 open_by_handle_at() is now prohibited for containers,
2821 closing a hole similar to a recently discussed vulnerability
2822 in docker regarding access to files on file hierarchies the
2823 container should normally not have access to. Note that, for
2824 nspawn, we generally make no security claims anyway (and
2825 this is explicitly documented in the man page), so this is
2826 just a fix for one of the most obvious problems.
2828 * A new man page file-hierarchy(7) has been added that
2829 contains a minimized, modernized version of the file system
2830 layout systemd expects, similar in style to the FHS
2831 specification or hier(5). A new tool systemd-path(1) has
2832 been added to query many of these paths for the local
2835 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
2836 longer done. Since the directory now has a per-user size
2837 limit, and is cleaned on logout this appears unnecessary,
2838 in particular since this now brings the lifecycle of this
2839 directory closer in line with how IPC objects are handled.
2841 * systemd.pc now exports a number of additional directories,
2842 including $libdir (which is useful to identify the library
2843 path for the primary architecture of the system), and a
2844 couple of drop-in directories.
2846 * udev's predictable network interface names now use the dev_port
2847 sysfs attribute, introduced in linux 3.15 instead of dev_id to
2848 distinguish between ports of the same PCI function. dev_id should
2849 only be used for ports using the same HW address, hence the need
2852 * machined has been updated to export the OS version of a
2853 container (read from /etc/os-release and
2854 /usr/lib/os-release) on the bus. This is now shown in
2855 "machinectl status" for a machine.
2857 * A new service setting RestartForceExitStatus= has been
2858 added. If configured to a set of exit signals or process
2859 return values, the service will be restarted when the main
2860 daemon process exits with any of them, regardless of the
2863 * systemctl's -H switch for connecting to remote systemd
2864 machines has been extended so that it may be used to
2865 directly connect to a specific container on the
2866 host. "systemctl -H root@foobar:waldi" will now connect as
2867 user "root" to host "foobar", and then proceed directly to
2868 the container named "waldi". Note that currently you have to
2869 authenticate as user "root" for this to work, as entering
2870 containers is a privileged operation.
2872 Contributions from: Andreas Henriksson, Benjamin Steinwender,
2873 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
2874 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
2875 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
2876 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
2877 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
2878 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
2879 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
2880 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
2881 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
2882 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
2883 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
2885 — Berlin, 2014-07-03
2889 * As an experimental feature, udev now tries to lock the
2890 disk device node (flock(LOCK_SH|LOCK_NB)) while it
2891 executes events for the disk or any of its partitions.
2892 Applications like partitioning programs can lock the
2893 disk device node (flock(LOCK_EX)) and claim temporary
2894 device ownership that way; udev will entirely skip all event
2895 handling for this disk and its partitions. If the disk
2896 was opened for writing, the close will trigger a partition
2897 table rescan in udev's "watch" facility, and if needed
2898 synthesize "change" events for the disk and all its partitions.
2899 This is now unconditionally enabled, and if it turns out to
2900 cause major problems, we might turn it on only for specific
2901 devices, or might need to disable it entirely. Device Mapper
2902 devices are excluded from this logic.
2904 * We temporarily dropped the "-l" switch for fsck invocations,
2905 since they collide with the flock() logic above. util-linux
2906 upstream has been changed already to avoid this conflict,
2907 and we will readd "-l" as soon as util-linux with this
2908 change has been released.
2910 * The dependency on libattr has been removed. Since a long
2911 time, the extended attribute calls have moved to glibc, and
2912 libattr is thus unnecessary.
2914 * Virtualization detection works without priviliges now. This
2915 means the systemd-detect-virt binary no longer requires
2916 CAP_SYS_PTRACE file capabilities, and our daemons can run
2917 with fewer privileges.
2919 * systemd-networkd now runs under its own "systemd-network"
2920 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
2921 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
2922 loses the ability to write to files owned by root this way.
2924 * Similarly, systemd-resolved now runs under its own
2925 "systemd-resolve" user with no capabilities remaining.
2927 * Similarly, systemd-bus-proxyd now runs under its own
2928 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
2930 * systemd-networkd gained support for setting up "veth"
2931 virtual Ethernet devices for container connectivity, as well
2932 as GRE and VTI tunnels.
2934 * systemd-networkd will no longer automatically attempt to
2935 manually load kernel modules necessary for certain tunnel
2936 transports. Instead, it is assumed the kernel loads them
2937 automatically when required. This only works correctly on
2938 very new kernels. On older kernels, please consider adding
2939 the kernel modules to /etc/modules-load.d/ as a work-around.
2941 * The resolv.conf file systemd-resolved generates has been
2942 moved to /run/systemd/resolve/. If you have a symlink from
2943 /etc/resolv.conf, it might be necessary to correct it.
2945 * Two new service settings, ProtectHome= and ProtectSystem=,
2946 have been added. When enabled, they will make the user data
2947 (such as /home) inaccessible or read-only and the system
2948 (such as /usr) read-only, for specific services. This allows
2949 very light-weight per-service sandboxing to avoid
2950 modifications of user data or system files from
2951 services. These two new switches have been enabled for all
2952 of systemd's long-running services, where appropriate.
2954 * Socket units gained new SocketUser= and SocketGroup=
2955 settings to set the owner user and group of AF_UNIX sockets
2956 and FIFOs in the file system.
2958 * Socket units gained a new RemoveOnStop= setting. If enabled,
2959 all FIFOS and sockets in the file system will be removed
2960 when the specific socket unit is stopped.
2962 * Socket units gained a new Symlinks= setting. It takes a list
2963 of symlinks to create to file system sockets or FIFOs
2964 created by the specific Unix sockets. This is useful to
2965 manage symlinks to socket nodes with the same life-cycle as
2968 * The /dev/log socket and /dev/initctl FIFO have been moved to
2969 /run, and have been replaced by symlinks. This allows
2970 connecting to these facilities even if PrivateDevices=yes is
2971 used for a service (which makes /dev/log itself unavailable,
2972 but /run is left). This also has the benefit of ensuring
2973 that /dev only contains device nodes, directories and
2974 symlinks, and nothing else.
2976 * sd-daemon gained two new calls sd_pid_notify() and
2977 sd_pid_notifyf(). They are similar to sd_notify() and
2978 sd_notifyf(), but allow overriding of the source PID of
2979 notification messages if permissions permit this. This is
2980 useful to send notify messages on behalf of a different
2981 process (for example, the parent process). The
2982 systemd-notify tool has been updated to make use of this
2983 when sending messages (so that notification messages now
2984 originate from the shell script invoking systemd-notify and
2985 not the systemd-notify process itself. This should minimize
2986 a race where systemd fails to associate notification
2987 messages to services when the originating process already
2990 * A new "on-abnormal" setting for Restart= has been added. If
2991 set, it will result in automatic restarts on all "abnormal"
2992 reasons for a process to exit, which includes unclean
2993 signals, core dumps, timeouts and watchdog timeouts, but
2994 does not include clean and unclean exit codes or clean
2995 signals. Restart=on-abnormal is an alternative for
2996 Restart=on-failure for services that shall be able to
2997 terminate and avoid restarts on certain errors, by
2998 indicating so with an unclean exit code. Restart=on-failure
2999 or Restart=on-abnormal is now the recommended setting for
3000 all long-running services.
3002 * If the InaccessibleDirectories= service setting points to a
3003 mount point (or if there are any submounts contained within
3004 it), it is now attempted to completely unmount it, to make
3005 the file systems truly unavailable for the respective
3008 * The ReadOnlyDirectories= service setting and
3009 systemd-nspawn's --read-only parameter are now recursively
3010 applied to all submounts, too.
3012 * Mount units may now be created transiently via the bus APIs.
3014 * The support for SysV and LSB init scripts has been removed
3015 from the systemd daemon itself. Instead, it is now
3016 implemented as a generator that creates native systemd units
3017 from these scripts when needed. This enables us to remove a
3018 substantial amount of legacy code from PID 1, following the
3019 fact that many distributions only ship a very small number
3020 of LSB/SysV init scripts nowadays.
3022 * Privileged Xen (dom0) domains are not considered
3023 virtualization anymore by the virtualization detection
3024 logic. After all, they generally have unrestricted access to
3025 the hardware and usually are used to manage the unprivileged
3028 * systemd-tmpfiles gained a new "C" line type, for copying
3029 files or entire directories.
3031 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
3032 lines. So far, they have been non-globbing versions of the
3033 latter, and have thus been redundant. In future, it is
3034 recommended to only use "z". "m" has hence been removed
3035 from the documentation, even though it stays supported.
3037 * A tmpfiles snippet to recreate the most basic structure in
3038 /var has been added. This is enough to create the /var/run →
3039 /run symlink and create a couple of structural
3040 directories. This allows systems to boot up with an empty or
3041 volatile /var. Of course, while with this change, the core OS
3042 now is capable with dealing with a volatile /var, not all
3043 user services are ready for it. However, we hope that sooner
3044 or later, many service daemons will be changed upstream so
3045 that they are able to automatically create their necessary
3046 directories in /var at boot, should they be missing. This is
3047 the first step to allow state-less systems that only require
3048 the vendor image for /usr to boot.
3050 * systemd-nspawn has gained a new --tmpfs= switch to mount an
3051 empty tmpfs instance to a specific directory. This is
3052 particularly useful for making use of the automatic
3053 reconstruction of /var (see above), by passing --tmpfs=/var.
3055 * Access modes specified in tmpfiles snippets may now be
3056 prefixed with "~", which indicates that they shall be masked
3057 by whether the existing file or directory is currently
3058 writable, readable or executable at all. Also, if specified,
3059 the sgid/suid/sticky bits will be masked for all
3062 * A new passive target unit "network-pre.target" has been
3063 added which is useful for services that shall run before any
3064 network is configured, for example firewall scripts.
3066 * The "floppy" group that previously owned the /dev/fd*
3067 devices is no longer used. The "disk" group is now used
3068 instead. Distributions should probably deprecate usage of
3071 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
3072 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
3073 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
3074 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
3075 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
3076 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
3079 — Berlin, 2014-06-11
3083 * A new "systemd-timesyncd" daemon has been added for
3084 synchronizing the system clock across the network. It
3085 implements an SNTP client. In contrast to NTP
3086 implementations such as chrony or the NTP reference server,
3087 this only implements a client side, and does not bother with
3088 the full NTP complexity, focusing only on querying time from
3089 one remote server and synchronizing the local clock to
3090 it. Unless you intend to serve NTP to networked clients or
3091 want to connect to local hardware clocks, this simple NTP
3092 client should be more than appropriate for most
3093 installations. The daemon runs with minimal privileges, and
3094 has been hooked up with networkd to only operate when
3095 network connectivity is available. The daemon saves the
3096 current clock to disk every time a new NTP sync has been
3097 acquired, and uses this to possibly correct the system clock
3098 early at bootup, in order to accommodate for systems that
3099 lack an RTC such as the Raspberry Pi and embedded devices,
3100 and to make sure that time monotonically progresses on these
3101 systems, even if it is not always correct. To make use of
3102 this daemon, a new system user and group "systemd-timesync"
3103 needs to be created on installation of systemd.
3105 * The queue "seqnum" interface of libudev has been disabled, as
3106 it was generally incompatible with device namespacing as
3107 sequence numbers of devices go "missing" if the devices are
3108 part of a different namespace.
3110 * "systemctl list-timers" and "systemctl list-sockets" gained
3111 a --recursive switch for showing units of these types also
3112 for all local containers, similar in style to the already
3113 supported --recursive switch for "systemctl list-units".
3115 * A new RebootArgument= setting has been added for service
3116 units, which may be used to specify a kernel reboot argument
3117 to use when triggering reboots with StartLimitAction=.
3119 * A new FailureAction= setting has been added for service
3120 units which may be used to specify an operation to trigger
3121 when a service fails. This works similarly to
3122 StartLimitAction=, but unlike it, controls what is done
3123 immediately rather than only after several attempts to
3124 restart the service in question.
3126 * hostnamed got updated to also expose the kernel name,
3127 release, and version on the bus. This is useful for
3128 executing commands like hostnamectl with the -H switch.
3129 systemd-analyze makes use of this to properly display
3130 details when running non-locally.
3132 * The bootchart tool can now show cgroup information in the
3133 graphs it generates.
3135 * The CFS CPU quota cgroup attribute is now exposed for
3136 services. The new CPUQuota= switch has been added for this
3137 which takes a percentage value. Setting this will have the
3138 result that a service may never get more CPU time than the
3139 specified percentage, even if the machine is otherwise idle.
3141 * systemd-networkd learned IPIP and SIT tunnel support.
3143 * LSB init scripts exposing a dependency on $network will now
3144 get a dependency on network-online.target rather than simply
3145 network.target. This should bring LSB handling closer to
3146 what it was on SysV systems.
3148 * A new fsck.repair= kernel option has been added to control
3149 how fsck shall deal with unclean file systems at boot.
3151 * The (.ini) configuration file parser will now silently
3152 ignore sections whose name begins with "X-". This may be
3153 used to maintain application-specific extension sections in unit
3156 * machined gained a new API to query the IP addresses of
3157 registered containers. "machinectl status" has been updated
3158 to show these addresses in its output.
3160 * A new call sd_uid_get_display() has been added to the
3161 sd-login APIs for querying the "primary" session of a
3162 user. The "primary" session of the user is elected from the
3163 user's sessions and generally a graphical session is
3164 preferred over a text one.
3166 * A minimal systemd-resolved daemon has been added. It
3167 currently simply acts as a companion to systemd-networkd and
3168 manages resolv.conf based on per-interface DNS
3169 configuration, possibly supplied via DHCP. In the long run
3170 we hope to extend this into a local DNSSEC enabled DNS and
3173 * The systemd-networkd-wait-online tool is now enabled by
3174 default. It will delay network-online.target until a network
3175 connection has been configured. The tool primarily integrates
3176 with networkd, but will also make a best effort to make sense
3177 of network configuration performed in some other way.
3179 * Two new service options StartupCPUShares= and
3180 StartupBlockIOWeight= have been added that work similarly to
3181 CPUShares= and BlockIOWeight= however only apply during
3182 system startup. This is useful to prioritize certain services
3183 differently during bootup than during normal runtime.
3185 * hostnamed has been changed to prefer the statically
3186 configured hostname in /etc/hostname (unless set to
3187 'localhost' or empty) over any dynamic one supplied by
3188 dhcp. With this change, the rules for picking the hostname
3189 match more closely the rules of other configuration settings
3190 where the local administrator's configuration in /etc always
3191 overrides any other settings.
3193 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
3194 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
3195 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
3196 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
3197 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
3198 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
3199 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
3200 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
3201 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
3202 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
3203 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
3204 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
3205 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
3206 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
3207 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
3208 Lindskog, WaLyong Cho, Will Woods, Zbigniew
3211 — Beijing, 2014-05-28
3215 * When restoring the screen brightness at boot, stay away from
3216 the darkest setting or from the lowest 5% of the available
3217 range, depending on which is the larger value of both. This
3218 should effectively protect the user from rebooting into a
3219 black screen, should the brightness have been set to minimum
3222 * sd-login gained a new sd_machine_get_class() call to
3223 determine the class ("vm" or "container") of a machine
3224 registered with machined.
3226 * sd-login gained new calls
3227 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
3228 to query the identity of the peer of a local AF_UNIX
3229 connection. They operate similarly to their sd_pid_get_xyz()
3232 * PID 1 will now maintain a system-wide system state engine
3233 with the states "starting", "running", "degraded",
3234 "maintenance", "stopping". These states are bound to system
3235 startup, normal runtime, runtime with at least one failed
3236 service, rescue/emergency mode and system shutdown. This
3237 state is shown in the "systemctl status" output when no unit
3238 name is passed. It is useful to determine system state, in
3239 particularly when doing so for many systems or containers at
3242 * A new command "list-machines" has been added to "systemctl"
3243 that lists all local OS containers and shows their system
3244 state (see above), if systemd runs inside of them.
3246 * systemctl gained a new "-r" switch to recursively enumerate
3247 units on all local containers, when used with the
3248 "list-unit" command (which is the default one that is
3249 executed when no parameters are specified).
3251 * The GPT automatic partition discovery logic will now honour
3252 two GPT partition flags: one may be set on a partition to
3253 cause it to be mounted read-only, and the other may be set
3254 on a partition to ignore it during automatic discovery.
3256 * Two new GPT type UUIDs have been added for automatic root
3257 partition discovery, for 32-bit and 64-bit ARM. This is not
3258 particularly useful for discovering the root directory on
3259 these architectures during bare-metal boots (since UEFI is
3260 not common there), but still very useful to allow booting of
3261 ARM disk images in nspawn with the -i option.
3263 * MAC addresses of interfaces created with nspawn's
3264 --network-interface= switch will now be generated from the
3265 machine name, and thus be stable between multiple invocations
3268 * logind will now automatically remove all IPC objects owned
3269 by a user if she or he fully logs out. This makes sure that
3270 users who are logged out cannot continue to consume IPC
3271 resources. This covers SysV memory, semaphores and message
3272 queues as well as POSIX shared memory and message
3273 queues. Traditionally, SysV and POSIX IPC had no life-cycle
3274 limits. With this functionality, that is corrected. This may
3275 be turned off by using the RemoveIPC= switch of logind.conf.
3277 * The systemd-machine-id-setup and tmpfiles tools gained a
3278 --root= switch to operate on a specific root directory,
3281 * journald can now forward logged messages to the TTYs of all
3282 logged in users ("wall"). This is the default for all
3283 emergency messages now.
3285 * A new tool systemd-journal-remote has been added to stream
3286 journal log messages across the network.
3288 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
3289 controller trees are mounted into it. Note that the
3290 directories mounted beneath it are not read-only. This is a
3291 security measure and is particularly useful because glibc
3292 actually includes a search logic to pick any tmpfs it can
3293 find to implement shm_open() if /dev/shm is not available
3294 (which it might very well be in namespaced setups).
3296 * machinectl gained a new "poweroff" command to cleanly power
3297 down a local OS container.
3299 * The PrivateDevices= unit file setting will now also drop the
3300 CAP_MKNOD capability from the capability bound set, and
3301 imply DevicePolicy=closed.
3303 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
3304 comprehensively on all long-running systemd services where
3305 this is appropriate.
3307 * systemd-udevd will now run in a disassociated mount
3308 namespace. To mount directories from udev rules, make sure to
3309 pull in mount units via SYSTEMD_WANTS properties.
3311 * The kdbus support gained support for uploading policy into
3312 the kernel. sd-bus gained support for creating "monitoring"
3313 connections that can eavesdrop into all bus communication
3314 for debugging purposes.
3316 * Timestamps may now be specified in seconds since the UNIX
3317 epoch Jan 1st, 1970 by specifying "@" followed by the value
3320 * Native tcpwrap support in systemd has been removed. tcpwrap
3321 is old code, not really maintained anymore and has serious
3322 shortcomings, and better options such as firewalls
3323 exist. For setups that require tcpwrap usage, please
3324 consider invoking your socket-activated service via tcpd,
3325 like on traditional inetd.
3327 * A new system.conf configuration option
3328 DefaultTimerAccuracySec= has been added that controls the
3329 default AccuracySec= setting of .timer units.
3331 * Timer units gained a new WakeSystem= switch. If enabled,
3332 timers configured this way will cause the system to resume
3333 from system suspend (if the system supports that, which most
3336 * Timer units gained a new Persistent= switch. If enabled,
3337 timers configured this way will save to disk when they have
3338 been last triggered. This information is then used on next
3339 reboot to possible execute overdue timer events, that
3340 could not take place because the system was powered off.
3341 This enables simple anacron-like behaviour for timer units.
3343 * systemctl's "list-timers" will now also list the time a
3344 timer unit was last triggered in addition to the next time
3345 it will be triggered.
3347 * systemd-networkd will now assign predictable IPv4LL
3348 addresses to its local interfaces.
3350 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
3351 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
3352 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
3353 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
3354 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
3355 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
3356 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
3357 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
3360 — Berlin, 2014-03-25
3364 * A new unit file setting RestrictAddressFamilies= has been
3365 added to restrict which socket address families unit
3366 processes gain access to. This takes address family names
3367 like "AF_INET" or "AF_UNIX", and is useful to minimize the
3368 attack surface of services via exotic protocol stacks. This
3369 is built on seccomp system call filters.
3371 * Two new unit file settings RuntimeDirectory= and
3372 RuntimeDirectoryMode= have been added that may be used to
3373 manage a per-daemon runtime directories below /run. This is
3374 an alternative for setting up directory permissions with
3375 tmpfiles snippets, and has the advantage that the runtime
3376 directory's lifetime is bound to the daemon runtime and that
3377 the daemon starts up with an empty directory each time. This
3378 is particularly useful when writing services that drop
3379 privileges using the User= or Group= setting.
3381 * The DeviceAllow= unit setting now supports globbing for
3382 matching against device group names.
3384 * The systemd configuration file system.conf gained new
3385 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
3386 DefaultMemoryAccounting= to globally turn on/off accounting
3387 for specific resources (cgroups) for all units. These
3388 settings may still be overridden individually in each unit
3391 * systemd-gpt-auto-generator is now able to discover /srv and
3392 root partitions in addition to /home and swap partitions. It
3393 also supports LUKS-encrypted partitions now. With this in
3394 place, automatic discovery of partitions to mount following
3395 the Discoverable Partitions Specification
3396 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
3397 is now a lot more complete. This allows booting without
3398 /etc/fstab and without root= on the kernel command line on
3399 systems prepared appropriately.
3401 * systemd-nspawn gained a new --image= switch which allows
3402 booting up disk images and Linux installations on any block
3403 device that follow the Discoverable Partitions Specification
3404 (see above). This means that installations made with
3405 appropriately updated installers may now be started and
3406 deployed using container managers, completely
3407 unmodified. (We hope that libvirt-lxc will add support for
3408 this feature soon, too.)
3410 * systemd-nspawn gained a new --network-macvlan= setting to
3411 set up a private macvlan interface for the
3412 container. Similarly, systemd-networkd gained a new
3413 Kind=macvlan setting in .netdev files.
3415 * systemd-networkd now supports configuring local addresses
3418 * A new tool systemd-network-wait-online has been added to
3419 synchronously wait for network connectivity using
3422 * The sd-bus.h bus API gained a new sd_bus_track object for
3423 tracking the life-cycle of bus peers. Note that sd-bus.h is
3424 still not a public API though (unless you specify
3425 --enable-kdbus on the configure command line, which however
3426 voids your warranty and you get no API stability guarantee).
3428 * The $XDG_RUNTIME_DIR runtime directories for each user are
3429 now individual tmpfs instances, which has the benefit of
3430 introducing separate pools for each user, with individual
3431 size limits, and thus making sure that unprivileged clients
3432 can no longer negatively impact the system or other users by
3433 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
3434 RuntimeDirectorySize= has been introduced that allows
3435 controlling the default size limit for all users. It
3436 defaults to 10% of the available physical memory. This is no
3437 replacement for quotas on tmpfs though (which the kernel
3438 still does not support), as /dev/shm and /tmp are still
3439 shared resources used by both the system and unprivileged
3442 * logind will now automatically turn off automatic suspending
3443 on laptop lid close when more than one display is
3444 connected. This was previously expected to be implemented
3445 individually in desktop environments (such as GNOME),
3446 however has been added to logind now, in order to fix a
3447 boot-time race where a desktop environment might not have
3448 been started yet and thus not been able to take an inhibitor
3449 lock at the time where logind already suspends the system
3450 due to a closed lid.
3452 * logind will now wait at least 30s after each system
3453 suspend/resume cycle, and 3min after system boot before
3454 suspending the system due to a closed laptop lid. This
3455 should give USB docking stations and similar enough time to
3456 be probed and configured after system resume and boot in
3457 order to then act as suspend blocker.
3459 * systemd-run gained a new --property= setting which allows
3460 initialization of resource control properties (and others)
3461 for the created scope or service unit. Example: "systemd-run
3462 --property=BlockIOWeight=10 updatedb" may be used to run
3463 updatedb at a low block IO scheduling weight.
3465 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
3466 now also work in --scope mode.
3468 * When systemd is compiled with kdbus support, basic support
3469 for enforced policies is now in place. (Note that enabling
3470 kdbus still voids your warranty and no API compatibility
3473 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
3474 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
3475 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
3476 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
3477 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
3478 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
3479 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
3480 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
3481 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
3482 Zbigniew Jędrzejewski-Szmek
3484 — Berlin, 2014-03-12
3488 * systemd will now relabel /dev after loading the SMACK policy
3489 according to SMACK rules.
3491 * A new unit file option AppArmorProfile= has been added to
3492 set the AppArmor profile for the processes of a unit.
3494 * A new condition check ConditionArchitecture= has been added
3495 to conditionalize units based on the system architecture, as
3496 reported by uname()'s "machine" field.
3498 * systemd-networkd now supports matching on the system
3499 virtualization, architecture, kernel command line, host name
3502 * logind is now a lot more aggressive when suspending the
3503 machine due to a closed laptop lid. Instead of acting only
3504 on the lid close action, it will continuously watch the lid
3505 status and act on it. This is useful for laptops where the
3506 power button is on the outside of the chassis so that it can
3507 be reached without opening the lid (such as the Lenovo
3508 Yoga). On those machines, logind will now immediately
3509 re-suspend the machine if the power button has been
3510 accidentally pressed while the laptop was suspended and in a
3511 backpack or similar.
3513 * logind will now watch SW_DOCK switches and inhibit reaction
3514 to the lid switch if it is pressed. This means that logind
3515 will not suspend the machine anymore if the lid is closed
3516 and the system is docked, if the laptop supports SW_DOCK
3517 notifications via the input layer. Note that ACPI docking
3518 stations do not generate this currently. Also note that this
3519 logic is usually not fully sufficient and Desktop
3520 Environments should take a lid switch inhibitor lock when an
3521 external display is connected, as systemd will not watch
3524 * nspawn will now make use of the devices cgroup controller by
3525 default, and only permit creation of and access to the usual
3526 API device nodes like /dev/null or /dev/random, as well as
3527 access to (but not creation of) the pty devices.
3529 * We will now ship a default .network file for
3530 systemd-networkd that automatically configures DHCP for
3531 network interfaces created by nspawn's --network-veth or
3532 --network-bridge= switches.
3534 * systemd will now understand the usual M, K, G, T suffixes
3535 according to SI conventions (i.e. to the base 1000) when
3536 referring to throughput and hardware metrics. It will stay
3537 with IEC conventions (i.e. to the base 1024) for software
3538 metrics, according to what is customary according to
3539 Wikipedia. We explicitly document which base applies for
3540 each configuration option.
3542 * The DeviceAllow= setting in unit files now supports a syntax
3543 to whitelist an entire group of devices node majors at once,
3544 based on the /proc/devices listing. For example, with the
3545 string "char-pts", it is now possible to whitelist all
3546 current and future pseudo-TTYs at once.
3548 * sd-event learned a new "post" event source. Event sources of
3549 this type are triggered by the dispatching of any event
3550 source of a type that is not "post". This is useful for
3551 implementing clean-up and check event sources that are
3552 triggered by other work being done in the program.
3554 * systemd-networkd is no longer statically enabled, but uses
3555 the usual [Install] sections so that it can be
3556 enabled/disabled using systemctl. It still is enabled by
3559 * When creating a veth interface pair with systemd-nspawn, the
3560 host side will now be prefixed with "vb-" if
3561 --network-bridge= is used, and with "ve-" if --network-veth
3562 is used. This way, it is easy to distinguish these cases on
3563 the host, for example to apply different configuration to
3564 them with systemd-networkd.
3566 * The compatibility libraries for libsystemd-journal.so,
3567 libsystem-id128.so, libsystemd-login.so and
3568 libsystemd-daemon.so do not make use of IFUNC
3569 anymore. Instead, we now build libsystemd.so multiple times
3570 under these alternative names. This means that the footprint
3571 is drastically increased, but given that these are
3572 transitional compatibility libraries, this should not matter
3573 much. This change has been made necessary to support the ARM
3574 platform for these compatibility libraries, as the ARM
3575 toolchain is not really at the same level as the toolchain
3576 for other architectures like x86 and does not support
3577 IFUNC. Please make sure to use --enable-compat-libs only
3578 during a transitional period!
3580 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
3581 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
3582 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
3583 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
3584 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
3585 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
3586 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
3587 Zbigniew Jędrzejewski-Szmek
3589 — Berlin, 2014-02-24
3593 * A new component "systemd-networkd" has been added that can
3594 be used to configure local network interfaces statically or
3595 via DHCP. It is capable of bringing up bridges, VLANs, and
3596 bonding. Currently, no hook-ups for interactive network
3597 configuration are provided. Use this for your initrd,
3598 container, embedded, or server setup if you need a simple,
3599 yet powerful, network configuration solution. This
3600 configuration subsystem is quite nifty, as it allows wildcard
3601 hotplug matching in interfaces. For example, with a single
3602 configuration snippet, you can configure that all Ethernet
3603 interfaces showing up are automatically added to a bridge,
3604 or similar. It supports link-sensing and more.
3606 * A new tool "systemd-socket-proxyd" has been added which can
3607 act as a bidirectional proxy for TCP sockets. This is
3608 useful for adding socket activation support to services that
3609 do not actually support socket activation, including virtual
3610 machines and the like.
3612 * Add a new tool to save/restore rfkill state on
3615 * Save/restore state of keyboard backlights in addition to
3616 display backlights on shutdown/boot.
3618 * udev learned a new SECLABEL{} construct to label device
3619 nodes with a specific security label when they appear. For
3620 now, only SECLABEL{selinux} is supported, but the syntax is
3621 prepared for additional security frameworks.
3623 * udev gained a new scheme to configure link-level attributes
3624 from files in /etc/systemd/network/*.link. These files can
3625 match against MAC address, device path, driver name and type,
3626 and will apply attributes like the naming policy, link speed,
3627 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
3628 address assignment policy (randomized, ...).
3630 * The configuration of network interface naming rules for
3631 "permanent interface names" has changed: a new NamePolicy=
3632 setting in the [Link] section of .link files determines the
3633 priority of possible naming schemes (onboard, slot, MAC,
3634 path). The default value of this setting is determined by
3635 /usr/lib/net/links/99-default.link. Old
3636 80-net-name-slot.rules udev configuration file has been
3637 removed, so local configuration overriding this file should
3638 be adapated to override 99-default.link instead.
3640 * When the User= switch is used in a unit file, also
3641 initialize $SHELL= based on the user database entry.
3643 * systemd no longer depends on libdbus. All communication is
3644 now done with sd-bus, systemd's low-level bus library
3647 * kdbus support has been added to PID 1 itself. When kdbus is
3648 enabled, this causes PID 1 to set up the system bus and
3649 enable support for a new ".busname" unit type that
3650 encapsulates bus name activation on kdbus. It works a little
3651 bit like ".socket" units, except for bus names. A new
3652 generator has been added that converts classic dbus1 service
3653 activation files automatically into native systemd .busname
3656 * sd-bus: add a light-weight vtable implementation that allows
3657 defining objects on the bus with a simple static const
3658 vtable array of its methods, signals and properties.
3660 * systemd will not generate or install static dbus
3661 introspection data anymore to /usr/share/dbus-1/interfaces,
3662 as the precise format of these files is unclear, and
3663 nothing makes use of it.
3665 * A proxy daemon is now provided to proxy clients connecting
3666 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
3667 compatibility with classic D-Bus.
3669 * A bus driver implementation has been added that supports the
3670 classic D-Bus bus driver calls on kdbus, also for
3671 compatibility purposes.
3673 * A new API "sd-event.h" has been added that implements a
3674 minimal event loop API built around epoll. It provides a
3675 couple of features that direct epoll usage is lacking:
3676 prioritization of events, scales to large numbers of timer
3677 events, per-event timer slack (accuracy), system-wide
3678 coalescing of timer events, exit handlers, watchdog
3679 supervision support using systemd's sd_notify() API, child
3682 * A new API "sd-rntl.h" has been added that provides an API
3683 around the route netlink interface of the kernel, similar in
3684 style to "sd-bus.h".
3686 * A new API "sd-dhcp-client.h" has been added that provides a
3687 small DHCPv4 client-side implementation. This is used by
3690 * There is a new kernel command line option
3691 "systemd.restore_state=0|1". When set to "0", none of the
3692 systemd tools will restore saved runtime state to hardware
3693 devices. More specifically, the rfkill and backlight states
3696 * The FsckPassNo= compatibility option in mount/service units
3697 has been removed. The fstab generator will now add the
3698 necessary dependencies automatically, and does not require
3699 PID1's support for that anymore.
3701 * journalctl gained a new switch, --list-boots, that lists
3702 recent boots with their times and boot IDs.
3704 * The various tools like systemctl, loginctl, timedatectl,
3705 busctl, systemd-run, ... have gained a new switch "-M" to
3706 connect to a specific, local OS container (as direct
3707 connection, without requiring SSH). This works on any
3708 container that is registered with machined, such as those
3709 created by libvirt-lxc or nspawn.
3711 * systemd-run and systemd-analyze also gained support for "-H"
3712 to connect to remote hosts via SSH. This is particularly
3713 useful for systemd-run because it enables queuing of jobs
3714 onto remote systems.
3716 * machinectl gained a new command "login" to open a getty
3717 login in any local container. This works with any container
3718 that is registered with machined (such as those created by
3719 libvirt-lxc or nspawn), and which runs systemd inside.
3721 * machinectl gained a new "reboot" command that may be used to
3722 trigger a reboot on a specific container that is registered
3723 with machined. This works on any container that runs an init
3724 system of some kind.
3726 * systemctl gained a new "list-timers" command to print a nice
3727 listing of installed timer units with the times they elapse
3730 * Alternative reboot() parameters may now be specified on the
3731 "systemctl reboot" command line and are passed to the
3732 reboot() system call.
3734 * systemctl gained a new --job-mode= switch to configure the
3735 mode to queue a job with. This is a more generic version of
3736 --fail, --irreversible, and --ignore-dependencies, which are
3737 still available but not advertised anymore.
3739 * /etc/systemd/system.conf gained new settings to configure
3740 various default timeouts of units, as well as the default
3741 start limit interval and burst. These may still be overridden
3744 * PID1 will now export on the bus profile data of the security
3745 policy upload process (such as the SELinux policy upload to
3748 * journald: when forwarding logs to the console, include
3749 timestamps (following the setting in
3750 /sys/module/printk/parameters/time).
3752 * OnCalendar= in timer units now understands the special
3753 strings "yearly" and "annually". (Both are equivalent)
3755 * The accuracy of timer units is now configurable with the new
3756 AccuracySec= setting. It defaults to 1min.
3758 * A new dependency type JoinsNamespaceOf= has been added that
3759 allows running two services within the same /tmp and network
3760 namespace, if PrivateNetwork= or PrivateTmp= are used.
3762 * A new command "cat" has been added to systemctl. It outputs
3763 the original unit file of a unit, and concatenates the
3764 contents of additional "drop-in" unit file snippets, so that
3765 the full configuration is shown.
3767 * systemctl now supports globbing on the various "list-xyz"
3768 commands, like "list-units" or "list-sockets", as well as on
3769 those commands which take multiple unit names.
3771 * journalctl's --unit= switch gained support for globbing.
3773 * All systemd daemons now make use of the watchdog logic so
3774 that systemd automatically notices when they hang.
3776 * If the $container_ttys environment variable is set,
3777 getty-generator will automatically spawn a getty for each
3778 listed tty. This is useful for container managers to request
3779 login gettys to be spawned on as many ttys as needed.
3781 * %h, %s, %U specifier support is not available anymore when
3782 used in unit files for PID 1. This is because NSS calls are
3783 not safe from PID 1. They stay available for --user
3784 instances of systemd, and as special case for the root user.
3786 * loginctl gained a new "--no-legend" switch to turn off output
3789 * The "sd-login.h" API gained three new calls:
3790 sd_session_is_remote(), sd_session_get_remote_user(),
3791 sd_session_get_remote_host() to query information about
3794 * The udev hardware database now also carries vendor/product
3795 information of SDIO devices.
3797 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
3798 determine whether watchdog notifications are requested by
3801 * Socket-activated per-connection services now include a
3802 short description of the connection parameters in the
3805 * tmpfiles gained a new "--boot" option. When this is not used,
3806 only lines where the command character is not suffixed with
3807 "!" are executed. When this option is specified, those
3808 options are executed too. This partitions tmpfiles
3809 directives into those that can be safely executed at any
3810 time, and those which should be run only at boot (for
3811 example, a line that creates /run/nologin).
3813 * A new API "sd-resolve.h" has been added which provides a simple
3814 asynchronous wrapper around glibc NSS host name resolution
3815 calls, such as getaddrinfo(). In contrast to glibc's
3816 getaddrinfo_a(), it does not use signals. In contrast to most
3817 other asynchronous name resolution libraries, this one does
3818 not reimplement DNS, but reuses NSS, so that alternate
3819 host name resolution systems continue to work, such as mDNS,
3820 LDAP, etc. This API is based on libasyncns, but it has been
3821 cleaned up for inclusion in systemd.
3823 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
3824 "sd-daemon.h" are no longer found in individual libraries
3825 libsystemd-journal.so, libsystemd-login.so,
3826 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
3827 merged them into a single library, libsystemd.so, which
3828 provides all symbols. The reason for this is cyclic
3829 dependencies, as these libraries tend to use each other's
3830 symbols. So far, we have managed to workaround that by linking
3831 a copy of a good part of our code into each of these
3832 libraries again and again, which, however, makes certain
3833 things hard to do, like sharing static variables. Also, it
3834 substantially increases footprint. With this change, there
3835 is only one library for the basic APIs systemd
3836 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
3837 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
3838 library as well, however are subject to the --enable-kdbus
3839 switch (see below). Note that "sd-dhcp-client.h" is not part
3840 of this library (this is because it only consumes, never
3841 provides, services of/to other APIs). To make the transition
3842 easy from the separate libraries to the unified one, we
3843 provide the --enable-compat-libs compile-time switch which
3844 will generate stub libraries that are compatible with the
3845 old ones but redirect all calls to the new one.
3847 * All of the kdbus logic and the new APIs "sd-bus.h",
3848 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
3849 and "sd-utf8.h" are compile-time optional via the
3850 "--enable-kdbus" switch, and they are not compiled in by
3851 default. To make use of kdbus, you have to explicitly enable
3852 the switch. Note however, that neither the kernel nor the
3853 userspace API for all of this is considered stable yet. We
3854 want to maintain the freedom to still change the APIs for
3855 now. By specifying this build-time switch, you acknowledge
3856 that you are aware of the instability of the current
3859 * Also, note that while kdbus is pretty much complete,
3860 it lacks one thing: proper policy support. This means you
3861 can build a fully working system with all features; however,
3862 it will be highly insecure. Policy support will be added in
3863 one of the next releases, at the same time that we will
3864 declare the APIs stable.
3866 * When the kernel command line argument "kdbus" is specified,
3867 systemd will automatically load the kdbus.ko kernel module. At
3868 this stage of development, it is only useful for testing kdbus
3869 and should not be used in production. Note: if "--enable-kdbus"
3870 is specified, and the kdbus.ko kernel module is available, and
3871 "kdbus" is added to the kernel command line, the entire system
3872 runs with kdbus instead of dbus-daemon, with the above mentioned
3873 problem of missing the system policy enforcement. Also a future
3874 version of kdbus.ko or a newer systemd will not be compatible with
3875 each other, and will unlikely be able to boot the machine if only
3876 one of them is updated.
3878 * systemctl gained a new "import-environment" command which
3879 uploads the caller's environment (or parts thereof) into the
3880 service manager so that it is inherited by services started
3881 by the manager. This is useful to upload variables like
3882 $DISPLAY into the user service manager.
3884 * A new PrivateDevices= switch has been added to service units
3885 which allows running a service with a namespaced /dev
3886 directory that does not contain any device nodes for
3887 physical devices. More specifically, it only includes devices
3888 such as /dev/null, /dev/urandom, and /dev/zero which are API
3891 * logind has been extended to support behaviour like VT
3892 switching on seats that do not support a VT. This makes
3893 multi-session available on seats that are not the first seat
3894 (seat0), and on systems where kernel support for VTs has
3895 been disabled at compile-time.
3897 * If a process holds a delay lock for system sleep or shutdown
3898 and fails to release it in time, we will now log its
3899 identity. This makes it easier to identify processes that
3900 cause slow suspends or power-offs.
3902 * When parsing /etc/crypttab, support for a new key-slot=
3903 option as supported by Debian is added. It allows indicating
3904 which LUKS slot to use on disk, speeding up key loading.
3906 * The sd_journald_sendv() API call has been checked and
3907 officially declared to be async-signal-safe so that it may
3908 be invoked from signal handlers for logging purposes.
3910 * Boot-time status output is now enabled automatically after a
3911 short timeout if boot does not progress, in order to give
3912 the user an indication what she or he is waiting for.
3914 * The boot-time output has been improved to show how much time
3915 remains until jobs expire.
3917 * The KillMode= switch in service units gained a new possible
3918 value "mixed". If set, and the unit is shut down, then the
3919 initial SIGTERM signal is sent only to the main daemon
3920 process, while the following SIGKILL signal is sent to
3921 all remaining processes of the service.
3923 * When a scope unit is registered, a new property "Controller"
3924 may be set. If set to a valid bus name, systemd will send a
3925 RequestStop() signal to this name when it would like to shut
3926 down the scope. This may be used to hook manager logic into
3927 the shutdown logic of scope units. Also, scope units may now
3928 be put in a special "abandoned" state, in which case the
3929 manager process which created them takes no further
3930 responsibilities for it.
3932 * When reading unit files, systemd will now verify
3933 the access mode of these files, and warn about certain
3934 suspicious combinations. This has been added to make it
3935 easier to track down packaging bugs where unit files are
3936 marked executable or world-writable.
3938 * systemd-nspawn gained a new "--setenv=" switch to set
3939 container-wide environment variables. The similar option in
3940 systemd-activate was renamed from "--environment=" to
3941 "--setenv=" for consistency.
3943 * systemd-nspawn has been updated to create a new kdbus domain
3944 for each container that is invoked, thus allowing each
3945 container to have its own set of system and user buses,
3946 independent of the host.
3948 * systemd-nspawn gained a new --drop-capability= switch to run
3949 the container with less capabilities than the default. Both
3950 --drop-capability= and --capability= now take the special
3951 string "all" for dropping or keeping all capabilities.
3953 * systemd-nspawn gained new switches for executing containers
3954 with specific SELinux labels set.
3956 * systemd-nspawn gained a new --quiet switch to not generate
3957 any additional output but the container's own console
3960 * systemd-nspawn gained a new --share-system switch to run a
3961 container without PID namespacing enabled.
3963 * systemd-nspawn gained a new --register= switch to control
3964 whether the container is registered with systemd-machined or
3965 not. This is useful for containers that do not run full
3966 OS images, but only specific apps.
3968 * systemd-nspawn gained a new --keep-unit which may be used
3969 when invoked as the only program from a service unit, and
3970 results in registration of the unit service itself in
3971 systemd-machined, instead of a newly opened scope unit.
3973 * systemd-nspawn gained a new --network-interface= switch for
3974 moving arbitrary interfaces to the container. The new
3975 --network-veth switch creates a virtual Ethernet connection
3976 between host and container. The new --network-bridge=
3977 switch then allows assigning the host side of this virtual
3978 Ethernet connection to a bridge device.
3980 * systemd-nspawn gained a new --personality= switch for
3981 setting the kernel personality for the container. This is
3982 useful when running a 32-bit container on a 64-bit host. A
3983 similar option Personality= is now also available for service
3986 * logind will now also track a "Desktop" identifier for each
3987 session which encodes the desktop environment of it. This is
3988 useful for desktop environments that want to identify
3989 multiple running sessions of itself easily.
3991 * A new SELinuxContext= setting for service units has been
3992 added that allows setting a specific SELinux execution
3993 context for a service.
3995 * Most systemd client tools will now honour $SYSTEMD_LESS for
3996 settings of the "less" pager. By default, these tools will
3997 override $LESS to allow certain operations to work, such as
3998 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
3999 influence this logic.
4001 * systemd's "seccomp" hook-up has been changed to make use of
4002 the libseccomp library instead of using its own
4003 implementation. This has benefits for portability among
4006 * For usage together with SystemCallFilter=, a new
4007 SystemCallErrorNumber= setting has been introduced that
4008 allows configuration of a system error number to be returned
4009 on filtered system calls, instead of immediately killing the
4010 process. Also, SystemCallArchitectures= has been added to
4011 limit access to system calls of a particular architecture
4012 (in order to turn off support for unused secondary
4013 architectures). There is also a global
4014 SystemCallArchitectures= setting in system.conf now to turn
4015 off support for non-native system calls system-wide.
4017 * systemd requires a kernel with a working name_to_handle_at(),
4018 please see the kernel config requirements in the README file.
4020 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
4021 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
4022 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
4023 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
4024 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
4025 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
4026 Elia Pinto, Florian Weimer, George McCollister, Goffredo
4027 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
4028 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
4029 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
4030 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
4031 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
4032 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
4033 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
4034 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
4035 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
4036 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
4037 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
4038 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
4039 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
4040 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
4041 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
4042 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
4043 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
4045 — Berlin, 2014-02-20
4049 * logind has gained support for facilitating privileged input
4050 and drm device access for unprivileged clients. This work is
4051 useful to allow Wayland display servers (and similar
4052 programs, such as kmscon) to run under the user's ID and
4053 access input and drm devices which are normally
4054 protected. When this is used (and the kernel is new enough)
4055 logind will "mute" IO on the file descriptors passed to
4056 Wayland as long as it is in the background and "unmute" it
4057 if it returns into the foreground. This allows secure
4058 session switching without allowing background sessions to
4059 eavesdrop on input and display data. This also introduces
4060 session switching support if VT support is turned off in the
4061 kernel, and on seats that are not seat0.
4063 * A new kernel command line option luks.options= is understood
4064 now which allows specifying LUKS options for usage for LUKS
4065 encrypted partitions specified with luks.uuid=.
4067 * tmpfiles.d(5) snippets may now use specifier expansion in
4068 path names. More specifically %m, %b, %H, %v, are now
4069 replaced by the local machine id, boot id, hostname, and
4070 kernel version number.
4072 * A new tmpfiles.d(5) command "m" has been introduced which
4073 may be used to change the owner/group/access mode of a file
4074 or directory if it exists, but do nothing if it does not.
4076 * This release removes high-level support for the
4077 MemorySoftLimit= cgroup setting. The underlying kernel
4078 cgroup attribute memory.soft_limit= is currently badly
4079 designed and likely to be removed from the kernel API in its
4080 current form, hence we should not expose it for now.
4082 * The memory.use_hierarchy cgroup attribute is now enabled for
4083 all cgroups systemd creates in the memory cgroup
4084 hierarchy. This option is likely to be come the built-in
4085 default in the kernel anyway, and the non-hierarchical mode
4086 never made much sense in the intrinsically hierarchical
4089 * A new field _SYSTEMD_SLICE= is logged along with all journal
4090 messages containing the slice a message was generated
4091 from. This is useful to allow easy per-customer filtering of
4092 logs among other things.
4094 * systemd-journald will no longer adjust the group of journal
4095 files it creates to the "systemd-journal" group. Instead we
4096 rely on the journal directory to be owned by the
4097 "systemd-journal" group, and its setgid bit set, so that the
4098 kernel file system layer will automatically enforce that
4099 journal files inherit this group assignment. The reason for
4100 this change is that we cannot allow NSS look-ups from
4101 journald which would be necessary to resolve
4102 "systemd-journal" to a numeric GID, because this might
4103 create deadlocks if NSS involves synchronous queries to
4104 other daemons (such as nscd, or sssd) which in turn are
4105 logging clients of journald and might block on it, which
4106 would then dead lock. A tmpfiles.d(5) snippet included in
4107 systemd will make sure the setgid bit and group are
4108 properly set on the journal directory if it exists on every
4109 boot. However, we recommend adjusting it manually after
4110 upgrades too (or from RPM scriptlets), so that the change is
4111 not delayed until next reboot.
4113 * Backlight and random seed files in /var/lib/ have moved into
4114 the /var/lib/systemd/ directory, in order to centralize all
4115 systemd generated files in one directory.
4117 * Boot time performance measurements (as displayed by
4118 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
4119 performance information if that's available to determine how
4120 much time BIOS and boot loader initialization required. With
4121 a sufficiently new BIOS you hence no longer need to boot
4122 with Gummiboot to get access to such information.
4124 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
4125 Cristian Rodríguez, Dave Reisner, David Herrmann, David
4126 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
4127 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
4128 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
4129 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
4130 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
4132 — Berlin, 2013-10-02
4136 * The Restart= option for services now understands a new
4137 on-watchdog setting, which will restart the service
4138 automatically if the service stops sending out watchdog keep
4139 alive messages (as configured with WatchdogSec=).
4141 * The getty generator (which is responsible for bringing up a
4142 getty on configured serial consoles) will no longer only
4143 start a getty on the primary kernel console but on all
4144 others, too. This makes the order in which console= is
4145 specified on the kernel command line less important.
4147 * libsystemd-logind gained a new sd_session_get_vt() call to
4148 retrieve the VT number of a session.
4150 * If the option "tries=0" is set for an entry of /etc/crypttab
4151 its passphrase is queried indefinitely instead of any
4152 maximum number of tries.
4154 * If a service with a configure PID file terminates its PID
4155 file will now be removed automatically if it still exists
4156 afterwards. This should put an end to stale PID files.
4158 * systemd-run will now also take relative binary path names
4159 for execution and no longer insists on absolute paths.
4161 * InaccessibleDirectories= and ReadOnlyDirectories= now take
4162 paths that are optionally prefixed with "-" to indicate that
4163 it should not be considered a failure if they do not exist.
4165 * journalctl -o (and similar commands) now understands a new
4166 output mode "short-precise", it is similar to "short" but
4167 shows timestamps with usec accuracy.
4169 * The option "discard" (as known from Debian) is now
4170 synonymous to "allow-discards" in /etc/crypttab. In fact,
4171 "discard" is preferred now (since it is easier to remember
4174 * Some licensing clean-ups were made, so that more code is now
4175 LGPL-2.1 licensed than before.
4177 * A minimal tool to save/restore the display backlight
4178 brightness across reboots has been added. It will store the
4179 backlight setting as late as possible at shutdown, and
4180 restore it as early as possible during reboot.
4182 * A logic to automatically discover and enable home and swap
4183 partitions on GPT disks has been added. With this in place
4184 /etc/fstab becomes optional for many setups as systemd can
4185 discover certain partitions located on the root disk
4186 automatically. Home partitions are recognized under their
4187 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
4188 partitions are recognized under their GPT type ID
4189 0657fd6da4ab43c484e50933c84b4f4f.
4191 * systemd will no longer pass any environment from the kernel
4192 or initrd to system services. If you want to set an
4193 environment for all services, do so via the kernel command
4194 line systemd.setenv= assignment.
4196 * The systemd-sysctl tool no longer natively reads the file
4197 /etc/sysctl.conf. If desired, the file should be symlinked
4198 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
4199 legacy support by a symlink rather than built-in code, it
4200 also makes the otherwise hidden order of application of the
4201 different files visible. (Note that this partly reverts to a
4202 pre-198 application order of sysctl knobs!)
4204 * The "systemctl set-log-level" and "systemctl dump" commands
4205 have been moved to systemd-analyze.
4207 * systemd-run learned the new --remain-after-exit switch,
4208 which causes the scope unit not to be cleaned up
4209 automatically after the process terminated.
4211 * tmpfiles learned a new --exclude-prefix= switch to exclude
4212 certain paths from operation.
4214 * journald will now automatically flush all messages to disk
4215 as soon as a message at the log level CRIT, ALERT or EMERG
4218 Contributions from: Andrew Cook, Brandon Philips, Christian
4219 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
4220 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
4221 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
4222 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
4223 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
4224 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
4225 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
4226 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
4227 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
4228 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
4229 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
4230 William Giokas, Zbigniew Jędrzejewski-Szmek
4232 — Berlin, 2013-09-13
4236 * The documentation has been updated to cover the various new
4237 concepts introduced with 205.
4239 * Unit files now understand the new %v specifier which
4240 resolves to the kernel version string as returned by "uname
4243 * systemctl now supports filtering the unit list output by
4244 load state, active state and sub state, using the new
4247 * "systemctl status" will now show the results of the
4248 condition checks (like ConditionPathExists= and similar) of
4249 the last start attempts of the unit. They are also logged to
4252 * "journalctl -b" may now be used to look for boot output of a
4253 specific boot. Try "journalctl -b -1" for the previous boot,
4254 but the syntax is substantially more powerful.
4256 * "journalctl --show-cursor" has been added which prints the
4257 cursor string the last shown log line. This may then be used
4258 with the new "journalctl --after-cursor=" switch to continue
4259 browsing logs from that point on.
4261 * "journalctl --force" may now be used to force regeneration
4264 * Creation of "dead" device nodes has been moved from udev
4265 into kmod and tmpfiles. Previously, udev would read the kmod
4266 databases to pre-generate dead device nodes based on meta
4267 information contained in kernel modules, so that these would
4268 be auto-loaded on access rather then at boot. As this
4269 does not really have much to do with the exposing actual
4270 kernel devices to userspace this has always been slightly
4271 alien in the udev codebase. Following the new scheme kmod
4272 will now generate a runtime snippet for tmpfiles from the
4273 module meta information and it now is tmpfiles' job to the
4274 create the nodes. This also allows overriding access and
4275 other parameters for the nodes using the usual tmpfiles
4276 facilities. As side effect this allows us to remove the
4277 CAP_SYS_MKNOD capability bit from udevd entirely.
4279 * logind's device ACLs may now be applied to these "dead"
4280 devices nodes too, thus finally allowing managed access to
4281 devices such as /dev/snd/sequencer whithout loading the
4282 backing module right-away.
4284 * A new RPM macro has been added that may be used to apply
4285 tmpfiles configuration during package installation.
4287 * systemd-detect-virt and ConditionVirtualization= now can
4288 detect User-Mode-Linux machines (UML).
4290 * journald will now implicitly log the effective capabilities
4291 set of processes in the message metadata.
4293 * systemd-cryptsetup has gained support for TrueCrypt volumes.
4295 * The initrd interface has been simplified (more specifically,
4296 support for passing performance data via environment
4297 variables and fsck results via files in /run has been
4298 removed). These features were non-essential, and are
4299 nowadays available in a much nicer way by having systemd in
4300 the initrd serialize its state and have the hosts systemd
4301 deserialize it again.
4303 * The udev "keymap" data files and tools to apply keyboard
4304 specific mappings of scan to key codes, and force-release
4305 scan code lists have been entirely replaced by a udev
4306 "keyboard" builtin and a hwdb data file.
4308 * systemd will now honour the kernel's "quiet" command line
4309 argument also during late shutdown, resulting in a
4310 completely silent shutdown when used.
4312 * There's now an option to control the SO_REUSEPORT socket
4313 option in .socket units.
4315 * Instance units will now automatically get a per-template
4316 subslice of system.slice unless something else is explicitly
4317 configured. For example, instances of sshd@.service will now
4318 implicitly be placed in system-sshd.slice rather than
4319 system.slice as before.
4321 * Test coverage support may now be enabled at build time.
4323 Contributions from: Dave Reisner, Frederic Crozat, Harald
4324 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
4325 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
4326 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
4327 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
4328 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
4329 Giokas, Zbigniew Jędrzejewski-Szmek
4331 — Berlin, 2013-07-23
4335 * Two new unit types have been introduced:
4337 Scope units are very similar to service units, however, are
4338 created out of pre-existing processes — instead of PID 1
4339 forking off the processes. By using scope units it is
4340 possible for system services and applications to group their
4341 own child processes (worker processes) in a powerful way
4342 which then maybe used to organize them, or kill them
4343 together, or apply resource limits on them.
4345 Slice units may be used to partition system resources in an
4346 hierarchical fashion and then assign other units to them. By
4347 default there are now three slices: system.slice (for all
4348 system services), user.slice (for all user sessions),
4349 machine.slice (for VMs and containers).
4351 Slices and scopes have been introduced primarily in
4352 context of the work to move cgroup handling to a
4353 single-writer scheme, where only PID 1
4354 creates/removes/manages cgroups.
4356 * There's a new concept of "transient" units. In contrast to
4357 normal units these units are created via an API at runtime,
4358 not from configuration from disk. More specifically this
4359 means it is now possible to run arbitrary programs as
4360 independent services, with all execution parameters passed
4361 in via bus APIs rather than read from disk. Transient units
4362 make systemd substantially more dynamic then it ever was,
4363 and useful as a general batch manager.
4365 * logind has been updated to make use of scope and slice units
4366 for managing user sessions. As a user logs in he will get
4367 his own private slice unit, to which all sessions are added
4368 as scope units. We also added support for automatically
4369 adding an instance of user@.service for the user into the
4370 slice. Effectively logind will no longer create cgroup
4371 hierarchies on its own now, it will defer entirely to PID 1
4372 for this by means of scope, service and slice units. Since
4373 user sessions this way become entities managed by PID 1
4374 the output of "systemctl" is now a lot more comprehensive.
4376 * A new mini-daemon "systemd-machined" has been added which
4377 may be used by virtualization managers to register local
4378 VMs/containers. nspawn has been updated accordingly, and
4379 libvirt will be updated shortly. machined will collect a bit
4380 of meta information about the VMs/containers, and assign
4381 them their own scope unit (see above). The collected
4382 meta-data is then made available via the "machinectl" tool,
4383 and exposed in "ps" and similar tools. machined/machinectl
4384 is compile-time optional.
4386 * As discussed earlier, the low-level cgroup configuration
4387 options ControlGroup=, ControlGroupModify=,
4388 ControlGroupPersistent=, ControlGroupAttribute= have been
4389 removed. Please use high-level attribute settings instead as
4390 well as slice units.
4392 * A new bus call SetUnitProperties() has been added to alter
4393 various runtime parameters of a unit. This is primarily
4394 useful to alter cgroup parameters dynamically in a nice way,
4395 but will be extended later on to make more properties
4396 modifiable at runtime. systemctl gained a new set-properties
4397 command that wraps this call.
4399 * A new tool "systemd-run" has been added which can be used to
4400 run arbitrary command lines as transient services or scopes,
4401 while configuring a number of settings via the command
4402 line. This tool is currently very basic, however already
4403 very useful. We plan to extend this tool to even allow
4404 queuing of execution jobs with time triggers from the
4405 command line, similar in fashion to "at".
4407 * nspawn will now inform the user explicitly that kernels with
4408 audit enabled break containers, and suggest the user to turn
4411 * Support for detecting the IMA and AppArmor security
4412 frameworks with ConditionSecurity= has been added.
4414 * journalctl gained a new "-k" switch for showing only kernel
4415 messages, mimicking dmesg output; in addition to "--user"
4416 and "--system" switches for showing only user's own logs
4419 * systemd-delta can now show information about drop-in
4420 snippets extending unit files.
4422 * libsystemd-bus has been substantially updated but is still
4423 not available as public API.
4425 * systemd will now look for the "debug" argument on the kernel
4426 command line and enable debug logging, similar to what
4427 "systemd.log_level=debug" already did before.
4429 * "systemctl set-default", "systemctl get-default" has been
4430 added to configure the default.target symlink, which
4431 controls what to boot into by default.
4433 * "systemctl set-log-level" has been added as a convenient
4434 way to raise and lower systemd logging threshold.
4436 * "systemd-analyze plot" will now show the time the various
4437 generators needed for execution, as well as information
4438 about the unit file loading.
4440 * libsystemd-journal gained a new sd_journal_open_files() call
4441 for opening specific journal files. journactl also gained a
4442 new switch to expose this new functionality. Previously we
4443 only supported opening all files from a directory, or all
4444 files from the system, as opening individual files only is
4445 racy due to journal file rotation.
4447 * systemd gained the new DefaultEnvironment= setting in
4448 /etc/systemd/system.conf to set environment variables for
4451 * If a privileged process logs a journal message with the
4452 OBJECT_PID= field set, then journald will automatically
4453 augment this with additional OBJECT_UID=, OBJECT_GID=,
4454 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
4455 system services want to log events about specific client
4456 processes. journactl/systemctl has been updated to make use
4457 of this information if all log messages regarding a specific
4460 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
4461 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
4462 Reisner, David Coppa, David King, David Strauss, Eelco
4463 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
4464 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
4465 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
4466 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
4467 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
4468 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
4469 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
4470 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
4471 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
4472 Łukasz Stelmach, 장동준
4476 * The Python bindings gained some minimal support for the APIs
4477 exposed by libsystemd-logind.
4479 * ConditionSecurity= gained support for detecting SMACK. Since
4480 this condition already supports SELinux and AppArmor we only
4481 miss IMA for this. Patches welcome!
4483 Contributions from: Karol Lewandowski, Lennart Poettering,
4484 Zbigniew Jędrzejewski-Szmek
4488 * systemd-nspawn will now create /etc/resolv.conf if
4489 necessary, before bind-mounting the host's file onto it.
4491 * systemd-nspawn will now store meta information about a
4492 container on the container's cgroup as extended attribute
4493 fields, including the root directory.
4495 * The cgroup hierarchy has been reworked in many ways. All
4496 objects any of the components systemd creates in the cgroup
4497 tree are now suffixed. More specifically, user sessions are
4498 now placed in cgroups suffixed with ".session", users in
4499 cgroups suffixed with ".user", and nspawn containers in
4500 cgroups suffixed with ".nspawn". Furthermore, all cgroup
4501 names are now escaped in a simple scheme to avoid collision
4502 of userspace object names with kernel filenames. This work
4503 is preparation for making these objects relocatable in the
4504 cgroup tree, in order to allow easy resource partitioning of
4505 these objects without causing naming conflicts.
4507 * systemctl list-dependencies gained the new switches
4508 --plain, --reverse, --after and --before.
4510 * systemd-inhibit now shows the process name of processes that
4511 have taken an inhibitor lock.
4513 * nss-myhostname will now also resolve "localhost"
4514 implicitly. This makes /etc/hosts an optional file and
4515 nicely handles that on IPv6 ::1 maps to both "localhost" and
4518 * libsystemd-logind.so gained a new call
4519 sd_get_machine_names() to enumerate running containers and
4520 VMs (currently only supported by very new libvirt and
4521 nspawn). sd_login_monitor can now be used to watch
4522 VMs/containers coming and going.
4524 * .include is not allowed recursively anymore, and only in
4525 unit files. Usually it is better to use drop-in snippets in
4526 .d/*.conf anyway, as introduced with systemd 198.
4528 * systemd-analyze gained a new "critical-chain" command that
4529 determines the slowest chain of units run during system
4530 boot-up. It is very useful for tracking down where
4531 optimizing boot time is the most beneficial.
4533 * systemd will no longer allow manipulating service paths in
4534 the name=systemd:/system cgroup tree using ControlGroup= in
4535 units. (But is still fine with it in all other dirs.)
4537 * There's a new systemd-nspawn@.service service file that may
4538 be used to easily run nspawn containers as system
4539 services. With the container's root directory in
4540 /var/lib/container/foobar it is now sufficient to run
4541 "systemctl start systemd-nspawn@foobar.service" to boot it.
4543 * systemd-cgls gained a new parameter "--machine" to list only
4544 the processes within a certain container.
4546 * ConditionSecurity= now can check for "apparmor". We still
4547 are lacking checks for SMACK and IMA for this condition
4548 check though. Patches welcome!
4550 * A new configuration file /etc/systemd/sleep.conf has been
4551 added that may be used to configure which kernel operation
4552 systemd is supposed to execute when "suspend", "hibernate"
4553 or "hybrid-sleep" is requested. This makes the new kernel
4554 "freeze" state accessible to the user.
4556 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
4557 the passed argument if applicable.
4559 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
4560 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
4561 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
4562 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
4563 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
4564 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
4565 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
4570 * The output of 'systemctl list-jobs' got some polishing. The
4571 '--type=' argument may now be passed more than once. A new
4572 command 'systemctl list-sockets' has been added which shows
4573 a list of kernel sockets systemd is listening on with the
4574 socket units they belong to, plus the units these socket
4577 * The experimental libsystemd-bus library got substantial
4578 updates to work in conjunction with the (also experimental)
4579 kdbus kernel project. It works well enough to exchange
4580 messages with some sophistication. Note that kdbus is not
4581 ready yet, and the library is mostly an elaborate test case
4582 for now, and not installable.
4584 * systemd gained a new unit 'systemd-static-nodes.service'
4585 that generates static device nodes earlier during boot, and
4586 can run in conjunction with udev.
4588 * libsystemd-login gained a new call sd_pid_get_user_unit()
4589 to retrieve the user systemd unit a process is running
4590 in. This is useful for systems where systemd is used as
4593 * systemd-nspawn now places all containers in the new /machine
4594 top-level cgroup directory in the name=systemd
4595 hierarchy. libvirt will soon do the same, so that we get a
4596 uniform separation of /system, /user and /machine for system
4597 services, user processes and containers/virtual
4598 machines. This new cgroup hierarchy is also useful to stick
4599 stable names to specific container instances, which can be
4600 recognized later this way (this name may be controlled
4601 via systemd-nspawn's new -M switch). libsystemd-login also
4602 gained a new call sd_pid_get_machine_name() to retrieve the
4603 name of the container/VM a specific process belongs to.
4605 * bootchart can now store its data in the journal.
4607 * libsystemd-journal gained a new call
4608 sd_journal_add_conjunction() for AND expressions to the
4609 matching logic. This can be used to express more complex
4610 logical expressions.
4612 * journactl can now take multiple --unit= and --user-unit=
4615 * The cryptsetup logic now understands the "luks.key=" kernel
4616 command line switch for specifying a file to read the
4617 decryption key from. Also, if a configured key file is not
4618 found the tool will now automatically fall back to prompting
4621 * Python systemd.journal module was updated to wrap recently
4622 added functions from libsystemd-journal. The interface was
4623 changed to bring the low level interface in s.j._Reader
4624 closer to the C API, and the high level interface in
4625 s.j.Reader was updated to wrap and convert all data about
4628 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
4629 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
4630 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
4631 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
4632 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
4633 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
4637 * journalctl --update-catalog now understands a new --root=
4638 option to operate on catalogs found in a different root
4641 * During shutdown after systemd has terminated all running
4642 services a final killing loop kills all remaining left-over
4643 processes. We will now print the name of these processes
4644 when we send SIGKILL to them, since this usually indicates a
4647 * If /etc/crypttab refers to password files stored on
4648 configured mount points automatic dependencies will now be
4649 generated to ensure the specific mount is established first
4650 before the key file is attempted to be read.
4652 * 'systemctl status' will now show information about the
4653 network sockets a socket unit is listening on.
4655 * 'systemctl status' will also shown information about any
4656 drop-in configuration file for units. (Drop-In configuration
4657 files in this context are files such as
4658 /etc/systemd/systemd/foobar.service.d/*.conf)
4660 * systemd-cgtop now optionally shows summed up CPU times of
4661 cgroups. Press '%' while running cgtop to switch between
4662 percentage and absolute mode. This is useful to determine
4663 which cgroups use up the most CPU time over the entire
4664 runtime of the system. systemd-cgtop has also been updated
4665 to be 'pipeable' for processing with further shell tools.
4667 * 'hostnamectl set-hostname' will now allow setting of FQDN
4670 * The formatting and parsing of time span values has been
4671 changed. The parser now understands fractional expressions
4672 such as "5.5h". The formatter will now output fractional
4673 expressions for all time spans under 1min, i.e. "5.123456s"
4674 rather than "5s 123ms 456us". For time spans under 1s
4675 millisecond values are shown, for those under 1ms
4676 microsecond values are shown. This should greatly improve
4677 all time-related output of systemd.
4679 * libsystemd-login and libsystemd-journal gained new
4680 functions for querying the poll() events mask and poll()
4681 timeout value for integration into arbitrary event
4684 * localectl gained the ability to list available X11 keymaps
4685 (models, layouts, variants, options).
4687 * 'systemd-analyze dot' gained the ability to filter for
4688 specific units via shell-style globs, to create smaller,
4689 more useful graphs. I.e. it is now possible to create simple
4690 graphs of all the dependencies between only target units, or
4691 of all units that Avahi has dependencies with.
4693 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
4694 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
4695 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
4696 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
4697 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
4698 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
4699 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
4703 * The boot-time readahead implementation for rotating media
4704 will now read the read-ahead data in multiple passes which
4705 consist of all read requests made in equidistant time
4706 intervals. This means instead of strictly reading read-ahead
4707 data in its physical order on disk we now try to find a
4708 middle ground between physical and access time order.
4710 * /etc/os-release files gained a new BUILD_ID= field for usage
4711 on operating systems that provide continuous builds of OS
4714 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
4715 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
4716 William Douglas, Zbigniew Jędrzejewski-Szmek
4720 * systemd-python gained an API exposing libsystemd-daemon.
4722 * The SMACK setup logic gained support for uploading CIPSO
4725 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
4726 ReadOnlyDirectories= and InaccessibleDirectories= has
4727 changed. The private /tmp and /var/tmp directories are now
4728 shared by all processes of a service (which means
4729 ExecStartPre= may now leave data in /tmp that ExecStart= of
4730 the same service can still access). When a service is
4731 stopped its temporary directories are immediately deleted
4732 (normal clean-up with tmpfiles is still done in addition to
4735 * By default, systemd will now set a couple of sysctl
4736 variables in the kernel: the safe sysrq options are turned
4737 on, IP route verification is turned on, and source routing
4738 disabled. The recently added hardlink and softlink
4739 protection of the kernel is turned on. These settings should
4740 be reasonably safe, and good defaults for all new systems.
4742 * The predictable network naming logic may now be turned off
4743 with a new kernel command line switch: net.ifnames=0.
4745 * A new libsystemd-bus module has been added that implements a
4746 pretty complete D-Bus client library. For details see:
4748 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
4750 * journald will now explicitly flush the journal files to disk
4751 at the latest 5min after each write. The file will then also
4752 be marked offline until the next write. This should increase
4753 reliability in case of a crash. The synchronization delay
4754 can be configured via SyncIntervalSec= in journald.conf.
4756 * There's a new remote-fs-setup.target unit that can be used
4757 to pull in specific services when at least one remote file
4758 system is to be mounted.
4760 * There are new targets timers.target and paths.target as
4761 canonical targets to pull user timer and path units in
4762 from. This complements sockets.target with a similar
4763 purpose for socket units.
4765 * libudev gained a new call udev_device_set_attribute_value()
4766 to set sysfs attributes of a device.
4768 * The udev daemon now sets the default number of worker
4769 processes executed in parallel based on the number of available
4770 CPUs instead of the amount of available RAM. This is supposed
4771 to provide a more reliable default and limit a too aggressive
4772 paralellism for setups with 1000s of devices connected.
4774 Contributions from: Auke Kok, Colin Walters, Cristian
4775 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
4776 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
4777 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
4778 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
4779 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
4780 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
4781 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
4782 Zbigniew Jędrzejewski-Szmek
4786 * Configuration of unit files may now be extended via drop-in
4787 files without having to edit/override the unit files
4788 themselves. More specifically, if the administrator wants to
4789 change one value for a service file foobar.service he can
4790 now do so by dropping in a configuration snippet into
4791 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
4792 will load all these snippets and apply them on top of the
4793 main unit configuration file, possibly extending or
4794 overriding its settings. Using these drop-in snippets is
4795 generally nicer than the two earlier options for changing
4796 unit files locally: copying the files from
4797 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
4798 them there; or creating a new file in /etc/systemd/system/
4799 that incorporates the original one via ".include". Drop-in
4800 snippets into these .d/ directories can be placed in any
4801 directory systemd looks for units in, and the usual
4802 overriding semantics between /usr/lib, /etc and /run apply
4805 * Most unit file settings which take lists of items can now be
4806 reset by assigning the empty string to them. For example,
4807 normally, settings such as Environment=FOO=BAR append a new
4808 environment variable assignment to the environment block,
4809 each time they are used. By assigning Environment= the empty
4810 string the environment block can be reset to empty. This is
4811 particularly useful with the .d/*.conf drop-in snippets
4812 mentioned above, since this adds the ability to reset list
4813 settings from vendor unit files via these drop-ins.
4815 * systemctl gained a new "list-dependencies" command for
4816 listing the dependencies of a unit recursively.
4818 * Inhibitors are now honored and listed by "systemctl
4819 suspend", "systemctl poweroff" (and similar) too, not only
4820 GNOME. These commands will also list active sessions by
4823 * Resource limits (as exposed by the various control group
4824 controllers) can now be controlled dynamically at runtime
4825 for all units. More specifically, you can now use a command
4826 like "systemctl set-cgroup-attr foobar.service cpu.shares
4827 2000" to alter the CPU shares a specific service gets. These
4828 settings are stored persistently on disk, and thus allow the
4829 administrator to easily adjust the resource usage of
4830 services with a few simple commands. This dynamic resource
4831 management logic is also available to other programs via the
4832 bus. Almost any kernel cgroup attribute and controller is
4835 * systemd-vconsole-setup will now copy all font settings to
4836 all allocated VTs, where it previously applied them only to
4839 * libsystemd-login gained the new sd_session_get_tty() API
4842 * This release drops support for a few legacy or
4843 distribution-specific LSB facility names when parsing init
4844 scripts: $x-display-manager, $mail-transfer-agent,
4845 $mail-transport-agent, $mail-transfer-agent, $smtp,
4846 $null. Also, the mail-transfer-agent.target unit backing
4847 this has been removed. Distributions which want to retain
4848 compatibility with this should carry the burden for
4849 supporting this themselves and patch support for these back
4850 in, if they really need to. Also, the facilities $syslog and
4851 $local_fs are now ignored, since systemd does not support
4852 early-boot LSB init scripts anymore, and these facilities
4853 are implied anyway for normal services. syslog.target has
4856 * There are new bus calls on PID1's Manager object for
4857 cancelling jobs, and removing snapshot units. Previously,
4858 both calls were only available on the Job and Snapshot
4861 * systemd-journal-gatewayd gained SSL support.
4863 * The various "environment" files, such as /etc/locale.conf
4864 now support continuation lines with a backslash ("\") as
4865 last character in the line, similarly in style (but different)
4866 to how this is supported in shells.
4868 * For normal user processes the _SYSTEMD_USER_UNIT= field is
4869 now implicitly appended to every log entry logged. systemctl
4870 has been updated to filter by this field when operating on a
4871 user systemd instance.
4873 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
4874 CAP_AUDIT_CONTROL capabilities to the capabilities set for
4875 the container. This makes it easier to boot unmodified
4876 Fedora systems in a container, which however still requires
4877 audit=0 to be passed on the kernel command line. Auditing in
4878 kernel and userspace is unfortunately still too broken in
4879 context of containers, hence we recommend compiling it out
4880 of the kernel or using audit=0. Hopefully this will be fixed
4881 one day for good in the kernel.
4883 * nspawn gained the new --bind= and --bind-ro= parameters to
4884 bind mount specific directories from the host into the
4887 * nspawn will now mount its own devpts file system instance
4888 into the container, in order not to leak pty devices from
4889 the host into the container.
4891 * systemd will now read the firmware boot time performance
4892 information from the EFI variables, if the used boot loader
4893 supports this, and takes it into account for boot performance
4894 analysis via "systemd-analyze". This is currently supported
4895 only in conjunction with Gummiboot, but could be supported
4896 by other boot loaders too. For details see:
4898 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
4900 * A new generator has been added that automatically mounts the
4901 EFI System Partition (ESP) to /boot, if that directory
4902 exists, is empty, and no other file system has been
4903 configured to be mounted there.
4905 * logind will now send out PrepareForSleep(false) out
4906 unconditionally, after coming back from suspend. This may be
4907 used by applications as asynchronous notification for
4908 system resume events.
4910 * "systemctl unlock-sessions" has been added, that allows
4911 unlocking the screens of all user sessions at once, similar
4912 to how "systemctl lock-sessions" already locked all users
4913 sessions. This is backed by a new D-Bus call UnlockSessions().
4915 * "loginctl seat-status" will now show the master device of a
4916 seat. (i.e. the device of a seat that needs to be around for
4917 the seat to be considered available, usually the graphics
4920 * tmpfiles gained a new "X" line type, that allows
4921 configuration of files and directories (with wildcards) that
4922 shall be excluded from automatic cleanup ("aging").
4924 * udev default rules set the device node permissions now only
4925 at "add" events, and do not change them any longer with a
4926 later "change" event.
4928 * The log messages for lid events and power/sleep keypresses
4929 now carry a message ID.
4931 * We now have a substantially larger unit test suite, but this
4932 continues to be work in progress.
4934 * udevadm hwdb gained a new --root= parameter to change the
4935 root directory to operate relative to.
4937 * logind will now issue a background sync() request to the kernel
4938 early at shutdown, so that dirty buffers are flushed to disk early
4939 instead of at the last moment, in order to optimize shutdown
4942 * A new bootctl tool has been added that is an interface for
4943 certain boot loader operations. This is currently a preview
4944 and is likely to be extended into a small mechanism daemon
4945 like timedated, localed, hostnamed, and can be used by
4946 graphical UIs to enumerate available boot options, and
4947 request boot into firmware operations.
4949 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
4950 the rest of the package. It also has been updated to work
4951 correctly in initrds.
4953 * Policykit previously has been runtime optional, and is now
4954 also compile time optional via a configure switch.
4956 * systemd-analyze has been reimplemented in C. Also "systemctl
4957 dot" has moved into systemd-analyze.
4959 * "systemctl status" with no further parameters will now print
4960 the status of all active or failed units.
4962 * Operations such as "systemctl start" can now be executed
4963 with a new mode "--irreversible" which may be used to queue
4964 operations that cannot accidentally be reversed by a later
4965 job queuing. This is by default used to make shutdown
4966 requests more robust.
4968 * The Python API of systemd now gained a new module for
4969 reading journal files.
4971 * A new tool kernel-install has been added that can install
4972 kernel images according to the Boot Loader Specification:
4974 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
4976 * Boot time console output has been improved to provide
4977 animated boot time output for hanging jobs.
4979 * A new tool systemd-activate has been added which can be used
4980 to test socket activation with, directly from the command
4981 line. This should make it much easier to test and debug
4982 socket activation in daemons.
4984 * journalctl gained a new "--reverse" (or -r) option to show
4985 journal output in reverse order (i.e. newest line first).
4987 * journalctl gained a new "--pager-end" (or -e) option to jump
4988 to immediately jump to the end of the journal in the
4989 pager. This is only supported in conjunction with "less".
4991 * journalctl gained a new "--user-unit=" option, that works
4992 similarly to "--unit=" but filters for user units rather than
4995 * A number of unit files to ease adoption of systemd in
4996 initrds has been added. This moves some minimal logic from
4997 the various initrd implementations into systemd proper.
4999 * The journal files are now owned by a new group
5000 "systemd-journal", which exists specifically to allow access
5001 to the journal, and nothing else. Previously, we used the
5002 "adm" group for that, which however possibly covers more
5003 than just journal/log file access. This new group is now
5004 already used by systemd-journal-gatewayd to ensure this
5005 daemon gets access to the journal files and as little else
5006 as possible. Note that "make install" will also set FS ACLs
5007 up for /var/log/journal to give "adm" and "wheel" read
5008 access to it, in addition to "systemd-journal" which owns
5009 the journal files. We recommend that packaging scripts also
5010 add read access to "adm" + "wheel" to /var/log/journal, and
5011 all existing/future journal files. To normal users and
5012 administrators little changes, however packagers need to
5013 ensure to create the "systemd-journal" system group at
5014 package installation time.
5016 * The systemd-journal-gatewayd now runs as unprivileged user
5017 systemd-journal-gateway:systemd-journal-gateway. Packaging
5018 scripts need to create these system user/group at
5021 * timedated now exposes a new boolean property CanNTP that
5022 indicates whether a local NTP service is available or not.
5024 * systemd-detect-virt will now also detect xen PVs
5026 * The pstore file system is now mounted by default, if it is
5029 * In addition to the SELinux and IMA policies we will now also
5030 load SMACK policies at early boot.
5032 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
5033 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
5034 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
5035 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
5036 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
5037 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
5038 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
5039 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
5040 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
5041 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
5042 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
5043 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
5044 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
5045 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
5049 * Timer units now support calendar time events in addition to
5050 monotonic time events. That means you can now trigger a unit
5051 based on a calendar time specification such as "Thu,Fri
5052 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
5053 or fifth day of any month of the year 2013, given that it is
5054 a thursday or friday. This brings timer event support
5055 considerably closer to cron's capabilities. For details on
5056 the supported calendar time specification language see
5059 * udev now supports a number of different naming policies for
5060 network interfaces for predictable names, and a combination
5061 of these policies is now the default. Please see this wiki
5062 document for details:
5064 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
5066 * Auke Kok's bootchart implementation has been added to the
5067 systemd tree. It is an optional component that can graph the
5068 boot in quite some detail. It is one of the best bootchart
5069 implementations around and minimal in its code and
5072 * nss-myhostname has been integrated into the systemd source
5073 tree. nss-myhostname guarantees that the local hostname
5074 always stays resolvable via NSS. It has been a weak
5075 requirement of systemd-hostnamed since a long time, and
5076 since its code is actually trivial we decided to just
5077 include it in systemd's source tree. It can be turned off
5078 with a configure switch.
5080 * The read-ahead logic is now capable of properly detecting
5081 whether a btrfs file system is on SSD or rotating media, in
5082 order to optimize the read-ahead scheme. Previously, it was
5083 only capable of detecting this on traditional file systems
5086 * In udev, additional device properties are now read from the
5087 IAB in addition to the OUI database. Also, Bluetooth company
5088 identities are attached to the devices as well.
5090 * In service files %U may be used as specifier that is
5091 replaced by the configured user name of the service.
5093 * nspawn may now be invoked without a controlling TTY. This
5094 makes it suitable for invocation as its own service. This
5095 may be used to set up a simple containerized server system
5096 using only core OS tools.
5098 * systemd and nspawn can now accept socket file descriptors
5099 when they are started for socket activation. This enables
5100 implementation of socket activated nspawn
5101 containers. i.e. think about autospawning an entire OS image
5102 when the first SSH or HTTP connection is received. We expect
5103 that similar functionality will also be added to libvirt-lxc
5106 * journalctl will now suppress ANSI color codes when
5107 presenting log data.
5109 * systemctl will no longer show control group information for
5110 a unit if a the control group is empty anyway.
5112 * logind can now automatically suspend/hibernate/shutdown the
5115 * /etc/machine-info and hostnamed now also expose the chassis
5116 type of the system. This can be used to determine whether
5117 the local system is a laptop, desktop, handset or
5118 tablet. This information may either be configured by the
5119 user/vendor or is automatically determined from ACPI and DMI
5120 information if possible.
5122 * A number of PolicyKit actions are now bound together with
5123 "imply" rules. This should simplify creating UIs because
5124 many actions will now authenticate similar ones as well.
5126 * Unit files learnt a new condition ConditionACPower= which
5127 may be used to conditionalize a unit depending on whether an
5128 AC power source is connected or not, of whether the system
5129 is running on battery power.
5131 * systemctl gained a new "is-failed" verb that may be used in
5132 shell scripts and suchlike to check whether a specific unit
5133 is in the "failed" state.
5135 * The EnvironmentFile= setting in unit files now supports file
5136 globbing, and can hence be used to easily read a number of
5137 environment files at once.
5139 * systemd will no longer detect and recognize specific
5140 distributions. All distribution-specific #ifdeffery has been
5141 removed, systemd is now fully generic and
5142 distribution-agnostic. Effectively, not too much is lost as
5143 a lot of the code is still accessible via explicit configure
5144 switches. However, support for some distribution specific
5145 legacy configuration file formats has been dropped. We
5146 recommend distributions to simply adopt the configuration
5147 files everybody else uses now and convert the old
5148 configuration from packaging scripts. Most distributions
5149 already did that. If that's not possible or desirable,
5150 distributions are welcome to forward port the specific
5151 pieces of code locally from the git history.
5153 * When logging a message about a unit systemd will now always
5154 log the unit name in the message meta data.
5156 * localectl will now also discover system locale data that is
5157 not stored in locale archives, but directly unpacked.
5159 * logind will no longer unconditionally use framebuffer
5160 devices as seat masters, i.e. as devices that are required
5161 to be existing before a seat is considered preset. Instead,
5162 it will now look for all devices that are tagged as
5163 "seat-master" in udev. By default, framebuffer devices will
5164 be marked as such, but depending on local systems, other
5165 devices might be marked as well. This may be used to
5166 integrate graphics cards using closed source drivers (such
5167 as NVidia ones) more nicely into logind. Note however, that
5168 we recommend using the open source NVidia drivers instead,
5169 and no udev rules for the closed-source drivers will be
5170 shipped from us upstream.
5172 Contributions from: Adam Williamson, Alessandro Crismani, Auke
5173 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
5174 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
5175 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
5176 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
5177 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
5178 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
5179 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
5180 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
5181 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
5182 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
5183 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
5188 * udev gained support for loading additional device properties
5189 from an indexed database that is keyed by vendor/product IDs
5190 and similar device identifiers. For the beginning this
5191 "hwdb" is populated with data from the well-known PCI and
5192 USB database, but also includes PNP, ACPI and OID data. In
5193 the longer run this indexed database shall grow into
5194 becoming the one central database for non-essential
5195 userspace device metadata. Previously, data from the PCI/USB
5196 database was only attached to select devices, since the
5197 lookup was a relatively expensive operation due to O(n) time
5198 complexity (with n being the number of entries in the
5199 database). Since this is now O(1), we decided to add in this
5200 data for all devices where this is available, by
5201 default. Note that the indexed database needs to be rebuilt
5202 when new data files are installed. To achieve this you need
5203 to update your packaging scripts to invoke "udevadm hwdb
5204 --update" after installation of hwdb data files. For
5205 RPM-based distributions we introduced the new
5206 %udev_hwdb_update macro for this purpose.
5208 * The Journal gained support for the "Message Catalog", an
5209 indexed database to link up additional information with
5210 journal entries. For further details please check:
5212 http://www.freedesktop.org/wiki/Software/systemd/catalog
5214 The indexed message catalog database also needs to be
5215 rebuilt after installation of message catalog files. Use
5216 "journalctl --update-catalog" for this. For RPM-based
5217 distributions we introduced the %journal_catalog_update
5218 macro for this purpose.
5220 * The Python Journal bindings gained support for the standard
5221 Python logging framework.
5223 * The Journal API gained new functions for checking whether
5224 the underlying file system of a journal file is capable of
5225 properly reporting file change notifications, or whether
5226 applications that want to reflect journal changes "live"
5227 need to recheck journal files continuously in appropriate
5230 * It is now possible to set the "age" field for tmpfiles
5231 entries to 0, indicating that files matching this entry
5232 shall always be removed when the directories are cleaned up.
5234 * coredumpctl gained a new "gdb" verb which invokes gdb
5235 right-away on the selected coredump.
5237 * There's now support for "hybrid sleep" on kernels that
5238 support this, in addition to "suspend" and "hibernate". Use
5239 "systemctl hybrid-sleep" to make use of this.
5241 * logind's HandleSuspendKey= setting (and related settings)
5242 now gained support for a new "lock" setting to simply
5243 request the screen lock on all local sessions, instead of
5244 actually executing a suspend or hibernation.
5246 * systemd will now mount the EFI variables file system by
5249 * Socket units now gained support for configuration of the
5250 SMACK security label.
5252 * timedatectl will now output the time of the last and next
5253 daylight saving change.
5255 * We dropped support for various legacy and distro-specific
5256 concepts, such as insserv, early-boot SysV services
5257 (i.e. those for non-standard runlevels such as 'b' or 'S')
5258 or ArchLinux /etc/rc.conf support. We recommend the
5259 distributions who still need support this to either continue
5260 to maintain the necessary patches downstream, or find a
5261 different solution. (Talk to us if you have questions!)
5263 * Various systemd components will now bypass PolicyKit checks
5264 for root and otherwise handle properly if PolicyKit is not
5265 found to be around. This should fix most issues for
5266 PolicyKit-less systems. Quite frankly this should have been
5267 this way since day one. It is absolutely our intention to
5268 make systemd work fine on PolicyKit-less systems, and we
5269 consider it a bug if something does not work as it should if
5270 PolicyKit is not around.
5272 * For embedded systems it is now possible to build udev and
5273 systemd without blkid and/or kmod support.
5275 * "systemctl switch-root" is now capable of switching root
5276 more than once. I.e. in addition to transitions from the
5277 initrd to the host OS it is now possible to transition to
5278 further OS images from the host. This is useful to implement
5279 offline updating tools.
5281 * Various other additions have been made to the RPM macros
5282 shipped with systemd. Use %udev_rules_update() after
5283 installing new udev rules files. %_udevhwdbdir,
5284 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
5285 %_sysctldir are now available which resolve to the right
5286 directories for packages to place various data files in.
5288 * journalctl gained the new --full switch (in addition to
5289 --all, to disable ellipsation for long messages.
5291 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
5292 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
5293 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
5294 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
5295 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
5296 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
5297 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
5298 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
5299 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
5303 * journalctl gained new --since= and --until= switches to
5304 filter by time. It also now supports nice filtering for
5305 units via --unit=/-u.
5307 * Type=oneshot services may use ExecReload= and do the
5310 * The journal daemon now supports time-based rotation and
5311 vacuuming, in addition to the usual disk-space based
5314 * The journal will now index the available field values for
5315 each field name. This enables clients to show pretty drop
5316 downs of available match values when filtering. The bash
5317 completion of journalctl has been updated
5318 accordingly. journalctl gained a new switch -F to list all
5319 values a certain field takes in the journal database.
5321 * More service events are now written as structured messages
5322 to the journal, and made recognizable via message IDs.
5324 * The timedated, localed and hostnamed mini-services which
5325 previously only provided support for changing time, locale
5326 and hostname settings from graphical DEs such as GNOME now
5327 also have a minimal (but very useful) text-based client
5328 utility each. This is probably the nicest way to changing
5329 these settings from the command line now, especially since
5330 it lists available options and is fully integrated with bash
5333 * There's now a new tool "systemd-coredumpctl" to list and
5334 extract coredumps from the journal.
5336 * We now install a README each in /var/log/ and
5337 /etc/rc.d/init.d explaining where the system logs and init
5338 scripts went. This hopefully should help folks who go to
5339 that dirs and look into the otherwise now empty void and
5340 scratch their heads.
5342 * When user-services are invoked (by systemd --user) the
5343 $MANAGERPID env var is set to the PID of systemd.
5345 * SIGRTMIN+24 when sent to a --user instance will now result
5346 in immediate termination of systemd.
5348 * gatewayd received numerous feature additions such as a
5349 "follow" mode, for live syncing and filtering.
5351 * browse.html now allows filtering and showing detailed
5352 information on specific entries. Keyboard navigation and
5353 mouse screen support has been added.
5355 * gatewayd/journalctl now supports HTML5/JSON
5356 Server-Sent-Events as output.
5358 * The SysV init script compatibility logic will now
5359 heuristically determine whether a script supports the
5360 "reload" verb, and only then make this available as
5363 * "systemctl status --follow" has been removed, use "journalctl
5366 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
5367 have been removed since they are hardly useful to be
5370 * And I'd like to take the opportunity to specifically mention
5371 Zbigniew for his great contributions. Zbigniew, you rock!
5373 Contributions from: Andrew Eikum, Christian Hesse, Colin
5374 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
5375 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
5376 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
5377 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
5378 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
5379 Jędrzejewski-Szmek, Сковорода Никита Андреевич
5383 * If /etc/vconsole.conf is non-existent or empty we will no
5384 longer load any console font or key map at boot by
5385 default. Instead the kernel defaults will be left
5386 intact. This is definitely the right thing to do, as no
5387 configuration should mean no configuration, and hard-coding
5388 font names that are different on all archs is probably a bad
5389 idea. Also, the kernel default key map and font should be
5390 good enough for most cases anyway, and mostly identical to
5391 the userspace fonts/key maps we previously overloaded them
5392 with. If distributions want to continue to default to a
5393 non-kernel font or key map they should ship a default
5394 /etc/vconsole.conf with the appropriate contents.
5396 Contributions from: Colin Walters, Daniel J Walsh, Dave
5397 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
5398 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5402 * journalctl gained a new --cursor= switch to show entries
5403 starting from the specified location in the journal.
5405 * We now enforce a size limit on journal entry fields exported
5406 with "-o json" in journalctl. Fields larger than 4K will be
5407 assigned null. This can be turned off with --all.
5409 * An (optional) journal gateway daemon is now available as
5410 "systemd-journal-gatewayd.service". This service provides
5411 access to the journal via HTTP and JSON. This functionality
5412 will be used to implement live log synchronization in both
5413 pull and push modes, but has various other users too, such
5414 as easy log access for debugging of embedded devices. Right
5415 now it is already useful to retrieve the journal via HTTP:
5417 # systemctl start systemd-journal-gatewayd.service
5418 # wget http://localhost:19531/entries
5420 This will download the journal contents in a
5421 /var/log/messages compatible format. The same as JSON:
5423 # curl -H"Accept: application/json" http://localhost:19531/entries
5425 This service is also accessible via a web browser where a
5426 single static HTML5 app is served that uses the JSON logic
5427 to enable the user to do some basic browsing of the
5428 journal. This will be extended later on. Here's an example
5429 screenshot of this app in its current state:
5431 http://0pointer.de/public/journal-gatewayd
5433 Contributions from: Kay Sievers, Lennart Poettering, Robert
5434 Milasan, Tom Gundersen
5438 * The bash completion logic is now available for journalctl
5441 * We do not mount the "cpuset" controller anymore together with
5442 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
5443 started if no parameters are assigned to it. "cpuset" hence
5444 broke code that assumed it could create "cpu" groups and
5447 * journalctl -f will now subscribe to terminal size changes,
5448 and line break accordingly.
5450 Contributions from: Dave Reisner, Kay Sievers, Lennart
5451 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
5455 * nspawn will now create a symlink /etc/localtime in the
5456 container environment, copying the host's timezone
5457 setting. Previously this has been done via a bind mount, but
5458 since symlinks cannot be bind mounted this has now been
5459 changed to create/update the appropriate symlink.
5461 * journalctl -n's line number argument is now optional, and
5462 will default to 10 if omitted.
5464 * journald will now log the maximum size the journal files may
5465 take up on disk. This is particularly useful if the default
5466 built-in logic of determining this parameter from the file
5467 system size is used. Use "systemctl status
5468 systemd-journald.service" to see this information.
5470 * The multi-seat X wrapper tool has been stripped down. As X
5471 is now capable of enumerating graphics devices via udev in a
5472 seat-aware way the wrapper is not strictly necessary
5473 anymore. A stripped down temporary stop-gap is still shipped
5474 until the upstream display managers have been updated to
5475 fully support the new X logic. Expect this wrapper to be
5476 removed entirely in one of the next releases.
5478 * HandleSleepKey= in logind.conf has been split up into
5479 HandleSuspendKey= and HandleHibernateKey=. The old setting
5480 is not available anymore. X11 and the kernel are
5481 distinguishing between these keys and we should too. This
5482 also means the inhibition lock for these keys has been split
5485 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
5486 Poettering, Lukas Nykryn, Václav Pavlín
5490 * Whenever a unit changes state we will now log this to the
5491 journal and show along the unit's own log output in
5494 * ConditionPathIsMountPoint= can now properly detect bind
5495 mount points too. (Previously, a bind mount of one file
5496 system to another place in the same file system could not be
5497 detected as mount, since they shared struct stat's st_dev
5500 * We will now mount the cgroup controllers cpu, cpuacct,
5501 cpuset and the controllers net_cls, net_prio together by
5504 * nspawn containers will now have a virtualized boot
5505 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
5506 over with a randomized ID at container initialization). This
5507 has the effect of making "journalctl -b" do the right thing
5510 * The JSON output journal serialization has been updated not
5511 to generate "endless" list objects anymore, but rather one
5512 JSON object per line. This is more in line how most JSON
5513 parsers expect JSON objects. The new output mode
5514 "json-pretty" has been added to provide similar output, but
5515 neatly aligned for readability by humans.
5517 * We dropped all explicit sync() invocations in the shutdown
5518 code. The kernel does this implicitly anyway in the kernel
5519 reboot() syscall. halt(8)'s -n option is now a compatibility
5522 * We now support virtualized reboot() in containers, as
5523 supported by newer kernels. We will fall back to exit() if
5524 CAP_SYS_REBOOT is not available to the container. Also,
5525 nspawn makes use of this now and will actually reboot the
5526 container if the containerized OS asks for that.
5528 * journalctl will only show local log output by default
5529 now. Use --merge (-m) to show remote log output, too.
5531 * libsystemd-journal gained the new sd_journal_get_usage()
5532 call to determine the current disk usage of all journal
5533 files. This is exposed in the new "journalctl --disk-usage"
5536 * journald gained a new configuration setting SplitMode= in
5537 journald.conf which may be used to control how user journals
5538 are split off. See journald.conf(5) for details.
5540 * A new condition type ConditionFileNotEmpty= has been added.
5542 * tmpfiles' "w" lines now support file globbing, to write
5543 multiple files at once.
5545 * We added Python bindings for the journal submission
5546 APIs. More Python APIs for a number of selected APIs will
5547 likely follow. Note that we intend to add native bindings
5548 only for the Python language, as we consider it common
5549 enough to deserve bindings shipped within systemd. There are
5550 various projects outside of systemd that provide bindings
5551 for languages such as PHP or Lua.
5553 * Many conditions will now resolve specifiers such as %i. In
5554 addition, PathChanged= and related directives of .path units
5555 now support specifiers as well.
5557 * There's now a new RPM macro definition for the system preset
5560 * journald will now warn if it ca not forward a message to the
5561 syslog daemon because its socket is full.
5563 * timedated will no longer write or process /etc/timezone,
5564 except on Debian. As we do not support late mounted /usr
5565 anymore /etc/localtime always being a symlink is now safe,
5566 and hence the information in /etc/timezone is not necessary
5569 * logind will now always reserve one VT for a text getty (VT6
5570 by default). Previously if more than 6 X sessions where
5571 started they took up all the VTs with auto-spawned gettys,
5572 so that no text gettys were available anymore.
5574 * udev will now automatically inform the btrfs kernel logic
5575 about btrfs RAID components showing up. This should make
5576 simple hotplug based btrfs RAID assembly work.
5578 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
5579 (but not for its children which will stay at the kernel
5580 default). This should allow setups with a lot more listening
5583 * systemd will now always pass the configured timezone to the
5584 kernel at boot. timedated will do the same when the timezone
5587 * logind's inhibition logic has been updated. By default,
5588 logind will now handle the lid switch, the power and sleep
5589 keys all the time, even in graphical sessions. If DEs want
5590 to handle these events on their own they should take the new
5591 handle-power-key, handle-sleep-key and handle-lid-switch
5592 inhibitors during their runtime. A simple way to achieve
5593 that is to invoke the DE wrapped in an invocation of:
5595 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
5597 * Access to unit operations is now checked via SELinux taking
5598 the unit file label and client process label into account.
5600 * systemd will now notify the administrator in the journal
5601 when he over-mounts a non-empty directory.
5603 * There are new specifiers that are resolved in unit files,
5604 for the host name (%H), the machine ID (%m) and the boot ID
5607 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
5608 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
5609 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
5610 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
5611 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
5612 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
5613 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
5617 * Support for reading structured kernel messages from
5618 /dev/kmsg has now been added and is enabled by default.
5620 * Support for reading kernel messages from /proc/kmsg has now
5621 been removed. If you want kernel messages in the journal
5622 make sure to run a recent kernel (>= 3.5) that supports
5623 reading structured messages from /dev/kmsg (see
5624 above). /proc/kmsg is now exclusive property of classic
5625 syslog daemons again.
5627 * The libudev API gained the new
5628 udev_device_new_from_device_id() call.
5630 * The logic for file system namespace (ReadOnlyDirectory=,
5631 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
5632 require pivot_root() anymore. This means fewer temporary
5633 directories are created below /tmp for this feature.
5635 * nspawn containers will now see and receive all submounts
5636 made on the host OS below the root file system of the
5639 * Forward Secure Sealing is now supported for Journal files,
5640 which provide cryptographical sealing of journal files so
5641 that attackers cannot alter log history anymore without this
5642 being detectable. Lennart will soon post a blog story about
5643 this explaining it in more detail.
5645 * There are two new service settings RestartPreventExitStatus=
5646 and SuccessExitStatus= which allow configuration of exit
5647 status (exit code or signal) which will be excepted from the
5648 restart logic, resp. consider successful.
5650 * journalctl gained the new --verify switch that can be used
5651 to check the integrity of the structure of journal files and
5652 (if Forward Secure Sealing is enabled) the contents of
5655 * nspawn containers will now be run with /dev/stdin, /dev/fd/
5656 and similar symlinks pre-created. This makes running shells
5657 as container init process a lot more fun.
5659 * The fstab support can now handle PARTUUID= and PARTLABEL=
5662 * A new ConditionHost= condition has been added to match
5663 against the hostname (with globs) and machine ID. This is
5664 useful for clusters where a single OS image is used to
5665 provision a large number of hosts which shall run slightly
5666 different sets of services.
5668 * Services which hit the restart limit will now be placed in a
5671 Contributions from: Bertram Poettering, Dave Reisner, Huang
5672 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
5673 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
5677 * When running in --user mode systemd will now become a
5678 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
5679 tree a lot more organized.
5681 * A new PartOf= unit dependency type has been introduced that
5682 may be used to group services in a natural way.
5684 * "systemctl enable" may now be used to enable instances of
5687 * journalctl now prints error log levels in red, and
5688 warning/notice log levels in bright white. It also supports
5689 filtering by log level now.
5691 * cgtop gained a new -n switch (similar to top), to configure
5692 the maximum number of iterations to run for. It also gained
5693 -b, to run in batch mode (accepting no input).
5695 * The suffix ".service" may now be omitted on most systemctl
5696 command lines involving service unit names.
5698 * There's a new bus call in logind to lock all sessions, as
5699 well as a loginctl verb for it "lock-sessions".
5701 * libsystemd-logind.so gained a new call sd_journal_perror()
5702 that works similar to libc perror() but logs to the journal
5703 and encodes structured information about the error number.
5705 * /etc/crypttab entries now understand the new keyfile-size=
5708 * shutdown(8) now can send a (configurable) wall message when
5709 a shutdown is cancelled.
5711 * The mount propagation mode for the root file system will now
5712 default to "shared", which is useful to make containers work
5713 nicely out-of-the-box so that they receive new mounts from
5714 the host. This can be undone locally by running "mount
5715 --make-rprivate /" if needed.
5717 * The prefdm.service file has been removed. Distributions
5718 should maintain this unit downstream if they intend to keep
5719 it around. However, we recommend writing normal unit files
5720 for display managers instead.
5722 * Since systemd is a crucial part of the OS we will now
5723 default to a number of compiler switches that improve
5724 security (hardening) such as read-only relocations, stack
5725 protection, and suchlike.
5727 * The TimeoutSec= setting for services is now split into
5728 TimeoutStartSec= and TimeoutStopSec= to allow configuration
5729 of individual time outs for the start and the stop phase of
5732 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
5733 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
5734 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
5735 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
5736 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
5737 Gundersen, Zbigniew Jędrzejewski-Szmek
5741 * The journal and id128 C APIs are now fully documented as man
5744 * Extra safety checks have been added when transitioning from
5745 the initial RAM disk to the main system to avoid accidental
5748 * /etc/crypttab entries now understand the new keyfile-offset=
5751 * systemctl -t can now be used to filter by unit load state.
5753 * The journal C API gained the new sd_journal_wait() call to
5754 make writing synchronous journal clients easier.
5756 * journalctl gained the new -D switch to show journals from a
5759 * journalctl now displays a special marker between log
5760 messages of two different boots.
5762 * The journal is now explicitly flushed to /var via a service
5763 systemd-journal-flush.service, rather than implicitly simply
5764 by seeing /var/log/journal to be writable.
5766 * journalctl (and the journal C APIs) can now match for much
5767 more complex expressions, with alternatives and
5770 * When transitioning from the initial RAM disk to the main
5771 system we will now kill all processes in a killing spree to
5772 ensure no processes stay around by accident.
5774 * Three new specifiers may be used in unit files: %u, %h, %s
5775 resolve to the user name, user home directory resp. user
5776 shell. This is useful for running systemd user instances.
5778 * We now automatically rotate journal files if their data
5779 object hash table gets a fill level > 75%. We also size the
5780 hash table based on the configured maximum file size. This
5781 together should lower hash collisions drastically and thus
5782 speed things up a bit.
5784 * journalctl gained the new "--header" switch to introspect
5785 header data of journal files.
5787 * A new setting SystemCallFilters= has been added to services
5788 which may be used to apply blacklists or whitelists to
5789 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
5791 * nspawn gained a new --link-journal= switch (and quicker: -j)
5792 to link the container journal with the host. This makes it
5793 very easy to centralize log viewing on the host for all
5794 guests while still keeping the journal files separated.
5796 * Many bugfixes and optimizations
5798 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
5799 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
5800 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
5805 * Several tools now understand kernel command line arguments,
5806 which are only read when run in an initial RAM disk. They
5807 usually follow closely their normal counterparts, but are
5810 * There's a new tool to analyze the readahead files that are
5811 automatically generated at boot. Use:
5813 /usr/lib/systemd/systemd-readahead analyze /.readahead
5815 * We now provide an early debug shell on tty9 if this enabled. Use:
5817 systemctl enable debug-shell.service
5819 * All plymouth related units have been moved into the Plymouth
5820 package. Please make sure to upgrade your Plymouth version
5823 * systemd-tmpfiles now supports getting passed the basename of
5824 a configuration file only, in which case it will look for it
5825 in all appropriate directories automatically.
5827 * udevadm info now takes a /dev or /sys path as argument, and
5828 does the right thing. Example:
5830 udevadm info /dev/sda
5831 udevadm info /sys/class/block/sda
5833 * systemctl now prints a warning if a unit is stopped but a
5834 unit that might trigger it continues to run. Example: a
5835 service is stopped but the socket that activates it is left
5838 * "systemctl status" will now mention if the log output was
5839 shortened due to rotation since a service has been started.
5841 * The journal API now exposes functions to determine the
5842 "cutoff" times due to rotation.
5844 * journald now understands SIGUSR1 and SIGUSR2 for triggering
5845 immediately flushing of runtime logs to /var if possible,
5846 resp. for triggering immediate rotation of the journal
5849 * It is now considered an error if a service is attempted to
5850 be stopped that is not loaded.
5852 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
5854 * systemd-analyze now supports Python 3
5856 * tmpfiles now supports cleaning up directories via aging
5857 where the first level dirs are always kept around but
5858 directories beneath it automatically aged. This is enabled
5859 by prefixing the age field with '~'.
5861 * Seat objects now expose CanGraphical, CanTTY properties
5862 which is required to deal with very fast bootups where the
5863 display manager might be running before the graphics drivers
5864 completed initialization.
5866 * Seat objects now expose a State property.
5868 * We now include RPM macros for service enabling/disabling
5869 based on the preset logic. We recommend RPM based
5870 distributions to make use of these macros if possible. This
5871 makes it simpler to reuse RPM spec files across
5874 * We now make sure that the collected systemd unit name is
5875 always valid when services log to the journal via
5878 * There's a new man page kernel-command-line(7) detailing all
5879 command line options we understand.
5881 * The fstab generator may now be disabled at boot by passing
5882 fstab=0 on the kernel command line.
5884 * A new kernel command line option modules-load= is now understood
5885 to load a specific kernel module statically, early at boot.
5887 * Unit names specified on the systemctl command line are now
5888 automatically escaped as needed. Also, if file system or
5889 device paths are specified they are automatically turned
5890 into the appropriate mount or device unit names. Example:
5892 systemctl status /home
5893 systemctl status /dev/sda
5895 * The SysVConsole= configuration option has been removed from
5896 system.conf parsing.
5898 * The SysV search path is no longer exported on the D-Bus
5901 * The Names= option is been removed from unit file parsing.
5903 * There's a new man page bootup(7) detailing the boot process.
5905 * Every unit and every generator we ship with systemd now
5906 comes with full documentation. The self-explanatory boot is
5909 * A couple of services gained "systemd-" prefixes in their
5910 name if they wrap systemd code, rather than only external
5911 code. Among them fsck@.service which is now
5912 systemd-fsck@.service.
5914 * The HaveWatchdog property has been removed from the D-Bus
5917 * systemd.confirm_spawn= on the kernel command line should now
5920 * There's a new man page crypttab(5) which details all options
5921 we actually understand.
5923 * systemd-nspawn gained a new --capability= switch to pass
5924 additional capabilities to the container.
5926 * timedated will now read known NTP implementation unit names
5927 from /usr/lib/systemd/ntp-units.d/*.list,
5928 systemd-timedated-ntp.target has been removed.
5930 * journalctl gained a new switch "-b" that lists log data of
5931 the current boot only.
5933 * The notify socket is in the abstract namespace again, in
5934 order to support daemons which chroot() at start-up.
5936 * There is a new Storage= configuration option for journald
5937 which allows configuration of where log data should go. This
5938 also provides a way to disable journal logging entirely, so
5939 that data collected is only forwarded to the console, the
5940 kernel log buffer or another syslog implementation.
5942 * Many bugfixes and optimizations
5944 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
5945 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
5946 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
5947 Shawn Landden, Tom Gundersen
5951 * "systemctl help <unit>" now shows the man page if one is
5954 * Several new man pages have been added.
5956 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
5957 MaxLevelConsole= can now be specified in
5958 journald.conf. These options allow reducing the amount of
5959 data stored on disk or forwarded by the log level.
5961 * TimerSlackNSec= can now be specified in system.conf for
5962 PID1. This allows system-wide power savings.
5964 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
5965 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
5970 * logind is now capable of (optionally) handling power and
5971 sleep keys as well as the lid switch.
5973 * journalctl now understands the syntax "journalctl
5974 /usr/bin/avahi-daemon" to get all log output of a specific
5977 * CapabilityBoundingSet= in system.conf now also influences
5978 the capability bound set of usermode helpers of the kernel.
5980 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
5981 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
5982 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
5983 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
5987 * Note that we skipped 139 releases here in order to set the
5988 new version to something that is greater than both udev's
5989 and systemd's most recent version number.
5991 * udev: all udev sources are merged into the systemd source tree now.
5992 All future udev development will happen in the systemd tree. It
5993 is still fully supported to use the udev daemon and tools without
5994 systemd running, like in initramfs or other init systems. Building
5995 udev though, will require the *build* of the systemd tree, but
5996 udev can be properly *run* without systemd.
5998 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
5999 should be used to create dead device nodes as workarounds for broken
6002 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
6003 no longer supported. udev_monitor_new_from_netlink() needs to be
6004 used to subscribe to events.
6006 * udev: when udevd is started by systemd, processes which are left
6007 behind by forking them off of udev rules, are unconditionally cleaned
6008 up and killed now after the event handling has finished. Services or
6009 daemons must be started as systemd services. Services can be
6010 pulled-in by udev to get started, but they can no longer be directly
6011 forked by udev rules.
6013 * udev: the daemon binary is called systemd-udevd now and installed
6014 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
6015 to adapt to that, create symlink, or rename the binary after building
6018 * libudev no longer provides these symbols:
6019 udev_monitor_from_socket()
6020 udev_queue_get_failed_list_entry()
6021 udev_get_{dev,sys,run}_path()
6022 The versions number was bumped and symbol versioning introduced.
6024 * systemd-loginctl and systemd-journalctl have been renamed
6025 to loginctl and journalctl to match systemctl.
6027 * The config files: /etc/systemd/systemd-logind.conf and
6028 /etc/systemd/systemd-journald.conf have been renamed to
6029 logind.conf and journald.conf. Package updates should rename
6030 the files to the new names on upgrade.
6032 * For almost all files the license is now LGPL2.1+, changed
6033 from the previous GPL2.0+. Exceptions are some minor stuff
6034 of udev (which will be changed to LGPL2.1 eventually, too),
6035 and the MIT licensed sd-daemon.[ch] library that is suitable
6036 to be used as drop-in files.
6038 * systemd and logind now handle system sleep states, in
6039 particular suspending and hibernating.
6041 * logind now implements a sleep/shutdown/idle inhibiting logic
6042 suitable for a variety of uses. Soonishly Lennart will blog
6043 about this in more detail.
6045 * var-run.mount and var-lock.mount are no longer provided
6046 (which prevously bind mounted these directories to their new
6047 places). Distributions which have not converted these
6048 directories to symlinks should consider stealing these files
6049 from git history and add them downstream.
6051 * We introduced the Documentation= field for units and added
6052 this to all our shipped units. This is useful to make it
6053 easier to explore the boot and the purpose of the various
6056 * All smaller setup units (such as
6057 systemd-vconsole-setup.service) now detect properly if they
6058 are run in a container and are skipped when
6059 appropriate. This guarantees an entirely noise-free boot in
6060 Linux container environments such as systemd-nspawn.
6062 * A framework for implementing offline system updates is now
6063 integrated, for details see:
6064 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
6066 * A new service type Type=idle is available now which helps us
6067 avoiding ugly interleaving of getty output and boot status
6070 * There's now a system-wide CapabilityBoundingSet= option to
6071 globally reduce the set of capabilities for the
6072 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
6073 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
6074 even CAP_NET_ADMIN system-wide for secure systems.
6076 * There are now system-wide DefaultLimitXXX= options to
6077 globally change the defaults of the various resource limits
6078 for all units started by PID 1.
6080 * Harald Hoyer's systemd test suite has been integrated into
6081 systemd which allows easy testing of systemd builds in qemu
6082 and nspawn. (This is really awesome! Ask us for details!)
6084 * The fstab parser is now implemented as generator, not inside
6087 * systemctl will now warn you if .mount units generated from
6088 /etc/fstab are out of date due to changes in fstab that
6089 have not been read by systemd yet.
6091 * systemd is now suitable for usage in initrds. Dracut has
6092 already been updated to make use of this. With this in place
6093 initrds get a slight bit faster but primarily are much
6094 easier to introspect and debug since "systemctl status" in
6095 the host system can be used to introspect initrd services,
6096 and the journal from the initrd is kept around too.
6098 * systemd-delta has been added, a tool to explore differences
6099 between user/admin configuration and vendor defaults.
6101 * PrivateTmp= now affects both /tmp and /var/tmp.
6103 * Boot time status messages are now much prettier and feature
6104 proper english language. Booting up systemd has never been
6107 * Read-ahead pack files now include the inode number of all
6108 files to pre-cache. When the inode changes the pre-caching
6109 is not attempted. This should be nicer to deal with updated
6110 packages which might result in changes of read-ahead
6113 * We now temporaritly lower the kernel's read_ahead_kb variable
6114 when collecting read-ahead data to ensure the kernel's
6115 built-in read-ahead does not add noise to our measurements
6116 of necessary blocks to pre-cache.
6118 * There's now RequiresMountsFor= to add automatic dependencies
6119 for all mounts necessary for a specific file system path.
6121 * MountAuto= and SwapAuto= have been removed from
6122 system.conf. Mounting file systems at boot has to take place
6125 * nspawn now learned a new switch --uuid= to set the machine
6126 ID on the command line.
6128 * nspawn now learned the -b switch to automatically search
6131 * vt102 is now the default TERM for serial TTYs, upgraded from
6134 * systemd-logind now works on VT-less systems.
6136 * The build tree has been reorganized. The individual
6137 components now have directories of their own.
6139 * A new condition type ConditionPathIsReadWrite= is now available.
6141 * nspawn learned the new -C switch to create cgroups for the
6142 container in other hierarchies.
6144 * We now have support for hardware watchdogs, configurable in
6147 * The scheduled shutdown logic now has a public API.
6149 * We now mount /tmp as tmpfs by default, but this can be
6150 masked and /etc/fstab can override it.
6152 * Since udisks does not make use of /media anymore we are not
6153 mounting a tmpfs on it anymore.
6155 * journalctl gained a new --local switch to only interleave
6156 locally generated journal files.
6158 * We can now load the IMA policy at boot automatically.
6160 * The GTK tools have been split off into a systemd-ui.
6162 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
6163 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
6164 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
6165 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
6166 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
6167 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
6168 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
6169 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
6170 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
6175 * This is mostly a bugfix release
6177 * Support optional initialization of the machine ID from the
6178 KVM or container configured UUID.
6180 * Support immediate reboots with "systemctl reboot -ff"
6182 * Show /etc/os-release data in systemd-analyze output
6184 * Many bugfixes for the journal, including endianness fixes and
6185 ensuring that disk space enforcement works
6187 * sd-login.h is C++ comptaible again
6189 * Extend the /etc/os-release format on request of the Debian
6192 * We now refuse non-UTF8 strings used in various configuration
6193 and unit files. This is done to ensure we do not pass invalid
6194 data over D-Bus or expose it elsewhere.
6196 * Register Mimo USB Screens as suitable for automatic seat
6199 * Read SELinux client context from journal clients in a race
6202 * Reorder configuration file lookup order. /etc now always
6203 overrides /run in order to allow the administrator to always
6204 and unconditionally override vendor-supplied or
6205 automatically generated data.
6207 * The various user visible bits of the journal now have man
6208 pages. We still lack man pages for the journal API calls
6211 * We now ship all man pages in HTML format again in the
6214 Contributions from: Dave Reisner, Dirk Eibach, Frederic
6215 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
6216 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
6221 * This is mostly a bugfix release
6223 * systems lacking /etc/os-release are no longer supported.
6225 * Various functionality updates to libsystemd-login.so
6227 * Track class of PAM logins to distinguish greeters from
6230 Contributions from: Kay Sievers, Lennart Poettering, Michael
6235 * This is an important bugfix release for v41.
6237 * Building man pages is now optional which should be useful
6238 for those building systemd from git but unwilling to install
6241 * Watchdog support for supervising services is now usable. In
6242 a future release support for hardware watchdogs
6243 (i.e. /dev/watchdog) will be added building on this.
6245 * Service start rate limiting is now configurable and can be
6246 turned off per service. When a start rate limit is hit a
6247 reboot can automatically be triggered.
6249 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
6251 Contributions from: Benjamin Franzke, Bill Nottingham,
6252 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
6253 Schmidt, Michał Górny, Piotr Drąg
6257 * The systemd binary is installed /usr/lib/systemd/systemd now;
6258 An existing /sbin/init symlink needs to be adapted with the
6261 * The code that loads kernel modules has been ported to invoke
6262 libkmod directly, instead of modprobe. This means we do not
6263 support systems with module-init-tools anymore.
6265 * Watchdog support is now already useful, but still not
6268 * A new kernel command line option systemd.setenv= is
6269 understood to set system wide environment variables
6270 dynamically at boot.
6272 * We now limit the set of capabilities of systemd-journald.
6274 * We now set SIGPIPE to ignore by default, since it only is
6275 useful in shell pipelines, and has little use in general
6276 code. This can be disabled with IgnoreSIPIPE=no in unit
6279 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
6280 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
6285 * This is mostly a bugfix release
6287 * We now expose the reason why a service failed in the
6288 "Result" D-Bus property.
6290 * Rudimentary service watchdog support (will be completed over
6291 the next few releases.)
6293 * When systemd forks off in order execute some service we will
6294 now immediately changes its argv[0] to reflect which process
6295 it will execute. This is useful to minimize the time window
6296 with a generic argv[0], which makes bootcharts more useful
6298 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
6299 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
6300 Mike Kazantsev, Ray Strode
6304 * This is mostly a test release, but incorporates many
6307 * New systemd-cgtop tool to show control groups by their
6310 * Linking against libacl for ACLs is optional again. If
6311 disabled, support tracking device access for active logins
6312 goes becomes unavailable, and so does access to the user
6313 journals by the respective users.
6315 * If a group "adm" exists, journal files are automatically
6316 owned by them, thus allow members of this group full access
6317 to the system journal as well as all user journals.
6319 * The journal now stores the SELinux context of the logging
6320 client for all entries.
6322 * Add C++ inclusion guards to all public headers
6324 * New output mode "cat" in the journal to print only text
6325 messages, without any meta data like date or time.
6327 * Include tiny X server wrapper as a temporary stop-gap to
6328 teach XOrg udev display enumeration. This is used by display
6329 managers such as gdm, and will go away as soon as XOrg
6330 learned native udev hotplugging for display devices.
6332 * Add new systemd-cat tool for executing arbitrary programs
6333 with STDERR/STDOUT connected to the journal. Can also act as
6334 BSD logger replacement, and does so by default.
6336 * Optionally store all locally generated coredumps in the
6337 journal along with meta data.
6339 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
6340 writing short strings to files (for usage for /sys), and for
6341 creating symlinks, character and block device nodes.
6343 * New unit file option ControlGroupPersistent= to make cgroups
6344 persistent, following the mechanisms outlined in
6345 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
6347 * Support multiple local RTCs in a sane way
6349 * No longer monopolize IO when replaying readahead data on
6350 rotating disks, since we might starve non-file-system IO to
6351 death, since fanotify() will not see accesses done by blkid,
6354 * Do not show kernel threads in systemd-cgls anymore, unless
6355 requested with new -k switch.
6357 Contributions from: Dan Horák, Kay Sievers, Lennart
6358 Poettering, Michal Schmidt
6362 * This is mostly a test release, but incorporates many
6365 * The git repository moved to:
6366 git://anongit.freedesktop.org/systemd/systemd
6367 ssh://git.freedesktop.org/git/systemd/systemd
6369 * First release with the journal
6370 http://0pointer.de/blog/projects/the-journal.html
6372 * The journal replaces both systemd-kmsg-syslogd and
6373 systemd-stdout-bridge.
6375 * New sd_pid_get_unit() API call in libsystemd-logind
6377 * Many systemadm clean-ups
6379 * Introduce remote-fs-pre.target which is ordered before all
6380 remote mounts and may be used to start services before all
6383 * Added Mageia support
6385 * Add bash completion for systemd-loginctl
6387 * Actively monitor PID file creation for daemons which exit in
6388 the parent process before having finished writing the PID
6389 file in the daemon process. Daemons which do this need to be
6390 fixed (i.e. PID file creation must have finished before the
6391 parent exits), but we now react a bit more gracefully to them.
6393 * Add colourful boot output, mimicking the well-known output
6394 of existing distributions.
6396 * New option PassCredentials= for socket units, for
6397 compatibility with a recent kernel ABI breakage.
6399 * /etc/rc.local is now hooked in via a generator binary, and
6400 thus will no longer act as synchronization point during
6403 * systemctl list-unit-files now supports --root=.
6405 * systemd-tmpfiles now understands two new commands: z, Z for
6406 relabelling files according to the SELinux database. This is
6407 useful to apply SELinux labels to specific files in /sys,
6410 * Output of SysV services is now forwarded to both the console
6411 and the journal by default, not only just the console.
6413 * New man pages for all APIs from libsystemd-login.
6415 * The build tree got reorganized and a the build system is a
6416 lot more modular allowing embedded setups to specifically
6417 select the components of systemd they are interested in.
6419 * Support for Linux systems lacking the kernel VT subsystem is
6422 * configure's --with-rootdir= got renamed to
6423 --with-rootprefix= to follow the naming used by udev and
6426 * Unless specified otherwise we will now install to /usr instead
6427 of /usr/local by default.
6429 * Processes with '@' in argv[0][0] are now excluded from the
6430 final shut-down killing spree, following the logic explained
6432 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
6434 * All processes remaining in a service cgroup when we enter
6435 the START or START_PRE states are now killed with
6436 SIGKILL. That means it is no longer possible to spawn
6437 background processes from ExecStart= lines (which was never
6438 supported anyway, and bad style).
6440 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
6441 reloading of units together.
6443 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
6444 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
6445 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
6446 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
6447 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek