1 systemd System and Service Manager
3 CHANGES WITH 232 in spe
5 * Journald's SplitMode=login setting has been deprecated. It has been
6 removed from documentation, and it's use is discouraged. In a future
7 release it will be completely removed, and made equivalent to current
8 default of SplitMode=uid.
12 * In service units the various ExecXYZ= settings have been extended
13 with an additional special character as first argument of the
14 assigned value: if the character '+' is used the specified command
15 line it will be run with full privileges, regardless of User=,
16 Group=, CapabilityBoundingSet= and similar options. The effect is
17 similar to the existing PermissionsStartOnly= option, but allows
18 configuration of this concept for each executed command line
21 * Services may now alter the service watchdog timeout at runtime by
22 sending a WATCHDOG_USEC= message via sd_notify().
24 * MemoryLimit= and related unit settings now optionally take percentage
25 specifications. The percentage is taken relative to the amount of
26 physical memory in the system (or in case of containers, the assigned
27 amount of memory). This allows scaling service resources neatly with
28 the amount of RAM available on the system. Similarly, systemd-logind's
29 RuntimeDirectorySize= option now also optionally takes percentage
32 * In similar fashion TasksMax= takes percentage values now, too. The
33 value is taken relative to the configured maximum number of processes
34 on the system. The per-service task maximum has been changed to 15%
35 using this functionality. (Effectively this is an increase of 512 →
36 4915 for service units, given the kernel's default pid_max setting.)
38 * Calendar time specifications in .timer units now understand a ".."
39 syntax for time ranges. Example: "4..7:10" may now be used for
40 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
43 * The InaccessableDirectories=, ReadOnlyDirectories= and
44 ReadWriteDirectories= unit file settings have been renamed to
45 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
46 applied to all kinds of file nodes, and not just directories, with
47 the exception of symlinks. Specifically these settings may now be
48 used on block and character device nodes, UNIX sockets and FIFOS as
49 well as regular files. The old names of these settings remain
50 available for compatibility.
52 * systemd will now log about all service processes it kills forcibly
53 (using SIGKILL) because they remained after the clean shutdown phase
54 of the service completed. This should help identifying services that
55 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
56 systemd-logind's configuration a similar log message is generated for
57 processes killed at the end of each session due to this setting.
59 * systemd will now set the $JOURNAL_STREAM environment variable for all
60 services whose stdout/stderr are connected to the Journal (which
61 effectively means by default: all services). The variable contains
62 the device and inode number of the file descriptor used for
63 stdout/stderr. This may be used by invoked programs to detect whether
64 their stdout/stderr is connected to the Journal, in which case they
65 can switch over to direct Journal communication, thus being able to
66 pass extended, structured metadata along with their log messages. As
67 one example, this is now used by glib's logging primitives.
69 * When using systemd's default tmp.mount unit for /tmp, the mount point
70 will now be established with the "nosuid" and "nodev" options. This
71 avoids privilege escalation attacks that put traps and exploits into
72 /tmp. However, this might cause problems if you e. g. put container
73 images or overlays into /tmp; if you need this, override tmp.mount's
74 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
77 * systemd now supports the "memory" cgroup controller also on
80 * The systemd-cgtop tool now optionally takes a control group path as
81 command line argument. If specified, the control group list shown is
82 limited to subgroups of that group.
84 * The SystemCallFilter= unit file setting gained support for
85 pre-defined, named system call filter sets. For example
86 SystemCallFilter=@clock is now an effective way to make all clock
87 changing-related system calls unavailable to a service. A number of
88 similar pre-defined groups are defined. Writing system call filters
89 for system services is simplified substantially with this new
90 concept. Accordingly, all of systemd's own, long-running services now
91 enable system call filtering based on this, by default.
93 * A new service setting MemoryDenyWriteExecute= has been added, taking
94 a boolean value. If turned on, a service may no longer create memory
95 mappings that are writable and executable at the same time. This
96 enhances security for services where this is enabled as it becomes
97 harder to dynamically write and then execute memory in exploited
98 service processes. This option has been enabled for all of systemd's
99 own long-running services.
101 * A new RestrictRealtime= service setting has been added, taking a
102 boolean argument. If set the service's processes may no longer
103 acquire realtime scheduling. This improves security as realtime
104 scheduling may otherwise be used to easily freeze the system.
106 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
107 value. This may be used for requesting that the system manager inside
108 of the container reports start-up completion to nspawn which then
109 propagates this notification further to the service manager
110 supervising nspawn itself. A related option NotifyReady= in .nspawn
111 files has been added too. This functionality allows ordering of the
112 start-up of multiple containers using the usual systemd ordering
115 * machinectl gained a new command "stop" that is an alias for
118 * systemd-resolved gained support for contacting DNS servers on
119 link-local IPv6 addresses.
121 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
122 its caches. A method call for requesting the same operation has been
123 added to the bus API too, and is made available via "systemd-resolve
126 * systemd-resolve gained a new --status switch. If passed a brief
127 summary of the used DNS configuration with per-interface information
130 * resolved.conf gained a new Cache= boolean option, defaulting to
131 on. If turned off local DNS caching is disabled. This comes with a
132 performance penalty in particular when DNSSEC is enabled. Note that
133 resolved disables its internal caching implicitly anyway, when the
134 configured DNS server is on a host-local IP address such as ::1 or
135 127.0.0.1, thus automatically avoiding double local caching.
137 * systemd-resolved now listens on the local IP address 127.0.0.53:53
138 for DNS requests. This improves compatibility with local programs
139 that do not use the libc NSS or systemd-resolved's bus APIs for name
140 resolution. This minimal DNS service is only available to local
141 programs and does not implement the full DNS protocol, but enough to
142 cover local DNS clients. A new, static resolv.conf file, listing just
143 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
144 now recommended to make /etc/resolv.conf a symlink to this file in
145 order to route all DNS lookups to systemd-resolved, regardless if
146 done via NSS, the bus API or raw DNS packets. Note that this local
147 DNS service is not as fully featured as the libc NSS or
148 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
149 used to deliver link-local address information (as this implies
150 sending a local interface index along), LLMNR/mDNS support via this
151 interface is severely restricted. It is thus strongly recommended for
152 all applications to use the libc NSS API or native systemd-resolved
155 * systemd-networkd's bridge support learned a new setting
156 VLANFiltering= for controlling VLAN filtering. Moreover a new section
157 in .network files has been added for configuring VLAN bridging in
158 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
160 * systemd-networkd's IPv6 Router Advertisement code now makes use of
161 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
162 now be acquired without relying on DHCPv6. Two new options
163 UseDomains= and UseDNS= have been added to configure this behaviour.
165 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
166 renamed IPv6AcceptRA=, without altering its behaviour. The old
167 setting name remains available for compatibility reasons.
169 * The systemd-networkd VTI/VTI6 tunneling support gained new options
170 Key=, InputKey= and OutputKey=.
172 * systemd-networkd gained support for VRF ("Virtual Routing Function")
173 interface configuration.
175 * "systemctl edit" may now be used to create new unit files by
176 specifying the --force switch.
178 * sd-event gained a new function sd_event_get_iteration() for
179 requesting the current iteration counter of the event loop. It starts
180 at zero and is increased by one with each event loop iteration.
182 * A new rpm macro %systemd_ordering is provided by the macros.systemd
183 file. It can be used in lieu of %systemd_requires in packages which
184 don't use any systemd functionality and are intended to be installed
185 in minimal containers without systemd present. This macro provides
186 ordering dependecies to ensure that if the package is installed in
187 the same rpm transaction as systemd, systemd will be installed before
188 the scriptlets for the package are executed, allowing unit presets
191 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
192 been added to simplify packaging of generators.
194 * The os-release file gained VERSION_CODENAME field for the
195 distribution nickname (e.g. VERSION_CODENAME=woody).
197 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
198 can be set to disable parsing of metadata and the creation
199 of persistent symlinks for that device.
201 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
202 to make them available to logged-in users has been reverted.
204 * Much of the common code of the various systemd components is now
205 built into an internal shared library libsystemd-shared-231.so
206 (incorporating the systemd version number in the name, to be updated
207 with future releases) that the components link to. This should
208 decrease systemd footprint both in memory during runtime and on
209 disk. Note that the shared library is not for public use, and is
210 neither API not ABI stable, but is likely to change with every new
211 released update. Packagers need to make sure that binaries
212 linking to libsystemd-shared.so are updated in step with the
215 * Configuration for "mkosi" is now part of the systemd
216 repository. mkosi is a tool to easily build legacy-free OS images,
217 and is available on github: https://github.com/systemd/mkosi. If
218 "mkosi" is invoked in the build tree a new raw OS image is generated
219 incorporating the systemd sources currently being worked on and a
220 clean, fresh distribution installation. The generated OS image may be
221 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physcial
222 UEFI PC. This functionality is particularly useful to easily test
223 local changes made to systemd in a pristine, defined environment. See
226 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
227 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
228 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
229 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
230 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
231 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
232 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
233 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
234 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
235 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
236 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
237 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
238 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
239 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
240 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
241 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
242 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
243 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
244 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
250 * DNSSEC is now turned on by default in systemd-resolved (in
251 "allow-downgrade" mode), but may be turned off during compile time by
252 passing "--with-default-dnssec=no" to "configure" (and of course,
253 during runtime with DNSSEC= in resolved.conf). We recommend
254 downstreams to leave this on at least during development cycles and
255 report any issues with the DNSSEC logic upstream. We are very
256 interested in collecting feedback about the DNSSEC validator and its
257 limitations in the wild. Note however, that DNSSEC support is
258 probably nothing downstreams should turn on in stable distros just
259 yet, as it might create incompatibilities with a few DNS servers and
260 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
261 automatically whenever we detect such incompatible setups, but there
262 might be systems we do not cover yet. Hence: please help us testing
263 the DNSSEC code, leave this on where you can, report back, but then
264 again don't consider turning this on in your stable, LTS or
265 production release just yet. (Note that you have to enable
266 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
267 and its DNSSEC mode for host name resolution from local
270 * systemd-resolve conveniently resolves DANE records with the --tlsa
271 option and OPENPGPKEY records with the --openpgp option. It also
272 supports dumping raw DNS record data via the new --raw= switch.
274 * systemd-logind will now by default terminate user processes that are
275 part of the user session scope unit (session-XX.scope) when the user
276 logs out. This behavior is controlled by the KillUserProcesses=
277 setting in logind.conf, and the previous default of "no" is now
278 changed to "yes". This means that user sessions will be properly
279 cleaned up after, but additional steps are necessary to allow
280 intentionally long-running processes to survive logout.
282 While the user is logged in at least once, user@.service is running,
283 and any service that should survive the end of any individual login
284 session can be started at a user service or scope using systemd-run.
285 systemd-run(1) man page has been extended with an example which shows
286 how to run screen in a scope unit underneath user@.service. The same
287 command works for tmux.
289 After the user logs out of all sessions, user@.service will be
290 terminated too, by default, unless the user has "lingering" enabled.
291 To effectively allow users to run long-term tasks even if they are
292 logged out, lingering must be enabled for them. See loginctl(1) for
293 details. The default polkit policy was modified to allow users to
294 set lingering for themselves without authentication.
296 Previous defaults can be restored at compile time by the
297 --without-kill-user-processes option to "configure".
299 * systemd-logind gained new configuration settings SessionsMax= and
300 InhibitorsMax=, both with a default of 8192. It will not register new
301 user sessions or inhibitors above this limit.
303 * systemd-logind will now reload configuration on SIGHUP.
305 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
306 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
307 enable. Also, support for the "io" cgroup controller in the unified
308 hierarchy has been added, so that the "memory", "pids" and "io" are
309 now the controllers that are supported on the unified hierarchy.
311 WARNING: it is not possible to use previous systemd versions with
312 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
313 is necessary to also update systemd in the initramfs if using the
314 unified hierarchy. An updated SELinux policy is also required.
316 * LLDP support has been extended, and both passive (receive-only) and
317 active (sender) modes are supported. Passive mode ("routers-only") is
318 enabled by default in systemd-networkd. Active LLDP mode is enabled
319 by default for containers on the internal network. The "networkctl
320 lldp" command may be used to list information gathered. "networkctl
321 status" will also show basic LLDP information on connected peers now.
323 * The IAID and DUID unique identifier sent in DHCP requests may now be
324 configured for the system and each .network file managed by
325 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
327 * systemd-networkd gained support for configuring proxy ARP support for
328 each interface, via the ProxyArp= setting in .network files. It also
329 gained support for configuring the multicast querier feature of
330 bridge devices, via the new MulticastQuerier= setting in .netdev
331 files. Similarly, snooping on the IGMP traffic can be controlled
332 via the new setting MulticastSnooping=.
334 A new setting PreferredLifetime= has been added for addresses
335 configured in .network file to configure the lifetime intended for an
338 The systemd-networkd DHCP server gained the option EmitRouter=, which
339 defaults to yes, to configure whether the DHCP Option 3 (Router)
342 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
343 systemd-socket-activate and installed into /usr/bin. It is now fully
346 * systemd-journald now uses separate threads to flush changes to disk
347 when closing journal files, thus reducing impact of slow disk I/O on
350 * The sd-journal API gained two new calls
351 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
352 can be used to open journal files using file descriptors instead of
353 file or directory paths. sd_journal_open_container() has been
354 deprecated, sd_journal_open_directory_fd() should be used instead
355 with the flag SD_JOURNAL_OS_ROOT.
357 * journalctl learned a new output mode "-o short-unix" that outputs log
358 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
359 UTC). It also gained support for a new --no-hostname setting to
360 suppress the hostname column in the family of "short" output modes.
362 * systemd-ask-password now optionally skips printing of the password to
363 stdout with --no-output which can be useful in scripts.
365 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
366 (devices tagged with ID_MAKER_TOOL) are now tagged with
367 "uaccess" and are available to logged in users.
369 * The DeviceAllow= unit setting now supports specifiers (with "%").
371 * "systemctl show" gained a new --value switch, which allows print a
372 only the contents of a specific unit property, without also printing
373 the property's name. Similar support was added to "show*" verbs
374 of loginctl and machinectl that output "key=value" lists.
376 * A new unit type "generated" was added for files dynamically generated
377 by generator tools. Similarly, a new unit type "transient" is used
378 for unit files created using the runtime API. "systemctl enable" will
379 refuse to operate on such files.
381 * A new command "systemctl revert" has been added that may be used to
382 revert to the vendor version of a unit file, in case local changes
383 have been made by adding drop-ins or overriding the unit file.
385 * "machinectl clean" gained a new verb to automatically remove all or
386 just hidden container images.
388 * systemd-tmpfiles gained support for a new line type "e" for emptying
389 directories, if they exist, without creating them if they don't.
391 * systemd-nspawn gained support for automatically patching the UID/GIDs
392 of the owners and the ACLs of all files and directories in a
393 container tree to match the UID/GID user namespacing range selected
394 for the container invocation. This mode is enabled via the new
395 --private-user-chown switch. It also gained support for automatically
396 choosing a free, previously unused UID/GID range when starting a
397 container, via the new --private-users=pick setting (which implies
398 --private-user-chown). Together, these options for the first time
399 make user namespacing for nspawn containers fully automatic and thus
400 deployable. The systemd-nspawn@.service template unit file has been
401 changed to use this functionality by default.
403 * systemd-nspawn gained a new --network-zone= switch, that allows
404 creating ad-hoc virtual Ethernet links between multiple containers,
405 that only exist as long as at least one container referencing them is
406 running. This allows easy connecting of multiple containers with a
407 common link that implements an Ethernet broadcast domain. Each of
408 these network "zones" may be named relatively freely by the user, and
409 may be referenced by any number of containers, but each container may
410 only reference one of these "zones". On the lower level, this is
411 implemented by an automatically managed bridge network interface for
412 each zone, that is created when the first container referencing its
413 zone is created and removed when the last one referencing its zone
416 * The default start timeout may now be configured on the kernel command
417 line via systemd.default_timeout_start_sec=. It was already
418 configurable via the DefaultTimeoutStartSec= option in
419 /etc/systemd/system.conf.
421 * Socket units gained a new TriggerLimitIntervalSec= and
422 TriggerLimitBurst= setting to configure a limit on the activation
423 rate of the socket unit.
425 * The LimitNICE= setting now optionally takes normal UNIX nice values
426 in addition to the raw integer limit value. If the specified
427 parameter is prefixed with "+" or "-" and is in the range -20..19 the
428 value is understood as UNIX nice value. If not prefixed like this it
429 is understood as raw RLIMIT_NICE limit.
431 * Note that the effect of the PrivateDevices= unit file setting changed
432 slightly with this release: the per-device /dev file system will be
433 mounted read-only from this version on, and will have "noexec"
434 set. This (minor) change of behavior might cause some (exceptional)
435 legacy software to break, when PrivateDevices=yes is set for its
436 service. Please leave PrivateDevices= off if you run into problems
439 * systemd-bootchart has been split out to a separate repository:
440 https://github.com/systemd/systemd-bootchart
442 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
443 merged into the kernel in its current form.
445 * The compatibility libraries libsystemd-daemon.so,
446 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
447 which have been deprecated since systemd-209 have been removed along
448 with the corresponding pkg-config files. All symbols provided by
449 those libraries are provided by libsystemd.so.
451 * The Capabilities= unit file setting has been removed (it is ignored
452 for backwards compatibility). AmbientCapabilities= and
453 CapabilityBoundingSet= should be used instead.
455 * A new special target has been added, initrd-root-device.target,
456 which creates a synchronization point for dependencies of the root
457 device in early userspace. Initramfs builders must ensure that this
458 target is now included in early userspace.
460 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
461 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
462 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
463 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
464 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
465 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
466 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
467 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
468 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
469 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
470 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
471 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
472 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
473 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
474 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
475 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
476 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
477 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
478 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
479 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
480 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
481 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
482 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
483 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
484 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
487 — Fairfax, 2016-05-21
491 * The systemd-resolved DNS resolver service has gained a substantial
492 set of new features, most prominently it may now act as a DNSSEC
493 validating stub resolver. DNSSEC mode is currently turned off by
494 default, but is expected to be turned on by default in one of the
495 next releases. For now, we invite everybody to test the DNSSEC logic
496 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
497 service also gained a full set of D-Bus interfaces, including calls
498 to configure DNS and DNSSEC settings per link (for use by external
499 network management software). systemd-resolved and systemd-networkd
500 now distinguish between "search" and "routing" domains. The former
501 are used to qualify single-label names, the latter are used purely
502 for routing lookups within certain domains to specific links.
503 resolved now also synthesizes RRs for all entries from /etc/hosts.
505 * The systemd-resolve tool (which is a client utility for
506 systemd-resolved) has been improved considerably and is now fully
507 supported and documented. Hence it has moved from /usr/lib/systemd to
510 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
513 * The coredump collection logic has been reworked: when a coredump is
514 collected it is now written to disk, compressed and processed
515 (including stacktrace extraction) from a new instantiated service
516 systemd-coredump@.service, instead of directly from the
517 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
518 processing large coredumps can take up a substantial amount of
519 resources and time, and this previously happened entirely outside of
520 systemd's service supervision. With the new logic the core_pattern
521 hook only does minimal metadata collection before passing off control
522 to the new instantiated service, which is configured with a time
523 limit, a nice level and other settings to minimize negative impact on
524 the rest of the system. Also note that the new logic will honour the
525 RLIMIT_CORE setting of the crashed process, which now allows users
526 and processes to turn off coredumping for their processes by setting
529 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
530 and all forked processes by default. Previously, PID 1 would leave
531 the setting at "0" for all processes, as set by the kernel. Note that
532 the resource limit traditionally has no effect on the generated
533 coredumps on the system if the /proc/sys/kernel/core_pattern hook
534 logic is used. Since the limit is now honoured (see above) its
535 default has been changed so that the coredumping logic is enabled by
536 default for all processes, while allowing specific opt-out.
538 * When the stacktrace is extracted from processes of system users, this
539 is now done as "systemd-coredump" user, in order to sandbox this
540 potentially security sensitive parsing operation. (Note that when
541 processing coredumps of normal users this is done under the user ID
542 of process that crashed, as before.) Packagers should take notice
543 that it is now necessary to create the "systemd-coredump" system user
544 and group at package installation time.
546 * The systemd-activate socket activation testing tool gained support
547 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
548 and --seqpacket switches. It also has been extended to support both
549 new-style and inetd-style file descriptor passing. Use the new
550 --inetd switch to request inetd-style file descriptor passing.
552 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
553 variable, which takes a boolean value. If set to false, ANSI color
554 output is disabled in the tools even when run on a terminal that
557 * The VXLAN support in networkd now supports two new settings
558 DestinationPort= and PortRange=.
560 * A new systemd.machine_id= kernel command line switch has been added,
561 that may be used to set the machine ID in /etc/machine-id if it is
562 not initialized yet. This command line option has no effect if the
563 file is already initialized.
565 * systemd-nspawn gained a new --as-pid2 switch that invokes any
566 specified command line as PID 2 rather than PID 1 in the
567 container. In this mode PID 1 is a minimal stub init process that
568 implements the special POSIX and Linux semantics of PID 1 regarding
569 signal and child process management. Note that this stub init process
570 is implemented in nspawn itself and requires no support from the
571 container image. This new logic is useful to support running
572 arbitrary commands in the container, as normal processes are
573 generally not prepared to run as PID 1.
575 * systemd-nspawn gained a new --chdir= switch for setting the current
576 working directory for the process started in the container.
578 * "journalctl /dev/sda" will now output all kernel log messages for
579 specified device from the current boot, in addition to all devices
580 that are parents of it. This should make log output about devices
581 pretty useful, as long as kernel drivers attach enough metadata to
582 the log messages. (The usual SATA drivers do.)
584 * The sd-journal API gained two new calls
585 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
586 that report whether log data from /run or /var has been found.
588 * journalctl gained a new switch "--fields" that prints all journal
589 record field names currently in use in the journal. This is backed
590 by two new sd-journal API calls sd_journal_enumerate_fields() and
591 sd_journal_restart_fields().
593 * Most configurable timeouts in systemd now expect an argument of
594 "infinity" to turn them off, instead of "0" as before. The semantics
595 from now on is that a timeout of "0" means "now", and "infinity"
596 means "never". To maintain backwards compatibility, "0" continues to
597 turn off previously existing timeout settings.
599 * "systemctl reload-or-try-restart" has been renamed to "systemctl
600 try-reload-or-restart" to clarify what it actually does: the "try"
601 logic applies to both reloading and restarting, not just restarting.
602 The old name continues to be accepted for compatibility.
604 * On boot-up, when PID 1 detects that the system clock is behind the
605 release date of the systemd version in use, the clock is now set
606 to the latter. Previously, this was already done in timesyncd, in order
607 to avoid running with clocks set to the various clock epochs such as
608 1902, 1938 or 1970. With this change the logic is now done in PID 1
609 in addition to timesyncd during early boot-up, so that it is enforced
610 before the first process is spawned by systemd. Note that the logic
611 in timesyncd remains, as it is more comprehensive and ensures
612 clock monotonicity by maintaining a persistent timestamp file in
613 /var. Since /var is generally not available in earliest boot or the
614 initrd, this part of the logic remains in timesyncd, and is not done
617 * Support for tweaking details in net_cls.class_id through the
618 NetClass= configuration directive has been removed, as the kernel
619 people have decided to deprecate that controller in cgroup v2.
620 Userspace tools such as nftables are moving over to setting rules
621 that are specific to the full cgroup path of a task, which obsoletes
622 these controllers anyway. The NetClass= directive is kept around for
623 legacy compatibility reasons. For a more in-depth description of the
624 kernel change, please refer to the respective upstream commit:
626 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
628 * A new service setting RuntimeMaxSec= has been added that may be used
629 to specify a maximum runtime for a service. If the timeout is hit, the
630 service is terminated and put into a failure state.
632 * A new service setting AmbientCapabilities= has been added. It allows
633 configuration of additional Linux process capabilities that are
634 passed to the activated processes. This is only available on very
637 * The process resource limit settings in service units may now be used
638 to configure hard and soft limits individually.
640 * The various libsystemd APIs such as sd-bus or sd-event now publicly
641 expose support for gcc's __attribute__((cleanup())) C extension.
642 Specifically, for many object destructor functions alternative
643 versions have been added that have names suffixed with "p" and take a
644 pointer to a pointer to the object to destroy, instead of just a
645 pointer to the object itself. This is useful because these destructor
646 functions may be used directly as parameters to the cleanup
647 construct. Internally, systemd has been a heavy user of this GCC
648 extension for a long time, and with this change similar support is
649 now available to consumers of the library outside of systemd. Note
650 that by using this extension in your sources compatibility with old
651 and strictly ANSI compatible C compilers is lost. However, all gcc or
652 LLVM versions of recent years support this extension.
654 * Timer units gained support for a new setting RandomizedDelaySec= that
655 allows configuring some additional randomized delay to the configured
656 time. This is useful to spread out timer events to avoid load peaks in
657 clusters or larger setups.
659 * Calendar time specifications now support sub-second accuracy.
661 * Socket units now support listening on SCTP and UDP-lite protocol
664 * The sd-event API now comes with a full set of man pages.
666 * Older versions of systemd contained experimental support for
667 compressing journal files and coredumps with the LZ4 compressor that
668 was not compatible with the lz4 binary (due to API limitations of the
669 lz4 library). This support has been removed; only support for files
670 compatible with the lz4 binary remains. This LZ4 logic is now
671 officially supported and no longer considered experimental.
673 * The dkr image import logic has been removed again from importd. dkr's
674 micro-services focus doesn't fit into the machine image focus of
675 importd, and quickly got out of date with the upstream dkr API.
677 * Creation of the /run/lock/lockdev/ directory was dropped from
678 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
679 been available for many years. If you still need this, you need to
680 create your own tmpfiles.d config file with:
682 d /run/lock/lockdev 0775 root lock -
684 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
685 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
686 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
687 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
688 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
689 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
690 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
691 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
692 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
693 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
694 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
695 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
696 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
697 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
698 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
699 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
700 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
701 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
702 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
708 * A number of properties previously only settable in unit
709 files are now also available as properties to set when
710 creating transient units programmatically via the bus, as it
711 is exposed with systemd-run's --property=
712 setting. Specifically, these are: SyslogIdentifier=,
713 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
714 EnvironmentFile=, ReadWriteDirectories=,
715 ReadOnlyDirectories=, InaccessibleDirectories=,
716 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
718 * When creating transient services via the bus API it is now
719 possible to pass in a set of file descriptors to use as
720 STDIN/STDOUT/STDERR for the invoked process.
722 * Slice units may now be created transiently via the bus APIs,
723 similar to the way service and scope units may already be
726 * Wherever systemd expects a calendar timestamp specification
727 (like in journalctl's --since= and --until= switches) UTC
728 timestamps are now supported. Timestamps suffixed with "UTC"
729 are now considered to be in Universal Time Coordinated
730 instead of the local timezone. Also, timestamps may now
731 optionally be specified with sub-second accuracy. Both of
732 these additions also apply to recurring calendar event
733 specification, such as OnCalendar= in timer units.
735 * journalctl gained a new "--sync" switch that asks the
736 journal daemon to write all so far unwritten log messages to
737 disk and sync the files, before returning.
739 * systemd-tmpfiles learned two new line types "q" and "Q" that
740 operate like "v", but also set up a basic btrfs quota
741 hierarchy when used on a btrfs file system with quota
744 * tmpfiles' "v", "q" and "Q" will now create a plain directory
745 instead of a subvolume (even on a btrfs file system) if the
746 root directory is a plain directory, and not a
747 subvolume. This should simplify things with certain chroot()
748 environments which are not aware of the concept of btrfs
751 * systemd-detect-virt gained a new --chroot switch to detect
752 whether execution takes place in a chroot() environment.
754 * CPUAffinity= now takes CPU index ranges in addition to
757 * The various memory-related resource limit settings (such as
758 LimitAS=) now understand the usual K, M, G, ... suffixes to
759 the base of 1024 (IEC). Similar, the time-related resource
760 limit settings understand the usual min, h, day, ...
763 * There's a new system.conf setting DefaultTasksMax= to
764 control the default TasksMax= setting for services and
765 scopes running on the system. (TasksMax= is the primary
766 setting that exposes the "pids" cgroup controller on systemd
767 and was introduced in the previous systemd release.) The
768 setting now defaults to 512, which means services that are
769 not explicitly configured otherwise will only be able to
770 create 512 processes or threads at maximum, from this
771 version on. Note that this means that thread- or
772 process-heavy services might need to be reconfigured to set
773 TasksMax= to a higher value. It is sufficient to set
774 TasksMax= in these specific unit files to a higher value, or
775 even "infinity". Similar, there's now a logind.conf setting
776 UserTasksMax= that defaults to 4096 and limits the total
777 number of processes or tasks each user may own
778 concurrently. nspawn containers also have the TasksMax=
779 value set by default now, to 8192. Note that all of this
780 only has an effect if the "pids" cgroup controller is
781 enabled in the kernel. The general benefit of these changes
782 should be a more robust and safer system, that provides a
783 certain amount of per-service fork() bomb protection.
785 * systemd-nspawn gained the new --network-veth-extra= switch
786 to define additional and arbitrarily-named virtual Ethernet
787 links between the host and the container.
789 * A new service execution setting PassEnvironment= has been
790 added that allows importing select environment variables
791 from PID1's environment block into the environment block of
794 * Timer units gained support for a new RemainAfterElapse=
795 setting which takes a boolean argument. It defaults to on,
796 exposing behaviour unchanged to previous releases. If set to
797 off, timer units are unloaded after they elapsed if they
798 cannot elapse again. This is particularly useful for
799 transient timer units, which shall not stay around longer
800 than until they first elapse.
802 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
803 default now (the kernel default is 16). This is beneficial
804 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
805 allows substantially larger numbers of queued
806 datagrams. This should increase the capability of systemd to
807 parallelize boot-up, as logging and sd_notify() are unlikely
808 to stall execution anymore. If you need to change the value
809 from the new defaults, use the usual sysctl.d/ snippets.
811 * The compression framing format used by the journal or
812 coredump processing has changed to be in line with what the
813 official LZ4 tools generate. LZ4 compression support in
814 systemd was considered unsupported previously, as the format
815 was not compatible with the normal tools. With this release
816 this has changed now, and it is hence safe for downstream
817 distributions to turn it on. While not compressing as well
818 as the XZ, LZ4 is substantially faster, which makes
819 it a good default choice for the compression logic in the
820 journal and in coredump handling.
822 * Any reference to /etc/mtab has been dropped from
823 systemd. The file has been obsolete since a while, but
824 systemd refused to work on systems where it was incorrectly
825 set up (it should be a symlink or non-existent). Please make
826 sure to update to util-linux 2.27.1 or newer in conjunction
827 with this systemd release, which also drops any reference to
828 /etc/mtab. If you maintain a distribution make sure that no
829 software you package still references it, as this is a
830 likely source of bugs. There's also a glibc bug pending,
831 asking for removal of any reference to this obsolete file:
833 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
835 Note that only util-linux versions built with
836 --enable-libmount-force-mountinfo are supported.
838 * Support for the ".snapshot" unit type has been removed. This
839 feature turned out to be little useful and little used, and
840 has now been removed from the core and from systemctl.
842 * The dependency types RequiresOverridable= and
843 RequisiteOverridable= have been removed from systemd. They
844 have been used only very sparingly to our knowledge and
845 other options that provide a similar effect (such as
846 systemctl --mode=ignore-dependencies) are much more useful
847 and commonly used. Moreover, they were only half-way
848 implemented as the option to control behaviour regarding
849 these dependencies was never added to systemctl. By removing
850 these dependency types the execution engine becomes a bit
851 simpler. Unit files that use these dependencies should be
852 changed to use the non-Overridable dependency types
853 instead. In fact, when parsing unit files with these
854 options, that's what systemd will automatically convert them
855 too, but it will also warn, asking users to fix the unit
856 files accordingly. Removal of these dependency types should
857 only affect a negligible number of unit files in the wild.
859 * Behaviour of networkd's IPForward= option changed
860 (again). It will no longer maintain a per-interface setting,
861 but propagate one way from interfaces where this is enabled
862 to the global kernel setting. The global setting will be
863 enabled when requested by a network that is set up, but
864 never be disabled again. This change was made to make sure
865 IPv4 and IPv6 behaviour regarding packet forwarding is
866 similar (as the Linux IPv6 stack does not support
867 per-interface control of this setting) and to minimize
870 * In unit files the behaviour of %u, %U, %h, %s has
871 changed. These specifiers will now unconditionally resolve
872 to the various user database fields of the user that the
873 systemd instance is running as, instead of the user
874 configured in the specific unit via User=. Note that this
875 effectively doesn't change much, as resolving of these
876 specifiers was already turned off in the --system instance
877 of systemd, as we cannot do NSS lookups from PID 1. In the
878 --user instance of systemd these specifiers where correctly
879 resolved, but hardly made any sense, since the user instance
880 lacks privileges to do user switches anyway, and User= is
881 hence useless. Morever, even in the --user instance of
882 systemd behaviour was awkward as it would only take settings
883 from User= assignment placed before the specifier into
884 account. In order to unify and simplify the logic around
885 this the specifiers will now always resolve to the
886 credentials of the user invoking the manager (which in case
887 of PID 1 is the root user).
889 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
890 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
891 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
892 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
893 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
894 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
895 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
896 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
897 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
898 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
899 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
900 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
901 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
902 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
909 * systemd now depends on util-linux v2.27. More specifically,
910 the newly added mount monitor feature in libmount now
911 replaces systemd's former own implementation.
913 * libmount mandates /etc/mtab not to be regular file, and
914 systemd now enforces this condition at early boot.
915 /etc/mtab has been deprecated and warned about for a very
916 long time, so systems running systemd should already have
917 stopped having this file around as anything else than a
918 symlink to /proc/self/mounts.
920 * Support for the "pids" cgroup controller has been added. It
921 allows accounting the number of tasks in a cgroup and
922 enforcing limits on it. This adds two new setting
923 TasksAccounting= and TasksMax= to each unit, as well as a
924 global option DefaultTasksAccounting=.
926 * Support for the "net_cls" cgroup controller has been added.
927 It allows assigning a net class ID to each task in the
928 cgroup, which can then be used in firewall rules and traffic
929 shaping configurations. Note that the kernel netfilter net
930 class code does not currently work reliably for ingress
931 packets on unestablished sockets.
933 This adds a new config directive called NetClass= to CGroup
934 enabled units. Allowed values are positive numbers for fixed
935 assignments and "auto" for picking a free value
938 * 'systemctl is-system-running' now returns 'offline' if the
939 system is not booted with systemd. This command can now be
940 used as a substitute for 'systemd-notify --booted'.
942 * Watchdog timeouts have been increased to 3 minutes for all
943 in-tree service files. Apparently, disk IO issues are more
944 frequent than we hoped, and user reported >1 minute waiting
947 * 'machine-id-commit' functionality has been merged into
948 'machine-id-setup --commit'. The separate binary has been
951 * The WorkingDirectory= directive in unit files may now be set
952 to the special value '~'. In this case, the working
953 directory is set to the home directory of the user
956 * "machinectl shell" will now open the shell in the home
957 directory of the selected user by default.
959 * The CrashChVT= configuration file setting is renamed to
960 CrashChangeVT=, following our usual logic of not
961 abbreviating unnecessarily. The old directive is still
962 supported for compat reasons. Also, this directive now takes
963 an integer value between 1 and 63, or a boolean value. The
964 formerly supported '-1' value for disabling stays around for
967 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
968 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
969 RootDirectory= properties can now be set for transient
972 * The systemd-analyze tool gained a new "set-log-target" verb
973 to change the logging target the system manager logs to
974 dynamically during runtime. This is similar to how
975 "systemd-analyze set-log-level" already changes the log
978 * In nspawn /sys is now mounted as tmpfs, with only a selected
979 set of subdirectories mounted in from the real sysfs. This
980 enhances security slightly, and is useful for ensuring user
981 namespaces work correctly.
983 * Support for USB FunctionFS activation has been added. This
984 allows implementation of USB gadget services that are
985 activated as soon as they are requested, so that they don't
986 have to run continuously, similar to classic socket
989 * The "systemctl exit" command now optionally takes an
990 additional parameter that sets the exit code to return from
991 the systemd manager when exiting. This is only relevant when
992 running the systemd user instance, or when running the
993 system instance in a container.
995 * sd-bus gained the new API calls sd_bus_path_encode_many()
996 and sd_bus_path_decode_many() that allow easy encoding and
997 decoding of multiple identifier strings inside a D-Bus
998 object path. Another new call sd_bus_default_flush_close()
999 has been added to flush and close per-thread default
1002 * systemd-cgtop gained support for a -M/--machine= switch to
1003 show the control groups within a certain container only.
1005 * "systemctl kill" gained support for an optional --fail
1006 switch. If specified the requested operation will fail of no
1007 processes have been killed, because the unit had no
1008 processes attached, or similar.
1010 * A new systemd.crash_reboot=1 kernel command line option has
1011 been added that triggers a reboot after crashing. This can
1012 also be set through CrashReboot= in systemd.conf.
1014 * The RuntimeDirectory= setting now understands unit
1015 specifiers like %i or %f.
1017 * A new (still internal) libary API sd-ipv4acd has been added,
1018 that implements address conflict detection for IPv4. It's
1019 based on code from sd-ipv4ll, and will be useful for
1020 detecting DHCP address conflicts.
1022 * File descriptors passed during socket activation may now be
1023 named. A new API sd_listen_fds_with_names() is added to
1024 access the names. The default names may be overridden,
1025 either in the .socket file using the FileDescriptorName=
1026 parameter, or by passing FDNAME= when storing the file
1027 descriptors using sd_notify().
1029 * systemd-networkd gained support for:
1031 - Setting the IPv6 Router Advertisement settings via
1032 IPv6AcceptRouterAdvertisements= in .network files.
1034 - Configuring the HelloTimeSec=, MaxAgeSec= and
1035 ForwardDelaySec= bridge parameters in .netdev files.
1037 - Configuring PreferredSource= for static routes in
1040 * The "ask-password" framework used to query for LUKS harddisk
1041 passwords or SSL passwords during boot gained support for
1042 caching passwords in the kernel keyring, if it is
1043 available. This makes sure that the user only has to type in
1044 a passphrase once if there are multiple objects to unlock
1045 with the same one. Previously, such password caching was
1046 available only when Plymouth was used; this moves the
1047 caching logic into the systemd codebase itself. The
1048 "systemd-ask-password" utility gained a new --keyname=
1049 switch to control which kernel keyring key to use for
1050 caching a password in. This functionality is also useful for
1051 enabling display managers such as gdm to automatically
1052 unlock the user's GNOME keyring if its passphrase, the
1053 user's password and the harddisk password are the same, if
1054 gdm-autologin is used.
1056 * When downloading tar or raw images using "machinectl
1057 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
1058 file is now also downloaded, if it is available and stored
1059 next to the image file.
1061 * Units of type ".socket" gained a new boolean setting
1062 Writable= which is only useful in conjunction with
1063 ListenSpecial=. If true, enables opening the specified
1064 special file in O_RDWR mode rather than O_RDONLY mode.
1066 * systemd-rfkill has been reworked to become a singleton
1067 service that is activated through /dev/rfkill on each rfkill
1068 state change and saves the settings to disk. This way,
1069 systemd-rfkill is now compatible with devices that exist
1070 only intermittendly, and even restores state if the previous
1071 system shutdown was abrupt rather than clean.
1073 * The journal daemon gained support for vacuuming old journal
1074 files controlled by the number of files that shall remain,
1075 in addition to the already existing control by size and by
1076 date. This is useful as journal interleaving performance
1077 degrades with too many separate journal files, and allows
1078 putting an effective limit on them. The new setting defaults
1079 to 100, but this may be changed by setting SystemMaxFiles=
1080 and RuntimeMaxFiles= in journald.conf. Also, the
1081 "journalctl" tool gained the new --vacuum-files= switch to
1082 manually vacuum journal files to leave only the specified
1083 number of files in place.
1085 * udev will now create /dev/disk/by-path links for ATA devices
1086 on kernels where that is supported.
1088 * Galician, Serbian, Turkish and Korean translations were added.
1090 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
1091 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
1092 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
1093 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
1094 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
1095 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
1096 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
1097 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
1098 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
1099 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
1100 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
1101 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
1102 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
1103 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
1104 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
1105 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
1106 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
1107 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
1109 — Berlin, 2015-10-07
1113 * The DHCP implementation of systemd-networkd gained a set of
1116 - The DHCP server now supports emitting DNS and NTP
1117 information. It may be enabled and configured via
1118 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
1119 and NTP information is enabled, but no servers are
1120 configured, the corresponding uplink information (if there
1121 is any) is propagated.
1123 - Server and client now support transmission and reception
1124 of timezone information. It can be configured via the
1125 newly introduced network options UseTimezone=,
1126 EmitTimezone=, and Timezone=. Transmission of timezone
1127 information is enabled between host and containers by
1128 default now: the container will change its local timezone
1129 to what the host has set.
1131 - Lease timeouts can now be configured via
1132 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
1134 - The DHCP server improved on the stability of
1135 leases. Clients are more likely to get the same lease
1136 information back, even if the server loses state.
1138 - The DHCP server supports two new configuration options to
1139 control the lease address pool metrics, PoolOffset= and
1142 * The encapsulation limit of tunnels in systemd-networkd may
1143 now be configured via 'EncapsulationLimit='. It allows
1144 modifying the maximum additional levels of encapsulation
1145 that are permitted to be prepended to a packet.
1147 * systemd now supports the concept of user buses replacing
1148 session buses, if used with dbus-1.10 (and enabled via dbus
1149 --enable-user-session). It previously only supported this on
1150 kdbus-enabled systems, and this release expands this to
1151 'dbus-daemon' systems.
1153 * systemd-networkd now supports predictable interface names
1156 * systemd now optionally supports the new Linux kernel
1157 "unified" control group hierarchy. If enabled via the kernel
1158 command-line option 'systemd.unified_cgroup_hierarchy=1',
1159 systemd will try to mount the unified cgroup hierarchy
1160 directly on /sys/fs/cgroup. If not enabled, or not
1161 available, systemd will fall back to the legacy cgroup
1162 hierarchy setup, as before. Host system and containers can
1163 mix and match legacy and unified hierarchies as they
1164 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
1165 environment variable to individually select the hierarchy to
1166 use for executed containers. By default, nspawn will use the
1167 unified hierarchy for the containers if the host uses the
1168 unified hierarchy, and the legacy hierarchy otherwise.
1169 Please note that at this point the unified hierarchy is an
1170 experimental kernel feature and is likely to change in one
1171 of the next kernel releases. Therefore, it should not be
1172 enabled by default in downstream distributions yet. The
1173 minimum required kernel version for the unified hierarchy to
1174 work is 4.2. Note that when the unified hierarchy is used
1175 for the first time delegated access to controllers is
1176 safe. Because of this systemd-nspawn containers will get
1177 access to controllers now, as will systemd user
1178 sessions. This means containers and user sessions may now
1179 manage their own resources, partitioning up what the system
1182 * A new special scope unit "init.scope" has been introduced
1183 that encapsulates PID 1 of the system. It may be used to
1184 determine resource usage and enforce resource limits on PID
1185 1 itself. PID 1 hence moved out of the root of the control
1188 * The cgtop tool gained support for filtering out kernel
1189 threads when counting tasks in a control group. Also, the
1190 count of processes is now recursively summed up by
1191 default. Two options -k and --recursive= have been added to
1192 revert to old behaviour. The tool has also been updated to
1193 work correctly in containers now.
1195 * systemd-nspawn's --bind= and --bind-ro= options have been
1196 extended to allow creation of non-recursive bind mounts.
1198 * libsystemd gained two new calls sd_pid_get_cgroup() and
1199 sd_peer_get_cgroup() which return the control group path of
1200 a process or peer of a connected AF_UNIX socket. This
1201 function call is particularly useful when implementing
1202 delegated subtrees support in the control group hierarchy.
1204 * The "sd-event" event loop API of libsystemd now supports
1205 correct dequeuing of real-time signals, without losing
1208 * When systemd requests a PolicyKit decision when managing
1209 units it will now add additional fields to the request,
1210 including unit name and desired operation. This enables more
1211 powerful PolicyKit policies, that make decisions depending
1212 on these parameters.
1214 * nspawn learnt support for .nspawn settings files, that may
1215 accompany the image files or directories of containers, and
1216 may contain additional settings for the container. This is
1217 an alternative to configuring container parameters via the
1218 nspawn command line.
1220 Contributions from: Cristian Rodríguez, Daniel Mack, David
1221 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
1222 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
1223 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
1224 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
1225 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
1226 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
1227 Andersen, Tom Gundersen, Torstein Husebø
1229 — Berlin, 2015-09-08
1233 * machinectl gained a new verb 'shell' which opens a fresh
1234 shell on the target container or the host. It is similar to
1235 the existing 'login' command of machinectl, but spawns the
1236 shell directly without prompting for username or
1237 password. The pseudo machine '.host' now refers to the local
1238 host and is used by default. Hence, 'machinectl shell' can
1239 be used as replacement for 'su -' which spawns a session as
1240 a fresh systemd unit in a way that is fully isolated from
1241 the originating session.
1243 * systemd-networkd learned to cope with private-zone DHCP
1244 options and allows other programs to query the values.
1246 * SELinux access control when enabling/disabling units is no
1247 longer enforced with this release. The previous
1248 implementation was incorrect, and a new corrected
1249 implementation is not yet available. As unit file operations
1250 are still protected via PolicyKit and D-Bus policy this is
1251 not a security problem. Yet, distributions which care about
1252 optimal SELinux support should probably not stabilize on
1255 * sd-bus gained support for matches of type "arg0has=", that
1256 test for membership of strings in string arrays sent in bus
1259 * systemd-resolved now dumps the contents of its DNS and LLMNR
1260 caches to the logs on reception of the SIGUSR1 signal. This
1261 is useful to debug DNS behaviour.
1263 * The coredumpctl tool gained a new --directory= option to
1264 operate on journal files in a specific directory.
1266 * "systemctl reboot" and related commands gained a new
1267 "--message=" option which may be used to set a free-text
1268 wall message when shutting down or rebooting the
1269 system. This message is also logged, which is useful for
1270 figuring out the reason for a reboot or shutdown a
1273 * The "systemd-resolve-host" tool's -i switch now takes
1274 network interface numbers as alternative to interface names.
1276 * A new unit file setting for services has been introduced:
1277 UtmpMode= allows configuration of how precisely systemd
1278 handles utmp and wtmp entries for the service if this is
1279 enabled. This allows writing services that appear similar to
1280 user sessions in the output of the "w", "who", "last" and
1283 * systemd-resolved will now locally synthesize DNS resource
1284 records for the "localhost" and "gateway" domains as well as
1285 the local hostname. This should ensure that clients querying
1286 RRs via resolved will get similar results as those going via
1287 NSS, if nss-myhostname is enabled.
1289 Contributions from: Alastair Hughes, Alex Crawford, Daniel
1290 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
1291 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
1292 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
1293 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
1294 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
1295 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
1296 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
1297 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
1298 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
1299 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
1300 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
1302 — Berlin, 2015-08-27
1306 * The systemd-efi-boot-generator functionality was merged into
1307 systemd-gpt-auto-generator.
1309 * systemd-networkd now supports Group Policy for vxlan
1310 devices. It can be enabled via the new boolean configuration
1311 option called 'GroupPolicyExtension='.
1313 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
1314 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
1315 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
1317 — Berlin, 2015-07-31
1321 * The python-systemd code has been removed from the systemd repository.
1322 A new repository has been created which accommodates the code from
1323 now on, and we kindly ask distributions to create a separate package
1324 for this: https://github.com/systemd/python-systemd
1326 * The systemd daemon will now reload its main configuration
1327 (/etc/systemd/system.conf) on daemon-reload.
1329 * sd-dhcp now exposes vendor specific extensions via
1330 sd_dhcp_lease_get_vendor_specific().
1332 * systemd-networkd gained a number of new configuration options.
1334 - A new boolean configuration option for TAP devices called
1335 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
1336 device, thus allowing to send and receive GSO packets.
1338 - A new tunnel configuration option called 'CopyDSCP='.
1339 If enabled, the DSCP field of ip6 tunnels is copied into the
1340 decapsulated packet.
1342 - A set of boolean bridge configuration options were added.
1343 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
1344 and 'UnicastFlood=' are now parsed by networkd and applied to the
1345 respective bridge link device via the respective IFLA_BRPORT_*
1348 - A new string configuration option to override the hostname sent
1349 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
1350 is true, networkd will use the configured hostname instead of the
1351 system hostname when sending DHCP requests.
1353 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
1354 networkd will configure the IPv6 flow-label of the tunnel device
1355 according to RFC2460.
1357 - The 'macvtap' virtual network devices are now supported, similar to
1358 the already supported 'macvlan' devices.
1360 * systemd-resolved now implements RFC5452 to improve resilience against
1361 cache poisoning. Additionally, source port randomization is enabled
1362 by default to further protect against DNS spoofing attacks.
1364 * nss-mymachines now supports translating UIDs and GIDs of running
1365 containers with user-namespaces enabled. If a container 'foo'
1366 translates a host uid 'UID' to the container uid 'TUID', then
1367 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
1368 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
1369 mapped as 'vg-foo-TGID'.
1371 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
1372 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
1373 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
1374 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
1375 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
1376 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
1377 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
1378 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
1379 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
1380 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
1382 — Berlin, 2015-07-29
1386 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
1387 There are no known issues with current sysfs, and udev does not need
1388 or should be used to work around such bugs.
1390 * udev does no longer enable USB HID power management. Several reports
1391 indicate, that some devices cannot handle that setting.
1393 * The udev accelerometer helper was removed. The functionality
1394 is now fully included in iio-sensor-proxy. But this means,
1395 older iio-sensor-proxy versions will no longer provide
1396 accelerometer/orientation data with this systemd version.
1397 Please upgrade iio-sensor-proxy to version 1.0.
1399 * networkd gained a new configuration option IPv6PrivacyExtensions=
1400 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
1401 for Stateless Address") on selected networks.
1403 * For the sake of fewer build-time dependencies and less code in the
1404 main repository, the python bindings are about to be removed in the
1405 next release. A new repository has been created which accommodates
1406 the code from now on, and we kindly ask distributions to create a
1407 separate package for this. The removal will take place in v223.
1409 https://github.com/systemd/python-systemd
1411 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
1412 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
1413 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
1414 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
1415 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
1416 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
1417 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
1418 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
1419 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
1420 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
1422 — Berlin, 2015-07-07
1426 * The sd-bus.h and sd-event.h APIs have now been declared
1427 stable and have been added to the official interface of
1428 libsystemd.so. sd-bus implements an alternative D-Bus client
1429 library, that is relatively easy to use, very efficient and
1430 supports both classic D-Bus as well as kdbus as transport
1431 backend. sd-event is a generic event loop abstraction that
1432 is built around Linux epoll, but adds features such as event
1433 prioritization or efficient timer handling. Both APIs are good
1434 choices for C programs looking for a bus and/or event loop
1435 implementation that is minimal and does not have to be
1436 portable to other kernels.
1438 * kdbus support is no longer compile-time optional. It is now
1439 always built-in. However, it can still be disabled at
1440 runtime using the kdbus=0 kernel command line setting, and
1441 that setting may be changed to default to off, by specifying
1442 --disable-kdbus at build-time. Note though that the kernel
1443 command line setting has no effect if the kdbus.ko kernel
1444 module is not installed, in which case kdbus is (obviously)
1445 also disabled. We encourage all downstream distributions to
1446 begin testing kdbus by adding it to the kernel images in the
1447 development distributions, and leaving kdbus support in
1450 * The minimal required util-linux version has been bumped to
1453 * Support for chkconfig (--enable-chkconfig) was removed in
1454 favor of calling an abstraction tool
1455 /lib/systemd/systemd-sysv-install. This needs to be
1456 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
1457 in README for details.
1459 * If there's a systemd unit and a SysV init script for the
1460 same service name, and the user executes "systemctl enable"
1461 for it (or a related call), then this will now enable both
1462 (or execute the related operation on both), not just the
1465 * The libudev API documentation has been converted from gtkdoc
1468 * gudev has been removed from the systemd tree, it is now an
1471 * The systemd-cgtop tool learnt a new --raw switch to generate
1472 "raw" (machine parsable) output.
1474 * networkd's IPForwarding= .network file setting learnt the
1475 new setting "kernel", which ensures that networkd does not
1476 change the IP forwarding sysctl from the default kernel
1479 * The systemd-logind bus API now exposes a new boolean
1480 property "Docked" that reports whether logind considers the
1481 system "docked", i.e. connected to a docking station or not.
1483 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
1484 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
1485 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
1486 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
1487 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
1488 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
1489 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
1490 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
1491 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
1492 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
1493 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
1494 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
1495 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
1496 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
1497 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
1498 Fink, Zbigniew Jędrzejewski-Szmek
1500 — Berlin, 2015-06-19
1504 * The gudev library has been extracted into a separate repository
1505 available at: https://git.gnome.org/browse/libgudev/
1506 It is now managed as part of the Gnome project. Distributions
1507 are recommended to pass --disable-gudev to systemd and use
1508 gudev from the Gnome project instead. gudev is still included
1509 in systemd, for now. It will be removed soon, though. Please
1510 also see the announcement-thread on systemd-devel:
1511 http://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
1513 * systemd now exposes a CPUUsageNSec= property for each
1514 service unit on the bus, that contains the overall consumed
1515 CPU time of a service (the sum of what each process of the
1516 service consumed). This value is only available if
1517 CPUAccounting= is turned on for a service, and is then shown
1518 in the "systemctl status" output.
1520 * Support for configuring alternative mappings of the old SysV
1521 runlevels to systemd targets has been removed. They are now
1522 hardcoded in a way that runlevels 2, 3, 4 all map to
1523 multi-user.target and 5 to graphical.target (which
1524 previously was already the default behaviour).
1526 * The auto-mounter logic gained support for mount point
1527 expiry, using a new TimeoutIdleSec= setting in .automount
1528 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
1530 * The EFI System Partition (ESP) as mounted to /boot by
1531 systemd-efi-boot-generator will now be unmounted
1532 automatically after 2 minutes of not being used. This should
1533 minimize the risk of ESP corruptions.
1535 * New /etc/fstab options x-systemd.requires= and
1536 x-systemd.requires-mounts-for= are now supported to express
1537 additional dependencies for mounts. This is useful for
1538 journalling file systems that support external journal
1539 devices or overlay file systems that require underlying file
1540 systems to be mounted.
1542 * systemd does not support direct live-upgrades (via systemctl
1543 daemon-reexec) from versions older than v44 anymore. As no
1544 distribution we are aware of shipped such old versions in a
1545 stable release this should not be problematic.
1547 * When systemd forks off a new per-connection service instance
1548 it will now set the $REMOTE_ADDR environment variable to the
1549 remote IP address, and $REMOTE_PORT environment variable to
1550 the remote IP port. This behaviour is similar to the
1551 corresponding environment variables defined by CGI.
1553 * systemd-networkd gained support for uplink failure
1554 detection. The BindCarrier= option allows binding interface
1555 configuration dynamically to the link sense of other
1556 interfaces. This is useful to achieve behaviour like in
1559 * systemd-networkd gained support for configuring the DHCP
1560 client identifier to use when requesting leases.
1562 * systemd-networkd now has a per-network UseNTP= option to
1563 configure whether NTP server information acquired via DHCP
1564 is passed on to services like systemd-timesyncd.
1566 * systemd-networkd gained support for vti6 tunnels.
1568 * Note that systemd-networkd manages the sysctl variable
1569 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
1570 it is configured for since v219. The variable controls IP
1571 forwarding, and is a per-interface alternative to the global
1572 /proc/sys/net/ipv[46]/ip_forward. This setting is
1573 configurable in the IPForward= option, which defaults to
1574 "no". This means if networkd is used for an interface it is
1575 no longer sufficient to set the global sysctl option to turn
1576 on IP forwarding! Instead, the .network file option
1577 IPForward= needs to be turned on! Note that the
1578 implementation of this behaviour was broken in v219 and has
1581 * Many bonding and vxlan options are now configurable in
1584 * systemd-nspawn gained a new --property= setting to set unit
1585 properties for the container scope. This is useful for
1586 setting resource parameters (e.g "CPUShares=500") on
1587 containers started from the command line.
1589 * systemd-nspawn gained a new --private-users= switch to make
1590 use of user namespacing available on recent Linux kernels.
1592 * systemd-nspawn may now be called as part of a shell pipeline
1593 in which case the pipes used for stdin and stdout are passed
1594 directly to the process invoked in the container, without
1595 indirection via a pseudo tty.
1597 * systemd-nspawn gained a new switch to control the UNIX
1598 signal to use when killing the init process of the container
1601 * systemd-nspawn gained a new --overlay= switch for mounting
1602 overlay file systems into the container using the new kernel
1605 * When a container image is imported via systemd-importd and
1606 the host file system is not btrfs, a loopback block device
1607 file is created in /var/lib/machines.raw with a btrfs file
1608 system inside. It is then mounted to /var/lib/machines to
1609 enable btrfs features for container management. The loopback
1610 file and btrfs file system is grown as needed when container
1611 images are imported via systemd-importd.
1613 * systemd-machined/systemd-importd gained support for btrfs
1614 quota, to enforce container disk space limits on disk. This
1615 is exposed in "machinectl set-limit".
1617 * systemd-importd now can import containers from local .tar,
1618 .raw and .qcow2 images, and export them to .tar and .raw. It
1619 can also import dkr v2 images now from the network (on top
1622 * systemd-importd gained support for verifying downloaded
1623 images with gpg2 (previously only gpg1 was supported).
1625 * systemd-machined, systemd-logind, systemd: most bus calls
1626 are now accessible to unprivileged processes via
1627 PolicyKit. Also, systemd-logind will now allow users to kill
1628 their own sessions without further privileges or
1631 * systemd-shutdownd has been removed. This service was
1632 previously responsible for implementing scheduled shutdowns
1633 as exposed in /usr/bin/shutdown's time parameter. This
1634 functionality has now been moved into systemd-logind and is
1635 accessible via a bus interface.
1637 * "systemctl reboot" gained a new switch --firmware-setup that
1638 can be used to reboot into the EFI firmware setup, if that
1639 is available. systemd-logind now exposes an API on the bus
1640 to trigger such reboots, in case graphical desktop UIs want
1641 to cover this functionality.
1643 * "systemctl enable", "systemctl disable" and "systemctl mask"
1644 now support a new "--now" switch. If specified the units
1645 that are enabled will also be started, and the ones
1646 disabled/masked also stopped.
1648 * The Gummiboot EFI boot loader tool has been merged into
1649 systemd, and renamed to "systemd-boot". The bootctl tool has been
1650 updated to support systemd-boot.
1652 * An EFI kernel stub has been added that may be used to create
1653 kernel EFI binaries that contain not only the actual kernel,
1654 but also an initrd, boot splash, command line and OS release
1655 information. This combined binary can then be signed as a
1656 single image, so that the firmware can verify it all in one
1657 step. systemd-boot has special support for EFI binaries created
1658 like this and can extract OS release information from them
1659 and show them in the boot menu. This functionality is useful
1660 to implement cryptographically verified boot schemes.
1662 * Optional support has been added to systemd-fsck to pass
1663 fsck's progress report to an AF_UNIX socket in the file
1666 * udev will no longer create device symlinks for all block
1667 devices by default. A blacklist for excluding special block
1668 devices from this logic has been turned into a whitelist
1669 that requires picking block devices explicitly that require
1672 * A new (currently still internal) API sd-device.h has been
1673 added to libsystemd. This modernized API is supposed to
1674 replace libudev eventually. In fact, already much of libudev
1675 is now just a wrapper around sd-device.h.
1677 * A new hwdb database for storing metadata about pointing
1678 stick devices has been added.
1680 * systemd-tmpfiles gained support for setting file attributes
1681 similar to the "chattr" tool with new 'h' and 'H' lines.
1683 * systemd-journald will no longer unconditionally set the
1684 btrfs NOCOW flag on new journal files. This is instead done
1685 with tmpfiles snippet using the new 'h' line type. This
1686 allows easy disabling of this logic, by masking the
1687 journal-nocow.conf tmpfiles file.
1689 * systemd-journald will now translate audit message types to
1690 human readable identifiers when writing them to the
1691 journal. This should improve readability of audit messages.
1693 * The LUKS logic gained support for the offset= and skip=
1694 options in /etc/crypttab, as previously implemented by
1697 * /usr/lib/os-release gained a new optional field VARIANT= for
1698 distributions that support multiple variants (such as a
1699 desktop edition, a server edition, ...)
1701 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
1702 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
1703 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
1704 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
1705 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
1706 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
1707 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
1708 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
1709 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
1710 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
1711 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
1712 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
1713 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
1714 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
1715 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
1716 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
1717 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
1718 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
1719 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
1720 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
1721 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
1722 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
1723 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
1724 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
1725 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
1726 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
1727 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
1729 — Berlin, 2015-05-22
1733 * Introduce a new API "sd-hwdb.h" for querying the hardware
1734 metadata database. With this minimal interface one can query
1735 and enumerate the udev hwdb, decoupled from the old libudev
1736 library. libudev's interface for this is now only a wrapper
1737 around sd-hwdb. A new tool systemd-hwdb has been added to
1738 interface with and update the database.
1740 * When any of systemd's tools copies files (for example due to
1741 tmpfiles' C lines) a btrfs reflink will attempted first,
1742 before bytewise copying is done.
1744 * systemd-nspawn gained a new --ephemeral switch. When
1745 specified a btrfs snapshot is taken of the container's root
1746 directory, and immediately removed when the container
1747 terminates again. Thus, a container can be started whose
1748 changes never alter the container's root directory, and are
1749 lost on container termination. This switch can also be used
1750 for starting a container off the root file system of the
1751 host without affecting the host OS. This switch is only
1752 available on btrfs file systems.
1754 * systemd-nspawn gained a new --template= switch. It takes the
1755 path to a container tree to use as template for the tree
1756 specified via --directory=, should that directory be
1757 missing. This allows instantiating containers dynamically,
1758 on first run. This switch is only available on btrfs file
1761 * When a .mount unit refers to a mount point on which multiple
1762 mounts are stacked, and the .mount unit is stopped all of
1763 the stacked mount points will now be unmounted until no
1764 mount point remains.
1766 * systemd now has an explicit notion of supported and
1767 unsupported unit types. Jobs enqueued for unsupported unit
1768 types will now fail with an "unsupported" error code. More
1769 specifically .swap, .automount and .device units are not
1770 supported in containers, .busname units are not supported on
1771 non-kdbus systems. .swap and .automount are also not
1772 supported if their respective kernel compile time options
1775 * machinectl gained support for two new "copy-from" and
1776 "copy-to" commands for copying files from a running
1777 container to the host or vice versa.
1779 * machinectl gained support for a new "bind" command to bind
1780 mount host directories into local containers. This is
1781 currently only supported for nspawn containers.
1783 * networkd gained support for configuring bridge forwarding
1784 database entries (fdb) from .network files.
1786 * A new tiny daemon "systemd-importd" has been added that can
1787 download container images in tar, raw, qcow2 or dkr formats,
1788 and make them available locally in /var/lib/machines, so
1789 that they can run as nspawn containers. The daemon can GPG
1790 verify the downloads (not supported for dkr, since it has no
1791 provisions for verifying downloads). It will transparently
1792 decompress bz2, xz, gzip compressed downloads if necessary,
1793 and restore sparse files on disk. The daemon uses privilege
1794 separation to ensure the actual download logic runs with
1795 fewer privileges than the daemon itself. machinectl has
1796 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
1797 make the functionality of importd available to the
1798 user. With this in place the Fedora and Ubuntu "Cloud"
1799 images can be downloaded and booted as containers unmodified
1800 (the Fedora images lack the appropriate GPG signature files
1801 currently, so they cannot be verified, but this will change
1802 soon, hopefully). Note that downloading images is currently
1803 only fully supported on btrfs.
1805 * machinectl is now able to list container images found in
1806 /var/lib/machines, along with some metadata about sizes of
1807 disk and similar. If the directory is located on btrfs and
1808 quota is enabled, this includes quota display. A new command
1809 "image-status" has been added that shows additional
1810 information about images.
1812 * machinectl is now able to clone container images
1813 efficiently, if the underlying file system (btrfs) supports
1814 it, with the new "machinectl clone" command. It also
1815 gained commands for renaming and removing images, as well as
1816 marking them read-only or read-write (supported also on
1817 legacy file systems).
1819 * networkd gained support for collecting LLDP network
1820 announcements, from hardware that supports this. This is
1821 shown in networkctl output.
1823 * systemd-run gained support for a new -t (--pty) switch for
1824 invoking a binary on a pty whose input and output is
1825 connected to the invoking terminal. This allows executing
1826 processes as system services while interactively
1827 communicating with them via the terminal. Most interestingly
1828 this is supported across container boundaries. Invoking
1829 "systemd-run -t /bin/bash" is an alternative to running a
1830 full login session, the difference being that the former
1831 will not register a session, nor go through the PAM session
1834 * tmpfiles gained support for a new "v" line type for creating
1835 btrfs subvolumes. If the underlying file system is a legacy
1836 file system, this automatically degrades to creating a
1837 normal directory. Among others /var/lib/machines is now
1838 created like this at boot, should it be missing.
1840 * The directory /var/lib/containers/ has been deprecated and
1841 been replaced by /var/lib/machines. The term "machines" has
1842 been used in the systemd context as generic term for both
1843 VMs and containers, and hence appears more appropriate for
1844 this, as the directory can also contain raw images bootable
1847 * systemd-nspawn when invoked with -M but without --directory=
1848 or --image= is now capable of searching for the container
1849 root directory, subvolume or disk image automatically, in
1850 /var/lib/machines. systemd-nspawn@.service has been updated
1851 to make use of this, thus allowing it to be used for raw
1854 * A new machines.target unit has been introduced that is
1855 supposed to group all containers/VMs invoked as services on
1856 the system. systemd-nspawn@.service has been updated to
1857 integrate with that.
1859 * machinectl gained a new "start" command, for invoking a
1860 container as a service. "machinectl start foo" is mostly
1861 equivalent to "systemctl start systemd-nspawn@foo.service",
1862 but handles escaping in a nicer way.
1864 * systemd-nspawn will now mount most of the cgroupfs tree
1865 read-only into each container, with the exception of the
1866 container's own subtree in the name=systemd hierarchy.
1868 * journald now sets the special FS_NOCOW file flag for its
1869 journal files. This should improve performance on btrfs, by
1870 avoiding heavy fragmentation when journald's write-pattern
1871 is used on COW file systems. It degrades btrfs' data
1872 integrity guarantees for the files to the same levels as for
1873 ext3/ext4 however. This should be OK though as journald does
1874 its own data integrity checks and all its objects are
1875 checksummed on disk. Also, journald should handle btrfs disk
1876 full events a lot more gracefully now, by processing SIGBUS
1877 errors, and not relying on fallocate() anymore.
1879 * When journald detects that journal files it is writing to
1880 have been deleted it will immediately start new journal
1883 * systemd now provides a way to store file descriptors
1884 per-service in PID 1.This is useful for daemons to ensure
1885 that fds they require are not lost during a daemon
1886 restart. The fds are passed to the daemon on the next
1887 invocation in the same way socket activation fds are
1888 passed. This is now used by journald to ensure that the
1889 various sockets connected to all the system's stdout/stderr
1890 are not lost when journald is restarted. File descriptors
1891 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
1892 an extension to sd_notify(). Note that a limit is enforced
1893 on the number of fds a service can store in PID 1, and it
1894 defaults to 0, so that no fds may be stored, unless this is
1895 explicitly turned on.
1897 * The default TERM variable to use for units connected to a
1898 terminal, when no other value is explicitly is set is now
1899 vt220 rather than vt102. This should be fairly safe still,
1900 but allows PgUp/PgDn work.
1902 * The /etc/crypttab option header= as known from Debian is now
1905 * "loginctl user-status" and "loginctl session-status" will
1906 now show the last 10 lines of log messages of the
1907 user/session following the status output. Similar,
1908 "machinectl status" will show the last 10 log lines
1909 associated with a virtual machine or container
1910 service. (Note that this is usually not the log messages
1911 done in the VM/container itself, but simply what the
1912 container manager logs. For nspawn this includes all console
1915 * "loginctl session-status" without further argument will now
1916 show the status of the session of the caller. Similar,
1917 "lock-session", "unlock-session", "activate",
1918 "enable-linger", "disable-linger" may now be called without
1919 session/user parameter in which case they apply to the
1920 caller's session/user.
1922 * An X11 session scriptlet is now shipped that uploads
1923 $DISPLAY and $XAUTHORITY into the environment of the systemd
1924 --user daemon if a session begins. This should improve
1925 compatibility with X11 enabled applications run as systemd
1928 * Generators are now subject to masking via /etc and /run, the
1929 same way as unit files.
1931 * networkd .network files gained support for configuring
1932 per-link IPv4/IPv6 packet forwarding as well as IPv4
1933 masquerading. This is by default turned on for veth links to
1934 containers, as registered by systemd-nspawn. This means that
1935 nspawn containers run with --network-veth will now get
1936 automatic routed access to the host's networks without any
1937 further configuration or setup, as long as networkd runs on
1940 * systemd-nspawn gained the --port= (-p) switch to expose TCP
1941 or UDP posts of a container on the host. With this in place
1942 it is possible to run containers with private veth links
1943 (--network-veth), and have their functionality exposed on
1944 the host as if their services were running directly on the
1947 * systemd-nspawn's --network-veth switch now gained a short
1948 version "-n", since with the changes above it is now truly
1949 useful out-of-the-box. The systemd-nspawn@.service has been
1950 updated to make use of it too by default.
1952 * systemd-nspawn will now maintain a per-image R/W lock, to
1953 ensure that the same image is not started more than once
1954 writable. (It's OK to run an image multiple times
1955 simultaneously in read-only mode.)
1957 * systemd-nspawn's --image= option is now capable of
1958 dissecting and booting MBR and GPT disk images that contain
1959 only a single active Linux partition. Previously it
1960 supported only GPT disk images with proper GPT type
1961 IDs. This allows running cloud images from major
1962 distributions directly with systemd-nspawn, without
1965 * In addition to collecting mouse dpi data in the udev
1966 hardware database, there's now support for collecting angle
1967 information for mouse scroll wheels. The database is
1968 supposed to guarantee similar scrolling behavior on mice
1969 that it knows about. There's also support for collecting
1970 information about Touchpad types.
1972 * udev's input_id built-in will now also collect touch screen
1973 dimension data and attach it to probed devices.
1975 * /etc/os-release gained support for a Distribution Privacy
1978 * networkd gained support for creating "ipvlan", "gretap",
1979 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
1981 * systemd-tmpfiles gained support for "a" lines for setting
1984 * systemd-nspawn will now mount /tmp in the container to
1985 tmpfs, automatically.
1987 * systemd now exposes the memory.usage_in_bytes cgroup
1988 attribute and shows it for each service in the "systemctl
1989 status" output, if available.
1991 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
1992 immediate reboot is triggered. This useful if shutdown is
1993 hung and is unable to complete, to expedite the
1994 operation. Note that this kind of reboot will still unmount
1995 all file systems, and hence should not result in fsck being
1998 * A .device unit for an optical block device will now be
1999 considered active only when a medium is in the drive. Also,
2000 mount units are now bound to their backing devices thus
2001 triggering automatic unmounting when devices become
2002 unavailable. With this in place systemd will now
2003 automatically unmount left-over mounts when a CD-ROM is
2004 ejected or an USB stick is yanked from the system.
2006 * networkd-wait-online now has support for waiting for
2007 specific interfaces only (with globbing), and for giving up
2008 after a configurable timeout.
2010 * networkd now exits when idle. It will be automatically
2011 restarted as soon as interfaces show up, are removed or
2012 change state. networkd will stay around as long as there is
2013 at least one DHCP state machine or similar around, that keep
2016 * networkd may now configure IPv6 link-local addressing in
2017 addition to IPv4 link-local addressing.
2019 * The IPv6 "token" for use in SLAAC may now be configured for
2020 each .network interface in networkd.
2022 * Routes configured with networkd may now be assigned a scope
2025 * networkd's [Match] sections now support globbing and lists
2026 of multiple space-separated matches per item.
2028 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
2029 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
2030 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
2031 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
2032 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
2033 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
2034 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
2035 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
2036 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
2037 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
2038 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
2039 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
2040 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
2041 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
2042 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
2043 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
2044 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
2045 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
2046 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
2047 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
2048 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
2049 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
2050 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
2051 Hoffmann, Zbigniew Jędrzejewski-Szmek
2053 — Berlin, 2015-02-16
2057 * When querying unit file enablement status (for example via
2058 "systemctl is-enabled"), a new state "indirect" is now known
2059 which indicates that a unit might not be enabled itself, but
2060 another unit listed in its Also= setting might be.
2062 * Similar to the various existing ConditionXYZ= settings for
2063 units, there are now matching AssertXYZ= settings. While
2064 failing conditions cause a unit to be skipped, but its job
2065 to succeed, failing assertions declared like this will cause
2066 a unit start operation and its job to fail.
2068 * hostnamed now knows a new chassis type "embedded".
2070 * systemctl gained a new "edit" command. When used on a unit
2071 file, this allows extending unit files with .d/ drop-in
2072 configuration snippets or editing the full file (after
2073 copying it from /usr/lib to /etc). This will invoke the
2074 user's editor (as configured with $EDITOR), and reload the
2075 modified configuration after editing.
2077 * "systemctl status" now shows the suggested enablement state
2078 for a unit, as declared in the (usually vendor-supplied)
2079 system preset files.
2081 * nss-myhostname will now resolve the single-label host name
2082 "gateway" to the locally configured default IP routing
2083 gateways, ordered by their metrics. This assigns a stable
2084 name to the used gateways, regardless which ones are
2085 currently configured. Note that the name will only be
2086 resolved after all other name sources (if nss-myhostname is
2087 configured properly) and should hence not negatively impact
2088 systems that use the single-label host name "gateway" in
2091 * systemd-inhibit now allows filtering by mode when listing
2094 * Scope and service units gained a new "Delegate" boolean
2095 property, which, when set, allows processes running inside the
2096 unit to further partition resources. This is primarily
2097 useful for systemd user instances as well as container
2100 * journald will now pick up audit messages directly from
2101 the kernel, and log them like any other log message. The
2102 audit fields are split up and fully indexed. This means that
2103 journalctl in many ways is now a (nicer!) alternative to
2104 ausearch, the traditional audit client. Note that this
2105 implements only a minimal audit client. If you want the
2106 special audit modes like reboot-on-log-overflow, please use
2107 the traditional auditd instead, which can be used in
2108 parallel to journald.
2110 * The ConditionSecurity= unit file option now understands the
2111 special string "audit" to check whether auditing is
2114 * journalctl gained two new commands --vacuum-size= and
2115 --vacuum-time= to delete old journal files until the
2116 remaining ones take up no more than the specified size on disk,
2117 or are not older than the specified time.
2119 * A new, native PPPoE library has been added to sd-network,
2120 systemd's library of light-weight networking protocols. This
2121 library will be used in a future version of networkd to
2122 enable PPPoE communication without an external pppd daemon.
2124 * The busctl tool now understands a new "capture" verb that
2125 works similar to "monitor", but writes a packet capture
2126 trace to STDOUT that can be redirected to a file which is
2127 compatible with libcap's capture file format. This can then
2128 be loaded in Wireshark and similar tools to inspect bus
2131 * The busctl tool now understands a new "tree" verb that shows
2132 the object trees of a specific service on the bus, or of all
2135 * The busctl tool now understands a new "introspect" verb that
2136 shows all interfaces and members of objects on the bus,
2137 including their signature and values. This is particularly
2138 useful to get more information about bus objects shown by
2139 the new "busctl tree" command.
2141 * The busctl tool now understands new verbs "call",
2142 "set-property" and "get-property" for invoking bus method
2143 calls, setting and getting bus object properties in a
2146 * busctl gained a new --augment-creds= argument that controls
2147 whether the tool shall augment credential information it
2148 gets from the bus with data from /proc, in a possibly
2151 * nspawn's --link-journal= switch gained two new values
2152 "try-guest" and "try-host" that work like "guest" and
2153 "host", but do not fail if the host has no persistent
2154 journalling enabled. -j is now equivalent to
2155 --link-journal=try-guest.
2157 * macvlan network devices created by nspawn will now have
2158 stable MAC addresses.
2160 * A new SmackProcessLabel= unit setting has been added, which
2161 controls the SMACK security label processes forked off by
2162 the respective unit shall use.
2164 * If compiled with --enable-xkbcommon, systemd-localed will
2165 verify x11 keymap settings by compiling the given keymap. It
2166 will spew out warnings if the compilation fails. This
2167 requires libxkbcommon to be installed.
2169 * When a coredump is collected, a larger number of metadata
2170 fields is now collected and included in the journal records
2171 created for it. More specifically, control group membership,
2172 environment variables, memory maps, working directory,
2173 chroot directory, /proc/$PID/status, and a list of open file
2174 descriptors is now stored in the log entry.
2176 * The udev hwdb now contains DPI information for mice. For
2179 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
2181 * All systemd programs that read standalone configuration
2182 files in /etc now also support a corresponding series of
2183 .conf.d configuration directories in /etc/, /run/,
2184 /usr/local/lib/, /usr/lib/, and (if configured with
2185 --enable-split-usr) /lib/. In particular, the following
2186 configuration files now have corresponding configuration
2187 directories: system.conf user.conf, logind.conf,
2188 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
2189 resolved.conf, timesyncd.conf, journal-remote.conf, and
2190 journal-upload.conf. Note that distributions should use the
2191 configuration directories in /usr/lib/; the directories in
2192 /etc/ are reserved for the system administrator.
2194 * systemd-rfkill will no longer take the rfkill device name
2195 into account when storing rfkill state on disk, as the name
2196 might be dynamically assigned and not stable. Instead, the
2197 ID_PATH udev variable combined with the rfkill type (wlan,
2198 bluetooth, ...) is used.
2200 * A new service systemd-machine-id-commit.service has been
2201 added. When used on systems where /etc is read-only during
2202 boot, and /etc/machine-id is not initialized (but an empty
2203 file), this service will copy the temporary machine ID
2204 created as replacement into /etc after the system is fully
2205 booted up. This is useful for systems that are freshly
2206 installed with a non-initialized machine ID, but should get
2207 a fixed machine ID for subsequent boots.
2209 * networkd's .netdev files now provide a large set of
2210 configuration parameters for VXLAN devices. Similarly, the
2211 bridge port cost parameter is now configurable in .network
2212 files. There's also new support for configuring IP source
2213 routing. networkd .link files gained support for a new
2214 OriginalName= match that is useful to match against the
2215 original interface name the kernel assigned. .network files
2216 may include MTU= and MACAddress= fields for altering the MTU
2217 and MAC address while being connected to a specific network
2220 * The LUKS logic gained supported for configuring
2221 UUID-specific key files. There's also new support for naming
2222 LUKS device from the kernel command line, using the new
2223 luks.name= argument.
2225 * Timer units may now be transiently created via the bus API
2226 (this was previously already available for scope and service
2227 units). In addition it is now possible to create multiple
2228 transient units at the same time with a single bus call. The
2229 "systemd-run" tool has been updated to make use of this for
2230 running commands on a specified time, in at(1)-style.
2232 * tmpfiles gained support for "t" lines, for assigning
2233 extended attributes to files. Among other uses this may be
2234 used to assign SMACK labels to files.
2236 Contributions from: Alin Rauta, Alison Chaiken, Andrej
2237 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
2238 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
2239 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
2240 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
2241 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
2242 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
2243 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
2244 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
2245 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
2246 Michael Biebl, Michael Chapman, Michael Marineau, Michal
2247 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
2248 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
2249 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
2250 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
2251 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
2252 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
2253 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
2255 — Berlin, 2014-12-10
2259 * journalctl gained the new options -t/--identifier= to match
2260 on the syslog identifier (aka "tag"), as well as --utc to
2261 show log timestamps in the UTC timezone. journalctl now also
2262 accepts -n/--lines=all to disable line capping in a pager.
2264 * journalctl gained a new switch, --flush, that synchronously
2265 flushes logs from /run/log/journal to /var/log/journal if
2266 persistent storage is enabled. systemd-journal-flush.service
2267 now waits until the operation is complete.
2269 * Services can notify the manager before they start a reload
2270 (by sending RELOADING=1) or shutdown (by sending
2271 STOPPING=1). This allows the manager to track and show the
2272 internal state of daemons and closes a race condition when
2273 the process is still running but has closed its D-Bus
2276 * Services with Type=oneshot do not have to have any ExecStart
2279 * User units are now loaded also from
2280 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
2281 /run/systemd/user directory that was already previously
2282 supported, but is under the control of the user.
2284 * Job timeouts (i.e. time-outs on the time a job that is
2285 queued stays in the run queue) can now optionally result in
2286 immediate reboot or power-off actions (JobTimeoutAction= and
2287 JobTimeoutRebootArgument=). This is useful on ".target"
2288 units, to limit the maximum time a target remains
2289 undispatched in the run queue, and to trigger an emergency
2290 operation in such a case. This is now used by default to
2291 turn off the system if boot-up (as defined by everything in
2292 basic.target) hangs and does not complete for at least
2293 15min. Also, if power-off or reboot hang for at least 30min
2294 an immediate power-off/reboot operation is triggered. This
2295 functionality is particularly useful to increase reliability
2296 on embedded devices, but also on laptops which might
2297 accidentally get powered on when carried in a backpack and
2298 whose boot stays stuck in a hard disk encryption passphrase
2301 * systemd-logind can be configured to also handle lid switch
2302 events even when the machine is docked or multiple displays
2303 are attached (HandleLidSwitchDocked= option).
2305 * A helper binary and a service have been added which can be
2306 used to resume from hibernation in the initramfs. A
2307 generator will parse the resume= option on the kernel
2308 command line to trigger resume.
2310 * A user console daemon systemd-consoled has been
2311 added. Currently, it is a preview, and will so far open a
2312 single terminal on each session of the user marked as
2313 Desktop=systemd-console.
2315 * Route metrics can be specified for DHCP routes added by
2318 * The SELinux context of socket-activated services can be set
2319 from the information provided by the networking stack
2320 (SELinuxContextFromNet= option).
2322 * Userspace firmware loading support has been removed and
2323 the minimum supported kernel version is thus bumped to 3.7.
2325 * Timeout for udev workers has been increased from 1 to 3
2326 minutes, but a warning will be printed after 1 minute to
2327 help diagnose kernel modules that take a long time to load.
2329 * Udev rules can now remove tags on devices with TAG-="foobar".
2331 * systemd's readahead implementation has been removed. In many
2332 circumstances it didn't give expected benefits even for
2333 rotational disk drives and was becoming less relevant in the
2334 age of SSDs. As none of the developers has been using
2335 rotating media anymore, and nobody stepped up to actively
2336 maintain this component of systemd it has now been removed.
2338 * Swap units can use Options= to specify discard options.
2339 Discard options specified for swaps in /etc/fstab are now
2342 * Docker containers are now detected as a separate type of
2345 * The Password Agent protocol gained support for queries where
2346 the user input is shown, useful e.g. for user names.
2347 systemd-ask-password gained a new --echo option to turn that
2350 * The default sysctl.d/ snippets will now set:
2352 net.core.default_qdisc = fq_codel
2354 This selects Fair Queuing Controlled Delay as the default
2355 queuing discipline for network interfaces. fq_codel helps
2356 fight the network bufferbloat problem. It is believed to be
2357 a good default with no tuning required for most workloads.
2358 Downstream distributions may override this choice. On 10Gbit
2359 servers that do not do forwarding, "fq" may perform better.
2360 Systems without a good clocksource should use "pfifo_fast".
2362 * If kdbus is enabled during build a new option BusPolicy= is
2363 available for service units, that allows locking all service
2364 processes into a stricter bus policy, in order to limit
2365 access to various bus services, or even hide most of them
2366 from the service's view entirely.
2368 * networkctl will now show the .network and .link file
2369 networkd has applied to a specific interface.
2371 * sd-login gained a new API call sd_session_get_desktop() to
2372 query which desktop environment has been selected for a
2375 * UNIX utmp support is now compile-time optional to support
2376 legacy-free systems.
2378 * systemctl gained two new commands "add-wants" and
2379 "add-requires" for pulling in units from specific targets
2382 * If the word "rescue" is specified on the kernel command line
2383 the system will now boot into rescue mode (aka
2384 rescue.target), which was previously available only by
2385 specifying "1" or "systemd.unit=rescue.target" on the kernel
2386 command line. This new kernel command line option nicely
2387 mirrors the already existing "emergency" kernel command line
2390 * New kernel command line options mount.usr=, mount.usrflags=,
2391 mount.usrfstype= have been added that match root=, rootflags=,
2392 rootfstype= but allow mounting a specific file system to
2395 * The $NOTIFY_SOCKET is now also passed to control processes of
2396 services, not only the main process.
2398 * This version reenables support for fsck's -l switch. This
2399 means at least version v2.25 of util-linux is required for
2400 operation, otherwise dead-locks on device nodes may
2401 occur. Again: you need to update util-linux to at least
2402 v2.25 when updating systemd to v217.
2404 * The "multi-seat-x" tool has been removed from systemd, as
2405 its functionality has been integrated into X servers 1.16,
2406 and the tool is hence redundant. It is recommended to update
2407 display managers invoking this tool to simply invoke X
2408 directly from now on, again.
2410 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
2411 message flag has been added for all of systemd's PolicyKit
2412 authenticated method calls has been added. In particular
2413 this now allows optional interactive authorization via
2414 PolicyKit for many of PID1's privileged operations such as
2415 unit file enabling and disabling.
2417 * "udevadm hwdb --update" learnt a new switch "--usr" for
2418 placing the rebuilt hardware database in /usr instead of
2419 /etc. When used only hardware database entries stored in
2420 /usr will be used, and any user database entries in /etc are
2421 ignored. This functionality is useful for vendors to ship a
2422 pre-built database on systems where local configuration is
2423 unnecessary or unlikely.
2425 * Calendar time specifications in .timer units now also
2426 understand the strings "semi-annually", "quarterly" and
2427 "minutely" as shortcuts (in addition to the preexisting
2428 "anually", "hourly", ...).
2430 * systemd-tmpfiles will now correctly create files in /dev
2431 at boot which are marked for creation only at boot. It is
2432 recommended to always create static device nodes with 'c!'
2433 and 'b!', so that they are created only at boot and not
2434 overwritten at runtime.
2436 * When the watchdog logic is used for a service (WatchdogSec=)
2437 and the watchdog timeout is hit the service will now be
2438 terminated with SIGABRT (instead of just SIGTERM), in order
2439 to make sure a proper coredump and backtrace is
2440 generated. This ensures that hanging services will result in
2441 similar coredump/backtrace behaviour as services that hit a
2444 Contributions from: Andreas Henriksson, Andrei Borzenkov,
2445 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
2446 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
2447 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
2448 Herrmann, David Sommerseth, David Strauss, Emil Renner
2449 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
2450 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
2451 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
2452 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
2453 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
2454 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
2455 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
2456 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
2457 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
2458 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
2459 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
2460 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
2461 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
2462 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
2463 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
2464 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
2467 — Berlin, 2014-10-28
2471 * timedated no longer reads NTP implementation unit names from
2472 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
2473 implementations should add a
2475 Conflicts=systemd-timesyncd.service
2477 to their unit files to take over and replace systemd's NTP
2478 default functionality.
2480 * systemd-sysusers gained a new line type "r" for configuring
2481 which UID/GID ranges to allocate system users/groups
2482 from. Lines of type "u" may now add an additional column
2483 that specifies the home directory for the system user to be
2484 created. Also, systemd-sysusers may now optionally read user
2485 information from STDIN instead of a file. This is useful for
2486 invoking it from RPM preinst scriptlets that need to create
2487 users before the first RPM file is installed since these
2488 files might need to be owned by them. A new
2489 %sysusers_create_inline RPM macro has been introduced to do
2490 just that. systemd-sysusers now updates the shadow files as
2491 well as the user/group databases, which should enhance
2492 compatibility with certain tools like grpck.
2494 * A number of bus APIs of PID 1 now optionally consult
2495 PolicyKit to permit access for otherwise unprivileged
2496 clients under certain conditions. Note that this currently
2497 doesn't support interactive authentication yet, but this is
2498 expected to be added eventually, too.
2500 * /etc/machine-info now has new fields for configuring the
2501 deployment environment of the machine, as well as the
2502 location of the machine. hostnamectl has been updated with
2503 new command to update these fields.
2505 * systemd-timesyncd has been updated to automatically acquire
2506 NTP server information from systemd-networkd, which might
2507 have been discovered via DHCP.
2509 * systemd-resolved now includes a caching DNS stub resolver
2510 and a complete LLMNR name resolution implementation. A new
2511 NSS module "nss-resolve" has been added which can be used
2512 instead of glibc's own "nss-dns" to resolve hostnames via
2513 systemd-resolved. Hostnames, addresses and arbitrary RRs may
2514 be resolved via systemd-resolved D-Bus APIs. In contrast to
2515 the glibc internal resolver systemd-resolved is aware of
2516 multi-homed system, and keeps DNS server and caches separate
2517 and per-interface. Queries are sent simultaneously on all
2518 interfaces that have DNS servers configured, in order to
2519 properly handle VPNs and local LANs which might resolve
2520 separate sets of domain names. systemd-resolved may acquire
2521 DNS server information from systemd-networkd automatically,
2522 which in turn might have discovered them via DHCP. A tool
2523 "systemd-resolve-host" has been added that may be used to
2524 query the DNS logic in resolved. systemd-resolved implements
2525 IDNA and automatically uses IDNA or UTF-8 encoding depending
2526 on whether classic DNS or LLMNR is used as transport. In the
2527 next releases we intend to add a DNSSEC and mDNS/DNS-SD
2528 implementation to systemd-resolved.
2530 * A new NSS module nss-mymachines has been added, that
2531 automatically resolves the names of all local registered
2532 containers to their respective IP addresses.
2534 * A new client tool "networkctl" for systemd-networkd has been
2535 added. It currently is entirely passive and will query
2536 networking configuration from udev, rtnetlink and networkd,
2537 and present it to the user in a very friendly
2538 way. Eventually, we hope to extend it to become a full
2539 control utility for networkd.
2541 * .socket units gained a new DeferAcceptSec= setting that
2542 controls the kernels' TCP_DEFER_ACCEPT sockopt for
2543 TCP. Similarly, support for controlling TCP keep-alive
2544 settings has been added (KeepAliveTimeSec=,
2545 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
2546 turning off Nagle's algorithm on TCP has been added
2549 * logind learned a new session type "web", for use in projects
2550 like Cockpit which register web clients as PAM sessions.
2552 * timer units with at least one OnCalendar= setting will now
2553 be started only after timer-sync.target has been
2554 reached. This way they will not elapse before the system
2555 clock has been corrected by a local NTP client or
2556 similar. This is particular useful on RTC-less embedded
2557 machines, that come up with an invalid system clock.
2559 * systemd-nspawn's --network-veth= switch should now result in
2560 stable MAC addresses for both the outer and the inner side
2563 * systemd-nspawn gained a new --volatile= switch for running
2564 container instances with /etc or /var unpopulated.
2566 * The kdbus client code has been updated to use the new Linux
2567 3.17 memfd subsystem instead of the old kdbus-specific one.
2569 * systemd-networkd's DHCP client and server now support
2570 FORCERENEW. There are also new configuration options to
2571 configure the vendor client identifier and broadcast mode
2574 * systemd will no longer inform the kernel about the current
2575 timezone, as this is necessarily incorrect and racy as the
2576 kernel has no understanding of DST and similar
2577 concepts. This hence means FAT timestamps will be always
2578 considered UTC, similar to what Android is already
2579 doing. Also, when the RTC is configured to the local time
2580 (rather than UTC) systemd will never synchronize back to it,
2581 as this might confuse Windows at a later boot.
2583 * systemd-analyze gained a new command "verify" for offline
2584 validation of unit files.
2586 * systemd-networkd gained support for a couple of additional
2587 settings for bonding networking setups. Also, the metric for
2588 statically configured routes may now be configured. For
2589 network interfaces where this is appropriate the peer IP
2590 address may now be configured.
2592 * systemd-networkd's DHCP client will no longer request
2593 broadcasting by default, as this tripped up some networks.
2594 For hardware where broadcast is required the feature should
2595 be switched back on using RequestBroadcast=yes.
2597 * systemd-networkd will now set up IPv4LL addresses (when
2598 enabled) even if DHCP is configured successfully.
2600 * udev will now default to respect network device names given
2601 by the kernel when the kernel indicates that these are
2602 predictable. This behavior can be tweaked by changing
2603 NamePolicy= in the relevant .link file.
2605 * A new library systemd-terminal has been added that
2606 implements full TTY stream parsing and rendering. This
2607 library is supposed to be used later on for implementing a
2608 full userspace VT subsystem, replacing the current kernel
2611 * A new tool systemd-journal-upload has been added to push
2612 journal data to a remote system running
2613 systemd-journal-remote.
2615 * journald will no longer forward all local data to another
2616 running syslog daemon. This change has been made because
2617 rsyslog (which appears to be the most commonly used syslog
2618 implementation these days) no longer makes use of this, and
2619 instead pulls the data out of the journal on its own. Since
2620 forwarding the messages to a non-existent syslog server is
2621 more expensive than we assumed we have now turned this
2622 off. If you run a syslog server that is not a recent rsyslog
2623 version, you have to turn this option on again
2624 (ForwardToSyslog= in journald.conf).
2626 * journald now optionally supports the LZ4 compressor for
2627 larger journal fields. This compressor should perform much
2628 better than XZ which was the previous default.
2630 * machinectl now shows the IP addresses of local containers,
2631 if it knows them, plus the interface name of the container.
2633 * A new tool "systemd-escape" has been added that makes it
2634 easy to escape strings to build unit names and similar.
2636 * sd_notify() messages may now include a new ERRNO= field
2637 which is parsed and collected by systemd and shown among the
2638 "systemctl status" output for a service.
2640 * A new component "systemd-firstboot" has been added that
2641 queries the most basic systemd information (timezone,
2642 hostname, root password) interactively on first
2643 boot. Alternatively it may also be used to provision these
2644 things offline on OS images installed into directories.
2646 * The default sysctl.d/ snippets will now set
2648 net.ipv4.conf.default.promote_secondaries=1
2650 This has the benefit of no flushing secondary IP addresses
2651 when primary addresses are removed.
2653 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
2654 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
2655 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
2656 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
2657 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
2658 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
2659 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
2660 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
2661 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
2662 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
2663 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
2664 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
2665 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
2666 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
2667 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
2669 — Berlin, 2014-08-19
2673 * A new tool systemd-sysusers has been added. This tool
2674 creates system users and groups in /etc/passwd and
2675 /etc/group, based on static declarative system user/group
2676 definitions in /usr/lib/sysusers.d/. This is useful to
2677 enable factory resets and volatile systems that boot up with
2678 an empty /etc directory, and thus need system users and
2679 groups created during early boot. systemd now also ships
2680 with two default sysusers.d/ files for the most basic
2681 users and groups systemd and the core operating system
2684 * A new tmpfiles snippet has been added that rebuilds the
2685 essential files in /etc on boot, should they be missing.
2687 * A directive for ensuring automatic clean-up of
2688 /var/cache/man/ has been removed from the default
2689 configuration. This line should now be shipped by the man
2690 implementation. The necessary change has been made to the
2691 man-db implementation. Note that you need to update your man
2692 implementation to one that ships this line, otherwise no
2693 automatic clean-up of /var/cache/man will take place.
2695 * A new condition ConditionNeedsUpdate= has been added that
2696 may conditionalize services to only run when /etc or /var
2697 are "older" than the vendor operating system resources in
2698 /usr. This is useful for reconstructing or updating /etc
2699 after an offline update of /usr or a factory reset, on the
2700 next reboot. Services that want to run once after such an
2701 update or reset should use this condition and order
2702 themselves before the new systemd-update-done.service, which
2703 will mark the two directories as fully updated. A number of
2704 service files have been added making use of this, to rebuild
2705 the udev hardware database, the journald message catalog and
2706 dynamic loader cache (ldconfig). The systemd-sysusers tool
2707 described above also makes use of this now. With this in
2708 place it is now possible to start up a minimal operating
2709 system with /etc empty cleanly. For more information on the
2710 concepts involved see this recent blog story:
2712 http://0pointer.de/blog/projects/stateless.html
2714 * A new system group "input" has been introduced, and all
2715 input device nodes get this group assigned. This is useful
2716 for system-level software to get access to input devices. It
2717 complements what is already done for "audio" and "video".
2719 * systemd-networkd learnt minimal DHCPv4 server support in
2720 addition to the existing DHCPv4 client support. It also
2721 learnt DHCPv6 client and IPv6 Router Solicitation client
2722 support. The DHCPv4 client gained support for static routes
2723 passed in from the server. Note that the [DHCPv4] section
2724 known in older systemd-networkd versions has been renamed to
2725 [DHCP] and is now also used by the DHCPv6 client. Existing
2726 .network files using settings of this section should be
2727 updated, though compatibility is maintained. Optionally, the
2728 client hostname may now be sent to the DHCP server.
2730 * networkd gained support for vxlan virtual networks as well
2731 as tun/tap and dummy devices.
2733 * networkd gained support for automatic allocation of address
2734 ranges for interfaces from a system-wide pool of
2735 addresses. This is useful for dynamically managing a large
2736 number of interfaces with a single network configuration
2737 file. In particular this is useful to easily assign
2738 appropriate IP addresses to the veth links of a large number
2739 of nspawn instances.
2741 * RPM macros for processing sysusers, sysctl and binfmt
2742 drop-in snippets at package installation time have been
2745 * The /etc/os-release file should now be placed in
2746 /usr/lib/os-release. The old location is automatically
2747 created as symlink. /usr/lib is the more appropriate
2748 location of this file, since it shall actually describe the
2749 vendor operating system shipped in /usr, and not the
2750 configuration stored in /etc.
2752 * .mount units gained a new boolean SloppyOptions= setting
2753 that maps to mount(8)'s -s option which enables permissive
2754 parsing of unknown mount options.
2756 * tmpfiles learnt a new "L+" directive which creates a symlink
2757 but (unlike "L") deletes a pre-existing file first, should
2758 it already exist and not already be the correct
2759 symlink. Similarly, "b+", "c+" and "p+" directives have been
2760 added as well, which create block and character devices, as
2761 well as fifos in the filesystem, possibly removing any
2762 pre-existing files of different types.
2764 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
2765 'argument' field (which so far specified the source to
2766 symlink/copy the files from) is now optional. If omitted the
2767 same file os copied from /usr/share/factory/ suffixed by the
2768 full destination path. This is useful for populating /etc
2769 with essential files, by copying them from vendor defaults
2770 shipped in /usr/share/factory/etc.
2772 * A new command "systemctl preset-all" has been added that
2773 applies the service preset settings to all installed unit
2774 files. A new switch --preset-mode= has been added that
2775 controls whether only enable or only disable operations
2778 * A new command "systemctl is-system-running" has been added
2779 that allows checking the overall state of the system, for
2780 example whether it is fully up and running.
2782 * When the system boots up with an empty /etc, the equivalent
2783 to "systemctl preset-all" is executed during early boot, to
2784 make sure all default services are enabled after a factory
2787 * systemd now contains a minimal preset file that enables the
2788 most basic services systemd ships by default.
2790 * Unit files' [Install] section gained a new DefaultInstance=
2791 field for defining the default instance to create if a
2792 template unit is enabled with no instance specified.
2794 * A new passive target cryptsetup-pre.target has been added
2795 that may be used by services that need to make they run and
2796 finish before the first LUKS cryptographic device is set up.
2798 * The /dev/loop-control and /dev/btrfs-control device nodes
2799 are now owned by the "disk" group by default, opening up
2800 access to this group.
2802 * systemd-coredump will now automatically generate a
2803 stack trace of all core dumps taking place on the system,
2804 based on elfutils' libdw library. This stack trace is logged
2807 * systemd-coredump may now optionally store coredumps directly
2808 on disk (in /var/lib/systemd/coredump, possibly compressed),
2809 instead of storing them unconditionally in the journal. This
2810 mode is the new default. A new configuration file
2811 /etc/systemd/coredump.conf has been added to configure this
2812 and other parameters of systemd-coredump.
2814 * coredumpctl gained a new "info" verb to show details about a
2815 specific coredump. A new switch "-1" has also been added
2816 that makes sure to only show information about the most
2817 recent entry instead of all entries. Also, as the tool is
2818 generally useful now the "systemd-" prefix of the binary
2819 name has been removed. Distributions that want to maintain
2820 compatibility with the old name should add a symlink from
2821 the old name to the new name.
2823 * journald's SplitMode= now defaults to "uid". This makes sure
2824 that unprivileged users can access their own coredumps with
2825 coredumpctl without restrictions.
2827 * New kernel command line options "systemd.wants=" (for
2828 pulling an additional unit during boot), "systemd.mask="
2829 (for masking a specific unit for the boot), and
2830 "systemd.debug-shell" (for enabling the debug shell on tty9)
2831 have been added. This is implemented in the new generator
2832 "systemd-debug-generator".
2834 * systemd-nspawn will now by default filter a couple of
2835 syscalls for containers, among them those required for
2836 kernel module loading, direct x86 IO port access, swap
2837 management, and kexec. Most importantly though
2838 open_by_handle_at() is now prohibited for containers,
2839 closing a hole similar to a recently discussed vulnerability
2840 in docker regarding access to files on file hierarchies the
2841 container should normally not have access to. Note that, for
2842 nspawn, we generally make no security claims anyway (and
2843 this is explicitly documented in the man page), so this is
2844 just a fix for one of the most obvious problems.
2846 * A new man page file-hierarchy(7) has been added that
2847 contains a minimized, modernized version of the file system
2848 layout systemd expects, similar in style to the FHS
2849 specification or hier(5). A new tool systemd-path(1) has
2850 been added to query many of these paths for the local
2853 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
2854 longer done. Since the directory now has a per-user size
2855 limit, and is cleaned on logout this appears unnecessary,
2856 in particular since this now brings the lifecycle of this
2857 directory closer in line with how IPC objects are handled.
2859 * systemd.pc now exports a number of additional directories,
2860 including $libdir (which is useful to identify the library
2861 path for the primary architecture of the system), and a
2862 couple of drop-in directories.
2864 * udev's predictable network interface names now use the dev_port
2865 sysfs attribute, introduced in linux 3.15 instead of dev_id to
2866 distinguish between ports of the same PCI function. dev_id should
2867 only be used for ports using the same HW address, hence the need
2870 * machined has been updated to export the OS version of a
2871 container (read from /etc/os-release and
2872 /usr/lib/os-release) on the bus. This is now shown in
2873 "machinectl status" for a machine.
2875 * A new service setting RestartForceExitStatus= has been
2876 added. If configured to a set of exit signals or process
2877 return values, the service will be restarted when the main
2878 daemon process exits with any of them, regardless of the
2881 * systemctl's -H switch for connecting to remote systemd
2882 machines has been extended so that it may be used to
2883 directly connect to a specific container on the
2884 host. "systemctl -H root@foobar:waldi" will now connect as
2885 user "root" to host "foobar", and then proceed directly to
2886 the container named "waldi". Note that currently you have to
2887 authenticate as user "root" for this to work, as entering
2888 containers is a privileged operation.
2890 Contributions from: Andreas Henriksson, Benjamin Steinwender,
2891 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
2892 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
2893 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
2894 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
2895 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
2896 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
2897 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
2898 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
2899 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
2900 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
2901 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
2903 — Berlin, 2014-07-03
2907 * As an experimental feature, udev now tries to lock the
2908 disk device node (flock(LOCK_SH|LOCK_NB)) while it
2909 executes events for the disk or any of its partitions.
2910 Applications like partitioning programs can lock the
2911 disk device node (flock(LOCK_EX)) and claim temporary
2912 device ownership that way; udev will entirely skip all event
2913 handling for this disk and its partitions. If the disk
2914 was opened for writing, the close will trigger a partition
2915 table rescan in udev's "watch" facility, and if needed
2916 synthesize "change" events for the disk and all its partitions.
2917 This is now unconditionally enabled, and if it turns out to
2918 cause major problems, we might turn it on only for specific
2919 devices, or might need to disable it entirely. Device Mapper
2920 devices are excluded from this logic.
2922 * We temporarily dropped the "-l" switch for fsck invocations,
2923 since they collide with the flock() logic above. util-linux
2924 upstream has been changed already to avoid this conflict,
2925 and we will readd "-l" as soon as util-linux with this
2926 change has been released.
2928 * The dependency on libattr has been removed. Since a long
2929 time, the extended attribute calls have moved to glibc, and
2930 libattr is thus unnecessary.
2932 * Virtualization detection works without priviliges now. This
2933 means the systemd-detect-virt binary no longer requires
2934 CAP_SYS_PTRACE file capabilities, and our daemons can run
2935 with fewer privileges.
2937 * systemd-networkd now runs under its own "systemd-network"
2938 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
2939 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
2940 loses the ability to write to files owned by root this way.
2942 * Similarly, systemd-resolved now runs under its own
2943 "systemd-resolve" user with no capabilities remaining.
2945 * Similarly, systemd-bus-proxyd now runs under its own
2946 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
2948 * systemd-networkd gained support for setting up "veth"
2949 virtual Ethernet devices for container connectivity, as well
2950 as GRE and VTI tunnels.
2952 * systemd-networkd will no longer automatically attempt to
2953 manually load kernel modules necessary for certain tunnel
2954 transports. Instead, it is assumed the kernel loads them
2955 automatically when required. This only works correctly on
2956 very new kernels. On older kernels, please consider adding
2957 the kernel modules to /etc/modules-load.d/ as a work-around.
2959 * The resolv.conf file systemd-resolved generates has been
2960 moved to /run/systemd/resolve/. If you have a symlink from
2961 /etc/resolv.conf, it might be necessary to correct it.
2963 * Two new service settings, ProtectHome= and ProtectSystem=,
2964 have been added. When enabled, they will make the user data
2965 (such as /home) inaccessible or read-only and the system
2966 (such as /usr) read-only, for specific services. This allows
2967 very light-weight per-service sandboxing to avoid
2968 modifications of user data or system files from
2969 services. These two new switches have been enabled for all
2970 of systemd's long-running services, where appropriate.
2972 * Socket units gained new SocketUser= and SocketGroup=
2973 settings to set the owner user and group of AF_UNIX sockets
2974 and FIFOs in the file system.
2976 * Socket units gained a new RemoveOnStop= setting. If enabled,
2977 all FIFOS and sockets in the file system will be removed
2978 when the specific socket unit is stopped.
2980 * Socket units gained a new Symlinks= setting. It takes a list
2981 of symlinks to create to file system sockets or FIFOs
2982 created by the specific Unix sockets. This is useful to
2983 manage symlinks to socket nodes with the same life-cycle as
2986 * The /dev/log socket and /dev/initctl FIFO have been moved to
2987 /run, and have been replaced by symlinks. This allows
2988 connecting to these facilities even if PrivateDevices=yes is
2989 used for a service (which makes /dev/log itself unavailable,
2990 but /run is left). This also has the benefit of ensuring
2991 that /dev only contains device nodes, directories and
2992 symlinks, and nothing else.
2994 * sd-daemon gained two new calls sd_pid_notify() and
2995 sd_pid_notifyf(). They are similar to sd_notify() and
2996 sd_notifyf(), but allow overriding of the source PID of
2997 notification messages if permissions permit this. This is
2998 useful to send notify messages on behalf of a different
2999 process (for example, the parent process). The
3000 systemd-notify tool has been updated to make use of this
3001 when sending messages (so that notification messages now
3002 originate from the shell script invoking systemd-notify and
3003 not the systemd-notify process itself. This should minimize
3004 a race where systemd fails to associate notification
3005 messages to services when the originating process already
3008 * A new "on-abnormal" setting for Restart= has been added. If
3009 set, it will result in automatic restarts on all "abnormal"
3010 reasons for a process to exit, which includes unclean
3011 signals, core dumps, timeouts and watchdog timeouts, but
3012 does not include clean and unclean exit codes or clean
3013 signals. Restart=on-abnormal is an alternative for
3014 Restart=on-failure for services that shall be able to
3015 terminate and avoid restarts on certain errors, by
3016 indicating so with an unclean exit code. Restart=on-failure
3017 or Restart=on-abnormal is now the recommended setting for
3018 all long-running services.
3020 * If the InaccessibleDirectories= service setting points to a
3021 mount point (or if there are any submounts contained within
3022 it), it is now attempted to completely unmount it, to make
3023 the file systems truly unavailable for the respective
3026 * The ReadOnlyDirectories= service setting and
3027 systemd-nspawn's --read-only parameter are now recursively
3028 applied to all submounts, too.
3030 * Mount units may now be created transiently via the bus APIs.
3032 * The support for SysV and LSB init scripts has been removed
3033 from the systemd daemon itself. Instead, it is now
3034 implemented as a generator that creates native systemd units
3035 from these scripts when needed. This enables us to remove a
3036 substantial amount of legacy code from PID 1, following the
3037 fact that many distributions only ship a very small number
3038 of LSB/SysV init scripts nowadays.
3040 * Privileged Xen (dom0) domains are not considered
3041 virtualization anymore by the virtualization detection
3042 logic. After all, they generally have unrestricted access to
3043 the hardware and usually are used to manage the unprivileged
3046 * systemd-tmpfiles gained a new "C" line type, for copying
3047 files or entire directories.
3049 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
3050 lines. So far, they have been non-globbing versions of the
3051 latter, and have thus been redundant. In future, it is
3052 recommended to only use "z". "m" has hence been removed
3053 from the documentation, even though it stays supported.
3055 * A tmpfiles snippet to recreate the most basic structure in
3056 /var has been added. This is enough to create the /var/run →
3057 /run symlink and create a couple of structural
3058 directories. This allows systems to boot up with an empty or
3059 volatile /var. Of course, while with this change, the core OS
3060 now is capable with dealing with a volatile /var, not all
3061 user services are ready for it. However, we hope that sooner
3062 or later, many service daemons will be changed upstream so
3063 that they are able to automatically create their necessary
3064 directories in /var at boot, should they be missing. This is
3065 the first step to allow state-less systems that only require
3066 the vendor image for /usr to boot.
3068 * systemd-nspawn has gained a new --tmpfs= switch to mount an
3069 empty tmpfs instance to a specific directory. This is
3070 particularly useful for making use of the automatic
3071 reconstruction of /var (see above), by passing --tmpfs=/var.
3073 * Access modes specified in tmpfiles snippets may now be
3074 prefixed with "~", which indicates that they shall be masked
3075 by whether the existing file or directory is currently
3076 writable, readable or executable at all. Also, if specified,
3077 the sgid/suid/sticky bits will be masked for all
3080 * A new passive target unit "network-pre.target" has been
3081 added which is useful for services that shall run before any
3082 network is configured, for example firewall scripts.
3084 * The "floppy" group that previously owned the /dev/fd*
3085 devices is no longer used. The "disk" group is now used
3086 instead. Distributions should probably deprecate usage of
3089 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
3090 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
3091 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
3092 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
3093 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
3094 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
3097 — Berlin, 2014-06-11
3101 * A new "systemd-timesyncd" daemon has been added for
3102 synchronizing the system clock across the network. It
3103 implements an SNTP client. In contrast to NTP
3104 implementations such as chrony or the NTP reference server,
3105 this only implements a client side, and does not bother with
3106 the full NTP complexity, focusing only on querying time from
3107 one remote server and synchronizing the local clock to
3108 it. Unless you intend to serve NTP to networked clients or
3109 want to connect to local hardware clocks, this simple NTP
3110 client should be more than appropriate for most
3111 installations. The daemon runs with minimal privileges, and
3112 has been hooked up with networkd to only operate when
3113 network connectivity is available. The daemon saves the
3114 current clock to disk every time a new NTP sync has been
3115 acquired, and uses this to possibly correct the system clock
3116 early at bootup, in order to accommodate for systems that
3117 lack an RTC such as the Raspberry Pi and embedded devices,
3118 and to make sure that time monotonically progresses on these
3119 systems, even if it is not always correct. To make use of
3120 this daemon, a new system user and group "systemd-timesync"
3121 needs to be created on installation of systemd.
3123 * The queue "seqnum" interface of libudev has been disabled, as
3124 it was generally incompatible with device namespacing as
3125 sequence numbers of devices go "missing" if the devices are
3126 part of a different namespace.
3128 * "systemctl list-timers" and "systemctl list-sockets" gained
3129 a --recursive switch for showing units of these types also
3130 for all local containers, similar in style to the already
3131 supported --recursive switch for "systemctl list-units".
3133 * A new RebootArgument= setting has been added for service
3134 units, which may be used to specify a kernel reboot argument
3135 to use when triggering reboots with StartLimitAction=.
3137 * A new FailureAction= setting has been added for service
3138 units which may be used to specify an operation to trigger
3139 when a service fails. This works similarly to
3140 StartLimitAction=, but unlike it, controls what is done
3141 immediately rather than only after several attempts to
3142 restart the service in question.
3144 * hostnamed got updated to also expose the kernel name,
3145 release, and version on the bus. This is useful for
3146 executing commands like hostnamectl with the -H switch.
3147 systemd-analyze makes use of this to properly display
3148 details when running non-locally.
3150 * The bootchart tool can now show cgroup information in the
3151 graphs it generates.
3153 * The CFS CPU quota cgroup attribute is now exposed for
3154 services. The new CPUQuota= switch has been added for this
3155 which takes a percentage value. Setting this will have the
3156 result that a service may never get more CPU time than the
3157 specified percentage, even if the machine is otherwise idle.
3159 * systemd-networkd learned IPIP and SIT tunnel support.
3161 * LSB init scripts exposing a dependency on $network will now
3162 get a dependency on network-online.target rather than simply
3163 network.target. This should bring LSB handling closer to
3164 what it was on SysV systems.
3166 * A new fsck.repair= kernel option has been added to control
3167 how fsck shall deal with unclean file systems at boot.
3169 * The (.ini) configuration file parser will now silently
3170 ignore sections whose name begins with "X-". This may be
3171 used to maintain application-specific extension sections in unit
3174 * machined gained a new API to query the IP addresses of
3175 registered containers. "machinectl status" has been updated
3176 to show these addresses in its output.
3178 * A new call sd_uid_get_display() has been added to the
3179 sd-login APIs for querying the "primary" session of a
3180 user. The "primary" session of the user is elected from the
3181 user's sessions and generally a graphical session is
3182 preferred over a text one.
3184 * A minimal systemd-resolved daemon has been added. It
3185 currently simply acts as a companion to systemd-networkd and
3186 manages resolv.conf based on per-interface DNS
3187 configuration, possibly supplied via DHCP. In the long run
3188 we hope to extend this into a local DNSSEC enabled DNS and
3191 * The systemd-networkd-wait-online tool is now enabled by
3192 default. It will delay network-online.target until a network
3193 connection has been configured. The tool primarily integrates
3194 with networkd, but will also make a best effort to make sense
3195 of network configuration performed in some other way.
3197 * Two new service options StartupCPUShares= and
3198 StartupBlockIOWeight= have been added that work similarly to
3199 CPUShares= and BlockIOWeight= however only apply during
3200 system startup. This is useful to prioritize certain services
3201 differently during bootup than during normal runtime.
3203 * hostnamed has been changed to prefer the statically
3204 configured hostname in /etc/hostname (unless set to
3205 'localhost' or empty) over any dynamic one supplied by
3206 dhcp. With this change, the rules for picking the hostname
3207 match more closely the rules of other configuration settings
3208 where the local administrator's configuration in /etc always
3209 overrides any other settings.
3211 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
3212 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
3213 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
3214 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
3215 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
3216 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
3217 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
3218 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
3219 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
3220 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
3221 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
3222 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
3223 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
3224 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
3225 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
3226 Lindskog, WaLyong Cho, Will Woods, Zbigniew
3229 — Beijing, 2014-05-28
3233 * When restoring the screen brightness at boot, stay away from
3234 the darkest setting or from the lowest 5% of the available
3235 range, depending on which is the larger value of both. This
3236 should effectively protect the user from rebooting into a
3237 black screen, should the brightness have been set to minimum
3240 * sd-login gained a new sd_machine_get_class() call to
3241 determine the class ("vm" or "container") of a machine
3242 registered with machined.
3244 * sd-login gained new calls
3245 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
3246 to query the identity of the peer of a local AF_UNIX
3247 connection. They operate similarly to their sd_pid_get_xyz()
3250 * PID 1 will now maintain a system-wide system state engine
3251 with the states "starting", "running", "degraded",
3252 "maintenance", "stopping". These states are bound to system
3253 startup, normal runtime, runtime with at least one failed
3254 service, rescue/emergency mode and system shutdown. This
3255 state is shown in the "systemctl status" output when no unit
3256 name is passed. It is useful to determine system state, in
3257 particularly when doing so for many systems or containers at
3260 * A new command "list-machines" has been added to "systemctl"
3261 that lists all local OS containers and shows their system
3262 state (see above), if systemd runs inside of them.
3264 * systemctl gained a new "-r" switch to recursively enumerate
3265 units on all local containers, when used with the
3266 "list-unit" command (which is the default one that is
3267 executed when no parameters are specified).
3269 * The GPT automatic partition discovery logic will now honour
3270 two GPT partition flags: one may be set on a partition to
3271 cause it to be mounted read-only, and the other may be set
3272 on a partition to ignore it during automatic discovery.
3274 * Two new GPT type UUIDs have been added for automatic root
3275 partition discovery, for 32-bit and 64-bit ARM. This is not
3276 particularly useful for discovering the root directory on
3277 these architectures during bare-metal boots (since UEFI is
3278 not common there), but still very useful to allow booting of
3279 ARM disk images in nspawn with the -i option.
3281 * MAC addresses of interfaces created with nspawn's
3282 --network-interface= switch will now be generated from the
3283 machine name, and thus be stable between multiple invocations
3286 * logind will now automatically remove all IPC objects owned
3287 by a user if she or he fully logs out. This makes sure that
3288 users who are logged out cannot continue to consume IPC
3289 resources. This covers SysV memory, semaphores and message
3290 queues as well as POSIX shared memory and message
3291 queues. Traditionally, SysV and POSIX IPC had no life-cycle
3292 limits. With this functionality, that is corrected. This may
3293 be turned off by using the RemoveIPC= switch of logind.conf.
3295 * The systemd-machine-id-setup and tmpfiles tools gained a
3296 --root= switch to operate on a specific root directory,
3299 * journald can now forward logged messages to the TTYs of all
3300 logged in users ("wall"). This is the default for all
3301 emergency messages now.
3303 * A new tool systemd-journal-remote has been added to stream
3304 journal log messages across the network.
3306 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
3307 controller trees are mounted into it. Note that the
3308 directories mounted beneath it are not read-only. This is a
3309 security measure and is particularly useful because glibc
3310 actually includes a search logic to pick any tmpfs it can
3311 find to implement shm_open() if /dev/shm is not available
3312 (which it might very well be in namespaced setups).
3314 * machinectl gained a new "poweroff" command to cleanly power
3315 down a local OS container.
3317 * The PrivateDevices= unit file setting will now also drop the
3318 CAP_MKNOD capability from the capability bound set, and
3319 imply DevicePolicy=closed.
3321 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
3322 comprehensively on all long-running systemd services where
3323 this is appropriate.
3325 * systemd-udevd will now run in a disassociated mount
3326 namespace. To mount directories from udev rules, make sure to
3327 pull in mount units via SYSTEMD_WANTS properties.
3329 * The kdbus support gained support for uploading policy into
3330 the kernel. sd-bus gained support for creating "monitoring"
3331 connections that can eavesdrop into all bus communication
3332 for debugging purposes.
3334 * Timestamps may now be specified in seconds since the UNIX
3335 epoch Jan 1st, 1970 by specifying "@" followed by the value
3338 * Native tcpwrap support in systemd has been removed. tcpwrap
3339 is old code, not really maintained anymore and has serious
3340 shortcomings, and better options such as firewalls
3341 exist. For setups that require tcpwrap usage, please
3342 consider invoking your socket-activated service via tcpd,
3343 like on traditional inetd.
3345 * A new system.conf configuration option
3346 DefaultTimerAccuracySec= has been added that controls the
3347 default AccuracySec= setting of .timer units.
3349 * Timer units gained a new WakeSystem= switch. If enabled,
3350 timers configured this way will cause the system to resume
3351 from system suspend (if the system supports that, which most
3354 * Timer units gained a new Persistent= switch. If enabled,
3355 timers configured this way will save to disk when they have
3356 been last triggered. This information is then used on next
3357 reboot to possible execute overdue timer events, that
3358 could not take place because the system was powered off.
3359 This enables simple anacron-like behaviour for timer units.
3361 * systemctl's "list-timers" will now also list the time a
3362 timer unit was last triggered in addition to the next time
3363 it will be triggered.
3365 * systemd-networkd will now assign predictable IPv4LL
3366 addresses to its local interfaces.
3368 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
3369 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
3370 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
3371 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
3372 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
3373 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
3374 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
3375 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
3378 — Berlin, 2014-03-25
3382 * A new unit file setting RestrictAddressFamilies= has been
3383 added to restrict which socket address families unit
3384 processes gain access to. This takes address family names
3385 like "AF_INET" or "AF_UNIX", and is useful to minimize the
3386 attack surface of services via exotic protocol stacks. This
3387 is built on seccomp system call filters.
3389 * Two new unit file settings RuntimeDirectory= and
3390 RuntimeDirectoryMode= have been added that may be used to
3391 manage a per-daemon runtime directories below /run. This is
3392 an alternative for setting up directory permissions with
3393 tmpfiles snippets, and has the advantage that the runtime
3394 directory's lifetime is bound to the daemon runtime and that
3395 the daemon starts up with an empty directory each time. This
3396 is particularly useful when writing services that drop
3397 privileges using the User= or Group= setting.
3399 * The DeviceAllow= unit setting now supports globbing for
3400 matching against device group names.
3402 * The systemd configuration file system.conf gained new
3403 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
3404 DefaultMemoryAccounting= to globally turn on/off accounting
3405 for specific resources (cgroups) for all units. These
3406 settings may still be overridden individually in each unit
3409 * systemd-gpt-auto-generator is now able to discover /srv and
3410 root partitions in addition to /home and swap partitions. It
3411 also supports LUKS-encrypted partitions now. With this in
3412 place, automatic discovery of partitions to mount following
3413 the Discoverable Partitions Specification
3414 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
3415 is now a lot more complete. This allows booting without
3416 /etc/fstab and without root= on the kernel command line on
3417 systems prepared appropriately.
3419 * systemd-nspawn gained a new --image= switch which allows
3420 booting up disk images and Linux installations on any block
3421 device that follow the Discoverable Partitions Specification
3422 (see above). This means that installations made with
3423 appropriately updated installers may now be started and
3424 deployed using container managers, completely
3425 unmodified. (We hope that libvirt-lxc will add support for
3426 this feature soon, too.)
3428 * systemd-nspawn gained a new --network-macvlan= setting to
3429 set up a private macvlan interface for the
3430 container. Similarly, systemd-networkd gained a new
3431 Kind=macvlan setting in .netdev files.
3433 * systemd-networkd now supports configuring local addresses
3436 * A new tool systemd-network-wait-online has been added to
3437 synchronously wait for network connectivity using
3440 * The sd-bus.h bus API gained a new sd_bus_track object for
3441 tracking the life-cycle of bus peers. Note that sd-bus.h is
3442 still not a public API though (unless you specify
3443 --enable-kdbus on the configure command line, which however
3444 voids your warranty and you get no API stability guarantee).
3446 * The $XDG_RUNTIME_DIR runtime directories for each user are
3447 now individual tmpfs instances, which has the benefit of
3448 introducing separate pools for each user, with individual
3449 size limits, and thus making sure that unprivileged clients
3450 can no longer negatively impact the system or other users by
3451 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
3452 RuntimeDirectorySize= has been introduced that allows
3453 controlling the default size limit for all users. It
3454 defaults to 10% of the available physical memory. This is no
3455 replacement for quotas on tmpfs though (which the kernel
3456 still does not support), as /dev/shm and /tmp are still
3457 shared resources used by both the system and unprivileged
3460 * logind will now automatically turn off automatic suspending
3461 on laptop lid close when more than one display is
3462 connected. This was previously expected to be implemented
3463 individually in desktop environments (such as GNOME),
3464 however has been added to logind now, in order to fix a
3465 boot-time race where a desktop environment might not have
3466 been started yet and thus not been able to take an inhibitor
3467 lock at the time where logind already suspends the system
3468 due to a closed lid.
3470 * logind will now wait at least 30s after each system
3471 suspend/resume cycle, and 3min after system boot before
3472 suspending the system due to a closed laptop lid. This
3473 should give USB docking stations and similar enough time to
3474 be probed and configured after system resume and boot in
3475 order to then act as suspend blocker.
3477 * systemd-run gained a new --property= setting which allows
3478 initialization of resource control properties (and others)
3479 for the created scope or service unit. Example: "systemd-run
3480 --property=BlockIOWeight=10 updatedb" may be used to run
3481 updatedb at a low block IO scheduling weight.
3483 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
3484 now also work in --scope mode.
3486 * When systemd is compiled with kdbus support, basic support
3487 for enforced policies is now in place. (Note that enabling
3488 kdbus still voids your warranty and no API compatibility
3491 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
3492 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
3493 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
3494 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
3495 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
3496 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
3497 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
3498 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
3499 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
3500 Zbigniew Jędrzejewski-Szmek
3502 — Berlin, 2014-03-12
3506 * systemd will now relabel /dev after loading the SMACK policy
3507 according to SMACK rules.
3509 * A new unit file option AppArmorProfile= has been added to
3510 set the AppArmor profile for the processes of a unit.
3512 * A new condition check ConditionArchitecture= has been added
3513 to conditionalize units based on the system architecture, as
3514 reported by uname()'s "machine" field.
3516 * systemd-networkd now supports matching on the system
3517 virtualization, architecture, kernel command line, host name
3520 * logind is now a lot more aggressive when suspending the
3521 machine due to a closed laptop lid. Instead of acting only
3522 on the lid close action, it will continuously watch the lid
3523 status and act on it. This is useful for laptops where the
3524 power button is on the outside of the chassis so that it can
3525 be reached without opening the lid (such as the Lenovo
3526 Yoga). On those machines, logind will now immediately
3527 re-suspend the machine if the power button has been
3528 accidentally pressed while the laptop was suspended and in a
3529 backpack or similar.
3531 * logind will now watch SW_DOCK switches and inhibit reaction
3532 to the lid switch if it is pressed. This means that logind
3533 will not suspend the machine anymore if the lid is closed
3534 and the system is docked, if the laptop supports SW_DOCK
3535 notifications via the input layer. Note that ACPI docking
3536 stations do not generate this currently. Also note that this
3537 logic is usually not fully sufficient and Desktop
3538 Environments should take a lid switch inhibitor lock when an
3539 external display is connected, as systemd will not watch
3542 * nspawn will now make use of the devices cgroup controller by
3543 default, and only permit creation of and access to the usual
3544 API device nodes like /dev/null or /dev/random, as well as
3545 access to (but not creation of) the pty devices.
3547 * We will now ship a default .network file for
3548 systemd-networkd that automatically configures DHCP for
3549 network interfaces created by nspawn's --network-veth or
3550 --network-bridge= switches.
3552 * systemd will now understand the usual M, K, G, T suffixes
3553 according to SI conventions (i.e. to the base 1000) when
3554 referring to throughput and hardware metrics. It will stay
3555 with IEC conventions (i.e. to the base 1024) for software
3556 metrics, according to what is customary according to
3557 Wikipedia. We explicitly document which base applies for
3558 each configuration option.
3560 * The DeviceAllow= setting in unit files now supports a syntax
3561 to whitelist an entire group of devices node majors at once,
3562 based on the /proc/devices listing. For example, with the
3563 string "char-pts", it is now possible to whitelist all
3564 current and future pseudo-TTYs at once.
3566 * sd-event learned a new "post" event source. Event sources of
3567 this type are triggered by the dispatching of any event
3568 source of a type that is not "post". This is useful for
3569 implementing clean-up and check event sources that are
3570 triggered by other work being done in the program.
3572 * systemd-networkd is no longer statically enabled, but uses
3573 the usual [Install] sections so that it can be
3574 enabled/disabled using systemctl. It still is enabled by
3577 * When creating a veth interface pair with systemd-nspawn, the
3578 host side will now be prefixed with "vb-" if
3579 --network-bridge= is used, and with "ve-" if --network-veth
3580 is used. This way, it is easy to distinguish these cases on
3581 the host, for example to apply different configuration to
3582 them with systemd-networkd.
3584 * The compatibility libraries for libsystemd-journal.so,
3585 libsystem-id128.so, libsystemd-login.so and
3586 libsystemd-daemon.so do not make use of IFUNC
3587 anymore. Instead, we now build libsystemd.so multiple times
3588 under these alternative names. This means that the footprint
3589 is drastically increased, but given that these are
3590 transitional compatibility libraries, this should not matter
3591 much. This change has been made necessary to support the ARM
3592 platform for these compatibility libraries, as the ARM
3593 toolchain is not really at the same level as the toolchain
3594 for other architectures like x86 and does not support
3595 IFUNC. Please make sure to use --enable-compat-libs only
3596 during a transitional period!
3598 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
3599 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
3600 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
3601 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
3602 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
3603 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
3604 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
3605 Zbigniew Jędrzejewski-Szmek
3607 — Berlin, 2014-02-24
3611 * A new component "systemd-networkd" has been added that can
3612 be used to configure local network interfaces statically or
3613 via DHCP. It is capable of bringing up bridges, VLANs, and
3614 bonding. Currently, no hook-ups for interactive network
3615 configuration are provided. Use this for your initrd,
3616 container, embedded, or server setup if you need a simple,
3617 yet powerful, network configuration solution. This
3618 configuration subsystem is quite nifty, as it allows wildcard
3619 hotplug matching in interfaces. For example, with a single
3620 configuration snippet, you can configure that all Ethernet
3621 interfaces showing up are automatically added to a bridge,
3622 or similar. It supports link-sensing and more.
3624 * A new tool "systemd-socket-proxyd" has been added which can
3625 act as a bidirectional proxy for TCP sockets. This is
3626 useful for adding socket activation support to services that
3627 do not actually support socket activation, including virtual
3628 machines and the like.
3630 * Add a new tool to save/restore rfkill state on
3633 * Save/restore state of keyboard backlights in addition to
3634 display backlights on shutdown/boot.
3636 * udev learned a new SECLABEL{} construct to label device
3637 nodes with a specific security label when they appear. For
3638 now, only SECLABEL{selinux} is supported, but the syntax is
3639 prepared for additional security frameworks.
3641 * udev gained a new scheme to configure link-level attributes
3642 from files in /etc/systemd/network/*.link. These files can
3643 match against MAC address, device path, driver name and type,
3644 and will apply attributes like the naming policy, link speed,
3645 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
3646 address assignment policy (randomized, ...).
3648 * The configuration of network interface naming rules for
3649 "permanent interface names" has changed: a new NamePolicy=
3650 setting in the [Link] section of .link files determines the
3651 priority of possible naming schemes (onboard, slot, MAC,
3652 path). The default value of this setting is determined by
3653 /usr/lib/net/links/99-default.link. Old
3654 80-net-name-slot.rules udev configuration file has been
3655 removed, so local configuration overriding this file should
3656 be adapated to override 99-default.link instead.
3658 * When the User= switch is used in a unit file, also
3659 initialize $SHELL= based on the user database entry.
3661 * systemd no longer depends on libdbus. All communication is
3662 now done with sd-bus, systemd's low-level bus library
3665 * kdbus support has been added to PID 1 itself. When kdbus is
3666 enabled, this causes PID 1 to set up the system bus and
3667 enable support for a new ".busname" unit type that
3668 encapsulates bus name activation on kdbus. It works a little
3669 bit like ".socket" units, except for bus names. A new
3670 generator has been added that converts classic dbus1 service
3671 activation files automatically into native systemd .busname
3674 * sd-bus: add a light-weight vtable implementation that allows
3675 defining objects on the bus with a simple static const
3676 vtable array of its methods, signals and properties.
3678 * systemd will not generate or install static dbus
3679 introspection data anymore to /usr/share/dbus-1/interfaces,
3680 as the precise format of these files is unclear, and
3681 nothing makes use of it.
3683 * A proxy daemon is now provided to proxy clients connecting
3684 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
3685 compatibility with classic D-Bus.
3687 * A bus driver implementation has been added that supports the
3688 classic D-Bus bus driver calls on kdbus, also for
3689 compatibility purposes.
3691 * A new API "sd-event.h" has been added that implements a
3692 minimal event loop API built around epoll. It provides a
3693 couple of features that direct epoll usage is lacking:
3694 prioritization of events, scales to large numbers of timer
3695 events, per-event timer slack (accuracy), system-wide
3696 coalescing of timer events, exit handlers, watchdog
3697 supervision support using systemd's sd_notify() API, child
3700 * A new API "sd-rntl.h" has been added that provides an API
3701 around the route netlink interface of the kernel, similar in
3702 style to "sd-bus.h".
3704 * A new API "sd-dhcp-client.h" has been added that provides a
3705 small DHCPv4 client-side implementation. This is used by
3708 * There is a new kernel command line option
3709 "systemd.restore_state=0|1". When set to "0", none of the
3710 systemd tools will restore saved runtime state to hardware
3711 devices. More specifically, the rfkill and backlight states
3714 * The FsckPassNo= compatibility option in mount/service units
3715 has been removed. The fstab generator will now add the
3716 necessary dependencies automatically, and does not require
3717 PID1's support for that anymore.
3719 * journalctl gained a new switch, --list-boots, that lists
3720 recent boots with their times and boot IDs.
3722 * The various tools like systemctl, loginctl, timedatectl,
3723 busctl, systemd-run, ... have gained a new switch "-M" to
3724 connect to a specific, local OS container (as direct
3725 connection, without requiring SSH). This works on any
3726 container that is registered with machined, such as those
3727 created by libvirt-lxc or nspawn.
3729 * systemd-run and systemd-analyze also gained support for "-H"
3730 to connect to remote hosts via SSH. This is particularly
3731 useful for systemd-run because it enables queuing of jobs
3732 onto remote systems.
3734 * machinectl gained a new command "login" to open a getty
3735 login in any local container. This works with any container
3736 that is registered with machined (such as those created by
3737 libvirt-lxc or nspawn), and which runs systemd inside.
3739 * machinectl gained a new "reboot" command that may be used to
3740 trigger a reboot on a specific container that is registered
3741 with machined. This works on any container that runs an init
3742 system of some kind.
3744 * systemctl gained a new "list-timers" command to print a nice
3745 listing of installed timer units with the times they elapse
3748 * Alternative reboot() parameters may now be specified on the
3749 "systemctl reboot" command line and are passed to the
3750 reboot() system call.
3752 * systemctl gained a new --job-mode= switch to configure the
3753 mode to queue a job with. This is a more generic version of
3754 --fail, --irreversible, and --ignore-dependencies, which are
3755 still available but not advertised anymore.
3757 * /etc/systemd/system.conf gained new settings to configure
3758 various default timeouts of units, as well as the default
3759 start limit interval and burst. These may still be overridden
3762 * PID1 will now export on the bus profile data of the security
3763 policy upload process (such as the SELinux policy upload to
3766 * journald: when forwarding logs to the console, include
3767 timestamps (following the setting in
3768 /sys/module/printk/parameters/time).
3770 * OnCalendar= in timer units now understands the special
3771 strings "yearly" and "annually". (Both are equivalent)
3773 * The accuracy of timer units is now configurable with the new
3774 AccuracySec= setting. It defaults to 1min.
3776 * A new dependency type JoinsNamespaceOf= has been added that
3777 allows running two services within the same /tmp and network
3778 namespace, if PrivateNetwork= or PrivateTmp= are used.
3780 * A new command "cat" has been added to systemctl. It outputs
3781 the original unit file of a unit, and concatenates the
3782 contents of additional "drop-in" unit file snippets, so that
3783 the full configuration is shown.
3785 * systemctl now supports globbing on the various "list-xyz"
3786 commands, like "list-units" or "list-sockets", as well as on
3787 those commands which take multiple unit names.
3789 * journalctl's --unit= switch gained support for globbing.
3791 * All systemd daemons now make use of the watchdog logic so
3792 that systemd automatically notices when they hang.
3794 * If the $container_ttys environment variable is set,
3795 getty-generator will automatically spawn a getty for each
3796 listed tty. This is useful for container managers to request
3797 login gettys to be spawned on as many ttys as needed.
3799 * %h, %s, %U specifier support is not available anymore when
3800 used in unit files for PID 1. This is because NSS calls are
3801 not safe from PID 1. They stay available for --user
3802 instances of systemd, and as special case for the root user.
3804 * loginctl gained a new "--no-legend" switch to turn off output
3807 * The "sd-login.h" API gained three new calls:
3808 sd_session_is_remote(), sd_session_get_remote_user(),
3809 sd_session_get_remote_host() to query information about
3812 * The udev hardware database now also carries vendor/product
3813 information of SDIO devices.
3815 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
3816 determine whether watchdog notifications are requested by
3819 * Socket-activated per-connection services now include a
3820 short description of the connection parameters in the
3823 * tmpfiles gained a new "--boot" option. When this is not used,
3824 only lines where the command character is not suffixed with
3825 "!" are executed. When this option is specified, those
3826 options are executed too. This partitions tmpfiles
3827 directives into those that can be safely executed at any
3828 time, and those which should be run only at boot (for
3829 example, a line that creates /run/nologin).
3831 * A new API "sd-resolve.h" has been added which provides a simple
3832 asynchronous wrapper around glibc NSS host name resolution
3833 calls, such as getaddrinfo(). In contrast to glibc's
3834 getaddrinfo_a(), it does not use signals. In contrast to most
3835 other asynchronous name resolution libraries, this one does
3836 not reimplement DNS, but reuses NSS, so that alternate
3837 host name resolution systems continue to work, such as mDNS,
3838 LDAP, etc. This API is based on libasyncns, but it has been
3839 cleaned up for inclusion in systemd.
3841 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
3842 "sd-daemon.h" are no longer found in individual libraries
3843 libsystemd-journal.so, libsystemd-login.so,
3844 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
3845 merged them into a single library, libsystemd.so, which
3846 provides all symbols. The reason for this is cyclic
3847 dependencies, as these libraries tend to use each other's
3848 symbols. So far, we have managed to workaround that by linking
3849 a copy of a good part of our code into each of these
3850 libraries again and again, which, however, makes certain
3851 things hard to do, like sharing static variables. Also, it
3852 substantially increases footprint. With this change, there
3853 is only one library for the basic APIs systemd
3854 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
3855 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
3856 library as well, however are subject to the --enable-kdbus
3857 switch (see below). Note that "sd-dhcp-client.h" is not part
3858 of this library (this is because it only consumes, never
3859 provides, services of/to other APIs). To make the transition
3860 easy from the separate libraries to the unified one, we
3861 provide the --enable-compat-libs compile-time switch which
3862 will generate stub libraries that are compatible with the
3863 old ones but redirect all calls to the new one.
3865 * All of the kdbus logic and the new APIs "sd-bus.h",
3866 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
3867 and "sd-utf8.h" are compile-time optional via the
3868 "--enable-kdbus" switch, and they are not compiled in by
3869 default. To make use of kdbus, you have to explicitly enable
3870 the switch. Note however, that neither the kernel nor the
3871 userspace API for all of this is considered stable yet. We
3872 want to maintain the freedom to still change the APIs for
3873 now. By specifying this build-time switch, you acknowledge
3874 that you are aware of the instability of the current
3877 * Also, note that while kdbus is pretty much complete,
3878 it lacks one thing: proper policy support. This means you
3879 can build a fully working system with all features; however,
3880 it will be highly insecure. Policy support will be added in
3881 one of the next releases, at the same time that we will
3882 declare the APIs stable.
3884 * When the kernel command line argument "kdbus" is specified,
3885 systemd will automatically load the kdbus.ko kernel module. At
3886 this stage of development, it is only useful for testing kdbus
3887 and should not be used in production. Note: if "--enable-kdbus"
3888 is specified, and the kdbus.ko kernel module is available, and
3889 "kdbus" is added to the kernel command line, the entire system
3890 runs with kdbus instead of dbus-daemon, with the above mentioned
3891 problem of missing the system policy enforcement. Also a future
3892 version of kdbus.ko or a newer systemd will not be compatible with
3893 each other, and will unlikely be able to boot the machine if only
3894 one of them is updated.
3896 * systemctl gained a new "import-environment" command which
3897 uploads the caller's environment (or parts thereof) into the
3898 service manager so that it is inherited by services started
3899 by the manager. This is useful to upload variables like
3900 $DISPLAY into the user service manager.
3902 * A new PrivateDevices= switch has been added to service units
3903 which allows running a service with a namespaced /dev
3904 directory that does not contain any device nodes for
3905 physical devices. More specifically, it only includes devices
3906 such as /dev/null, /dev/urandom, and /dev/zero which are API
3909 * logind has been extended to support behaviour like VT
3910 switching on seats that do not support a VT. This makes
3911 multi-session available on seats that are not the first seat
3912 (seat0), and on systems where kernel support for VTs has
3913 been disabled at compile-time.
3915 * If a process holds a delay lock for system sleep or shutdown
3916 and fails to release it in time, we will now log its
3917 identity. This makes it easier to identify processes that
3918 cause slow suspends or power-offs.
3920 * When parsing /etc/crypttab, support for a new key-slot=
3921 option as supported by Debian is added. It allows indicating
3922 which LUKS slot to use on disk, speeding up key loading.
3924 * The sd_journald_sendv() API call has been checked and
3925 officially declared to be async-signal-safe so that it may
3926 be invoked from signal handlers for logging purposes.
3928 * Boot-time status output is now enabled automatically after a
3929 short timeout if boot does not progress, in order to give
3930 the user an indication what she or he is waiting for.
3932 * The boot-time output has been improved to show how much time
3933 remains until jobs expire.
3935 * The KillMode= switch in service units gained a new possible
3936 value "mixed". If set, and the unit is shut down, then the
3937 initial SIGTERM signal is sent only to the main daemon
3938 process, while the following SIGKILL signal is sent to
3939 all remaining processes of the service.
3941 * When a scope unit is registered, a new property "Controller"
3942 may be set. If set to a valid bus name, systemd will send a
3943 RequestStop() signal to this name when it would like to shut
3944 down the scope. This may be used to hook manager logic into
3945 the shutdown logic of scope units. Also, scope units may now
3946 be put in a special "abandoned" state, in which case the
3947 manager process which created them takes no further
3948 responsibilities for it.
3950 * When reading unit files, systemd will now verify
3951 the access mode of these files, and warn about certain
3952 suspicious combinations. This has been added to make it
3953 easier to track down packaging bugs where unit files are
3954 marked executable or world-writable.
3956 * systemd-nspawn gained a new "--setenv=" switch to set
3957 container-wide environment variables. The similar option in
3958 systemd-activate was renamed from "--environment=" to
3959 "--setenv=" for consistency.
3961 * systemd-nspawn has been updated to create a new kdbus domain
3962 for each container that is invoked, thus allowing each
3963 container to have its own set of system and user buses,
3964 independent of the host.
3966 * systemd-nspawn gained a new --drop-capability= switch to run
3967 the container with less capabilities than the default. Both
3968 --drop-capability= and --capability= now take the special
3969 string "all" for dropping or keeping all capabilities.
3971 * systemd-nspawn gained new switches for executing containers
3972 with specific SELinux labels set.
3974 * systemd-nspawn gained a new --quiet switch to not generate
3975 any additional output but the container's own console
3978 * systemd-nspawn gained a new --share-system switch to run a
3979 container without PID namespacing enabled.
3981 * systemd-nspawn gained a new --register= switch to control
3982 whether the container is registered with systemd-machined or
3983 not. This is useful for containers that do not run full
3984 OS images, but only specific apps.
3986 * systemd-nspawn gained a new --keep-unit which may be used
3987 when invoked as the only program from a service unit, and
3988 results in registration of the unit service itself in
3989 systemd-machined, instead of a newly opened scope unit.
3991 * systemd-nspawn gained a new --network-interface= switch for
3992 moving arbitrary interfaces to the container. The new
3993 --network-veth switch creates a virtual Ethernet connection
3994 between host and container. The new --network-bridge=
3995 switch then allows assigning the host side of this virtual
3996 Ethernet connection to a bridge device.
3998 * systemd-nspawn gained a new --personality= switch for
3999 setting the kernel personality for the container. This is
4000 useful when running a 32-bit container on a 64-bit host. A
4001 similar option Personality= is now also available for service
4004 * logind will now also track a "Desktop" identifier for each
4005 session which encodes the desktop environment of it. This is
4006 useful for desktop environments that want to identify
4007 multiple running sessions of itself easily.
4009 * A new SELinuxContext= setting for service units has been
4010 added that allows setting a specific SELinux execution
4011 context for a service.
4013 * Most systemd client tools will now honour $SYSTEMD_LESS for
4014 settings of the "less" pager. By default, these tools will
4015 override $LESS to allow certain operations to work, such as
4016 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
4017 influence this logic.
4019 * systemd's "seccomp" hook-up has been changed to make use of
4020 the libseccomp library instead of using its own
4021 implementation. This has benefits for portability among
4024 * For usage together with SystemCallFilter=, a new
4025 SystemCallErrorNumber= setting has been introduced that
4026 allows configuration of a system error number to be returned
4027 on filtered system calls, instead of immediately killing the
4028 process. Also, SystemCallArchitectures= has been added to
4029 limit access to system calls of a particular architecture
4030 (in order to turn off support for unused secondary
4031 architectures). There is also a global
4032 SystemCallArchitectures= setting in system.conf now to turn
4033 off support for non-native system calls system-wide.
4035 * systemd requires a kernel with a working name_to_handle_at(),
4036 please see the kernel config requirements in the README file.
4038 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
4039 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
4040 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
4041 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
4042 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
4043 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
4044 Elia Pinto, Florian Weimer, George McCollister, Goffredo
4045 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
4046 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
4047 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
4048 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
4049 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
4050 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
4051 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
4052 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
4053 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
4054 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
4055 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
4056 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
4057 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
4058 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
4059 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
4060 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
4061 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
4063 — Berlin, 2014-02-20
4067 * logind has gained support for facilitating privileged input
4068 and drm device access for unprivileged clients. This work is
4069 useful to allow Wayland display servers (and similar
4070 programs, such as kmscon) to run under the user's ID and
4071 access input and drm devices which are normally
4072 protected. When this is used (and the kernel is new enough)
4073 logind will "mute" IO on the file descriptors passed to
4074 Wayland as long as it is in the background and "unmute" it
4075 if it returns into the foreground. This allows secure
4076 session switching without allowing background sessions to
4077 eavesdrop on input and display data. This also introduces
4078 session switching support if VT support is turned off in the
4079 kernel, and on seats that are not seat0.
4081 * A new kernel command line option luks.options= is understood
4082 now which allows specifying LUKS options for usage for LUKS
4083 encrypted partitions specified with luks.uuid=.
4085 * tmpfiles.d(5) snippets may now use specifier expansion in
4086 path names. More specifically %m, %b, %H, %v, are now
4087 replaced by the local machine id, boot id, hostname, and
4088 kernel version number.
4090 * A new tmpfiles.d(5) command "m" has been introduced which
4091 may be used to change the owner/group/access mode of a file
4092 or directory if it exists, but do nothing if it does not.
4094 * This release removes high-level support for the
4095 MemorySoftLimit= cgroup setting. The underlying kernel
4096 cgroup attribute memory.soft_limit= is currently badly
4097 designed and likely to be removed from the kernel API in its
4098 current form, hence we should not expose it for now.
4100 * The memory.use_hierarchy cgroup attribute is now enabled for
4101 all cgroups systemd creates in the memory cgroup
4102 hierarchy. This option is likely to be come the built-in
4103 default in the kernel anyway, and the non-hierarchical mode
4104 never made much sense in the intrinsically hierarchical
4107 * A new field _SYSTEMD_SLICE= is logged along with all journal
4108 messages containing the slice a message was generated
4109 from. This is useful to allow easy per-customer filtering of
4110 logs among other things.
4112 * systemd-journald will no longer adjust the group of journal
4113 files it creates to the "systemd-journal" group. Instead we
4114 rely on the journal directory to be owned by the
4115 "systemd-journal" group, and its setgid bit set, so that the
4116 kernel file system layer will automatically enforce that
4117 journal files inherit this group assignment. The reason for
4118 this change is that we cannot allow NSS look-ups from
4119 journald which would be necessary to resolve
4120 "systemd-journal" to a numeric GID, because this might
4121 create deadlocks if NSS involves synchronous queries to
4122 other daemons (such as nscd, or sssd) which in turn are
4123 logging clients of journald and might block on it, which
4124 would then dead lock. A tmpfiles.d(5) snippet included in
4125 systemd will make sure the setgid bit and group are
4126 properly set on the journal directory if it exists on every
4127 boot. However, we recommend adjusting it manually after
4128 upgrades too (or from RPM scriptlets), so that the change is
4129 not delayed until next reboot.
4131 * Backlight and random seed files in /var/lib/ have moved into
4132 the /var/lib/systemd/ directory, in order to centralize all
4133 systemd generated files in one directory.
4135 * Boot time performance measurements (as displayed by
4136 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
4137 performance information if that's available to determine how
4138 much time BIOS and boot loader initialization required. With
4139 a sufficiently new BIOS you hence no longer need to boot
4140 with Gummiboot to get access to such information.
4142 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
4143 Cristian Rodríguez, Dave Reisner, David Herrmann, David
4144 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
4145 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
4146 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
4147 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
4148 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
4150 — Berlin, 2013-10-02
4154 * The Restart= option for services now understands a new
4155 on-watchdog setting, which will restart the service
4156 automatically if the service stops sending out watchdog keep
4157 alive messages (as configured with WatchdogSec=).
4159 * The getty generator (which is responsible for bringing up a
4160 getty on configured serial consoles) will no longer only
4161 start a getty on the primary kernel console but on all
4162 others, too. This makes the order in which console= is
4163 specified on the kernel command line less important.
4165 * libsystemd-logind gained a new sd_session_get_vt() call to
4166 retrieve the VT number of a session.
4168 * If the option "tries=0" is set for an entry of /etc/crypttab
4169 its passphrase is queried indefinitely instead of any
4170 maximum number of tries.
4172 * If a service with a configure PID file terminates its PID
4173 file will now be removed automatically if it still exists
4174 afterwards. This should put an end to stale PID files.
4176 * systemd-run will now also take relative binary path names
4177 for execution and no longer insists on absolute paths.
4179 * InaccessibleDirectories= and ReadOnlyDirectories= now take
4180 paths that are optionally prefixed with "-" to indicate that
4181 it should not be considered a failure if they do not exist.
4183 * journalctl -o (and similar commands) now understands a new
4184 output mode "short-precise", it is similar to "short" but
4185 shows timestamps with usec accuracy.
4187 * The option "discard" (as known from Debian) is now
4188 synonymous to "allow-discards" in /etc/crypttab. In fact,
4189 "discard" is preferred now (since it is easier to remember
4192 * Some licensing clean-ups were made, so that more code is now
4193 LGPL-2.1 licensed than before.
4195 * A minimal tool to save/restore the display backlight
4196 brightness across reboots has been added. It will store the
4197 backlight setting as late as possible at shutdown, and
4198 restore it as early as possible during reboot.
4200 * A logic to automatically discover and enable home and swap
4201 partitions on GPT disks has been added. With this in place
4202 /etc/fstab becomes optional for many setups as systemd can
4203 discover certain partitions located on the root disk
4204 automatically. Home partitions are recognized under their
4205 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
4206 partitions are recognized under their GPT type ID
4207 0657fd6da4ab43c484e50933c84b4f4f.
4209 * systemd will no longer pass any environment from the kernel
4210 or initrd to system services. If you want to set an
4211 environment for all services, do so via the kernel command
4212 line systemd.setenv= assignment.
4214 * The systemd-sysctl tool no longer natively reads the file
4215 /etc/sysctl.conf. If desired, the file should be symlinked
4216 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
4217 legacy support by a symlink rather than built-in code, it
4218 also makes the otherwise hidden order of application of the
4219 different files visible. (Note that this partly reverts to a
4220 pre-198 application order of sysctl knobs!)
4222 * The "systemctl set-log-level" and "systemctl dump" commands
4223 have been moved to systemd-analyze.
4225 * systemd-run learned the new --remain-after-exit switch,
4226 which causes the scope unit not to be cleaned up
4227 automatically after the process terminated.
4229 * tmpfiles learned a new --exclude-prefix= switch to exclude
4230 certain paths from operation.
4232 * journald will now automatically flush all messages to disk
4233 as soon as a message at the log level CRIT, ALERT or EMERG
4236 Contributions from: Andrew Cook, Brandon Philips, Christian
4237 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
4238 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
4239 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
4240 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
4241 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
4242 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
4243 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
4244 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
4245 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
4246 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
4247 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
4248 William Giokas, Zbigniew Jędrzejewski-Szmek
4250 — Berlin, 2013-09-13
4254 * The documentation has been updated to cover the various new
4255 concepts introduced with 205.
4257 * Unit files now understand the new %v specifier which
4258 resolves to the kernel version string as returned by "uname
4261 * systemctl now supports filtering the unit list output by
4262 load state, active state and sub state, using the new
4265 * "systemctl status" will now show the results of the
4266 condition checks (like ConditionPathExists= and similar) of
4267 the last start attempts of the unit. They are also logged to
4270 * "journalctl -b" may now be used to look for boot output of a
4271 specific boot. Try "journalctl -b -1" for the previous boot,
4272 but the syntax is substantially more powerful.
4274 * "journalctl --show-cursor" has been added which prints the
4275 cursor string the last shown log line. This may then be used
4276 with the new "journalctl --after-cursor=" switch to continue
4277 browsing logs from that point on.
4279 * "journalctl --force" may now be used to force regeneration
4282 * Creation of "dead" device nodes has been moved from udev
4283 into kmod and tmpfiles. Previously, udev would read the kmod
4284 databases to pre-generate dead device nodes based on meta
4285 information contained in kernel modules, so that these would
4286 be auto-loaded on access rather then at boot. As this
4287 does not really have much to do with the exposing actual
4288 kernel devices to userspace this has always been slightly
4289 alien in the udev codebase. Following the new scheme kmod
4290 will now generate a runtime snippet for tmpfiles from the
4291 module meta information and it now is tmpfiles' job to the
4292 create the nodes. This also allows overriding access and
4293 other parameters for the nodes using the usual tmpfiles
4294 facilities. As side effect this allows us to remove the
4295 CAP_SYS_MKNOD capability bit from udevd entirely.
4297 * logind's device ACLs may now be applied to these "dead"
4298 devices nodes too, thus finally allowing managed access to
4299 devices such as /dev/snd/sequencer whithout loading the
4300 backing module right-away.
4302 * A new RPM macro has been added that may be used to apply
4303 tmpfiles configuration during package installation.
4305 * systemd-detect-virt and ConditionVirtualization= now can
4306 detect User-Mode-Linux machines (UML).
4308 * journald will now implicitly log the effective capabilities
4309 set of processes in the message metadata.
4311 * systemd-cryptsetup has gained support for TrueCrypt volumes.
4313 * The initrd interface has been simplified (more specifically,
4314 support for passing performance data via environment
4315 variables and fsck results via files in /run has been
4316 removed). These features were non-essential, and are
4317 nowadays available in a much nicer way by having systemd in
4318 the initrd serialize its state and have the hosts systemd
4319 deserialize it again.
4321 * The udev "keymap" data files and tools to apply keyboard
4322 specific mappings of scan to key codes, and force-release
4323 scan code lists have been entirely replaced by a udev
4324 "keyboard" builtin and a hwdb data file.
4326 * systemd will now honour the kernel's "quiet" command line
4327 argument also during late shutdown, resulting in a
4328 completely silent shutdown when used.
4330 * There's now an option to control the SO_REUSEPORT socket
4331 option in .socket units.
4333 * Instance units will now automatically get a per-template
4334 subslice of system.slice unless something else is explicitly
4335 configured. For example, instances of sshd@.service will now
4336 implicitly be placed in system-sshd.slice rather than
4337 system.slice as before.
4339 * Test coverage support may now be enabled at build time.
4341 Contributions from: Dave Reisner, Frederic Crozat, Harald
4342 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
4343 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
4344 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
4345 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
4346 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
4347 Giokas, Zbigniew Jędrzejewski-Szmek
4349 — Berlin, 2013-07-23
4353 * Two new unit types have been introduced:
4355 Scope units are very similar to service units, however, are
4356 created out of pre-existing processes — instead of PID 1
4357 forking off the processes. By using scope units it is
4358 possible for system services and applications to group their
4359 own child processes (worker processes) in a powerful way
4360 which then maybe used to organize them, or kill them
4361 together, or apply resource limits on them.
4363 Slice units may be used to partition system resources in an
4364 hierarchical fashion and then assign other units to them. By
4365 default there are now three slices: system.slice (for all
4366 system services), user.slice (for all user sessions),
4367 machine.slice (for VMs and containers).
4369 Slices and scopes have been introduced primarily in
4370 context of the work to move cgroup handling to a
4371 single-writer scheme, where only PID 1
4372 creates/removes/manages cgroups.
4374 * There's a new concept of "transient" units. In contrast to
4375 normal units these units are created via an API at runtime,
4376 not from configuration from disk. More specifically this
4377 means it is now possible to run arbitrary programs as
4378 independent services, with all execution parameters passed
4379 in via bus APIs rather than read from disk. Transient units
4380 make systemd substantially more dynamic then it ever was,
4381 and useful as a general batch manager.
4383 * logind has been updated to make use of scope and slice units
4384 for managing user sessions. As a user logs in he will get
4385 his own private slice unit, to which all sessions are added
4386 as scope units. We also added support for automatically
4387 adding an instance of user@.service for the user into the
4388 slice. Effectively logind will no longer create cgroup
4389 hierarchies on its own now, it will defer entirely to PID 1
4390 for this by means of scope, service and slice units. Since
4391 user sessions this way become entities managed by PID 1
4392 the output of "systemctl" is now a lot more comprehensive.
4394 * A new mini-daemon "systemd-machined" has been added which
4395 may be used by virtualization managers to register local
4396 VMs/containers. nspawn has been updated accordingly, and
4397 libvirt will be updated shortly. machined will collect a bit
4398 of meta information about the VMs/containers, and assign
4399 them their own scope unit (see above). The collected
4400 meta-data is then made available via the "machinectl" tool,
4401 and exposed in "ps" and similar tools. machined/machinectl
4402 is compile-time optional.
4404 * As discussed earlier, the low-level cgroup configuration
4405 options ControlGroup=, ControlGroupModify=,
4406 ControlGroupPersistent=, ControlGroupAttribute= have been
4407 removed. Please use high-level attribute settings instead as
4408 well as slice units.
4410 * A new bus call SetUnitProperties() has been added to alter
4411 various runtime parameters of a unit. This is primarily
4412 useful to alter cgroup parameters dynamically in a nice way,
4413 but will be extended later on to make more properties
4414 modifiable at runtime. systemctl gained a new set-properties
4415 command that wraps this call.
4417 * A new tool "systemd-run" has been added which can be used to
4418 run arbitrary command lines as transient services or scopes,
4419 while configuring a number of settings via the command
4420 line. This tool is currently very basic, however already
4421 very useful. We plan to extend this tool to even allow
4422 queuing of execution jobs with time triggers from the
4423 command line, similar in fashion to "at".
4425 * nspawn will now inform the user explicitly that kernels with
4426 audit enabled break containers, and suggest the user to turn
4429 * Support for detecting the IMA and AppArmor security
4430 frameworks with ConditionSecurity= has been added.
4432 * journalctl gained a new "-k" switch for showing only kernel
4433 messages, mimicking dmesg output; in addition to "--user"
4434 and "--system" switches for showing only user's own logs
4437 * systemd-delta can now show information about drop-in
4438 snippets extending unit files.
4440 * libsystemd-bus has been substantially updated but is still
4441 not available as public API.
4443 * systemd will now look for the "debug" argument on the kernel
4444 command line and enable debug logging, similar to what
4445 "systemd.log_level=debug" already did before.
4447 * "systemctl set-default", "systemctl get-default" has been
4448 added to configure the default.target symlink, which
4449 controls what to boot into by default.
4451 * "systemctl set-log-level" has been added as a convenient
4452 way to raise and lower systemd logging threshold.
4454 * "systemd-analyze plot" will now show the time the various
4455 generators needed for execution, as well as information
4456 about the unit file loading.
4458 * libsystemd-journal gained a new sd_journal_open_files() call
4459 for opening specific journal files. journactl also gained a
4460 new switch to expose this new functionality. Previously we
4461 only supported opening all files from a directory, or all
4462 files from the system, as opening individual files only is
4463 racy due to journal file rotation.
4465 * systemd gained the new DefaultEnvironment= setting in
4466 /etc/systemd/system.conf to set environment variables for
4469 * If a privileged process logs a journal message with the
4470 OBJECT_PID= field set, then journald will automatically
4471 augment this with additional OBJECT_UID=, OBJECT_GID=,
4472 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
4473 system services want to log events about specific client
4474 processes. journactl/systemctl has been updated to make use
4475 of this information if all log messages regarding a specific
4478 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
4479 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
4480 Reisner, David Coppa, David King, David Strauss, Eelco
4481 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
4482 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
4483 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
4484 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
4485 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
4486 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
4487 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
4488 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
4489 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
4490 Łukasz Stelmach, 장동준
4494 * The Python bindings gained some minimal support for the APIs
4495 exposed by libsystemd-logind.
4497 * ConditionSecurity= gained support for detecting SMACK. Since
4498 this condition already supports SELinux and AppArmor we only
4499 miss IMA for this. Patches welcome!
4501 Contributions from: Karol Lewandowski, Lennart Poettering,
4502 Zbigniew Jędrzejewski-Szmek
4506 * systemd-nspawn will now create /etc/resolv.conf if
4507 necessary, before bind-mounting the host's file onto it.
4509 * systemd-nspawn will now store meta information about a
4510 container on the container's cgroup as extended attribute
4511 fields, including the root directory.
4513 * The cgroup hierarchy has been reworked in many ways. All
4514 objects any of the components systemd creates in the cgroup
4515 tree are now suffixed. More specifically, user sessions are
4516 now placed in cgroups suffixed with ".session", users in
4517 cgroups suffixed with ".user", and nspawn containers in
4518 cgroups suffixed with ".nspawn". Furthermore, all cgroup
4519 names are now escaped in a simple scheme to avoid collision
4520 of userspace object names with kernel filenames. This work
4521 is preparation for making these objects relocatable in the
4522 cgroup tree, in order to allow easy resource partitioning of
4523 these objects without causing naming conflicts.
4525 * systemctl list-dependencies gained the new switches
4526 --plain, --reverse, --after and --before.
4528 * systemd-inhibit now shows the process name of processes that
4529 have taken an inhibitor lock.
4531 * nss-myhostname will now also resolve "localhost"
4532 implicitly. This makes /etc/hosts an optional file and
4533 nicely handles that on IPv6 ::1 maps to both "localhost" and
4536 * libsystemd-logind.so gained a new call
4537 sd_get_machine_names() to enumerate running containers and
4538 VMs (currently only supported by very new libvirt and
4539 nspawn). sd_login_monitor can now be used to watch
4540 VMs/containers coming and going.
4542 * .include is not allowed recursively anymore, and only in
4543 unit files. Usually it is better to use drop-in snippets in
4544 .d/*.conf anyway, as introduced with systemd 198.
4546 * systemd-analyze gained a new "critical-chain" command that
4547 determines the slowest chain of units run during system
4548 boot-up. It is very useful for tracking down where
4549 optimizing boot time is the most beneficial.
4551 * systemd will no longer allow manipulating service paths in
4552 the name=systemd:/system cgroup tree using ControlGroup= in
4553 units. (But is still fine with it in all other dirs.)
4555 * There's a new systemd-nspawn@.service service file that may
4556 be used to easily run nspawn containers as system
4557 services. With the container's root directory in
4558 /var/lib/container/foobar it is now sufficient to run
4559 "systemctl start systemd-nspawn@foobar.service" to boot it.
4561 * systemd-cgls gained a new parameter "--machine" to list only
4562 the processes within a certain container.
4564 * ConditionSecurity= now can check for "apparmor". We still
4565 are lacking checks for SMACK and IMA for this condition
4566 check though. Patches welcome!
4568 * A new configuration file /etc/systemd/sleep.conf has been
4569 added that may be used to configure which kernel operation
4570 systemd is supposed to execute when "suspend", "hibernate"
4571 or "hybrid-sleep" is requested. This makes the new kernel
4572 "freeze" state accessible to the user.
4574 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
4575 the passed argument if applicable.
4577 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
4578 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
4579 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
4580 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
4581 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
4582 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
4583 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
4588 * The output of 'systemctl list-jobs' got some polishing. The
4589 '--type=' argument may now be passed more than once. A new
4590 command 'systemctl list-sockets' has been added which shows
4591 a list of kernel sockets systemd is listening on with the
4592 socket units they belong to, plus the units these socket
4595 * The experimental libsystemd-bus library got substantial
4596 updates to work in conjunction with the (also experimental)
4597 kdbus kernel project. It works well enough to exchange
4598 messages with some sophistication. Note that kdbus is not
4599 ready yet, and the library is mostly an elaborate test case
4600 for now, and not installable.
4602 * systemd gained a new unit 'systemd-static-nodes.service'
4603 that generates static device nodes earlier during boot, and
4604 can run in conjunction with udev.
4606 * libsystemd-login gained a new call sd_pid_get_user_unit()
4607 to retrieve the user systemd unit a process is running
4608 in. This is useful for systems where systemd is used as
4611 * systemd-nspawn now places all containers in the new /machine
4612 top-level cgroup directory in the name=systemd
4613 hierarchy. libvirt will soon do the same, so that we get a
4614 uniform separation of /system, /user and /machine for system
4615 services, user processes and containers/virtual
4616 machines. This new cgroup hierarchy is also useful to stick
4617 stable names to specific container instances, which can be
4618 recognized later this way (this name may be controlled
4619 via systemd-nspawn's new -M switch). libsystemd-login also
4620 gained a new call sd_pid_get_machine_name() to retrieve the
4621 name of the container/VM a specific process belongs to.
4623 * bootchart can now store its data in the journal.
4625 * libsystemd-journal gained a new call
4626 sd_journal_add_conjunction() for AND expressions to the
4627 matching logic. This can be used to express more complex
4628 logical expressions.
4630 * journactl can now take multiple --unit= and --user-unit=
4633 * The cryptsetup logic now understands the "luks.key=" kernel
4634 command line switch for specifying a file to read the
4635 decryption key from. Also, if a configured key file is not
4636 found the tool will now automatically fall back to prompting
4639 * Python systemd.journal module was updated to wrap recently
4640 added functions from libsystemd-journal. The interface was
4641 changed to bring the low level interface in s.j._Reader
4642 closer to the C API, and the high level interface in
4643 s.j.Reader was updated to wrap and convert all data about
4646 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
4647 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
4648 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
4649 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
4650 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
4651 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
4655 * journalctl --update-catalog now understands a new --root=
4656 option to operate on catalogs found in a different root
4659 * During shutdown after systemd has terminated all running
4660 services a final killing loop kills all remaining left-over
4661 processes. We will now print the name of these processes
4662 when we send SIGKILL to them, since this usually indicates a
4665 * If /etc/crypttab refers to password files stored on
4666 configured mount points automatic dependencies will now be
4667 generated to ensure the specific mount is established first
4668 before the key file is attempted to be read.
4670 * 'systemctl status' will now show information about the
4671 network sockets a socket unit is listening on.
4673 * 'systemctl status' will also shown information about any
4674 drop-in configuration file for units. (Drop-In configuration
4675 files in this context are files such as
4676 /etc/systemd/systemd/foobar.service.d/*.conf)
4678 * systemd-cgtop now optionally shows summed up CPU times of
4679 cgroups. Press '%' while running cgtop to switch between
4680 percentage and absolute mode. This is useful to determine
4681 which cgroups use up the most CPU time over the entire
4682 runtime of the system. systemd-cgtop has also been updated
4683 to be 'pipeable' for processing with further shell tools.
4685 * 'hostnamectl set-hostname' will now allow setting of FQDN
4688 * The formatting and parsing of time span values has been
4689 changed. The parser now understands fractional expressions
4690 such as "5.5h". The formatter will now output fractional
4691 expressions for all time spans under 1min, i.e. "5.123456s"
4692 rather than "5s 123ms 456us". For time spans under 1s
4693 millisecond values are shown, for those under 1ms
4694 microsecond values are shown. This should greatly improve
4695 all time-related output of systemd.
4697 * libsystemd-login and libsystemd-journal gained new
4698 functions for querying the poll() events mask and poll()
4699 timeout value for integration into arbitrary event
4702 * localectl gained the ability to list available X11 keymaps
4703 (models, layouts, variants, options).
4705 * 'systemd-analyze dot' gained the ability to filter for
4706 specific units via shell-style globs, to create smaller,
4707 more useful graphs. I.e. it is now possible to create simple
4708 graphs of all the dependencies between only target units, or
4709 of all units that Avahi has dependencies with.
4711 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
4712 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
4713 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
4714 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
4715 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
4716 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
4717 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
4721 * The boot-time readahead implementation for rotating media
4722 will now read the read-ahead data in multiple passes which
4723 consist of all read requests made in equidistant time
4724 intervals. This means instead of strictly reading read-ahead
4725 data in its physical order on disk we now try to find a
4726 middle ground between physical and access time order.
4728 * /etc/os-release files gained a new BUILD_ID= field for usage
4729 on operating systems that provide continuous builds of OS
4732 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
4733 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
4734 William Douglas, Zbigniew Jędrzejewski-Szmek
4738 * systemd-python gained an API exposing libsystemd-daemon.
4740 * The SMACK setup logic gained support for uploading CIPSO
4743 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
4744 ReadOnlyDirectories= and InaccessibleDirectories= has
4745 changed. The private /tmp and /var/tmp directories are now
4746 shared by all processes of a service (which means
4747 ExecStartPre= may now leave data in /tmp that ExecStart= of
4748 the same service can still access). When a service is
4749 stopped its temporary directories are immediately deleted
4750 (normal clean-up with tmpfiles is still done in addition to
4753 * By default, systemd will now set a couple of sysctl
4754 variables in the kernel: the safe sysrq options are turned
4755 on, IP route verification is turned on, and source routing
4756 disabled. The recently added hardlink and softlink
4757 protection of the kernel is turned on. These settings should
4758 be reasonably safe, and good defaults for all new systems.
4760 * The predictable network naming logic may now be turned off
4761 with a new kernel command line switch: net.ifnames=0.
4763 * A new libsystemd-bus module has been added that implements a
4764 pretty complete D-Bus client library. For details see:
4766 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
4768 * journald will now explicitly flush the journal files to disk
4769 at the latest 5min after each write. The file will then also
4770 be marked offline until the next write. This should increase
4771 reliability in case of a crash. The synchronization delay
4772 can be configured via SyncIntervalSec= in journald.conf.
4774 * There's a new remote-fs-setup.target unit that can be used
4775 to pull in specific services when at least one remote file
4776 system is to be mounted.
4778 * There are new targets timers.target and paths.target as
4779 canonical targets to pull user timer and path units in
4780 from. This complements sockets.target with a similar
4781 purpose for socket units.
4783 * libudev gained a new call udev_device_set_attribute_value()
4784 to set sysfs attributes of a device.
4786 * The udev daemon now sets the default number of worker
4787 processes executed in parallel based on the number of available
4788 CPUs instead of the amount of available RAM. This is supposed
4789 to provide a more reliable default and limit a too aggressive
4790 paralellism for setups with 1000s of devices connected.
4792 Contributions from: Auke Kok, Colin Walters, Cristian
4793 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
4794 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
4795 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
4796 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
4797 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
4798 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
4799 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
4800 Zbigniew Jędrzejewski-Szmek
4804 * Configuration of unit files may now be extended via drop-in
4805 files without having to edit/override the unit files
4806 themselves. More specifically, if the administrator wants to
4807 change one value for a service file foobar.service he can
4808 now do so by dropping in a configuration snippet into
4809 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
4810 will load all these snippets and apply them on top of the
4811 main unit configuration file, possibly extending or
4812 overriding its settings. Using these drop-in snippets is
4813 generally nicer than the two earlier options for changing
4814 unit files locally: copying the files from
4815 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
4816 them there; or creating a new file in /etc/systemd/system/
4817 that incorporates the original one via ".include". Drop-in
4818 snippets into these .d/ directories can be placed in any
4819 directory systemd looks for units in, and the usual
4820 overriding semantics between /usr/lib, /etc and /run apply
4823 * Most unit file settings which take lists of items can now be
4824 reset by assigning the empty string to them. For example,
4825 normally, settings such as Environment=FOO=BAR append a new
4826 environment variable assignment to the environment block,
4827 each time they are used. By assigning Environment= the empty
4828 string the environment block can be reset to empty. This is
4829 particularly useful with the .d/*.conf drop-in snippets
4830 mentioned above, since this adds the ability to reset list
4831 settings from vendor unit files via these drop-ins.
4833 * systemctl gained a new "list-dependencies" command for
4834 listing the dependencies of a unit recursively.
4836 * Inhibitors are now honored and listed by "systemctl
4837 suspend", "systemctl poweroff" (and similar) too, not only
4838 GNOME. These commands will also list active sessions by
4841 * Resource limits (as exposed by the various control group
4842 controllers) can now be controlled dynamically at runtime
4843 for all units. More specifically, you can now use a command
4844 like "systemctl set-cgroup-attr foobar.service cpu.shares
4845 2000" to alter the CPU shares a specific service gets. These
4846 settings are stored persistently on disk, and thus allow the
4847 administrator to easily adjust the resource usage of
4848 services with a few simple commands. This dynamic resource
4849 management logic is also available to other programs via the
4850 bus. Almost any kernel cgroup attribute and controller is
4853 * systemd-vconsole-setup will now copy all font settings to
4854 all allocated VTs, where it previously applied them only to
4857 * libsystemd-login gained the new sd_session_get_tty() API
4860 * This release drops support for a few legacy or
4861 distribution-specific LSB facility names when parsing init
4862 scripts: $x-display-manager, $mail-transfer-agent,
4863 $mail-transport-agent, $mail-transfer-agent, $smtp,
4864 $null. Also, the mail-transfer-agent.target unit backing
4865 this has been removed. Distributions which want to retain
4866 compatibility with this should carry the burden for
4867 supporting this themselves and patch support for these back
4868 in, if they really need to. Also, the facilities $syslog and
4869 $local_fs are now ignored, since systemd does not support
4870 early-boot LSB init scripts anymore, and these facilities
4871 are implied anyway for normal services. syslog.target has
4874 * There are new bus calls on PID1's Manager object for
4875 cancelling jobs, and removing snapshot units. Previously,
4876 both calls were only available on the Job and Snapshot
4879 * systemd-journal-gatewayd gained SSL support.
4881 * The various "environment" files, such as /etc/locale.conf
4882 now support continuation lines with a backslash ("\") as
4883 last character in the line, similarly in style (but different)
4884 to how this is supported in shells.
4886 * For normal user processes the _SYSTEMD_USER_UNIT= field is
4887 now implicitly appended to every log entry logged. systemctl
4888 has been updated to filter by this field when operating on a
4889 user systemd instance.
4891 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
4892 CAP_AUDIT_CONTROL capabilities to the capabilities set for
4893 the container. This makes it easier to boot unmodified
4894 Fedora systems in a container, which however still requires
4895 audit=0 to be passed on the kernel command line. Auditing in
4896 kernel and userspace is unfortunately still too broken in
4897 context of containers, hence we recommend compiling it out
4898 of the kernel or using audit=0. Hopefully this will be fixed
4899 one day for good in the kernel.
4901 * nspawn gained the new --bind= and --bind-ro= parameters to
4902 bind mount specific directories from the host into the
4905 * nspawn will now mount its own devpts file system instance
4906 into the container, in order not to leak pty devices from
4907 the host into the container.
4909 * systemd will now read the firmware boot time performance
4910 information from the EFI variables, if the used boot loader
4911 supports this, and takes it into account for boot performance
4912 analysis via "systemd-analyze". This is currently supported
4913 only in conjunction with Gummiboot, but could be supported
4914 by other boot loaders too. For details see:
4916 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
4918 * A new generator has been added that automatically mounts the
4919 EFI System Partition (ESP) to /boot, if that directory
4920 exists, is empty, and no other file system has been
4921 configured to be mounted there.
4923 * logind will now send out PrepareForSleep(false) out
4924 unconditionally, after coming back from suspend. This may be
4925 used by applications as asynchronous notification for
4926 system resume events.
4928 * "systemctl unlock-sessions" has been added, that allows
4929 unlocking the screens of all user sessions at once, similar
4930 to how "systemctl lock-sessions" already locked all users
4931 sessions. This is backed by a new D-Bus call UnlockSessions().
4933 * "loginctl seat-status" will now show the master device of a
4934 seat. (i.e. the device of a seat that needs to be around for
4935 the seat to be considered available, usually the graphics
4938 * tmpfiles gained a new "X" line type, that allows
4939 configuration of files and directories (with wildcards) that
4940 shall be excluded from automatic cleanup ("aging").
4942 * udev default rules set the device node permissions now only
4943 at "add" events, and do not change them any longer with a
4944 later "change" event.
4946 * The log messages for lid events and power/sleep keypresses
4947 now carry a message ID.
4949 * We now have a substantially larger unit test suite, but this
4950 continues to be work in progress.
4952 * udevadm hwdb gained a new --root= parameter to change the
4953 root directory to operate relative to.
4955 * logind will now issue a background sync() request to the kernel
4956 early at shutdown, so that dirty buffers are flushed to disk early
4957 instead of at the last moment, in order to optimize shutdown
4960 * A new bootctl tool has been added that is an interface for
4961 certain boot loader operations. This is currently a preview
4962 and is likely to be extended into a small mechanism daemon
4963 like timedated, localed, hostnamed, and can be used by
4964 graphical UIs to enumerate available boot options, and
4965 request boot into firmware operations.
4967 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
4968 the rest of the package. It also has been updated to work
4969 correctly in initrds.
4971 * Policykit previously has been runtime optional, and is now
4972 also compile time optional via a configure switch.
4974 * systemd-analyze has been reimplemented in C. Also "systemctl
4975 dot" has moved into systemd-analyze.
4977 * "systemctl status" with no further parameters will now print
4978 the status of all active or failed units.
4980 * Operations such as "systemctl start" can now be executed
4981 with a new mode "--irreversible" which may be used to queue
4982 operations that cannot accidentally be reversed by a later
4983 job queuing. This is by default used to make shutdown
4984 requests more robust.
4986 * The Python API of systemd now gained a new module for
4987 reading journal files.
4989 * A new tool kernel-install has been added that can install
4990 kernel images according to the Boot Loader Specification:
4992 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
4994 * Boot time console output has been improved to provide
4995 animated boot time output for hanging jobs.
4997 * A new tool systemd-activate has been added which can be used
4998 to test socket activation with, directly from the command
4999 line. This should make it much easier to test and debug
5000 socket activation in daemons.
5002 * journalctl gained a new "--reverse" (or -r) option to show
5003 journal output in reverse order (i.e. newest line first).
5005 * journalctl gained a new "--pager-end" (or -e) option to jump
5006 to immediately jump to the end of the journal in the
5007 pager. This is only supported in conjunction with "less".
5009 * journalctl gained a new "--user-unit=" option, that works
5010 similarly to "--unit=" but filters for user units rather than
5013 * A number of unit files to ease adoption of systemd in
5014 initrds has been added. This moves some minimal logic from
5015 the various initrd implementations into systemd proper.
5017 * The journal files are now owned by a new group
5018 "systemd-journal", which exists specifically to allow access
5019 to the journal, and nothing else. Previously, we used the
5020 "adm" group for that, which however possibly covers more
5021 than just journal/log file access. This new group is now
5022 already used by systemd-journal-gatewayd to ensure this
5023 daemon gets access to the journal files and as little else
5024 as possible. Note that "make install" will also set FS ACLs
5025 up for /var/log/journal to give "adm" and "wheel" read
5026 access to it, in addition to "systemd-journal" which owns
5027 the journal files. We recommend that packaging scripts also
5028 add read access to "adm" + "wheel" to /var/log/journal, and
5029 all existing/future journal files. To normal users and
5030 administrators little changes, however packagers need to
5031 ensure to create the "systemd-journal" system group at
5032 package installation time.
5034 * The systemd-journal-gatewayd now runs as unprivileged user
5035 systemd-journal-gateway:systemd-journal-gateway. Packaging
5036 scripts need to create these system user/group at
5039 * timedated now exposes a new boolean property CanNTP that
5040 indicates whether a local NTP service is available or not.
5042 * systemd-detect-virt will now also detect xen PVs
5044 * The pstore file system is now mounted by default, if it is
5047 * In addition to the SELinux and IMA policies we will now also
5048 load SMACK policies at early boot.
5050 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
5051 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
5052 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
5053 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
5054 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
5055 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
5056 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
5057 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
5058 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
5059 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
5060 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
5061 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
5062 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
5063 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
5067 * Timer units now support calendar time events in addition to
5068 monotonic time events. That means you can now trigger a unit
5069 based on a calendar time specification such as "Thu,Fri
5070 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
5071 or fifth day of any month of the year 2013, given that it is
5072 a thursday or friday. This brings timer event support
5073 considerably closer to cron's capabilities. For details on
5074 the supported calendar time specification language see
5077 * udev now supports a number of different naming policies for
5078 network interfaces for predictable names, and a combination
5079 of these policies is now the default. Please see this wiki
5080 document for details:
5082 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
5084 * Auke Kok's bootchart implementation has been added to the
5085 systemd tree. It is an optional component that can graph the
5086 boot in quite some detail. It is one of the best bootchart
5087 implementations around and minimal in its code and
5090 * nss-myhostname has been integrated into the systemd source
5091 tree. nss-myhostname guarantees that the local hostname
5092 always stays resolvable via NSS. It has been a weak
5093 requirement of systemd-hostnamed since a long time, and
5094 since its code is actually trivial we decided to just
5095 include it in systemd's source tree. It can be turned off
5096 with a configure switch.
5098 * The read-ahead logic is now capable of properly detecting
5099 whether a btrfs file system is on SSD or rotating media, in
5100 order to optimize the read-ahead scheme. Previously, it was
5101 only capable of detecting this on traditional file systems
5104 * In udev, additional device properties are now read from the
5105 IAB in addition to the OUI database. Also, Bluetooth company
5106 identities are attached to the devices as well.
5108 * In service files %U may be used as specifier that is
5109 replaced by the configured user name of the service.
5111 * nspawn may now be invoked without a controlling TTY. This
5112 makes it suitable for invocation as its own service. This
5113 may be used to set up a simple containerized server system
5114 using only core OS tools.
5116 * systemd and nspawn can now accept socket file descriptors
5117 when they are started for socket activation. This enables
5118 implementation of socket activated nspawn
5119 containers. i.e. think about autospawning an entire OS image
5120 when the first SSH or HTTP connection is received. We expect
5121 that similar functionality will also be added to libvirt-lxc
5124 * journalctl will now suppress ANSI color codes when
5125 presenting log data.
5127 * systemctl will no longer show control group information for
5128 a unit if a the control group is empty anyway.
5130 * logind can now automatically suspend/hibernate/shutdown the
5133 * /etc/machine-info and hostnamed now also expose the chassis
5134 type of the system. This can be used to determine whether
5135 the local system is a laptop, desktop, handset or
5136 tablet. This information may either be configured by the
5137 user/vendor or is automatically determined from ACPI and DMI
5138 information if possible.
5140 * A number of PolicyKit actions are now bound together with
5141 "imply" rules. This should simplify creating UIs because
5142 many actions will now authenticate similar ones as well.
5144 * Unit files learnt a new condition ConditionACPower= which
5145 may be used to conditionalize a unit depending on whether an
5146 AC power source is connected or not, of whether the system
5147 is running on battery power.
5149 * systemctl gained a new "is-failed" verb that may be used in
5150 shell scripts and suchlike to check whether a specific unit
5151 is in the "failed" state.
5153 * The EnvironmentFile= setting in unit files now supports file
5154 globbing, and can hence be used to easily read a number of
5155 environment files at once.
5157 * systemd will no longer detect and recognize specific
5158 distributions. All distribution-specific #ifdeffery has been
5159 removed, systemd is now fully generic and
5160 distribution-agnostic. Effectively, not too much is lost as
5161 a lot of the code is still accessible via explicit configure
5162 switches. However, support for some distribution specific
5163 legacy configuration file formats has been dropped. We
5164 recommend distributions to simply adopt the configuration
5165 files everybody else uses now and convert the old
5166 configuration from packaging scripts. Most distributions
5167 already did that. If that's not possible or desirable,
5168 distributions are welcome to forward port the specific
5169 pieces of code locally from the git history.
5171 * When logging a message about a unit systemd will now always
5172 log the unit name in the message meta data.
5174 * localectl will now also discover system locale data that is
5175 not stored in locale archives, but directly unpacked.
5177 * logind will no longer unconditionally use framebuffer
5178 devices as seat masters, i.e. as devices that are required
5179 to be existing before a seat is considered preset. Instead,
5180 it will now look for all devices that are tagged as
5181 "seat-master" in udev. By default, framebuffer devices will
5182 be marked as such, but depending on local systems, other
5183 devices might be marked as well. This may be used to
5184 integrate graphics cards using closed source drivers (such
5185 as NVidia ones) more nicely into logind. Note however, that
5186 we recommend using the open source NVidia drivers instead,
5187 and no udev rules for the closed-source drivers will be
5188 shipped from us upstream.
5190 Contributions from: Adam Williamson, Alessandro Crismani, Auke
5191 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
5192 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
5193 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
5194 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
5195 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
5196 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
5197 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
5198 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
5199 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
5200 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
5201 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
5206 * udev gained support for loading additional device properties
5207 from an indexed database that is keyed by vendor/product IDs
5208 and similar device identifiers. For the beginning this
5209 "hwdb" is populated with data from the well-known PCI and
5210 USB database, but also includes PNP, ACPI and OID data. In
5211 the longer run this indexed database shall grow into
5212 becoming the one central database for non-essential
5213 userspace device metadata. Previously, data from the PCI/USB
5214 database was only attached to select devices, since the
5215 lookup was a relatively expensive operation due to O(n) time
5216 complexity (with n being the number of entries in the
5217 database). Since this is now O(1), we decided to add in this
5218 data for all devices where this is available, by
5219 default. Note that the indexed database needs to be rebuilt
5220 when new data files are installed. To achieve this you need
5221 to update your packaging scripts to invoke "udevadm hwdb
5222 --update" after installation of hwdb data files. For
5223 RPM-based distributions we introduced the new
5224 %udev_hwdb_update macro for this purpose.
5226 * The Journal gained support for the "Message Catalog", an
5227 indexed database to link up additional information with
5228 journal entries. For further details please check:
5230 http://www.freedesktop.org/wiki/Software/systemd/catalog
5232 The indexed message catalog database also needs to be
5233 rebuilt after installation of message catalog files. Use
5234 "journalctl --update-catalog" for this. For RPM-based
5235 distributions we introduced the %journal_catalog_update
5236 macro for this purpose.
5238 * The Python Journal bindings gained support for the standard
5239 Python logging framework.
5241 * The Journal API gained new functions for checking whether
5242 the underlying file system of a journal file is capable of
5243 properly reporting file change notifications, or whether
5244 applications that want to reflect journal changes "live"
5245 need to recheck journal files continuously in appropriate
5248 * It is now possible to set the "age" field for tmpfiles
5249 entries to 0, indicating that files matching this entry
5250 shall always be removed when the directories are cleaned up.
5252 * coredumpctl gained a new "gdb" verb which invokes gdb
5253 right-away on the selected coredump.
5255 * There's now support for "hybrid sleep" on kernels that
5256 support this, in addition to "suspend" and "hibernate". Use
5257 "systemctl hybrid-sleep" to make use of this.
5259 * logind's HandleSuspendKey= setting (and related settings)
5260 now gained support for a new "lock" setting to simply
5261 request the screen lock on all local sessions, instead of
5262 actually executing a suspend or hibernation.
5264 * systemd will now mount the EFI variables file system by
5267 * Socket units now gained support for configuration of the
5268 SMACK security label.
5270 * timedatectl will now output the time of the last and next
5271 daylight saving change.
5273 * We dropped support for various legacy and distro-specific
5274 concepts, such as insserv, early-boot SysV services
5275 (i.e. those for non-standard runlevels such as 'b' or 'S')
5276 or ArchLinux /etc/rc.conf support. We recommend the
5277 distributions who still need support this to either continue
5278 to maintain the necessary patches downstream, or find a
5279 different solution. (Talk to us if you have questions!)
5281 * Various systemd components will now bypass PolicyKit checks
5282 for root and otherwise handle properly if PolicyKit is not
5283 found to be around. This should fix most issues for
5284 PolicyKit-less systems. Quite frankly this should have been
5285 this way since day one. It is absolutely our intention to
5286 make systemd work fine on PolicyKit-less systems, and we
5287 consider it a bug if something does not work as it should if
5288 PolicyKit is not around.
5290 * For embedded systems it is now possible to build udev and
5291 systemd without blkid and/or kmod support.
5293 * "systemctl switch-root" is now capable of switching root
5294 more than once. I.e. in addition to transitions from the
5295 initrd to the host OS it is now possible to transition to
5296 further OS images from the host. This is useful to implement
5297 offline updating tools.
5299 * Various other additions have been made to the RPM macros
5300 shipped with systemd. Use %udev_rules_update() after
5301 installing new udev rules files. %_udevhwdbdir,
5302 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
5303 %_sysctldir are now available which resolve to the right
5304 directories for packages to place various data files in.
5306 * journalctl gained the new --full switch (in addition to
5307 --all, to disable ellipsation for long messages.
5309 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
5310 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
5311 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
5312 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
5313 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
5314 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
5315 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
5316 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
5317 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
5321 * journalctl gained new --since= and --until= switches to
5322 filter by time. It also now supports nice filtering for
5323 units via --unit=/-u.
5325 * Type=oneshot services may use ExecReload= and do the
5328 * The journal daemon now supports time-based rotation and
5329 vacuuming, in addition to the usual disk-space based
5332 * The journal will now index the available field values for
5333 each field name. This enables clients to show pretty drop
5334 downs of available match values when filtering. The bash
5335 completion of journalctl has been updated
5336 accordingly. journalctl gained a new switch -F to list all
5337 values a certain field takes in the journal database.
5339 * More service events are now written as structured messages
5340 to the journal, and made recognizable via message IDs.
5342 * The timedated, localed and hostnamed mini-services which
5343 previously only provided support for changing time, locale
5344 and hostname settings from graphical DEs such as GNOME now
5345 also have a minimal (but very useful) text-based client
5346 utility each. This is probably the nicest way to changing
5347 these settings from the command line now, especially since
5348 it lists available options and is fully integrated with bash
5351 * There's now a new tool "systemd-coredumpctl" to list and
5352 extract coredumps from the journal.
5354 * We now install a README each in /var/log/ and
5355 /etc/rc.d/init.d explaining where the system logs and init
5356 scripts went. This hopefully should help folks who go to
5357 that dirs and look into the otherwise now empty void and
5358 scratch their heads.
5360 * When user-services are invoked (by systemd --user) the
5361 $MANAGERPID env var is set to the PID of systemd.
5363 * SIGRTMIN+24 when sent to a --user instance will now result
5364 in immediate termination of systemd.
5366 * gatewayd received numerous feature additions such as a
5367 "follow" mode, for live syncing and filtering.
5369 * browse.html now allows filtering and showing detailed
5370 information on specific entries. Keyboard navigation and
5371 mouse screen support has been added.
5373 * gatewayd/journalctl now supports HTML5/JSON
5374 Server-Sent-Events as output.
5376 * The SysV init script compatibility logic will now
5377 heuristically determine whether a script supports the
5378 "reload" verb, and only then make this available as
5381 * "systemctl status --follow" has been removed, use "journalctl
5384 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
5385 have been removed since they are hardly useful to be
5388 * And I'd like to take the opportunity to specifically mention
5389 Zbigniew for his great contributions. Zbigniew, you rock!
5391 Contributions from: Andrew Eikum, Christian Hesse, Colin
5392 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
5393 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
5394 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
5395 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
5396 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
5397 Jędrzejewski-Szmek, Сковорода Никита Андреевич
5401 * If /etc/vconsole.conf is non-existent or empty we will no
5402 longer load any console font or key map at boot by
5403 default. Instead the kernel defaults will be left
5404 intact. This is definitely the right thing to do, as no
5405 configuration should mean no configuration, and hard-coding
5406 font names that are different on all archs is probably a bad
5407 idea. Also, the kernel default key map and font should be
5408 good enough for most cases anyway, and mostly identical to
5409 the userspace fonts/key maps we previously overloaded them
5410 with. If distributions want to continue to default to a
5411 non-kernel font or key map they should ship a default
5412 /etc/vconsole.conf with the appropriate contents.
5414 Contributions from: Colin Walters, Daniel J Walsh, Dave
5415 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
5416 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5420 * journalctl gained a new --cursor= switch to show entries
5421 starting from the specified location in the journal.
5423 * We now enforce a size limit on journal entry fields exported
5424 with "-o json" in journalctl. Fields larger than 4K will be
5425 assigned null. This can be turned off with --all.
5427 * An (optional) journal gateway daemon is now available as
5428 "systemd-journal-gatewayd.service". This service provides
5429 access to the journal via HTTP and JSON. This functionality
5430 will be used to implement live log synchronization in both
5431 pull and push modes, but has various other users too, such
5432 as easy log access for debugging of embedded devices. Right
5433 now it is already useful to retrieve the journal via HTTP:
5435 # systemctl start systemd-journal-gatewayd.service
5436 # wget http://localhost:19531/entries
5438 This will download the journal contents in a
5439 /var/log/messages compatible format. The same as JSON:
5441 # curl -H"Accept: application/json" http://localhost:19531/entries
5443 This service is also accessible via a web browser where a
5444 single static HTML5 app is served that uses the JSON logic
5445 to enable the user to do some basic browsing of the
5446 journal. This will be extended later on. Here's an example
5447 screenshot of this app in its current state:
5449 http://0pointer.de/public/journal-gatewayd
5451 Contributions from: Kay Sievers, Lennart Poettering, Robert
5452 Milasan, Tom Gundersen
5456 * The bash completion logic is now available for journalctl
5459 * We do not mount the "cpuset" controller anymore together with
5460 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
5461 started if no parameters are assigned to it. "cpuset" hence
5462 broke code that assumed it could create "cpu" groups and
5465 * journalctl -f will now subscribe to terminal size changes,
5466 and line break accordingly.
5468 Contributions from: Dave Reisner, Kay Sievers, Lennart
5469 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
5473 * nspawn will now create a symlink /etc/localtime in the
5474 container environment, copying the host's timezone
5475 setting. Previously this has been done via a bind mount, but
5476 since symlinks cannot be bind mounted this has now been
5477 changed to create/update the appropriate symlink.
5479 * journalctl -n's line number argument is now optional, and
5480 will default to 10 if omitted.
5482 * journald will now log the maximum size the journal files may
5483 take up on disk. This is particularly useful if the default
5484 built-in logic of determining this parameter from the file
5485 system size is used. Use "systemctl status
5486 systemd-journald.service" to see this information.
5488 * The multi-seat X wrapper tool has been stripped down. As X
5489 is now capable of enumerating graphics devices via udev in a
5490 seat-aware way the wrapper is not strictly necessary
5491 anymore. A stripped down temporary stop-gap is still shipped
5492 until the upstream display managers have been updated to
5493 fully support the new X logic. Expect this wrapper to be
5494 removed entirely in one of the next releases.
5496 * HandleSleepKey= in logind.conf has been split up into
5497 HandleSuspendKey= and HandleHibernateKey=. The old setting
5498 is not available anymore. X11 and the kernel are
5499 distinguishing between these keys and we should too. This
5500 also means the inhibition lock for these keys has been split
5503 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
5504 Poettering, Lukas Nykryn, Václav Pavlín
5508 * Whenever a unit changes state we will now log this to the
5509 journal and show along the unit's own log output in
5512 * ConditionPathIsMountPoint= can now properly detect bind
5513 mount points too. (Previously, a bind mount of one file
5514 system to another place in the same file system could not be
5515 detected as mount, since they shared struct stat's st_dev
5518 * We will now mount the cgroup controllers cpu, cpuacct,
5519 cpuset and the controllers net_cls, net_prio together by
5522 * nspawn containers will now have a virtualized boot
5523 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
5524 over with a randomized ID at container initialization). This
5525 has the effect of making "journalctl -b" do the right thing
5528 * The JSON output journal serialization has been updated not
5529 to generate "endless" list objects anymore, but rather one
5530 JSON object per line. This is more in line how most JSON
5531 parsers expect JSON objects. The new output mode
5532 "json-pretty" has been added to provide similar output, but
5533 neatly aligned for readability by humans.
5535 * We dropped all explicit sync() invocations in the shutdown
5536 code. The kernel does this implicitly anyway in the kernel
5537 reboot() syscall. halt(8)'s -n option is now a compatibility
5540 * We now support virtualized reboot() in containers, as
5541 supported by newer kernels. We will fall back to exit() if
5542 CAP_SYS_REBOOT is not available to the container. Also,
5543 nspawn makes use of this now and will actually reboot the
5544 container if the containerized OS asks for that.
5546 * journalctl will only show local log output by default
5547 now. Use --merge (-m) to show remote log output, too.
5549 * libsystemd-journal gained the new sd_journal_get_usage()
5550 call to determine the current disk usage of all journal
5551 files. This is exposed in the new "journalctl --disk-usage"
5554 * journald gained a new configuration setting SplitMode= in
5555 journald.conf which may be used to control how user journals
5556 are split off. See journald.conf(5) for details.
5558 * A new condition type ConditionFileNotEmpty= has been added.
5560 * tmpfiles' "w" lines now support file globbing, to write
5561 multiple files at once.
5563 * We added Python bindings for the journal submission
5564 APIs. More Python APIs for a number of selected APIs will
5565 likely follow. Note that we intend to add native bindings
5566 only for the Python language, as we consider it common
5567 enough to deserve bindings shipped within systemd. There are
5568 various projects outside of systemd that provide bindings
5569 for languages such as PHP or Lua.
5571 * Many conditions will now resolve specifiers such as %i. In
5572 addition, PathChanged= and related directives of .path units
5573 now support specifiers as well.
5575 * There's now a new RPM macro definition for the system preset
5578 * journald will now warn if it ca not forward a message to the
5579 syslog daemon because its socket is full.
5581 * timedated will no longer write or process /etc/timezone,
5582 except on Debian. As we do not support late mounted /usr
5583 anymore /etc/localtime always being a symlink is now safe,
5584 and hence the information in /etc/timezone is not necessary
5587 * logind will now always reserve one VT for a text getty (VT6
5588 by default). Previously if more than 6 X sessions where
5589 started they took up all the VTs with auto-spawned gettys,
5590 so that no text gettys were available anymore.
5592 * udev will now automatically inform the btrfs kernel logic
5593 about btrfs RAID components showing up. This should make
5594 simple hotplug based btrfs RAID assembly work.
5596 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
5597 (but not for its children which will stay at the kernel
5598 default). This should allow setups with a lot more listening
5601 * systemd will now always pass the configured timezone to the
5602 kernel at boot. timedated will do the same when the timezone
5605 * logind's inhibition logic has been updated. By default,
5606 logind will now handle the lid switch, the power and sleep
5607 keys all the time, even in graphical sessions. If DEs want
5608 to handle these events on their own they should take the new
5609 handle-power-key, handle-sleep-key and handle-lid-switch
5610 inhibitors during their runtime. A simple way to achieve
5611 that is to invoke the DE wrapped in an invocation of:
5613 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
5615 * Access to unit operations is now checked via SELinux taking
5616 the unit file label and client process label into account.
5618 * systemd will now notify the administrator in the journal
5619 when he over-mounts a non-empty directory.
5621 * There are new specifiers that are resolved in unit files,
5622 for the host name (%H), the machine ID (%m) and the boot ID
5625 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
5626 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
5627 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
5628 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
5629 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
5630 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
5631 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
5635 * Support for reading structured kernel messages from
5636 /dev/kmsg has now been added and is enabled by default.
5638 * Support for reading kernel messages from /proc/kmsg has now
5639 been removed. If you want kernel messages in the journal
5640 make sure to run a recent kernel (>= 3.5) that supports
5641 reading structured messages from /dev/kmsg (see
5642 above). /proc/kmsg is now exclusive property of classic
5643 syslog daemons again.
5645 * The libudev API gained the new
5646 udev_device_new_from_device_id() call.
5648 * The logic for file system namespace (ReadOnlyDirectory=,
5649 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
5650 require pivot_root() anymore. This means fewer temporary
5651 directories are created below /tmp for this feature.
5653 * nspawn containers will now see and receive all submounts
5654 made on the host OS below the root file system of the
5657 * Forward Secure Sealing is now supported for Journal files,
5658 which provide cryptographical sealing of journal files so
5659 that attackers cannot alter log history anymore without this
5660 being detectable. Lennart will soon post a blog story about
5661 this explaining it in more detail.
5663 * There are two new service settings RestartPreventExitStatus=
5664 and SuccessExitStatus= which allow configuration of exit
5665 status (exit code or signal) which will be excepted from the
5666 restart logic, resp. consider successful.
5668 * journalctl gained the new --verify switch that can be used
5669 to check the integrity of the structure of journal files and
5670 (if Forward Secure Sealing is enabled) the contents of
5673 * nspawn containers will now be run with /dev/stdin, /dev/fd/
5674 and similar symlinks pre-created. This makes running shells
5675 as container init process a lot more fun.
5677 * The fstab support can now handle PARTUUID= and PARTLABEL=
5680 * A new ConditionHost= condition has been added to match
5681 against the hostname (with globs) and machine ID. This is
5682 useful for clusters where a single OS image is used to
5683 provision a large number of hosts which shall run slightly
5684 different sets of services.
5686 * Services which hit the restart limit will now be placed in a
5689 Contributions from: Bertram Poettering, Dave Reisner, Huang
5690 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
5691 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
5695 * When running in --user mode systemd will now become a
5696 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
5697 tree a lot more organized.
5699 * A new PartOf= unit dependency type has been introduced that
5700 may be used to group services in a natural way.
5702 * "systemctl enable" may now be used to enable instances of
5705 * journalctl now prints error log levels in red, and
5706 warning/notice log levels in bright white. It also supports
5707 filtering by log level now.
5709 * cgtop gained a new -n switch (similar to top), to configure
5710 the maximum number of iterations to run for. It also gained
5711 -b, to run in batch mode (accepting no input).
5713 * The suffix ".service" may now be omitted on most systemctl
5714 command lines involving service unit names.
5716 * There's a new bus call in logind to lock all sessions, as
5717 well as a loginctl verb for it "lock-sessions".
5719 * libsystemd-logind.so gained a new call sd_journal_perror()
5720 that works similar to libc perror() but logs to the journal
5721 and encodes structured information about the error number.
5723 * /etc/crypttab entries now understand the new keyfile-size=
5726 * shutdown(8) now can send a (configurable) wall message when
5727 a shutdown is cancelled.
5729 * The mount propagation mode for the root file system will now
5730 default to "shared", which is useful to make containers work
5731 nicely out-of-the-box so that they receive new mounts from
5732 the host. This can be undone locally by running "mount
5733 --make-rprivate /" if needed.
5735 * The prefdm.service file has been removed. Distributions
5736 should maintain this unit downstream if they intend to keep
5737 it around. However, we recommend writing normal unit files
5738 for display managers instead.
5740 * Since systemd is a crucial part of the OS we will now
5741 default to a number of compiler switches that improve
5742 security (hardening) such as read-only relocations, stack
5743 protection, and suchlike.
5745 * The TimeoutSec= setting for services is now split into
5746 TimeoutStartSec= and TimeoutStopSec= to allow configuration
5747 of individual time outs for the start and the stop phase of
5750 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
5751 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
5752 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
5753 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
5754 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
5755 Gundersen, Zbigniew Jędrzejewski-Szmek
5759 * The journal and id128 C APIs are now fully documented as man
5762 * Extra safety checks have been added when transitioning from
5763 the initial RAM disk to the main system to avoid accidental
5766 * /etc/crypttab entries now understand the new keyfile-offset=
5769 * systemctl -t can now be used to filter by unit load state.
5771 * The journal C API gained the new sd_journal_wait() call to
5772 make writing synchronous journal clients easier.
5774 * journalctl gained the new -D switch to show journals from a
5777 * journalctl now displays a special marker between log
5778 messages of two different boots.
5780 * The journal is now explicitly flushed to /var via a service
5781 systemd-journal-flush.service, rather than implicitly simply
5782 by seeing /var/log/journal to be writable.
5784 * journalctl (and the journal C APIs) can now match for much
5785 more complex expressions, with alternatives and
5788 * When transitioning from the initial RAM disk to the main
5789 system we will now kill all processes in a killing spree to
5790 ensure no processes stay around by accident.
5792 * Three new specifiers may be used in unit files: %u, %h, %s
5793 resolve to the user name, user home directory resp. user
5794 shell. This is useful for running systemd user instances.
5796 * We now automatically rotate journal files if their data
5797 object hash table gets a fill level > 75%. We also size the
5798 hash table based on the configured maximum file size. This
5799 together should lower hash collisions drastically and thus
5800 speed things up a bit.
5802 * journalctl gained the new "--header" switch to introspect
5803 header data of journal files.
5805 * A new setting SystemCallFilters= has been added to services
5806 which may be used to apply blacklists or whitelists to
5807 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
5809 * nspawn gained a new --link-journal= switch (and quicker: -j)
5810 to link the container journal with the host. This makes it
5811 very easy to centralize log viewing on the host for all
5812 guests while still keeping the journal files separated.
5814 * Many bugfixes and optimizations
5816 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
5817 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
5818 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
5823 * Several tools now understand kernel command line arguments,
5824 which are only read when run in an initial RAM disk. They
5825 usually follow closely their normal counterparts, but are
5828 * There's a new tool to analyze the readahead files that are
5829 automatically generated at boot. Use:
5831 /usr/lib/systemd/systemd-readahead analyze /.readahead
5833 * We now provide an early debug shell on tty9 if this enabled. Use:
5835 systemctl enable debug-shell.service
5837 * All plymouth related units have been moved into the Plymouth
5838 package. Please make sure to upgrade your Plymouth version
5841 * systemd-tmpfiles now supports getting passed the basename of
5842 a configuration file only, in which case it will look for it
5843 in all appropriate directories automatically.
5845 * udevadm info now takes a /dev or /sys path as argument, and
5846 does the right thing. Example:
5848 udevadm info /dev/sda
5849 udevadm info /sys/class/block/sda
5851 * systemctl now prints a warning if a unit is stopped but a
5852 unit that might trigger it continues to run. Example: a
5853 service is stopped but the socket that activates it is left
5856 * "systemctl status" will now mention if the log output was
5857 shortened due to rotation since a service has been started.
5859 * The journal API now exposes functions to determine the
5860 "cutoff" times due to rotation.
5862 * journald now understands SIGUSR1 and SIGUSR2 for triggering
5863 immediately flushing of runtime logs to /var if possible,
5864 resp. for triggering immediate rotation of the journal
5867 * It is now considered an error if a service is attempted to
5868 be stopped that is not loaded.
5870 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
5872 * systemd-analyze now supports Python 3
5874 * tmpfiles now supports cleaning up directories via aging
5875 where the first level dirs are always kept around but
5876 directories beneath it automatically aged. This is enabled
5877 by prefixing the age field with '~'.
5879 * Seat objects now expose CanGraphical, CanTTY properties
5880 which is required to deal with very fast bootups where the
5881 display manager might be running before the graphics drivers
5882 completed initialization.
5884 * Seat objects now expose a State property.
5886 * We now include RPM macros for service enabling/disabling
5887 based on the preset logic. We recommend RPM based
5888 distributions to make use of these macros if possible. This
5889 makes it simpler to reuse RPM spec files across
5892 * We now make sure that the collected systemd unit name is
5893 always valid when services log to the journal via
5896 * There's a new man page kernel-command-line(7) detailing all
5897 command line options we understand.
5899 * The fstab generator may now be disabled at boot by passing
5900 fstab=0 on the kernel command line.
5902 * A new kernel command line option modules-load= is now understood
5903 to load a specific kernel module statically, early at boot.
5905 * Unit names specified on the systemctl command line are now
5906 automatically escaped as needed. Also, if file system or
5907 device paths are specified they are automatically turned
5908 into the appropriate mount or device unit names. Example:
5910 systemctl status /home
5911 systemctl status /dev/sda
5913 * The SysVConsole= configuration option has been removed from
5914 system.conf parsing.
5916 * The SysV search path is no longer exported on the D-Bus
5919 * The Names= option is been removed from unit file parsing.
5921 * There's a new man page bootup(7) detailing the boot process.
5923 * Every unit and every generator we ship with systemd now
5924 comes with full documentation. The self-explanatory boot is
5927 * A couple of services gained "systemd-" prefixes in their
5928 name if they wrap systemd code, rather than only external
5929 code. Among them fsck@.service which is now
5930 systemd-fsck@.service.
5932 * The HaveWatchdog property has been removed from the D-Bus
5935 * systemd.confirm_spawn= on the kernel command line should now
5938 * There's a new man page crypttab(5) which details all options
5939 we actually understand.
5941 * systemd-nspawn gained a new --capability= switch to pass
5942 additional capabilities to the container.
5944 * timedated will now read known NTP implementation unit names
5945 from /usr/lib/systemd/ntp-units.d/*.list,
5946 systemd-timedated-ntp.target has been removed.
5948 * journalctl gained a new switch "-b" that lists log data of
5949 the current boot only.
5951 * The notify socket is in the abstract namespace again, in
5952 order to support daemons which chroot() at start-up.
5954 * There is a new Storage= configuration option for journald
5955 which allows configuration of where log data should go. This
5956 also provides a way to disable journal logging entirely, so
5957 that data collected is only forwarded to the console, the
5958 kernel log buffer or another syslog implementation.
5960 * Many bugfixes and optimizations
5962 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
5963 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
5964 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
5965 Shawn Landden, Tom Gundersen
5969 * "systemctl help <unit>" now shows the man page if one is
5972 * Several new man pages have been added.
5974 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
5975 MaxLevelConsole= can now be specified in
5976 journald.conf. These options allow reducing the amount of
5977 data stored on disk or forwarded by the log level.
5979 * TimerSlackNSec= can now be specified in system.conf for
5980 PID1. This allows system-wide power savings.
5982 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
5983 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
5988 * logind is now capable of (optionally) handling power and
5989 sleep keys as well as the lid switch.
5991 * journalctl now understands the syntax "journalctl
5992 /usr/bin/avahi-daemon" to get all log output of a specific
5995 * CapabilityBoundingSet= in system.conf now also influences
5996 the capability bound set of usermode helpers of the kernel.
5998 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
5999 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
6000 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
6001 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
6005 * Note that we skipped 139 releases here in order to set the
6006 new version to something that is greater than both udev's
6007 and systemd's most recent version number.
6009 * udev: all udev sources are merged into the systemd source tree now.
6010 All future udev development will happen in the systemd tree. It
6011 is still fully supported to use the udev daemon and tools without
6012 systemd running, like in initramfs or other init systems. Building
6013 udev though, will require the *build* of the systemd tree, but
6014 udev can be properly *run* without systemd.
6016 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
6017 should be used to create dead device nodes as workarounds for broken
6020 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
6021 no longer supported. udev_monitor_new_from_netlink() needs to be
6022 used to subscribe to events.
6024 * udev: when udevd is started by systemd, processes which are left
6025 behind by forking them off of udev rules, are unconditionally cleaned
6026 up and killed now after the event handling has finished. Services or
6027 daemons must be started as systemd services. Services can be
6028 pulled-in by udev to get started, but they can no longer be directly
6029 forked by udev rules.
6031 * udev: the daemon binary is called systemd-udevd now and installed
6032 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
6033 to adapt to that, create symlink, or rename the binary after building
6036 * libudev no longer provides these symbols:
6037 udev_monitor_from_socket()
6038 udev_queue_get_failed_list_entry()
6039 udev_get_{dev,sys,run}_path()
6040 The versions number was bumped and symbol versioning introduced.
6042 * systemd-loginctl and systemd-journalctl have been renamed
6043 to loginctl and journalctl to match systemctl.
6045 * The config files: /etc/systemd/systemd-logind.conf and
6046 /etc/systemd/systemd-journald.conf have been renamed to
6047 logind.conf and journald.conf. Package updates should rename
6048 the files to the new names on upgrade.
6050 * For almost all files the license is now LGPL2.1+, changed
6051 from the previous GPL2.0+. Exceptions are some minor stuff
6052 of udev (which will be changed to LGPL2.1 eventually, too),
6053 and the MIT licensed sd-daemon.[ch] library that is suitable
6054 to be used as drop-in files.
6056 * systemd and logind now handle system sleep states, in
6057 particular suspending and hibernating.
6059 * logind now implements a sleep/shutdown/idle inhibiting logic
6060 suitable for a variety of uses. Soonishly Lennart will blog
6061 about this in more detail.
6063 * var-run.mount and var-lock.mount are no longer provided
6064 (which prevously bind mounted these directories to their new
6065 places). Distributions which have not converted these
6066 directories to symlinks should consider stealing these files
6067 from git history and add them downstream.
6069 * We introduced the Documentation= field for units and added
6070 this to all our shipped units. This is useful to make it
6071 easier to explore the boot and the purpose of the various
6074 * All smaller setup units (such as
6075 systemd-vconsole-setup.service) now detect properly if they
6076 are run in a container and are skipped when
6077 appropriate. This guarantees an entirely noise-free boot in
6078 Linux container environments such as systemd-nspawn.
6080 * A framework for implementing offline system updates is now
6081 integrated, for details see:
6082 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
6084 * A new service type Type=idle is available now which helps us
6085 avoiding ugly interleaving of getty output and boot status
6088 * There's now a system-wide CapabilityBoundingSet= option to
6089 globally reduce the set of capabilities for the
6090 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
6091 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
6092 even CAP_NET_ADMIN system-wide for secure systems.
6094 * There are now system-wide DefaultLimitXXX= options to
6095 globally change the defaults of the various resource limits
6096 for all units started by PID 1.
6098 * Harald Hoyer's systemd test suite has been integrated into
6099 systemd which allows easy testing of systemd builds in qemu
6100 and nspawn. (This is really awesome! Ask us for details!)
6102 * The fstab parser is now implemented as generator, not inside
6105 * systemctl will now warn you if .mount units generated from
6106 /etc/fstab are out of date due to changes in fstab that
6107 have not been read by systemd yet.
6109 * systemd is now suitable for usage in initrds. Dracut has
6110 already been updated to make use of this. With this in place
6111 initrds get a slight bit faster but primarily are much
6112 easier to introspect and debug since "systemctl status" in
6113 the host system can be used to introspect initrd services,
6114 and the journal from the initrd is kept around too.
6116 * systemd-delta has been added, a tool to explore differences
6117 between user/admin configuration and vendor defaults.
6119 * PrivateTmp= now affects both /tmp and /var/tmp.
6121 * Boot time status messages are now much prettier and feature
6122 proper english language. Booting up systemd has never been
6125 * Read-ahead pack files now include the inode number of all
6126 files to pre-cache. When the inode changes the pre-caching
6127 is not attempted. This should be nicer to deal with updated
6128 packages which might result in changes of read-ahead
6131 * We now temporaritly lower the kernel's read_ahead_kb variable
6132 when collecting read-ahead data to ensure the kernel's
6133 built-in read-ahead does not add noise to our measurements
6134 of necessary blocks to pre-cache.
6136 * There's now RequiresMountsFor= to add automatic dependencies
6137 for all mounts necessary for a specific file system path.
6139 * MountAuto= and SwapAuto= have been removed from
6140 system.conf. Mounting file systems at boot has to take place
6143 * nspawn now learned a new switch --uuid= to set the machine
6144 ID on the command line.
6146 * nspawn now learned the -b switch to automatically search
6149 * vt102 is now the default TERM for serial TTYs, upgraded from
6152 * systemd-logind now works on VT-less systems.
6154 * The build tree has been reorganized. The individual
6155 components now have directories of their own.
6157 * A new condition type ConditionPathIsReadWrite= is now available.
6159 * nspawn learned the new -C switch to create cgroups for the
6160 container in other hierarchies.
6162 * We now have support for hardware watchdogs, configurable in
6165 * The scheduled shutdown logic now has a public API.
6167 * We now mount /tmp as tmpfs by default, but this can be
6168 masked and /etc/fstab can override it.
6170 * Since udisks does not make use of /media anymore we are not
6171 mounting a tmpfs on it anymore.
6173 * journalctl gained a new --local switch to only interleave
6174 locally generated journal files.
6176 * We can now load the IMA policy at boot automatically.
6178 * The GTK tools have been split off into a systemd-ui.
6180 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
6181 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
6182 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
6183 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
6184 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
6185 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
6186 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
6187 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
6188 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
6193 * This is mostly a bugfix release
6195 * Support optional initialization of the machine ID from the
6196 KVM or container configured UUID.
6198 * Support immediate reboots with "systemctl reboot -ff"
6200 * Show /etc/os-release data in systemd-analyze output
6202 * Many bugfixes for the journal, including endianness fixes and
6203 ensuring that disk space enforcement works
6205 * sd-login.h is C++ comptaible again
6207 * Extend the /etc/os-release format on request of the Debian
6210 * We now refuse non-UTF8 strings used in various configuration
6211 and unit files. This is done to ensure we do not pass invalid
6212 data over D-Bus or expose it elsewhere.
6214 * Register Mimo USB Screens as suitable for automatic seat
6217 * Read SELinux client context from journal clients in a race
6220 * Reorder configuration file lookup order. /etc now always
6221 overrides /run in order to allow the administrator to always
6222 and unconditionally override vendor-supplied or
6223 automatically generated data.
6225 * The various user visible bits of the journal now have man
6226 pages. We still lack man pages for the journal API calls
6229 * We now ship all man pages in HTML format again in the
6232 Contributions from: Dave Reisner, Dirk Eibach, Frederic
6233 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
6234 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
6239 * This is mostly a bugfix release
6241 * systems lacking /etc/os-release are no longer supported.
6243 * Various functionality updates to libsystemd-login.so
6245 * Track class of PAM logins to distinguish greeters from
6248 Contributions from: Kay Sievers, Lennart Poettering, Michael
6253 * This is an important bugfix release for v41.
6255 * Building man pages is now optional which should be useful
6256 for those building systemd from git but unwilling to install
6259 * Watchdog support for supervising services is now usable. In
6260 a future release support for hardware watchdogs
6261 (i.e. /dev/watchdog) will be added building on this.
6263 * Service start rate limiting is now configurable and can be
6264 turned off per service. When a start rate limit is hit a
6265 reboot can automatically be triggered.
6267 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
6269 Contributions from: Benjamin Franzke, Bill Nottingham,
6270 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
6271 Schmidt, Michał Górny, Piotr Drąg
6275 * The systemd binary is installed /usr/lib/systemd/systemd now;
6276 An existing /sbin/init symlink needs to be adapted with the
6279 * The code that loads kernel modules has been ported to invoke
6280 libkmod directly, instead of modprobe. This means we do not
6281 support systems with module-init-tools anymore.
6283 * Watchdog support is now already useful, but still not
6286 * A new kernel command line option systemd.setenv= is
6287 understood to set system wide environment variables
6288 dynamically at boot.
6290 * We now limit the set of capabilities of systemd-journald.
6292 * We now set SIGPIPE to ignore by default, since it only is
6293 useful in shell pipelines, and has little use in general
6294 code. This can be disabled with IgnoreSIPIPE=no in unit
6297 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
6298 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
6303 * This is mostly a bugfix release
6305 * We now expose the reason why a service failed in the
6306 "Result" D-Bus property.
6308 * Rudimentary service watchdog support (will be completed over
6309 the next few releases.)
6311 * When systemd forks off in order execute some service we will
6312 now immediately changes its argv[0] to reflect which process
6313 it will execute. This is useful to minimize the time window
6314 with a generic argv[0], which makes bootcharts more useful
6316 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
6317 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
6318 Mike Kazantsev, Ray Strode
6322 * This is mostly a test release, but incorporates many
6325 * New systemd-cgtop tool to show control groups by their
6328 * Linking against libacl for ACLs is optional again. If
6329 disabled, support tracking device access for active logins
6330 goes becomes unavailable, and so does access to the user
6331 journals by the respective users.
6333 * If a group "adm" exists, journal files are automatically
6334 owned by them, thus allow members of this group full access
6335 to the system journal as well as all user journals.
6337 * The journal now stores the SELinux context of the logging
6338 client for all entries.
6340 * Add C++ inclusion guards to all public headers
6342 * New output mode "cat" in the journal to print only text
6343 messages, without any meta data like date or time.
6345 * Include tiny X server wrapper as a temporary stop-gap to
6346 teach XOrg udev display enumeration. This is used by display
6347 managers such as gdm, and will go away as soon as XOrg
6348 learned native udev hotplugging for display devices.
6350 * Add new systemd-cat tool for executing arbitrary programs
6351 with STDERR/STDOUT connected to the journal. Can also act as
6352 BSD logger replacement, and does so by default.
6354 * Optionally store all locally generated coredumps in the
6355 journal along with meta data.
6357 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
6358 writing short strings to files (for usage for /sys), and for
6359 creating symlinks, character and block device nodes.
6361 * New unit file option ControlGroupPersistent= to make cgroups
6362 persistent, following the mechanisms outlined in
6363 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
6365 * Support multiple local RTCs in a sane way
6367 * No longer monopolize IO when replaying readahead data on
6368 rotating disks, since we might starve non-file-system IO to
6369 death, since fanotify() will not see accesses done by blkid,
6372 * Do not show kernel threads in systemd-cgls anymore, unless
6373 requested with new -k switch.
6375 Contributions from: Dan Horák, Kay Sievers, Lennart
6376 Poettering, Michal Schmidt
6380 * This is mostly a test release, but incorporates many
6383 * The git repository moved to:
6384 git://anongit.freedesktop.org/systemd/systemd
6385 ssh://git.freedesktop.org/git/systemd/systemd
6387 * First release with the journal
6388 http://0pointer.de/blog/projects/the-journal.html
6390 * The journal replaces both systemd-kmsg-syslogd and
6391 systemd-stdout-bridge.
6393 * New sd_pid_get_unit() API call in libsystemd-logind
6395 * Many systemadm clean-ups
6397 * Introduce remote-fs-pre.target which is ordered before all
6398 remote mounts and may be used to start services before all
6401 * Added Mageia support
6403 * Add bash completion for systemd-loginctl
6405 * Actively monitor PID file creation for daemons which exit in
6406 the parent process before having finished writing the PID
6407 file in the daemon process. Daemons which do this need to be
6408 fixed (i.e. PID file creation must have finished before the
6409 parent exits), but we now react a bit more gracefully to them.
6411 * Add colourful boot output, mimicking the well-known output
6412 of existing distributions.
6414 * New option PassCredentials= for socket units, for
6415 compatibility with a recent kernel ABI breakage.
6417 * /etc/rc.local is now hooked in via a generator binary, and
6418 thus will no longer act as synchronization point during
6421 * systemctl list-unit-files now supports --root=.
6423 * systemd-tmpfiles now understands two new commands: z, Z for
6424 relabelling files according to the SELinux database. This is
6425 useful to apply SELinux labels to specific files in /sys,
6428 * Output of SysV services is now forwarded to both the console
6429 and the journal by default, not only just the console.
6431 * New man pages for all APIs from libsystemd-login.
6433 * The build tree got reorganized and a the build system is a
6434 lot more modular allowing embedded setups to specifically
6435 select the components of systemd they are interested in.
6437 * Support for Linux systems lacking the kernel VT subsystem is
6440 * configure's --with-rootdir= got renamed to
6441 --with-rootprefix= to follow the naming used by udev and
6444 * Unless specified otherwise we will now install to /usr instead
6445 of /usr/local by default.
6447 * Processes with '@' in argv[0][0] are now excluded from the
6448 final shut-down killing spree, following the logic explained
6450 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
6452 * All processes remaining in a service cgroup when we enter
6453 the START or START_PRE states are now killed with
6454 SIGKILL. That means it is no longer possible to spawn
6455 background processes from ExecStart= lines (which was never
6456 supported anyway, and bad style).
6458 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
6459 reloading of units together.
6461 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
6462 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
6463 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
6464 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
6465 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek