1 systemd System and Service Manager
3 CHANGES WITH 230 in spe:
5 * DNSSEC is now turned on by default in systemd-resolved (in
6 "allow-downgrade" mode), but may be turned off during compile time by
7 passing "--with-default-dnssec=no" to "configure" (and of course,
8 during runtime with DNSSEC= in resolved.conf). We recommend
9 downstreams to leave this on at least during development cycles and
10 report any issues with the DNSSEC logic upstream. We are very
11 interested in collecting feedback about the DNSSEC validator and its
12 limitations in the wild. Note however, that DNSSEC support is
13 probably nothing downstreams should turn on in stable distros just
14 yet, as it might create incompatibilities with a few DNS servers and
15 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
16 automatically whenever we detect such incompatible setups, but there
17 might be systems we do not cover yet. Hence: please help us testing
18 the DNSSEC code, leave this on where you can, report back, but then
19 again don't consider turning this on in your stable, LTS or
20 production release just yet. (Note that you have to enable
21 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
22 and its DNSSEC mode for host name resolution from local
25 * systemd-resolve conveniently resolves DANE records with the --tlsa
26 option and OPENPGPKEY records with the --openpgp option. It also
27 supports dumping raw DNS record data via the new --raw= switch.
29 * systemd-logind will now by default terminate user processes that are
30 part of the user session scope unit (session-XX.scope) when the user
31 logs out. This behavior is controlled by the KillUserProcesses=
32 setting in logind.conf, and the previous default of "no" is now
33 changed to "yes". This means that user sessions will be properly
34 cleaned up after, but additional steps are necessary to allow
35 intentionally long-running processes to survive logout.
37 While the user is logged in at least once, user@.service is running,
38 and any service that should survive the end of any individual login
39 session can be started at a user service or scope using systemd-run.
40 systemd-run(1) man page has been extended with an example which shows
41 how to run screen in a scope unit underneath user@.service. The same
42 command works for tmux.
44 After the user logs out of all sessions, user@.service will be
45 terminated too, by default, unless the user has "lingering" enabled.
46 To effectively allow users to run long-term tasks even if they are
47 logged out, lingering must be enabled for them. See loginctl(1) for
48 details. The default polkit policy was modified to allow users to
49 set lingering for themselves without authentication.
51 Previous defaults can be restored at compile time by the
52 --without-kill-user-processes option to "configure".
54 * systemd-logind gained new configuration settings SessionsMax= and
55 InhibitorsMax=, both with a default of 8192. It will not register new
56 user sessions or inhibitors above this limit.
58 * systemd-logind will now reload configuration on SIGHUP.
60 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
61 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
62 enable. Also, support for the "io" cgroup controller in the unified
63 hierarchy has been added, so that the "memory", "pids" and "io" are
64 now the controllers that are supported on the unified hierarchy.
66 WARNING: it is not possible to use previous systemd versions with
67 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
68 is necessary to also update systemd in the initramfs if using the
69 unified hierarchy. An updated SELinux policy is also required.
71 * LLDP support has been extended, and both passive (receive-only) and
72 active (sender) modes are supported. Passive mode ("routers-only") is
73 enabled by default in systemd-networkd. Active LLDP mode is enabled
74 by default for containers on the internal network. The "networkctl
75 lldp" command may be used to list information gathered. "networkctl
76 status" will also show basic LLDP information on connected peers now.
78 * The IAID and DUID unique identifier sent in DHCP requests may now be
79 configured for the system and each .network file managed by
80 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
82 * systemd-networkd gained support for configuring proxy ARP support for
83 each interface, via the ProxyArp= setting in .network files. It also
84 gained support for configuring the multicast querier feature of
85 bridge devices, via the new MulticastQuerier= setting in .netdev
86 files. Similarly, snooping on the IGMP traffic can be controlled
87 via the new setting MulticastSnooping=.
89 A new setting PreferredLifetime= has been added for addresses
90 configured in .network file to configure the lifetime intended for an
93 The systemd-networkd DHCP server gained the option EmitRouter=, which
94 defaults to yes, to configure whether the DHCP Option 3 (Router)
97 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
98 systemd-socket-activate and installed into /usr/bin. It is now fully
101 * systemd-journald now uses separate threads to flush changes to disk
102 when closing journal files, thus reducing impact of slow disk I/O on
105 * The sd-journal API gained two new calls
106 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
107 can be used to open journal files using file descriptors instead of
108 file or directory paths. sd_journal_open_container() has been
109 deprecated, sd_journal_open_directory_fd() should be used instead
110 with the flag SD_JOURNAL_OS_ROOT.
112 * journalctl learned a new output mode "-o short-unix" that outputs log
113 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
114 UTC). It also gained support for a new --no-hostname setting to
115 suppress the hostname column in the family of "short" output modes.
117 * systemd-ask-password now optionally skips printing of the password to
118 stdout with --no-output which can be useful in scripts.
120 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
121 (devices tagged with ID_MAKER_TOOL) are now tagged with
122 "uaccess" and are available to logged in users.
124 * The DeviceAllow= unit setting now supports specifiers (with "%").
126 * "systemctl show" gained a new --value switch, which allows print a
127 only the contents of a specific unit property, without also printing
128 the property's name. Similar support was added to "show*" verbs
129 of loginctl and machinectl that output "key=value" lists.
131 * A new unit type "generated" was added for files dynamically generated
132 by generator tools. Similarly, a new unit type "transient" is used
133 for unit files created using the runtime API. "systemctl enable" will
134 refuse to operate on such files.
136 * A new command "systemctl revert" has been added that may be used to
137 revert to the vendor version of a unit file, in case local changes
138 have been made by adding drop-ins or overriding the unit file.
140 * "machinectl clean" gained a new verb to automatically remove all or
141 just hidden container images.
143 * systemd-tmpfiles gained support for a new line type "e" for emptying
144 directories, if they exist, without creating them if they don't.
146 * systemd-nspawn gained support for automatically patching the UID/GIDs
147 of the owners and the ACLs of all files and directories in a
148 container tree to match the UID/GID user namespacing range selected
149 for the container invocation. This mode is enabled via the new
150 --private-user-chown switch. It also gained support for automatically
151 choosing a free, previously unused UID/GID range when starting a
152 container, via the new --private-users=pick setting (which implies
153 --private-user-chown). Together, these options for the first time
154 make user namespacing for nspawn containers fully automatic and thus
155 deployable. The systemd-nspaw@.service template unit file has been
156 changed to use this functionality by default.
158 * systemd-nspawn gained a new --network-zone= switch, that allows
159 creating ad-hoc virtual Ethernet links between multiple containers,
160 that only exist as long as at least one container referencing them is
161 running. This allows easy connecting of multiple containers with a
162 common link that implements an Ethernet broadcast domain. Each of
163 these network "zones" may be named relatively freely by the user, and
164 may be referenced by any number of containers, but each container may
165 only reference one of these "zones". On the lower level, this is
166 implemented by an automatically managed bridge network interface for
167 each zone, that is created when the first container referencing its
168 zone is created and removed when the last one referencing its zone
171 * The default start timeout may now be configured on the kernel command
172 line via systemd.default_timeout_start_sec=. It was already
173 configurable via the DefaultTimeoutStartSec= option in
174 /etc/systemd/system.conf.
176 * Socket units gained a new TriggerLimitIntervalSec= and
177 TriggerLimitBurst= setting to configure a limit on the activation
178 rate of the socket unit.
180 * The LimitNICE= setting now optionally takes normal UNIX nice values
181 in addition to the raw integer limit value. If the specified
182 parameter is prefixed with "+" or "-" and is in the range -20..19 the
183 value is understood as UNIX nice value. If not prefixed like this it
184 is understood as raw RLIMIT_NICE limit.
186 * Note that the effect of the PrivateDevices= unit file setting changed
187 slightly with this release: the per-device /dev file system will be
188 mounted read-only from this version on, and will have "noexec"
189 set. This (minor) change of behavior might cause some (exceptional)
190 legacy software to break, when PrivateDevices=yes is set for its
191 service. Please leave PrivateDevices= off if you run into problems
194 * systemd-bootchart has been split out to a separate repository:
195 https://github.com/systemd/systemd-bootchart
197 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
198 merged into the kernel in its current form.
200 * The compatibility libraries libsystemd-daemon.so,
201 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
202 which have been deprecated since systemd-209 have been removed along
203 with the corresponding pkg-config files. All symbols provided by
204 those libraries are provided by libsystemd.so.
206 * The Capabilities= unit file setting has been removed (it is ignored
207 for backwards compatibility). AmbientCapabilities= and
208 CapabilityBoundingSet= should be used instead.
210 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
211 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
212 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
213 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
214 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
215 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
216 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
217 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
218 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
219 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
220 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
221 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
222 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
223 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
224 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
225 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
226 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
227 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
228 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
229 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
230 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
231 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
232 Torstein Husebø, Umut Tezduyar Lindskog, Victor Toso, Vinay Kulkarni,
233 Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev, Wieland
234 Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
240 * The systemd-resolved DNS resolver service has gained a substantial
241 set of new features, most prominently it may now act as a DNSSEC
242 validating stub resolver. DNSSEC mode is currently turned off by
243 default, but is expected to be turned on by default in one of the
244 next releases. For now, we invite everybody to test the DNSSEC logic
245 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
246 service also gained a full set of D-Bus interfaces, including calls
247 to configure DNS and DNSSEC settings per link (for use by external
248 network management software). systemd-resolved and systemd-networkd
249 now distinguish between "search" and "routing" domains. The former
250 are used to qualify single-label names, the latter are used purely
251 for routing lookups within certain domains to specific links.
252 resolved now also synthesizes RRs for all entries from /etc/hosts.
254 * The systemd-resolve tool (which is a client utility for
255 systemd-resolved) has been improved considerably and is now fully
256 supported and documented. Hence it has moved from /usr/lib/systemd to
259 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
262 * The coredump collection logic has been reworked: when a coredump is
263 collected it is now written to disk, compressed and processed
264 (including stacktrace extraction) from a new instantiated service
265 systemd-coredump@.service, instead of directly from the
266 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
267 processing large coredumps can take up a substantial amount of
268 resources and time, and this previously happened entirely outside of
269 systemd's service supervision. With the new logic the core_pattern
270 hook only does minimal metadata collection before passing off control
271 to the new instantiated service, which is configured with a time
272 limit, a nice level and other settings to minimize negative impact on
273 the rest of the system. Also note that the new logic will honour the
274 RLIMIT_CORE setting of the crashed process, which now allows users
275 and processes to turn off coredumping for their processes by setting
278 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
279 and all forked processes by default. Previously, PID 1 would leave
280 the setting at "0" for all processes, as set by the kernel. Note that
281 the resource limit traditionally has no effect on the generated
282 coredumps on the system if the /proc/sys/kernel/core_pattern hook
283 logic is used. Since the limit is now honoured (see above) its
284 default has been changed so that the coredumping logic is enabled by
285 default for all processes, while allowing specific opt-out.
287 * When the stacktrace is extracted from processes of system users, this
288 is now done as "systemd-coredump" user, in order to sandbox this
289 potentially security sensitive parsing operation. (Note that when
290 processing coredumps of normal users this is done under the user ID
291 of process that crashed, as before.) Packagers should take notice
292 that it is now necessary to create the "systemd-coredump" system user
293 and group at package installation time.
295 * The systemd-activate socket activation testing tool gained support
296 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
297 and --seqpacket switches. It also has been extended to support both
298 new-style and inetd-style file descriptor passing. Use the new
299 --inetd switch to request inetd-style file descriptor passing.
301 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
302 variable, which takes a boolean value. If set to false, ANSI color
303 output is disabled in the tools even when run on a terminal that
306 * The VXLAN support in networkd now supports two new settings
307 DestinationPort= and PortRange=.
309 * A new systemd.machine_id= kernel command line switch has been added,
310 that may be used to set the machine ID in /etc/machine-id if it is
311 not initialized yet. This command line option has no effect if the
312 file is already initialized.
314 * systemd-nspawn gained a new --as-pid2 switch that invokes any
315 specified command line as PID 2 rather than PID 1 in the
316 container. In this mode PID 1 is a minimal stub init process that
317 implements the special POSIX and Linux semantics of PID 1 regarding
318 signal and child process management. Note that this stub init process
319 is implemented in nspawn itself and requires no support from the
320 container image. This new logic is useful to support running
321 arbitrary commands in the container, as normal processes are
322 generally not prepared to run as PID 1.
324 * systemd-nspawn gained a new --chdir= switch for setting the current
325 working directory for the process started in the container.
327 * "journalctl /dev/sda" will now output all kernel log messages for
328 specified device from the current boot, in addition to all devices
329 that are parents of it. This should make log output about devices
330 pretty useful, as long as kernel drivers attach enough metadata to
331 the log messages. (The usual SATA drivers do.)
333 * The sd-journal API gained two new calls
334 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
335 that report whether log data from /run or /var has been found.
337 * journalctl gained a new switch "--fields" that prints all journal
338 record field names currently in use in the journal. This is backed
339 by two new sd-journal API calls sd_journal_enumerate_fields() and
340 sd_journal_restart_fields().
342 * Most configurable timeouts in systemd now expect an argument of
343 "infinity" to turn them off, instead of "0" as before. The semantics
344 from now on is that a timeout of "0" means "now", and "infinity"
345 means "never". To maintain backwards compatibility, "0" continues to
346 turn off previously existing timeout settings.
348 * "systemctl reload-or-try-restart" has been renamed to "systemctl
349 try-reload-or-restart" to clarify what it actually does: the "try"
350 logic applies to both reloading and restarting, not just restarting.
351 The old name continues to be accepted for compatibility.
353 * On boot-up, when PID 1 detects that the system clock is behind the
354 release date of the systemd version in use, the clock is now set
355 to the latter. Previously, this was already done in timesyncd, in order
356 to avoid running with clocks set to the various clock epochs such as
357 1902, 1938 or 1970. With this change the logic is now done in PID 1
358 in addition to timesyncd during early boot-up, so that it is enforced
359 before the first process is spawned by systemd. Note that the logic
360 in timesyncd remains, as it is more comprehensive and ensures
361 clock monotonicity by maintaining a persistent timestamp file in
362 /var. Since /var is generally not available in earliest boot or the
363 initrd, this part of the logic remains in timesyncd, and is not done
366 * Support for tweaking details in net_cls.class_id through the
367 NetClass= configuration directive has been removed, as the kernel
368 people have decided to deprecate that controller in cgroup v2.
369 Userspace tools such as nftables are moving over to setting rules
370 that are specific to the full cgroup path of a task, which obsoletes
371 these controllers anyway. The NetClass= directive is kept around for
372 legacy compatibility reasons. For a more in-depth description of the
373 kernel change, please refer to the respective upstream commit:
375 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
377 * A new service setting RuntimeMaxSec= has been added that may be used
378 to specify a maximum runtime for a service. If the timeout is hit, the
379 service is terminated and put into a failure state.
381 * A new service setting AmbientCapabilities= has been added. It allows
382 configuration of additional Linux process capabilities that are
383 passed to the activated processes. This is only available on very
386 * The process resource limit settings in service units may now be used
387 to configure hard and soft limits individually.
389 * The various libsystemd APIs such as sd-bus or sd-event now publicly
390 expose support for gcc's __attribute__((cleanup())) C extension.
391 Specifically, for many object destructor functions alternative
392 versions have been added that have names suffixed with "p" and take a
393 pointer to a pointer to the object to destroy, instead of just a
394 pointer to the object itself. This is useful because these destructor
395 functions may be used directly as parameters to the cleanup
396 construct. Internally, systemd has been a heavy user of this GCC
397 extension for a long time, and with this change similar support is
398 now available to consumers of the library outside of systemd. Note
399 that by using this extension in your sources compatibility with old
400 and strictly ANSI compatible C compilers is lost. However, all gcc or
401 LLVM versions of recent years support this extension.
403 * Timer units gained support for a new setting RandomizedDelaySec= that
404 allows configuring some additional randomized delay to the configured
405 time. This is useful to spread out timer events to avoid load peaks in
406 clusters or larger setups.
408 * Calendar time specifications now support sub-second accuracy.
410 * Socket units now support listening on SCTP and UDP-lite protocol
413 * The sd-event API now comes with a full set of man pages.
415 * Older versions of systemd contained experimental support for
416 compressing journal files and coredumps with the LZ4 compressor that
417 was not compatible with the lz4 binary (due to API limitations of the
418 lz4 library). This support has been removed; only support for files
419 compatible with the lz4 binary remains. This LZ4 logic is now
420 officially supported and no longer considered experimental.
422 * The dkr image import logic has been removed again from importd. dkr's
423 micro-services focus doesn't fit into the machine image focus of
424 importd, and quickly got out of date with the upstream dkr API.
426 * Creation of the /run/lock/lockdev/ directory was dropped from
427 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
428 been available for many years. If you still need this, you need to
429 create your own tmpfiles.d config file with:
431 d /run/lock/lockdev 0775 root lock -
433 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
434 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
435 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
436 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
437 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
438 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
439 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
440 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
441 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
442 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
443 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
444 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
445 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
446 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
447 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
448 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
449 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
450 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
451 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
457 * A number of properties previously only settable in unit
458 files are now also available as properties to set when
459 creating transient units programmatically via the bus, as it
460 is exposed with systemd-run's --property=
461 setting. Specifically, these are: SyslogIdentifier=,
462 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
463 EnvironmentFile=, ReadWriteDirectories=,
464 ReadOnlyDirectories=, InaccessibleDirectories=,
465 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
467 * When creating transient services via the bus API it is now
468 possible to pass in a set of file descriptors to use as
469 STDIN/STDOUT/STDERR for the invoked process.
471 * Slice units may now be created transiently via the bus APIs,
472 similar to the way service and scope units may already be
475 * Wherever systemd expects a calendar timestamp specification
476 (like in journalctl's --since= and --until= switches) UTC
477 timestamps are now supported. Timestamps suffixed with "UTC"
478 are now considered to be in Universal Time Coordinated
479 instead of the local timezone. Also, timestamps may now
480 optionally be specified with sub-second accuracy. Both of
481 these additions also apply to recurring calendar event
482 specification, such as OnCalendar= in timer units.
484 * journalctl gained a new "--sync" switch that asks the
485 journal daemon to write all so far unwritten log messages to
486 disk and sync the files, before returning.
488 * systemd-tmpfiles learned two new line types "q" and "Q" that
489 operate like "v", but also set up a basic btrfs quota
490 hierarchy when used on a btrfs file system with quota
493 * tmpfiles' "v", "q" and "Q" will now create a plain directory
494 instead of a subvolume (even on a btrfs file system) if the
495 root directory is a plain directory, and not a
496 subvolume. This should simplify things with certain chroot()
497 environments which are not aware of the concept of btrfs
500 * systemd-detect-virt gained a new --chroot switch to detect
501 whether execution takes place in a chroot() environment.
503 * CPUAffinity= now takes CPU index ranges in addition to
506 * The various memory-related resource limit settings (such as
507 LimitAS=) now understand the usual K, M, G, ... suffixes to
508 the base of 1024 (IEC). Similar, the time-related resource
509 limit settings understand the usual min, h, day, ...
512 * There's a new system.conf setting DefaultTasksMax= to
513 control the default TasksMax= setting for services and
514 scopes running on the system. (TasksMax= is the primary
515 setting that exposes the "pids" cgroup controller on systemd
516 and was introduced in the previous systemd release.) The
517 setting now defaults to 512, which means services that are
518 not explicitly configured otherwise will only be able to
519 create 512 processes or threads at maximum, from this
520 version on. Note that this means that thread- or
521 process-heavy services might need to be reconfigured to set
522 TasksMax= to a higher value. It is sufficient to set
523 TasksMax= in these specific unit files to a higher value, or
524 even "infinity". Similar, there's now a logind.conf setting
525 UserTasksMax= that defaults to 4096 and limits the total
526 number of processes or tasks each user may own
527 concurrently. nspawn containers also have the TasksMax=
528 value set by default now, to 8192. Note that all of this
529 only has an effect if the "pids" cgroup controller is
530 enabled in the kernel. The general benefit of these changes
531 should be a more robust and safer system, that provides a
532 certain amount of per-service fork() bomb protection.
534 * systemd-nspawn gained the new --network-veth-extra= switch
535 to define additional and arbitrarily-named virtual Ethernet
536 links between the host and the container.
538 * A new service execution setting PassEnvironment= has been
539 added that allows importing select environment variables
540 from PID1's environment block into the environment block of
543 * Timer units gained support for a new RemainAfterElapse=
544 setting which takes a boolean argument. It defaults on on,
545 exposing behaviour unchanged to previous releases. If set to
546 off, timer units are unloaded after they elapsed if they
547 cannot elapse again. This is particularly useful for
548 transient timer units, which shall not stay around longer
549 than until they first elapse.
551 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
552 default now (the kernel default is 16). This is beneficial
553 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
554 allows substantially larger numbers of queued
555 datagrams. This should increase the capability of systemd to
556 parallelize boot-up, as logging and sd_notify() are unlikely
557 to stall execution anymore. If you need to change the value
558 from the new defaults, use the usual sysctl.d/ snippets.
560 * The compression framing format used by the journal or
561 coredump processing has changed to be in line with what the
562 official LZ4 tools generate. LZ4 compression support in
563 systemd was considered unsupported previously, as the format
564 was not compatible with the normal tools. With this release
565 this has changed now, and it is hence safe for downstream
566 distributions to turn it on. While not compressing as well
567 as the XZ, LZ4 is substantially faster, which makes
568 it a good default choice for the compression logic in the
569 journal and in coredump handling.
571 * Any reference to /etc/mtab has been dropped from
572 systemd. The file has been obsolete since a while, but
573 systemd refused to work on systems where it was incorrectly
574 set up (it should be a symlink or non-existent). Please make
575 sure to update to util-linux 2.27.1 or newer in conjunction
576 with this systemd release, which also drops any reference to
577 /etc/mtab. If you maintain a distribution make sure that no
578 software you package still references it, as this is a
579 likely source of bugs. There's also a glibc bug pending,
580 asking for removal of any reference to this obsolete file:
582 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
584 Note that only util-linux versions built with
585 --enable-libmount-force-mountinfo are supported.
587 * Support for the ".snapshot" unit type has been removed. This
588 feature turned out to be little useful and little used, and
589 has now been removed from the core and from systemctl.
591 * The dependency types RequiresOverridable= and
592 RequisiteOverridable= have been removed from systemd. They
593 have been used only very sparingly to our knowledge and
594 other options that provide a similar effect (such as
595 systemctl --mode=ignore-dependencies) are much more useful
596 and commonly used. Moreover, they were only half-way
597 implemented as the option to control behaviour regarding
598 these dependencies was never added to systemctl. By removing
599 these dependency types the execution engine becomes a bit
600 simpler. Unit files that use these dependencies should be
601 changed to use the non-Overridable dependency types
602 instead. In fact, when parsing unit files with these
603 options, that's what systemd will automatically convert them
604 too, but it will also warn, asking users to fix the unit
605 files accordingly. Removal of these dependency types should
606 only affect a negligible number of unit files in the wild.
608 * Behaviour of networkd's IPForward= option changed
609 (again). It will no longer maintain a per-interface setting,
610 but propagate one way from interfaces where this is enabled
611 to the global kernel setting. The global setting will be
612 enabled when requested by a network that is set up, but
613 never be disabled again. This change was made to make sure
614 IPv4 and IPv6 behaviour regarding packet forwarding is
615 similar (as the Linux IPv6 stack does not support
616 per-interface control of this setting) and to minimize
619 * In unit files the behaviour of %u, %U, %h, %s has
620 changed. These specifiers will now unconditionally resolve
621 to the various user database fields of the user that the
622 systemd instance is running as, instead of the user
623 configured in the specific unit via User=. Note that this
624 effectively doesn't change much, as resolving of these
625 specifiers was already turned off in the --system instance
626 of systemd, as we cannot do NSS lookups from PID 1. In the
627 --user instance of systemd these specifiers where correctly
628 resolved, but hardly made any sense, since the user instance
629 lacks privileges to do user switches anyway, and User= is
630 hence useless. Morever, even in the --user instance of
631 systemd behaviour was awkward as it would only take settings
632 from User= assignment placed before the specifier into
633 account. In order to unify and simplify the logic around
634 this the specifiers will now always resolve to the
635 credentials of the user invoking the manager (which in case
636 of PID 1 is the root user).
638 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
639 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
640 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
641 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
642 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
643 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
644 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
645 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
646 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
647 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
648 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
649 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
650 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
651 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
658 * systemd now depends on util-linux v2.27. More specifically,
659 the newly added mount monitor feature in libmount now
660 replaces systemd's former own implementation.
662 * libmount mandates /etc/mtab not to be regular file, and
663 systemd now enforces this condition at early boot.
664 /etc/mtab has been deprecated and warned about for a very
665 long time, so systems running systemd should already have
666 stopped having this file around as anything else than a
667 symlink to /proc/self/mounts.
669 * Support for the "pids" cgroup controller has been added. It
670 allows accounting the number of tasks in a cgroup and
671 enforcing limits on it. This adds two new setting
672 TasksAccounting= and TasksMax= to each unit, as well as a
673 global option DefaultTasksAccounting=.
675 * Support for the "net_cls" cgroup controller has been added.
676 It allows assigning a net class ID to each task in the
677 cgroup, which can then be used in firewall rules and traffic
678 shaping configurations. Note that the kernel netfilter net
679 class code does not currently work reliably for ingress
680 packets on unestablished sockets.
682 This adds a new config directive called NetClass= to CGroup
683 enabled units. Allowed values are positive numbers for fixed
684 assignments and "auto" for picking a free value
687 * 'systemctl is-system-running' now returns 'offline' if the
688 system is not booted with systemd. This command can now be
689 used as a substitute for 'systemd-notify --booted'.
691 * Watchdog timeouts have been increased to 3 minutes for all
692 in-tree service files. Apparently, disk IO issues are more
693 frequent than we hoped, and user reported >1 minute waiting
696 * 'machine-id-commit' functionality has been merged into
697 'machine-id-setup --commit'. The separate binary has been
700 * The WorkingDirectory= directive in unit files may now be set
701 to the special value '~'. In this case, the working
702 directory is set to the home directory of the user
705 * "machinectl shell" will now open the shell in the home
706 directory of the selected user by default.
708 * The CrashChVT= configuration file setting is renamed to
709 CrashChangeVT=, following our usual logic of not
710 abbreviating unnecessarily. The old directive is still
711 supported for compat reasons. Also, this directive now takes
712 an integer value between 1 and 63, or a boolean value. The
713 formerly supported '-1' value for disabling stays around for
716 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
717 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
718 RootDirectory= properties can now be set for transient
721 * The systemd-analyze tool gained a new "set-log-target" verb
722 to change the logging target the system manager logs to
723 dynamically during runtime. This is similar to how
724 "systemd-analyze set-log-level" already changes the log
727 * In nspawn /sys is now mounted as tmpfs, with only a selected
728 set of subdirectories mounted in from the real sysfs. This
729 enhances security slightly, and is useful for ensuring user
730 namespaces work correctly.
732 * Support for USB FunctionFS activation has been added. This
733 allows implementation of USB gadget services that are
734 activated as soon as they are requested, so that they don't
735 have to run continously, similar to classic socket
738 * The "systemctl exit" command now optionally takes an
739 additional parameter that sets the exit code to return from
740 the systemd manager when exiting. This is only relevant when
741 running the systemd user instance, or when running the
742 system instance in a container.
744 * sd-bus gained the new API calls sd_bus_path_encode_many()
745 and sd_bus_path_decode_many() that allow easy encoding and
746 decoding of multiple identifier strings inside a D-Bus
747 object path. Another new call sd_bus_default_flush_close()
748 has been added to flush and close per-thread default
751 * systemd-cgtop gained support for a -M/--machine= switch to
752 show the control groups within a certain container only.
754 * "systemctl kill" gained support for an optional --fail
755 switch. If specified the requested operation will fail of no
756 processes have been killed, because the unit had no
757 processes attached, or similar.
759 * A new systemd.crash_reboot=1 kernel command line option has
760 been added that triggers a reboot after crashing. This can
761 also be set through CrashReboot= in systemd.conf.
763 * The RuntimeDirectory= setting now understands unit
764 specifiers like %i or %f.
766 * A new (still internal) libary API sd-ipv4acd has been added,
767 that implements address conflict detection for IPv4. It's
768 based on code from sd-ipv4ll, and will be useful for
769 detecting DHCP address conflicts.
771 * File descriptors passed during socket activation may now be
772 named. A new API sd_listen_fds_with_names() is added to
773 access the names. The default names may be overridden,
774 either in the .socket file using the FileDescriptorName=
775 parameter, or by passing FDNAME= when storing the file
776 descriptors using sd_notify().
778 * systemd-networkd gained support for:
780 - Setting the IPv6 Router Advertisment settings via
781 IPv6AcceptRouterAdvertisements= in .network files.
783 - Configuring the HelloTimeSec=, MaxAgeSec= and
784 ForwardDelaySec= bridge parameters in .netdev files.
786 - Configuring PreferredSource= for static routes in
789 * The "ask-password" framework used to query for LUKS harddisk
790 passwords or SSL passwords during boot gained support for
791 caching passwords in the kernel keyring, if it is
792 available. This makes sure that the user only has to type in
793 a passphrase once if there are multiple objects to unlock
794 with the same one. Previously, such password caching was
795 available only when Plymouth was used; this moves the
796 caching logic into the systemd codebase itself. The
797 "systemd-ask-password" utility gained a new --keyname=
798 switch to control which kernel keyring key to use for
799 caching a password in. This functionality is also useful for
800 enabling display managers such as gdm to automatically
801 unlock the user's GNOME keyring if its passphrase, the
802 user's password and the harddisk password are the same, if
803 gdm-autologin is used.
805 * When downloading tar or raw images using "machinectl
806 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
807 file is now also downloaded, if it is available and stored
808 next to the image file.
810 * Units of type ".socket" gained a new boolean setting
811 Writable= which is only useful in conjunction with
812 ListenSpecial=. If true, enables opening the specified
813 special file in O_RDWR mode rather than O_RDONLY mode.
815 * systemd-rfkill has been reworked to become a singleton
816 service that is activated through /dev/rfkill on each rfkill
817 state change and saves the settings to disk. This way,
818 systemd-rfkill is now compatible with devices that exist
819 only intermittendly, and even restores state if the previous
820 system shutdown was abrupt rather than clean.
822 * The journal daemon gained support for vacuuming old journal
823 files controlled by the number of files that shall remain,
824 in addition to the already existing control by size and by
825 date. This is useful as journal interleaving performance
826 degrades with too many seperate journal files, and allows
827 putting an effective limit on them. The new setting defaults
828 to 100, but this may be changed by setting SystemMaxFiles=
829 and RuntimeMaxFiles= in journald.conf. Also, the
830 "journalctl" tool gained the new --vacuum-files= switch to
831 manually vacuum journal files to leave only the specified
832 number of files in place.
834 * udev will now create /dev/disk/by-path links for ATA devices
835 on kernels where that is supported.
837 * Galician, Serbian, Turkish and Korean translations were added.
839 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
840 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
841 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
842 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
843 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
844 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
845 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
846 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
847 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
848 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
849 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
850 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
851 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
852 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
853 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
854 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
855 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
856 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
862 * The DHCP implementation of systemd-networkd gained a set of
865 - The DHCP server now supports emitting DNS and NTP
866 information. It may be enabled and configured via
867 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
868 and NTP information is enabled, but no servers are
869 configured, the corresponding uplink information (if there
870 is any) is propagated.
872 - Server and client now support transmission and reception
873 of timezone information. It can be configured via the
874 newly introduced network options UseTimezone=,
875 EmitTimezone=, and Timezone=. Transmission of timezone
876 information is enabled between host and containers by
877 default now: the container will change its local timezone
878 to what the host has set.
880 - Lease timeouts can now be configured via
881 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
883 - The DHCP server improved on the stability of
884 leases. Clients are more likely to get the same lease
885 information back, even if the server loses state.
887 - The DHCP server supports two new configuration options to
888 control the lease address pool metrics, PoolOffset= and
891 * The encapsulation limit of tunnels in systemd-networkd may
892 now be configured via 'EncapsulationLimit='. It allows
893 modifying the maximum additional levels of encapsulation
894 that are permitted to be prepended to a packet.
896 * systemd now supports the concept of user buses replacing
897 session buses, if used with dbus-1.10 (and enabled via dbus
898 --enable-user-session). It previously only supported this on
899 kdbus-enabled systems, and this release expands this to
900 'dbus-daemon' systems.
902 * systemd-networkd now supports predictable interface names
905 * systemd now optionally supports the new Linux kernel
906 "unified" control group hierarchy. If enabled via the kernel
907 command-line option 'systemd.unified_cgroup_hierarchy=1',
908 systemd will try to mount the unified cgroup hierarchy
909 directly on /sys/fs/cgroup. If not enabled, or not
910 available, systemd will fall back to the legacy cgroup
911 hierarchy setup, as before. Host system and containers can
912 mix and match legacy and unified hierarchies as they
913 wish. nspawn understands the $UNIFIED_CROUP_HIERARCHY
914 environment variable to individually select the hierarchy to
915 use for executed containers. By default, nspawn will use the
916 unified hierarchy for the containers if the host uses the
917 unified hierarchy, and the legacy hierarchy otherwise.
918 Please note that at this point the unified hierarchy is an
919 experimental kernel feature and is likely to change in one
920 of the next kernel releases. Therefore, it should not be
921 enabled by default in downstream distributions yet. The
922 minimum required kernel version for the unified hierarchy to
923 work is 4.2. Note that when the unified hierarchy is used
924 for the first time delegated access to controllers is
925 safe. Because of this systemd-nspawn containers will get
926 access to controllers now, as will systemd user
927 sessions. This means containers and user sessions may now
928 manage their own resources, partitioning up what the system
931 * A new special scope unit "init.scope" has been introduced
932 that encapsulates PID 1 of the system. It may be used to
933 determine resource usage and enforce resource limits on PID
934 1 itself. PID 1 hence moved out of the root of the control
937 * The cgtop tool gained support for filtering out kernel
938 threads when counting tasks in a control group. Also, the
939 count of processes is now recursively summed up by
940 default. Two options -k and --recursive= have been added to
941 revert to old behaviour. The tool has also been updated to
942 work correctly in containers now.
944 * systemd-nspawn's --bind= and --bind-ro= options have been
945 extended to allow creation of non-recursive bind mounts.
947 * libsystemd gained two new calls sd_pid_get_cgroup() and
948 sd_peer_get_cgroup() which return the control group path of
949 a process or peer of a connected AF_UNIX socket. This
950 function call is particularly useful when implementing
951 delegated subtrees support in the control group hierarchy.
953 * The "sd-event" event loop API of libsystemd now supports
954 correct dequeuing of real-time signals, without losing
957 * When systemd requests a PolicyKit decision when managing
958 units it will now add additional fields to the request,
959 including unit name and desired operation. This enables more
960 powerful PolicyKit policies, that make decisions depending
963 * nspawn learnt support for .nspawn settings files, that may
964 accompany the image files or directories of containers, and
965 may contain additional settings for the container. This is
966 an alternative to configuring container parameters via the
969 Contributions from: Cristian Rodríguez, Daniel Mack, David
970 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
971 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
972 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
973 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
974 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
975 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
976 Andersen, Tom Gundersen, Torstein Husebø
982 * machinectl gained a new verb 'shell' which opens a fresh
983 shell on the target container or the host. It is similar to
984 the existing 'login' command of machinectl, but spawns the
985 shell directly without prompting for username or
986 password. The pseudo machine '.host' now refers to the local
987 host and is used by default. Hence, 'machinectl shell' can
988 be used as replacement for 'su -' which spawns a session as
989 a fresh systemd unit in a way that is fully isolated from
990 the originating session.
992 * systemd-networkd learned to cope with private-zone DHCP
993 options and allows other programs to query the values.
995 * SELinux access control when enabling/disabling units is no
996 longer enforced with this release. The previous
997 implementation was incorrect, and a new corrected
998 implementation is not yet available. As unit file operations
999 are still protected via PolicyKit and D-Bus policy this is
1000 not a security problem. Yet, distributions which care about
1001 optimal SELinux support should probably not stabilize on
1004 * sd-bus gained support for matches of type "arg0has=", that
1005 test for membership of strings in string arrays sent in bus
1008 * systemd-resolved now dumps the contents of its DNS and LLMNR
1009 caches to the logs on reception of the SIGUSR1 signal. This
1010 is useful to debug DNS behaviour.
1012 * The coredumpctl tool gained a new --directory= option to
1013 operate on journal files in a specific directory.
1015 * "systemctl reboot" and related commands gained a new
1016 "--message=" option which may be used to set a free-text
1017 wall message when shutting down or rebooting the
1018 system. This message is also logged, which is useful for
1019 figuring out the reason for a reboot or shutdown a
1022 * The "systemd-resolve-host" tool's -i switch now takes
1023 network interface numbers as alternative to interface names.
1025 * A new unit file setting for services has been introduced:
1026 UtmpMode= allows configuration of how precisely systemd
1027 handles utmp and wtmp entries for the service if this is
1028 enabled. This allows writing services that appear similar to
1029 user sessions in the output of the "w", "who", "last" and
1032 * systemd-resolved will now locally synthesize DNS resource
1033 records for the "localhost" and "gateway" domains as well as
1034 the local hostname. This should ensure that clients querying
1035 RRs via resolved will get similar results as those going via
1036 NSS, if nss-myhostname is enabled.
1038 Contributions from: Alastair Hughes, Alex Crawford, Daniel
1039 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
1040 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
1041 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
1042 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
1043 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
1044 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
1045 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
1046 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
1047 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
1048 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
1049 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
1051 — Berlin, 2015-08-27
1055 * The systemd-efi-boot-generator functionality was merged into
1056 systemd-gpt-auto-generator.
1058 * systemd-networkd now supports Group Policy for vxlan
1059 devices. It can be enabled via the new boolean configuration
1060 option called 'GroupPolicyExtension='.
1062 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
1063 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
1064 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
1066 — Berlin, 2015-07-31
1070 * The python-systemd code has been removed from the systemd repository.
1071 A new repository has been created which accommodates the code from
1072 now on, and we kindly ask distributions to create a separate package
1073 for this: https://github.com/systemd/python-systemd
1075 * The systemd daemon will now reload its main configuration
1076 (/etc/systemd/system.conf) on daemon-reload.
1078 * sd-dhcp now exposes vendor specific extensions via
1079 sd_dhcp_lease_get_vendor_specific().
1081 * systemd-networkd gained a number of new configuration options.
1083 - A new boolean configuration option for TAP devices called
1084 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
1085 device, thus allowing to send and receive GSO packets.
1087 - A new tunnel configuration option called 'CopyDSCP='.
1088 If enabled, the DSCP field of ip6 tunnels is copied into the
1089 decapsulated packet.
1091 - A set of boolean bridge configuration options were added.
1092 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
1093 and 'UnicastFlood=' are now parsed by networkd and applied to the
1094 respective bridge link device via the respective IFLA_BRPORT_*
1097 - A new string configuration option to override the hostname sent
1098 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
1099 is true, networkd will use the configured hostname instead of the
1100 system hostname when sending DHCP requests.
1102 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
1103 networkd will configure the IPv6 flow-label of the tunnel device
1104 according to RFC2460.
1106 - The 'macvtap' virtual network devices are now supported, similar to
1107 the already supported 'macvlan' devices.
1109 * systemd-resolved now implements RFC5452 to improve resilience against
1110 cache poisoning. Additionally, source port randomization is enabled
1111 by default to further protect against DNS spoofing attacks.
1113 * nss-mymachines now supports translating UIDs and GIDs of running
1114 containers with user-namespaces enabled. If a container 'foo'
1115 translates a host uid 'UID' to the container uid 'TUID', then
1116 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
1117 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
1118 mapped as 'vg-foo-TGID'.
1120 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
1121 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
1122 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
1123 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
1124 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
1125 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
1126 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
1127 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
1128 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
1129 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
1131 — Berlin, 2015-07-29
1135 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
1136 There are no known issues with current sysfs, and udev does not need
1137 or should be used to work around such bugs.
1139 * udev does no longer enable USB HID power management. Several reports
1140 indicate, that some devices cannot handle that setting.
1142 * The udev accelerometer helper was removed. The functionality
1143 is now fully included in iio-sensor-proxy. But this means,
1144 older iio-sensor-proxy versions will no longer provide
1145 accelerometer/orientation data with this systemd version.
1146 Please upgrade iio-sensor-proxy to version 1.0.
1148 * networkd gained a new configuration option IPv6PrivacyExtensions=
1149 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
1150 for Stateless Address") on selected networks.
1152 * For the sake of fewer build-time dependencies and less code in the
1153 main repository, the python bindings are about to be removed in the
1154 next release. A new repository has been created which accommodates
1155 the code from now on, and we kindly ask distributions to create a
1156 separate package for this. The removal will take place in v223.
1158 https://github.com/systemd/python-systemd
1160 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
1161 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
1162 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
1163 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
1164 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
1165 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
1166 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
1167 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
1168 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
1169 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
1171 — Berlin, 2015-07-07
1175 * The sd-bus.h and sd-event.h APIs have now been declared
1176 stable and have been added to the official interface of
1177 libsystemd.so. sd-bus implements an alternative D-Bus client
1178 library, that is relatively easy to use, very efficient and
1179 supports both classic D-Bus as well as kdbus as transport
1180 backend. sd-event is a generic event loop abstraction that
1181 is built around Linux epoll, but adds features such as event
1182 prioritization or efficient timer handling. Both APIs are good
1183 choices for C programs looking for a bus and/or event loop
1184 implementation that is minimal and does not have to be
1185 portable to other kernels.
1187 * kdbus support is no longer compile-time optional. It is now
1188 always built-in. However, it can still be disabled at
1189 runtime using the kdbus=0 kernel command line setting, and
1190 that setting may be changed to default to off, by specifying
1191 --disable-kdbus at build-time. Note though that the kernel
1192 command line setting has no effect if the kdbus.ko kernel
1193 module is not installed, in which case kdbus is (obviously)
1194 also disabled. We encourage all downstream distributions to
1195 begin testing kdbus by adding it to the kernel images in the
1196 development distributions, and leaving kdbus support in
1199 * The minimal required util-linux version has been bumped to
1202 * Support for chkconfig (--enable-chkconfig) was removed in
1203 favor of calling an abstraction tool
1204 /lib/systemd/systemd-sysv-install. This needs to be
1205 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
1206 in README for details.
1208 * If there's a systemd unit and a SysV init script for the
1209 same service name, and the user executes "systemctl enable"
1210 for it (or a related call), then this will now enable both
1211 (or execute the related operation on both), not just the
1214 * The libudev API documentation has been converted from gtkdoc
1217 * gudev has been removed from the systemd tree, it is now an
1220 * The systemd-cgtop tool learnt a new --raw switch to generate
1221 "raw" (machine parsable) output.
1223 * networkd's IPForwarding= .network file setting learnt the
1224 new setting "kernel", which ensures that networkd does not
1225 change the IP forwarding sysctl from the default kernel
1228 * The systemd-logind bus API now exposes a new boolean
1229 property "Docked" that reports whether logind considers the
1230 system "docked", i.e. connected to a docking station or not.
1232 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
1233 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
1234 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
1235 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
1236 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
1237 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
1238 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
1239 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
1240 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
1241 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
1242 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
1243 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
1244 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
1245 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
1246 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
1247 Fink, Zbigniew Jędrzejewski-Szmek
1249 — Berlin, 2015-06-19
1253 * The gudev library has been extracted into a separate repository
1254 available at: https://git.gnome.org/browse/libgudev/
1255 It is now managed as part of the Gnome project. Distributions
1256 are recommended to pass --disable-gudev to systemd and use
1257 gudev from the Gnome project instead. gudev is still included
1258 in systemd, for now. It will be removed soon, though. Please
1259 also see the announcement-thread on systemd-devel:
1260 http://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
1262 * systemd now exposes a CPUUsageNSec= property for each
1263 service unit on the bus, that contains the overall consumed
1264 CPU time of a service (the sum of what each process of the
1265 service consumed). This value is only available if
1266 CPUAccounting= is turned on for a service, and is then shown
1267 in the "systemctl status" output.
1269 * Support for configuring alternative mappings of the old SysV
1270 runlevels to systemd targets has been removed. They are now
1271 hardcoded in a way that runlevels 2, 3, 4 all map to
1272 multi-user.target and 5 to graphical.target (which
1273 previously was already the default behaviour).
1275 * The auto-mounter logic gained support for mount point
1276 expiry, using a new TimeoutIdleSec= setting in .automount
1277 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
1279 * The EFI System Partition (ESP) as mounted to /boot by
1280 systemd-efi-boot-generator will now be unmounted
1281 automatically after 2 minutes of not being used. This should
1282 minimize the risk of ESP corruptions.
1284 * New /etc/fstab options x-systemd.requires= and
1285 x-systemd.requires-mounts-for= are now supported to express
1286 additional dependencies for mounts. This is useful for
1287 journalling file systems that support external journal
1288 devices or overlay file systems that require underlying file
1289 systems to be mounted.
1291 * systemd does not support direct live-upgrades (via systemctl
1292 daemon-reexec) from versions older than v44 anymore. As no
1293 distribution we are aware of shipped such old versions in a
1294 stable release this should not be problematic.
1296 * When systemd forks off a new per-connection service instance
1297 it will now set the $REMOTE_ADDR environment variable to the
1298 remote IP address, and $REMOTE_PORT environment variable to
1299 the remote IP port. This behaviour is similar to the
1300 corresponding environment variables defined by CGI.
1302 * systemd-networkd gained support for uplink failure
1303 detection. The BindCarrier= option allows binding interface
1304 configuration dynamically to the link sense of other
1305 interfaces. This is useful to achieve behaviour like in
1308 * systemd-networkd gained support for configuring the DHCP
1309 client identifier to use when requesting leases.
1311 * systemd-networkd now has a per-network UseNTP= option to
1312 configure whether NTP server information acquired via DHCP
1313 is passed on to services like systemd-timesyncd.
1315 * systemd-networkd gained support for vti6 tunnels.
1317 * Note that systemd-networkd manages the sysctl variable
1318 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
1319 it is configured for since v219. The variable controls IP
1320 forwarding, and is a per-interface alternative to the global
1321 /proc/sys/net/ipv[46]/ip_forward. This setting is
1322 configurable in the IPForward= option, which defaults to
1323 "no". This means if networkd is used for an interface it is
1324 no longer sufficient to set the global sysctl option to turn
1325 on IP forwarding! Instead, the .network file option
1326 IPForward= needs to be turned on! Note that the
1327 implementation of this behaviour was broken in v219 and has
1330 * Many bonding and vxlan options are now configurable in
1333 * systemd-nspawn gained a new --property= setting to set unit
1334 properties for the container scope. This is useful for
1335 setting resource parameters (e.g "CPUShares=500") on
1336 containers started from the command line.
1338 * systemd-nspawn gained a new --private-users= switch to make
1339 use of user namespacing available on recent Linux kernels.
1341 * systemd-nspawn may now be called as part of a shell pipeline
1342 in which case the pipes used for stdin and stdout are passed
1343 directly to the process invoked in the container, without
1344 indirection via a pseudo tty.
1346 * systemd-nspawn gained a new switch to control the UNIX
1347 signal to use when killing the init process of the container
1350 * systemd-nspawn gained a new --overlay= switch for mounting
1351 overlay file systems into the container using the new kernel
1354 * When a container image is imported via systemd-importd and
1355 the host file system is not btrfs, a loopback block device
1356 file is created in /var/lib/machines.raw with a btrfs file
1357 system inside. It is then mounted to /var/lib/machines to
1358 enable btrfs features for container management. The loopback
1359 file and btrfs file system is grown as needed when container
1360 images are imported via systemd-importd.
1362 * systemd-machined/systemd-importd gained support for btrfs
1363 quota, to enforce container disk space limits on disk. This
1364 is exposed in "machinectl set-limit".
1366 * systemd-importd now can import containers from local .tar,
1367 .raw and .qcow2 images, and export them to .tar and .raw. It
1368 can also import dkr v2 images now from the network (on top
1371 * systemd-importd gained support for verifying downloaded
1372 images with gpg2 (previously only gpg1 was supported).
1374 * systemd-machined, systemd-logind, systemd: most bus calls
1375 are now accessible to unprivileged processes via
1376 PolicyKit. Also, systemd-logind will now allow users to kill
1377 their own sessions without further privileges or
1380 * systemd-shutdownd has been removed. This service was
1381 previously responsible for implementing scheduled shutdowns
1382 as exposed in /usr/bin/shutdown's time parameter. This
1383 functionality has now been moved into systemd-logind and is
1384 accessible via a bus interface.
1386 * "systemctl reboot" gained a new switch --firmware-setup that
1387 can be used to reboot into the EFI firmware setup, if that
1388 is available. systemd-logind now exposes an API on the bus
1389 to trigger such reboots, in case graphical desktop UIs want
1390 to cover this functionality.
1392 * "systemctl enable", "systemctl disable" and "systemctl mask"
1393 now support a new "--now" switch. If specified the units
1394 that are enabled will also be started, and the ones
1395 disabled/masked also stopped.
1397 * The Gummiboot EFI boot loader tool has been merged into
1398 systemd, and renamed to "systemd-boot". The bootctl tool has been
1399 updated to support systemd-boot.
1401 * An EFI kernel stub has been added that may be used to create
1402 kernel EFI binaries that contain not only the actual kernel,
1403 but also an initrd, boot splash, command line and OS release
1404 information. This combined binary can then be signed as a
1405 single image, so that the firmware can verify it all in one
1406 step. systemd-boot has special support for EFI binaries created
1407 like this and can extract OS release information from them
1408 and show them in the boot menu. This functionality is useful
1409 to implement cryptographically verified boot schemes.
1411 * Optional support has been added to systemd-fsck to pass
1412 fsck's progress report to an AF_UNIX socket in the file
1415 * udev will no longer create device symlinks for all block
1416 devices by default. A blacklist for excluding special block
1417 devices from this logic has been turned into a whitelist
1418 that requires picking block devices explicitly that require
1421 * A new (currently still internal) API sd-device.h has been
1422 added to libsystemd. This modernized API is supposed to
1423 replace libudev eventually. In fact, already much of libudev
1424 is now just a wrapper around sd-device.h.
1426 * A new hwdb database for storing metadata about pointing
1427 stick devices has been added.
1429 * systemd-tmpfiles gained support for setting file attributes
1430 similar to the "chattr" tool with new 'h' and 'H' lines.
1432 * systemd-journald will no longer unconditionally set the
1433 btrfs NOCOW flag on new journal files. This is instead done
1434 with tmpfiles snippet using the new 'h' line type. This
1435 allows easy disabling of this logic, by masking the
1436 journal-nocow.conf tmpfiles file.
1438 * systemd-journald will now translate audit message types to
1439 human readable identifiers when writing them to the
1440 journal. This should improve readability of audit messages.
1442 * The LUKS logic gained support for the offset= and skip=
1443 options in /etc/crypttab, as previously implemented by
1446 * /usr/lib/os-release gained a new optional field VARIANT= for
1447 distributions that support multiple variants (such as a
1448 desktop edition, a server edition, ...)
1450 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
1451 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
1452 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
1453 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
1454 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
1455 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
1456 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
1457 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
1458 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
1459 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
1460 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
1461 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
1462 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
1463 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
1464 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
1465 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
1466 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
1467 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
1468 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
1469 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
1470 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
1471 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
1472 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
1473 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
1474 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
1475 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
1476 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
1478 — Berlin, 2015-05-22
1482 * Introduce a new API "sd-hwdb.h" for querying the hardware
1483 metadata database. With this minimal interface one can query
1484 and enumerate the udev hwdb, decoupled from the old libudev
1485 library. libudev's interface for this is now only a wrapper
1486 around sd-hwdb. A new tool systemd-hwdb has been added to
1487 interface with and update the database.
1489 * When any of systemd's tools copies files (for example due to
1490 tmpfiles' C lines) a btrfs reflink will attempted first,
1491 before bytewise copying is done.
1493 * systemd-nspawn gained a new --ephemeral switch. When
1494 specified a btrfs snapshot is taken of the container's root
1495 directory, and immediately removed when the container
1496 terminates again. Thus, a container can be started whose
1497 changes never alter the container's root directory, and are
1498 lost on container termination. This switch can also be used
1499 for starting a container off the root file system of the
1500 host without affecting the host OS. This switch is only
1501 available on btrfs file systems.
1503 * systemd-nspawn gained a new --template= switch. It takes the
1504 path to a container tree to use as template for the tree
1505 specified via --directory=, should that directory be
1506 missing. This allows instantiating containers dynamically,
1507 on first run. This switch is only available on btrfs file
1510 * When a .mount unit refers to a mount point on which multiple
1511 mounts are stacked, and the .mount unit is stopped all of
1512 the stacked mount points will now be unmounted until no
1513 mount point remains.
1515 * systemd now has an explicit notion of supported and
1516 unsupported unit types. Jobs enqueued for unsupported unit
1517 types will now fail with an "unsupported" error code. More
1518 specifically .swap, .automount and .device units are not
1519 supported in containers, .busname units are not supported on
1520 non-kdbus systems. .swap and .automount are also not
1521 supported if their respective kernel compile time options
1524 * machinectl gained support for two new "copy-from" and
1525 "copy-to" commands for copying files from a running
1526 container to the host or vice versa.
1528 * machinectl gained support for a new "bind" command to bind
1529 mount host directories into local containers. This is
1530 currently only supported for nspawn containers.
1532 * networkd gained support for configuring bridge forwarding
1533 database entries (fdb) from .network files.
1535 * A new tiny daemon "systemd-importd" has been added that can
1536 download container images in tar, raw, qcow2 or dkr formats,
1537 and make them available locally in /var/lib/machines, so
1538 that they can run as nspawn containers. The daemon can GPG
1539 verify the downloads (not supported for dkr, since it has no
1540 provisions for verifying downloads). It will transparently
1541 decompress bz2, xz, gzip compressed downloads if necessary,
1542 and restore sparse files on disk. The daemon uses privilege
1543 separation to ensure the actual download logic runs with
1544 fewer privileges than the daemon itself. machinectl has
1545 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
1546 make the functionality of importd available to the
1547 user. With this in place the Fedora and Ubuntu "Cloud"
1548 images can be downloaded and booted as containers unmodified
1549 (the Fedora images lack the appropriate GPG signature files
1550 currently, so they cannot be verified, but this will change
1551 soon, hopefully). Note that downloading images is currently
1552 only fully supported on btrfs.
1554 * machinectl is now able to list container images found in
1555 /var/lib/machines, along with some metadata about sizes of
1556 disk and similar. If the directory is located on btrfs and
1557 quota is enabled, this includes quota display. A new command
1558 "image-status" has been added that shows additional
1559 information about images.
1561 * machinectl is now able to clone container images
1562 efficiently, if the underlying file system (btrfs) supports
1563 it, with the new "machinectl clone" command. It also
1564 gained commands for renaming and removing images, as well as
1565 marking them read-only or read-write (supported also on
1566 legacy file systems).
1568 * networkd gained support for collecting LLDP network
1569 announcements, from hardware that supports this. This is
1570 shown in networkctl output.
1572 * systemd-run gained support for a new -t (--pty) switch for
1573 invoking a binary on a pty whose input and output is
1574 connected to the invoking terminal. This allows executing
1575 processes as system services while interactively
1576 communicating with them via the terminal. Most interestingly
1577 this is supported across container boundaries. Invoking
1578 "systemd-run -t /bin/bash" is an alternative to running a
1579 full login session, the difference being that the former
1580 will not register a session, nor go through the PAM session
1583 * tmpfiles gained support for a new "v" line type for creating
1584 btrfs subvolumes. If the underlying file system is a legacy
1585 file system, this automatically degrades to creating a
1586 normal directory. Among others /var/lib/machines is now
1587 created like this at boot, should it be missing.
1589 * The directory /var/lib/containers/ has been deprecated and
1590 been replaced by /var/lib/machines. The term "machines" has
1591 been used in the systemd context as generic term for both
1592 VMs and containers, and hence appears more appropriate for
1593 this, as the directory can also contain raw images bootable
1596 * systemd-nspawn when invoked with -M but without --directory=
1597 or --image= is now capable of searching for the container
1598 root directory, subvolume or disk image automatically, in
1599 /var/lib/machines. systemd-nspawn@.service has been updated
1600 to make use of this, thus allowing it to be used for raw
1603 * A new machines.target unit has been introduced that is
1604 supposed to group all containers/VMs invoked as services on
1605 the system. systemd-nspawn@.service has been updated to
1606 integrate with that.
1608 * machinectl gained a new "start" command, for invoking a
1609 container as a service. "machinectl start foo" is mostly
1610 equivalent to "systemctl start systemd-nspawn@foo.service",
1611 but handles escaping in a nicer way.
1613 * systemd-nspawn will now mount most of the cgroupfs tree
1614 read-only into each container, with the exception of the
1615 container's own subtree in the name=systemd hierarchy.
1617 * journald now sets the special FS_NOCOW file flag for its
1618 journal files. This should improve performance on btrfs, by
1619 avoiding heavy fragmentation when journald's write-pattern
1620 is used on COW file systems. It degrades btrfs' data
1621 integrity guarantees for the files to the same levels as for
1622 ext3/ext4 however. This should be OK though as journald does
1623 its own data integrity checks and all its objects are
1624 checksummed on disk. Also, journald should handle btrfs disk
1625 full events a lot more gracefully now, by processing SIGBUS
1626 errors, and not relying on fallocate() anymore.
1628 * When journald detects that journal files it is writing to
1629 have been deleted it will immediately start new journal
1632 * systemd now provides a way to store file descriptors
1633 per-service in PID 1.This is useful for daemons to ensure
1634 that fds they require are not lost during a daemon
1635 restart. The fds are passed to the daemon on the next
1636 invocation in the same way socket activation fds are
1637 passed. This is now used by journald to ensure that the
1638 various sockets connected to all the system's stdout/stderr
1639 are not lost when journald is restarted. File descriptors
1640 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
1641 an extension to sd_notify(). Note that a limit is enforced
1642 on the number of fds a service can store in PID 1, and it
1643 defaults to 0, so that no fds may be stored, unless this is
1644 explicitly turned on.
1646 * The default TERM variable to use for units connected to a
1647 terminal, when no other value is explicitly is set is now
1648 vt220 rather than vt102. This should be fairly safe still,
1649 but allows PgUp/PgDn work.
1651 * The /etc/crypttab option header= as known from Debian is now
1654 * "loginctl user-status" and "loginctl session-status" will
1655 now show the last 10 lines of log messages of the
1656 user/session following the status output. Similar,
1657 "machinectl status" will show the last 10 log lines
1658 associated with a virtual machine or container
1659 service. (Note that this is usually not the log messages
1660 done in the VM/container itself, but simply what the
1661 container manager logs. For nspawn this includes all console
1664 * "loginctl session-status" without further argument will now
1665 show the status of the session of the caller. Similar,
1666 "lock-session", "unlock-session", "activate",
1667 "enable-linger", "disable-linger" may now be called without
1668 session/user parameter in which case they apply to the
1669 caller's session/user.
1671 * An X11 session scriptlet is now shipped that uploads
1672 $DISPLAY and $XAUTHORITY into the environment of the systemd
1673 --user daemon if a session begins. This should improve
1674 compatibility with X11 enabled applications run as systemd
1677 * Generators are now subject to masking via /etc and /run, the
1678 same way as unit files.
1680 * networkd .network files gained support for configuring
1681 per-link IPv4/IPv6 packet forwarding as well as IPv4
1682 masquerading. This is by default turned on for veth links to
1683 containers, as registered by systemd-nspawn. This means that
1684 nspawn containers run with --network-veth will now get
1685 automatic routed access to the host's networks without any
1686 further configuration or setup, as long as networkd runs on
1689 * systemd-nspawn gained the --port= (-p) switch to expose TCP
1690 or UDP posts of a container on the host. With this in place
1691 it is possible to run containers with private veth links
1692 (--network-veth), and have their functionality exposed on
1693 the host as if their services were running directly on the
1696 * systemd-nspawn's --network-veth switch now gained a short
1697 version "-n", since with the changes above it is now truly
1698 useful out-of-the-box. The systemd-nspawn@.service has been
1699 updated to make use of it too by default.
1701 * systemd-nspawn will now maintain a per-image R/W lock, to
1702 ensure that the same image is not started more than once
1703 writable. (It's OK to run an image multiple times
1704 simultaneously in read-only mode.)
1706 * systemd-nspawn's --image= option is now capable of
1707 dissecting and booting MBR and GPT disk images that contain
1708 only a single active Linux partition. Previously it
1709 supported only GPT disk images with proper GPT type
1710 IDs. This allows running cloud images from major
1711 distributions directly with systemd-nspawn, without
1714 * In addition to collecting mouse dpi data in the udev
1715 hardware database, there's now support for collecting angle
1716 information for mouse scroll wheels. The database is
1717 supposed to guarantee similar scrolling behavior on mice
1718 that it knows about. There's also support for collecting
1719 information about Touchpad types.
1721 * udev's input_id built-in will now also collect touch screen
1722 dimension data and attach it to probed devices.
1724 * /etc/os-release gained support for a Distribution Privacy
1727 * networkd gained support for creating "ipvlan", "gretap",
1728 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
1730 * systemd-tmpfiles gained support for "a" lines for setting
1733 * systemd-nspawn will now mount /tmp in the container to
1734 tmpfs, automatically.
1736 * systemd now exposes the memory.usage_in_bytes cgroup
1737 attribute and shows it for each service in the "systemctl
1738 status" output, if available.
1740 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
1741 immediate reboot is triggered. This useful if shutdown is
1742 hung and is unable to complete, to expedite the
1743 operation. Note that this kind of reboot will still unmount
1744 all file systems, and hence should not result in fsck being
1747 * A .device unit for an optical block device will now be
1748 considered active only when a medium is in the drive. Also,
1749 mount units are now bound to their backing devices thus
1750 triggering automatic unmounting when devices become
1751 unavailable. With this in place systemd will now
1752 automatically unmount left-over mounts when a CD-ROM is
1753 ejected or an USB stick is yanked from the system.
1755 * networkd-wait-online now has support for waiting for
1756 specific interfaces only (with globbing), and for giving up
1757 after a configurable timeout.
1759 * networkd now exits when idle. It will be automatically
1760 restarted as soon as interfaces show up, are removed or
1761 change state. networkd will stay around as long as there is
1762 at least one DHCP state machine or similar around, that keep
1765 * networkd may now configure IPv6 link-local addressing in
1766 addition to IPv4 link-local addressing.
1768 * The IPv6 "token" for use in SLAAC may now be configured for
1769 each .network interface in networkd.
1771 * Routes configured with networkd may now be assigned a scope
1774 * networkd's [Match] sections now support globbing and lists
1775 of multiple space-separated matches per item.
1777 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
1778 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
1779 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
1780 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
1781 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
1782 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
1783 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
1784 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
1785 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
1786 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
1787 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
1788 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
1789 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
1790 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
1791 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
1792 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
1793 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
1794 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
1795 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
1796 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
1797 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
1798 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
1799 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
1800 Hoffmann, Zbigniew Jędrzejewski-Szmek
1802 — Berlin, 2015-02-16
1806 * When querying unit file enablement status (for example via
1807 "systemctl is-enabled"), a new state "indirect" is now known
1808 which indicates that a unit might not be enabled itself, but
1809 another unit listed in its Also= setting might be.
1811 * Similar to the various existing ConditionXYZ= settings for
1812 units, there are now matching AssertXYZ= settings. While
1813 failing conditions cause a unit to be skipped, but its job
1814 to succeed, failing assertions declared like this will cause
1815 a unit start operation and its job to fail.
1817 * hostnamed now knows a new chassis type "embedded".
1819 * systemctl gained a new "edit" command. When used on a unit
1820 file, this allows extending unit files with .d/ drop-in
1821 configuration snippets or editing the full file (after
1822 copying it from /usr/lib to /etc). This will invoke the
1823 user's editor (as configured with $EDITOR), and reload the
1824 modified configuration after editing.
1826 * "systemctl status" now shows the suggested enablement state
1827 for a unit, as declared in the (usually vendor-supplied)
1828 system preset files.
1830 * nss-myhostname will now resolve the single-label host name
1831 "gateway" to the locally configured default IP routing
1832 gateways, ordered by their metrics. This assigns a stable
1833 name to the used gateways, regardless which ones are
1834 currently configured. Note that the name will only be
1835 resolved after all other name sources (if nss-myhostname is
1836 configured properly) and should hence not negatively impact
1837 systems that use the single-label host name "gateway" in
1840 * systemd-inhibit now allows filtering by mode when listing
1843 * Scope and service units gained a new "Delegate" boolean
1844 property, which, when set, allows processes running inside the
1845 unit to further partition resources. This is primarily
1846 useful for systemd user instances as well as container
1849 * journald will now pick up audit messages directly from
1850 the kernel, and log them like any other log message. The
1851 audit fields are split up and fully indexed. This means that
1852 journalctl in many ways is now a (nicer!) alternative to
1853 ausearch, the traditional audit client. Note that this
1854 implements only a minimal audit client. If you want the
1855 special audit modes like reboot-on-log-overflow, please use
1856 the traditional auditd instead, which can be used in
1857 parallel to journald.
1859 * The ConditionSecurity= unit file option now understands the
1860 special string "audit" to check whether auditing is
1863 * journalctl gained two new commands --vacuum-size= and
1864 --vacuum-time= to delete old journal files until the
1865 remaining ones take up no more than the specified size on disk,
1866 or are not older than the specified time.
1868 * A new, native PPPoE library has been added to sd-network,
1869 systemd's library of light-weight networking protocols. This
1870 library will be used in a future version of networkd to
1871 enable PPPoE communication without an external pppd daemon.
1873 * The busctl tool now understands a new "capture" verb that
1874 works similar to "monitor", but writes a packet capture
1875 trace to STDOUT that can be redirected to a file which is
1876 compatible with libcap's capture file format. This can then
1877 be loaded in Wireshark and similar tools to inspect bus
1880 * The busctl tool now understands a new "tree" verb that shows
1881 the object trees of a specific service on the bus, or of all
1884 * The busctl tool now understands a new "introspect" verb that
1885 shows all interfaces and members of objects on the bus,
1886 including their signature and values. This is particularly
1887 useful to get more information about bus objects shown by
1888 the new "busctl tree" command.
1890 * The busctl tool now understands new verbs "call",
1891 "set-property" and "get-property" for invoking bus method
1892 calls, setting and getting bus object properties in a
1895 * busctl gained a new --augment-creds= argument that controls
1896 whether the tool shall augment credential information it
1897 gets from the bus with data from /proc, in a possibly
1900 * nspawn's --link-journal= switch gained two new values
1901 "try-guest" and "try-host" that work like "guest" and
1902 "host", but do not fail if the host has no persistent
1903 journalling enabled. -j is now equivalent to
1904 --link-journal=try-guest.
1906 * macvlan network devices created by nspawn will now have
1907 stable MAC addresses.
1909 * A new SmackProcessLabel= unit setting has been added, which
1910 controls the SMACK security label processes forked off by
1911 the respective unit shall use.
1913 * If compiled with --enable-xkbcommon, systemd-localed will
1914 verify x11 keymap settings by compiling the given keymap. It
1915 will spew out warnings if the compilation fails. This
1916 requires libxkbcommon to be installed.
1918 * When a coredump is collected, a larger number of metadata
1919 fields is now collected and included in the journal records
1920 created for it. More specifically, control group membership,
1921 environment variables, memory maps, working directory,
1922 chroot directory, /proc/$PID/status, and a list of open file
1923 descriptors is now stored in the log entry.
1925 * The udev hwdb now contains DPI information for mice. For
1928 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
1930 * All systemd programs that read standalone configuration
1931 files in /etc now also support a corresponding series of
1932 .conf.d configuration directories in /etc/, /run/,
1933 /usr/local/lib/, /usr/lib/, and (if configured with
1934 --enable-split-usr) /lib/. In particular, the following
1935 configuration files now have corresponding configuration
1936 directories: system.conf user.conf, logind.conf,
1937 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
1938 resolved.conf, timesyncd.conf, journal-remote.conf, and
1939 journal-upload.conf. Note that distributions should use the
1940 configuration directories in /usr/lib/; the directories in
1941 /etc/ are reserved for the system administrator.
1943 * systemd-rfkill will no longer take the rfkill device name
1944 into account when storing rfkill state on disk, as the name
1945 might be dynamically assigned and not stable. Instead, the
1946 ID_PATH udev variable combined with the rfkill type (wlan,
1947 bluetooth, ...) is used.
1949 * A new service systemd-machine-id-commit.service has been
1950 added. When used on systems where /etc is read-only during
1951 boot, and /etc/machine-id is not initialized (but an empty
1952 file), this service will copy the temporary machine ID
1953 created as replacement into /etc after the system is fully
1954 booted up. This is useful for systems that are freshly
1955 installed with a non-initialized machine ID, but should get
1956 a fixed machine ID for subsequent boots.
1958 * networkd's .netdev files now provide a large set of
1959 configuration parameters for VXLAN devices. Similarly, the
1960 bridge port cost parameter is now configurable in .network
1961 files. There's also new support for configuring IP source
1962 routing. networkd .link files gained support for a new
1963 OriginalName= match that is useful to match against the
1964 original interface name the kernel assigned. .network files
1965 may include MTU= and MACAddress= fields for altering the MTU
1966 and MAC address while being connected to a specific network
1969 * The LUKS logic gained supported for configuring
1970 UUID-specific key files. There's also new support for naming
1971 LUKS device from the kernel command line, using the new
1972 luks.name= argument.
1974 * Timer units may now be transiently created via the bus API
1975 (this was previously already available for scope and service
1976 units). In addition it is now possible to create multiple
1977 transient units at the same time with a single bus call. The
1978 "systemd-run" tool has been updated to make use of this for
1979 running commands on a specified time, in at(1)-style.
1981 * tmpfiles gained support for "t" lines, for assigning
1982 extended attributes to files. Among other uses this may be
1983 used to assign SMACK labels to files.
1985 Contributions from: Alin Rauta, Alison Chaiken, Andrej
1986 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
1987 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
1988 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
1989 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
1990 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
1991 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
1992 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
1993 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
1994 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
1995 Michael Biebl, Michael Chapman, Michael Marineau, Michal
1996 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
1997 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
1998 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
1999 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
2000 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
2001 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
2002 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
2004 — Berlin, 2014-12-10
2008 * journalctl gained the new options -t/--identifier= to match
2009 on the syslog identifier (aka "tag"), as well as --utc to
2010 show log timestamps in the UTC timezone. journalctl now also
2011 accepts -n/--lines=all to disable line capping in a pager.
2013 * journalctl gained a new switch, --flush, that synchronously
2014 flushes logs from /run/log/journal to /var/log/journal if
2015 persistent storage is enabled. systemd-journal-flush.service
2016 now waits until the operation is complete.
2018 * Services can notify the manager before they start a reload
2019 (by sending RELOADING=1) or shutdown (by sending
2020 STOPPING=1). This allows the manager to track and show the
2021 internal state of daemons and closes a race condition when
2022 the process is still running but has closed its D-Bus
2025 * Services with Type=oneshot do not have to have any ExecStart
2028 * User units are now loaded also from
2029 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
2030 /run/systemd/user directory that was already previously
2031 supported, but is under the control of the user.
2033 * Job timeouts (i.e. time-outs on the time a job that is
2034 queued stays in the run queue) can now optionally result in
2035 immediate reboot or power-off actions (JobTimeoutAction= and
2036 JobTimeoutRebootArgument=). This is useful on ".target"
2037 units, to limit the maximum time a target remains
2038 undispatched in the run queue, and to trigger an emergency
2039 operation in such a case. This is now used by default to
2040 turn off the system if boot-up (as defined by everything in
2041 basic.target) hangs and does not complete for at least
2042 15min. Also, if power-off or reboot hang for at least 30min
2043 an immediate power-off/reboot operation is triggered. This
2044 functionality is particularly useful to increase reliability
2045 on embedded devices, but also on laptops which might
2046 accidentally get powered on when carried in a backpack and
2047 whose boot stays stuck in a hard disk encryption passphrase
2050 * systemd-logind can be configured to also handle lid switch
2051 events even when the machine is docked or multiple displays
2052 are attached (HandleLidSwitchDocked= option).
2054 * A helper binary and a service have been added which can be
2055 used to resume from hibernation in the initramfs. A
2056 generator will parse the resume= option on the kernel
2057 command line to trigger resume.
2059 * A user console daemon systemd-consoled has been
2060 added. Currently, it is a preview, and will so far open a
2061 single terminal on each session of the user marked as
2062 Desktop=systemd-console.
2064 * Route metrics can be specified for DHCP routes added by
2067 * The SELinux context of socket-activated services can be set
2068 from the information provided by the networking stack
2069 (SELinuxContextFromNet= option).
2071 * Userspace firmware loading support has been removed and
2072 the minimum supported kernel version is thus bumped to 3.7.
2074 * Timeout for udev workers has been increased from 1 to 3
2075 minutes, but a warning will be printed after 1 minute to
2076 help diagnose kernel modules that take a long time to load.
2078 * Udev rules can now remove tags on devices with TAG-="foobar".
2080 * systemd's readahead implementation has been removed. In many
2081 circumstances it didn't give expected benefits even for
2082 rotational disk drives and was becoming less relevant in the
2083 age of SSDs. As none of the developers has been using
2084 rotating media anymore, and nobody stepped up to actively
2085 maintain this component of systemd it has now been removed.
2087 * Swap units can use Options= to specify discard options.
2088 Discard options specified for swaps in /etc/fstab are now
2091 * Docker containers are now detected as a separate type of
2094 * The Password Agent protocol gained support for queries where
2095 the user input is shown, useful e.g. for user names.
2096 systemd-ask-password gained a new --echo option to turn that
2099 * The default sysctl.d/ snippets will now set:
2101 net.core.default_qdisc = fq_codel
2103 This selects Fair Queuing Controlled Delay as the default
2104 queuing discipline for network interfaces. fq_codel helps
2105 fight the network bufferbloat problem. It is believed to be
2106 a good default with no tuning required for most workloads.
2107 Downstream distributions may override this choice. On 10Gbit
2108 servers that do not do forwarding, "fq" may perform better.
2109 Systems without a good clocksource should use "pfifo_fast".
2111 * If kdbus is enabled during build a new option BusPolicy= is
2112 available for service units, that allows locking all service
2113 processes into a stricter bus policy, in order to limit
2114 access to various bus services, or even hide most of them
2115 from the service's view entirely.
2117 * networkctl will now show the .network and .link file
2118 networkd has applied to a specific interface.
2120 * sd-login gained a new API call sd_session_get_desktop() to
2121 query which desktop environment has been selected for a
2124 * UNIX utmp support is now compile-time optional to support
2125 legacy-free systems.
2127 * systemctl gained two new commands "add-wants" and
2128 "add-requires" for pulling in units from specific targets
2131 * If the word "rescue" is specified on the kernel command line
2132 the system will now boot into rescue mode (aka
2133 rescue.target), which was previously available only by
2134 specifying "1" or "systemd.unit=rescue.target" on the kernel
2135 command line. This new kernel command line option nicely
2136 mirrors the already existing "emergency" kernel command line
2139 * New kernel command line options mount.usr=, mount.usrflags=,
2140 mount.usrfstype= have been added that match root=, rootflags=,
2141 rootfstype= but allow mounting a specific file system to
2144 * The $NOTIFY_SOCKET is now also passed to control processes of
2145 services, not only the main process.
2147 * This version reenables support for fsck's -l switch. This
2148 means at least version v2.25 of util-linux is required for
2149 operation, otherwise dead-locks on device nodes may
2150 occur. Again: you need to update util-linux to at least
2151 v2.25 when updating systemd to v217.
2153 * The "multi-seat-x" tool has been removed from systemd, as
2154 its functionality has been integrated into X servers 1.16,
2155 and the tool is hence redundant. It is recommended to update
2156 display managers invoking this tool to simply invoke X
2157 directly from now on, again.
2159 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
2160 message flag has been added for all of systemd's PolicyKit
2161 authenticated method calls has been added. In particular
2162 this now allows optional interactive authorization via
2163 PolicyKit for many of PID1's privileged operations such as
2164 unit file enabling and disabling.
2166 * "udevadm hwdb --update" learnt a new switch "--usr" for
2167 placing the rebuilt hardware database in /usr instead of
2168 /etc. When used only hardware database entries stored in
2169 /usr will be used, and any user database entries in /etc are
2170 ignored. This functionality is useful for vendors to ship a
2171 pre-built database on systems where local configuration is
2172 unnecessary or unlikely.
2174 * Calendar time specifications in .timer units now also
2175 understand the strings "semi-annually", "quarterly" and
2176 "minutely" as shortcuts (in addition to the preexisting
2177 "anually", "hourly", ...).
2179 * systemd-tmpfiles will now correctly create files in /dev
2180 at boot which are marked for creation only at boot. It is
2181 recommended to always create static device nodes with 'c!'
2182 and 'b!', so that they are created only at boot and not
2183 overwritten at runtime.
2185 * When the watchdog logic is used for a service (WatchdogSec=)
2186 and the watchdog timeout is hit the service will now be
2187 terminated with SIGABRT (instead of just SIGTERM), in order
2188 to make sure a proper coredump and backtrace is
2189 generated. This ensures that hanging services will result in
2190 similar coredump/backtrace behaviour as services that hit a
2193 Contributions from: Andreas Henriksson, Andrei Borzenkov,
2194 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
2195 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
2196 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
2197 Herrmann, David Sommerseth, David Strauss, Emil Renner
2198 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
2199 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
2200 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
2201 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
2202 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
2203 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
2204 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
2205 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
2206 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
2207 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
2208 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
2209 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
2210 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
2211 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
2212 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
2213 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
2216 — Berlin, 2014-10-28
2220 * timedated no longer reads NTP implementation unit names from
2221 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
2222 implementations should add a
2224 Conflicts=systemd-timesyncd.service
2226 to their unit files to take over and replace systemd's NTP
2227 default functionality.
2229 * systemd-sysusers gained a new line type "r" for configuring
2230 which UID/GID ranges to allocate system users/groups
2231 from. Lines of type "u" may now add an additional column
2232 that specifies the home directory for the system user to be
2233 created. Also, systemd-sysusers may now optionally read user
2234 information from STDIN instead of a file. This is useful for
2235 invoking it from RPM preinst scriptlets that need to create
2236 users before the first RPM file is installed since these
2237 files might need to be owned by them. A new
2238 %sysusers_create_inline RPM macro has been introduced to do
2239 just that. systemd-sysusers now updates the shadow files as
2240 well as the user/group databases, which should enhance
2241 compatibility with certain tools like grpck.
2243 * A number of bus APIs of PID 1 now optionally consult
2244 PolicyKit to permit access for otherwise unprivileged
2245 clients under certain conditions. Note that this currently
2246 doesn't support interactive authentication yet, but this is
2247 expected to be added eventually, too.
2249 * /etc/machine-info now has new fields for configuring the
2250 deployment environment of the machine, as well as the
2251 location of the machine. hostnamectl has been updated with
2252 new command to update these fields.
2254 * systemd-timesyncd has been updated to automatically acquire
2255 NTP server information from systemd-networkd, which might
2256 have been discovered via DHCP.
2258 * systemd-resolved now includes a caching DNS stub resolver
2259 and a complete LLMNR name resolution implementation. A new
2260 NSS module "nss-resolve" has been added which can be used
2261 instead of glibc's own "nss-dns" to resolve hostnames via
2262 systemd-resolved. Hostnames, addresses and arbitrary RRs may
2263 be resolved via systemd-resolved D-Bus APIs. In contrast to
2264 the glibc internal resolver systemd-resolved is aware of
2265 multi-homed system, and keeps DNS server and caches separate
2266 and per-interface. Queries are sent simultaneously on all
2267 interfaces that have DNS servers configured, in order to
2268 properly handle VPNs and local LANs which might resolve
2269 separate sets of domain names. systemd-resolved may acquire
2270 DNS server information from systemd-networkd automatically,
2271 which in turn might have discovered them via DHCP. A tool
2272 "systemd-resolve-host" has been added that may be used to
2273 query the DNS logic in resolved. systemd-resolved implements
2274 IDNA and automatically uses IDNA or UTF-8 encoding depending
2275 on whether classic DNS or LLMNR is used as transport. In the
2276 next releases we intend to add a DNSSEC and mDNS/DNS-SD
2277 implementation to systemd-resolved.
2279 * A new NSS module nss-mymachines has been added, that
2280 automatically resolves the names of all local registered
2281 containers to their respective IP addresses.
2283 * A new client tool "networkctl" for systemd-networkd has been
2284 added. It currently is entirely passive and will query
2285 networking configuration from udev, rtnetlink and networkd,
2286 and present it to the user in a very friendly
2287 way. Eventually, we hope to extend it to become a full
2288 control utility for networkd.
2290 * .socket units gained a new DeferAcceptSec= setting that
2291 controls the kernels' TCP_DEFER_ACCEPT sockopt for
2292 TCP. Similarly, support for controlling TCP keep-alive
2293 settings has been added (KeepAliveTimeSec=,
2294 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
2295 turning off Nagle's algorithm on TCP has been added
2298 * logind learned a new session type "web", for use in projects
2299 like Cockpit which register web clients as PAM sessions.
2301 * timer units with at least one OnCalendar= setting will now
2302 be started only after timer-sync.target has been
2303 reached. This way they will not elapse before the system
2304 clock has been corrected by a local NTP client or
2305 similar. This is particular useful on RTC-less embedded
2306 machines, that come up with an invalid system clock.
2308 * systemd-nspawn's --network-veth= switch should now result in
2309 stable MAC addresses for both the outer and the inner side
2312 * systemd-nspawn gained a new --volatile= switch for running
2313 container instances with /etc or /var unpopulated.
2315 * The kdbus client code has been updated to use the new Linux
2316 3.17 memfd subsystem instead of the old kdbus-specific one.
2318 * systemd-networkd's DHCP client and server now support
2319 FORCERENEW. There are also new configuration options to
2320 configure the vendor client identifier and broadcast mode
2323 * systemd will no longer inform the kernel about the current
2324 timezone, as this is necessarily incorrect and racy as the
2325 kernel has no understanding of DST and similar
2326 concepts. This hence means FAT timestamps will be always
2327 considered UTC, similar to what Android is already
2328 doing. Also, when the RTC is configured to the local time
2329 (rather than UTC) systemd will never synchronize back to it,
2330 as this might confuse Windows at a later boot.
2332 * systemd-analyze gained a new command "verify" for offline
2333 validation of unit files.
2335 * systemd-networkd gained support for a couple of additional
2336 settings for bonding networking setups. Also, the metric for
2337 statically configured routes may now be configured. For
2338 network interfaces where this is appropriate the peer IP
2339 address may now be configured.
2341 * systemd-networkd's DHCP client will no longer request
2342 broadcasting by default, as this tripped up some networks.
2343 For hardware where broadcast is required the feature should
2344 be switched back on using RequestBroadcast=yes.
2346 * systemd-networkd will now set up IPv4LL addresses (when
2347 enabled) even if DHCP is configured successfully.
2349 * udev will now default to respect network device names given
2350 by the kernel when the kernel indicates that these are
2351 predictable. This behavior can be tweaked by changing
2352 NamePolicy= in the relevant .link file.
2354 * A new library systemd-terminal has been added that
2355 implements full TTY stream parsing and rendering. This
2356 library is supposed to be used later on for implementing a
2357 full userspace VT subsystem, replacing the current kernel
2360 * A new tool systemd-journal-upload has been added to push
2361 journal data to a remote system running
2362 systemd-journal-remote.
2364 * journald will no longer forward all local data to another
2365 running syslog daemon. This change has been made because
2366 rsyslog (which appears to be the most commonly used syslog
2367 implementation these days) no longer makes use of this, and
2368 instead pulls the data out of the journal on its own. Since
2369 forwarding the messages to a non-existent syslog server is
2370 more expensive than we assumed we have now turned this
2371 off. If you run a syslog server that is not a recent rsyslog
2372 version, you have to turn this option on again
2373 (ForwardToSyslog= in journald.conf).
2375 * journald now optionally supports the LZ4 compressor for
2376 larger journal fields. This compressor should perform much
2377 better than XZ which was the previous default.
2379 * machinectl now shows the IP addresses of local containers,
2380 if it knows them, plus the interface name of the container.
2382 * A new tool "systemd-escape" has been added that makes it
2383 easy to escape strings to build unit names and similar.
2385 * sd_notify() messages may now include a new ERRNO= field
2386 which is parsed and collected by systemd and shown among the
2387 "systemctl status" output for a service.
2389 * A new component "systemd-firstboot" has been added that
2390 queries the most basic systemd information (timezone,
2391 hostname, root password) interactively on first
2392 boot. Alternatively it may also be used to provision these
2393 things offline on OS images installed into directories.
2395 * The default sysctl.d/ snippets will now set
2397 net.ipv4.conf.default.promote_secondaries=1
2399 This has the benefit of no flushing secondary IP addresses
2400 when primary addresses are removed.
2402 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
2403 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
2404 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
2405 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
2406 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
2407 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
2408 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
2409 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
2410 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
2411 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
2412 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
2413 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
2414 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
2415 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
2416 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
2418 — Berlin, 2014-08-19
2422 * A new tool systemd-sysusers has been added. This tool
2423 creates system users and groups in /etc/passwd and
2424 /etc/group, based on static declarative system user/group
2425 definitions in /usr/lib/sysusers.d/. This is useful to
2426 enable factory resets and volatile systems that boot up with
2427 an empty /etc directory, and thus need system users and
2428 groups created during early boot. systemd now also ships
2429 with two default sysusers.d/ files for the most basic
2430 users and groups systemd and the core operating system
2433 * A new tmpfiles snippet has been added that rebuilds the
2434 essential files in /etc on boot, should they be missing.
2436 * A directive for ensuring automatic clean-up of
2437 /var/cache/man/ has been removed from the default
2438 configuration. This line should now be shipped by the man
2439 implementation. The necessary change has been made to the
2440 man-db implementation. Note that you need to update your man
2441 implementation to one that ships this line, otherwise no
2442 automatic clean-up of /var/cache/man will take place.
2444 * A new condition ConditionNeedsUpdate= has been added that
2445 may conditionalize services to only run when /etc or /var
2446 are "older" than the vendor operating system resources in
2447 /usr. This is useful for reconstructing or updating /etc
2448 after an offline update of /usr or a factory reset, on the
2449 next reboot. Services that want to run once after such an
2450 update or reset should use this condition and order
2451 themselves before the new systemd-update-done.service, which
2452 will mark the two directories as fully updated. A number of
2453 service files have been added making use of this, to rebuild
2454 the udev hardware database, the journald message catalog and
2455 dynamic loader cache (ldconfig). The systemd-sysusers tool
2456 described above also makes use of this now. With this in
2457 place it is now possible to start up a minimal operating
2458 system with /etc empty cleanly. For more information on the
2459 concepts involved see this recent blog story:
2461 http://0pointer.de/blog/projects/stateless.html
2463 * A new system group "input" has been introduced, and all
2464 input device nodes get this group assigned. This is useful
2465 for system-level software to get access to input devices. It
2466 complements what is already done for "audio" and "video".
2468 * systemd-networkd learnt minimal DHCPv4 server support in
2469 addition to the existing DHCPv4 client support. It also
2470 learnt DHCPv6 client and IPv6 Router Solicitation client
2471 support. The DHCPv4 client gained support for static routes
2472 passed in from the server. Note that the [DHCPv4] section
2473 known in older systemd-networkd versions has been renamed to
2474 [DHCP] and is now also used by the DHCPv6 client. Existing
2475 .network files using settings of this section should be
2476 updated, though compatibility is maintained. Optionally, the
2477 client hostname may now be sent to the DHCP server.
2479 * networkd gained support for vxlan virtual networks as well
2480 as tun/tap and dummy devices.
2482 * networkd gained support for automatic allocation of address
2483 ranges for interfaces from a system-wide pool of
2484 addresses. This is useful for dynamically managing a large
2485 number of interfaces with a single network configuration
2486 file. In particular this is useful to easily assign
2487 appropriate IP addresses to the veth links of a large number
2488 of nspawn instances.
2490 * RPM macros for processing sysusers, sysctl and binfmt
2491 drop-in snippets at package installation time have been
2494 * The /etc/os-release file should now be placed in
2495 /usr/lib/os-release. The old location is automatically
2496 created as symlink. /usr/lib is the more appropriate
2497 location of this file, since it shall actually describe the
2498 vendor operating system shipped in /usr, and not the
2499 configuration stored in /etc.
2501 * .mount units gained a new boolean SloppyOptions= setting
2502 that maps to mount(8)'s -s option which enables permissive
2503 parsing of unknown mount options.
2505 * tmpfiles learnt a new "L+" directive which creates a symlink
2506 but (unlike "L") deletes a pre-existing file first, should
2507 it already exist and not already be the correct
2508 symlink. Similarly, "b+", "c+" and "p+" directives have been
2509 added as well, which create block and character devices, as
2510 well as fifos in the filesystem, possibly removing any
2511 pre-existing files of different types.
2513 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
2514 'argument' field (which so far specified the source to
2515 symlink/copy the files from) is now optional. If omitted the
2516 same file os copied from /usr/share/factory/ suffixed by the
2517 full destination path. This is useful for populating /etc
2518 with essential files, by copying them from vendor defaults
2519 shipped in /usr/share/factory/etc.
2521 * A new command "systemctl preset-all" has been added that
2522 applies the service preset settings to all installed unit
2523 files. A new switch --preset-mode= has been added that
2524 controls whether only enable or only disable operations
2527 * A new command "systemctl is-system-running" has been added
2528 that allows checking the overall state of the system, for
2529 example whether it is fully up and running.
2531 * When the system boots up with an empty /etc, the equivalent
2532 to "systemctl preset-all" is executed during early boot, to
2533 make sure all default services are enabled after a factory
2536 * systemd now contains a minimal preset file that enables the
2537 most basic services systemd ships by default.
2539 * Unit files' [Install] section gained a new DefaultInstance=
2540 field for defining the default instance to create if a
2541 template unit is enabled with no instance specified.
2543 * A new passive target cryptsetup-pre.target has been added
2544 that may be used by services that need to make they run and
2545 finish before the first LUKS cryptographic device is set up.
2547 * The /dev/loop-control and /dev/btrfs-control device nodes
2548 are now owned by the "disk" group by default, opening up
2549 access to this group.
2551 * systemd-coredump will now automatically generate a
2552 stack trace of all core dumps taking place on the system,
2553 based on elfutils' libdw library. This stack trace is logged
2556 * systemd-coredump may now optionally store coredumps directly
2557 on disk (in /var/lib/systemd/coredump, possibly compressed),
2558 instead of storing them unconditionally in the journal. This
2559 mode is the new default. A new configuration file
2560 /etc/systemd/coredump.conf has been added to configure this
2561 and other parameters of systemd-coredump.
2563 * coredumpctl gained a new "info" verb to show details about a
2564 specific coredump. A new switch "-1" has also been added
2565 that makes sure to only show information about the most
2566 recent entry instead of all entries. Also, as the tool is
2567 generally useful now the "systemd-" prefix of the binary
2568 name has been removed. Distributions that want to maintain
2569 compatibility with the old name should add a symlink from
2570 the old name to the new name.
2572 * journald's SplitMode= now defaults to "uid". This makes sure
2573 that unprivileged users can access their own coredumps with
2574 coredumpctl without restrictions.
2576 * New kernel command line options "systemd.wants=" (for
2577 pulling an additional unit during boot), "systemd.mask="
2578 (for masking a specific unit for the boot), and
2579 "systemd.debug-shell" (for enabling the debug shell on tty9)
2580 have been added. This is implemented in the new generator
2581 "systemd-debug-generator".
2583 * systemd-nspawn will now by default filter a couple of
2584 syscalls for containers, among them those required for
2585 kernel module loading, direct x86 IO port access, swap
2586 management, and kexec. Most importantly though
2587 open_by_handle_at() is now prohibited for containers,
2588 closing a hole similar to a recently discussed vulnerability
2589 in docker regarding access to files on file hierarchies the
2590 container should normally not have access to. Note that, for
2591 nspawn, we generally make no security claims anyway (and
2592 this is explicitly documented in the man page), so this is
2593 just a fix for one of the most obvious problems.
2595 * A new man page file-hierarchy(7) has been added that
2596 contains a minimized, modernized version of the file system
2597 layout systemd expects, similar in style to the FHS
2598 specification or hier(5). A new tool systemd-path(1) has
2599 been added to query many of these paths for the local
2602 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
2603 longer done. Since the directory now has a per-user size
2604 limit, and is cleaned on logout this appears unnecessary,
2605 in particular since this now brings the lifecycle of this
2606 directory closer in line with how IPC objects are handled.
2608 * systemd.pc now exports a number of additional directories,
2609 including $libdir (which is useful to identify the library
2610 path for the primary architecture of the system), and a
2611 couple of drop-in directories.
2613 * udev's predictable network interface names now use the dev_port
2614 sysfs attribute, introduced in linux 3.15 instead of dev_id to
2615 distinguish between ports of the same PCI function. dev_id should
2616 only be used for ports using the same HW address, hence the need
2619 * machined has been updated to export the OS version of a
2620 container (read from /etc/os-release and
2621 /usr/lib/os-release) on the bus. This is now shown in
2622 "machinectl status" for a machine.
2624 * A new service setting RestartForceExitStatus= has been
2625 added. If configured to a set of exit signals or process
2626 return values, the service will be restarted when the main
2627 daemon process exits with any of them, regardless of the
2630 * systemctl's -H switch for connecting to remote systemd
2631 machines has been extended so that it may be used to
2632 directly connect to a specific container on the
2633 host. "systemctl -H root@foobar:waldi" will now connect as
2634 user "root" to host "foobar", and then proceed directly to
2635 the container named "waldi". Note that currently you have to
2636 authenticate as user "root" for this to work, as entering
2637 containers is a privileged operation.
2639 Contributions from: Andreas Henriksson, Benjamin Steinwender,
2640 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
2641 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
2642 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
2643 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
2644 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
2645 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
2646 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
2647 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
2648 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
2649 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
2650 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
2652 — Berlin, 2014-07-03
2656 * As an experimental feature, udev now tries to lock the
2657 disk device node (flock(LOCK_SH|LOCK_NB)) while it
2658 executes events for the disk or any of its partitions.
2659 Applications like partitioning programs can lock the
2660 disk device node (flock(LOCK_EX)) and claim temporary
2661 device ownership that way; udev will entirely skip all event
2662 handling for this disk and its partitions. If the disk
2663 was opened for writing, the close will trigger a partition
2664 table rescan in udev's "watch" facility, and if needed
2665 synthesize "change" events for the disk and all its partitions.
2666 This is now unconditionally enabled, and if it turns out to
2667 cause major problems, we might turn it on only for specific
2668 devices, or might need to disable it entirely. Device Mapper
2669 devices are excluded from this logic.
2671 * We temporarily dropped the "-l" switch for fsck invocations,
2672 since they collide with the flock() logic above. util-linux
2673 upstream has been changed already to avoid this conflict,
2674 and we will readd "-l" as soon as util-linux with this
2675 change has been released.
2677 * The dependency on libattr has been removed. Since a long
2678 time, the extended attribute calls have moved to glibc, and
2679 libattr is thus unnecessary.
2681 * Virtualization detection works without priviliges now. This
2682 means the systemd-detect-virt binary no longer requires
2683 CAP_SYS_PTRACE file capabilities, and our daemons can run
2684 with fewer privileges.
2686 * systemd-networkd now runs under its own "systemd-network"
2687 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
2688 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
2689 loses the ability to write to files owned by root this way.
2691 * Similarly, systemd-resolved now runs under its own
2692 "systemd-resolve" user with no capabilities remaining.
2694 * Similarly, systemd-bus-proxyd now runs under its own
2695 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
2697 * systemd-networkd gained support for setting up "veth"
2698 virtual Ethernet devices for container connectivity, as well
2699 as GRE and VTI tunnels.
2701 * systemd-networkd will no longer automatically attempt to
2702 manually load kernel modules necessary for certain tunnel
2703 transports. Instead, it is assumed the kernel loads them
2704 automatically when required. This only works correctly on
2705 very new kernels. On older kernels, please consider adding
2706 the kernel modules to /etc/modules-load.d/ as a work-around.
2708 * The resolv.conf file systemd-resolved generates has been
2709 moved to /run/systemd/resolve/. If you have a symlink from
2710 /etc/resolv.conf, it might be necessary to correct it.
2712 * Two new service settings, ProtectHome= and ProtectSystem=,
2713 have been added. When enabled, they will make the user data
2714 (such as /home) inaccessible or read-only and the system
2715 (such as /usr) read-only, for specific services. This allows
2716 very light-weight per-service sandboxing to avoid
2717 modifications of user data or system files from
2718 services. These two new switches have been enabled for all
2719 of systemd's long-running services, where appropriate.
2721 * Socket units gained new SocketUser= and SocketGroup=
2722 settings to set the owner user and group of AF_UNIX sockets
2723 and FIFOs in the file system.
2725 * Socket units gained a new RemoveOnStop= setting. If enabled,
2726 all FIFOS and sockets in the file system will be removed
2727 when the specific socket unit is stopped.
2729 * Socket units gained a new Symlinks= setting. It takes a list
2730 of symlinks to create to file system sockets or FIFOs
2731 created by the specific Unix sockets. This is useful to
2732 manage symlinks to socket nodes with the same life-cycle as
2735 * The /dev/log socket and /dev/initctl FIFO have been moved to
2736 /run, and have been replaced by symlinks. This allows
2737 connecting to these facilities even if PrivateDevices=yes is
2738 used for a service (which makes /dev/log itself unavailable,
2739 but /run is left). This also has the benefit of ensuring
2740 that /dev only contains device nodes, directories and
2741 symlinks, and nothing else.
2743 * sd-daemon gained two new calls sd_pid_notify() and
2744 sd_pid_notifyf(). They are similar to sd_notify() and
2745 sd_notifyf(), but allow overriding of the source PID of
2746 notification messages if permissions permit this. This is
2747 useful to send notify messages on behalf of a different
2748 process (for example, the parent process). The
2749 systemd-notify tool has been updated to make use of this
2750 when sending messages (so that notification messages now
2751 originate from the shell script invoking systemd-notify and
2752 not the systemd-notify process itself. This should minimize
2753 a race where systemd fails to associate notification
2754 messages to services when the originating process already
2757 * A new "on-abnormal" setting for Restart= has been added. If
2758 set, it will result in automatic restarts on all "abnormal"
2759 reasons for a process to exit, which includes unclean
2760 signals, core dumps, timeouts and watchdog timeouts, but
2761 does not include clean and unclean exit codes or clean
2762 signals. Restart=on-abnormal is an alternative for
2763 Restart=on-failure for services that shall be able to
2764 terminate and avoid restarts on certain errors, by
2765 indicating so with an unclean exit code. Restart=on-failure
2766 or Restart=on-abnormal is now the recommended setting for
2767 all long-running services.
2769 * If the InaccessibleDirectories= service setting points to a
2770 mount point (or if there are any submounts contained within
2771 it), it is now attempted to completely unmount it, to make
2772 the file systems truly unavailable for the respective
2775 * The ReadOnlyDirectories= service setting and
2776 systemd-nspawn's --read-only parameter are now recursively
2777 applied to all submounts, too.
2779 * Mount units may now be created transiently via the bus APIs.
2781 * The support for SysV and LSB init scripts has been removed
2782 from the systemd daemon itself. Instead, it is now
2783 implemented as a generator that creates native systemd units
2784 from these scripts when needed. This enables us to remove a
2785 substantial amount of legacy code from PID 1, following the
2786 fact that many distributions only ship a very small number
2787 of LSB/SysV init scripts nowadays.
2789 * Privileged Xen (dom0) domains are not considered
2790 virtualization anymore by the virtualization detection
2791 logic. After all, they generally have unrestricted access to
2792 the hardware and usually are used to manage the unprivileged
2795 * systemd-tmpfiles gained a new "C" line type, for copying
2796 files or entire directories.
2798 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
2799 lines. So far, they have been non-globbing versions of the
2800 latter, and have thus been redundant. In future, it is
2801 recommended to only use "z". "m" has hence been removed
2802 from the documentation, even though it stays supported.
2804 * A tmpfiles snippet to recreate the most basic structure in
2805 /var has been added. This is enough to create the /var/run →
2806 /run symlink and create a couple of structural
2807 directories. This allows systems to boot up with an empty or
2808 volatile /var. Of course, while with this change, the core OS
2809 now is capable with dealing with a volatile /var, not all
2810 user services are ready for it. However, we hope that sooner
2811 or later, many service daemons will be changed upstream so
2812 that they are able to automatically create their necessary
2813 directories in /var at boot, should they be missing. This is
2814 the first step to allow state-less systems that only require
2815 the vendor image for /usr to boot.
2817 * systemd-nspawn has gained a new --tmpfs= switch to mount an
2818 empty tmpfs instance to a specific directory. This is
2819 particularly useful for making use of the automatic
2820 reconstruction of /var (see above), by passing --tmpfs=/var.
2822 * Access modes specified in tmpfiles snippets may now be
2823 prefixed with "~", which indicates that they shall be masked
2824 by whether the existing file or directory is currently
2825 writable, readable or executable at all. Also, if specified,
2826 the sgid/suid/sticky bits will be masked for all
2829 * A new passive target unit "network-pre.target" has been
2830 added which is useful for services that shall run before any
2831 network is configured, for example firewall scripts.
2833 * The "floppy" group that previously owned the /dev/fd*
2834 devices is no longer used. The "disk" group is now used
2835 instead. Distributions should probably deprecate usage of
2838 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
2839 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
2840 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
2841 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
2842 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
2843 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
2846 — Berlin, 2014-06-11
2850 * A new "systemd-timesyncd" daemon has been added for
2851 synchronizing the system clock across the network. It
2852 implements an SNTP client. In contrast to NTP
2853 implementations such as chrony or the NTP reference server,
2854 this only implements a client side, and does not bother with
2855 the full NTP complexity, focusing only on querying time from
2856 one remote server and synchronizing the local clock to
2857 it. Unless you intend to serve NTP to networked clients or
2858 want to connect to local hardware clocks, this simple NTP
2859 client should be more than appropriate for most
2860 installations. The daemon runs with minimal privileges, and
2861 has been hooked up with networkd to only operate when
2862 network connectivity is available. The daemon saves the
2863 current clock to disk every time a new NTP sync has been
2864 acquired, and uses this to possibly correct the system clock
2865 early at bootup, in order to accommodate for systems that
2866 lack an RTC such as the Raspberry Pi and embedded devices,
2867 and to make sure that time monotonically progresses on these
2868 systems, even if it is not always correct. To make use of
2869 this daemon, a new system user and group "systemd-timesync"
2870 needs to be created on installation of systemd.
2872 * The queue "seqnum" interface of libudev has been disabled, as
2873 it was generally incompatible with device namespacing as
2874 sequence numbers of devices go "missing" if the devices are
2875 part of a different namespace.
2877 * "systemctl list-timers" and "systemctl list-sockets" gained
2878 a --recursive switch for showing units of these types also
2879 for all local containers, similar in style to the already
2880 supported --recursive switch for "systemctl list-units".
2882 * A new RebootArgument= setting has been added for service
2883 units, which may be used to specify a kernel reboot argument
2884 to use when triggering reboots with StartLimitAction=.
2886 * A new FailureAction= setting has been added for service
2887 units which may be used to specify an operation to trigger
2888 when a service fails. This works similarly to
2889 StartLimitAction=, but unlike it, controls what is done
2890 immediately rather than only after several attempts to
2891 restart the service in question.
2893 * hostnamed got updated to also expose the kernel name,
2894 release, and version on the bus. This is useful for
2895 executing commands like hostnamectl with the -H switch.
2896 systemd-analyze makes use of this to properly display
2897 details when running non-locally.
2899 * The bootchart tool can now show cgroup information in the
2900 graphs it generates.
2902 * The CFS CPU quota cgroup attribute is now exposed for
2903 services. The new CPUQuota= switch has been added for this
2904 which takes a percentage value. Setting this will have the
2905 result that a service may never get more CPU time than the
2906 specified percentage, even if the machine is otherwise idle.
2908 * systemd-networkd learned IPIP and SIT tunnel support.
2910 * LSB init scripts exposing a dependency on $network will now
2911 get a dependency on network-online.target rather than simply
2912 network.target. This should bring LSB handling closer to
2913 what it was on SysV systems.
2915 * A new fsck.repair= kernel option has been added to control
2916 how fsck shall deal with unclean file systems at boot.
2918 * The (.ini) configuration file parser will now silently
2919 ignore sections whose name begins with "X-". This may be
2920 used to maintain application-specific extension sections in unit
2923 * machined gained a new API to query the IP addresses of
2924 registered containers. "machinectl status" has been updated
2925 to show these addresses in its output.
2927 * A new call sd_uid_get_display() has been added to the
2928 sd-login APIs for querying the "primary" session of a
2929 user. The "primary" session of the user is elected from the
2930 user's sessions and generally a graphical session is
2931 preferred over a text one.
2933 * A minimal systemd-resolved daemon has been added. It
2934 currently simply acts as a companion to systemd-networkd and
2935 manages resolv.conf based on per-interface DNS
2936 configuration, possibly supplied via DHCP. In the long run
2937 we hope to extend this into a local DNSSEC enabled DNS and
2940 * The systemd-networkd-wait-online tool is now enabled by
2941 default. It will delay network-online.target until a network
2942 connection has been configured. The tool primarily integrates
2943 with networkd, but will also make a best effort to make sense
2944 of network configuration performed in some other way.
2946 * Two new service options StartupCPUShares= and
2947 StartupBlockIOWeight= have been added that work similarly to
2948 CPUShares= and BlockIOWeight= however only apply during
2949 system startup. This is useful to prioritize certain services
2950 differently during bootup than during normal runtime.
2952 * hostnamed has been changed to prefer the statically
2953 configured hostname in /etc/hostname (unless set to
2954 'localhost' or empty) over any dynamic one supplied by
2955 dhcp. With this change, the rules for picking the hostname
2956 match more closely the rules of other configuration settings
2957 where the local administrator's configuration in /etc always
2958 overrides any other settings.
2960 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
2961 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
2962 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
2963 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
2964 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
2965 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
2966 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
2967 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
2968 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
2969 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
2970 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
2971 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
2972 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
2973 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
2974 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
2975 Lindskog, WaLyong Cho, Will Woods, Zbigniew
2978 — Beijing, 2014-05-28
2982 * When restoring the screen brightness at boot, stay away from
2983 the darkest setting or from the lowest 5% of the available
2984 range, depending on which is the larger value of both. This
2985 should effectively protect the user from rebooting into a
2986 black screen, should the brightness have been set to minimum
2989 * sd-login gained a new sd_machine_get_class() call to
2990 determine the class ("vm" or "container") of a machine
2991 registered with machined.
2993 * sd-login gained new calls
2994 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
2995 to query the identity of the peer of a local AF_UNIX
2996 connection. They operate similarly to their sd_pid_get_xyz()
2999 * PID 1 will now maintain a system-wide system state engine
3000 with the states "starting", "running", "degraded",
3001 "maintenance", "stopping". These states are bound to system
3002 startup, normal runtime, runtime with at least one failed
3003 service, rescue/emergency mode and system shutdown. This
3004 state is shown in the "systemctl status" output when no unit
3005 name is passed. It is useful to determine system state, in
3006 particularly when doing so for many systems or containers at
3009 * A new command "list-machines" has been added to "systemctl"
3010 that lists all local OS containers and shows their system
3011 state (see above), if systemd runs inside of them.
3013 * systemctl gained a new "-r" switch to recursively enumerate
3014 units on all local containers, when used with the
3015 "list-unit" command (which is the default one that is
3016 executed when no parameters are specified).
3018 * The GPT automatic partition discovery logic will now honour
3019 two GPT partition flags: one may be set on a partition to
3020 cause it to be mounted read-only, and the other may be set
3021 on a partition to ignore it during automatic discovery.
3023 * Two new GPT type UUIDs have been added for automatic root
3024 partition discovery, for 32-bit and 64-bit ARM. This is not
3025 particularly useful for discovering the root directory on
3026 these architectures during bare-metal boots (since UEFI is
3027 not common there), but still very useful to allow booting of
3028 ARM disk images in nspawn with the -i option.
3030 * MAC addresses of interfaces created with nspawn's
3031 --network-interface= switch will now be generated from the
3032 machine name, and thus be stable between multiple invocations
3035 * logind will now automatically remove all IPC objects owned
3036 by a user if she or he fully logs out. This makes sure that
3037 users who are logged out cannot continue to consume IPC
3038 resources. This covers SysV memory, semaphores and message
3039 queues as well as POSIX shared memory and message
3040 queues. Traditionally, SysV and POSIX IPC had no life-cycle
3041 limits. With this functionality, that is corrected. This may
3042 be turned off by using the RemoveIPC= switch of logind.conf.
3044 * The systemd-machine-id-setup and tmpfiles tools gained a
3045 --root= switch to operate on a specific root directory,
3048 * journald can now forward logged messages to the TTYs of all
3049 logged in users ("wall"). This is the default for all
3050 emergency messages now.
3052 * A new tool systemd-journal-remote has been added to stream
3053 journal log messages across the network.
3055 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
3056 controller trees are mounted into it. Note that the
3057 directories mounted beneath it are not read-only. This is a
3058 security measure and is particularly useful because glibc
3059 actually includes a search logic to pick any tmpfs it can
3060 find to implement shm_open() if /dev/shm is not available
3061 (which it might very well be in namespaced setups).
3063 * machinectl gained a new "poweroff" command to cleanly power
3064 down a local OS container.
3066 * The PrivateDevices= unit file setting will now also drop the
3067 CAP_MKNOD capability from the capability bound set, and
3068 imply DevicePolicy=closed.
3070 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
3071 comprehensively on all long-running systemd services where
3072 this is appropriate.
3074 * systemd-udevd will now run in a disassociated mount
3075 namespace. To mount directories from udev rules, make sure to
3076 pull in mount units via SYSTEMD_WANTS properties.
3078 * The kdbus support gained support for uploading policy into
3079 the kernel. sd-bus gained support for creating "monitoring"
3080 connections that can eavesdrop into all bus communication
3081 for debugging purposes.
3083 * Timestamps may now be specified in seconds since the UNIX
3084 epoch Jan 1st, 1970 by specifying "@" followed by the value
3087 * Native tcpwrap support in systemd has been removed. tcpwrap
3088 is old code, not really maintained anymore and has serious
3089 shortcomings, and better options such as firewalls
3090 exist. For setups that require tcpwrap usage, please
3091 consider invoking your socket-activated service via tcpd,
3092 like on traditional inetd.
3094 * A new system.conf configuration option
3095 DefaultTimerAccuracySec= has been added that controls the
3096 default AccuracySec= setting of .timer units.
3098 * Timer units gained a new WakeSystem= switch. If enabled,
3099 timers configured this way will cause the system to resume
3100 from system suspend (if the system supports that, which most
3103 * Timer units gained a new Persistent= switch. If enabled,
3104 timers configured this way will save to disk when they have
3105 been last triggered. This information is then used on next
3106 reboot to possible execute overdue timer events, that
3107 could not take place because the system was powered off.
3108 This enables simple anacron-like behaviour for timer units.
3110 * systemctl's "list-timers" will now also list the time a
3111 timer unit was last triggered in addition to the next time
3112 it will be triggered.
3114 * systemd-networkd will now assign predictable IPv4LL
3115 addresses to its local interfaces.
3117 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
3118 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
3119 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
3120 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
3121 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
3122 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
3123 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
3124 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
3127 — Berlin, 2014-03-25
3131 * A new unit file setting RestrictAddressFamilies= has been
3132 added to restrict which socket address families unit
3133 processes gain access to. This takes address family names
3134 like "AF_INET" or "AF_UNIX", and is useful to minimize the
3135 attack surface of services via exotic protocol stacks. This
3136 is built on seccomp system call filters.
3138 * Two new unit file settings RuntimeDirectory= and
3139 RuntimeDirectoryMode= have been added that may be used to
3140 manage a per-daemon runtime directories below /run. This is
3141 an alternative for setting up directory permissions with
3142 tmpfiles snippets, and has the advantage that the runtime
3143 directory's lifetime is bound to the daemon runtime and that
3144 the daemon starts up with an empty directory each time. This
3145 is particularly useful when writing services that drop
3146 privileges using the User= or Group= setting.
3148 * The DeviceAllow= unit setting now supports globbing for
3149 matching against device group names.
3151 * The systemd configuration file system.conf gained new
3152 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
3153 DefaultMemoryAccounting= to globally turn on/off accounting
3154 for specific resources (cgroups) for all units. These
3155 settings may still be overridden individually in each unit
3158 * systemd-gpt-auto-generator is now able to discover /srv and
3159 root partitions in addition to /home and swap partitions. It
3160 also supports LUKS-encrypted partitions now. With this in
3161 place, automatic discovery of partitions to mount following
3162 the Discoverable Partitions Specification
3163 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
3164 is now a lot more complete. This allows booting without
3165 /etc/fstab and without root= on the kernel command line on
3166 systems prepared appropriately.
3168 * systemd-nspawn gained a new --image= switch which allows
3169 booting up disk images and Linux installations on any block
3170 device that follow the Discoverable Partitions Specification
3171 (see above). This means that installations made with
3172 appropriately updated installers may now be started and
3173 deployed using container managers, completely
3174 unmodified. (We hope that libvirt-lxc will add support for
3175 this feature soon, too.)
3177 * systemd-nspawn gained a new --network-macvlan= setting to
3178 set up a private macvlan interface for the
3179 container. Similarly, systemd-networkd gained a new
3180 Kind=macvlan setting in .netdev files.
3182 * systemd-networkd now supports configuring local addresses
3185 * A new tool systemd-network-wait-online has been added to
3186 synchronously wait for network connectivity using
3189 * The sd-bus.h bus API gained a new sd_bus_track object for
3190 tracking the life-cycle of bus peers. Note that sd-bus.h is
3191 still not a public API though (unless you specify
3192 --enable-kdbus on the configure command line, which however
3193 voids your warranty and you get no API stability guarantee).
3195 * The $XDG_RUNTIME_DIR runtime directories for each user are
3196 now individual tmpfs instances, which has the benefit of
3197 introducing separate pools for each user, with individual
3198 size limits, and thus making sure that unprivileged clients
3199 can no longer negatively impact the system or other users by
3200 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
3201 RuntimeDirectorySize= has been introduced that allows
3202 controlling the default size limit for all users. It
3203 defaults to 10% of the available physical memory. This is no
3204 replacement for quotas on tmpfs though (which the kernel
3205 still does not support), as /dev/shm and /tmp are still
3206 shared resources used by both the system and unprivileged
3209 * logind will now automatically turn off automatic suspending
3210 on laptop lid close when more than one display is
3211 connected. This was previously expected to be implemented
3212 individually in desktop environments (such as GNOME),
3213 however has been added to logind now, in order to fix a
3214 boot-time race where a desktop environment might not have
3215 been started yet and thus not been able to take an inhibitor
3216 lock at the time where logind already suspends the system
3217 due to a closed lid.
3219 * logind will now wait at least 30s after each system
3220 suspend/resume cycle, and 3min after system boot before
3221 suspending the system due to a closed laptop lid. This
3222 should give USB docking stations and similar enough time to
3223 be probed and configured after system resume and boot in
3224 order to then act as suspend blocker.
3226 * systemd-run gained a new --property= setting which allows
3227 initialization of resource control properties (and others)
3228 for the created scope or service unit. Example: "systemd-run
3229 --property=BlockIOWeight=10 updatedb" may be used to run
3230 updatedb at a low block IO scheduling weight.
3232 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
3233 now also work in --scope mode.
3235 * When systemd is compiled with kdbus support, basic support
3236 for enforced policies is now in place. (Note that enabling
3237 kdbus still voids your warranty and no API compatibility
3240 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
3241 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
3242 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
3243 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
3244 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
3245 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
3246 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
3247 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
3248 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
3249 Zbigniew Jędrzejewski-Szmek
3251 — Berlin, 2014-03-12
3255 * systemd will now relabel /dev after loading the SMACK policy
3256 according to SMACK rules.
3258 * A new unit file option AppArmorProfile= has been added to
3259 set the AppArmor profile for the processes of a unit.
3261 * A new condition check ConditionArchitecture= has been added
3262 to conditionalize units based on the system architecture, as
3263 reported by uname()'s "machine" field.
3265 * systemd-networkd now supports matching on the system
3266 virtualization, architecture, kernel command line, host name
3269 * logind is now a lot more aggressive when suspending the
3270 machine due to a closed laptop lid. Instead of acting only
3271 on the lid close action, it will continuously watch the lid
3272 status and act on it. This is useful for laptops where the
3273 power button is on the outside of the chassis so that it can
3274 be reached without opening the lid (such as the Lenovo
3275 Yoga). On those machines, logind will now immediately
3276 re-suspend the machine if the power button has been
3277 accidentally pressed while the laptop was suspended and in a
3278 backpack or similar.
3280 * logind will now watch SW_DOCK switches and inhibit reaction
3281 to the lid switch if it is pressed. This means that logind
3282 will not suspend the machine anymore if the lid is closed
3283 and the system is docked, if the laptop supports SW_DOCK
3284 notifications via the input layer. Note that ACPI docking
3285 stations do not generate this currently. Also note that this
3286 logic is usually not fully sufficient and Desktop
3287 Environments should take a lid switch inhibitor lock when an
3288 external display is connected, as systemd will not watch
3291 * nspawn will now make use of the devices cgroup controller by
3292 default, and only permit creation of and access to the usual
3293 API device nodes like /dev/null or /dev/random, as well as
3294 access to (but not creation of) the pty devices.
3296 * We will now ship a default .network file for
3297 systemd-networkd that automatically configures DHCP for
3298 network interfaces created by nspawn's --network-veth or
3299 --network-bridge= switches.
3301 * systemd will now understand the usual M, K, G, T suffixes
3302 according to SI conventions (i.e. to the base 1000) when
3303 referring to throughput and hardware metrics. It will stay
3304 with IEC conventions (i.e. to the base 1024) for software
3305 metrics, according to what is customary according to
3306 Wikipedia. We explicitly document which base applies for
3307 each configuration option.
3309 * The DeviceAllow= setting in unit files now supports a syntax
3310 to whitelist an entire group of devices node majors at once,
3311 based on the /proc/devices listing. For example, with the
3312 string "char-pts", it is now possible to whitelist all
3313 current and future pseudo-TTYs at once.
3315 * sd-event learned a new "post" event source. Event sources of
3316 this type are triggered by the dispatching of any event
3317 source of a type that is not "post". This is useful for
3318 implementing clean-up and check event sources that are
3319 triggered by other work being done in the program.
3321 * systemd-networkd is no longer statically enabled, but uses
3322 the usual [Install] sections so that it can be
3323 enabled/disabled using systemctl. It still is enabled by
3326 * When creating a veth interface pair with systemd-nspawn, the
3327 host side will now be prefixed with "vb-" if
3328 --network-bridge= is used, and with "ve-" if --network-veth
3329 is used. This way, it is easy to distinguish these cases on
3330 the host, for example to apply different configuration to
3331 them with systemd-networkd.
3333 * The compatibility libraries for libsystemd-journal.so,
3334 libsystem-id128.so, libsystemd-login.so and
3335 libsystemd-daemon.so do not make use of IFUNC
3336 anymore. Instead, we now build libsystemd.so multiple times
3337 under these alternative names. This means that the footprint
3338 is drastically increased, but given that these are
3339 transitional compatibility libraries, this should not matter
3340 much. This change has been made necessary to support the ARM
3341 platform for these compatibility libraries, as the ARM
3342 toolchain is not really at the same level as the toolchain
3343 for other architectures like x86 and does not support
3344 IFUNC. Please make sure to use --enable-compat-libs only
3345 during a transitional period!
3347 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
3348 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
3349 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
3350 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
3351 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
3352 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
3353 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
3354 Zbigniew Jędrzejewski-Szmek
3356 — Berlin, 2014-02-24
3360 * A new component "systemd-networkd" has been added that can
3361 be used to configure local network interfaces statically or
3362 via DHCP. It is capable of bringing up bridges, VLANs, and
3363 bonding. Currently, no hook-ups for interactive network
3364 configuration are provided. Use this for your initrd,
3365 container, embedded, or server setup if you need a simple,
3366 yet powerful, network configuration solution. This
3367 configuration subsystem is quite nifty, as it allows wildcard
3368 hotplug matching in interfaces. For example, with a single
3369 configuration snippet, you can configure that all Ethernet
3370 interfaces showing up are automatically added to a bridge,
3371 or similar. It supports link-sensing and more.
3373 * A new tool "systemd-socket-proxyd" has been added which can
3374 act as a bidirectional proxy for TCP sockets. This is
3375 useful for adding socket activation support to services that
3376 do not actually support socket activation, including virtual
3377 machines and the like.
3379 * Add a new tool to save/restore rfkill state on
3382 * Save/restore state of keyboard backlights in addition to
3383 display backlights on shutdown/boot.
3385 * udev learned a new SECLABEL{} construct to label device
3386 nodes with a specific security label when they appear. For
3387 now, only SECLABEL{selinux} is supported, but the syntax is
3388 prepared for additional security frameworks.
3390 * udev gained a new scheme to configure link-level attributes
3391 from files in /etc/systemd/network/*.link. These files can
3392 match against MAC address, device path, driver name and type,
3393 and will apply attributes like the naming policy, link speed,
3394 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
3395 address assignment policy (randomized, ...).
3397 * The configuration of network interface naming rules for
3398 "permanent interface names" has changed: a new NamePolicy=
3399 setting in the [Link] section of .link files determines the
3400 priority of possible naming schemes (onboard, slot, MAC,
3401 path). The default value of this setting is determined by
3402 /usr/lib/net/links/99-default.link. Old
3403 80-net-name-slot.rules udev configuration file has been
3404 removed, so local configuration overriding this file should
3405 be adapated to override 99-default.link instead.
3407 * When the User= switch is used in a unit file, also
3408 initialize $SHELL= based on the user database entry.
3410 * systemd no longer depends on libdbus. All communication is
3411 now done with sd-bus, systemd's low-level bus library
3414 * kdbus support has been added to PID 1 itself. When kdbus is
3415 enabled, this causes PID 1 to set up the system bus and
3416 enable support for a new ".busname" unit type that
3417 encapsulates bus name activation on kdbus. It works a little
3418 bit like ".socket" units, except for bus names. A new
3419 generator has been added that converts classic dbus1 service
3420 activation files automatically into native systemd .busname
3423 * sd-bus: add a light-weight vtable implementation that allows
3424 defining objects on the bus with a simple static const
3425 vtable array of its methods, signals and properties.
3427 * systemd will not generate or install static dbus
3428 introspection data anymore to /usr/share/dbus-1/interfaces,
3429 as the precise format of these files is unclear, and
3430 nothing makes use of it.
3432 * A proxy daemon is now provided to proxy clients connecting
3433 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
3434 compatibility with classic D-Bus.
3436 * A bus driver implementation has been added that supports the
3437 classic D-Bus bus driver calls on kdbus, also for
3438 compatibility purposes.
3440 * A new API "sd-event.h" has been added that implements a
3441 minimal event loop API built around epoll. It provides a
3442 couple of features that direct epoll usage is lacking:
3443 prioritization of events, scales to large numbers of timer
3444 events, per-event timer slack (accuracy), system-wide
3445 coalescing of timer events, exit handlers, watchdog
3446 supervision support using systemd's sd_notify() API, child
3449 * A new API "sd-rntl.h" has been added that provides an API
3450 around the route netlink interface of the kernel, similar in
3451 style to "sd-bus.h".
3453 * A new API "sd-dhcp-client.h" has been added that provides a
3454 small DHCPv4 client-side implementation. This is used by
3457 * There is a new kernel command line option
3458 "systemd.restore_state=0|1". When set to "0", none of the
3459 systemd tools will restore saved runtime state to hardware
3460 devices. More specifically, the rfkill and backlight states
3463 * The FsckPassNo= compatibility option in mount/service units
3464 has been removed. The fstab generator will now add the
3465 necessary dependencies automatically, and does not require
3466 PID1's support for that anymore.
3468 * journalctl gained a new switch, --list-boots, that lists
3469 recent boots with their times and boot IDs.
3471 * The various tools like systemctl, loginctl, timedatectl,
3472 busctl, systemd-run, ... have gained a new switch "-M" to
3473 connect to a specific, local OS container (as direct
3474 connection, without requiring SSH). This works on any
3475 container that is registered with machined, such as those
3476 created by libvirt-lxc or nspawn.
3478 * systemd-run and systemd-analyze also gained support for "-H"
3479 to connect to remote hosts via SSH. This is particularly
3480 useful for systemd-run because it enables queuing of jobs
3481 onto remote systems.
3483 * machinectl gained a new command "login" to open a getty
3484 login in any local container. This works with any container
3485 that is registered with machined (such as those created by
3486 libvirt-lxc or nspawn), and which runs systemd inside.
3488 * machinectl gained a new "reboot" command that may be used to
3489 trigger a reboot on a specific container that is registered
3490 with machined. This works on any container that runs an init
3491 system of some kind.
3493 * systemctl gained a new "list-timers" command to print a nice
3494 listing of installed timer units with the times they elapse
3497 * Alternative reboot() parameters may now be specified on the
3498 "systemctl reboot" command line and are passed to the
3499 reboot() system call.
3501 * systemctl gained a new --job-mode= switch to configure the
3502 mode to queue a job with. This is a more generic version of
3503 --fail, --irreversible, and --ignore-dependencies, which are
3504 still available but not advertised anymore.
3506 * /etc/systemd/system.conf gained new settings to configure
3507 various default timeouts of units, as well as the default
3508 start limit interval and burst. These may still be overridden
3511 * PID1 will now export on the bus profile data of the security
3512 policy upload process (such as the SELinux policy upload to
3515 * journald: when forwarding logs to the console, include
3516 timestamps (following the setting in
3517 /sys/module/printk/parameters/time).
3519 * OnCalendar= in timer units now understands the special
3520 strings "yearly" and "annually". (Both are equivalent)
3522 * The accuracy of timer units is now configurable with the new
3523 AccuracySec= setting. It defaults to 1min.
3525 * A new dependency type JoinsNamespaceOf= has been added that
3526 allows running two services within the same /tmp and network
3527 namespace, if PrivateNetwork= or PrivateTmp= are used.
3529 * A new command "cat" has been added to systemctl. It outputs
3530 the original unit file of a unit, and concatenates the
3531 contents of additional "drop-in" unit file snippets, so that
3532 the full configuration is shown.
3534 * systemctl now supports globbing on the various "list-xyz"
3535 commands, like "list-units" or "list-sockets", as well as on
3536 those commands which take multiple unit names.
3538 * journalctl's --unit= switch gained support for globbing.
3540 * All systemd daemons now make use of the watchdog logic so
3541 that systemd automatically notices when they hang.
3543 * If the $container_ttys environment variable is set,
3544 getty-generator will automatically spawn a getty for each
3545 listed tty. This is useful for container managers to request
3546 login gettys to be spawned on as many ttys as needed.
3548 * %h, %s, %U specifier support is not available anymore when
3549 used in unit files for PID 1. This is because NSS calls are
3550 not safe from PID 1. They stay available for --user
3551 instances of systemd, and as special case for the root user.
3553 * loginctl gained a new "--no-legend" switch to turn off output
3556 * The "sd-login.h" API gained three new calls:
3557 sd_session_is_remote(), sd_session_get_remote_user(),
3558 sd_session_get_remote_host() to query information about
3561 * The udev hardware database now also carries vendor/product
3562 information of SDIO devices.
3564 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
3565 determine whether watchdog notifications are requested by
3568 * Socket-activated per-connection services now include a
3569 short description of the connection parameters in the
3572 * tmpfiles gained a new "--boot" option. When this is not used,
3573 only lines where the command character is not suffixed with
3574 "!" are executed. When this option is specified, those
3575 options are executed too. This partitions tmpfiles
3576 directives into those that can be safely executed at any
3577 time, and those which should be run only at boot (for
3578 example, a line that creates /run/nologin).
3580 * A new API "sd-resolve.h" has been added which provides a simple
3581 asynchronous wrapper around glibc NSS host name resolution
3582 calls, such as getaddrinfo(). In contrast to glibc's
3583 getaddrinfo_a(), it does not use signals. In contrast to most
3584 other asynchronous name resolution libraries, this one does
3585 not reimplement DNS, but reuses NSS, so that alternate
3586 host name resolution systems continue to work, such as mDNS,
3587 LDAP, etc. This API is based on libasyncns, but it has been
3588 cleaned up for inclusion in systemd.
3590 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
3591 "sd-daemon.h" are no longer found in individual libraries
3592 libsystemd-journal.so, libsystemd-login.so,
3593 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
3594 merged them into a single library, libsystemd.so, which
3595 provides all symbols. The reason for this is cyclic
3596 dependencies, as these libraries tend to use each other's
3597 symbols. So far, we have managed to workaround that by linking
3598 a copy of a good part of our code into each of these
3599 libraries again and again, which, however, makes certain
3600 things hard to do, like sharing static variables. Also, it
3601 substantially increases footprint. With this change, there
3602 is only one library for the basic APIs systemd
3603 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
3604 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
3605 library as well, however are subject to the --enable-kdbus
3606 switch (see below). Note that "sd-dhcp-client.h" is not part
3607 of this library (this is because it only consumes, never
3608 provides, services of/to other APIs). To make the transition
3609 easy from the separate libraries to the unified one, we
3610 provide the --enable-compat-libs compile-time switch which
3611 will generate stub libraries that are compatible with the
3612 old ones but redirect all calls to the new one.
3614 * All of the kdbus logic and the new APIs "sd-bus.h",
3615 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
3616 and "sd-utf8.h" are compile-time optional via the
3617 "--enable-kdbus" switch, and they are not compiled in by
3618 default. To make use of kdbus, you have to explicitly enable
3619 the switch. Note however, that neither the kernel nor the
3620 userspace API for all of this is considered stable yet. We
3621 want to maintain the freedom to still change the APIs for
3622 now. By specifying this build-time switch, you acknowledge
3623 that you are aware of the instability of the current
3626 * Also, note that while kdbus is pretty much complete,
3627 it lacks one thing: proper policy support. This means you
3628 can build a fully working system with all features; however,
3629 it will be highly insecure. Policy support will be added in
3630 one of the next releases, at the same time that we will
3631 declare the APIs stable.
3633 * When the kernel command line argument "kdbus" is specified,
3634 systemd will automatically load the kdbus.ko kernel module. At
3635 this stage of development, it is only useful for testing kdbus
3636 and should not be used in production. Note: if "--enable-kdbus"
3637 is specified, and the kdbus.ko kernel module is available, and
3638 "kdbus" is added to the kernel command line, the entire system
3639 runs with kdbus instead of dbus-daemon, with the above mentioned
3640 problem of missing the system policy enforcement. Also a future
3641 version of kdbus.ko or a newer systemd will not be compatible with
3642 each other, and will unlikely be able to boot the machine if only
3643 one of them is updated.
3645 * systemctl gained a new "import-environment" command which
3646 uploads the caller's environment (or parts thereof) into the
3647 service manager so that it is inherited by services started
3648 by the manager. This is useful to upload variables like
3649 $DISPLAY into the user service manager.
3651 * A new PrivateDevices= switch has been added to service units
3652 which allows running a service with a namespaced /dev
3653 directory that does not contain any device nodes for
3654 physical devices. More specifically, it only includes devices
3655 such as /dev/null, /dev/urandom, and /dev/zero which are API
3658 * logind has been extended to support behaviour like VT
3659 switching on seats that do not support a VT. This makes
3660 multi-session available on seats that are not the first seat
3661 (seat0), and on systems where kernel support for VTs has
3662 been disabled at compile-time.
3664 * If a process holds a delay lock for system sleep or shutdown
3665 and fails to release it in time, we will now log its
3666 identity. This makes it easier to identify processes that
3667 cause slow suspends or power-offs.
3669 * When parsing /etc/crypttab, support for a new key-slot=
3670 option as supported by Debian is added. It allows indicating
3671 which LUKS slot to use on disk, speeding up key loading.
3673 * The sd_journald_sendv() API call has been checked and
3674 officially declared to be async-signal-safe so that it may
3675 be invoked from signal handlers for logging purposes.
3677 * Boot-time status output is now enabled automatically after a
3678 short timeout if boot does not progress, in order to give
3679 the user an indication what she or he is waiting for.
3681 * The boot-time output has been improved to show how much time
3682 remains until jobs expire.
3684 * The KillMode= switch in service units gained a new possible
3685 value "mixed". If set, and the unit is shut down, then the
3686 initial SIGTERM signal is sent only to the main daemon
3687 process, while the following SIGKILL signal is sent to
3688 all remaining processes of the service.
3690 * When a scope unit is registered, a new property "Controller"
3691 may be set. If set to a valid bus name, systemd will send a
3692 RequestStop() signal to this name when it would like to shut
3693 down the scope. This may be used to hook manager logic into
3694 the shutdown logic of scope units. Also, scope units may now
3695 be put in a special "abandoned" state, in which case the
3696 manager process which created them takes no further
3697 responsibilities for it.
3699 * When reading unit files, systemd will now verify
3700 the access mode of these files, and warn about certain
3701 suspicious combinations. This has been added to make it
3702 easier to track down packaging bugs where unit files are
3703 marked executable or world-writable.
3705 * systemd-nspawn gained a new "--setenv=" switch to set
3706 container-wide environment variables. The similar option in
3707 systemd-activate was renamed from "--environment=" to
3708 "--setenv=" for consistency.
3710 * systemd-nspawn has been updated to create a new kdbus domain
3711 for each container that is invoked, thus allowing each
3712 container to have its own set of system and user buses,
3713 independent of the host.
3715 * systemd-nspawn gained a new --drop-capability= switch to run
3716 the container with less capabilities than the default. Both
3717 --drop-capability= and --capability= now take the special
3718 string "all" for dropping or keeping all capabilities.
3720 * systemd-nspawn gained new switches for executing containers
3721 with specific SELinux labels set.
3723 * systemd-nspawn gained a new --quiet switch to not generate
3724 any additional output but the container's own console
3727 * systemd-nspawn gained a new --share-system switch to run a
3728 container without PID namespacing enabled.
3730 * systemd-nspawn gained a new --register= switch to control
3731 whether the container is registered with systemd-machined or
3732 not. This is useful for containers that do not run full
3733 OS images, but only specific apps.
3735 * systemd-nspawn gained a new --keep-unit which may be used
3736 when invoked as the only program from a service unit, and
3737 results in registration of the unit service itself in
3738 systemd-machined, instead of a newly opened scope unit.
3740 * systemd-nspawn gained a new --network-interface= switch for
3741 moving arbitrary interfaces to the container. The new
3742 --network-veth switch creates a virtual Ethernet connection
3743 between host and container. The new --network-bridge=
3744 switch then allows assigning the host side of this virtual
3745 Ethernet connection to a bridge device.
3747 * systemd-nspawn gained a new --personality= switch for
3748 setting the kernel personality for the container. This is
3749 useful when running a 32-bit container on a 64-bit host. A
3750 similar option Personality= is now also available for service
3753 * logind will now also track a "Desktop" identifier for each
3754 session which encodes the desktop environment of it. This is
3755 useful for desktop environments that want to identify
3756 multiple running sessions of itself easily.
3758 * A new SELinuxContext= setting for service units has been
3759 added that allows setting a specific SELinux execution
3760 context for a service.
3762 * Most systemd client tools will now honour $SYSTEMD_LESS for
3763 settings of the "less" pager. By default, these tools will
3764 override $LESS to allow certain operations to work, such as
3765 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
3766 influence this logic.
3768 * systemd's "seccomp" hook-up has been changed to make use of
3769 the libseccomp library instead of using its own
3770 implementation. This has benefits for portability among
3773 * For usage together with SystemCallFilter=, a new
3774 SystemCallErrorNumber= setting has been introduced that
3775 allows configuration of a system error number to be returned
3776 on filtered system calls, instead of immediately killing the
3777 process. Also, SystemCallArchitectures= has been added to
3778 limit access to system calls of a particular architecture
3779 (in order to turn off support for unused secondary
3780 architectures). There is also a global
3781 SystemCallArchitectures= setting in system.conf now to turn
3782 off support for non-native system calls system-wide.
3784 * systemd requires a kernel with a working name_to_handle_at(),
3785 please see the kernel config requirements in the README file.
3787 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
3788 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
3789 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
3790 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
3791 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
3792 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
3793 Elia Pinto, Florian Weimer, George McCollister, Goffredo
3794 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
3795 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
3796 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
3797 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
3798 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
3799 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
3800 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
3801 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
3802 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
3803 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
3804 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
3805 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
3806 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
3807 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
3808 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
3809 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
3810 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
3812 — Berlin, 2014-02-20
3816 * logind has gained support for facilitating privileged input
3817 and drm device access for unprivileged clients. This work is
3818 useful to allow Wayland display servers (and similar
3819 programs, such as kmscon) to run under the user's ID and
3820 access input and drm devices which are normally
3821 protected. When this is used (and the kernel is new enough)
3822 logind will "mute" IO on the file descriptors passed to
3823 Wayland as long as it is in the background and "unmute" it
3824 if it returns into the foreground. This allows secure
3825 session switching without allowing background sessions to
3826 eavesdrop on input and display data. This also introduces
3827 session switching support if VT support is turned off in the
3828 kernel, and on seats that are not seat0.
3830 * A new kernel command line option luks.options= is understood
3831 now which allows specifying LUKS options for usage for LUKS
3832 encrypted partitions specified with luks.uuid=.
3834 * tmpfiles.d(5) snippets may now use specifier expansion in
3835 path names. More specifically %m, %b, %H, %v, are now
3836 replaced by the local machine id, boot id, hostname, and
3837 kernel version number.
3839 * A new tmpfiles.d(5) command "m" has been introduced which
3840 may be used to change the owner/group/access mode of a file
3841 or directory if it exists, but do nothing if it does not.
3843 * This release removes high-level support for the
3844 MemorySoftLimit= cgroup setting. The underlying kernel
3845 cgroup attribute memory.soft_limit= is currently badly
3846 designed and likely to be removed from the kernel API in its
3847 current form, hence we should not expose it for now.
3849 * The memory.use_hierarchy cgroup attribute is now enabled for
3850 all cgroups systemd creates in the memory cgroup
3851 hierarchy. This option is likely to be come the built-in
3852 default in the kernel anyway, and the non-hierarchical mode
3853 never made much sense in the intrinsically hierarchical
3856 * A new field _SYSTEMD_SLICE= is logged along with all journal
3857 messages containing the slice a message was generated
3858 from. This is useful to allow easy per-customer filtering of
3859 logs among other things.
3861 * systemd-journald will no longer adjust the group of journal
3862 files it creates to the "systemd-journal" group. Instead we
3863 rely on the journal directory to be owned by the
3864 "systemd-journal" group, and its setgid bit set, so that the
3865 kernel file system layer will automatically enforce that
3866 journal files inherit this group assignment. The reason for
3867 this change is that we cannot allow NSS look-ups from
3868 journald which would be necessary to resolve
3869 "systemd-journal" to a numeric GID, because this might
3870 create deadlocks if NSS involves synchronous queries to
3871 other daemons (such as nscd, or sssd) which in turn are
3872 logging clients of journald and might block on it, which
3873 would then dead lock. A tmpfiles.d(5) snippet included in
3874 systemd will make sure the setgid bit and group are
3875 properly set on the journal directory if it exists on every
3876 boot. However, we recommend adjusting it manually after
3877 upgrades too (or from RPM scriptlets), so that the change is
3878 not delayed until next reboot.
3880 * Backlight and random seed files in /var/lib/ have moved into
3881 the /var/lib/systemd/ directory, in order to centralize all
3882 systemd generated files in one directory.
3884 * Boot time performance measurements (as displayed by
3885 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
3886 performance information if that's available to determine how
3887 much time BIOS and boot loader initialization required. With
3888 a sufficiently new BIOS you hence no longer need to boot
3889 with Gummiboot to get access to such information.
3891 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
3892 Cristian Rodríguez, Dave Reisner, David Herrmann, David
3893 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
3894 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
3895 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
3896 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
3897 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3899 — Berlin, 2013-10-02
3903 * The Restart= option for services now understands a new
3904 on-watchdog setting, which will restart the service
3905 automatically if the service stops sending out watchdog keep
3906 alive messages (as configured with WatchdogSec=).
3908 * The getty generator (which is responsible for bringing up a
3909 getty on configured serial consoles) will no longer only
3910 start a getty on the primary kernel console but on all
3911 others, too. This makes the order in which console= is
3912 specified on the kernel command line less important.
3914 * libsystemd-logind gained a new sd_session_get_vt() call to
3915 retrieve the VT number of a session.
3917 * If the option "tries=0" is set for an entry of /etc/crypttab
3918 its passphrase is queried indefinitely instead of any
3919 maximum number of tries.
3921 * If a service with a configure PID file terminates its PID
3922 file will now be removed automatically if it still exists
3923 afterwards. This should put an end to stale PID files.
3925 * systemd-run will now also take relative binary path names
3926 for execution and no longer insists on absolute paths.
3928 * InaccessibleDirectories= and ReadOnlyDirectories= now take
3929 paths that are optionally prefixed with "-" to indicate that
3930 it should not be considered a failure if they do not exist.
3932 * journalctl -o (and similar commands) now understands a new
3933 output mode "short-precise", it is similar to "short" but
3934 shows timestamps with usec accuracy.
3936 * The option "discard" (as known from Debian) is now
3937 synonymous to "allow-discards" in /etc/crypttab. In fact,
3938 "discard" is preferred now (since it is easier to remember
3941 * Some licensing clean-ups were made, so that more code is now
3942 LGPL-2.1 licensed than before.
3944 * A minimal tool to save/restore the display backlight
3945 brightness across reboots has been added. It will store the
3946 backlight setting as late as possible at shutdown, and
3947 restore it as early as possible during reboot.
3949 * A logic to automatically discover and enable home and swap
3950 partitions on GPT disks has been added. With this in place
3951 /etc/fstab becomes optional for many setups as systemd can
3952 discover certain partitions located on the root disk
3953 automatically. Home partitions are recognized under their
3954 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
3955 partitions are recognized under their GPT type ID
3956 0657fd6da4ab43c484e50933c84b4f4f.
3958 * systemd will no longer pass any environment from the kernel
3959 or initrd to system services. If you want to set an
3960 environment for all services, do so via the kernel command
3961 line systemd.setenv= assignment.
3963 * The systemd-sysctl tool no longer natively reads the file
3964 /etc/sysctl.conf. If desired, the file should be symlinked
3965 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
3966 legacy support by a symlink rather than built-in code, it
3967 also makes the otherwise hidden order of application of the
3968 different files visible. (Note that this partly reverts to a
3969 pre-198 application order of sysctl knobs!)
3971 * The "systemctl set-log-level" and "systemctl dump" commands
3972 have been moved to systemd-analyze.
3974 * systemd-run learned the new --remain-after-exit switch,
3975 which causes the scope unit not to be cleaned up
3976 automatically after the process terminated.
3978 * tmpfiles learned a new --exclude-prefix= switch to exclude
3979 certain paths from operation.
3981 * journald will now automatically flush all messages to disk
3982 as soon as a message at the log level CRIT, ALERT or EMERG
3985 Contributions from: Andrew Cook, Brandon Philips, Christian
3986 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
3987 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
3988 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
3989 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
3990 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
3991 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
3992 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
3993 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
3994 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
3995 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
3996 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
3997 William Giokas, Zbigniew Jędrzejewski-Szmek
3999 — Berlin, 2013-09-13
4003 * The documentation has been updated to cover the various new
4004 concepts introduced with 205.
4006 * Unit files now understand the new %v specifier which
4007 resolves to the kernel version string as returned by "uname
4010 * systemctl now supports filtering the unit list output by
4011 load state, active state and sub state, using the new
4014 * "systemctl status" will now show the results of the
4015 condition checks (like ConditionPathExists= and similar) of
4016 the last start attempts of the unit. They are also logged to
4019 * "journalctl -b" may now be used to look for boot output of a
4020 specific boot. Try "journalctl -b -1" for the previous boot,
4021 but the syntax is substantially more powerful.
4023 * "journalctl --show-cursor" has been added which prints the
4024 cursor string the last shown log line. This may then be used
4025 with the new "journalctl --after-cursor=" switch to continue
4026 browsing logs from that point on.
4028 * "journalctl --force" may now be used to force regeneration
4031 * Creation of "dead" device nodes has been moved from udev
4032 into kmod and tmpfiles. Previously, udev would read the kmod
4033 databases to pre-generate dead device nodes based on meta
4034 information contained in kernel modules, so that these would
4035 be auto-loaded on access rather then at boot. As this
4036 does not really have much to do with the exposing actual
4037 kernel devices to userspace this has always been slightly
4038 alien in the udev codebase. Following the new scheme kmod
4039 will now generate a runtime snippet for tmpfiles from the
4040 module meta information and it now is tmpfiles' job to the
4041 create the nodes. This also allows overriding access and
4042 other parameters for the nodes using the usual tmpfiles
4043 facilities. As side effect this allows us to remove the
4044 CAP_SYS_MKNOD capability bit from udevd entirely.
4046 * logind's device ACLs may now be applied to these "dead"
4047 devices nodes too, thus finally allowing managed access to
4048 devices such as /dev/snd/sequencer whithout loading the
4049 backing module right-away.
4051 * A new RPM macro has been added that may be used to apply
4052 tmpfiles configuration during package installation.
4054 * systemd-detect-virt and ConditionVirtualization= now can
4055 detect User-Mode-Linux machines (UML).
4057 * journald will now implicitly log the effective capabilities
4058 set of processes in the message metadata.
4060 * systemd-cryptsetup has gained support for TrueCrypt volumes.
4062 * The initrd interface has been simplified (more specifically,
4063 support for passing performance data via environment
4064 variables and fsck results via files in /run has been
4065 removed). These features were non-essential, and are
4066 nowadays available in a much nicer way by having systemd in
4067 the initrd serialize its state and have the hosts systemd
4068 deserialize it again.
4070 * The udev "keymap" data files and tools to apply keyboard
4071 specific mappings of scan to key codes, and force-release
4072 scan code lists have been entirely replaced by a udev
4073 "keyboard" builtin and a hwdb data file.
4075 * systemd will now honour the kernel's "quiet" command line
4076 argument also during late shutdown, resulting in a
4077 completely silent shutdown when used.
4079 * There's now an option to control the SO_REUSEPORT socket
4080 option in .socket units.
4082 * Instance units will now automatically get a per-template
4083 subslice of system.slice unless something else is explicitly
4084 configured. For example, instances of sshd@.service will now
4085 implicitly be placed in system-sshd.slice rather than
4086 system.slice as before.
4088 * Test coverage support may now be enabled at build time.
4090 Contributions from: Dave Reisner, Frederic Crozat, Harald
4091 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
4092 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
4093 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
4094 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
4095 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
4096 Giokas, Zbigniew Jędrzejewski-Szmek
4098 — Berlin, 2013-07-23
4102 * Two new unit types have been introduced:
4104 Scope units are very similar to service units, however, are
4105 created out of pre-existing processes — instead of PID 1
4106 forking off the processes. By using scope units it is
4107 possible for system services and applications to group their
4108 own child processes (worker processes) in a powerful way
4109 which then maybe used to organize them, or kill them
4110 together, or apply resource limits on them.
4112 Slice units may be used to partition system resources in an
4113 hierarchical fashion and then assign other units to them. By
4114 default there are now three slices: system.slice (for all
4115 system services), user.slice (for all user sessions),
4116 machine.slice (for VMs and containers).
4118 Slices and scopes have been introduced primarily in
4119 context of the work to move cgroup handling to a
4120 single-writer scheme, where only PID 1
4121 creates/removes/manages cgroups.
4123 * There's a new concept of "transient" units. In contrast to
4124 normal units these units are created via an API at runtime,
4125 not from configuration from disk. More specifically this
4126 means it is now possible to run arbitrary programs as
4127 independent services, with all execution parameters passed
4128 in via bus APIs rather than read from disk. Transient units
4129 make systemd substantially more dynamic then it ever was,
4130 and useful as a general batch manager.
4132 * logind has been updated to make use of scope and slice units
4133 for managing user sessions. As a user logs in he will get
4134 his own private slice unit, to which all sessions are added
4135 as scope units. We also added support for automatically
4136 adding an instance of user@.service for the user into the
4137 slice. Effectively logind will no longer create cgroup
4138 hierarchies on its own now, it will defer entirely to PID 1
4139 for this by means of scope, service and slice units. Since
4140 user sessions this way become entities managed by PID 1
4141 the output of "systemctl" is now a lot more comprehensive.
4143 * A new mini-daemon "systemd-machined" has been added which
4144 may be used by virtualization managers to register local
4145 VMs/containers. nspawn has been updated accordingly, and
4146 libvirt will be updated shortly. machined will collect a bit
4147 of meta information about the VMs/containers, and assign
4148 them their own scope unit (see above). The collected
4149 meta-data is then made available via the "machinectl" tool,
4150 and exposed in "ps" and similar tools. machined/machinectl
4151 is compile-time optional.
4153 * As discussed earlier, the low-level cgroup configuration
4154 options ControlGroup=, ControlGroupModify=,
4155 ControlGroupPersistent=, ControlGroupAttribute= have been
4156 removed. Please use high-level attribute settings instead as
4157 well as slice units.
4159 * A new bus call SetUnitProperties() has been added to alter
4160 various runtime parameters of a unit. This is primarily
4161 useful to alter cgroup parameters dynamically in a nice way,
4162 but will be extended later on to make more properties
4163 modifiable at runtime. systemctl gained a new set-properties
4164 command that wraps this call.
4166 * A new tool "systemd-run" has been added which can be used to
4167 run arbitrary command lines as transient services or scopes,
4168 while configuring a number of settings via the command
4169 line. This tool is currently very basic, however already
4170 very useful. We plan to extend this tool to even allow
4171 queuing of execution jobs with time triggers from the
4172 command line, similar in fashion to "at".
4174 * nspawn will now inform the user explicitly that kernels with
4175 audit enabled break containers, and suggest the user to turn
4178 * Support for detecting the IMA and AppArmor security
4179 frameworks with ConditionSecurity= has been added.
4181 * journalctl gained a new "-k" switch for showing only kernel
4182 messages, mimicking dmesg output; in addition to "--user"
4183 and "--system" switches for showing only user's own logs
4186 * systemd-delta can now show information about drop-in
4187 snippets extending unit files.
4189 * libsystemd-bus has been substantially updated but is still
4190 not available as public API.
4192 * systemd will now look for the "debug" argument on the kernel
4193 command line and enable debug logging, similar to what
4194 "systemd.log_level=debug" already did before.
4196 * "systemctl set-default", "systemctl get-default" has been
4197 added to configure the default.target symlink, which
4198 controls what to boot into by default.
4200 * "systemctl set-log-level" has been added as a convenient
4201 way to raise and lower systemd logging threshold.
4203 * "systemd-analyze plot" will now show the time the various
4204 generators needed for execution, as well as information
4205 about the unit file loading.
4207 * libsystemd-journal gained a new sd_journal_open_files() call
4208 for opening specific journal files. journactl also gained a
4209 new switch to expose this new functionality. Previously we
4210 only supported opening all files from a directory, or all
4211 files from the system, as opening individual files only is
4212 racy due to journal file rotation.
4214 * systemd gained the new DefaultEnvironment= setting in
4215 /etc/systemd/system.conf to set environment variables for
4218 * If a privileged process logs a journal message with the
4219 OBJECT_PID= field set, then journald will automatically
4220 augment this with additional OBJECT_UID=, OBJECT_GID=,
4221 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
4222 system services want to log events about specific client
4223 processes. journactl/systemctl has been updated to make use
4224 of this information if all log messages regarding a specific
4227 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
4228 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
4229 Reisner, David Coppa, David King, David Strauss, Eelco
4230 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
4231 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
4232 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
4233 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
4234 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
4235 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
4236 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
4237 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
4238 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
4239 Łukasz Stelmach, 장동준
4243 * The Python bindings gained some minimal support for the APIs
4244 exposed by libsystemd-logind.
4246 * ConditionSecurity= gained support for detecting SMACK. Since
4247 this condition already supports SELinux and AppArmor we only
4248 miss IMA for this. Patches welcome!
4250 Contributions from: Karol Lewandowski, Lennart Poettering,
4251 Zbigniew Jędrzejewski-Szmek
4255 * systemd-nspawn will now create /etc/resolv.conf if
4256 necessary, before bind-mounting the host's file onto it.
4258 * systemd-nspawn will now store meta information about a
4259 container on the container's cgroup as extended attribute
4260 fields, including the root directory.
4262 * The cgroup hierarchy has been reworked in many ways. All
4263 objects any of the components systemd creates in the cgroup
4264 tree are now suffixed. More specifically, user sessions are
4265 now placed in cgroups suffixed with ".session", users in
4266 cgroups suffixed with ".user", and nspawn containers in
4267 cgroups suffixed with ".nspawn". Furthermore, all cgroup
4268 names are now escaped in a simple scheme to avoid collision
4269 of userspace object names with kernel filenames. This work
4270 is preparation for making these objects relocatable in the
4271 cgroup tree, in order to allow easy resource partitioning of
4272 these objects without causing naming conflicts.
4274 * systemctl list-dependencies gained the new switches
4275 --plain, --reverse, --after and --before.
4277 * systemd-inhibit now shows the process name of processes that
4278 have taken an inhibitor lock.
4280 * nss-myhostname will now also resolve "localhost"
4281 implicitly. This makes /etc/hosts an optional file and
4282 nicely handles that on IPv6 ::1 maps to both "localhost" and
4285 * libsystemd-logind.so gained a new call
4286 sd_get_machine_names() to enumerate running containers and
4287 VMs (currently only supported by very new libvirt and
4288 nspawn). sd_login_monitor can now be used to watch
4289 VMs/containers coming and going.
4291 * .include is not allowed recursively anymore, and only in
4292 unit files. Usually it is better to use drop-in snippets in
4293 .d/*.conf anyway, as introduced with systemd 198.
4295 * systemd-analyze gained a new "critical-chain" command that
4296 determines the slowest chain of units run during system
4297 boot-up. It is very useful for tracking down where
4298 optimizing boot time is the most beneficial.
4300 * systemd will no longer allow manipulating service paths in
4301 the name=systemd:/system cgroup tree using ControlGroup= in
4302 units. (But is still fine with it in all other dirs.)
4304 * There's a new systemd-nspawn@.service service file that may
4305 be used to easily run nspawn containers as system
4306 services. With the container's root directory in
4307 /var/lib/container/foobar it is now sufficient to run
4308 "systemctl start systemd-nspawn@foobar.service" to boot it.
4310 * systemd-cgls gained a new parameter "--machine" to list only
4311 the processes within a certain container.
4313 * ConditionSecurity= now can check for "apparmor". We still
4314 are lacking checks for SMACK and IMA for this condition
4315 check though. Patches welcome!
4317 * A new configuration file /etc/systemd/sleep.conf has been
4318 added that may be used to configure which kernel operation
4319 systemd is supposed to execute when "suspend", "hibernate"
4320 or "hybrid-sleep" is requested. This makes the new kernel
4321 "freeze" state accessible to the user.
4323 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
4324 the passed argument if applicable.
4326 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
4327 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
4328 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
4329 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
4330 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
4331 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
4332 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
4337 * The output of 'systemctl list-jobs' got some polishing. The
4338 '--type=' argument may now be passed more than once. A new
4339 command 'systemctl list-sockets' has been added which shows
4340 a list of kernel sockets systemd is listening on with the
4341 socket units they belong to, plus the units these socket
4344 * The experimental libsystemd-bus library got substantial
4345 updates to work in conjunction with the (also experimental)
4346 kdbus kernel project. It works well enough to exchange
4347 messages with some sophistication. Note that kdbus is not
4348 ready yet, and the library is mostly an elaborate test case
4349 for now, and not installable.
4351 * systemd gained a new unit 'systemd-static-nodes.service'
4352 that generates static device nodes earlier during boot, and
4353 can run in conjunction with udev.
4355 * libsystemd-login gained a new call sd_pid_get_user_unit()
4356 to retrieve the user systemd unit a process is running
4357 in. This is useful for systems where systemd is used as
4360 * systemd-nspawn now places all containers in the new /machine
4361 top-level cgroup directory in the name=systemd
4362 hierarchy. libvirt will soon do the same, so that we get a
4363 uniform separation of /system, /user and /machine for system
4364 services, user processes and containers/virtual
4365 machines. This new cgroup hierarchy is also useful to stick
4366 stable names to specific container instances, which can be
4367 recognized later this way (this name may be controlled
4368 via systemd-nspawn's new -M switch). libsystemd-login also
4369 gained a new call sd_pid_get_machine_name() to retrieve the
4370 name of the container/VM a specific process belongs to.
4372 * bootchart can now store its data in the journal.
4374 * libsystemd-journal gained a new call
4375 sd_journal_add_conjunction() for AND expressions to the
4376 matching logic. This can be used to express more complex
4377 logical expressions.
4379 * journactl can now take multiple --unit= and --user-unit=
4382 * The cryptsetup logic now understands the "luks.key=" kernel
4383 command line switch for specifying a file to read the
4384 decryption key from. Also, if a configured key file is not
4385 found the tool will now automatically fall back to prompting
4388 * Python systemd.journal module was updated to wrap recently
4389 added functions from libsystemd-journal. The interface was
4390 changed to bring the low level interface in s.j._Reader
4391 closer to the C API, and the high level interface in
4392 s.j.Reader was updated to wrap and convert all data about
4395 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
4396 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
4397 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
4398 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
4399 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
4400 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
4404 * journalctl --update-catalog now understands a new --root=
4405 option to operate on catalogs found in a different root
4408 * During shutdown after systemd has terminated all running
4409 services a final killing loop kills all remaining left-over
4410 processes. We will now print the name of these processes
4411 when we send SIGKILL to them, since this usually indicates a
4414 * If /etc/crypttab refers to password files stored on
4415 configured mount points automatic dependencies will now be
4416 generated to ensure the specific mount is established first
4417 before the key file is attempted to be read.
4419 * 'systemctl status' will now show information about the
4420 network sockets a socket unit is listening on.
4422 * 'systemctl status' will also shown information about any
4423 drop-in configuration file for units. (Drop-In configuration
4424 files in this context are files such as
4425 /etc/systemd/systemd/foobar.service.d/*.conf)
4427 * systemd-cgtop now optionally shows summed up CPU times of
4428 cgroups. Press '%' while running cgtop to switch between
4429 percentage and absolute mode. This is useful to determine
4430 which cgroups use up the most CPU time over the entire
4431 runtime of the system. systemd-cgtop has also been updated
4432 to be 'pipeable' for processing with further shell tools.
4434 * 'hostnamectl set-hostname' will now allow setting of FQDN
4437 * The formatting and parsing of time span values has been
4438 changed. The parser now understands fractional expressions
4439 such as "5.5h". The formatter will now output fractional
4440 expressions for all time spans under 1min, i.e. "5.123456s"
4441 rather than "5s 123ms 456us". For time spans under 1s
4442 millisecond values are shown, for those under 1ms
4443 microsecond values are shown. This should greatly improve
4444 all time-related output of systemd.
4446 * libsystemd-login and libsystemd-journal gained new
4447 functions for querying the poll() events mask and poll()
4448 timeout value for integration into arbitrary event
4451 * localectl gained the ability to list available X11 keymaps
4452 (models, layouts, variants, options).
4454 * 'systemd-analyze dot' gained the ability to filter for
4455 specific units via shell-style globs, to create smaller,
4456 more useful graphs. I.e. it is now possible to create simple
4457 graphs of all the dependencies between only target units, or
4458 of all units that Avahi has dependencies with.
4460 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
4461 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
4462 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
4463 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
4464 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
4465 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
4466 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
4470 * The boot-time readahead implementation for rotating media
4471 will now read the read-ahead data in multiple passes which
4472 consist of all read requests made in equidistant time
4473 intervals. This means instead of strictly reading read-ahead
4474 data in its physical order on disk we now try to find a
4475 middle ground between physical and access time order.
4477 * /etc/os-release files gained a new BUILD_ID= field for usage
4478 on operating systems that provide continuous builds of OS
4481 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
4482 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
4483 William Douglas, Zbigniew Jędrzejewski-Szmek
4487 * systemd-python gained an API exposing libsystemd-daemon.
4489 * The SMACK setup logic gained support for uploading CIPSO
4492 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
4493 ReadOnlyDirectories= and InaccessibleDirectories= has
4494 changed. The private /tmp and /var/tmp directories are now
4495 shared by all processes of a service (which means
4496 ExecStartPre= may now leave data in /tmp that ExecStart= of
4497 the same service can still access). When a service is
4498 stopped its temporary directories are immediately deleted
4499 (normal clean-up with tmpfiles is still done in addition to
4502 * By default, systemd will now set a couple of sysctl
4503 variables in the kernel: the safe sysrq options are turned
4504 on, IP route verification is turned on, and source routing
4505 disabled. The recently added hardlink and softlink
4506 protection of the kernel is turned on. These settings should
4507 be reasonably safe, and good defaults for all new systems.
4509 * The predictable network naming logic may now be turned off
4510 with a new kernel command line switch: net.ifnames=0.
4512 * A new libsystemd-bus module has been added that implements a
4513 pretty complete D-Bus client library. For details see:
4515 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
4517 * journald will now explicitly flush the journal files to disk
4518 at the latest 5min after each write. The file will then also
4519 be marked offline until the next write. This should increase
4520 reliability in case of a crash. The synchronization delay
4521 can be configured via SyncIntervalSec= in journald.conf.
4523 * There's a new remote-fs-setup.target unit that can be used
4524 to pull in specific services when at least one remote file
4525 system is to be mounted.
4527 * There are new targets timers.target and paths.target as
4528 canonical targets to pull user timer and path units in
4529 from. This complements sockets.target with a similar
4530 purpose for socket units.
4532 * libudev gained a new call udev_device_set_attribute_value()
4533 to set sysfs attributes of a device.
4535 * The udev daemon now sets the default number of worker
4536 processes executed in parallel based on the number of available
4537 CPUs instead of the amount of available RAM. This is supposed
4538 to provide a more reliable default and limit a too aggressive
4539 paralellism for setups with 1000s of devices connected.
4541 Contributions from: Auke Kok, Colin Walters, Cristian
4542 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
4543 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
4544 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
4545 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
4546 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
4547 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
4548 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
4549 Zbigniew Jędrzejewski-Szmek
4553 * Configuration of unit files may now be extended via drop-in
4554 files without having to edit/override the unit files
4555 themselves. More specifically, if the administrator wants to
4556 change one value for a service file foobar.service he can
4557 now do so by dropping in a configuration snippet into
4558 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
4559 will load all these snippets and apply them on top of the
4560 main unit configuration file, possibly extending or
4561 overriding its settings. Using these drop-in snippets is
4562 generally nicer than the two earlier options for changing
4563 unit files locally: copying the files from
4564 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
4565 them there; or creating a new file in /etc/systemd/system/
4566 that incorporates the original one via ".include". Drop-in
4567 snippets into these .d/ directories can be placed in any
4568 directory systemd looks for units in, and the usual
4569 overriding semantics between /usr/lib, /etc and /run apply
4572 * Most unit file settings which take lists of items can now be
4573 reset by assigning the empty string to them. For example,
4574 normally, settings such as Environment=FOO=BAR append a new
4575 environment variable assignment to the environment block,
4576 each time they are used. By assigning Environment= the empty
4577 string the environment block can be reset to empty. This is
4578 particularly useful with the .d/*.conf drop-in snippets
4579 mentioned above, since this adds the ability to reset list
4580 settings from vendor unit files via these drop-ins.
4582 * systemctl gained a new "list-dependencies" command for
4583 listing the dependencies of a unit recursively.
4585 * Inhibitors are now honored and listed by "systemctl
4586 suspend", "systemctl poweroff" (and similar) too, not only
4587 GNOME. These commands will also list active sessions by
4590 * Resource limits (as exposed by the various control group
4591 controllers) can now be controlled dynamically at runtime
4592 for all units. More specifically, you can now use a command
4593 like "systemctl set-cgroup-attr foobar.service cpu.shares
4594 2000" to alter the CPU shares a specific service gets. These
4595 settings are stored persistently on disk, and thus allow the
4596 administrator to easily adjust the resource usage of
4597 services with a few simple commands. This dynamic resource
4598 management logic is also available to other programs via the
4599 bus. Almost any kernel cgroup attribute and controller is
4602 * systemd-vconsole-setup will now copy all font settings to
4603 all allocated VTs, where it previously applied them only to
4606 * libsystemd-login gained the new sd_session_get_tty() API
4609 * This release drops support for a few legacy or
4610 distribution-specific LSB facility names when parsing init
4611 scripts: $x-display-manager, $mail-transfer-agent,
4612 $mail-transport-agent, $mail-transfer-agent, $smtp,
4613 $null. Also, the mail-transfer-agent.target unit backing
4614 this has been removed. Distributions which want to retain
4615 compatibility with this should carry the burden for
4616 supporting this themselves and patch support for these back
4617 in, if they really need to. Also, the facilities $syslog and
4618 $local_fs are now ignored, since systemd does not support
4619 early-boot LSB init scripts anymore, and these facilities
4620 are implied anyway for normal services. syslog.target has
4623 * There are new bus calls on PID1's Manager object for
4624 cancelling jobs, and removing snapshot units. Previously,
4625 both calls were only available on the Job and Snapshot
4628 * systemd-journal-gatewayd gained SSL support.
4630 * The various "environment" files, such as /etc/locale.conf
4631 now support continuation lines with a backslash ("\") as
4632 last character in the line, similarly in style (but different)
4633 to how this is supported in shells.
4635 * For normal user processes the _SYSTEMD_USER_UNIT= field is
4636 now implicitly appended to every log entry logged. systemctl
4637 has been updated to filter by this field when operating on a
4638 user systemd instance.
4640 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
4641 CAP_AUDIT_CONTROL capabilities to the capabilities set for
4642 the container. This makes it easier to boot unmodified
4643 Fedora systems in a container, which however still requires
4644 audit=0 to be passed on the kernel command line. Auditing in
4645 kernel and userspace is unfortunately still too broken in
4646 context of containers, hence we recommend compiling it out
4647 of the kernel or using audit=0. Hopefully this will be fixed
4648 one day for good in the kernel.
4650 * nspawn gained the new --bind= and --bind-ro= parameters to
4651 bind mount specific directories from the host into the
4654 * nspawn will now mount its own devpts file system instance
4655 into the container, in order not to leak pty devices from
4656 the host into the container.
4658 * systemd will now read the firmware boot time performance
4659 information from the EFI variables, if the used boot loader
4660 supports this, and takes it into account for boot performance
4661 analysis via "systemd-analyze". This is currently supported
4662 only in conjunction with Gummiboot, but could be supported
4663 by other boot loaders too. For details see:
4665 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
4667 * A new generator has been added that automatically mounts the
4668 EFI System Partition (ESP) to /boot, if that directory
4669 exists, is empty, and no other file system has been
4670 configured to be mounted there.
4672 * logind will now send out PrepareForSleep(false) out
4673 unconditionally, after coming back from suspend. This may be
4674 used by applications as asynchronous notification for
4675 system resume events.
4677 * "systemctl unlock-sessions" has been added, that allows
4678 unlocking the screens of all user sessions at once, similar
4679 to how "systemctl lock-sessions" already locked all users
4680 sessions. This is backed by a new D-Bus call UnlockSessions().
4682 * "loginctl seat-status" will now show the master device of a
4683 seat. (i.e. the device of a seat that needs to be around for
4684 the seat to be considered available, usually the graphics
4687 * tmpfiles gained a new "X" line type, that allows
4688 configuration of files and directories (with wildcards) that
4689 shall be excluded from automatic cleanup ("aging").
4691 * udev default rules set the device node permissions now only
4692 at "add" events, and do not change them any longer with a
4693 later "change" event.
4695 * The log messages for lid events and power/sleep keypresses
4696 now carry a message ID.
4698 * We now have a substantially larger unit test suite, but this
4699 continues to be work in progress.
4701 * udevadm hwdb gained a new --root= parameter to change the
4702 root directory to operate relative to.
4704 * logind will now issue a background sync() request to the kernel
4705 early at shutdown, so that dirty buffers are flushed to disk early
4706 instead of at the last moment, in order to optimize shutdown
4709 * A new bootctl tool has been added that is an interface for
4710 certain boot loader operations. This is currently a preview
4711 and is likely to be extended into a small mechanism daemon
4712 like timedated, localed, hostnamed, and can be used by
4713 graphical UIs to enumerate available boot options, and
4714 request boot into firmware operations.
4716 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
4717 the rest of the package. It also has been updated to work
4718 correctly in initrds.
4720 * Policykit previously has been runtime optional, and is now
4721 also compile time optional via a configure switch.
4723 * systemd-analyze has been reimplemented in C. Also "systemctl
4724 dot" has moved into systemd-analyze.
4726 * "systemctl status" with no further parameters will now print
4727 the status of all active or failed units.
4729 * Operations such as "systemctl start" can now be executed
4730 with a new mode "--irreversible" which may be used to queue
4731 operations that cannot accidentally be reversed by a later
4732 job queuing. This is by default used to make shutdown
4733 requests more robust.
4735 * The Python API of systemd now gained a new module for
4736 reading journal files.
4738 * A new tool kernel-install has been added that can install
4739 kernel images according to the Boot Loader Specification:
4741 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
4743 * Boot time console output has been improved to provide
4744 animated boot time output for hanging jobs.
4746 * A new tool systemd-activate has been added which can be used
4747 to test socket activation with, directly from the command
4748 line. This should make it much easier to test and debug
4749 socket activation in daemons.
4751 * journalctl gained a new "--reverse" (or -r) option to show
4752 journal output in reverse order (i.e. newest line first).
4754 * journalctl gained a new "--pager-end" (or -e) option to jump
4755 to immediately jump to the end of the journal in the
4756 pager. This is only supported in conjunction with "less".
4758 * journalctl gained a new "--user-unit=" option, that works
4759 similarly to "--unit=" but filters for user units rather than
4762 * A number of unit files to ease adoption of systemd in
4763 initrds has been added. This moves some minimal logic from
4764 the various initrd implementations into systemd proper.
4766 * The journal files are now owned by a new group
4767 "systemd-journal", which exists specifically to allow access
4768 to the journal, and nothing else. Previously, we used the
4769 "adm" group for that, which however possibly covers more
4770 than just journal/log file access. This new group is now
4771 already used by systemd-journal-gatewayd to ensure this
4772 daemon gets access to the journal files and as little else
4773 as possible. Note that "make install" will also set FS ACLs
4774 up for /var/log/journal to give "adm" and "wheel" read
4775 access to it, in addition to "systemd-journal" which owns
4776 the journal files. We recommend that packaging scripts also
4777 add read access to "adm" + "wheel" to /var/log/journal, and
4778 all existing/future journal files. To normal users and
4779 administrators little changes, however packagers need to
4780 ensure to create the "systemd-journal" system group at
4781 package installation time.
4783 * The systemd-journal-gatewayd now runs as unprivileged user
4784 systemd-journal-gateway:systemd-journal-gateway. Packaging
4785 scripts need to create these system user/group at
4788 * timedated now exposes a new boolean property CanNTP that
4789 indicates whether a local NTP service is available or not.
4791 * systemd-detect-virt will now also detect xen PVs
4793 * The pstore file system is now mounted by default, if it is
4796 * In addition to the SELinux and IMA policies we will now also
4797 load SMACK policies at early boot.
4799 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
4800 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
4801 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
4802 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
4803 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
4804 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
4805 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
4806 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
4807 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
4808 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
4809 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
4810 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
4811 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
4812 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
4816 * Timer units now support calendar time events in addition to
4817 monotonic time events. That means you can now trigger a unit
4818 based on a calendar time specification such as "Thu,Fri
4819 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
4820 or fifth day of any month of the year 2013, given that it is
4821 a thursday or friday. This brings timer event support
4822 considerably closer to cron's capabilities. For details on
4823 the supported calendar time specification language see
4826 * udev now supports a number of different naming policies for
4827 network interfaces for predictable names, and a combination
4828 of these policies is now the default. Please see this wiki
4829 document for details:
4831 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
4833 * Auke Kok's bootchart implementation has been added to the
4834 systemd tree. It is an optional component that can graph the
4835 boot in quite some detail. It is one of the best bootchart
4836 implementations around and minimal in its code and
4839 * nss-myhostname has been integrated into the systemd source
4840 tree. nss-myhostname guarantees that the local hostname
4841 always stays resolvable via NSS. It has been a weak
4842 requirement of systemd-hostnamed since a long time, and
4843 since its code is actually trivial we decided to just
4844 include it in systemd's source tree. It can be turned off
4845 with a configure switch.
4847 * The read-ahead logic is now capable of properly detecting
4848 whether a btrfs file system is on SSD or rotating media, in
4849 order to optimize the read-ahead scheme. Previously, it was
4850 only capable of detecting this on traditional file systems
4853 * In udev, additional device properties are now read from the
4854 IAB in addition to the OUI database. Also, Bluetooth company
4855 identities are attached to the devices as well.
4857 * In service files %U may be used as specifier that is
4858 replaced by the configured user name of the service.
4860 * nspawn may now be invoked without a controlling TTY. This
4861 makes it suitable for invocation as its own service. This
4862 may be used to set up a simple containerized server system
4863 using only core OS tools.
4865 * systemd and nspawn can now accept socket file descriptors
4866 when they are started for socket activation. This enables
4867 implementation of socket activated nspawn
4868 containers. i.e. think about autospawning an entire OS image
4869 when the first SSH or HTTP connection is received. We expect
4870 that similar functionality will also be added to libvirt-lxc
4873 * journalctl will now suppress ANSI color codes when
4874 presenting log data.
4876 * systemctl will no longer show control group information for
4877 a unit if a the control group is empty anyway.
4879 * logind can now automatically suspend/hibernate/shutdown the
4882 * /etc/machine-info and hostnamed now also expose the chassis
4883 type of the system. This can be used to determine whether
4884 the local system is a laptop, desktop, handset or
4885 tablet. This information may either be configured by the
4886 user/vendor or is automatically determined from ACPI and DMI
4887 information if possible.
4889 * A number of PolicyKit actions are now bound together with
4890 "imply" rules. This should simplify creating UIs because
4891 many actions will now authenticate similar ones as well.
4893 * Unit files learnt a new condition ConditionACPower= which
4894 may be used to conditionalize a unit depending on whether an
4895 AC power source is connected or not, of whether the system
4896 is running on battery power.
4898 * systemctl gained a new "is-failed" verb that may be used in
4899 shell scripts and suchlike to check whether a specific unit
4900 is in the "failed" state.
4902 * The EnvironmentFile= setting in unit files now supports file
4903 globbing, and can hence be used to easily read a number of
4904 environment files at once.
4906 * systemd will no longer detect and recognize specific
4907 distributions. All distribution-specific #ifdeffery has been
4908 removed, systemd is now fully generic and
4909 distribution-agnostic. Effectively, not too much is lost as
4910 a lot of the code is still accessible via explicit configure
4911 switches. However, support for some distribution specific
4912 legacy configuration file formats has been dropped. We
4913 recommend distributions to simply adopt the configuration
4914 files everybody else uses now and convert the old
4915 configuration from packaging scripts. Most distributions
4916 already did that. If that's not possible or desirable,
4917 distributions are welcome to forward port the specific
4918 pieces of code locally from the git history.
4920 * When logging a message about a unit systemd will now always
4921 log the unit name in the message meta data.
4923 * localectl will now also discover system locale data that is
4924 not stored in locale archives, but directly unpacked.
4926 * logind will no longer unconditionally use framebuffer
4927 devices as seat masters, i.e. as devices that are required
4928 to be existing before a seat is considered preset. Instead,
4929 it will now look for all devices that are tagged as
4930 "seat-master" in udev. By default, framebuffer devices will
4931 be marked as such, but depending on local systems, other
4932 devices might be marked as well. This may be used to
4933 integrate graphics cards using closed source drivers (such
4934 as NVidia ones) more nicely into logind. Note however, that
4935 we recommend using the open source NVidia drivers instead,
4936 and no udev rules for the closed-source drivers will be
4937 shipped from us upstream.
4939 Contributions from: Adam Williamson, Alessandro Crismani, Auke
4940 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
4941 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
4942 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
4943 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
4944 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
4945 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
4946 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
4947 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
4948 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
4949 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
4950 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
4955 * udev gained support for loading additional device properties
4956 from an indexed database that is keyed by vendor/product IDs
4957 and similar device identifiers. For the beginning this
4958 "hwdb" is populated with data from the well-known PCI and
4959 USB database, but also includes PNP, ACPI and OID data. In
4960 the longer run this indexed database shall grow into
4961 becoming the one central database for non-essential
4962 userspace device metadata. Previously, data from the PCI/USB
4963 database was only attached to select devices, since the
4964 lookup was a relatively expensive operation due to O(n) time
4965 complexity (with n being the number of entries in the
4966 database). Since this is now O(1), we decided to add in this
4967 data for all devices where this is available, by
4968 default. Note that the indexed database needs to be rebuilt
4969 when new data files are installed. To achieve this you need
4970 to update your packaging scripts to invoke "udevadm hwdb
4971 --update" after installation of hwdb data files. For
4972 RPM-based distributions we introduced the new
4973 %udev_hwdb_update macro for this purpose.
4975 * The Journal gained support for the "Message Catalog", an
4976 indexed database to link up additional information with
4977 journal entries. For further details please check:
4979 http://www.freedesktop.org/wiki/Software/systemd/catalog
4981 The indexed message catalog database also needs to be
4982 rebuilt after installation of message catalog files. Use
4983 "journalctl --update-catalog" for this. For RPM-based
4984 distributions we introduced the %journal_catalog_update
4985 macro for this purpose.
4987 * The Python Journal bindings gained support for the standard
4988 Python logging framework.
4990 * The Journal API gained new functions for checking whether
4991 the underlying file system of a journal file is capable of
4992 properly reporting file change notifications, or whether
4993 applications that want to reflect journal changes "live"
4994 need to recheck journal files continuously in appropriate
4997 * It is now possible to set the "age" field for tmpfiles
4998 entries to 0, indicating that files matching this entry
4999 shall always be removed when the directories are cleaned up.
5001 * coredumpctl gained a new "gdb" verb which invokes gdb
5002 right-away on the selected coredump.
5004 * There's now support for "hybrid sleep" on kernels that
5005 support this, in addition to "suspend" and "hibernate". Use
5006 "systemctl hybrid-sleep" to make use of this.
5008 * logind's HandleSuspendKey= setting (and related settings)
5009 now gained support for a new "lock" setting to simply
5010 request the screen lock on all local sessions, instead of
5011 actually executing a suspend or hibernation.
5013 * systemd will now mount the EFI variables file system by
5016 * Socket units now gained support for configuration of the
5017 SMACK security label.
5019 * timedatectl will now output the time of the last and next
5020 daylight saving change.
5022 * We dropped support for various legacy and distro-specific
5023 concepts, such as insserv, early-boot SysV services
5024 (i.e. those for non-standard runlevels such as 'b' or 'S')
5025 or ArchLinux /etc/rc.conf support. We recommend the
5026 distributions who still need support this to either continue
5027 to maintain the necessary patches downstream, or find a
5028 different solution. (Talk to us if you have questions!)
5030 * Various systemd components will now bypass PolicyKit checks
5031 for root and otherwise handle properly if PolicyKit is not
5032 found to be around. This should fix most issues for
5033 PolicyKit-less systems. Quite frankly this should have been
5034 this way since day one. It is absolutely our intention to
5035 make systemd work fine on PolicyKit-less systems, and we
5036 consider it a bug if something does not work as it should if
5037 PolicyKit is not around.
5039 * For embedded systems it is now possible to build udev and
5040 systemd without blkid and/or kmod support.
5042 * "systemctl switch-root" is now capable of switching root
5043 more than once. I.e. in addition to transitions from the
5044 initrd to the host OS it is now possible to transition to
5045 further OS images from the host. This is useful to implement
5046 offline updating tools.
5048 * Various other additions have been made to the RPM macros
5049 shipped with systemd. Use %udev_rules_update() after
5050 installing new udev rules files. %_udevhwdbdir,
5051 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
5052 %_sysctldir are now available which resolve to the right
5053 directories for packages to place various data files in.
5055 * journalctl gained the new --full switch (in addition to
5056 --all, to disable ellipsation for long messages.
5058 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
5059 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
5060 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
5061 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
5062 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
5063 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
5064 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
5065 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
5066 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
5070 * journalctl gained new --since= and --until= switches to
5071 filter by time. It also now supports nice filtering for
5072 units via --unit=/-u.
5074 * Type=oneshot services may use ExecReload= and do the
5077 * The journal daemon now supports time-based rotation and
5078 vacuuming, in addition to the usual disk-space based
5081 * The journal will now index the available field values for
5082 each field name. This enables clients to show pretty drop
5083 downs of available match values when filtering. The bash
5084 completion of journalctl has been updated
5085 accordingly. journalctl gained a new switch -F to list all
5086 values a certain field takes in the journal database.
5088 * More service events are now written as structured messages
5089 to the journal, and made recognizable via message IDs.
5091 * The timedated, localed and hostnamed mini-services which
5092 previously only provided support for changing time, locale
5093 and hostname settings from graphical DEs such as GNOME now
5094 also have a minimal (but very useful) text-based client
5095 utility each. This is probably the nicest way to changing
5096 these settings from the command line now, especially since
5097 it lists available options and is fully integrated with bash
5100 * There's now a new tool "systemd-coredumpctl" to list and
5101 extract coredumps from the journal.
5103 * We now install a README each in /var/log/ and
5104 /etc/rc.d/init.d explaining where the system logs and init
5105 scripts went. This hopefully should help folks who go to
5106 that dirs and look into the otherwise now empty void and
5107 scratch their heads.
5109 * When user-services are invoked (by systemd --user) the
5110 $MANAGERPID env var is set to the PID of systemd.
5112 * SIGRTMIN+24 when sent to a --user instance will now result
5113 in immediate termination of systemd.
5115 * gatewayd received numerous feature additions such as a
5116 "follow" mode, for live syncing and filtering.
5118 * browse.html now allows filtering and showing detailed
5119 information on specific entries. Keyboard navigation and
5120 mouse screen support has been added.
5122 * gatewayd/journalctl now supports HTML5/JSON
5123 Server-Sent-Events as output.
5125 * The SysV init script compatibility logic will now
5126 heuristically determine whether a script supports the
5127 "reload" verb, and only then make this available as
5130 * "systemctl status --follow" has been removed, use "journalctl
5133 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
5134 have been removed since they are hardly useful to be
5137 * And I'd like to take the opportunity to specifically mention
5138 Zbigniew for his great contributions. Zbigniew, you rock!
5140 Contributions from: Andrew Eikum, Christian Hesse, Colin
5141 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
5142 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
5143 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
5144 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
5145 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
5146 Jędrzejewski-Szmek, Сковорода Никита Андреевич
5150 * If /etc/vconsole.conf is non-existent or empty we will no
5151 longer load any console font or key map at boot by
5152 default. Instead the kernel defaults will be left
5153 intact. This is definitely the right thing to do, as no
5154 configuration should mean no configuration, and hard-coding
5155 font names that are different on all archs is probably a bad
5156 idea. Also, the kernel default key map and font should be
5157 good enough for most cases anyway, and mostly identical to
5158 the userspace fonts/key maps we previously overloaded them
5159 with. If distributions want to continue to default to a
5160 non-kernel font or key map they should ship a default
5161 /etc/vconsole.conf with the appropriate contents.
5163 Contributions from: Colin Walters, Daniel J Walsh, Dave
5164 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
5165 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5169 * journalctl gained a new --cursor= switch to show entries
5170 starting from the specified location in the journal.
5172 * We now enforce a size limit on journal entry fields exported
5173 with "-o json" in journalctl. Fields larger than 4K will be
5174 assigned null. This can be turned off with --all.
5176 * An (optional) journal gateway daemon is now available as
5177 "systemd-journal-gatewayd.service". This service provides
5178 access to the journal via HTTP and JSON. This functionality
5179 will be used to implement live log synchronization in both
5180 pull and push modes, but has various other users too, such
5181 as easy log access for debugging of embedded devices. Right
5182 now it is already useful to retrieve the journal via HTTP:
5184 # systemctl start systemd-journal-gatewayd.service
5185 # wget http://localhost:19531/entries
5187 This will download the journal contents in a
5188 /var/log/messages compatible format. The same as JSON:
5190 # curl -H"Accept: application/json" http://localhost:19531/entries
5192 This service is also accessible via a web browser where a
5193 single static HTML5 app is served that uses the JSON logic
5194 to enable the user to do some basic browsing of the
5195 journal. This will be extended later on. Here's an example
5196 screenshot of this app in its current state:
5198 http://0pointer.de/public/journal-gatewayd
5200 Contributions from: Kay Sievers, Lennart Poettering, Robert
5201 Milasan, Tom Gundersen
5205 * The bash completion logic is now available for journalctl
5208 * We do not mount the "cpuset" controller anymore together with
5209 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
5210 started if no parameters are assigned to it. "cpuset" hence
5211 broke code that assumed it it could create "cpu" groups and
5214 * journalctl -f will now subscribe to terminal size changes,
5215 and line break accordingly.
5217 Contributions from: Dave Reisner, Kay Sievers, Lennart
5218 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
5222 * nspawn will now create a symlink /etc/localtime in the
5223 container environment, copying the host's timezone
5224 setting. Previously this has been done via a bind mount, but
5225 since symlinks cannot be bind mounted this has now been
5226 changed to create/update the appropriate symlink.
5228 * journalctl -n's line number argument is now optional, and
5229 will default to 10 if omitted.
5231 * journald will now log the maximum size the journal files may
5232 take up on disk. This is particularly useful if the default
5233 built-in logic of determining this parameter from the file
5234 system size is used. Use "systemctl status
5235 systemd-journald.service" to see this information.
5237 * The multi-seat X wrapper tool has been stripped down. As X
5238 is now capable of enumerating graphics devices via udev in a
5239 seat-aware way the wrapper is not strictly necessary
5240 anymore. A stripped down temporary stop-gap is still shipped
5241 until the upstream display managers have been updated to
5242 fully support the new X logic. Expect this wrapper to be
5243 removed entirely in one of the next releases.
5245 * HandleSleepKey= in logind.conf has been split up into
5246 HandleSuspendKey= and HandleHibernateKey=. The old setting
5247 is not available anymore. X11 and the kernel are
5248 distinguishing between these keys and we should too. This
5249 also means the inhibition lock for these keys has been split
5252 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
5253 Poettering, Lukas Nykryn, Václav Pavlín
5257 * Whenever a unit changes state we will now log this to the
5258 journal and show along the unit's own log output in
5261 * ConditionPathIsMountPoint= can now properly detect bind
5262 mount points too. (Previously, a bind mount of one file
5263 system to another place in the same file system could not be
5264 detected as mount, since they shared struct stat's st_dev
5267 * We will now mount the cgroup controllers cpu, cpuacct,
5268 cpuset and the controllers net_cls, net_prio together by
5271 * nspawn containers will now have a virtualized boot
5272 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
5273 over with a randomized ID at container initialization). This
5274 has the effect of making "journalctl -b" do the right thing
5277 * The JSON output journal serialization has been updated not
5278 to generate "endless" list objects anymore, but rather one
5279 JSON object per line. This is more in line how most JSON
5280 parsers expect JSON objects. The new output mode
5281 "json-pretty" has been added to provide similar output, but
5282 neatly aligned for readability by humans.
5284 * We dropped all explicit sync() invocations in the shutdown
5285 code. The kernel does this implicitly anyway in the kernel
5286 reboot() syscall. halt(8)'s -n option is now a compatibility
5289 * We now support virtualized reboot() in containers, as
5290 supported by newer kernels. We will fall back to exit() if
5291 CAP_SYS_REBOOT is not available to the container. Also,
5292 nspawn makes use of this now and will actually reboot the
5293 container if the containerized OS asks for that.
5295 * journalctl will only show local log output by default
5296 now. Use --merge (-m) to show remote log output, too.
5298 * libsystemd-journal gained the new sd_journal_get_usage()
5299 call to determine the current disk usage of all journal
5300 files. This is exposed in the new "journalctl --disk-usage"
5303 * journald gained a new configuration setting SplitMode= in
5304 journald.conf which may be used to control how user journals
5305 are split off. See journald.conf(5) for details.
5307 * A new condition type ConditionFileNotEmpty= has been added.
5309 * tmpfiles' "w" lines now support file globbing, to write
5310 multiple files at once.
5312 * We added Python bindings for the journal submission
5313 APIs. More Python APIs for a number of selected APIs will
5314 likely follow. Note that we intend to add native bindings
5315 only for the Python language, as we consider it common
5316 enough to deserve bindings shipped within systemd. There are
5317 various projects outside of systemd that provide bindings
5318 for languages such as PHP or Lua.
5320 * Many conditions will now resolve specifiers such as %i. In
5321 addition, PathChanged= and related directives of .path units
5322 now support specifiers as well.
5324 * There's now a new RPM macro definition for the system preset
5327 * journald will now warn if it ca not forward a message to the
5328 syslog daemon because its socket is full.
5330 * timedated will no longer write or process /etc/timezone,
5331 except on Debian. As we do not support late mounted /usr
5332 anymore /etc/localtime always being a symlink is now safe,
5333 and hence the information in /etc/timezone is not necessary
5336 * logind will now always reserve one VT for a text getty (VT6
5337 by default). Previously if more than 6 X sessions where
5338 started they took up all the VTs with auto-spawned gettys,
5339 so that no text gettys were available anymore.
5341 * udev will now automatically inform the btrfs kernel logic
5342 about btrfs RAID components showing up. This should make
5343 simple hotplug based btrfs RAID assembly work.
5345 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
5346 (but not for its children which will stay at the kernel
5347 default). This should allow setups with a lot more listening
5350 * systemd will now always pass the configured timezone to the
5351 kernel at boot. timedated will do the same when the timezone
5354 * logind's inhibition logic has been updated. By default,
5355 logind will now handle the lid switch, the power and sleep
5356 keys all the time, even in graphical sessions. If DEs want
5357 to handle these events on their own they should take the new
5358 handle-power-key, handle-sleep-key and handle-lid-switch
5359 inhibitors during their runtime. A simple way to achieve
5360 that is to invoke the DE wrapped in an invocation of:
5362 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
5364 * Access to unit operations is now checked via SELinux taking
5365 the unit file label and client process label into account.
5367 * systemd will now notify the administrator in the journal
5368 when he over-mounts a non-empty directory.
5370 * There are new specifiers that are resolved in unit files,
5371 for the host name (%H), the machine ID (%m) and the boot ID
5374 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
5375 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
5376 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
5377 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
5378 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
5379 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
5380 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
5384 * Support for reading structured kernel messages from
5385 /dev/kmsg has now been added and is enabled by default.
5387 * Support for reading kernel messages from /proc/kmsg has now
5388 been removed. If you want kernel messages in the journal
5389 make sure to run a recent kernel (>= 3.5) that supports
5390 reading structured messages from /dev/kmsg (see
5391 above). /proc/kmsg is now exclusive property of classic
5392 syslog daemons again.
5394 * The libudev API gained the new
5395 udev_device_new_from_device_id() call.
5397 * The logic for file system namespace (ReadOnlyDirectory=,
5398 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
5399 require pivot_root() anymore. This means fewer temporary
5400 directories are created below /tmp for this feature.
5402 * nspawn containers will now see and receive all submounts
5403 made on the host OS below the root file system of the
5406 * Forward Secure Sealing is now supported for Journal files,
5407 which provide cryptographical sealing of journal files so
5408 that attackers cannot alter log history anymore without this
5409 being detectable. Lennart will soon post a blog story about
5410 this explaining it in more detail.
5412 * There are two new service settings RestartPreventExitStatus=
5413 and SuccessExitStatus= which allow configuration of exit
5414 status (exit code or signal) which will be excepted from the
5415 restart logic, resp. consider successful.
5417 * journalctl gained the new --verify switch that can be used
5418 to check the integrity of the structure of journal files and
5419 (if Forward Secure Sealing is enabled) the contents of
5422 * nspawn containers will now be run with /dev/stdin, /dev/fd/
5423 and similar symlinks pre-created. This makes running shells
5424 as container init process a lot more fun.
5426 * The fstab support can now handle PARTUUID= and PARTLABEL=
5429 * A new ConditionHost= condition has been added to match
5430 against the hostname (with globs) and machine ID. This is
5431 useful for clusters where a single OS image is used to
5432 provision a large number of hosts which shall run slightly
5433 different sets of services.
5435 * Services which hit the restart limit will now be placed in a
5438 Contributions from: Bertram Poettering, Dave Reisner, Huang
5439 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
5440 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
5444 * When running in --user mode systemd will now become a
5445 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
5446 tree a lot more organized.
5448 * A new PartOf= unit dependency type has been introduced that
5449 may be used to group services in a natural way.
5451 * "systemctl enable" may now be used to enable instances of
5454 * journalctl now prints error log levels in red, and
5455 warning/notice log levels in bright white. It also supports
5456 filtering by log level now.
5458 * cgtop gained a new -n switch (similar to top), to configure
5459 the maximum number of iterations to run for. It also gained
5460 -b, to run in batch mode (accepting no input).
5462 * The suffix ".service" may now be omitted on most systemctl
5463 command lines involving service unit names.
5465 * There's a new bus call in logind to lock all sessions, as
5466 well as a loginctl verb for it "lock-sessions".
5468 * libsystemd-logind.so gained a new call sd_journal_perror()
5469 that works similar to libc perror() but logs to the journal
5470 and encodes structured information about the error number.
5472 * /etc/crypttab entries now understand the new keyfile-size=
5475 * shutdown(8) now can send a (configurable) wall message when
5476 a shutdown is cancelled.
5478 * The mount propagation mode for the root file system will now
5479 default to "shared", which is useful to make containers work
5480 nicely out-of-the-box so that they receive new mounts from
5481 the host. This can be undone locally by running "mount
5482 --make-rprivate /" if needed.
5484 * The prefdm.service file has been removed. Distributions
5485 should maintain this unit downstream if they intend to keep
5486 it around. However, we recommend writing normal unit files
5487 for display managers instead.
5489 * Since systemd is a crucial part of the OS we will now
5490 default to a number of compiler switches that improve
5491 security (hardening) such as read-only relocations, stack
5492 protection, and suchlike.
5494 * The TimeoutSec= setting for services is now split into
5495 TimeoutStartSec= and TimeoutStopSec= to allow configuration
5496 of individual time outs for the start and the stop phase of
5499 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
5500 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
5501 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
5502 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
5503 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
5504 Gundersen, Zbigniew Jędrzejewski-Szmek
5508 * The journal and id128 C APIs are now fully documented as man
5511 * Extra safety checks have been added when transitioning from
5512 the initial RAM disk to the main system to avoid accidental
5515 * /etc/crypttab entries now understand the new keyfile-offset=
5518 * systemctl -t can now be used to filter by unit load state.
5520 * The journal C API gained the new sd_journal_wait() call to
5521 make writing synchronous journal clients easier.
5523 * journalctl gained the new -D switch to show journals from a
5526 * journalctl now displays a special marker between log
5527 messages of two different boots.
5529 * The journal is now explicitly flushed to /var via a service
5530 systemd-journal-flush.service, rather than implicitly simply
5531 by seeing /var/log/journal to be writable.
5533 * journalctl (and the journal C APIs) can now match for much
5534 more complex expressions, with alternatives and
5537 * When transitioning from the initial RAM disk to the main
5538 system we will now kill all processes in a killing spree to
5539 ensure no processes stay around by accident.
5541 * Three new specifiers may be used in unit files: %u, %h, %s
5542 resolve to the user name, user home directory resp. user
5543 shell. This is useful for running systemd user instances.
5545 * We now automatically rotate journal files if their data
5546 object hash table gets a fill level > 75%. We also size the
5547 hash table based on the configured maximum file size. This
5548 together should lower hash collisions drastically and thus
5549 speed things up a bit.
5551 * journalctl gained the new "--header" switch to introspect
5552 header data of journal files.
5554 * A new setting SystemCallFilters= has been added to services
5555 which may be used to apply blacklists or whitelists to
5556 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
5558 * nspawn gained a new --link-journal= switch (and quicker: -j)
5559 to link the container journal with the host. This makes it
5560 very easy to centralize log viewing on the host for all
5561 guests while still keeping the journal files separated.
5563 * Many bugfixes and optimizations
5565 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
5566 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
5567 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
5572 * Several tools now understand kernel command line arguments,
5573 which are only read when run in an initial RAM disk. They
5574 usually follow closely their normal counterparts, but are
5577 * There's a new tool to analyze the readahead files that are
5578 automatically generated at boot. Use:
5580 /usr/lib/systemd/systemd-readahead analyze /.readahead
5582 * We now provide an early debug shell on tty9 if this enabled. Use:
5584 systemctl enable debug-shell.service
5586 * All plymouth related units have been moved into the Plymouth
5587 package. Please make sure to upgrade your Plymouth version
5590 * systemd-tmpfiles now supports getting passed the basename of
5591 a configuration file only, in which case it will look for it
5592 in all appropriate directories automatically.
5594 * udevadm info now takes a /dev or /sys path as argument, and
5595 does the right thing. Example:
5597 udevadm info /dev/sda
5598 udevadm info /sys/class/block/sda
5600 * systemctl now prints a warning if a unit is stopped but a
5601 unit that might trigger it continues to run. Example: a
5602 service is stopped but the socket that activates it is left
5605 * "systemctl status" will now mention if the log output was
5606 shortened due to rotation since a service has been started.
5608 * The journal API now exposes functions to determine the
5609 "cutoff" times due to rotation.
5611 * journald now understands SIGUSR1 and SIGUSR2 for triggering
5612 immediately flushing of runtime logs to /var if possible,
5613 resp. for triggering immediate rotation of the journal
5616 * It is now considered an error if a service is attempted to
5617 be stopped that is not loaded.
5619 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
5621 * systemd-analyze now supports Python 3
5623 * tmpfiles now supports cleaning up directories via aging
5624 where the first level dirs are always kept around but
5625 directories beneath it automatically aged. This is enabled
5626 by prefixing the age field with '~'.
5628 * Seat objects now expose CanGraphical, CanTTY properties
5629 which is required to deal with very fast bootups where the
5630 display manager might be running before the graphics drivers
5631 completed initialization.
5633 * Seat objects now expose a State property.
5635 * We now include RPM macros for service enabling/disabling
5636 based on the preset logic. We recommend RPM based
5637 distributions to make use of these macros if possible. This
5638 makes it simpler to reuse RPM spec files across
5641 * We now make sure that the collected systemd unit name is
5642 always valid when services log to the journal via
5645 * There's a new man page kernel-command-line(7) detailing all
5646 command line options we understand.
5648 * The fstab generator may now be disabled at boot by passing
5649 fstab=0 on the kernel command line.
5651 * A new kernel command line option modules-load= is now understood
5652 to load a specific kernel module statically, early at boot.
5654 * Unit names specified on the systemctl command line are now
5655 automatically escaped as needed. Also, if file system or
5656 device paths are specified they are automatically turned
5657 into the appropriate mount or device unit names. Example:
5659 systemctl status /home
5660 systemctl status /dev/sda
5662 * The SysVConsole= configuration option has been removed from
5663 system.conf parsing.
5665 * The SysV search path is no longer exported on the D-Bus
5668 * The Names= option is been removed from unit file parsing.
5670 * There's a new man page bootup(7) detailing the boot process.
5672 * Every unit and every generator we ship with systemd now
5673 comes with full documentation. The self-explanatory boot is
5676 * A couple of services gained "systemd-" prefixes in their
5677 name if they wrap systemd code, rather than only external
5678 code. Among them fsck@.service which is now
5679 systemd-fsck@.service.
5681 * The HaveWatchdog property has been removed from the D-Bus
5684 * systemd.confirm_spawn= on the kernel command line should now
5687 * There's a new man page crypttab(5) which details all options
5688 we actually understand.
5690 * systemd-nspawn gained a new --capability= switch to pass
5691 additional capabilities to the container.
5693 * timedated will now read known NTP implementation unit names
5694 from /usr/lib/systemd/ntp-units.d/*.list,
5695 systemd-timedated-ntp.target has been removed.
5697 * journalctl gained a new switch "-b" that lists log data of
5698 the current boot only.
5700 * The notify socket is in the abstract namespace again, in
5701 order to support daemons which chroot() at start-up.
5703 * There is a new Storage= configuration option for journald
5704 which allows configuration of where log data should go. This
5705 also provides a way to disable journal logging entirely, so
5706 that data collected is only forwarded to the console, the
5707 kernel log buffer or another syslog implementation.
5709 * Many bugfixes and optimizations
5711 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
5712 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
5713 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
5714 Shawn Landden, Tom Gundersen
5718 * "systemctl help <unit>" now shows the man page if one is
5721 * Several new man pages have been added.
5723 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
5724 MaxLevelConsole= can now be specified in
5725 journald.conf. These options allow reducing the amount of
5726 data stored on disk or forwarded by the log level.
5728 * TimerSlackNSec= can now be specified in system.conf for
5729 PID1. This allows system-wide power savings.
5731 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
5732 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
5737 * logind is now capable of (optionally) handling power and
5738 sleep keys as well as the lid switch.
5740 * journalctl now understands the syntax "journalctl
5741 /usr/bin/avahi-daemon" to get all log output of a specific
5744 * CapabilityBoundingSet= in system.conf now also influences
5745 the capability bound set of usermode helpers of the kernel.
5747 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
5748 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
5749 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
5750 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
5754 * Note that we skipped 139 releases here in order to set the
5755 new version to something that is greater than both udev's
5756 and systemd's most recent version number.
5758 * udev: all udev sources are merged into the systemd source tree now.
5759 All future udev development will happen in the systemd tree. It
5760 is still fully supported to use the udev daemon and tools without
5761 systemd running, like in initramfs or other init systems. Building
5762 udev though, will require the *build* of the systemd tree, but
5763 udev can be properly *run* without systemd.
5765 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
5766 should be used to create dead device nodes as workarounds for broken
5769 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
5770 no longer supported. udev_monitor_new_from_netlink() needs to be
5771 used to subscribe to events.
5773 * udev: when udevd is started by systemd, processes which are left
5774 behind by forking them off of udev rules, are unconditionally cleaned
5775 up and killed now after the event handling has finished. Services or
5776 daemons must be started as systemd services. Services can be
5777 pulled-in by udev to get started, but they can no longer be directly
5778 forked by udev rules.
5780 * udev: the daemon binary is called systemd-udevd now and installed
5781 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
5782 to adapt to that, create symlink, or rename the binary after building
5785 * libudev no longer provides these symbols:
5786 udev_monitor_from_socket()
5787 udev_queue_get_failed_list_entry()
5788 udev_get_{dev,sys,run}_path()
5789 The versions number was bumped and symbol versioning introduced.
5791 * systemd-loginctl and systemd-journalctl have been renamed
5792 to loginctl and journalctl to match systemctl.
5794 * The config files: /etc/systemd/systemd-logind.conf and
5795 /etc/systemd/systemd-journald.conf have been renamed to
5796 logind.conf and journald.conf. Package updates should rename
5797 the files to the new names on upgrade.
5799 * For almost all files the license is now LGPL2.1+, changed
5800 from the previous GPL2.0+. Exceptions are some minor stuff
5801 of udev (which will be changed to LGPL2.1 eventually, too),
5802 and the MIT licensed sd-daemon.[ch] library that is suitable
5803 to be used as drop-in files.
5805 * systemd and logind now handle system sleep states, in
5806 particular suspending and hibernating.
5808 * logind now implements a sleep/shutdown/idle inhibiting logic
5809 suitable for a variety of uses. Soonishly Lennart will blog
5810 about this in more detail.
5812 * var-run.mount and var-lock.mount are no longer provided
5813 (which prevously bind mounted these directories to their new
5814 places). Distributions which have not converted these
5815 directories to symlinks should consider stealing these files
5816 from git history and add them downstream.
5818 * We introduced the Documentation= field for units and added
5819 this to all our shipped units. This is useful to make it
5820 easier to explore the boot and the purpose of the various
5823 * All smaller setup units (such as
5824 systemd-vconsole-setup.service) now detect properly if they
5825 are run in a container and are skipped when
5826 appropriate. This guarantees an entirely noise-free boot in
5827 Linux container environments such as systemd-nspawn.
5829 * A framework for implementing offline system updates is now
5830 integrated, for details see:
5831 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
5833 * A new service type Type=idle is available now which helps us
5834 avoiding ugly interleaving of getty output and boot status
5837 * There's now a system-wide CapabilityBoundingSet= option to
5838 globally reduce the set of capabilities for the
5839 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
5840 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
5841 even CAP_NET_ADMIN system-wide for secure systems.
5843 * There are now system-wide DefaultLimitXXX= options to
5844 globally change the defaults of the various resource limits
5845 for all units started by PID 1.
5847 * Harald Hoyer's systemd test suite has been integrated into
5848 systemd which allows easy testing of systemd builds in qemu
5849 and nspawn. (This is really awesome! Ask us for details!)
5851 * The fstab parser is now implemented as generator, not inside
5854 * systemctl will now warn you if .mount units generated from
5855 /etc/fstab are out of date due to changes in fstab that
5856 have not been read by systemd yet.
5858 * systemd is now suitable for usage in initrds. Dracut has
5859 already been updated to make use of this. With this in place
5860 initrds get a slight bit faster but primarily are much
5861 easier to introspect and debug since "systemctl status" in
5862 the host system can be used to introspect initrd services,
5863 and the journal from the initrd is kept around too.
5865 * systemd-delta has been added, a tool to explore differences
5866 between user/admin configuration and vendor defaults.
5868 * PrivateTmp= now affects both /tmp and /var/tmp.
5870 * Boot time status messages are now much prettier and feature
5871 proper english language. Booting up systemd has never been
5874 * Read-ahead pack files now include the inode number of all
5875 files to pre-cache. When the inode changes the pre-caching
5876 is not attempted. This should be nicer to deal with updated
5877 packages which might result in changes of read-ahead
5880 * We now temporaritly lower the kernel's read_ahead_kb variable
5881 when collecting read-ahead data to ensure the kernel's
5882 built-in read-ahead does not add noise to our measurements
5883 of necessary blocks to pre-cache.
5885 * There's now RequiresMountsFor= to add automatic dependencies
5886 for all mounts necessary for a specific file system path.
5888 * MountAuto= and SwapAuto= have been removed from
5889 system.conf. Mounting file systems at boot has to take place
5892 * nspawn now learned a new switch --uuid= to set the machine
5893 ID on the command line.
5895 * nspawn now learned the -b switch to automatically search
5898 * vt102 is now the default TERM for serial TTYs, upgraded from
5901 * systemd-logind now works on VT-less systems.
5903 * The build tree has been reorganized. The individual
5904 components now have directories of their own.
5906 * A new condition type ConditionPathIsReadWrite= is now available.
5908 * nspawn learned the new -C switch to create cgroups for the
5909 container in other hierarchies.
5911 * We now have support for hardware watchdogs, configurable in
5914 * The scheduled shutdown logic now has a public API.
5916 * We now mount /tmp as tmpfs by default, but this can be
5917 masked and /etc/fstab can override it.
5919 * Since udisks does not make use of /media anymore we are not
5920 mounting a tmpfs on it anymore.
5922 * journalctl gained a new --local switch to only interleave
5923 locally generated journal files.
5925 * We can now load the IMA policy at boot automatically.
5927 * The GTK tools have been split off into a systemd-ui.
5929 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
5930 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
5931 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
5932 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
5933 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
5934 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
5935 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
5936 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
5937 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
5942 * This is mostly a bugfix release
5944 * Support optional initialization of the machine ID from the
5945 KVM or container configured UUID.
5947 * Support immediate reboots with "systemctl reboot -ff"
5949 * Show /etc/os-release data in systemd-analyze output
5951 * Many bugfixes for the journal, including endianness fixes and
5952 ensuring that disk space enforcement works
5954 * sd-login.h is C++ comptaible again
5956 * Extend the /etc/os-release format on request of the Debian
5959 * We now refuse non-UTF8 strings used in various configuration
5960 and unit files. This is done to ensure we do not pass invalid
5961 data over D-Bus or expose it elsewhere.
5963 * Register Mimo USB Screens as suitable for automatic seat
5966 * Read SELinux client context from journal clients in a race
5969 * Reorder configuration file lookup order. /etc now always
5970 overrides /run in order to allow the administrator to always
5971 and unconditionally override vendor-supplied or
5972 automatically generated data.
5974 * The various user visible bits of the journal now have man
5975 pages. We still lack man pages for the journal API calls
5978 * We now ship all man pages in HTML format again in the
5981 Contributions from: Dave Reisner, Dirk Eibach, Frederic
5982 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
5983 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
5988 * This is mostly a bugfix release
5990 * systems lacking /etc/os-release are no longer supported.
5992 * Various functionality updates to libsystemd-login.so
5994 * Track class of PAM logins to distinguish greeters from
5997 Contributions from: Kay Sievers, Lennart Poettering, Michael
6002 * This is an important bugfix release for v41.
6004 * Building man pages is now optional which should be useful
6005 for those building systemd from git but unwilling to install
6008 * Watchdog support for supervising services is now usable. In
6009 a future release support for hardware watchdogs
6010 (i.e. /dev/watchdog) will be added building on this.
6012 * Service start rate limiting is now configurable and can be
6013 turned off per service. When a start rate limit is hit a
6014 reboot can automatically be triggered.
6016 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
6018 Contributions from: Benjamin Franzke, Bill Nottingham,
6019 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
6020 Schmidt, Michał Górny, Piotr Drąg
6024 * The systemd binary is installed /usr/lib/systemd/systemd now;
6025 An existing /sbin/init symlink needs to be adapted with the
6028 * The code that loads kernel modules has been ported to invoke
6029 libkmod directly, instead of modprobe. This means we do not
6030 support systems with module-init-tools anymore.
6032 * Watchdog support is now already useful, but still not
6035 * A new kernel command line option systemd.setenv= is
6036 understood to set system wide environment variables
6037 dynamically at boot.
6039 * We now limit the set of capabilities of systemd-journald.
6041 * We now set SIGPIPE to ignore by default, since it only is
6042 useful in shell pipelines, and has little use in general
6043 code. This can be disabled with IgnoreSIPIPE=no in unit
6046 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
6047 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
6052 * This is mostly a bugfix release
6054 * We now expose the reason why a service failed in the
6055 "Result" D-Bus property.
6057 * Rudimentary service watchdog support (will be completed over
6058 the next few releases.)
6060 * When systemd forks off in order execute some service we will
6061 now immediately changes its argv[0] to reflect which process
6062 it will execute. This is useful to minimize the time window
6063 with a generic argv[0], which makes bootcharts more useful
6065 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
6066 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
6067 Mike Kazantsev, Ray Strode
6071 * This is mostly a test release, but incorporates many
6074 * New systemd-cgtop tool to show control groups by their
6077 * Linking against libacl for ACLs is optional again. If
6078 disabled, support tracking device access for active logins
6079 goes becomes unavailable, and so does access to the user
6080 journals by the respective users.
6082 * If a group "adm" exists, journal files are automatically
6083 owned by them, thus allow members of this group full access
6084 to the system journal as well as all user journals.
6086 * The journal now stores the SELinux context of the logging
6087 client for all entries.
6089 * Add C++ inclusion guards to all public headers
6091 * New output mode "cat" in the journal to print only text
6092 messages, without any meta data like date or time.
6094 * Include tiny X server wrapper as a temporary stop-gap to
6095 teach XOrg udev display enumeration. This is used by display
6096 managers such as gdm, and will go away as soon as XOrg
6097 learned native udev hotplugging for display devices.
6099 * Add new systemd-cat tool for executing arbitrary programs
6100 with STDERR/STDOUT connected to the journal. Can also act as
6101 BSD logger replacement, and does so by default.
6103 * Optionally store all locally generated coredumps in the
6104 journal along with meta data.
6106 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
6107 writing short strings to files (for usage for /sys), and for
6108 creating symlinks, character and block device nodes.
6110 * New unit file option ControlGroupPersistent= to make cgroups
6111 persistent, following the mechanisms outlined in
6112 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
6114 * Support multiple local RTCs in a sane way
6116 * No longer monopolize IO when replaying readahead data on
6117 rotating disks, since we might starve non-file-system IO to
6118 death, since fanotify() will not see accesses done by blkid,
6121 * Do not show kernel threads in systemd-cgls anymore, unless
6122 requested with new -k switch.
6124 Contributions from: Dan Horák, Kay Sievers, Lennart
6125 Poettering, Michal Schmidt
6129 * This is mostly a test release, but incorporates many
6132 * The git repository moved to:
6133 git://anongit.freedesktop.org/systemd/systemd
6134 ssh://git.freedesktop.org/git/systemd/systemd
6136 * First release with the journal
6137 http://0pointer.de/blog/projects/the-journal.html
6139 * The journal replaces both systemd-kmsg-syslogd and
6140 systemd-stdout-bridge.
6142 * New sd_pid_get_unit() API call in libsystemd-logind
6144 * Many systemadm clean-ups
6146 * Introduce remote-fs-pre.target which is ordered before all
6147 remote mounts and may be used to start services before all
6150 * Added Mageia support
6152 * Add bash completion for systemd-loginctl
6154 * Actively monitor PID file creation for daemons which exit in
6155 the parent process before having finished writing the PID
6156 file in the daemon process. Daemons which do this need to be
6157 fixed (i.e. PID file creation must have finished before the
6158 parent exits), but we now react a bit more gracefully to them.
6160 * Add colourful boot output, mimicking the well-known output
6161 of existing distributions.
6163 * New option PassCredentials= for socket units, for
6164 compatibility with a recent kernel ABI breakage.
6166 * /etc/rc.local is now hooked in via a generator binary, and
6167 thus will no longer act as synchronization point during
6170 * systemctl list-unit-files now supports --root=.
6172 * systemd-tmpfiles now understands two new commands: z, Z for
6173 relabelling files according to the SELinux database. This is
6174 useful to apply SELinux labels to specific files in /sys,
6177 * Output of SysV services is now forwarded to both the console
6178 and the journal by default, not only just the console.
6180 * New man pages for all APIs from libsystemd-login.
6182 * The build tree got reorganized and a the build system is a
6183 lot more modular allowing embedded setups to specifically
6184 select the components of systemd they are interested in.
6186 * Support for Linux systems lacking the kernel VT subsystem is
6189 * configure's --with-rootdir= got renamed to
6190 --with-rootprefix= to follow the naming used by udev and
6193 * Unless specified otherwise we will now install to /usr instead
6194 of /usr/local by default.
6196 * Processes with '@' in argv[0][0] are now excluded from the
6197 final shut-down killing spree, following the logic explained
6199 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
6201 * All processes remaining in a service cgroup when we enter
6202 the START or START_PRE states are now killed with
6203 SIGKILL. That means it is no longer possible to spawn
6204 background processes from ExecStart= lines (which was never
6205 supported anyway, and bad style).
6207 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
6208 reloading of units together.
6210 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
6211 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
6212 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
6213 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
6214 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
projects / platform / upstream / systemd.git / blob