1 systemd System and Service Manager
5 * Creation of the legacy /run/lock/lockdev/ directory was
6 dropped from tmpfiles.d/legacy.conf. Hardly any software uses
7 that any more, and better locking mechanisms like flock() have
8 been available for many years. If you still need this, you need to
9 create your own tmpfiles.d config file with:
10 d /run/lock/lockdev 0775 root lock -
12 Contributions from: ...
18 * A number of properties previously only settable in unit
19 files are now also available as properties to set when
20 creating transient units programmatically via the bus, as it
21 is exposed with systemd-run's --property=
22 setting. Specifically, these are: SyslogIdentifier=,
23 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
24 EnvironmentFile=, ReadWriteDirectories=,
25 ReadOnlyDirectories=, InaccessibleDirectories=,
26 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
28 * When creating transient services via the bus API it is now
29 possible to pass in a set of file descriptors to use as
30 STDIN/STDOUT/STDERR for the invoked process.
32 * Slice units may now be created transiently via the bus APIs,
33 similar to the way service and scope units may already be
36 * Wherever systemd expects a calendar timestamp specification
37 (like in journalctl's --since= and --until= switches) UTC
38 timestamps are now supported. Timestamps suffixed with "UTC"
39 are now considered to be in Universal Time Coordinated
40 instead of the local timezone. Also, timestamps may now
41 optionally be specified with sub-second accuracy. Both of
42 these additions also apply to recurring calendar event
43 specification, such as OnCalendar= in timer units.
45 * journalctl gained a new "--sync" switch that asks the
46 journal daemon to write all so far unwritten log messages to
47 disk and sync the files, before returning.
49 * systemd-tmpfiles learned two new line types "q" and "Q" that
50 operate like "v", but also set up a basic btrfs quota
51 hierarchy when used on a btrfs file system with quota
54 * tmpfiles' "v", "q" and "Q" will now create a plain directory
55 instead of a subvolume (even on a btrfs file system) if the
56 root directory is a plain directory, and not a
57 subvolume. This should simplify things with certain chroot()
58 environments which are not aware of the concept of btrfs
61 * systemd-detect-virt gained a new --chroot switch to detect
62 whether execution takes place in a chroot() environment.
64 * CPUAffinity= now takes CPU index ranges in addition to
67 * The various memory-related resource limit settings (such as
68 LimitAS=) now understand the usual K, M, G, ... suffixes to
69 the base of 1024 (IEC). Similar, the time-related resource
70 limit settings understand the usual min, h, day, ...
73 * There's a new system.conf setting DefaultTasksMax= to
74 control the default TasksMax= setting for services and
75 scopes running on the system. (TasksMax= is the primary
76 setting that exposes the "pids" cgroup controller on systemd
77 and was introduced in the previous systemd release.) The
78 setting now defaults to 512, which means services that are
79 not explicitly configured otherwise will only be able to
80 create 512 processes or threads at maximum, from this
81 version on. Note that this means that thread- or
82 process-heavy services might need to be reconfigured to set
83 TasksMax= to a higher value. It is sufficient to set
84 TasksMax= in these specific unit files to a higher value, or
85 even "infinity". Similar, there's now a logind.conf setting
86 UserTasksMax= that defaults to 4096 and limits the total
87 number of processes or tasks each user may own
88 concurrently. nspawn containers also have the TasksMax=
89 value set by default now, to 8192. Note that all of this
90 only has an effect if the "pids" cgroup controller is
91 enabled in the kernel. The general benefit of these changes
92 should be a more robust and safer system, that provides a
93 certain amount of per-service fork() bomb protection.
95 * systemd-nspawn gained the new --network-veth-extra= switch
96 to define additional and arbitrarily-named virtual Ethernet
97 links between the host and the container.
99 * A new service execution setting PassEnvironment= has been
100 added that allows importing select environment variables
101 from PID1's environment block into the environment block of
104 * Timer units gained support for a new RemainAfterElapse=
105 setting which takes a boolean argument. It defaults on on,
106 exposing behaviour unchanged to previous releases. If set to
107 off, timer units are unloaded after they elapsed if they
108 cannot elapse again. This is particularly useful for
109 transient timer units, which shall not stay around longer
110 than until they first elapse.
112 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
113 default now (the kernel default is 16). This is beneficial
114 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
115 allows substantially larger numbers of queued
116 datagrams. This should increase the capability of systemd to
117 parallelize boot-up, as logging and sd_notify() are unlikely
118 to stall execution anymore. If you need to change the value
119 from the new defaults, use the usual sysctl.d/ snippets.
121 * The compression framing format used by the journal or
122 coredump processing has changed to be in line with what the
123 official LZ4 tools generate. LZ4 compression support in
124 systemd was considered unsupported previously, as the format
125 was not compatible with the normal tools. With this release
126 this has changed now, and it is hence safe for downstream
127 distributions to turn it on. While not compressing as well
128 as the XZ, LZ4 is substantially faster, which makes
129 it a good default choice for the compression logic in the
130 journal and in coredump handling.
132 * Any reference to /etc/mtab has been dropped from
133 systemd. The file has been obsolete since a while, but
134 systemd refused to work on systems where it was incorrectly
135 set up (it should be a symlink or non-existent). Please make
136 sure to update to util-linux 2.27.1 or newer in conjunction
137 with this systemd release, which also drops any reference to
138 /etc/mtab. If you maintain a distribution make sure that no
139 software you package still references it, as this is a
140 likely source of bugs. There's also a glibc bug pending,
141 asking for removal of any reference to this obsolete file:
143 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
145 * Support for the ".snapshot" unit type has been removed. This
146 feature turned out to be little useful and little used, and
147 has now been removed from the core and from systemctl.
149 * The dependency types RequiresOverridable= and
150 RequisiteOverridable= have been removed from systemd. They
151 have been used only very sparingly to our knowledge and
152 other options that provide a similar effect (such as
153 systemctl --mode=ignore-dependencies) are much more useful
154 and commonly used. Moreover, they were only half-way
155 implemented as the option to control behaviour regarding
156 these dependencies was never added to systemctl. By removing
157 these dependency types the execution engine becomes a bit
158 simpler. Unit files that use these dependencies should be
159 changed to use the non-Overridable dependency types
160 instead. In fact, when parsing unit files with these
161 options, that's what systemd will automatically convert them
162 too, but it will also warn, asking users to fix the unit
163 files accordingly. Removal of these dependency types should
164 only affect a negligible number of unit files in the wild.
166 * Behaviour of networkd's IPForward= option changed
167 (again). It will no longer maintain a per-interface setting,
168 but propagate one way from interfaces where this is enabled
169 to the global kernel setting. The global setting will be
170 enabled when requested by a network that is set up, but
171 never be disabled again. This change was made to make sure
172 IPv4 and IPv6 behaviour regarding packet forwarding is
173 similar (as the Linux IPv6 stack does not support
174 per-interface control of this setting) and to minimize
177 * In unit files the behaviour of %u, %U, %h, %s has
178 changed. These specifiers will now unconditionally resolve
179 to the various user database fields of the user that the
180 systemd instance is running as, instead of the user
181 configured in the specific unit via User=. Note that this
182 effectively doesn't change much, as resolving of these
183 specifiers was already turned off in the --system instance
184 of systemd, as we cannot do NSS lookups from PID 1. In the
185 --user instance of systemd these specifiers where correctly
186 resolved, but hardly made any sense, since the user instance
187 lacks privileges to do user switches anyway, and User= is
188 hence useless. Morever, even in the --user instance of
189 systemd behaviour was awkward as it would only take settings
190 from User= assignment placed before the specifier into
191 account. In order to unify and simplify the logic around
192 this the specifiers will now always resolve to the
193 credentials of the user invoking the manager (which in case
194 of PID 1 is the root user).
196 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
197 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
198 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
199 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
200 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
201 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
202 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
203 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
204 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
205 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
206 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
207 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
208 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
209 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
212 -- Berlin, 2015-11-18
216 * systemd now depends on util-linux v2.27. More specifically,
217 the newly added mount monitor feature in libmount now
218 replaces systemd's former own implementation.
220 * libmount mandates /etc/mtab not to be regular file, and
221 systemd now enforces this condition at early boot.
222 /etc/mtab has been deprecated and warned about for a very
223 long time, so systems running systemd should already have
224 stopped having this file around as anything else than a
225 symlink to /proc/self/mounts.
227 * Support for the "pids" cgroup controller has been added. It
228 allows accounting the number of tasks in a cgroup and
229 enforcing limits on it. This adds two new setting
230 TasksAccounting= and TasksMax= to each unit, as well as a
231 global option DefaultTasksAccounting=.
233 * Support for the "net_cls" cgroup controller has been added.
234 It allows assigning a net class ID to each task in the
235 cgroup, which can then be used in firewall rules and traffic
236 shaping configurations. Note that the kernel netfilter net
237 class code does not currently work reliably for ingress
238 packets on unestablished sockets.
240 This adds a new config directive called NetClass= to CGroup
241 enabled units. Allowed values are positive numbers for fixed
242 assignments and "auto" for picking a free value
245 * 'systemctl is-system-running' now returns 'offline' if the
246 system is not booted with systemd. This command can now be
247 used as a substitute for 'systemd-notify --booted'.
249 * Watchdog timeouts have been increased to 3 minutes for all
250 in-tree service files. Apparently, disk IO issues are more
251 frequent than we hoped, and user reported >1 minute waiting
254 * 'machine-id-commit' functionality has been merged into
255 'machine-id-setup --commit'. The separate binary has been
258 * The WorkingDirectory= directive in unit files may now be set
259 to the special value '~'. In this case, the working
260 directory is set to the home directory of the user
263 * "machinectl shell" will now open the shell in the home
264 directory of the selected user by default.
266 * The CrashChVT= configuration file setting is renamed to
267 CrashChangeVT=, following our usual logic of not
268 abbreviating unnecessarily. The old directive is still
269 supported for compat reasons. Also, this directive now takes
270 an integer value between 1 and 63, or a boolean value. The
271 formerly supported '-1' value for disabling stays around for
274 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
275 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
276 RootDirectory= properties can now be set for transient
279 * The systemd-analyze tool gained a new "set-log-target" verb
280 to change the logging target the system manager logs to
281 dynamically during runtime. This is similar to how
282 "systemd-analyze set-log-level" already changes the log
285 * In nspawn /sys is now mounted as tmpfs, with only a selected
286 set of subdirectories mounted in from the real sysfs. This
287 enhances security slightly, and is useful for ensuring user
288 namespaces work correctly.
290 * Support for USB FunctionFS activation has been added. This
291 allows implementation of USB gadget services that are
292 activated as soon as they are requested, so that they don't
293 have to run continously, similar to classic socket
296 * The "systemctl exit" command now optionally takes an
297 additional parameter that sets the exit code to return from
298 the systemd manager when exiting. This is only relevant when
299 running the systemd user instance, or when running the
300 system instance in a container.
302 * sd-bus gained the new API calls sd_bus_path_encode_many()
303 and sd_bus_path_decode_many() that allow easy encoding and
304 decoding of multiple identifier strings inside a D-Bus
305 object path. Another new call sd_bus_default_flush_close()
306 has been added to flush and close per-thread default
309 * systemd-cgtop gained support for a -M/--machine= switch to
310 show the control groups within a certain container only.
312 * "systemctl kill" gained support for an optional --fail
313 switch. If specified the requested operation will fail of no
314 processes have been killed, because the unit had no
315 processes attached, or similar.
317 * A new systemd.crash_reboot=1 kernel command line option has
318 been added that triggers a reboot after crashing. This can
319 also be set through CrashReboot= in systemd.conf.
321 * The RuntimeDirectory= setting now understands unit
322 specifiers like %i or %f.
324 * A new (still internal) libary API sd-ipv4acd has been added,
325 that implements address conflict detection for IPv4. It's
326 based on code from sd-ipv4ll, and will be useful for
327 detecting DHCP address conflicts.
329 * File descriptors passed during socket activation may now be
330 named. A new API sd_listen_fds_with_names() is added to
331 access the names. The default names may be overridden,
332 either in the .socket file using the FileDescriptorName=
333 parameter, or by passing FDNAME= when storing the file
334 descriptors using sd_notify().
336 * systemd-networkd gained support for:
338 - Setting the IPv6 Router Advertisment settings via
339 IPv6AcceptRouterAdvertisements= in .network files.
341 - Configuring the HelloTimeSec=, MaxAgeSec= and
342 ForwardDelaySec= bridge parameters in .netdev files.
344 - Configuring PreferredSource= for static routes in
347 * The "ask-password" framework used to query for LUKS harddisk
348 passwords or SSL passwords during boot gained support for
349 caching passwords in the kernel keyring, if it is
350 available. This makes sure that the user only has to type in
351 a passphrase once if there are multiple objects to unlock
352 with the same one. Previously, such password caching was
353 available only when Plymouth was used; this moves the
354 caching logic into the systemd codebase itself. The
355 "systemd-ask-password" utility gained a new --keyname=
356 switch to control which kernel keyring key to use for
357 caching a password in. This functionality is also useful for
358 enabling display managers such as gdm to automatically
359 unlock the user's GNOME keyring if its passphrase, the
360 user's password and the harddisk password are the same, if
361 gdm-autologin is used.
363 * When downloading tar or raw images using "machinectl
364 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
365 file is now also downloaded, if it is available and stored
366 next to the image file.
368 * Units of type ".socket" gained a new boolean setting
369 Writable= which is only useful in conjunction with
370 ListenSpecial=. If true, enables opening the specified
371 special file in O_RDWR mode rather than O_RDONLY mode.
373 * systemd-rfkill has been reworked to become a singleton
374 service that is activated through /dev/rfkill on each rfkill
375 state change and saves the settings to disk. This way,
376 systemd-rfkill is now compatible with devices that exist
377 only intermittendly, and even restores state if the previous
378 system shutdown was abrupt rather than clean.
380 * The journal daemon gained support for vacuuming old journal
381 files controlled by the number of files that shall remain,
382 in addition to the already existing control by size and by
383 date. This is useful as journal interleaving performance
384 degrades with too many seperate journal files, and allows
385 putting an effective limit on them. The new setting defaults
386 to 100, but this may be changed by setting SystemMaxFiles=
387 and RuntimeMaxFiles= in journald.conf. Also, the
388 "journalctl" tool gained the new --vacuum-files= switch to
389 manually vacuum journal files to leave only the specified
390 number of files in place.
392 * udev will now create /dev/disk/by-path links for ATA devices
393 on kernels where that is supported.
395 * Galician, Serbian, Turkish and Korean translations were added.
397 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
398 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
399 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
400 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
401 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
402 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
403 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
404 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
405 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
406 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
407 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
408 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
409 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
410 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
411 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
412 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
413 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
414 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
416 -- Berlin, 2015-10-07
420 * The DHCP implementation of systemd-networkd gained a set of
423 - The DHCP server now supports emitting DNS and NTP
424 information. It may be enabled and configured via
425 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
426 and NTP information is enabled, but no servers are
427 configured, the corresponding uplink information (if there
428 is any) is propagated.
430 - Server and client now support transmission and reception
431 of timezone information. It can be configured via the
432 newly introduced network options UseTimezone=,
433 EmitTimezone=, and Timezone=. Transmission of timezone
434 information is enabled between host and containers by
435 default now: the container will change its local timezone
436 to what the host has set.
438 - Lease timeouts can now be configured via
439 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
441 - The DHCP server improved on the stability of
442 leases. Clients are more likely to get the same lease
443 information back, even if the server loses state.
445 - The DHCP server supports two new configuration options to
446 control the lease address pool metrics, PoolOffset= and
449 * The encapsulation limit of tunnels in systemd-networkd may
450 now be configured via 'EncapsulationLimit='. It allows
451 modifying the maximum additional levels of encapsulation
452 that are permitted to be prepended to a packet.
454 * systemd now supports the concept of user buses replacing
455 session buses, if used with dbus-1.10 (and enabled via dbus
456 --enable-user-session). It previously only supported this on
457 kdbus-enabled systems, and this release expands this to
458 'dbus-daemon' systems.
460 * systemd-networkd now supports predictable interface names
463 * systemd now optionally supports the new Linux kernel
464 "unified" control group hierarchy. If enabled via the kernel
465 command-line option 'systemd.unified_cgroup_hierarchy=1',
466 systemd will try to mount the unified cgroup hierarchy
467 directly on /sys/fs/cgroup. If not enabled, or not
468 available, systemd will fall back to the legacy cgroup
469 hierarchy setup, as before. Host system and containers can
470 mix and match legacy and unified hierarchies as they
471 wish. nspawn understands the $UNIFIED_CROUP_HIERARCHY
472 environment variable to individually select the hierarchy to
473 use for executed containers. By default, nspawn will use the
474 unified hierarchy for the containers if the host uses the
475 unified hierarchy, and the legacy hierarchy otherwise.
476 Please note that at this point the unified hierarchy is an
477 experimental kernel feature and is likely to change in one
478 of the next kernel releases. Therefore, it should not be
479 enabled by default in downstream distributions yet. The
480 minimum required kernel version for the unified hierarchy to
481 work is 4.2. Note that when the unified hierarchy is used
482 for the first time delegated access to controllers is
483 safe. Because of this systemd-nspawn containers will get
484 access to controllers now, as will systemd user
485 sessions. This means containers and user sessions may now
486 manage their own resources, partitioning up what the system
489 * A new special scope unit "init.scope" has been introduced
490 that encapsulates PID 1 of the system. It may be used to
491 determine resource usage and enforce resource limits on PID
492 1 itself. PID 1 hence moved out of the root of the control
495 * The cgtop tool gained support for filtering out kernel
496 threads when counting tasks in a control group. Also, the
497 count of processes is now recursively summed up by
498 default. Two options -k and --recursive= have been added to
499 revert to old behaviour. The tool has also been updated to
500 work correctly in containers now.
502 * systemd-nspawn's --bind= and --bind-ro= options have been
503 extended to allow creation of non-recursive bind mounts.
505 * libsystemd gained two new calls sd_pid_get_cgroup() and
506 sd_peer_get_cgroup() which return the control group path of
507 a process or peer of a connected AF_UNIX socket. This
508 function call is particularly useful when implementing
509 delegated subtrees support in the control group hierarchy.
511 * The "sd-event" event loop API of libsystemd now supports
512 correct dequeuing of real-time signals, without losing
515 * When systemd requests a PolicyKit decision when managing
516 units it will now add additional fields to the request,
517 including unit name and desired operation. This enables more
518 powerful PolicyKit policies, that make decisions depending
521 * nspawn learnt support for .nspawn settings files, that may
522 accompany the image files or directories of containers, and
523 may contain additional settings for the container. This is
524 an alternative to configuring container parameters via the
527 Contributions from: Cristian Rodríguez, Daniel Mack, David
528 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
529 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
530 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
531 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
532 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
533 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
534 Andersen, Tom Gundersen, Torstein Husebø
536 -- Berlin, 2015-09-08
540 * machinectl gained a new verb 'shell' which opens a fresh
541 shell on the target container or the host. It is similar to
542 the existing 'login' command of machinectl, but spawns the
543 shell directly without prompting for username or
544 password. The pseudo machine '.host' now refers to the local
545 host and is used by default. Hence, 'machinectl shell' can
546 be used as replacement for 'su -' which spawns a session as
547 a fresh systemd unit in a way that is fully isolated from
548 the originating session.
550 * systemd-networkd learned to cope with private-zone DHCP
551 options and allows other programs to query the values.
553 * SELinux access control when enabling/disabling units is no
554 longer enforced with this release. The previous
555 implementation was incorrect, and a new corrected
556 implementation is not yet available. As unit file operations
557 are still protected via PolicyKit and D-Bus policy this is
558 not a security problem. Yet, distributions which care about
559 optimal SELinux support should probably not stabilize on
562 * sd-bus gained support for matches of type "arg0has=", that
563 test for membership of strings in string arrays sent in bus
566 * systemd-resolved now dumps the contents of its DNS and LLMNR
567 caches to the logs on reception of the SIGUSR1 signal. This
568 is useful to debug DNS behaviour.
570 * The coredumpctl tool gained a new --directory= option to
571 operate on journal files in a specific directory.
573 * "systemctl reboot" and related commands gained a new
574 "--message=" option which may be used to set a free-text
575 wall message when shutting down or rebooting the
576 system. This message is also logged, which is useful for
577 figuring out the reason for a reboot or shutdown a
580 * The "systemd-resolve-host" tool's -i switch now takes
581 network interface numbers as alternative to interface names.
583 * A new unit file setting for services has been introduced:
584 UtmpMode= allows configuration of how precisely systemd
585 handles utmp and wtmp entries for the service if this is
586 enabled. This allows writing services that appear similar to
587 user sessions in the output of the "w", "who", "last" and
590 * systemd-resolved will now locally synthesize DNS resource
591 records for the "localhost" and "gateway" domains as well as
592 the local hostname. This should ensure that clients querying
593 RRs via resolved will get similar results as those going via
594 NSS, if nss-myhostname is enabled.
596 Contributions from: Alastair Hughes, Alex Crawford, Daniel
597 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
598 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
599 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
600 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
601 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
602 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
603 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
604 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
605 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
606 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
607 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
609 -- Berlin, 2015-08-27
613 * The systemd-efi-boot-generator functionality was merged into
614 systemd-gpt-auto-generator.
616 * systemd-networkd now supports Group Policy for vxlan
617 devices. It can be enabled via the new boolean configuration
618 option called 'GroupPolicyExtension='.
620 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
621 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
622 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
624 -- Berlin, 2015-07-31
628 * The python-systemd code has been removed from the systemd repository.
629 A new repository has been created which accommodates the code from
630 now on, and we kindly ask distributions to create a separate package
631 for this: https://github.com/systemd/python-systemd
633 * The systemd daemon will now reload its main configuration
634 (/etc/systemd/system.conf) on daemon-reload.
636 * sd-dhcp now exposes vendor specific extensions via
637 sd_dhcp_lease_get_vendor_specific().
639 * systemd-networkd gained a number of new configuration options.
641 - A new boolean configuration option for TAP devices called
642 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
643 device, thus allowing to send and receive GSO packets.
645 - A new tunnel configuration option called 'CopyDSCP='.
646 If enabled, the DSCP field of ip6 tunnels is copied into the
649 - A set of boolean bridge configuration options were added.
650 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
651 and 'UnicastFlood=' are now parsed by networkd and applied to the
652 respective bridge link device via the respective IFLA_BRPORT_*
655 - A new string configuration option to override the hostname sent
656 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
657 is true, networkd will use the configured hostname instead of the
658 system hostname when sending DHCP requests.
660 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
661 networkd will configure the IPv6 flow-label of the tunnel device
662 according to RFC2460.
664 - The 'macvtap' virtual network devices are now supported, similar to
665 the already supported 'macvlan' devices.
667 * systemd-resolved now implements RFC5452 to improve resilience against
668 cache poisoning. Additionally, source port randomization is enabled
669 by default to further protect against DNS spoofing attacks.
671 * nss-mymachines now supports translating UIDs and GIDs of running
672 containers with user-namespaces enabled. If a container 'foo'
673 translates a host uid 'UID' to the container uid 'TUID', then
674 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
675 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
676 mapped as 'vg-foo-TGID'.
678 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
679 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
680 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
681 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
682 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
683 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
684 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
685 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
686 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
687 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
689 -- Berlin, 2015-07-29
693 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
694 There are no known issues with current sysfs, and udev does not need
695 or should be used to work around such bugs.
697 * udev does no longer enable USB HID power management. Several reports
698 indicate, that some devices cannot handle that setting.
700 * The udev accelerometer helper was removed. The functionality
701 is now fully included in iio-sensor-proxy. But this means,
702 older iio-sensor-proxy versions will no longer provide
703 accelerometer/orientation data with this systemd version.
704 Please upgrade iio-sensor-proxy to version 1.0.
706 * networkd gained a new configuration option IPv6PrivacyExtensions=
707 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
708 for Stateless Address") on selected networks.
710 * For the sake of fewer build-time dependencies and less code in the
711 main repository, the python bindings are about to be removed in the
712 next release. A new repository has been created which accommodates
713 the code from now on, and we kindly ask distributions to create a
714 separate package for this. The removal will take place in v223.
716 https://github.com/systemd/python-systemd
718 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
719 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
720 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
721 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
722 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
723 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
724 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
725 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
726 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
727 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
729 -- Berlin, 2015-07-07
733 * The sd-bus.h and sd-event.h APIs have now been declared
734 stable and have been added to the official interface of
735 libsystemd.so. sd-bus implements an alternative D-Bus client
736 library, that is relatively easy to use, very efficient and
737 supports both classic D-Bus as well as kdbus as transport
738 backend. sd-event is a generic event loop abstraction that
739 is built around Linux epoll, but adds features such as event
740 prioritization or efficient timer handling. Both APIs are good
741 choices for C programs looking for a bus and/or event loop
742 implementation that is minimal and does not have to be
743 portable to other kernels.
745 * kdbus support is no longer compile-time optional. It is now
746 always built-in. However, it can still be disabled at
747 runtime using the kdbus=0 kernel command line setting, and
748 that setting may be changed to default to off, by specifying
749 --disable-kdbus at build-time. Note though that the kernel
750 command line setting has no effect if the kdbus.ko kernel
751 module is not installed, in which case kdbus is (obviously)
752 also disabled. We encourage all downstream distributions to
753 begin testing kdbus by adding it to the kernel images in the
754 development distributions, and leaving kdbus support in
757 * The minimal required util-linux version has been bumped to
760 * Support for chkconfig (--enable-chkconfig) was removed in
761 favor of calling an abstraction tool
762 /lib/systemd/systemd-sysv-install. This needs to be
763 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
764 in README for details.
766 * If there's a systemd unit and a SysV init script for the
767 same service name, and the user executes "systemctl enable"
768 for it (or a related call), then this will now enable both
769 (or execute the related operation on both), not just the
772 * The libudev API documentation has been converted from gtkdoc
775 * gudev has been removed from the systemd tree, it is now an
778 * The systemd-cgtop tool learnt a new --raw switch to generate
779 "raw" (machine parsable) output.
781 * networkd's IPForwarding= .network file setting learnt the
782 new setting "kernel", which ensures that networkd does not
783 change the IP forwarding sysctl from the default kernel
786 * The systemd-logind bus API now exposes a new boolean
787 property "Docked" that reports whether logind considers the
788 system "docked", i.e. connected to a docking station or not.
790 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
791 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
792 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
793 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
794 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
795 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
796 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
797 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
798 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
799 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
800 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
801 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
802 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
803 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
804 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
805 Fink, Zbigniew Jędrzejewski-Szmek
807 -- Berlin, 2015-06-19
811 * The gudev library has been extracted into a separate repository
812 available at: https://git.gnome.org/browse/libgudev/
813 It is now managed as part of the Gnome project. Distributions
814 are recommended to pass --disable-gudev to systemd and use
815 gudev from the Gnome project instead. gudev is still included
816 in systemd, for now. It will be removed soon, though. Please
817 also see the announcement-thread on systemd-devel:
818 http://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
820 * systemd now exposes a CPUUsageNSec= property for each
821 service unit on the bus, that contains the overall consumed
822 CPU time of a service (the sum of what each process of the
823 service consumed). This value is only available if
824 CPUAccounting= is turned on for a service, and is then shown
825 in the "systemctl status" output.
827 * Support for configuring alternative mappings of the old SysV
828 runlevels to systemd targets has been removed. They are now
829 hardcoded in a way that runlevels 2, 3, 4 all map to
830 multi-user.target and 5 to graphical.target (which
831 previously was already the default behaviour).
833 * The auto-mounter logic gained support for mount point
834 expiry, using a new TimeoutIdleSec= setting in .automount
835 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
837 * The EFI System Partition (ESP) as mounted to /boot by
838 systemd-efi-boot-generator will now be unmounted
839 automatically after 2 minutes of not being used. This should
840 minimize the risk of ESP corruptions.
842 * New /etc/fstab options x-systemd.requires= and
843 x-systemd.requires-mounts-for= are now supported to express
844 additional dependencies for mounts. This is useful for
845 journalling file systems that support external journal
846 devices or overlay file systems that require underlying file
847 systems to be mounted.
849 * systemd does not support direct live-upgrades (via systemctl
850 daemon-reexec) from versions older than v44 anymore. As no
851 distribution we are aware of shipped such old versions in a
852 stable release this should not be problematic.
854 * When systemd forks off a new per-connection service instance
855 it will now set the $REMOTE_ADDR environment variable to the
856 remote IP address, and $REMOTE_PORT environment variable to
857 the remote IP port. This behaviour is similar to the
858 corresponding environment variables defined by CGI.
860 * systemd-networkd gained support for uplink failure
861 detection. The BindCarrier= option allows binding interface
862 configuration dynamically to the link sense of other
863 interfaces. This is useful to achieve behaviour like in
866 * systemd-networkd gained support for configuring the DHCP
867 client identifier to use when requesting leases.
869 * systemd-networkd now has a per-network UseNTP= option to
870 configure whether NTP server information acquired via DHCP
871 is passed on to services like systemd-timesyncd.
873 * systemd-networkd gained support for vti6 tunnels.
875 * Note that systemd-networkd manages the sysctl variable
876 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
877 it is configured for since v219. The variable controls IP
878 forwarding, and is a per-interface alternative to the global
879 /proc/sys/net/ipv[46]/ip_forward. This setting is
880 configurable in the IPForward= option, which defaults to
881 "no". This means if networkd is used for an interface it is
882 no longer sufficient to set the global sysctl option to turn
883 on IP forwarding! Instead, the .network file option
884 IPForward= needs to be turned on! Note that the
885 implementation of this behaviour was broken in v219 and has
888 * Many bonding and vxlan options are now configurable in
891 * systemd-nspawn gained a new --property= setting to set unit
892 properties for the container scope. This is useful for
893 setting resource parameters (e.g "CPUShares=500") on
894 containers started from the command line.
896 * systemd-nspawn gained a new --private-users= switch to make
897 use of user namespacing available on recent Linux kernels.
899 * systemd-nspawn may now be called as part of a shell pipeline
900 in which case the pipes used for stdin and stdout are passed
901 directly to the process invoked in the container, without
902 indirection via a pseudo tty.
904 * systemd-nspawn gained a new switch to control the UNIX
905 signal to use when killing the init process of the container
908 * systemd-nspawn gained a new --overlay= switch for mounting
909 overlay file systems into the container using the new kernel
912 * When a container image is imported via systemd-importd and
913 the host file system is not btrfs, a loopback block device
914 file is created in /var/lib/machines.raw with a btrfs file
915 system inside. It is then mounted to /var/lib/machines to
916 enable btrfs features for container management. The loopback
917 file and btrfs file system is grown as needed when container
918 images are imported via systemd-importd.
920 * systemd-machined/systemd-importd gained support for btrfs
921 quota, to enforce container disk space limits on disk. This
922 is exposed in "machinectl set-limit".
924 * systemd-importd now can import containers from local .tar,
925 .raw and .qcow2 images, and export them to .tar and .raw. It
926 can also import dkr v2 images now from the network (on top
929 * systemd-importd gained support for verifying downloaded
930 images with gpg2 (previously only gpg1 was supported).
932 * systemd-machined, systemd-logind, systemd: most bus calls
933 are now accessible to unprivileged processes via
934 PolicyKit. Also, systemd-logind will now allow users to kill
935 their own sessions without further privileges or
938 * systemd-shutdownd has been removed. This service was
939 previously responsible for implementing scheduled shutdowns
940 as exposed in /usr/bin/shutdown's time parameter. This
941 functionality has now been moved into systemd-logind and is
942 accessible via a bus interface.
944 * "systemctl reboot" gained a new switch --firmware-setup that
945 can be used to reboot into the EFI firmware setup, if that
946 is available. systemd-logind now exposes an API on the bus
947 to trigger such reboots, in case graphical desktop UIs want
948 to cover this functionality.
950 * "systemctl enable", "systemctl disable" and "systemctl mask"
951 now support a new "--now" switch. If specified the units
952 that are enabled will also be started, and the ones
953 disabled/masked also stopped.
955 * The Gummiboot EFI boot loader tool has been merged into
956 systemd, and renamed to "systemd-boot". The bootctl tool has been
957 updated to support systemd-boot.
959 * An EFI kernel stub has been added that may be used to create
960 kernel EFI binaries that contain not only the actual kernel,
961 but also an initrd, boot splash, command line and OS release
962 information. This combined binary can then be signed as a
963 single image, so that the firmware can verify it all in one
964 step. systemd-boot has special support for EFI binaries created
965 like this and can extract OS release information from them
966 and show them in the boot menu. This functionality is useful
967 to implement cryptographically verified boot schemes.
969 * Optional support has been added to systemd-fsck to pass
970 fsck's progress report to an AF_UNIX socket in the file
973 * udev will no longer create device symlinks for all block
974 devices by default. A blacklist for excluding special block
975 devices from this logic has been turned into a whitelist
976 that requires picking block devices explicitly that require
979 * A new (currently still internal) API sd-device.h has been
980 added to libsystemd. This modernized API is supposed to
981 replace libudev eventually. In fact, already much of libudev
982 is now just a wrapper around sd-device.h.
984 * A new hwdb database for storing metadata about pointing
985 stick devices has been added.
987 * systemd-tmpfiles gained support for setting file attributes
988 similar to the "chattr" tool with new 'h' and 'H' lines.
990 * systemd-journald will no longer unconditionally set the
991 btrfs NOCOW flag on new journal files. This is instead done
992 with tmpfiles snippet using the new 'h' line type. This
993 allows easy disabling of this logic, by masking the
994 journal-nocow.conf tmpfiles file.
996 * systemd-journald will now translate audit message types to
997 human readable identifiers when writing them to the
998 journal. This should improve readability of audit messages.
1000 * The LUKS logic gained support for the offset= and skip=
1001 options in /etc/crypttab, as previously implemented by
1004 * /usr/lib/os-release gained a new optional field VARIANT= for
1005 distributions that support multiple variants (such as a
1006 desktop edition, a server edition, ...)
1008 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
1009 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
1010 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
1011 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
1012 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
1013 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
1014 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
1015 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
1016 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
1017 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
1018 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
1019 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
1020 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
1021 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
1022 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
1023 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
1024 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
1025 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
1026 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
1027 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
1028 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
1029 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
1030 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
1031 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
1032 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
1033 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
1034 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
1036 -- Berlin, 2015-05-22
1040 * Introduce a new API "sd-hwdb.h" for querying the hardware
1041 metadata database. With this minimal interface one can query
1042 and enumerate the udev hwdb, decoupled from the old libudev
1043 library. libudev's interface for this is now only a wrapper
1044 around sd-hwdb. A new tool systemd-hwdb has been added to
1045 interface with and update the database.
1047 * When any of systemd's tools copies files (for example due to
1048 tmpfiles' C lines) a btrfs reflink will attempted first,
1049 before bytewise copying is done.
1051 * systemd-nspawn gained a new --ephemeral switch. When
1052 specified a btrfs snapshot is taken of the container's root
1053 directory, and immediately removed when the container
1054 terminates again. Thus, a container can be started whose
1055 changes never alter the container's root directory, and are
1056 lost on container termination. This switch can also be used
1057 for starting a container off the root file system of the
1058 host without affecting the host OS. This switch is only
1059 available on btrfs file systems.
1061 * systemd-nspawn gained a new --template= switch. It takes the
1062 path to a container tree to use as template for the tree
1063 specified via --directory=, should that directory be
1064 missing. This allows instantiating containers dynamically,
1065 on first run. This switch is only available on btrfs file
1068 * When a .mount unit refers to a mount point on which multiple
1069 mounts are stacked, and the .mount unit is stopped all of
1070 the stacked mount points will now be unmounted until no
1071 mount point remains.
1073 * systemd now has an explicit notion of supported and
1074 unsupported unit types. Jobs enqueued for unsupported unit
1075 types will now fail with an "unsupported" error code. More
1076 specifically .swap, .automount and .device units are not
1077 supported in containers, .busname units are not supported on
1078 non-kdbus systems. .swap and .automount are also not
1079 supported if their respective kernel compile time options
1082 * machinectl gained support for two new "copy-from" and
1083 "copy-to" commands for copying files from a running
1084 container to the host or vice versa.
1086 * machinectl gained support for a new "bind" command to bind
1087 mount host directories into local containers. This is
1088 currently only supported for nspawn containers.
1090 * networkd gained support for configuring bridge forwarding
1091 database entries (fdb) from .network files.
1093 * A new tiny daemon "systemd-importd" has been added that can
1094 download container images in tar, raw, qcow2 or dkr formats,
1095 and make them available locally in /var/lib/machines, so
1096 that they can run as nspawn containers. The daemon can GPG
1097 verify the downloads (not supported for dkr, since it has no
1098 provisions for verifying downloads). It will transparently
1099 decompress bz2, xz, gzip compressed downloads if necessary,
1100 and restore sparse files on disk. The daemon uses privilege
1101 separation to ensure the actual download logic runs with
1102 fewer privileges than the daemon itself. machinectl has
1103 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
1104 make the functionality of importd available to the
1105 user. With this in place the Fedora and Ubuntu "Cloud"
1106 images can be downloaded and booted as containers unmodified
1107 (the Fedora images lack the appropriate GPG signature files
1108 currently, so they cannot be verified, but this will change
1109 soon, hopefully). Note that downloading images is currently
1110 only fully supported on btrfs.
1112 * machinectl is now able to list container images found in
1113 /var/lib/machines, along with some metadata about sizes of
1114 disk and similar. If the directory is located on btrfs and
1115 quota is enabled, this includes quota display. A new command
1116 "image-status" has been added that shows additional
1117 information about images.
1119 * machinectl is now able to clone container images
1120 efficiently, if the underlying file system (btrfs) supports
1121 it, with the new "machinectl list-images" command. It also
1122 gained commands for renaming and removing images, as well as
1123 marking them read-only or read-write (supported also on
1124 legacy file systems).
1126 * networkd gained support for collecting LLDP network
1127 announcements, from hardware that supports this. This is
1128 shown in networkctl output.
1130 * systemd-run gained support for a new -t (--pty) switch for
1131 invoking a binary on a pty whose input and output is
1132 connected to the invoking terminal. This allows executing
1133 processes as system services while interactively
1134 communicating with them via the terminal. Most interestingly
1135 this is supported across container boundaries. Invoking
1136 "systemd-run -t /bin/bash" is an alternative to running a
1137 full login session, the difference being that the former
1138 will not register a session, nor go through the PAM session
1141 * tmpfiles gained support for a new "v" line type for creating
1142 btrfs subvolumes. If the underlying file system is a legacy
1143 file system, this automatically degrades to creating a
1144 normal directory. Among others /var/lib/machines is now
1145 created like this at boot, should it be missing.
1147 * The directory /var/lib/containers/ has been deprecated and
1148 been replaced by /var/lib/machines. The term "machines" has
1149 been used in the systemd context as generic term for both
1150 VMs and containers, and hence appears more appropriate for
1151 this, as the directory can also contain raw images bootable
1154 * systemd-nspawn when invoked with -M but without --directory=
1155 or --image= is now capable of searching for the container
1156 root directory, subvolume or disk image automatically, in
1157 /var/lib/machines. systemd-nspawn@.service has been updated
1158 to make use of this, thus allowing it to be used for raw
1161 * A new machines.target unit has been introduced that is
1162 supposed to group all containers/VMs invoked as services on
1163 the system. systemd-nspawn@.service has been updated to
1164 integrate with that.
1166 * machinectl gained a new "start" command, for invoking a
1167 container as a service. "machinectl start foo" is mostly
1168 equivalent to "systemctl start systemd-nspawn@foo.service",
1169 but handles escaping in a nicer way.
1171 * systemd-nspawn will now mount most of the cgroupfs tree
1172 read-only into each container, with the exception of the
1173 container's own subtree in the name=systemd hierarchy.
1175 * journald now sets the special FS_NOCOW file flag for its
1176 journal files. This should improve performance on btrfs, by
1177 avoiding heavy fragmentation when journald's write-pattern
1178 is used on COW file systems. It degrades btrfs' data
1179 integrity guarantees for the files to the same levels as for
1180 ext3/ext4 however. This should be OK though as journald does
1181 its own data integrity checks and all its objects are
1182 checksummed on disk. Also, journald should handle btrfs disk
1183 full events a lot more gracefully now, by processing SIGBUS
1184 errors, and not relying on fallocate() anymore.
1186 * When journald detects that journal files it is writing to
1187 have been deleted it will immediately start new journal
1190 * systemd now provides a way to store file descriptors
1191 per-service in PID 1.This is useful for daemons to ensure
1192 that fds they require are not lost during a daemon
1193 restart. The fds are passed to the daemon on the next
1194 invocation in the same way socket activation fds are
1195 passed. This is now used by journald to ensure that the
1196 various sockets connected to all the system's stdout/stderr
1197 are not lost when journald is restarted. File descriptors
1198 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
1199 an extension to sd_notify(). Note that a limit is enforced
1200 on the number of fds a service can store in PID 1, and it
1201 defaults to 0, so that no fds may be stored, unless this is
1202 explicitly turned on.
1204 * The default TERM variable to use for units connected to a
1205 terminal, when no other value is explicitly is set is now
1206 vt220 rather than vt102. This should be fairly safe still,
1207 but allows PgUp/PgDn work.
1209 * The /etc/crypttab option header= as known from Debian is now
1212 * "loginctl user-status" and "loginctl session-status" will
1213 now show the last 10 lines of log messages of the
1214 user/session following the status output. Similar,
1215 "machinectl status" will show the last 10 log lines
1216 associated with a virtual machine or container
1217 service. (Note that this is usually not the log messages
1218 done in the VM/container itself, but simply what the
1219 container manager logs. For nspawn this includes all console
1222 * "loginctl session-status" without further argument will now
1223 show the status of the session of the caller. Similar,
1224 "lock-session", "unlock-session", "activate",
1225 "enable-linger", "disable-linger" may now be called without
1226 session/user parameter in which case they apply to the
1227 caller's session/user.
1229 * An X11 session scriptlet is now shipped that uploads
1230 $DISPLAY and $XAUTHORITY into the environment of the systemd
1231 --user daemon if a session begins. This should improve
1232 compatibility with X11 enabled applications run as systemd
1235 * Generators are now subject to masking via /etc and /run, the
1236 same way as unit files.
1238 * networkd .network files gained support for configuring
1239 per-link IPv4/IPv6 packet forwarding as well as IPv4
1240 masquerading. This is by default turned on for veth links to
1241 containers, as registered by systemd-nspawn. This means that
1242 nspawn containers run with --network-veth will now get
1243 automatic routed access to the host's networks without any
1244 further configuration or setup, as long as networkd runs on
1247 * systemd-nspawn gained the --port= (-p) switch to expose TCP
1248 or UDP posts of a container on the host. With this in place
1249 it is possible to run containers with private veth links
1250 (--network-veth), and have their functionality exposed on
1251 the host as if their services were running directly on the
1254 * systemd-nspawn's --network-veth switch now gained a short
1255 version "-n", since with the changes above it is now truly
1256 useful out-of-the-box. The systemd-nspawn@.service has been
1257 updated to make use of it too by default.
1259 * systemd-nspawn will now maintain a per-image R/W lock, to
1260 ensure that the same image is not started more than once
1261 writable. (It's OK to run an image multiple times
1262 simultaneously in read-only mode.)
1264 * systemd-nspawn's --image= option is now capable of
1265 dissecting and booting MBR and GPT disk images that contain
1266 only a single active Linux partition. Previously it
1267 supported only GPT disk images with proper GPT type
1268 IDs. This allows running cloud images from major
1269 distributions directly with systemd-nspawn, without
1272 * In addition to collecting mouse dpi data in the udev
1273 hardware database, there's now support for collecting angle
1274 information for mouse scroll wheels. The database is
1275 supposed to guarantee similar scrolling behavior on mice
1276 that it knows about. There's also support for collecting
1277 information about Touchpad types.
1279 * udev's input_id built-in will now also collect touch screen
1280 dimension data and attach it to probed devices.
1282 * /etc/os-release gained support for a Distribution Privacy
1285 * networkd gained support for creating "ipvlan", "gretap",
1286 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
1288 * systemd-tmpfiles gained support for "a" lines for setting
1291 * systemd-nspawn will now mount /tmp in the container to
1292 tmpfs, automatically.
1294 * systemd now exposes the memory.usage_in_bytes cgroup
1295 attribute and shows it for each service in the "systemctl
1296 status" output, if available.
1298 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
1299 immediate reboot is triggered. This useful if shutdown is
1300 hung and is unable to complete, to expedite the
1301 operation. Note that this kind of reboot will still unmount
1302 all file systems, and hence should not result in fsck being
1305 * A .device unit for an optical block device will now be
1306 considered active only when a medium is in the drive. Also,
1307 mount units are now bound to their backing devices thus
1308 triggering automatic unmounting when devices become
1309 unavailable. With this in place systemd will now
1310 automatically unmount left-over mounts when a CD-ROM is
1311 ejected or an USB stick is yanked from the system.
1313 * networkd-wait-online now has support for waiting for
1314 specific interfaces only (with globbing), and for giving up
1315 after a configurable timeout.
1317 * networkd now exits when idle. It will be automatically
1318 restarted as soon as interfaces show up, are removed or
1319 change state. networkd will stay around as long as there is
1320 at least one DHCP state machine or similar around, that keep
1323 * networkd may now configure IPv6 link-local addressing in
1324 addition to IPv4 link-local addressing.
1326 * The IPv6 "token" for use in SLAAC may now be configured for
1327 each .network interface in networkd.
1329 * Routes configured with networkd may now be assigned a scope
1332 * networkd's [Match] sections now support globbing and lists
1333 of multiple space-separated matches per item.
1335 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
1336 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
1337 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
1338 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
1339 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
1340 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
1341 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
1342 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
1343 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
1344 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
1345 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
1346 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
1347 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
1348 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
1349 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
1350 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
1351 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
1352 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
1353 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
1354 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
1355 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
1356 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
1357 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
1358 Hoffmann, Zbigniew Jędrzejewski-Szmek
1360 -- Berlin, 2015-02-16
1364 * When querying unit file enablement status (for example via
1365 "systemctl is-enabled"), a new state "indirect" is now known
1366 which indicates that a unit might not be enabled itself, but
1367 another unit listed in its Also= setting might be.
1369 * Similar to the various existing ConditionXYZ= settings for
1370 units, there are now matching AssertXYZ= settings. While
1371 failing conditions cause a unit to be skipped, but its job
1372 to succeed, failing assertions declared like this will cause
1373 a unit start operation and its job to fail.
1375 * hostnamed now knows a new chassis type "embedded".
1377 * systemctl gained a new "edit" command. When used on a unit
1378 file, this allows extending unit files with .d/ drop-in
1379 configuration snippets or editing the full file (after
1380 copying it from /usr/lib to /etc). This will invoke the
1381 user's editor (as configured with $EDITOR), and reload the
1382 modified configuration after editing.
1384 * "systemctl status" now shows the suggested enablement state
1385 for a unit, as declared in the (usually vendor-supplied)
1386 system preset files.
1388 * nss-myhostname will now resolve the single-label host name
1389 "gateway" to the locally configured default IP routing
1390 gateways, ordered by their metrics. This assigns a stable
1391 name to the used gateways, regardless which ones are
1392 currently configured. Note that the name will only be
1393 resolved after all other name sources (if nss-myhostname is
1394 configured properly) and should hence not negatively impact
1395 systems that use the single-label host name "gateway" in
1398 * systemd-inhibit now allows filtering by mode when listing
1401 * Scope and service units gained a new "Delegate" boolean
1402 property, which, when set, allows processes running inside the
1403 unit to further partition resources. This is primarily
1404 useful for systemd user instances as well as container
1407 * journald will now pick up audit messages directly from
1408 the kernel, and log them like any other log message. The
1409 audit fields are split up and fully indexed. This means that
1410 journalctl in many ways is now a (nicer!) alternative to
1411 ausearch, the traditional audit client. Note that this
1412 implements only a minimal audit client. If you want the
1413 special audit modes like reboot-on-log-overflow, please use
1414 the traditional auditd instead, which can be used in
1415 parallel to journald.
1417 * The ConditionSecurity= unit file option now understands the
1418 special string "audit" to check whether auditing is
1421 * journalctl gained two new commands --vacuum-size= and
1422 --vacuum-time= to delete old journal files until the
1423 remaining ones take up no more than the specified size on disk,
1424 or are not older than the specified time.
1426 * A new, native PPPoE library has been added to sd-network,
1427 systemd's library of light-weight networking protocols. This
1428 library will be used in a future version of networkd to
1429 enable PPPoE communication without an external pppd daemon.
1431 * The busctl tool now understands a new "capture" verb that
1432 works similar to "monitor", but writes a packet capture
1433 trace to STDOUT that can be redirected to a file which is
1434 compatible with libcap's capture file format. This can then
1435 be loaded in Wireshark and similar tools to inspect bus
1438 * The busctl tool now understands a new "tree" verb that shows
1439 the object trees of a specific service on the bus, or of all
1442 * The busctl tool now understands a new "introspect" verb that
1443 shows all interfaces and members of objects on the bus,
1444 including their signature and values. This is particularly
1445 useful to get more information about bus objects shown by
1446 the new "busctl tree" command.
1448 * The busctl tool now understands new verbs "call",
1449 "set-property" and "get-property" for invoking bus method
1450 calls, setting and getting bus object properties in a
1453 * busctl gained a new --augment-creds= argument that controls
1454 whether the tool shall augment credential information it
1455 gets from the bus with data from /proc, in a possibly
1458 * nspawn's --link-journal= switch gained two new values
1459 "try-guest" and "try-host" that work like "guest" and
1460 "host", but do not fail if the host has no persistent
1461 journalling enabled. -j is now equivalent to
1462 --link-journal=try-guest.
1464 * macvlan network devices created by nspawn will now have
1465 stable MAC addresses.
1467 * A new SmackProcessLabel= unit setting has been added, which
1468 controls the SMACK security label processes forked off by
1469 the respective unit shall use.
1471 * If compiled with --enable-xkbcommon, systemd-localed will
1472 verify x11 keymap settings by compiling the given keymap. It
1473 will spew out warnings if the compilation fails. This
1474 requires libxkbcommon to be installed.
1476 * When a coredump is collected, a larger number of metadata
1477 fields is now collected and included in the journal records
1478 created for it. More specifically, control group membership,
1479 environment variables, memory maps, working directory,
1480 chroot directory, /proc/$PID/status, and a list of open file
1481 descriptors is now stored in the log entry.
1483 * The udev hwdb now contains DPI information for mice. For
1486 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
1488 * All systemd programs that read standalone configuration
1489 files in /etc now also support a corresponding series of
1490 .conf.d configuration directories in /etc/, /run/,
1491 /usr/local/lib/, /usr/lib/, and (if configured with
1492 --enable-split-usr) /lib/. In particular, the following
1493 configuration files now have corresponding configuration
1494 directories: system.conf user.conf, logind.conf,
1495 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
1496 resolved.conf, timesyncd.conf, journal-remote.conf, and
1497 journal-upload.conf. Note that distributions should use the
1498 configuration directories in /usr/lib/; the directories in
1499 /etc/ are reserved for the system administrator.
1501 * systemd-rfkill will no longer take the rfkill device name
1502 into account when storing rfkill state on disk, as the name
1503 might be dynamically assigned and not stable. Instead, the
1504 ID_PATH udev variable combined with the rfkill type (wlan,
1505 bluetooth, ...) is used.
1507 * A new service systemd-machine-id-commit.service has been
1508 added. When used on systems where /etc is read-only during
1509 boot, and /etc/machine-id is not initialized (but an empty
1510 file), this service will copy the temporary machine ID
1511 created as replacement into /etc after the system is fully
1512 booted up. This is useful for systems that are freshly
1513 installed with a non-initialized machine ID, but should get
1514 a fixed machine ID for subsequent boots.
1516 * networkd's .netdev files now provide a large set of
1517 configuration parameters for VXLAN devices. Similarly, the
1518 bridge port cost parameter is now configurable in .network
1519 files. There's also new support for configuring IP source
1520 routing. networkd .link files gained support for a new
1521 OriginalName= match that is useful to match against the
1522 original interface name the kernel assigned. .network files
1523 may include MTU= and MACAddress= fields for altering the MTU
1524 and MAC address while being connected to a specific network
1527 * The LUKS logic gained supported for configuring
1528 UUID-specific key files. There's also new support for naming
1529 LUKS device from the kernel command line, using the new
1530 luks.name= argument.
1532 * Timer units may now be transiently created via the bus API
1533 (this was previously already available for scope and service
1534 units). In addition it is now possible to create multiple
1535 transient units at the same time with a single bus call. The
1536 "systemd-run" tool has been updated to make use of this for
1537 running commands on a specified time, in at(1)-style.
1539 * tmpfiles gained support for "t" lines, for assigning
1540 extended attributes to files. Among other uses this may be
1541 used to assign SMACK labels to files.
1543 Contributions from: Alin Rauta, Alison Chaiken, Andrej
1544 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
1545 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
1546 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
1547 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
1548 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
1549 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
1550 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
1551 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
1552 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
1553 Michael Biebl, Michael Chapman, Michael Marineau, Michal
1554 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
1555 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
1556 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
1557 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
1558 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
1559 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
1560 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
1562 -- Berlin, 2014-12-10
1566 * journalctl gained the new options -t/--identifier= to match
1567 on the syslog identifier (aka "tag"), as well as --utc to
1568 show log timestamps in the UTC timezone. journalctl now also
1569 accepts -n/--lines=all to disable line capping in a pager.
1571 * journalctl gained a new switch, --flush, that synchronously
1572 flushes logs from /run/log/journal to /var/log/journal if
1573 persistent storage is enabled. systemd-journal-flush.service
1574 now waits until the operation is complete.
1576 * Services can notify the manager before they start a reload
1577 (by sending RELOADING=1) or shutdown (by sending
1578 STOPPING=1). This allows the manager to track and show the
1579 internal state of daemons and closes a race condition when
1580 the process is still running but has closed its D-Bus
1583 * Services with Type=oneshot do not have to have any ExecStart
1586 * User units are now loaded also from
1587 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
1588 /run/systemd/user directory that was already previously
1589 supported, but is under the control of the user.
1591 * Job timeouts (i.e. time-outs on the time a job that is
1592 queued stays in the run queue) can now optionally result in
1593 immediate reboot or power-off actions (JobTimeoutAction= and
1594 JobTimeoutRebootArgument=). This is useful on ".target"
1595 units, to limit the maximum time a target remains
1596 undispatched in the run queue, and to trigger an emergency
1597 operation in such a case. This is now used by default to
1598 turn off the system if boot-up (as defined by everything in
1599 basic.target) hangs and does not complete for at least
1600 15min. Also, if power-off or reboot hang for at least 30min
1601 an immediate power-off/reboot operation is triggered. This
1602 functionality is particularly useful to increase reliability
1603 on embedded devices, but also on laptops which might
1604 accidentally get powered on when carried in a backpack and
1605 whose boot stays stuck in a hard disk encryption passphrase
1608 * systemd-logind can be configured to also handle lid switch
1609 events even when the machine is docked or multiple displays
1610 are attached (HandleLidSwitchDocked= option).
1612 * A helper binary and a service have been added which can be
1613 used to resume from hibernation in the initramfs. A
1614 generator will parse the resume= option on the kernel
1615 command line to trigger resume.
1617 * A user console daemon systemd-consoled has been
1618 added. Currently, it is a preview, and will so far open a
1619 single terminal on each session of the user marked as
1620 Desktop=systemd-console.
1622 * Route metrics can be specified for DHCP routes added by
1625 * The SELinux context of socket-activated services can be set
1626 from the information provided by the networking stack
1627 (SELinuxContextFromNet= option).
1629 * Userspace firmware loading support has been removed and
1630 the minimum supported kernel version is thus bumped to 3.7.
1632 * Timeout for udev workers has been increased from 1 to 3
1633 minutes, but a warning will be printed after 1 minute to
1634 help diagnose kernel modules that take a long time to load.
1636 * Udev rules can now remove tags on devices with TAG-="foobar".
1638 * systemd's readahead implementation has been removed. In many
1639 circumstances it didn't give expected benefits even for
1640 rotational disk drives and was becoming less relevant in the
1641 age of SSDs. As none of the developers has been using
1642 rotating media anymore, and nobody stepped up to actively
1643 maintain this component of systemd it has now been removed.
1645 * Swap units can use Options= to specify discard options.
1646 Discard options specified for swaps in /etc/fstab are now
1649 * Docker containers are now detected as a separate type of
1652 * The Password Agent protocol gained support for queries where
1653 the user input is shown, useful e.g. for user names.
1654 systemd-ask-password gained a new --echo option to turn that
1657 * The default sysctl.d/ snippets will now set:
1659 net.core.default_qdisc = fq_codel
1661 This selects Fair Queuing Controlled Delay as the default
1662 queuing discipline for network interfaces. fq_codel helps
1663 fight the network bufferbloat problem. It is believed to be
1664 a good default with no tuning required for most workloads.
1665 Downstream distributions may override this choice. On 10Gbit
1666 servers that do not do forwarding, "fq" may perform better.
1667 Systems without a good clocksource should use "pfifo_fast".
1669 * If kdbus is enabled during build a new option BusPolicy= is
1670 available for service units, that allows locking all service
1671 processes into a stricter bus policy, in order to limit
1672 access to various bus services, or even hide most of them
1673 from the service's view entirely.
1675 * networkctl will now show the .network and .link file
1676 networkd has applied to a specific interface.
1678 * sd-login gained a new API call sd_session_get_desktop() to
1679 query which desktop environment has been selected for a
1682 * UNIX utmp support is now compile-time optional to support
1683 legacy-free systems.
1685 * systemctl gained two new commands "add-wants" and
1686 "add-requires" for pulling in units from specific targets
1689 * If the word "rescue" is specified on the kernel command line
1690 the system will now boot into rescue mode (aka
1691 rescue.target), which was previously available only by
1692 specifying "1" or "systemd.unit=rescue.target" on the kernel
1693 command line. This new kernel command line option nicely
1694 mirrors the already existing "emergency" kernel command line
1697 * New kernel command line options mount.usr=, mount.usrflags=,
1698 mount.usrfstype= have been added that match root=, rootflags=,
1699 rootfstype= but allow mounting a specific file system to
1702 * The $NOTIFY_SOCKET is now also passed to control processes of
1703 services, not only the main process.
1705 * This version reenables support for fsck's -l switch. This
1706 means at least version v2.25 of util-linux is required for
1707 operation, otherwise dead-locks on device nodes may
1708 occur. Again: you need to update util-linux to at least
1709 v2.25 when updating systemd to v217.
1711 * The "multi-seat-x" tool has been removed from systemd, as
1712 its functionality has been integrated into X servers 1.16,
1713 and the tool is hence redundant. It is recommended to update
1714 display managers invoking this tool to simply invoke X
1715 directly from now on, again.
1717 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
1718 message flag has been added for all of systemd's PolicyKit
1719 authenticated method calls has been added. In particular
1720 this now allows optional interactive authorization via
1721 PolicyKit for many of PID1's privileged operations such as
1722 unit file enabling and disabling.
1724 * "udevadm hwdb --update" learnt a new switch "--usr" for
1725 placing the rebuilt hardware database in /usr instead of
1726 /etc. When used only hardware database entries stored in
1727 /usr will be used, and any user database entries in /etc are
1728 ignored. This functionality is useful for vendors to ship a
1729 pre-built database on systems where local configuration is
1730 unnecessary or unlikely.
1732 * Calendar time specifications in .timer units now also
1733 understand the strings "semi-annually", "quarterly" and
1734 "minutely" as shortcuts (in addition to the preexisting
1735 "anually", "hourly", ...).
1737 * systemd-tmpfiles will now correctly create files in /dev
1738 at boot which are marked for creation only at boot. It is
1739 recommended to always create static device nodes with 'c!'
1740 and 'b!', so that they are created only at boot and not
1741 overwritten at runtime.
1743 * When the watchdog logic is used for a service (WatchdogSec=)
1744 and the watchdog timeout is hit the service will now be
1745 terminated with SIGABRT (instead of just SIGTERM), in order
1746 to make sure a proper coredump and backtrace is
1747 generated. This ensures that hanging services will result in
1748 similar coredump/backtrace behaviour as services that hit a
1751 Contributions from: Andreas Henriksson, Andrei Borzenkov,
1752 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
1753 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
1754 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
1755 Herrmann, David Sommerseth, David Strauss, Emil Renner
1756 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
1757 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
1758 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
1759 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
1760 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
1761 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
1762 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
1763 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
1764 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
1765 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
1766 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
1767 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
1768 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
1769 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
1770 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
1771 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
1774 -- Berlin, 2014-10-28
1778 * timedated no longer reads NTP implementation unit names from
1779 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
1780 implementations should add a
1782 Conflicts=systemd-timesyncd.service
1784 to their unit files to take over and replace systemd's NTP
1785 default functionality.
1787 * systemd-sysusers gained a new line type "r" for configuring
1788 which UID/GID ranges to allocate system users/groups
1789 from. Lines of type "u" may now add an additional column
1790 that specifies the home directory for the system user to be
1791 created. Also, systemd-sysusers may now optionally read user
1792 information from STDIN instead of a file. This is useful for
1793 invoking it from RPM preinst scriptlets that need to create
1794 users before the first RPM file is installed since these
1795 files might need to be owned by them. A new
1796 %sysusers_create_inline RPM macro has been introduced to do
1797 just that. systemd-sysusers now updates the shadow files as
1798 well as the user/group databases, which should enhance
1799 compatibility with certain tools like grpck.
1801 * A number of bus APIs of PID 1 now optionally consult
1802 PolicyKit to permit access for otherwise unprivileged
1803 clients under certain conditions. Note that this currently
1804 doesn't support interactive authentication yet, but this is
1805 expected to be added eventually, too.
1807 * /etc/machine-info now has new fields for configuring the
1808 deployment environment of the machine, as well as the
1809 location of the machine. hostnamectl has been updated with
1810 new command to update these fields.
1812 * systemd-timesyncd has been updated to automatically acquire
1813 NTP server information from systemd-networkd, which might
1814 have been discovered via DHCP.
1816 * systemd-resolved now includes a caching DNS stub resolver
1817 and a complete LLMNR name resolution implementation. A new
1818 NSS module "nss-resolve" has been added which can be used
1819 instead of glibc's own "nss-dns" to resolve hostnames via
1820 systemd-resolved. Hostnames, addresses and arbitrary RRs may
1821 be resolved via systemd-resolved D-Bus APIs. In contrast to
1822 the glibc internal resolver systemd-resolved is aware of
1823 multi-homed system, and keeps DNS server and caches separate
1824 and per-interface. Queries are sent simultaneously on all
1825 interfaces that have DNS servers configured, in order to
1826 properly handle VPNs and local LANs which might resolve
1827 separate sets of domain names. systemd-resolved may acquire
1828 DNS server information from systemd-networkd automatically,
1829 which in turn might have discovered them via DHCP. A tool
1830 "systemd-resolve-host" has been added that may be used to
1831 query the DNS logic in resolved. systemd-resolved implements
1832 IDNA and automatically uses IDNA or UTF-8 encoding depending
1833 on whether classic DNS or LLMNR is used as transport. In the
1834 next releases we intend to add a DNSSEC and mDNS/DNS-SD
1835 implementation to systemd-resolved.
1837 * A new NSS module nss-mymachines has been added, that
1838 automatically resolves the names of all local registered
1839 containers to their respective IP addresses.
1841 * A new client tool "networkctl" for systemd-networkd has been
1842 added. It currently is entirely passive and will query
1843 networking configuration from udev, rtnetlink and networkd,
1844 and present it to the user in a very friendly
1845 way. Eventually, we hope to extend it to become a full
1846 control utility for networkd.
1848 * .socket units gained a new DeferAcceptSec= setting that
1849 controls the kernels' TCP_DEFER_ACCEPT sockopt for
1850 TCP. Similarly, support for controlling TCP keep-alive
1851 settings has been added (KeepAliveTimeSec=,
1852 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
1853 turning off Nagle's algorithm on TCP has been added
1856 * logind learned a new session type "web", for use in projects
1857 like Cockpit which register web clients as PAM sessions.
1859 * timer units with at least one OnCalendar= setting will now
1860 be started only after timer-sync.target has been
1861 reached. This way they will not elapse before the system
1862 clock has been corrected by a local NTP client or
1863 similar. This is particular useful on RTC-less embedded
1864 machines, that come up with an invalid system clock.
1866 * systemd-nspawn's --network-veth= switch should now result in
1867 stable MAC addresses for both the outer and the inner side
1870 * systemd-nspawn gained a new --volatile= switch for running
1871 container instances with /etc or /var unpopulated.
1873 * The kdbus client code has been updated to use the new Linux
1874 3.17 memfd subsystem instead of the old kdbus-specific one.
1876 * systemd-networkd's DHCP client and server now support
1877 FORCERENEW. There are also new configuration options to
1878 configure the vendor client identifier and broadcast mode
1881 * systemd will no longer inform the kernel about the current
1882 timezone, as this is necessarily incorrect and racy as the
1883 kernel has no understanding of DST and similar
1884 concepts. This hence means FAT timestamps will be always
1885 considered UTC, similar to what Android is already
1886 doing. Also, when the RTC is configured to the local time
1887 (rather than UTC) systemd will never synchronize back to it,
1888 as this might confuse Windows at a later boot.
1890 * systemd-analyze gained a new command "verify" for offline
1891 validation of unit files.
1893 * systemd-networkd gained support for a couple of additional
1894 settings for bonding networking setups. Also, the metric for
1895 statically configured routes may now be configured. For
1896 network interfaces where this is appropriate the peer IP
1897 address may now be configured.
1899 * systemd-networkd's DHCP client will no longer request
1900 broadcasting by default, as this tripped up some networks.
1901 For hardware where broadcast is required the feature should
1902 be switched back on using RequestBroadcast=yes.
1904 * systemd-networkd will now set up IPv4LL addresses (when
1905 enabled) even if DHCP is configured successfully.
1907 * udev will now default to respect network device names given
1908 by the kernel when the kernel indicates that these are
1909 predictable. This behavior can be tweaked by changing
1910 NamePolicy= in the relevant .link file.
1912 * A new library systemd-terminal has been added that
1913 implements full TTY stream parsing and rendering. This
1914 library is supposed to be used later on for implementing a
1915 full userspace VT subsystem, replacing the current kernel
1918 * A new tool systemd-journal-upload has been added to push
1919 journal data to a remote system running
1920 systemd-journal-remote.
1922 * journald will no longer forward all local data to another
1923 running syslog daemon. This change has been made because
1924 rsyslog (which appears to be the most commonly used syslog
1925 implementation these days) no longer makes use of this, and
1926 instead pulls the data out of the journal on its own. Since
1927 forwarding the messages to a non-existent syslog server is
1928 more expensive than we assumed we have now turned this
1929 off. If you run a syslog server that is not a recent rsyslog
1930 version, you have to turn this option on again
1931 (ForwardToSyslog= in journald.conf).
1933 * journald now optionally supports the LZ4 compressor for
1934 larger journal fields. This compressor should perform much
1935 better than XZ which was the previous default.
1937 * machinectl now shows the IP addresses of local containers,
1938 if it knows them, plus the interface name of the container.
1940 * A new tool "systemd-escape" has been added that makes it
1941 easy to escape strings to build unit names and similar.
1943 * sd_notify() messages may now include a new ERRNO= field
1944 which is parsed and collected by systemd and shown among the
1945 "systemctl status" output for a service.
1947 * A new component "systemd-firstboot" has been added that
1948 queries the most basic systemd information (timezone,
1949 hostname, root password) interactively on first
1950 boot. Alternatively it may also be used to provision these
1951 things offline on OS images installed into directories.
1953 * The default sysctl.d/ snippets will now set
1955 net.ipv4.conf.default.promote_secondaries=1
1957 This has the benefit of no flushing secondary IP addresses
1958 when primary addresses are removed.
1960 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
1961 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
1962 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
1963 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
1964 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
1965 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
1966 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
1967 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
1968 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
1969 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
1970 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
1971 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
1972 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
1973 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
1974 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
1976 -- Berlin, 2014-08-19
1980 * A new tool systemd-sysusers has been added. This tool
1981 creates system users and groups in /etc/passwd and
1982 /etc/group, based on static declarative system user/group
1983 definitions in /usr/lib/sysusers.d/. This is useful to
1984 enable factory resets and volatile systems that boot up with
1985 an empty /etc directory, and thus need system users and
1986 groups created during early boot. systemd now also ships
1987 with two default sysusers.d/ files for the most basic
1988 users and groups systemd and the core operating system
1991 * A new tmpfiles snippet has been added that rebuilds the
1992 essential files in /etc on boot, should they be missing.
1994 * A directive for ensuring automatic clean-up of
1995 /var/cache/man/ has been removed from the default
1996 configuration. This line should now be shipped by the man
1997 implementation. The necessary change has been made to the
1998 man-db implementation. Note that you need to update your man
1999 implementation to one that ships this line, otherwise no
2000 automatic clean-up of /var/cache/man will take place.
2002 * A new condition ConditionNeedsUpdate= has been added that
2003 may conditionalize services to only run when /etc or /var
2004 are "older" than the vendor operating system resources in
2005 /usr. This is useful for reconstructing or updating /etc
2006 after an offline update of /usr or a factory reset, on the
2007 next reboot. Services that want to run once after such an
2008 update or reset should use this condition and order
2009 themselves before the new systemd-update-done.service, which
2010 will mark the two directories as fully updated. A number of
2011 service files have been added making use of this, to rebuild
2012 the udev hardware database, the journald message catalog and
2013 dynamic loader cache (ldconfig). The systemd-sysusers tool
2014 described above also makes use of this now. With this in
2015 place it is now possible to start up a minimal operating
2016 system with /etc empty cleanly. For more information on the
2017 concepts involved see this recent blog story:
2019 http://0pointer.de/blog/projects/stateless.html
2021 * A new system group "input" has been introduced, and all
2022 input device nodes get this group assigned. This is useful
2023 for system-level software to get access to input devices. It
2024 complements what is already done for "audio" and "video".
2026 * systemd-networkd learnt minimal DHCPv4 server support in
2027 addition to the existing DHCPv4 client support. It also
2028 learnt DHCPv6 client and IPv6 Router Solicitation client
2029 support. The DHCPv4 client gained support for static routes
2030 passed in from the server. Note that the [DHCPv4] section
2031 known in older systemd-networkd versions has been renamed to
2032 [DHCP] and is now also used by the DHCPv6 client. Existing
2033 .network files using settings of this section should be
2034 updated, though compatibility is maintained. Optionally, the
2035 client hostname may now be sent to the DHCP server.
2037 * networkd gained support for vxlan virtual networks as well
2038 as tun/tap and dummy devices.
2040 * networkd gained support for automatic allocation of address
2041 ranges for interfaces from a system-wide pool of
2042 addresses. This is useful for dynamically managing a large
2043 number of interfaces with a single network configuration
2044 file. In particular this is useful to easily assign
2045 appropriate IP addresses to the veth links of a large number
2046 of nspawn instances.
2048 * RPM macros for processing sysusers, sysctl and binfmt
2049 drop-in snippets at package installation time have been
2052 * The /etc/os-release file should now be placed in
2053 /usr/lib/os-release. The old location is automatically
2054 created as symlink. /usr/lib is the more appropriate
2055 location of this file, since it shall actually describe the
2056 vendor operating system shipped in /usr, and not the
2057 configuration stored in /etc.
2059 * .mount units gained a new boolean SloppyOptions= setting
2060 that maps to mount(8)'s -s option which enables permissive
2061 parsing of unknown mount options.
2063 * tmpfiles learnt a new "L+" directive which creates a symlink
2064 but (unlike "L") deletes a pre-existing file first, should
2065 it already exist and not already be the correct
2066 symlink. Similarly, "b+", "c+" and "p+" directives have been
2067 added as well, which create block and character devices, as
2068 well as fifos in the filesystem, possibly removing any
2069 pre-existing files of different types.
2071 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
2072 'argument' field (which so far specified the source to
2073 symlink/copy the files from) is now optional. If omitted the
2074 same file os copied from /usr/share/factory/ suffixed by the
2075 full destination path. This is useful for populating /etc
2076 with essential files, by copying them from vendor defaults
2077 shipped in /usr/share/factory/etc.
2079 * A new command "systemctl preset-all" has been added that
2080 applies the service preset settings to all installed unit
2081 files. A new switch --preset-mode= has been added that
2082 controls whether only enable or only disable operations
2085 * A new command "systemctl is-system-running" has been added
2086 that allows checking the overall state of the system, for
2087 example whether it is fully up and running.
2089 * When the system boots up with an empty /etc, the equivalent
2090 to "systemctl preset-all" is executed during early boot, to
2091 make sure all default services are enabled after a factory
2094 * systemd now contains a minimal preset file that enables the
2095 most basic services systemd ships by default.
2097 * Unit files' [Install] section gained a new DefaultInstance=
2098 field for defining the default instance to create if a
2099 template unit is enabled with no instance specified.
2101 * A new passive target cryptsetup-pre.target has been added
2102 that may be used by services that need to make they run and
2103 finish before the first LUKS cryptographic device is set up.
2105 * The /dev/loop-control and /dev/btrfs-control device nodes
2106 are now owned by the "disk" group by default, opening up
2107 access to this group.
2109 * systemd-coredump will now automatically generate a
2110 stack trace of all core dumps taking place on the system,
2111 based on elfutils' libdw library. This stack trace is logged
2114 * systemd-coredump may now optionally store coredumps directly
2115 on disk (in /var/lib/systemd/coredump, possibly compressed),
2116 instead of storing them unconditionally in the journal. This
2117 mode is the new default. A new configuration file
2118 /etc/systemd/coredump.conf has been added to configure this
2119 and other parameters of systemd-coredump.
2121 * coredumpctl gained a new "info" verb to show details about a
2122 specific coredump. A new switch "-1" has also been added
2123 that makes sure to only show information about the most
2124 recent entry instead of all entries. Also, as the tool is
2125 generally useful now the "systemd-" prefix of the binary
2126 name has been removed. Distributions that want to maintain
2127 compatibility with the old name should add a symlink from
2128 the old name to the new name.
2130 * journald's SplitMode= now defaults to "uid". This makes sure
2131 that unprivileged users can access their own coredumps with
2132 coredumpctl without restrictions.
2134 * New kernel command line options "systemd.wants=" (for
2135 pulling an additional unit during boot), "systemd.mask="
2136 (for masking a specific unit for the boot), and
2137 "systemd.debug-shell" (for enabling the debug shell on tty9)
2138 have been added. This is implemented in the new generator
2139 "systemd-debug-generator".
2141 * systemd-nspawn will now by default filter a couple of
2142 syscalls for containers, among them those required for
2143 kernel module loading, direct x86 IO port access, swap
2144 management, and kexec. Most importantly though
2145 open_by_handle_at() is now prohibited for containers,
2146 closing a hole similar to a recently discussed vulnerability
2147 in docker regarding access to files on file hierarchies the
2148 container should normally not have access to. Note that, for
2149 nspawn, we generally make no security claims anyway (and
2150 this is explicitly documented in the man page), so this is
2151 just a fix for one of the most obvious problems.
2153 * A new man page file-hierarchy(7) has been added that
2154 contains a minimized, modernized version of the file system
2155 layout systemd expects, similar in style to the FHS
2156 specification or hier(5). A new tool systemd-path(1) has
2157 been added to query many of these paths for the local
2160 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
2161 longer done. Since the directory now has a per-user size
2162 limit, and is cleaned on logout this appears unnecessary,
2163 in particular since this now brings the lifecycle of this
2164 directory closer in line with how IPC objects are handled.
2166 * systemd.pc now exports a number of additional directories,
2167 including $libdir (which is useful to identify the library
2168 path for the primary architecture of the system), and a
2169 couple of drop-in directories.
2171 * udev's predictable network interface names now use the dev_port
2172 sysfs attribute, introduced in linux 3.15 instead of dev_id to
2173 distinguish between ports of the same PCI function. dev_id should
2174 only be used for ports using the same HW address, hence the need
2177 * machined has been updated to export the OS version of a
2178 container (read from /etc/os-release and
2179 /usr/lib/os-release) on the bus. This is now shown in
2180 "machinectl status" for a machine.
2182 * A new service setting RestartForceExitStatus= has been
2183 added. If configured to a set of exit signals or process
2184 return values, the service will be restarted when the main
2185 daemon process exits with any of them, regardless of the
2188 * systemctl's -H switch for connecting to remote systemd
2189 machines has been extended so that it may be used to
2190 directly connect to a specific container on the
2191 host. "systemctl -H root@foobar:waldi" will now connect as
2192 user "root" to host "foobar", and then proceed directly to
2193 the container named "waldi". Note that currently you have to
2194 authenticate as user "root" for this to work, as entering
2195 containers is a privileged operation.
2197 Contributions from: Andreas Henriksson, Benjamin Steinwender,
2198 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
2199 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
2200 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
2201 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
2202 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
2203 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
2204 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
2205 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
2206 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
2207 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
2208 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
2210 -- Berlin, 2014-07-03
2214 * As an experimental feature, udev now tries to lock the
2215 disk device node (flock(LOCK_SH|LOCK_NB)) while it
2216 executes events for the disk or any of its partitions.
2217 Applications like partitioning programs can lock the
2218 disk device node (flock(LOCK_EX)) and claim temporary
2219 device ownership that way; udev will entirely skip all event
2220 handling for this disk and its partitions. If the disk
2221 was opened for writing, the close will trigger a partition
2222 table rescan in udev's "watch" facility, and if needed
2223 synthesize "change" events for the disk and all its partitions.
2224 This is now unconditionally enabled, and if it turns out to
2225 cause major problems, we might turn it on only for specific
2226 devices, or might need to disable it entirely. Device Mapper
2227 devices are excluded from this logic.
2229 * We temporarily dropped the "-l" switch for fsck invocations,
2230 since they collide with the flock() logic above. util-linux
2231 upstream has been changed already to avoid this conflict,
2232 and we will readd "-l" as soon as util-linux with this
2233 change has been released.
2235 * The dependency on libattr has been removed. Since a long
2236 time, the extended attribute calls have moved to glibc, and
2237 libattr is thus unnecessary.
2239 * Virtualization detection works without priviliges now. This
2240 means the systemd-detect-virt binary no longer requires
2241 CAP_SYS_PTRACE file capabilities, and our daemons can run
2242 with fewer privileges.
2244 * systemd-networkd now runs under its own "systemd-network"
2245 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
2246 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
2247 loses the ability to write to files owned by root this way.
2249 * Similarly, systemd-resolved now runs under its own
2250 "systemd-resolve" user with no capabilities remaining.
2252 * Similarly, systemd-bus-proxyd now runs under its own
2253 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
2255 * systemd-networkd gained support for setting up "veth"
2256 virtual Ethernet devices for container connectivity, as well
2257 as GRE and VTI tunnels.
2259 * systemd-networkd will no longer automatically attempt to
2260 manually load kernel modules necessary for certain tunnel
2261 transports. Instead, it is assumed the kernel loads them
2262 automatically when required. This only works correctly on
2263 very new kernels. On older kernels, please consider adding
2264 the kernel modules to /etc/modules-load.d/ as a work-around.
2266 * The resolv.conf file systemd-resolved generates has been
2267 moved to /run/systemd/resolve/. If you have a symlink from
2268 /etc/resolv.conf, it might be necessary to correct it.
2270 * Two new service settings, ProtectHome= and ProtectSystem=,
2271 have been added. When enabled, they will make the user data
2272 (such as /home) inaccessible or read-only and the system
2273 (such as /usr) read-only, for specific services. This allows
2274 very light-weight per-service sandboxing to avoid
2275 modifications of user data or system files from
2276 services. These two new switches have been enabled for all
2277 of systemd's long-running services, where appropriate.
2279 * Socket units gained new SocketUser= and SocketGroup=
2280 settings to set the owner user and group of AF_UNIX sockets
2281 and FIFOs in the file system.
2283 * Socket units gained a new RemoveOnStop= setting. If enabled,
2284 all FIFOS and sockets in the file system will be removed
2285 when the specific socket unit is stopped.
2287 * Socket units gained a new Symlinks= setting. It takes a list
2288 of symlinks to create to file system sockets or FIFOs
2289 created by the specific Unix sockets. This is useful to
2290 manage symlinks to socket nodes with the same life-cycle as
2293 * The /dev/log socket and /dev/initctl FIFO have been moved to
2294 /run, and have been replaced by symlinks. This allows
2295 connecting to these facilities even if PrivateDevices=yes is
2296 used for a service (which makes /dev/log itself unavailable,
2297 but /run is left). This also has the benefit of ensuring
2298 that /dev only contains device nodes, directories and
2299 symlinks, and nothing else.
2301 * sd-daemon gained two new calls sd_pid_notify() and
2302 sd_pid_notifyf(). They are similar to sd_notify() and
2303 sd_notifyf(), but allow overriding of the source PID of
2304 notification messages if permissions permit this. This is
2305 useful to send notify messages on behalf of a different
2306 process (for example, the parent process). The
2307 systemd-notify tool has been updated to make use of this
2308 when sending messages (so that notification messages now
2309 originate from the shell script invoking systemd-notify and
2310 not the systemd-notify process itself. This should minimize
2311 a race where systemd fails to associate notification
2312 messages to services when the originating process already
2315 * A new "on-abnormal" setting for Restart= has been added. If
2316 set, it will result in automatic restarts on all "abnormal"
2317 reasons for a process to exit, which includes unclean
2318 signals, core dumps, timeouts and watchdog timeouts, but
2319 does not include clean and unclean exit codes or clean
2320 signals. Restart=on-abnormal is an alternative for
2321 Restart=on-failure for services that shall be able to
2322 terminate and avoid restarts on certain errors, by
2323 indicating so with an unclean exit code. Restart=on-failure
2324 or Restart=on-abnormal is now the recommended setting for
2325 all long-running services.
2327 * If the InaccessibleDirectories= service setting points to a
2328 mount point (or if there are any submounts contained within
2329 it), it is now attempted to completely unmount it, to make
2330 the file systems truly unavailable for the respective
2333 * The ReadOnlyDirectories= service setting and
2334 systemd-nspawn's --read-only parameter are now recursively
2335 applied to all submounts, too.
2337 * Mount units may now be created transiently via the bus APIs.
2339 * The support for SysV and LSB init scripts has been removed
2340 from the systemd daemon itself. Instead, it is now
2341 implemented as a generator that creates native systemd units
2342 from these scripts when needed. This enables us to remove a
2343 substantial amount of legacy code from PID 1, following the
2344 fact that many distributions only ship a very small number
2345 of LSB/SysV init scripts nowadays.
2347 * Privileged Xen (dom0) domains are not considered
2348 virtualization anymore by the virtualization detection
2349 logic. After all, they generally have unrestricted access to
2350 the hardware and usually are used to manage the unprivileged
2353 * systemd-tmpfiles gained a new "C" line type, for copying
2354 files or entire directories.
2356 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
2357 lines. So far, they have been non-globbing versions of the
2358 latter, and have thus been redundant. In future, it is
2359 recommended to only use "z". "m" has hence been removed
2360 from the documentation, even though it stays supported.
2362 * A tmpfiles snippet to recreate the most basic structure in
2363 /var has been added. This is enough to create the /var/run →
2364 /run symlink and create a couple of structural
2365 directories. This allows systems to boot up with an empty or
2366 volatile /var. Of course, while with this change, the core OS
2367 now is capable with dealing with a volatile /var, not all
2368 user services are ready for it. However, we hope that sooner
2369 or later, many service daemons will be changed upstream so
2370 that they are able to automatically create their necessary
2371 directories in /var at boot, should they be missing. This is
2372 the first step to allow state-less systems that only require
2373 the vendor image for /usr to boot.
2375 * systemd-nspawn has gained a new --tmpfs= switch to mount an
2376 empty tmpfs instance to a specific directory. This is
2377 particularly useful for making use of the automatic
2378 reconstruction of /var (see above), by passing --tmpfs=/var.
2380 * Access modes specified in tmpfiles snippets may now be
2381 prefixed with "~", which indicates that they shall be masked
2382 by whether the existing file or directory is currently
2383 writable, readable or executable at all. Also, if specified,
2384 the sgid/suid/sticky bits will be masked for all
2387 * A new passive target unit "network-pre.target" has been
2388 added which is useful for services that shall run before any
2389 network is configured, for example firewall scripts.
2391 * The "floppy" group that previously owned the /dev/fd*
2392 devices is no longer used. The "disk" group is now used
2393 instead. Distributions should probably deprecate usage of
2396 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
2397 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
2398 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
2399 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
2400 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
2401 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
2404 -- Berlin, 2014-06-11
2408 * A new "systemd-timesyncd" daemon has been added for
2409 synchronizing the system clock across the network. It
2410 implements an SNTP client. In contrast to NTP
2411 implementations such as chrony or the NTP reference server,
2412 this only implements a client side, and does not bother with
2413 the full NTP complexity, focusing only on querying time from
2414 one remote server and synchronizing the local clock to
2415 it. Unless you intend to serve NTP to networked clients or
2416 want to connect to local hardware clocks, this simple NTP
2417 client should be more than appropriate for most
2418 installations. The daemon runs with minimal privileges, and
2419 has been hooked up with networkd to only operate when
2420 network connectivity is available. The daemon saves the
2421 current clock to disk every time a new NTP sync has been
2422 acquired, and uses this to possibly correct the system clock
2423 early at bootup, in order to accommodate for systems that
2424 lack an RTC such as the Raspberry Pi and embedded devices,
2425 and to make sure that time monotonically progresses on these
2426 systems, even if it is not always correct. To make use of
2427 this daemon, a new system user and group "systemd-timesync"
2428 needs to be created on installation of systemd.
2430 * The queue "seqnum" interface of libudev has been disabled, as
2431 it was generally incompatible with device namespacing as
2432 sequence numbers of devices go "missing" if the devices are
2433 part of a different namespace.
2435 * "systemctl list-timers" and "systemctl list-sockets" gained
2436 a --recursive switch for showing units of these types also
2437 for all local containers, similar in style to the already
2438 supported --recursive switch for "systemctl list-units".
2440 * A new RebootArgument= setting has been added for service
2441 units, which may be used to specify a kernel reboot argument
2442 to use when triggering reboots with StartLimitAction=.
2444 * A new FailureAction= setting has been added for service
2445 units which may be used to specify an operation to trigger
2446 when a service fails. This works similarly to
2447 StartLimitAction=, but unlike it, controls what is done
2448 immediately rather than only after several attempts to
2449 restart the service in question.
2451 * hostnamed got updated to also expose the kernel name,
2452 release, and version on the bus. This is useful for
2453 executing commands like hostnamectl with the -H switch.
2454 systemd-analyze makes use of this to properly display
2455 details when running non-locally.
2457 * The bootchart tool can now show cgroup information in the
2458 graphs it generates.
2460 * The CFS CPU quota cgroup attribute is now exposed for
2461 services. The new CPUQuota= switch has been added for this
2462 which takes a percentage value. Setting this will have the
2463 result that a service may never get more CPU time than the
2464 specified percentage, even if the machine is otherwise idle.
2466 * systemd-networkd learned IPIP and SIT tunnel support.
2468 * LSB init scripts exposing a dependency on $network will now
2469 get a dependency on network-online.target rather than simply
2470 network.target. This should bring LSB handling closer to
2471 what it was on SysV systems.
2473 * A new fsck.repair= kernel option has been added to control
2474 how fsck shall deal with unclean file systems at boot.
2476 * The (.ini) configuration file parser will now silently
2477 ignore sections whose name begins with "X-". This may be
2478 used to maintain application-specific extension sections in unit
2481 * machined gained a new API to query the IP addresses of
2482 registered containers. "machinectl status" has been updated
2483 to show these addresses in its output.
2485 * A new call sd_uid_get_display() has been added to the
2486 sd-login APIs for querying the "primary" session of a
2487 user. The "primary" session of the user is elected from the
2488 user's sessions and generally a graphical session is
2489 preferred over a text one.
2491 * A minimal systemd-resolved daemon has been added. It
2492 currently simply acts as a companion to systemd-networkd and
2493 manages resolv.conf based on per-interface DNS
2494 configuration, possibly supplied via DHCP. In the long run
2495 we hope to extend this into a local DNSSEC enabled DNS and
2498 * The systemd-networkd-wait-online tool is now enabled by
2499 default. It will delay network-online.target until a network
2500 connection has been configured. The tool primarily integrates
2501 with networkd, but will also make a best effort to make sense
2502 of network configuration performed in some other way.
2504 * Two new service options StartupCPUShares= and
2505 StartupBlockIOWeight= have been added that work similarly to
2506 CPUShares= and BlockIOWeight= however only apply during
2507 system startup. This is useful to prioritize certain services
2508 differently during bootup than during normal runtime.
2510 * hostnamed has been changed to prefer the statically
2511 configured hostname in /etc/hostname (unless set to
2512 'localhost' or empty) over any dynamic one supplied by
2513 dhcp. With this change, the rules for picking the hostname
2514 match more closely the rules of other configuration settings
2515 where the local administrator's configuration in /etc always
2516 overrides any other settings.
2518 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
2519 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
2520 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
2521 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
2522 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
2523 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
2524 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
2525 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
2526 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
2527 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
2528 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
2529 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
2530 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
2531 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
2532 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
2533 Lindskog, WaLyong Cho, Will Woods, Zbigniew
2536 -- Beijing, 2014-05-28
2540 * When restoring the screen brightness at boot, stay away from
2541 the darkest setting or from the lowest 5% of the available
2542 range, depending on which is the larger value of both. This
2543 should effectively protect the user from rebooting into a
2544 black screen, should the brightness have been set to minimum
2547 * sd-login gained a new sd_machine_get_class() call to
2548 determine the class ("vm" or "container") of a machine
2549 registered with machined.
2551 * sd-login gained new calls
2552 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
2553 to query the identity of the peer of a local AF_UNIX
2554 connection. They operate similarly to their sd_pid_get_xyz()
2557 * PID 1 will now maintain a system-wide system state engine
2558 with the states "starting", "running", "degraded",
2559 "maintenance", "stopping". These states are bound to system
2560 startup, normal runtime, runtime with at least one failed
2561 service, rescue/emergency mode and system shutdown. This
2562 state is shown in the "systemctl status" output when no unit
2563 name is passed. It is useful to determine system state, in
2564 particularly when doing so for many systems or containers at
2567 * A new command "list-machines" has been added to "systemctl"
2568 that lists all local OS containers and shows their system
2569 state (see above), if systemd runs inside of them.
2571 * systemctl gained a new "-r" switch to recursively enumerate
2572 units on all local containers, when used with the
2573 "list-unit" command (which is the default one that is
2574 executed when no parameters are specified).
2576 * The GPT automatic partition discovery logic will now honour
2577 two GPT partition flags: one may be set on a partition to
2578 cause it to be mounted read-only, and the other may be set
2579 on a partition to ignore it during automatic discovery.
2581 * Two new GPT type UUIDs have been added for automatic root
2582 partition discovery, for 32-bit and 64-bit ARM. This is not
2583 particularly useful for discovering the root directory on
2584 these architectures during bare-metal boots (since UEFI is
2585 not common there), but still very useful to allow booting of
2586 ARM disk images in nspawn with the -i option.
2588 * MAC addresses of interfaces created with nspawn's
2589 --network-interface= switch will now be generated from the
2590 machine name, and thus be stable between multiple invocations
2593 * logind will now automatically remove all IPC objects owned
2594 by a user if she or he fully logs out. This makes sure that
2595 users who are logged out cannot continue to consume IPC
2596 resources. This covers SysV memory, semaphores and message
2597 queues as well as POSIX shared memory and message
2598 queues. Traditionally, SysV and POSIX IPC had no life-cycle
2599 limits. With this functionality, that is corrected. This may
2600 be turned off by using the RemoveIPC= switch of logind.conf.
2602 * The systemd-machine-id-setup and tmpfiles tools gained a
2603 --root= switch to operate on a specific root directory,
2606 * journald can now forward logged messages to the TTYs of all
2607 logged in users ("wall"). This is the default for all
2608 emergency messages now.
2610 * A new tool systemd-journal-remote has been added to stream
2611 journal log messages across the network.
2613 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
2614 controller trees are mounted into it. Note that the
2615 directories mounted beneath it are not read-only. This is a
2616 security measure and is particularly useful because glibc
2617 actually includes a search logic to pick any tmpfs it can
2618 find to implement shm_open() if /dev/shm is not available
2619 (which it might very well be in namespaced setups).
2621 * machinectl gained a new "poweroff" command to cleanly power
2622 down a local OS container.
2624 * The PrivateDevices= unit file setting will now also drop the
2625 CAP_MKNOD capability from the capability bound set, and
2626 imply DevicePolicy=closed.
2628 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
2629 comprehensively on all long-running systemd services where
2630 this is appropriate.
2632 * systemd-udevd will now run in a disassociated mount
2633 namespace. To mount directories from udev rules, make sure to
2634 pull in mount units via SYSTEMD_WANTS properties.
2636 * The kdbus support gained support for uploading policy into
2637 the kernel. sd-bus gained support for creating "monitoring"
2638 connections that can eavesdrop into all bus communication
2639 for debugging purposes.
2641 * Timestamps may now be specified in seconds since the UNIX
2642 epoch Jan 1st, 1970 by specifying "@" followed by the value
2645 * Native tcpwrap support in systemd has been removed. tcpwrap
2646 is old code, not really maintained anymore and has serious
2647 shortcomings, and better options such as firewalls
2648 exist. For setups that require tcpwrap usage, please
2649 consider invoking your socket-activated service via tcpd,
2650 like on traditional inetd.
2652 * A new system.conf configuration option
2653 DefaultTimerAccuracySec= has been added that controls the
2654 default AccuracySec= setting of .timer units.
2656 * Timer units gained a new WakeSystem= switch. If enabled,
2657 timers configured this way will cause the system to resume
2658 from system suspend (if the system supports that, which most
2661 * Timer units gained a new Persistent= switch. If enabled,
2662 timers configured this way will save to disk when they have
2663 been last triggered. This information is then used on next
2664 reboot to possible execute overdue timer events, that
2665 could not take place because the system was powered off.
2666 This enables simple anacron-like behaviour for timer units.
2668 * systemctl's "list-timers" will now also list the time a
2669 timer unit was last triggered in addition to the next time
2670 it will be triggered.
2672 * systemd-networkd will now assign predictable IPv4LL
2673 addresses to its local interfaces.
2675 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
2676 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
2677 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
2678 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
2679 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
2680 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
2681 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
2682 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
2685 -- Berlin, 2014-03-25
2689 * A new unit file setting RestrictAddressFamilies= has been
2690 added to restrict which socket address families unit
2691 processes gain access to. This takes address family names
2692 like "AF_INET" or "AF_UNIX", and is useful to minimize the
2693 attack surface of services via exotic protocol stacks. This
2694 is built on seccomp system call filters.
2696 * Two new unit file settings RuntimeDirectory= and
2697 RuntimeDirectoryMode= have been added that may be used to
2698 manage a per-daemon runtime directories below /run. This is
2699 an alternative for setting up directory permissions with
2700 tmpfiles snippets, and has the advantage that the runtime
2701 directory's lifetime is bound to the daemon runtime and that
2702 the daemon starts up with an empty directory each time. This
2703 is particularly useful when writing services that drop
2704 privileges using the User= or Group= setting.
2706 * The DeviceAllow= unit setting now supports globbing for
2707 matching against device group names.
2709 * The systemd configuration file system.conf gained new
2710 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
2711 DefaultMemoryAccounting= to globally turn on/off accounting
2712 for specific resources (cgroups) for all units. These
2713 settings may still be overridden individually in each unit
2716 * systemd-gpt-auto-generator is now able to discover /srv and
2717 root partitions in addition to /home and swap partitions. It
2718 also supports LUKS-encrypted partitions now. With this in
2719 place, automatic discovery of partitions to mount following
2720 the Discoverable Partitions Specification
2721 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
2722 is now a lot more complete. This allows booting without
2723 /etc/fstab and without root= on the kernel command line on
2724 systems prepared appropriately.
2726 * systemd-nspawn gained a new --image= switch which allows
2727 booting up disk images and Linux installations on any block
2728 device that follow the Discoverable Partitions Specification
2729 (see above). This means that installations made with
2730 appropriately updated installers may now be started and
2731 deployed using container managers, completely
2732 unmodified. (We hope that libvirt-lxc will add support for
2733 this feature soon, too.)
2735 * systemd-nspawn gained a new --network-macvlan= setting to
2736 set up a private macvlan interface for the
2737 container. Similarly, systemd-networkd gained a new
2738 Kind=macvlan setting in .netdev files.
2740 * systemd-networkd now supports configuring local addresses
2743 * A new tool systemd-network-wait-online has been added to
2744 synchronously wait for network connectivity using
2747 * The sd-bus.h bus API gained a new sd_bus_track object for
2748 tracking the life-cycle of bus peers. Note that sd-bus.h is
2749 still not a public API though (unless you specify
2750 --enable-kdbus on the configure command line, which however
2751 voids your warranty and you get no API stability guarantee).
2753 * The $XDG_RUNTIME_DIR runtime directories for each user are
2754 now individual tmpfs instances, which has the benefit of
2755 introducing separate pools for each user, with individual
2756 size limits, and thus making sure that unprivileged clients
2757 can no longer negatively impact the system or other users by
2758 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
2759 RuntimeDirectorySize= has been introduced that allows
2760 controlling the default size limit for all users. It
2761 defaults to 10% of the available physical memory. This is no
2762 replacement for quotas on tmpfs though (which the kernel
2763 still does not support), as /dev/shm and /tmp are still
2764 shared resources used by both the system and unprivileged
2767 * logind will now automatically turn off automatic suspending
2768 on laptop lid close when more than one display is
2769 connected. This was previously expected to be implemented
2770 individually in desktop environments (such as GNOME),
2771 however has been added to logind now, in order to fix a
2772 boot-time race where a desktop environment might not have
2773 been started yet and thus not been able to take an inhibitor
2774 lock at the time where logind already suspends the system
2775 due to a closed lid.
2777 * logind will now wait at least 30s after each system
2778 suspend/resume cycle, and 3min after system boot before
2779 suspending the system due to a closed laptop lid. This
2780 should give USB docking stations and similar enough time to
2781 be probed and configured after system resume and boot in
2782 order to then act as suspend blocker.
2784 * systemd-run gained a new --property= setting which allows
2785 initialization of resource control properties (and others)
2786 for the created scope or service unit. Example: "systemd-run
2787 --property=BlockIOWeight=10 updatedb" may be used to run
2788 updatedb at a low block IO scheduling weight.
2790 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
2791 now also work in --scope mode.
2793 * When systemd is compiled with kdbus support, basic support
2794 for enforced policies is now in place. (Note that enabling
2795 kdbus still voids your warranty and no API compatibility
2798 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
2799 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
2800 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
2801 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
2802 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
2803 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
2804 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
2805 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
2806 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
2807 Zbigniew Jędrzejewski-Szmek
2809 -- Berlin, 2014-03-12
2813 * systemd will now relabel /dev after loading the SMACK policy
2814 according to SMACK rules.
2816 * A new unit file option AppArmorProfile= has been added to
2817 set the AppArmor profile for the processes of a unit.
2819 * A new condition check ConditionArchitecture= has been added
2820 to conditionalize units based on the system architecture, as
2821 reported by uname()'s "machine" field.
2823 * systemd-networkd now supports matching on the system
2824 virtualization, architecture, kernel command line, host name
2827 * logind is now a lot more aggressive when suspending the
2828 machine due to a closed laptop lid. Instead of acting only
2829 on the lid close action, it will continuously watch the lid
2830 status and act on it. This is useful for laptops where the
2831 power button is on the outside of the chassis so that it can
2832 be reached without opening the lid (such as the Lenovo
2833 Yoga). On those machines, logind will now immediately
2834 re-suspend the machine if the power button has been
2835 accidentally pressed while the laptop was suspended and in a
2836 backpack or similar.
2838 * logind will now watch SW_DOCK switches and inhibit reaction
2839 to the lid switch if it is pressed. This means that logind
2840 will not suspend the machine anymore if the lid is closed
2841 and the system is docked, if the laptop supports SW_DOCK
2842 notifications via the input layer. Note that ACPI docking
2843 stations do not generate this currently. Also note that this
2844 logic is usually not fully sufficient and Desktop
2845 Environments should take a lid switch inhibitor lock when an
2846 external display is connected, as systemd will not watch
2849 * nspawn will now make use of the devices cgroup controller by
2850 default, and only permit creation of and access to the usual
2851 API device nodes like /dev/null or /dev/random, as well as
2852 access to (but not creation of) the pty devices.
2854 * We will now ship a default .network file for
2855 systemd-networkd that automatically configures DHCP for
2856 network interfaces created by nspawn's --network-veth or
2857 --network-bridge= switches.
2859 * systemd will now understand the usual M, K, G, T suffixes
2860 according to SI conventions (i.e. to the base 1000) when
2861 referring to throughput and hardware metrics. It will stay
2862 with IEC conventions (i.e. to the base 1024) for software
2863 metrics, according to what is customary according to
2864 Wikipedia. We explicitly document which base applies for
2865 each configuration option.
2867 * The DeviceAllow= setting in unit files now supports a syntax
2868 to whitelist an entire group of devices node majors at once,
2869 based on the /proc/devices listing. For example, with the
2870 string "char-pts", it is now possible to whitelist all
2871 current and future pseudo-TTYs at once.
2873 * sd-event learned a new "post" event source. Event sources of
2874 this type are triggered by the dispatching of any event
2875 source of a type that is not "post". This is useful for
2876 implementing clean-up and check event sources that are
2877 triggered by other work being done in the program.
2879 * systemd-networkd is no longer statically enabled, but uses
2880 the usual [Install] sections so that it can be
2881 enabled/disabled using systemctl. It still is enabled by
2884 * When creating a veth interface pair with systemd-nspawn, the
2885 host side will now be prefixed with "vb-" if
2886 --network-bridge= is used, and with "ve-" if --network-veth
2887 is used. This way, it is easy to distinguish these cases on
2888 the host, for example to apply different configuration to
2889 them with systemd-networkd.
2891 * The compatibility libraries for libsystemd-journal.so,
2892 libsystem-id128.so, libsystemd-login.so and
2893 libsystemd-daemon.so do not make use of IFUNC
2894 anymore. Instead, we now build libsystemd.so multiple times
2895 under these alternative names. This means that the footprint
2896 is drastically increased, but given that these are
2897 transitional compatibility libraries, this should not matter
2898 much. This change has been made necessary to support the ARM
2899 platform for these compatibility libraries, as the ARM
2900 toolchain is not really at the same level as the toolchain
2901 for other architectures like x86 and does not support
2902 IFUNC. Please make sure to use --enable-compat-libs only
2903 during a transitional period!
2905 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
2906 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
2907 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
2908 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
2909 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
2910 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
2911 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
2912 Zbigniew Jędrzejewski-Szmek
2914 -- Berlin, 2014-02-24
2918 * A new component "systemd-networkd" has been added that can
2919 be used to configure local network interfaces statically or
2920 via DHCP. It is capable of bringing up bridges, VLANs, and
2921 bonding. Currently, no hook-ups for interactive network
2922 configuration are provided. Use this for your initrd,
2923 container, embedded, or server setup if you need a simple,
2924 yet powerful, network configuration solution. This
2925 configuration subsystem is quite nifty, as it allows wildcard
2926 hotplug matching in interfaces. For example, with a single
2927 configuration snippet, you can configure that all Ethernet
2928 interfaces showing up are automatically added to a bridge,
2929 or similar. It supports link-sensing and more.
2931 * A new tool "systemd-socket-proxyd" has been added which can
2932 act as a bidirectional proxy for TCP sockets. This is
2933 useful for adding socket activation support to services that
2934 do not actually support socket activation, including virtual
2935 machines and the like.
2937 * Add a new tool to save/restore rfkill state on
2940 * Save/restore state of keyboard backlights in addition to
2941 display backlights on shutdown/boot.
2943 * udev learned a new SECLABEL{} construct to label device
2944 nodes with a specific security label when they appear. For
2945 now, only SECLABEL{selinux} is supported, but the syntax is
2946 prepared for additional security frameworks.
2948 * udev gained a new scheme to configure link-level attributes
2949 from files in /etc/systemd/network/*.link. These files can
2950 match against MAC address, device path, driver name and type,
2951 and will apply attributes like the naming policy, link speed,
2952 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
2953 address assignment policy (randomized, ...).
2955 * The configuration of network interface naming rules for
2956 "permanent interface names" has changed: a new NamePolicy=
2957 setting in the [Link] section of .link files determines the
2958 priority of possible naming schemes (onboard, slot, MAC,
2959 path). The default value of this setting is determined by
2960 /usr/lib/net/links/99-default.link. Old
2961 80-net-name-slot.rules udev configuration file has been
2962 removed, so local configuration overriding this file should
2963 be adapated to override 99-default.link instead.
2965 * When the User= switch is used in a unit file, also
2966 initialize $SHELL= based on the user database entry.
2968 * systemd no longer depends on libdbus. All communication is
2969 now done with sd-bus, systemd's low-level bus library
2972 * kdbus support has been added to PID 1 itself. When kdbus is
2973 enabled, this causes PID 1 to set up the system bus and
2974 enable support for a new ".busname" unit type that
2975 encapsulates bus name activation on kdbus. It works a little
2976 bit like ".socket" units, except for bus names. A new
2977 generator has been added that converts classic dbus1 service
2978 activation files automatically into native systemd .busname
2981 * sd-bus: add a light-weight vtable implementation that allows
2982 defining objects on the bus with a simple static const
2983 vtable array of its methods, signals and properties.
2985 * systemd will not generate or install static dbus
2986 introspection data anymore to /usr/share/dbus-1/interfaces,
2987 as the precise format of these files is unclear, and
2988 nothing makes use of it.
2990 * A proxy daemon is now provided to proxy clients connecting
2991 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
2992 compatibility with classic D-Bus.
2994 * A bus driver implementation has been added that supports the
2995 classic D-Bus bus driver calls on kdbus, also for
2996 compatibility purposes.
2998 * A new API "sd-event.h" has been added that implements a
2999 minimal event loop API built around epoll. It provides a
3000 couple of features that direct epoll usage is lacking:
3001 prioritization of events, scales to large numbers of timer
3002 events, per-event timer slack (accuracy), system-wide
3003 coalescing of timer events, exit handlers, watchdog
3004 supervision support using systemd's sd_notify() API, child
3007 * A new API "sd-rntl.h" has been added that provides an API
3008 around the route netlink interface of the kernel, similar in
3009 style to "sd-bus.h".
3011 * A new API "sd-dhcp-client.h" has been added that provides a
3012 small DHCPv4 client-side implementation. This is used by
3015 * There is a new kernel command line option
3016 "systemd.restore_state=0|1". When set to "0", none of the
3017 systemd tools will restore saved runtime state to hardware
3018 devices. More specifically, the rfkill and backlight states
3021 * The FsckPassNo= compatibility option in mount/service units
3022 has been removed. The fstab generator will now add the
3023 necessary dependencies automatically, and does not require
3024 PID1's support for that anymore.
3026 * journalctl gained a new switch, --list-boots, that lists
3027 recent boots with their times and boot IDs.
3029 * The various tools like systemctl, loginctl, timedatectl,
3030 busctl, systemd-run, ... have gained a new switch "-M" to
3031 connect to a specific, local OS container (as direct
3032 connection, without requiring SSH). This works on any
3033 container that is registered with machined, such as those
3034 created by libvirt-lxc or nspawn.
3036 * systemd-run and systemd-analyze also gained support for "-H"
3037 to connect to remote hosts via SSH. This is particularly
3038 useful for systemd-run because it enables queuing of jobs
3039 onto remote systems.
3041 * machinectl gained a new command "login" to open a getty
3042 login in any local container. This works with any container
3043 that is registered with machined (such as those created by
3044 libvirt-lxc or nspawn), and which runs systemd inside.
3046 * machinectl gained a new "reboot" command that may be used to
3047 trigger a reboot on a specific container that is registered
3048 with machined. This works on any container that runs an init
3049 system of some kind.
3051 * systemctl gained a new "list-timers" command to print a nice
3052 listing of installed timer units with the times they elapse
3055 * Alternative reboot() parameters may now be specified on the
3056 "systemctl reboot" command line and are passed to the
3057 reboot() system call.
3059 * systemctl gained a new --job-mode= switch to configure the
3060 mode to queue a job with. This is a more generic version of
3061 --fail, --irreversible, and --ignore-dependencies, which are
3062 still available but not advertised anymore.
3064 * /etc/systemd/system.conf gained new settings to configure
3065 various default timeouts of units, as well as the default
3066 start limit interval and burst. These may still be overridden
3069 * PID1 will now export on the bus profile data of the security
3070 policy upload process (such as the SELinux policy upload to
3073 * journald: when forwarding logs to the console, include
3074 timestamps (following the setting in
3075 /sys/module/printk/parameters/time).
3077 * OnCalendar= in timer units now understands the special
3078 strings "yearly" and "annually". (Both are equivalent)
3080 * The accuracy of timer units is now configurable with the new
3081 AccuracySec= setting. It defaults to 1min.
3083 * A new dependency type JoinsNamespaceOf= has been added that
3084 allows running two services within the same /tmp and network
3085 namespace, if PrivateNetwork= or PrivateTmp= are used.
3087 * A new command "cat" has been added to systemctl. It outputs
3088 the original unit file of a unit, and concatenates the
3089 contents of additional "drop-in" unit file snippets, so that
3090 the full configuration is shown.
3092 * systemctl now supports globbing on the various "list-xyz"
3093 commands, like "list-units" or "list-sockets", as well as on
3094 those commands which take multiple unit names.
3096 * journalctl's --unit= switch gained support for globbing.
3098 * All systemd daemons now make use of the watchdog logic so
3099 that systemd automatically notices when they hang.
3101 * If the $container_ttys environment variable is set,
3102 getty-generator will automatically spawn a getty for each
3103 listed tty. This is useful for container managers to request
3104 login gettys to be spawned on as many ttys as needed.
3106 * %h, %s, %U specifier support is not available anymore when
3107 used in unit files for PID 1. This is because NSS calls are
3108 not safe from PID 1. They stay available for --user
3109 instances of systemd, and as special case for the root user.
3111 * loginctl gained a new "--no-legend" switch to turn off output
3114 * The "sd-login.h" API gained three new calls:
3115 sd_session_is_remote(), sd_session_get_remote_user(),
3116 sd_session_get_remote_host() to query information about
3119 * The udev hardware database now also carries vendor/product
3120 information of SDIO devices.
3122 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
3123 determine whether watchdog notifications are requested by
3126 * Socket-activated per-connection services now include a
3127 short description of the connection parameters in the
3130 * tmpfiles gained a new "--boot" option. When this is not used,
3131 only lines where the command character is not suffixed with
3132 "!" are executed. When this option is specified, those
3133 options are executed too. This partitions tmpfiles
3134 directives into those that can be safely executed at any
3135 time, and those which should be run only at boot (for
3136 example, a line that creates /run/nologin).
3138 * A new API "sd-resolve.h" has been added which provides a simple
3139 asynchronous wrapper around glibc NSS host name resolution
3140 calls, such as getaddrinfo(). In contrast to glibc's
3141 getaddrinfo_a(), it does not use signals. In contrast to most
3142 other asynchronous name resolution libraries, this one does
3143 not reimplement DNS, but reuses NSS, so that alternate
3144 host name resolution systems continue to work, such as mDNS,
3145 LDAP, etc. This API is based on libasyncns, but it has been
3146 cleaned up for inclusion in systemd.
3148 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
3149 "sd-daemon.h" are no longer found in individual libraries
3150 libsystemd-journal.so, libsystemd-login.so,
3151 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
3152 merged them into a single library, libsystemd.so, which
3153 provides all symbols. The reason for this is cyclic
3154 dependencies, as these libraries tend to use each other's
3155 symbols. So far, we have managed to workaround that by linking
3156 a copy of a good part of our code into each of these
3157 libraries again and again, which, however, makes certain
3158 things hard to do, like sharing static variables. Also, it
3159 substantially increases footprint. With this change, there
3160 is only one library for the basic APIs systemd
3161 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
3162 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
3163 library as well, however are subject to the --enable-kdbus
3164 switch (see below). Note that "sd-dhcp-client.h" is not part
3165 of this library (this is because it only consumes, never
3166 provides, services of/to other APIs). To make the transition
3167 easy from the separate libraries to the unified one, we
3168 provide the --enable-compat-libs compile-time switch which
3169 will generate stub libraries that are compatible with the
3170 old ones but redirect all calls to the new one.
3172 * All of the kdbus logic and the new APIs "sd-bus.h",
3173 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
3174 and "sd-utf8.h" are compile-time optional via the
3175 "--enable-kdbus" switch, and they are not compiled in by
3176 default. To make use of kdbus, you have to explicitly enable
3177 the switch. Note however, that neither the kernel nor the
3178 userspace API for all of this is considered stable yet. We
3179 want to maintain the freedom to still change the APIs for
3180 now. By specifying this build-time switch, you acknowledge
3181 that you are aware of the instability of the current
3184 * Also, note that while kdbus is pretty much complete,
3185 it lacks one thing: proper policy support. This means you
3186 can build a fully working system with all features; however,
3187 it will be highly insecure. Policy support will be added in
3188 one of the next releases, at the same time that we will
3189 declare the APIs stable.
3191 * When the kernel command line argument "kdbus" is specified,
3192 systemd will automatically load the kdbus.ko kernel module. At
3193 this stage of development, it is only useful for testing kdbus
3194 and should not be used in production. Note: if "--enable-kdbus"
3195 is specified, and the kdbus.ko kernel module is available, and
3196 "kdbus" is added to the kernel command line, the entire system
3197 runs with kdbus instead of dbus-daemon, with the above mentioned
3198 problem of missing the system policy enforcement. Also a future
3199 version of kdbus.ko or a newer systemd will not be compatible with
3200 each other, and will unlikely be able to boot the machine if only
3201 one of them is updated.
3203 * systemctl gained a new "import-environment" command which
3204 uploads the caller's environment (or parts thereof) into the
3205 service manager so that it is inherited by services started
3206 by the manager. This is useful to upload variables like
3207 $DISPLAY into the user service manager.
3209 * A new PrivateDevices= switch has been added to service units
3210 which allows running a service with a namespaced /dev
3211 directory that does not contain any device nodes for
3212 physical devices. More specifically, it only includes devices
3213 such as /dev/null, /dev/urandom, and /dev/zero which are API
3216 * logind has been extended to support behaviour like VT
3217 switching on seats that do not support a VT. This makes
3218 multi-session available on seats that are not the first seat
3219 (seat0), and on systems where kernel support for VTs has
3220 been disabled at compile-time.
3222 * If a process holds a delay lock for system sleep or shutdown
3223 and fails to release it in time, we will now log its
3224 identity. This makes it easier to identify processes that
3225 cause slow suspends or power-offs.
3227 * When parsing /etc/crypttab, support for a new key-slot=
3228 option as supported by Debian is added. It allows indicating
3229 which LUKS slot to use on disk, speeding up key loading.
3231 * The sd_journald_sendv() API call has been checked and
3232 officially declared to be async-signal-safe so that it may
3233 be invoked from signal handlers for logging purposes.
3235 * Boot-time status output is now enabled automatically after a
3236 short timeout if boot does not progress, in order to give
3237 the user an indication what she or he is waiting for.
3239 * The boot-time output has been improved to show how much time
3240 remains until jobs expire.
3242 * The KillMode= switch in service units gained a new possible
3243 value "mixed". If set, and the unit is shut down, then the
3244 initial SIGTERM signal is sent only to the main daemon
3245 process, while the following SIGKILL signal is sent to
3246 all remaining processes of the service.
3248 * When a scope unit is registered, a new property "Controller"
3249 may be set. If set to a valid bus name, systemd will send a
3250 RequestStop() signal to this name when it would like to shut
3251 down the scope. This may be used to hook manager logic into
3252 the shutdown logic of scope units. Also, scope units may now
3253 be put in a special "abandoned" state, in which case the
3254 manager process which created them takes no further
3255 responsibilities for it.
3257 * When reading unit files, systemd will now verify
3258 the access mode of these files, and warn about certain
3259 suspicious combinations. This has been added to make it
3260 easier to track down packaging bugs where unit files are
3261 marked executable or world-writable.
3263 * systemd-nspawn gained a new "--setenv=" switch to set
3264 container-wide environment variables. The similar option in
3265 systemd-activate was renamed from "--environment=" to
3266 "--setenv=" for consistency.
3268 * systemd-nspawn has been updated to create a new kdbus domain
3269 for each container that is invoked, thus allowing each
3270 container to have its own set of system and user buses,
3271 independent of the host.
3273 * systemd-nspawn gained a new --drop-capability= switch to run
3274 the container with less capabilities than the default. Both
3275 --drop-capability= and --capability= now take the special
3276 string "all" for dropping or keeping all capabilities.
3278 * systemd-nspawn gained new switches for executing containers
3279 with specific SELinux labels set.
3281 * systemd-nspawn gained a new --quiet switch to not generate
3282 any additional output but the container's own console
3285 * systemd-nspawn gained a new --share-system switch to run a
3286 container without PID namespacing enabled.
3288 * systemd-nspawn gained a new --register= switch to control
3289 whether the container is registered with systemd-machined or
3290 not. This is useful for containers that do not run full
3291 OS images, but only specific apps.
3293 * systemd-nspawn gained a new --keep-unit which may be used
3294 when invoked as the only program from a service unit, and
3295 results in registration of the unit service itself in
3296 systemd-machined, instead of a newly opened scope unit.
3298 * systemd-nspawn gained a new --network-interface= switch for
3299 moving arbitrary interfaces to the container. The new
3300 --network-veth switch creates a virtual Ethernet connection
3301 between host and container. The new --network-bridge=
3302 switch then allows assigning the host side of this virtual
3303 Ethernet connection to a bridge device.
3305 * systemd-nspawn gained a new --personality= switch for
3306 setting the kernel personality for the container. This is
3307 useful when running a 32-bit container on a 64-bit host. A
3308 similar option Personality= is now also available for service
3311 * logind will now also track a "Desktop" identifier for each
3312 session which encodes the desktop environment of it. This is
3313 useful for desktop environments that want to identify
3314 multiple running sessions of itself easily.
3316 * A new SELinuxContext= setting for service units has been
3317 added that allows setting a specific SELinux execution
3318 context for a service.
3320 * Most systemd client tools will now honour $SYSTEMD_LESS for
3321 settings of the "less" pager. By default, these tools will
3322 override $LESS to allow certain operations to work, such as
3323 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
3324 influence this logic.
3326 * systemd's "seccomp" hook-up has been changed to make use of
3327 the libseccomp library instead of using its own
3328 implementation. This has benefits for portability among
3331 * For usage together with SystemCallFilter=, a new
3332 SystemCallErrorNumber= setting has been introduced that
3333 allows configuration of a system error number to be returned
3334 on filtered system calls, instead of immediately killing the
3335 process. Also, SystemCallArchitectures= has been added to
3336 limit access to system calls of a particular architecture
3337 (in order to turn off support for unused secondary
3338 architectures). There is also a global
3339 SystemCallArchitectures= setting in system.conf now to turn
3340 off support for non-native system calls system-wide.
3342 * systemd requires a kernel with a working name_to_handle_at(),
3343 please see the kernel config requirements in the README file.
3345 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
3346 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
3347 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
3348 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
3349 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
3350 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
3351 Elia Pinto, Florian Weimer, George McCollister, Goffredo
3352 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
3353 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
3354 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
3355 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
3356 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
3357 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
3358 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
3359 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
3360 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
3361 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
3362 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
3363 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
3364 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
3365 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
3366 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
3367 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
3368 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
3370 -- Berlin, 2014-02-20
3374 * logind has gained support for facilitating privileged input
3375 and drm device access for unprivileged clients. This work is
3376 useful to allow Wayland display servers (and similar
3377 programs, such as kmscon) to run under the user's ID and
3378 access input and drm devices which are normally
3379 protected. When this is used (and the kernel is new enough)
3380 logind will "mute" IO on the file descriptors passed to
3381 Wayland as long as it is in the background and "unmute" it
3382 if it returns into the foreground. This allows secure
3383 session switching without allowing background sessions to
3384 eavesdrop on input and display data. This also introduces
3385 session switching support if VT support is turned off in the
3386 kernel, and on seats that are not seat0.
3388 * A new kernel command line option luks.options= is understood
3389 now which allows specifying LUKS options for usage for LUKS
3390 encrypted partitions specified with luks.uuid=.
3392 * tmpfiles.d(5) snippets may now use specifier expansion in
3393 path names. More specifically %m, %b, %H, %v, are now
3394 replaced by the local machine id, boot id, hostname, and
3395 kernel version number.
3397 * A new tmpfiles.d(5) command "m" has been introduced which
3398 may be used to change the owner/group/access mode of a file
3399 or directory if it exists, but do nothing if it does not.
3401 * This release removes high-level support for the
3402 MemorySoftLimit= cgroup setting. The underlying kernel
3403 cgroup attribute memory.soft_limit= is currently badly
3404 designed and likely to be removed from the kernel API in its
3405 current form, hence we should not expose it for now.
3407 * The memory.use_hierarchy cgroup attribute is now enabled for
3408 all cgroups systemd creates in the memory cgroup
3409 hierarchy. This option is likely to be come the built-in
3410 default in the kernel anyway, and the non-hierarchical mode
3411 never made much sense in the intrinsically hierarchical
3414 * A new field _SYSTEMD_SLICE= is logged along with all journal
3415 messages containing the slice a message was generated
3416 from. This is useful to allow easy per-customer filtering of
3417 logs among other things.
3419 * systemd-journald will no longer adjust the group of journal
3420 files it creates to the "systemd-journal" group. Instead we
3421 rely on the journal directory to be owned by the
3422 "systemd-journal" group, and its setgid bit set, so that the
3423 kernel file system layer will automatically enforce that
3424 journal files inherit this group assignment. The reason for
3425 this change is that we cannot allow NSS look-ups from
3426 journald which would be necessary to resolve
3427 "systemd-journal" to a numeric GID, because this might
3428 create deadlocks if NSS involves synchronous queries to
3429 other daemons (such as nscd, or sssd) which in turn are
3430 logging clients of journald and might block on it, which
3431 would then dead lock. A tmpfiles.d(5) snippet included in
3432 systemd will make sure the setgid bit and group are
3433 properly set on the journal directory if it exists on every
3434 boot. However, we recommend adjusting it manually after
3435 upgrades too (or from RPM scriptlets), so that the change is
3436 not delayed until next reboot.
3438 * Backlight and random seed files in /var/lib/ have moved into
3439 the /var/lib/systemd/ directory, in order to centralize all
3440 systemd generated files in one directory.
3442 * Boot time performance measurements (as displayed by
3443 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
3444 performance information if that's available to determine how
3445 much time BIOS and boot loader initialization required. With
3446 a sufficiently new BIOS you hence no longer need to boot
3447 with Gummiboot to get access to such information.
3449 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
3450 Cristian Rodríguez, Dave Reisner, David Herrmann, David
3451 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
3452 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
3453 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
3454 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
3455 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3457 -- Berlin, 2013-10-02
3461 * The Restart= option for services now understands a new
3462 on-watchdog setting, which will restart the service
3463 automatically if the service stops sending out watchdog keep
3464 alive messages (as configured with WatchdogSec=).
3466 * The getty generator (which is responsible for bringing up a
3467 getty on configured serial consoles) will no longer only
3468 start a getty on the primary kernel console but on all
3469 others, too. This makes the order in which console= is
3470 specified on the kernel command line less important.
3472 * libsystemd-logind gained a new sd_session_get_vt() call to
3473 retrieve the VT number of a session.
3475 * If the option "tries=0" is set for an entry of /etc/crypttab
3476 its passphrase is queried indefinitely instead of any
3477 maximum number of tries.
3479 * If a service with a configure PID file terminates its PID
3480 file will now be removed automatically if it still exists
3481 afterwards. This should put an end to stale PID files.
3483 * systemd-run will now also take relative binary path names
3484 for execution and no longer insists on absolute paths.
3486 * InaccessibleDirectories= and ReadOnlyDirectories= now take
3487 paths that are optionally prefixed with "-" to indicate that
3488 it should not be considered a failure if they do not exist.
3490 * journalctl -o (and similar commands) now understands a new
3491 output mode "short-precise", it is similar to "short" but
3492 shows timestamps with usec accuracy.
3494 * The option "discard" (as known from Debian) is now
3495 synonymous to "allow-discards" in /etc/crypttab. In fact,
3496 "discard" is preferred now (since it is easier to remember
3499 * Some licensing clean-ups were made, so that more code is now
3500 LGPL-2.1 licensed than before.
3502 * A minimal tool to save/restore the display backlight
3503 brightness across reboots has been added. It will store the
3504 backlight setting as late as possible at shutdown, and
3505 restore it as early as possible during reboot.
3507 * A logic to automatically discover and enable home and swap
3508 partitions on GPT disks has been added. With this in place
3509 /etc/fstab becomes optional for many setups as systemd can
3510 discover certain partitions located on the root disk
3511 automatically. Home partitions are recognized under their
3512 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
3513 partitions are recognized under their GPT type ID
3514 0657fd6da4ab43c484e50933c84b4f4f.
3516 * systemd will no longer pass any environment from the kernel
3517 or initrd to system services. If you want to set an
3518 environment for all services, do so via the kernel command
3519 line systemd.setenv= assignment.
3521 * The systemd-sysctl tool no longer natively reads the file
3522 /etc/sysctl.conf. If desired, the file should be symlinked
3523 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
3524 legacy support by a symlink rather than built-in code, it
3525 also makes the otherwise hidden order of application of the
3526 different files visible. (Note that this partly reverts to a
3527 pre-198 application order of sysctl knobs!)
3529 * The "systemctl set-log-level" and "systemctl dump" commands
3530 have been moved to systemd-analyze.
3532 * systemd-run learned the new --remain-after-exit switch,
3533 which causes the scope unit not to be cleaned up
3534 automatically after the process terminated.
3536 * tmpfiles learned a new --exclude-prefix= switch to exclude
3537 certain paths from operation.
3539 * journald will now automatically flush all messages to disk
3540 as soon as a message at the log level CRIT, ALERT or EMERG
3543 Contributions from: Andrew Cook, Brandon Philips, Christian
3544 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
3545 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
3546 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
3547 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
3548 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
3549 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
3550 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
3551 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
3552 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
3553 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
3554 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
3555 William Giokas, Zbigniew Jędrzejewski-Szmek
3557 -- Berlin, 2013-09-13
3561 * The documentation has been updated to cover the various new
3562 concepts introduced with 205.
3564 * Unit files now understand the new %v specifier which
3565 resolves to the kernel version string as returned by "uname
3568 * systemctl now supports filtering the unit list output by
3569 load state, active state and sub state, using the new
3572 * "systemctl status" will now show the results of the
3573 condition checks (like ConditionPathExists= and similar) of
3574 the last start attempts of the unit. They are also logged to
3577 * "journalctl -b" may now be used to look for boot output of a
3578 specific boot. Try "journalctl -b -1" for the previous boot,
3579 but the syntax is substantially more powerful.
3581 * "journalctl --show-cursor" has been added which prints the
3582 cursor string the last shown log line. This may then be used
3583 with the new "journalctl --after-cursor=" switch to continue
3584 browsing logs from that point on.
3586 * "journalctl --force" may now be used to force regeneration
3589 * Creation of "dead" device nodes has been moved from udev
3590 into kmod and tmpfiles. Previously, udev would read the kmod
3591 databases to pre-generate dead device nodes based on meta
3592 information contained in kernel modules, so that these would
3593 be auto-loaded on access rather then at boot. As this
3594 does not really have much to do with the exposing actual
3595 kernel devices to userspace this has always been slightly
3596 alien in the udev codebase. Following the new scheme kmod
3597 will now generate a runtime snippet for tmpfiles from the
3598 module meta information and it now is tmpfiles' job to the
3599 create the nodes. This also allows overriding access and
3600 other parameters for the nodes using the usual tmpfiles
3601 facilities. As side effect this allows us to remove the
3602 CAP_SYS_MKNOD capability bit from udevd entirely.
3604 * logind's device ACLs may now be applied to these "dead"
3605 devices nodes too, thus finally allowing managed access to
3606 devices such as /dev/snd/sequencer whithout loading the
3607 backing module right-away.
3609 * A new RPM macro has been added that may be used to apply
3610 tmpfiles configuration during package installation.
3612 * systemd-detect-virt and ConditionVirtualization= now can
3613 detect User-Mode-Linux machines (UML).
3615 * journald will now implicitly log the effective capabilities
3616 set of processes in the message metadata.
3618 * systemd-cryptsetup has gained support for TrueCrypt volumes.
3620 * The initrd interface has been simplified (more specifically,
3621 support for passing performance data via environment
3622 variables and fsck results via files in /run has been
3623 removed). These features were non-essential, and are
3624 nowadays available in a much nicer way by having systemd in
3625 the initrd serialize its state and have the hosts systemd
3626 deserialize it again.
3628 * The udev "keymap" data files and tools to apply keyboard
3629 specific mappings of scan to key codes, and force-release
3630 scan code lists have been entirely replaced by a udev
3631 "keyboard" builtin and a hwdb data file.
3633 * systemd will now honour the kernel's "quiet" command line
3634 argument also during late shutdown, resulting in a
3635 completely silent shutdown when used.
3637 * There's now an option to control the SO_REUSEPORT socket
3638 option in .socket units.
3640 * Instance units will now automatically get a per-template
3641 subslice of system.slice unless something else is explicitly
3642 configured. For example, instances of sshd@.service will now
3643 implicitly be placed in system-sshd.slice rather than
3644 system.slice as before.
3646 * Test coverage support may now be enabled at build time.
3648 Contributions from: Dave Reisner, Frederic Crozat, Harald
3649 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
3650 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
3651 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
3652 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
3653 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
3654 Giokas, Zbigniew Jędrzejewski-Szmek
3656 -- Berlin, 2013-07-23
3660 * Two new unit types have been introduced:
3662 Scope units are very similar to service units, however, are
3663 created out of pre-existing processes -- instead of PID 1
3664 forking off the processes. By using scope units it is
3665 possible for system services and applications to group their
3666 own child processes (worker processes) in a powerful way
3667 which then maybe used to organize them, or kill them
3668 together, or apply resource limits on them.
3670 Slice units may be used to partition system resources in an
3671 hierarchical fashion and then assign other units to them. By
3672 default there are now three slices: system.slice (for all
3673 system services), user.slice (for all user sessions),
3674 machine.slice (for VMs and containers).
3676 Slices and scopes have been introduced primarily in
3677 context of the work to move cgroup handling to a
3678 single-writer scheme, where only PID 1
3679 creates/removes/manages cgroups.
3681 * There's a new concept of "transient" units. In contrast to
3682 normal units these units are created via an API at runtime,
3683 not from configuration from disk. More specifically this
3684 means it is now possible to run arbitrary programs as
3685 independent services, with all execution parameters passed
3686 in via bus APIs rather than read from disk. Transient units
3687 make systemd substantially more dynamic then it ever was,
3688 and useful as a general batch manager.
3690 * logind has been updated to make use of scope and slice units
3691 for managing user sessions. As a user logs in he will get
3692 his own private slice unit, to which all sessions are added
3693 as scope units. We also added support for automatically
3694 adding an instance of user@.service for the user into the
3695 slice. Effectively logind will no longer create cgroup
3696 hierarchies on its own now, it will defer entirely to PID 1
3697 for this by means of scope, service and slice units. Since
3698 user sessions this way become entities managed by PID 1
3699 the output of "systemctl" is now a lot more comprehensive.
3701 * A new mini-daemon "systemd-machined" has been added which
3702 may be used by virtualization managers to register local
3703 VMs/containers. nspawn has been updated accordingly, and
3704 libvirt will be updated shortly. machined will collect a bit
3705 of meta information about the VMs/containers, and assign
3706 them their own scope unit (see above). The collected
3707 meta-data is then made available via the "machinectl" tool,
3708 and exposed in "ps" and similar tools. machined/machinectl
3709 is compile-time optional.
3711 * As discussed earlier, the low-level cgroup configuration
3712 options ControlGroup=, ControlGroupModify=,
3713 ControlGroupPersistent=, ControlGroupAttribute= have been
3714 removed. Please use high-level attribute settings instead as
3715 well as slice units.
3717 * A new bus call SetUnitProperties() has been added to alter
3718 various runtime parameters of a unit. This is primarily
3719 useful to alter cgroup parameters dynamically in a nice way,
3720 but will be extended later on to make more properties
3721 modifiable at runtime. systemctl gained a new set-properties
3722 command that wraps this call.
3724 * A new tool "systemd-run" has been added which can be used to
3725 run arbitrary command lines as transient services or scopes,
3726 while configuring a number of settings via the command
3727 line. This tool is currently very basic, however already
3728 very useful. We plan to extend this tool to even allow
3729 queuing of execution jobs with time triggers from the
3730 command line, similar in fashion to "at".
3732 * nspawn will now inform the user explicitly that kernels with
3733 audit enabled break containers, and suggest the user to turn
3736 * Support for detecting the IMA and AppArmor security
3737 frameworks with ConditionSecurity= has been added.
3739 * journalctl gained a new "-k" switch for showing only kernel
3740 messages, mimicking dmesg output; in addition to "--user"
3741 and "--system" switches for showing only user's own logs
3744 * systemd-delta can now show information about drop-in
3745 snippets extending unit files.
3747 * libsystemd-bus has been substantially updated but is still
3748 not available as public API.
3750 * systemd will now look for the "debug" argument on the kernel
3751 command line and enable debug logging, similar to what
3752 "systemd.log_level=debug" already did before.
3754 * "systemctl set-default", "systemctl get-default" has been
3755 added to configure the default.target symlink, which
3756 controls what to boot into by default.
3758 * "systemctl set-log-level" has been added as a convenient
3759 way to raise and lower systemd logging threshold.
3761 * "systemd-analyze plot" will now show the time the various
3762 generators needed for execution, as well as information
3763 about the unit file loading.
3765 * libsystemd-journal gained a new sd_journal_open_files() call
3766 for opening specific journal files. journactl also gained a
3767 new switch to expose this new functionality. Previously we
3768 only supported opening all files from a directory, or all
3769 files from the system, as opening individual files only is
3770 racy due to journal file rotation.
3772 * systemd gained the new DefaultEnvironment= setting in
3773 /etc/systemd/system.conf to set environment variables for
3776 * If a privileged process logs a journal message with the
3777 OBJECT_PID= field set, then journald will automatically
3778 augment this with additional OBJECT_UID=, OBJECT_GID=,
3779 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
3780 system services want to log events about specific client
3781 processes. journactl/systemctl has been updated to make use
3782 of this information if all log messages regarding a specific
3785 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
3786 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
3787 Reisner, David Coppa, David King, David Strauss, Eelco
3788 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
3789 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
3790 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
3791 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
3792 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
3793 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
3794 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
3795 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
3796 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
3797 Łukasz Stelmach, 장동준
3801 * The Python bindings gained some minimal support for the APIs
3802 exposed by libsystemd-logind.
3804 * ConditionSecurity= gained support for detecting SMACK. Since
3805 this condition already supports SELinux and AppArmor we only
3806 miss IMA for this. Patches welcome!
3808 Contributions from: Karol Lewandowski, Lennart Poettering,
3809 Zbigniew Jędrzejewski-Szmek
3813 * systemd-nspawn will now create /etc/resolv.conf if
3814 necessary, before bind-mounting the host's file onto it.
3816 * systemd-nspawn will now store meta information about a
3817 container on the container's cgroup as extended attribute
3818 fields, including the root directory.
3820 * The cgroup hierarchy has been reworked in many ways. All
3821 objects any of the components systemd creates in the cgroup
3822 tree are now suffixed. More specifically, user sessions are
3823 now placed in cgroups suffixed with ".session", users in
3824 cgroups suffixed with ".user", and nspawn containers in
3825 cgroups suffixed with ".nspawn". Furthermore, all cgroup
3826 names are now escaped in a simple scheme to avoid collision
3827 of userspace object names with kernel filenames. This work
3828 is preparation for making these objects relocatable in the
3829 cgroup tree, in order to allow easy resource partitioning of
3830 these objects without causing naming conflicts.
3832 * systemctl list-dependencies gained the new switches
3833 --plain, --reverse, --after and --before.
3835 * systemd-inhibit now shows the process name of processes that
3836 have taken an inhibitor lock.
3838 * nss-myhostname will now also resolve "localhost"
3839 implicitly. This makes /etc/hosts an optional file and
3840 nicely handles that on IPv6 ::1 maps to both "localhost" and
3843 * libsystemd-logind.so gained a new call
3844 sd_get_machine_names() to enumerate running containers and
3845 VMs (currently only supported by very new libvirt and
3846 nspawn). sd_login_monitor can now be used to watch
3847 VMs/containers coming and going.
3849 * .include is not allowed recursively anymore, and only in
3850 unit files. Usually it is better to use drop-in snippets in
3851 .d/*.conf anyway, as introduced with systemd 198.
3853 * systemd-analyze gained a new "critical-chain" command that
3854 determines the slowest chain of units run during system
3855 boot-up. It is very useful for tracking down where
3856 optimizing boot time is the most beneficial.
3858 * systemd will no longer allow manipulating service paths in
3859 the name=systemd:/system cgroup tree using ControlGroup= in
3860 units. (But is still fine with it in all other dirs.)
3862 * There's a new systemd-nspawn@.service service file that may
3863 be used to easily run nspawn containers as system
3864 services. With the container's root directory in
3865 /var/lib/container/foobar it is now sufficient to run
3866 "systemctl start systemd-nspawn@foobar.service" to boot it.
3868 * systemd-cgls gained a new parameter "--machine" to list only
3869 the processes within a certain container.
3871 * ConditionSecurity= now can check for "apparmor". We still
3872 are lacking checks for SMACK and IMA for this condition
3873 check though. Patches welcome!
3875 * A new configuration file /etc/systemd/sleep.conf has been
3876 added that may be used to configure which kernel operation
3877 systemd is supposed to execute when "suspend", "hibernate"
3878 or "hybrid-sleep" is requested. This makes the new kernel
3879 "freeze" state accessible to the user.
3881 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
3882 the passed argument if applicable.
3884 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
3885 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
3886 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
3887 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
3888 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
3889 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
3890 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
3895 * The output of 'systemctl list-jobs' got some polishing. The
3896 '--type=' argument may now be passed more than once. A new
3897 command 'systemctl list-sockets' has been added which shows
3898 a list of kernel sockets systemd is listening on with the
3899 socket units they belong to, plus the units these socket
3902 * The experimental libsystemd-bus library got substantial
3903 updates to work in conjunction with the (also experimental)
3904 kdbus kernel project. It works well enough to exchange
3905 messages with some sophistication. Note that kdbus is not
3906 ready yet, and the library is mostly an elaborate test case
3907 for now, and not installable.
3909 * systemd gained a new unit 'systemd-static-nodes.service'
3910 that generates static device nodes earlier during boot, and
3911 can run in conjunction with udev.
3913 * libsystemd-login gained a new call sd_pid_get_user_unit()
3914 to retrieve the user systemd unit a process is running
3915 in. This is useful for systems where systemd is used as
3918 * systemd-nspawn now places all containers in the new /machine
3919 top-level cgroup directory in the name=systemd
3920 hierarchy. libvirt will soon do the same, so that we get a
3921 uniform separation of /system, /user and /machine for system
3922 services, user processes and containers/virtual
3923 machines. This new cgroup hierarchy is also useful to stick
3924 stable names to specific container instances, which can be
3925 recognized later this way (this name may be controlled
3926 via systemd-nspawn's new -M switch). libsystemd-login also
3927 gained a new call sd_pid_get_machine_name() to retrieve the
3928 name of the container/VM a specific process belongs to.
3930 * bootchart can now store its data in the journal.
3932 * libsystemd-journal gained a new call
3933 sd_journal_add_conjunction() for AND expressions to the
3934 matching logic. This can be used to express more complex
3935 logical expressions.
3937 * journactl can now take multiple --unit= and --user-unit=
3940 * The cryptsetup logic now understands the "luks.key=" kernel
3941 command line switch for specifying a file to read the
3942 decryption key from. Also, if a configured key file is not
3943 found the tool will now automatically fall back to prompting
3946 * Python systemd.journal module was updated to wrap recently
3947 added functions from libsystemd-journal. The interface was
3948 changed to bring the low level interface in s.j._Reader
3949 closer to the C API, and the high level interface in
3950 s.j.Reader was updated to wrap and convert all data about
3953 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
3954 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
3955 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
3956 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
3957 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
3958 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3962 * journalctl --update-catalog now understands a new --root=
3963 option to operate on catalogs found in a different root
3966 * During shutdown after systemd has terminated all running
3967 services a final killing loop kills all remaining left-over
3968 processes. We will now print the name of these processes
3969 when we send SIGKILL to them, since this usually indicates a
3972 * If /etc/crypttab refers to password files stored on
3973 configured mount points automatic dependencies will now be
3974 generated to ensure the specific mount is established first
3975 before the key file is attempted to be read.
3977 * 'systemctl status' will now show information about the
3978 network sockets a socket unit is listening on.
3980 * 'systemctl status' will also shown information about any
3981 drop-in configuration file for units. (Drop-In configuration
3982 files in this context are files such as
3983 /etc/systemd/systemd/foobar.service.d/*.conf)
3985 * systemd-cgtop now optionally shows summed up CPU times of
3986 cgroups. Press '%' while running cgtop to switch between
3987 percentage and absolute mode. This is useful to determine
3988 which cgroups use up the most CPU time over the entire
3989 runtime of the system. systemd-cgtop has also been updated
3990 to be 'pipeable' for processing with further shell tools.
3992 * 'hostnamectl set-hostname' will now allow setting of FQDN
3995 * The formatting and parsing of time span values has been
3996 changed. The parser now understands fractional expressions
3997 such as "5.5h". The formatter will now output fractional
3998 expressions for all time spans under 1min, i.e. "5.123456s"
3999 rather than "5s 123ms 456us". For time spans under 1s
4000 millisecond values are shown, for those under 1ms
4001 microsecond values are shown. This should greatly improve
4002 all time-related output of systemd.
4004 * libsystemd-login and libsystemd-journal gained new
4005 functions for querying the poll() events mask and poll()
4006 timeout value for integration into arbitrary event
4009 * localectl gained the ability to list available X11 keymaps
4010 (models, layouts, variants, options).
4012 * 'systemd-analyze dot' gained the ability to filter for
4013 specific units via shell-style globs, to create smaller,
4014 more useful graphs. I.e. it is now possible to create simple
4015 graphs of all the dependencies between only target units, or
4016 of all units that Avahi has dependencies with.
4018 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
4019 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
4020 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
4021 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
4022 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
4023 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
4024 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
4028 * The boot-time readahead implementation for rotating media
4029 will now read the read-ahead data in multiple passes which
4030 consist of all read requests made in equidistant time
4031 intervals. This means instead of strictly reading read-ahead
4032 data in its physical order on disk we now try to find a
4033 middle ground between physical and access time order.
4035 * /etc/os-release files gained a new BUILD_ID= field for usage
4036 on operating systems that provide continuous builds of OS
4039 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
4040 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
4041 William Douglas, Zbigniew Jędrzejewski-Szmek
4045 * systemd-python gained an API exposing libsystemd-daemon.
4047 * The SMACK setup logic gained support for uploading CIPSO
4050 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
4051 ReadOnlyDirectories= and InaccessibleDirectories= has
4052 changed. The private /tmp and /var/tmp directories are now
4053 shared by all processes of a service (which means
4054 ExecStartPre= may now leave data in /tmp that ExecStart= of
4055 the same service can still access). When a service is
4056 stopped its temporary directories are immediately deleted
4057 (normal clean-up with tmpfiles is still done in addition to
4060 * By default, systemd will now set a couple of sysctl
4061 variables in the kernel: the safe sysrq options are turned
4062 on, IP route verification is turned on, and source routing
4063 disabled. The recently added hardlink and softlink
4064 protection of the kernel is turned on. These settings should
4065 be reasonably safe, and good defaults for all new systems.
4067 * The predictable network naming logic may now be turned off
4068 with a new kernel command line switch: net.ifnames=0.
4070 * A new libsystemd-bus module has been added that implements a
4071 pretty complete D-Bus client library. For details see:
4073 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
4075 * journald will now explicitly flush the journal files to disk
4076 at the latest 5min after each write. The file will then also
4077 be marked offline until the next write. This should increase
4078 reliability in case of a crash. The synchronization delay
4079 can be configured via SyncIntervalSec= in journald.conf.
4081 * There's a new remote-fs-setup.target unit that can be used
4082 to pull in specific services when at least one remote file
4083 system is to be mounted.
4085 * There are new targets timers.target and paths.target as
4086 canonical targets to pull user timer and path units in
4087 from. This complements sockets.target with a similar
4088 purpose for socket units.
4090 * libudev gained a new call udev_device_set_attribute_value()
4091 to set sysfs attributes of a device.
4093 * The udev daemon now sets the default number of worker
4094 processes executed in parallel based on the number of available
4095 CPUs instead of the amount of available RAM. This is supposed
4096 to provide a more reliable default and limit a too aggressive
4097 paralellism for setups with 1000s of devices connected.
4099 Contributions from: Auke Kok, Colin Walters, Cristian
4100 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
4101 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
4102 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
4103 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
4104 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
4105 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
4106 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
4107 Zbigniew Jędrzejewski-Szmek
4111 * Configuration of unit files may now be extended via drop-in
4112 files without having to edit/override the unit files
4113 themselves. More specifically, if the administrator wants to
4114 change one value for a service file foobar.service he can
4115 now do so by dropping in a configuration snippet into
4116 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
4117 will load all these snippets and apply them on top of the
4118 main unit configuration file, possibly extending or
4119 overriding its settings. Using these drop-in snippets is
4120 generally nicer than the two earlier options for changing
4121 unit files locally: copying the files from
4122 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
4123 them there; or creating a new file in /etc/systemd/system/
4124 that incorporates the original one via ".include". Drop-in
4125 snippets into these .d/ directories can be placed in any
4126 directory systemd looks for units in, and the usual
4127 overriding semantics between /usr/lib, /etc and /run apply
4130 * Most unit file settings which take lists of items can now be
4131 reset by assigning the empty string to them. For example,
4132 normally, settings such as Environment=FOO=BAR append a new
4133 environment variable assignment to the environment block,
4134 each time they are used. By assigning Environment= the empty
4135 string the environment block can be reset to empty. This is
4136 particularly useful with the .d/*.conf drop-in snippets
4137 mentioned above, since this adds the ability to reset list
4138 settings from vendor unit files via these drop-ins.
4140 * systemctl gained a new "list-dependencies" command for
4141 listing the dependencies of a unit recursively.
4143 * Inhibitors are now honored and listed by "systemctl
4144 suspend", "systemctl poweroff" (and similar) too, not only
4145 GNOME. These commands will also list active sessions by
4148 * Resource limits (as exposed by the various control group
4149 controllers) can now be controlled dynamically at runtime
4150 for all units. More specifically, you can now use a command
4151 like "systemctl set-cgroup-attr foobar.service cpu.shares
4152 2000" to alter the CPU shares a specific service gets. These
4153 settings are stored persistently on disk, and thus allow the
4154 administrator to easily adjust the resource usage of
4155 services with a few simple commands. This dynamic resource
4156 management logic is also available to other programs via the
4157 bus. Almost any kernel cgroup attribute and controller is
4160 * systemd-vconsole-setup will now copy all font settings to
4161 all allocated VTs, where it previously applied them only to
4164 * libsystemd-login gained the new sd_session_get_tty() API
4167 * This release drops support for a few legacy or
4168 distribution-specific LSB facility names when parsing init
4169 scripts: $x-display-manager, $mail-transfer-agent,
4170 $mail-transport-agent, $mail-transfer-agent, $smtp,
4171 $null. Also, the mail-transfer-agent.target unit backing
4172 this has been removed. Distributions which want to retain
4173 compatibility with this should carry the burden for
4174 supporting this themselves and patch support for these back
4175 in, if they really need to. Also, the facilities $syslog and
4176 $local_fs are now ignored, since systemd does not support
4177 early-boot LSB init scripts anymore, and these facilities
4178 are implied anyway for normal services. syslog.target has
4181 * There are new bus calls on PID1's Manager object for
4182 cancelling jobs, and removing snapshot units. Previously,
4183 both calls were only available on the Job and Snapshot
4186 * systemd-journal-gatewayd gained SSL support.
4188 * The various "environment" files, such as /etc/locale.conf
4189 now support continuation lines with a backslash ("\") as
4190 last character in the line, similarly in style (but different)
4191 to how this is supported in shells.
4193 * For normal user processes the _SYSTEMD_USER_UNIT= field is
4194 now implicitly appended to every log entry logged. systemctl
4195 has been updated to filter by this field when operating on a
4196 user systemd instance.
4198 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
4199 CAP_AUDIT_CONTROL capabilities to the capabilities set for
4200 the container. This makes it easier to boot unmodified
4201 Fedora systems in a container, which however still requires
4202 audit=0 to be passed on the kernel command line. Auditing in
4203 kernel and userspace is unfortunately still too broken in
4204 context of containers, hence we recommend compiling it out
4205 of the kernel or using audit=0. Hopefully this will be fixed
4206 one day for good in the kernel.
4208 * nspawn gained the new --bind= and --bind-ro= parameters to
4209 bind mount specific directories from the host into the
4212 * nspawn will now mount its own devpts file system instance
4213 into the container, in order not to leak pty devices from
4214 the host into the container.
4216 * systemd will now read the firmware boot time performance
4217 information from the EFI variables, if the used boot loader
4218 supports this, and takes it into account for boot performance
4219 analysis via "systemd-analyze". This is currently supported
4220 only in conjunction with Gummiboot, but could be supported
4221 by other boot loaders too. For details see:
4223 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
4225 * A new generator has been added that automatically mounts the
4226 EFI System Partition (ESP) to /boot, if that directory
4227 exists, is empty, and no other file system has been
4228 configured to be mounted there.
4230 * logind will now send out PrepareForSleep(false) out
4231 unconditionally, after coming back from suspend. This may be
4232 used by applications as asynchronous notification for
4233 system resume events.
4235 * "systemctl unlock-sessions" has been added, that allows
4236 unlocking the screens of all user sessions at once, similar
4237 to how "systemctl lock-sessions" already locked all users
4238 sessions. This is backed by a new D-Bus call UnlockSessions().
4240 * "loginctl seat-status" will now show the master device of a
4241 seat. (i.e. the device of a seat that needs to be around for
4242 the seat to be considered available, usually the graphics
4245 * tmpfiles gained a new "X" line type, that allows
4246 configuration of files and directories (with wildcards) that
4247 shall be excluded from automatic cleanup ("aging").
4249 * udev default rules set the device node permissions now only
4250 at "add" events, and do not change them any longer with a
4251 later "change" event.
4253 * The log messages for lid events and power/sleep keypresses
4254 now carry a message ID.
4256 * We now have a substantially larger unit test suite, but this
4257 continues to be work in progress.
4259 * udevadm hwdb gained a new --root= parameter to change the
4260 root directory to operate relative to.
4262 * logind will now issue a background sync() request to the kernel
4263 early at shutdown, so that dirty buffers are flushed to disk early
4264 instead of at the last moment, in order to optimize shutdown
4267 * A new bootctl tool has been added that is an interface for
4268 certain boot loader operations. This is currently a preview
4269 and is likely to be extended into a small mechanism daemon
4270 like timedated, localed, hostnamed, and can be used by
4271 graphical UIs to enumerate available boot options, and
4272 request boot into firmware operations.
4274 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
4275 the rest of the package. It also has been updated to work
4276 correctly in initrds.
4278 * Policykit previously has been runtime optional, and is now
4279 also compile time optional via a configure switch.
4281 * systemd-analyze has been reimplemented in C. Also "systemctl
4282 dot" has moved into systemd-analyze.
4284 * "systemctl status" with no further parameters will now print
4285 the status of all active or failed units.
4287 * Operations such as "systemctl start" can now be executed
4288 with a new mode "--irreversible" which may be used to queue
4289 operations that cannot accidentally be reversed by a later
4290 job queuing. This is by default used to make shutdown
4291 requests more robust.
4293 * The Python API of systemd now gained a new module for
4294 reading journal files.
4296 * A new tool kernel-install has been added that can install
4297 kernel images according to the Boot Loader Specification:
4299 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
4301 * Boot time console output has been improved to provide
4302 animated boot time output for hanging jobs.
4304 * A new tool systemd-activate has been added which can be used
4305 to test socket activation with, directly from the command
4306 line. This should make it much easier to test and debug
4307 socket activation in daemons.
4309 * journalctl gained a new "--reverse" (or -r) option to show
4310 journal output in reverse order (i.e. newest line first).
4312 * journalctl gained a new "--pager-end" (or -e) option to jump
4313 to immediately jump to the end of the journal in the
4314 pager. This is only supported in conjunction with "less".
4316 * journalctl gained a new "--user-unit=" option, that works
4317 similarly to "--unit=" but filters for user units rather than
4320 * A number of unit files to ease adoption of systemd in
4321 initrds has been added. This moves some minimal logic from
4322 the various initrd implementations into systemd proper.
4324 * The journal files are now owned by a new group
4325 "systemd-journal", which exists specifically to allow access
4326 to the journal, and nothing else. Previously, we used the
4327 "adm" group for that, which however possibly covers more
4328 than just journal/log file access. This new group is now
4329 already used by systemd-journal-gatewayd to ensure this
4330 daemon gets access to the journal files and as little else
4331 as possible. Note that "make install" will also set FS ACLs
4332 up for /var/log/journal to give "adm" and "wheel" read
4333 access to it, in addition to "systemd-journal" which owns
4334 the journal files. We recommend that packaging scripts also
4335 add read access to "adm" + "wheel" to /var/log/journal, and
4336 all existing/future journal files. To normal users and
4337 administrators little changes, however packagers need to
4338 ensure to create the "systemd-journal" system group at
4339 package installation time.
4341 * The systemd-journal-gatewayd now runs as unprivileged user
4342 systemd-journal-gateway:systemd-journal-gateway. Packaging
4343 scripts need to create these system user/group at
4346 * timedated now exposes a new boolean property CanNTP that
4347 indicates whether a local NTP service is available or not.
4349 * systemd-detect-virt will now also detect xen PVs
4351 * The pstore file system is now mounted by default, if it is
4354 * In addition to the SELinux and IMA policies we will now also
4355 load SMACK policies at early boot.
4357 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
4358 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
4359 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
4360 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
4361 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
4362 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
4363 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
4364 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
4365 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
4366 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
4367 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
4368 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
4369 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
4370 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
4374 * Timer units now support calendar time events in addition to
4375 monotonic time events. That means you can now trigger a unit
4376 based on a calendar time specification such as "Thu,Fri
4377 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
4378 or fifth day of any month of the year 2013, given that it is
4379 a thursday or friday. This brings timer event support
4380 considerably closer to cron's capabilities. For details on
4381 the supported calendar time specification language see
4384 * udev now supports a number of different naming policies for
4385 network interfaces for predictable names, and a combination
4386 of these policies is now the default. Please see this wiki
4387 document for details:
4389 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
4391 * Auke Kok's bootchart implementation has been added to the
4392 systemd tree. It is an optional component that can graph the
4393 boot in quite some detail. It is one of the best bootchart
4394 implementations around and minimal in its code and
4397 * nss-myhostname has been integrated into the systemd source
4398 tree. nss-myhostname guarantees that the local hostname
4399 always stays resolvable via NSS. It has been a weak
4400 requirement of systemd-hostnamed since a long time, and
4401 since its code is actually trivial we decided to just
4402 include it in systemd's source tree. It can be turned off
4403 with a configure switch.
4405 * The read-ahead logic is now capable of properly detecting
4406 whether a btrfs file system is on SSD or rotating media, in
4407 order to optimize the read-ahead scheme. Previously, it was
4408 only capable of detecting this on traditional file systems
4411 * In udev, additional device properties are now read from the
4412 IAB in addition to the OUI database. Also, Bluetooth company
4413 identities are attached to the devices as well.
4415 * In service files %U may be used as specifier that is
4416 replaced by the configured user name of the service.
4418 * nspawn may now be invoked without a controlling TTY. This
4419 makes it suitable for invocation as its own service. This
4420 may be used to set up a simple containerized server system
4421 using only core OS tools.
4423 * systemd and nspawn can now accept socket file descriptors
4424 when they are started for socket activation. This enables
4425 implementation of socket activated nspawn
4426 containers. i.e. think about autospawning an entire OS image
4427 when the first SSH or HTTP connection is received. We expect
4428 that similar functionality will also be added to libvirt-lxc
4431 * journalctl will now suppress ANSI color codes when
4432 presenting log data.
4434 * systemctl will no longer show control group information for
4435 a unit if a the control group is empty anyway.
4437 * logind can now automatically suspend/hibernate/shutdown the
4440 * /etc/machine-info and hostnamed now also expose the chassis
4441 type of the system. This can be used to determine whether
4442 the local system is a laptop, desktop, handset or
4443 tablet. This information may either be configured by the
4444 user/vendor or is automatically determined from ACPI and DMI
4445 information if possible.
4447 * A number of PolicyKit actions are now bound together with
4448 "imply" rules. This should simplify creating UIs because
4449 many actions will now authenticate similar ones as well.
4451 * Unit files learnt a new condition ConditionACPower= which
4452 may be used to conditionalize a unit depending on whether an
4453 AC power source is connected or not, of whether the system
4454 is running on battery power.
4456 * systemctl gained a new "is-failed" verb that may be used in
4457 shell scripts and suchlike to check whether a specific unit
4458 is in the "failed" state.
4460 * The EnvironmentFile= setting in unit files now supports file
4461 globbing, and can hence be used to easily read a number of
4462 environment files at once.
4464 * systemd will no longer detect and recognize specific
4465 distributions. All distribution-specific #ifdeffery has been
4466 removed, systemd is now fully generic and
4467 distribution-agnostic. Effectively, not too much is lost as
4468 a lot of the code is still accessible via explicit configure
4469 switches. However, support for some distribution specific
4470 legacy configuration file formats has been dropped. We
4471 recommend distributions to simply adopt the configuration
4472 files everybody else uses now and convert the old
4473 configuration from packaging scripts. Most distributions
4474 already did that. If that's not possible or desirable,
4475 distributions are welcome to forward port the specific
4476 pieces of code locally from the git history.
4478 * When logging a message about a unit systemd will now always
4479 log the unit name in the message meta data.
4481 * localectl will now also discover system locale data that is
4482 not stored in locale archives, but directly unpacked.
4484 * logind will no longer unconditionally use framebuffer
4485 devices as seat masters, i.e. as devices that are required
4486 to be existing before a seat is considered preset. Instead,
4487 it will now look for all devices that are tagged as
4488 "seat-master" in udev. By default, framebuffer devices will
4489 be marked as such, but depending on local systems, other
4490 devices might be marked as well. This may be used to
4491 integrate graphics cards using closed source drivers (such
4492 as NVidia ones) more nicely into logind. Note however, that
4493 we recommend using the open source NVidia drivers instead,
4494 and no udev rules for the closed-source drivers will be
4495 shipped from us upstream.
4497 Contributions from: Adam Williamson, Alessandro Crismani, Auke
4498 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
4499 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
4500 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
4501 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
4502 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
4503 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
4504 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
4505 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
4506 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
4507 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
4508 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
4513 * udev gained support for loading additional device properties
4514 from an indexed database that is keyed by vendor/product IDs
4515 and similar device identifiers. For the beginning this
4516 "hwdb" is populated with data from the well-known PCI and
4517 USB database, but also includes PNP, ACPI and OID data. In
4518 the longer run this indexed database shall grow into
4519 becoming the one central database for non-essential
4520 userspace device metadata. Previously, data from the PCI/USB
4521 database was only attached to select devices, since the
4522 lookup was a relatively expensive operation due to O(n) time
4523 complexity (with n being the number of entries in the
4524 database). Since this is now O(1), we decided to add in this
4525 data for all devices where this is available, by
4526 default. Note that the indexed database needs to be rebuilt
4527 when new data files are installed. To achieve this you need
4528 to update your packaging scripts to invoke "udevadm hwdb
4529 --update" after installation of hwdb data files. For
4530 RPM-based distributions we introduced the new
4531 %udev_hwdb_update macro for this purpose.
4533 * The Journal gained support for the "Message Catalog", an
4534 indexed database to link up additional information with
4535 journal entries. For further details please check:
4537 http://www.freedesktop.org/wiki/Software/systemd/catalog
4539 The indexed message catalog database also needs to be
4540 rebuilt after installation of message catalog files. Use
4541 "journalctl --update-catalog" for this. For RPM-based
4542 distributions we introduced the %journal_catalog_update
4543 macro for this purpose.
4545 * The Python Journal bindings gained support for the standard
4546 Python logging framework.
4548 * The Journal API gained new functions for checking whether
4549 the underlying file system of a journal file is capable of
4550 properly reporting file change notifications, or whether
4551 applications that want to reflect journal changes "live"
4552 need to recheck journal files continuously in appropriate
4555 * It is now possible to set the "age" field for tmpfiles
4556 entries to 0, indicating that files matching this entry
4557 shall always be removed when the directories are cleaned up.
4559 * coredumpctl gained a new "gdb" verb which invokes gdb
4560 right-away on the selected coredump.
4562 * There's now support for "hybrid sleep" on kernels that
4563 support this, in addition to "suspend" and "hibernate". Use
4564 "systemctl hybrid-sleep" to make use of this.
4566 * logind's HandleSuspendKey= setting (and related settings)
4567 now gained support for a new "lock" setting to simply
4568 request the screen lock on all local sessions, instead of
4569 actually executing a suspend or hibernation.
4571 * systemd will now mount the EFI variables file system by
4574 * Socket units now gained support for configuration of the
4575 SMACK security label.
4577 * timedatectl will now output the time of the last and next
4578 daylight saving change.
4580 * We dropped support for various legacy and distro-specific
4581 concepts, such as insserv, early-boot SysV services
4582 (i.e. those for non-standard runlevels such as 'b' or 'S')
4583 or ArchLinux /etc/rc.conf support. We recommend the
4584 distributions who still need support this to either continue
4585 to maintain the necessary patches downstream, or find a
4586 different solution. (Talk to us if you have questions!)
4588 * Various systemd components will now bypass PolicyKit checks
4589 for root and otherwise handle properly if PolicyKit is not
4590 found to be around. This should fix most issues for
4591 PolicyKit-less systems. Quite frankly this should have been
4592 this way since day one. It is absolutely our intention to
4593 make systemd work fine on PolicyKit-less systems, and we
4594 consider it a bug if something does not work as it should if
4595 PolicyKit is not around.
4597 * For embedded systems it is now possible to build udev and
4598 systemd without blkid and/or kmod support.
4600 * "systemctl switch-root" is now capable of switching root
4601 more than once. I.e. in addition to transitions from the
4602 initrd to the host OS it is now possible to transition to
4603 further OS images from the host. This is useful to implement
4604 offline updating tools.
4606 * Various other additions have been made to the RPM macros
4607 shipped with systemd. Use %udev_rules_update() after
4608 installing new udev rules files. %_udevhwdbdir,
4609 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
4610 %_sysctldir are now available which resolve to the right
4611 directories for packages to place various data files in.
4613 * journalctl gained the new --full switch (in addition to
4614 --all, to disable ellipsation for long messages.
4616 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
4617 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
4618 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
4619 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
4620 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
4621 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
4622 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
4623 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
4624 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
4628 * journalctl gained new --since= and --until= switches to
4629 filter by time. It also now supports nice filtering for
4630 units via --unit=/-u.
4632 * Type=oneshot services may use ExecReload= and do the
4635 * The journal daemon now supports time-based rotation and
4636 vacuuming, in addition to the usual disk-space based
4639 * The journal will now index the available field values for
4640 each field name. This enables clients to show pretty drop
4641 downs of available match values when filtering. The bash
4642 completion of journalctl has been updated
4643 accordingly. journalctl gained a new switch -F to list all
4644 values a certain field takes in the journal database.
4646 * More service events are now written as structured messages
4647 to the journal, and made recognizable via message IDs.
4649 * The timedated, localed and hostnamed mini-services which
4650 previously only provided support for changing time, locale
4651 and hostname settings from graphical DEs such as GNOME now
4652 also have a minimal (but very useful) text-based client
4653 utility each. This is probably the nicest way to changing
4654 these settings from the command line now, especially since
4655 it lists available options and is fully integrated with bash
4658 * There's now a new tool "systemd-coredumpctl" to list and
4659 extract coredumps from the journal.
4661 * We now install a README each in /var/log/ and
4662 /etc/rc.d/init.d explaining where the system logs and init
4663 scripts went. This hopefully should help folks who go to
4664 that dirs and look into the otherwise now empty void and
4665 scratch their heads.
4667 * When user-services are invoked (by systemd --user) the
4668 $MANAGERPID env var is set to the PID of systemd.
4670 * SIGRTMIN+24 when sent to a --user instance will now result
4671 in immediate termination of systemd.
4673 * gatewayd received numerous feature additions such as a
4674 "follow" mode, for live syncing and filtering.
4676 * browse.html now allows filtering and showing detailed
4677 information on specific entries. Keyboard navigation and
4678 mouse screen support has been added.
4680 * gatewayd/journalctl now supports HTML5/JSON
4681 Server-Sent-Events as output.
4683 * The SysV init script compatibility logic will now
4684 heuristically determine whether a script supports the
4685 "reload" verb, and only then make this available as
4688 * "systemctl status --follow" has been removed, use "journalctl
4691 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
4692 have been removed since they are hardly useful to be
4695 * And I'd like to take the opportunity to specifically mention
4696 Zbigniew for his great contributions. Zbigniew, you rock!
4698 Contributions from: Andrew Eikum, Christian Hesse, Colin
4699 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
4700 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
4701 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
4702 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
4703 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
4704 Jędrzejewski-Szmek, Сковорода Никита Андреевич
4708 * If /etc/vconsole.conf is non-existent or empty we will no
4709 longer load any console font or key map at boot by
4710 default. Instead the kernel defaults will be left
4711 intact. This is definitely the right thing to do, as no
4712 configuration should mean no configuration, and hard-coding
4713 font names that are different on all archs is probably a bad
4714 idea. Also, the kernel default key map and font should be
4715 good enough for most cases anyway, and mostly identical to
4716 the userspace fonts/key maps we previously overloaded them
4717 with. If distributions want to continue to default to a
4718 non-kernel font or key map they should ship a default
4719 /etc/vconsole.conf with the appropriate contents.
4721 Contributions from: Colin Walters, Daniel J Walsh, Dave
4722 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
4723 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
4727 * journalctl gained a new --cursor= switch to show entries
4728 starting from the specified location in the journal.
4730 * We now enforce a size limit on journal entry fields exported
4731 with "-o json" in journalctl. Fields larger than 4K will be
4732 assigned null. This can be turned off with --all.
4734 * An (optional) journal gateway daemon is now available as
4735 "systemd-journal-gatewayd.service". This service provides
4736 access to the journal via HTTP and JSON. This functionality
4737 will be used to implement live log synchronization in both
4738 pull and push modes, but has various other users too, such
4739 as easy log access for debugging of embedded devices. Right
4740 now it is already useful to retrieve the journal via HTTP:
4742 # systemctl start systemd-journal-gatewayd.service
4743 # wget http://localhost:19531/entries
4745 This will download the journal contents in a
4746 /var/log/messages compatible format. The same as JSON:
4748 # curl -H"Accept: application/json" http://localhost:19531/entries
4750 This service is also accessible via a web browser where a
4751 single static HTML5 app is served that uses the JSON logic
4752 to enable the user to do some basic browsing of the
4753 journal. This will be extended later on. Here's an example
4754 screenshot of this app in its current state:
4756 http://0pointer.de/public/journal-gatewayd
4758 Contributions from: Kay Sievers, Lennart Poettering, Robert
4759 Milasan, Tom Gundersen
4763 * The bash completion logic is now available for journalctl
4766 * We do not mount the "cpuset" controller anymore together with
4767 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
4768 started if no parameters are assigned to it. "cpuset" hence
4769 broke code that assumed it it could create "cpu" groups and
4772 * journalctl -f will now subscribe to terminal size changes,
4773 and line break accordingly.
4775 Contributions from: Dave Reisner, Kay Sievers, Lennart
4776 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
4780 * nspawn will now create a symlink /etc/localtime in the
4781 container environment, copying the host's timezone
4782 setting. Previously this has been done via a bind mount, but
4783 since symlinks cannot be bind mounted this has now been
4784 changed to create/update the appropriate symlink.
4786 * journalctl -n's line number argument is now optional, and
4787 will default to 10 if omitted.
4789 * journald will now log the maximum size the journal files may
4790 take up on disk. This is particularly useful if the default
4791 built-in logic of determining this parameter from the file
4792 system size is used. Use "systemctl status
4793 systemd-journald.service" to see this information.
4795 * The multi-seat X wrapper tool has been stripped down. As X
4796 is now capable of enumerating graphics devices via udev in a
4797 seat-aware way the wrapper is not strictly necessary
4798 anymore. A stripped down temporary stop-gap is still shipped
4799 until the upstream display managers have been updated to
4800 fully support the new X logic. Expect this wrapper to be
4801 removed entirely in one of the next releases.
4803 * HandleSleepKey= in logind.conf has been split up into
4804 HandleSuspendKey= and HandleHibernateKey=. The old setting
4805 is not available anymore. X11 and the kernel are
4806 distinguishing between these keys and we should too. This
4807 also means the inhibition lock for these keys has been split
4810 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
4811 Poettering, Lukas Nykryn, Václav Pavlín
4815 * Whenever a unit changes state we will now log this to the
4816 journal and show along the unit's own log output in
4819 * ConditionPathIsMountPoint= can now properly detect bind
4820 mount points too. (Previously, a bind mount of one file
4821 system to another place in the same file system could not be
4822 detected as mount, since they shared struct stat's st_dev
4825 * We will now mount the cgroup controllers cpu, cpuacct,
4826 cpuset and the controllers net_cls, net_prio together by
4829 * nspawn containers will now have a virtualized boot
4830 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
4831 over with a randomized ID at container initialization). This
4832 has the effect of making "journalctl -b" do the right thing
4835 * The JSON output journal serialization has been updated not
4836 to generate "endless" list objects anymore, but rather one
4837 JSON object per line. This is more in line how most JSON
4838 parsers expect JSON objects. The new output mode
4839 "json-pretty" has been added to provide similar output, but
4840 neatly aligned for readability by humans.
4842 * We dropped all explicit sync() invocations in the shutdown
4843 code. The kernel does this implicitly anyway in the kernel
4844 reboot() syscall. halt(8)'s -n option is now a compatibility
4847 * We now support virtualized reboot() in containers, as
4848 supported by newer kernels. We will fall back to exit() if
4849 CAP_SYS_REBOOT is not available to the container. Also,
4850 nspawn makes use of this now and will actually reboot the
4851 container if the containerized OS asks for that.
4853 * journalctl will only show local log output by default
4854 now. Use --merge (-m) to show remote log output, too.
4856 * libsystemd-journal gained the new sd_journal_get_usage()
4857 call to determine the current disk usage of all journal
4858 files. This is exposed in the new "journalctl --disk-usage"
4861 * journald gained a new configuration setting SplitMode= in
4862 journald.conf which may be used to control how user journals
4863 are split off. See journald.conf(5) for details.
4865 * A new condition type ConditionFileNotEmpty= has been added.
4867 * tmpfiles' "w" lines now support file globbing, to write
4868 multiple files at once.
4870 * We added Python bindings for the journal submission
4871 APIs. More Python APIs for a number of selected APIs will
4872 likely follow. Note that we intend to add native bindings
4873 only for the Python language, as we consider it common
4874 enough to deserve bindings shipped within systemd. There are
4875 various projects outside of systemd that provide bindings
4876 for languages such as PHP or Lua.
4878 * Many conditions will now resolve specifiers such as %i. In
4879 addition, PathChanged= and related directives of .path units
4880 now support specifiers as well.
4882 * There's now a new RPM macro definition for the system preset
4885 * journald will now warn if it ca not forward a message to the
4886 syslog daemon because its socket is full.
4888 * timedated will no longer write or process /etc/timezone,
4889 except on Debian. As we do not support late mounted /usr
4890 anymore /etc/localtime always being a symlink is now safe,
4891 and hence the information in /etc/timezone is not necessary
4894 * logind will now always reserve one VT for a text getty (VT6
4895 by default). Previously if more than 6 X sessions where
4896 started they took up all the VTs with auto-spawned gettys,
4897 so that no text gettys were available anymore.
4899 * udev will now automatically inform the btrfs kernel logic
4900 about btrfs RAID components showing up. This should make
4901 simple hotplug based btrfs RAID assembly work.
4903 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
4904 (but not for its children which will stay at the kernel
4905 default). This should allow setups with a lot more listening
4908 * systemd will now always pass the configured timezone to the
4909 kernel at boot. timedated will do the same when the timezone
4912 * logind's inhibition logic has been updated. By default,
4913 logind will now handle the lid switch, the power and sleep
4914 keys all the time, even in graphical sessions. If DEs want
4915 to handle these events on their own they should take the new
4916 handle-power-key, handle-sleep-key and handle-lid-switch
4917 inhibitors during their runtime. A simple way to achieve
4918 that is to invoke the DE wrapped in an invocation of:
4920 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
4922 * Access to unit operations is now checked via SELinux taking
4923 the unit file label and client process label into account.
4925 * systemd will now notify the administrator in the journal
4926 when he over-mounts a non-empty directory.
4928 * There are new specifiers that are resolved in unit files,
4929 for the host name (%H), the machine ID (%m) and the boot ID
4932 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
4933 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
4934 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
4935 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
4936 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
4937 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
4938 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
4942 * Support for reading structured kernel messages from
4943 /dev/kmsg has now been added and is enabled by default.
4945 * Support for reading kernel messages from /proc/kmsg has now
4946 been removed. If you want kernel messages in the journal
4947 make sure to run a recent kernel (>= 3.5) that supports
4948 reading structured messages from /dev/kmsg (see
4949 above). /proc/kmsg is now exclusive property of classic
4950 syslog daemons again.
4952 * The libudev API gained the new
4953 udev_device_new_from_device_id() call.
4955 * The logic for file system namespace (ReadOnlyDirectory=,
4956 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
4957 require pivot_root() anymore. This means fewer temporary
4958 directories are created below /tmp for this feature.
4960 * nspawn containers will now see and receive all submounts
4961 made on the host OS below the root file system of the
4964 * Forward Secure Sealing is now supported for Journal files,
4965 which provide cryptographical sealing of journal files so
4966 that attackers cannot alter log history anymore without this
4967 being detectable. Lennart will soon post a blog story about
4968 this explaining it in more detail.
4970 * There are two new service settings RestartPreventExitStatus=
4971 and SuccessExitStatus= which allow configuration of exit
4972 status (exit code or signal) which will be excepted from the
4973 restart logic, resp. consider successful.
4975 * journalctl gained the new --verify switch that can be used
4976 to check the integrity of the structure of journal files and
4977 (if Forward Secure Sealing is enabled) the contents of
4980 * nspawn containers will now be run with /dev/stdin, /dev/fd/
4981 and similar symlinks pre-created. This makes running shells
4982 as container init process a lot more fun.
4984 * The fstab support can now handle PARTUUID= and PARTLABEL=
4987 * A new ConditionHost= condition has been added to match
4988 against the hostname (with globs) and machine ID. This is
4989 useful for clusters where a single OS image is used to
4990 provision a large number of hosts which shall run slightly
4991 different sets of services.
4993 * Services which hit the restart limit will now be placed in a
4996 Contributions from: Bertram Poettering, Dave Reisner, Huang
4997 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
4998 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
5002 * When running in --user mode systemd will now become a
5003 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
5004 tree a lot more organized.
5006 * A new PartOf= unit dependency type has been introduced that
5007 may be used to group services in a natural way.
5009 * "systemctl enable" may now be used to enable instances of
5012 * journalctl now prints error log levels in red, and
5013 warning/notice log levels in bright white. It also supports
5014 filtering by log level now.
5016 * cgtop gained a new -n switch (similar to top), to configure
5017 the maximum number of iterations to run for. It also gained
5018 -b, to run in batch mode (accepting no input).
5020 * The suffix ".service" may now be omitted on most systemctl
5021 command lines involving service unit names.
5023 * There's a new bus call in logind to lock all sessions, as
5024 well as a loginctl verb for it "lock-sessions".
5026 * libsystemd-logind.so gained a new call sd_journal_perror()
5027 that works similar to libc perror() but logs to the journal
5028 and encodes structured information about the error number.
5030 * /etc/crypttab entries now understand the new keyfile-size=
5033 * shutdown(8) now can send a (configurable) wall message when
5034 a shutdown is cancelled.
5036 * The mount propagation mode for the root file system will now
5037 default to "shared", which is useful to make containers work
5038 nicely out-of-the-box so that they receive new mounts from
5039 the host. This can be undone locally by running "mount
5040 --make-rprivate /" if needed.
5042 * The prefdm.service file has been removed. Distributions
5043 should maintain this unit downstream if they intend to keep
5044 it around. However, we recommend writing normal unit files
5045 for display managers instead.
5047 * Since systemd is a crucial part of the OS we will now
5048 default to a number of compiler switches that improve
5049 security (hardening) such as read-only relocations, stack
5050 protection, and suchlike.
5052 * The TimeoutSec= setting for services is now split into
5053 TimeoutStartSec= and TimeoutStopSec= to allow configuration
5054 of individual time outs for the start and the stop phase of
5057 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
5058 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
5059 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
5060 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
5061 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
5062 Gundersen, Zbigniew Jędrzejewski-Szmek
5066 * The journal and id128 C APIs are now fully documented as man
5069 * Extra safety checks have been added when transitioning from
5070 the initial RAM disk to the main system to avoid accidental
5073 * /etc/crypttab entries now understand the new keyfile-offset=
5076 * systemctl -t can now be used to filter by unit load state.
5078 * The journal C API gained the new sd_journal_wait() call to
5079 make writing synchronous journal clients easier.
5081 * journalctl gained the new -D switch to show journals from a
5084 * journalctl now displays a special marker between log
5085 messages of two different boots.
5087 * The journal is now explicitly flushed to /var via a service
5088 systemd-journal-flush.service, rather than implicitly simply
5089 by seeing /var/log/journal to be writable.
5091 * journalctl (and the journal C APIs) can now match for much
5092 more complex expressions, with alternatives and
5095 * When transitioning from the initial RAM disk to the main
5096 system we will now kill all processes in a killing spree to
5097 ensure no processes stay around by accident.
5099 * Three new specifiers may be used in unit files: %u, %h, %s
5100 resolve to the user name, user home directory resp. user
5101 shell. This is useful for running systemd user instances.
5103 * We now automatically rotate journal files if their data
5104 object hash table gets a fill level > 75%. We also size the
5105 hash table based on the configured maximum file size. This
5106 together should lower hash collisions drastically and thus
5107 speed things up a bit.
5109 * journalctl gained the new "--header" switch to introspect
5110 header data of journal files.
5112 * A new setting SystemCallFilters= has been added to services
5113 which may be used to apply blacklists or whitelists to
5114 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
5116 * nspawn gained a new --link-journal= switch (and quicker: -j)
5117 to link the container journal with the host. This makes it
5118 very easy to centralize log viewing on the host for all
5119 guests while still keeping the journal files separated.
5121 * Many bugfixes and optimizations
5123 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
5124 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
5125 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
5130 * Several tools now understand kernel command line arguments,
5131 which are only read when run in an initial RAM disk. They
5132 usually follow closely their normal counterparts, but are
5135 * There's a new tool to analyze the readahead files that are
5136 automatically generated at boot. Use:
5138 /usr/lib/systemd/systemd-readahead analyze /.readahead
5140 * We now provide an early debug shell on tty9 if this enabled. Use:
5142 systemctl enable debug-shell.service
5144 * All plymouth related units have been moved into the Plymouth
5145 package. Please make sure to upgrade your Plymouth version
5148 * systemd-tmpfiles now supports getting passed the basename of
5149 a configuration file only, in which case it will look for it
5150 in all appropriate directories automatically.
5152 * udevadm info now takes a /dev or /sys path as argument, and
5153 does the right thing. Example:
5155 udevadm info /dev/sda
5156 udevadm info /sys/class/block/sda
5158 * systemctl now prints a warning if a unit is stopped but a
5159 unit that might trigger it continues to run. Example: a
5160 service is stopped but the socket that activates it is left
5163 * "systemctl status" will now mention if the log output was
5164 shortened due to rotation since a service has been started.
5166 * The journal API now exposes functions to determine the
5167 "cutoff" times due to rotation.
5169 * journald now understands SIGUSR1 and SIGUSR2 for triggering
5170 immediately flushing of runtime logs to /var if possible,
5171 resp. for triggering immediate rotation of the journal
5174 * It is now considered an error if a service is attempted to
5175 be stopped that is not loaded.
5177 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
5179 * systemd-analyze now supports Python 3
5181 * tmpfiles now supports cleaning up directories via aging
5182 where the first level dirs are always kept around but
5183 directories beneath it automatically aged. This is enabled
5184 by prefixing the age field with '~'.
5186 * Seat objects now expose CanGraphical, CanTTY properties
5187 which is required to deal with very fast bootups where the
5188 display manager might be running before the graphics drivers
5189 completed initialization.
5191 * Seat objects now expose a State property.
5193 * We now include RPM macros for service enabling/disabling
5194 based on the preset logic. We recommend RPM based
5195 distributions to make use of these macros if possible. This
5196 makes it simpler to reuse RPM spec files across
5199 * We now make sure that the collected systemd unit name is
5200 always valid when services log to the journal via
5203 * There's a new man page kernel-command-line(7) detailing all
5204 command line options we understand.
5206 * The fstab generator may now be disabled at boot by passing
5207 fstab=0 on the kernel command line.
5209 * A new kernel command line option modules-load= is now understood
5210 to load a specific kernel module statically, early at boot.
5212 * Unit names specified on the systemctl command line are now
5213 automatically escaped as needed. Also, if file system or
5214 device paths are specified they are automatically turned
5215 into the appropriate mount or device unit names. Example:
5217 systemctl status /home
5218 systemctl status /dev/sda
5220 * The SysVConsole= configuration option has been removed from
5221 system.conf parsing.
5223 * The SysV search path is no longer exported on the D-Bus
5226 * The Names= option is been removed from unit file parsing.
5228 * There's a new man page bootup(7) detailing the boot process.
5230 * Every unit and every generator we ship with systemd now
5231 comes with full documentation. The self-explanatory boot is
5234 * A couple of services gained "systemd-" prefixes in their
5235 name if they wrap systemd code, rather than only external
5236 code. Among them fsck@.service which is now
5237 systemd-fsck@.service.
5239 * The HaveWatchdog property has been removed from the D-Bus
5242 * systemd.confirm_spawn= on the kernel command line should now
5245 * There's a new man page crypttab(5) which details all options
5246 we actually understand.
5248 * systemd-nspawn gained a new --capability= switch to pass
5249 additional capabilities to the container.
5251 * timedated will now read known NTP implementation unit names
5252 from /usr/lib/systemd/ntp-units.d/*.list,
5253 systemd-timedated-ntp.target has been removed.
5255 * journalctl gained a new switch "-b" that lists log data of
5256 the current boot only.
5258 * The notify socket is in the abstract namespace again, in
5259 order to support daemons which chroot() at start-up.
5261 * There is a new Storage= configuration option for journald
5262 which allows configuration of where log data should go. This
5263 also provides a way to disable journal logging entirely, so
5264 that data collected is only forwarded to the console, the
5265 kernel log buffer or another syslog implementation.
5267 * Many bugfixes and optimizations
5269 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
5270 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
5271 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
5272 Shawn Landden, Tom Gundersen
5276 * "systemctl help <unit>" now shows the man page if one is
5279 * Several new man pages have been added.
5281 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
5282 MaxLevelConsole= can now be specified in
5283 journald.conf. These options allow reducing the amount of
5284 data stored on disk or forwarded by the log level.
5286 * TimerSlackNSec= can now be specified in system.conf for
5287 PID1. This allows system-wide power savings.
5289 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
5290 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
5295 * logind is now capable of (optionally) handling power and
5296 sleep keys as well as the lid switch.
5298 * journalctl now understands the syntax "journalctl
5299 /usr/bin/avahi-daemon" to get all log output of a specific
5302 * CapabilityBoundingSet= in system.conf now also influences
5303 the capability bound set of usermode helpers of the kernel.
5305 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
5306 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
5307 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
5308 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
5312 * Note that we skipped 139 releases here in order to set the
5313 new version to something that is greater than both udev's
5314 and systemd's most recent version number.
5316 * udev: all udev sources are merged into the systemd source tree now.
5317 All future udev development will happen in the systemd tree. It
5318 is still fully supported to use the udev daemon and tools without
5319 systemd running, like in initramfs or other init systems. Building
5320 udev though, will require the *build* of the systemd tree, but
5321 udev can be properly *run* without systemd.
5323 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
5324 should be used to create dead device nodes as workarounds for broken
5327 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
5328 no longer supported. udev_monitor_new_from_netlink() needs to be
5329 used to subscribe to events.
5331 * udev: when udevd is started by systemd, processes which are left
5332 behind by forking them off of udev rules, are unconditionally cleaned
5333 up and killed now after the event handling has finished. Services or
5334 daemons must be started as systemd services. Services can be
5335 pulled-in by udev to get started, but they can no longer be directly
5336 forked by udev rules.
5338 * udev: the daemon binary is called systemd-udevd now and installed
5339 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
5340 to adapt to that, create symlink, or rename the binary after building
5343 * libudev no longer provides these symbols:
5344 udev_monitor_from_socket()
5345 udev_queue_get_failed_list_entry()
5346 udev_get_{dev,sys,run}_path()
5347 The versions number was bumped and symbol versioning introduced.
5349 * systemd-loginctl and systemd-journalctl have been renamed
5350 to loginctl and journalctl to match systemctl.
5352 * The config files: /etc/systemd/systemd-logind.conf and
5353 /etc/systemd/systemd-journald.conf have been renamed to
5354 logind.conf and journald.conf. Package updates should rename
5355 the files to the new names on upgrade.
5357 * For almost all files the license is now LGPL2.1+, changed
5358 from the previous GPL2.0+. Exceptions are some minor stuff
5359 of udev (which will be changed to LGPL2.1 eventually, too),
5360 and the MIT licensed sd-daemon.[ch] library that is suitable
5361 to be used as drop-in files.
5363 * systemd and logind now handle system sleep states, in
5364 particular suspending and hibernating.
5366 * logind now implements a sleep/shutdown/idle inhibiting logic
5367 suitable for a variety of uses. Soonishly Lennart will blog
5368 about this in more detail.
5370 * var-run.mount and var-lock.mount are no longer provided
5371 (which prevously bind mounted these directories to their new
5372 places). Distributions which have not converted these
5373 directories to symlinks should consider stealing these files
5374 from git history and add them downstream.
5376 * We introduced the Documentation= field for units and added
5377 this to all our shipped units. This is useful to make it
5378 easier to explore the boot and the purpose of the various
5381 * All smaller setup units (such as
5382 systemd-vconsole-setup.service) now detect properly if they
5383 are run in a container and are skipped when
5384 appropriate. This guarantees an entirely noise-free boot in
5385 Linux container environments such as systemd-nspawn.
5387 * A framework for implementing offline system updates is now
5388 integrated, for details see:
5389 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
5391 * A new service type Type=idle is available now which helps us
5392 avoiding ugly interleaving of getty output and boot status
5395 * There's now a system-wide CapabilityBoundingSet= option to
5396 globally reduce the set of capabilities for the
5397 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
5398 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
5399 even CAP_NET_ADMIN system-wide for secure systems.
5401 * There are now system-wide DefaultLimitXXX= options to
5402 globally change the defaults of the various resource limits
5403 for all units started by PID 1.
5405 * Harald Hoyer's systemd test suite has been integrated into
5406 systemd which allows easy testing of systemd builds in qemu
5407 and nspawn. (This is really awesome! Ask us for details!)
5409 * The fstab parser is now implemented as generator, not inside
5412 * systemctl will now warn you if .mount units generated from
5413 /etc/fstab are out of date due to changes in fstab that
5414 have not been read by systemd yet.
5416 * systemd is now suitable for usage in initrds. Dracut has
5417 already been updated to make use of this. With this in place
5418 initrds get a slight bit faster but primarily are much
5419 easier to introspect and debug since "systemctl status" in
5420 the host system can be used to introspect initrd services,
5421 and the journal from the initrd is kept around too.
5423 * systemd-delta has been added, a tool to explore differences
5424 between user/admin configuration and vendor defaults.
5426 * PrivateTmp= now affects both /tmp and /var/tmp.
5428 * Boot time status messages are now much prettier and feature
5429 proper english language. Booting up systemd has never been
5432 * Read-ahead pack files now include the inode number of all
5433 files to pre-cache. When the inode changes the pre-caching
5434 is not attempted. This should be nicer to deal with updated
5435 packages which might result in changes of read-ahead
5438 * We now temporaritly lower the kernel's read_ahead_kb variable
5439 when collecting read-ahead data to ensure the kernel's
5440 built-in read-ahead does not add noise to our measurements
5441 of necessary blocks to pre-cache.
5443 * There's now RequiresMountsFor= to add automatic dependencies
5444 for all mounts necessary for a specific file system path.
5446 * MountAuto= and SwapAuto= have been removed from
5447 system.conf. Mounting file systems at boot has to take place
5450 * nspawn now learned a new switch --uuid= to set the machine
5451 ID on the command line.
5453 * nspawn now learned the -b switch to automatically search
5456 * vt102 is now the default TERM for serial TTYs, upgraded from
5459 * systemd-logind now works on VT-less systems.
5461 * The build tree has been reorganized. The individual
5462 components now have directories of their own.
5464 * A new condition type ConditionPathIsReadWrite= is now available.
5466 * nspawn learned the new -C switch to create cgroups for the
5467 container in other hierarchies.
5469 * We now have support for hardware watchdogs, configurable in
5472 * The scheduled shutdown logic now has a public API.
5474 * We now mount /tmp as tmpfs by default, but this can be
5475 masked and /etc/fstab can override it.
5477 * Since udisks does not make use of /media anymore we are not
5478 mounting a tmpfs on it anymore.
5480 * journalctl gained a new --local switch to only interleave
5481 locally generated journal files.
5483 * We can now load the IMA policy at boot automatically.
5485 * The GTK tools have been split off into a systemd-ui.
5487 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
5488 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
5489 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
5490 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
5491 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
5492 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
5493 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
5494 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
5495 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
5500 * This is mostly a bugfix release
5502 * Support optional initialization of the machine ID from the
5503 KVM or container configured UUID.
5505 * Support immediate reboots with "systemctl reboot -ff"
5507 * Show /etc/os-release data in systemd-analyze output
5509 * Many bugfixes for the journal, including endianness fixes and
5510 ensuring that disk space enforcement works
5512 * sd-login.h is C++ comptaible again
5514 * Extend the /etc/os-release format on request of the Debian
5517 * We now refuse non-UTF8 strings used in various configuration
5518 and unit files. This is done to ensure we do not pass invalid
5519 data over D-Bus or expose it elsewhere.
5521 * Register Mimo USB Screens as suitable for automatic seat
5524 * Read SELinux client context from journal clients in a race
5527 * Reorder configuration file lookup order. /etc now always
5528 overrides /run in order to allow the administrator to always
5529 and unconditionally override vendor-supplied or
5530 automatically generated data.
5532 * The various user visible bits of the journal now have man
5533 pages. We still lack man pages for the journal API calls
5536 * We now ship all man pages in HTML format again in the
5539 Contributions from: Dave Reisner, Dirk Eibach, Frederic
5540 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
5541 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
5546 * This is mostly a bugfix release
5548 * systems lacking /etc/os-release are no longer supported.
5550 * Various functionality updates to libsystemd-login.so
5552 * Track class of PAM logins to distinguish greeters from
5555 Contributions from: Kay Sievers, Lennart Poettering, Michael
5560 * This is an important bugfix release for v41.
5562 * Building man pages is now optional which should be useful
5563 for those building systemd from git but unwilling to install
5566 * Watchdog support for supervising services is now usable. In
5567 a future release support for hardware watchdogs
5568 (i.e. /dev/watchdog) will be added building on this.
5570 * Service start rate limiting is now configurable and can be
5571 turned off per service. When a start rate limit is hit a
5572 reboot can automatically be triggered.
5574 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
5576 Contributions from: Benjamin Franzke, Bill Nottingham,
5577 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
5578 Schmidt, Michał Górny, Piotr Drąg
5582 * The systemd binary is installed /usr/lib/systemd/systemd now;
5583 An existing /sbin/init symlink needs to be adapted with the
5586 * The code that loads kernel modules has been ported to invoke
5587 libkmod directly, instead of modprobe. This means we do not
5588 support systems with module-init-tools anymore.
5590 * Watchdog support is now already useful, but still not
5593 * A new kernel command line option systemd.setenv= is
5594 understood to set system wide environment variables
5595 dynamically at boot.
5597 * We now limit the set of capabilities of systemd-journald.
5599 * We now set SIGPIPE to ignore by default, since it only is
5600 useful in shell pipelines, and has little use in general
5601 code. This can be disabled with IgnoreSIPIPE=no in unit
5604 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
5605 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
5610 * This is mostly a bugfix release
5612 * We now expose the reason why a service failed in the
5613 "Result" D-Bus property.
5615 * Rudimentary service watchdog support (will be completed over
5616 the next few releases.)
5618 * When systemd forks off in order execute some service we will
5619 now immediately changes its argv[0] to reflect which process
5620 it will execute. This is useful to minimize the time window
5621 with a generic argv[0], which makes bootcharts more useful
5623 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
5624 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
5625 Mike Kazantsev, Ray Strode
5629 * This is mostly a test release, but incorporates many
5632 * New systemd-cgtop tool to show control groups by their
5635 * Linking against libacl for ACLs is optional again. If
5636 disabled, support tracking device access for active logins
5637 goes becomes unavailable, and so does access to the user
5638 journals by the respective users.
5640 * If a group "adm" exists, journal files are automatically
5641 owned by them, thus allow members of this group full access
5642 to the system journal as well as all user journals.
5644 * The journal now stores the SELinux context of the logging
5645 client for all entries.
5647 * Add C++ inclusion guards to all public headers
5649 * New output mode "cat" in the journal to print only text
5650 messages, without any meta data like date or time.
5652 * Include tiny X server wrapper as a temporary stop-gap to
5653 teach XOrg udev display enumeration. This is used by display
5654 managers such as gdm, and will go away as soon as XOrg
5655 learned native udev hotplugging for display devices.
5657 * Add new systemd-cat tool for executing arbitrary programs
5658 with STDERR/STDOUT connected to the journal. Can also act as
5659 BSD logger replacement, and does so by default.
5661 * Optionally store all locally generated coredumps in the
5662 journal along with meta data.
5664 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
5665 writing short strings to files (for usage for /sys), and for
5666 creating symlinks, character and block device nodes.
5668 * New unit file option ControlGroupPersistent= to make cgroups
5669 persistent, following the mechanisms outlined in
5670 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
5672 * Support multiple local RTCs in a sane way
5674 * No longer monopolize IO when replaying readahead data on
5675 rotating disks, since we might starve non-file-system IO to
5676 death, since fanotify() will not see accesses done by blkid,
5679 * Do not show kernel threads in systemd-cgls anymore, unless
5680 requested with new -k switch.
5682 Contributions from: Dan Horák, Kay Sievers, Lennart
5683 Poettering, Michal Schmidt
5687 * This is mostly a test release, but incorporates many
5690 * The git repository moved to:
5691 git://anongit.freedesktop.org/systemd/systemd
5692 ssh://git.freedesktop.org/git/systemd/systemd
5694 * First release with the journal
5695 http://0pointer.de/blog/projects/the-journal.html
5697 * The journal replaces both systemd-kmsg-syslogd and
5698 systemd-stdout-bridge.
5700 * New sd_pid_get_unit() API call in libsystemd-logind
5702 * Many systemadm clean-ups
5704 * Introduce remote-fs-pre.target which is ordered before all
5705 remote mounts and may be used to start services before all
5708 * Added Mageia support
5710 * Add bash completion for systemd-loginctl
5712 * Actively monitor PID file creation for daemons which exit in
5713 the parent process before having finished writing the PID
5714 file in the daemon process. Daemons which do this need to be
5715 fixed (i.e. PID file creation must have finished before the
5716 parent exits), but we now react a bit more gracefully to them.
5718 * Add colourful boot output, mimicking the well-known output
5719 of existing distributions.
5721 * New option PassCredentials= for socket units, for
5722 compatibility with a recent kernel ABI breakage.
5724 * /etc/rc.local is now hooked in via a generator binary, and
5725 thus will no longer act as synchronization point during
5728 * systemctl list-unit-files now supports --root=.
5730 * systemd-tmpfiles now understands two new commands: z, Z for
5731 relabelling files according to the SELinux database. This is
5732 useful to apply SELinux labels to specific files in /sys,
5735 * Output of SysV services is now forwarded to both the console
5736 and the journal by default, not only just the console.
5738 * New man pages for all APIs from libsystemd-login.
5740 * The build tree got reorganized and a the build system is a
5741 lot more modular allowing embedded setups to specifically
5742 select the components of systemd they are interested in.
5744 * Support for Linux systems lacking the kernel VT subsystem is
5747 * configure's --with-rootdir= got renamed to
5748 --with-rootprefix= to follow the naming used by udev and
5751 * Unless specified otherwise we will now install to /usr instead
5752 of /usr/local by default.
5754 * Processes with '@' in argv[0][0] are now excluded from the
5755 final shut-down killing spree, following the logic explained
5757 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
5759 * All processes remaining in a service cgroup when we enter
5760 the START or START_PRE states are now killed with
5761 SIGKILL. That means it is no longer possible to spawn
5762 background processes from ExecStart= lines (which was never
5763 supported anyway, and bad style).
5765 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
5766 reloading of units together.
5768 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
5769 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
5770 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
5771 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
5772 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek