1 systemd System and Service Manager
5 * A number of properties previously only settable in unit
6 files are now also available as properties to set when
7 creating transient units programmatically via the bus, as it
8 is exposed with systemd-run's --property=
9 setting. Specifically, these are: SyslogIdentifier=,
10 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
11 EnvironmentFile=, ReadWriteDirectories=,
12 ReadOnlyDirectories=, InaccessibleDirectories=,
13 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
15 * When creating transient services via the bus API it is now
16 possible to pass in a set of file descriptors to use as
17 STDIN/STDOUT/STDERR for the invoked process.
19 * Slice units may now be created transiently via the bus APIs,
20 similar to the way service and scope units may already be
23 * Wherever systemd expects a calendar timestamp specification
24 (like in journalctl's --since= and --until= switches) UTC
25 timestamps are now supported. Timestamps suffixed with "UTC"
26 are now considered to be in Universal Time Coordinated
27 instead of the local timezone. Also, timestamps may now
28 optionally be specified with sub-second accuracy. Both of
29 these additions also apply to recurring calendar event
30 specification, such as OnCalendar= in timer units.
32 * journalctl gained a new "--sync" switch that asks the
33 journal daemon to write all so far unwritten log messages to
34 disk and sync the files, before returning.
36 * systemd-tmpfiles learned two new line types "q" and "Q" that
37 operate like "v", but also set up a basic btrfs quota
38 hierarchy when used on a btrfs file system with quota
41 * tmpfiles' "v", "q" and "Q" will now create a plain directory
42 instead of a subvolume (even on a btrfs file system) if the
43 root directory is a plain directory, and not a
44 subvolume. This should simplify things with certain chroot()
45 environments which are not aware of the concept of btrfs
48 * systemd-detect-virt gained a new --chroot switch to detect
49 whether execution takes place in a chroot() environment.
51 * CPUAffinity= now takes CPU index ranges in addition to
54 * The various memory-related resource limit settings (such as
55 LimitAS=) now understand the usual K, M, G, ... suffixes to
56 the base of 1024 (IEC). Similar, the time-related resource
57 limit settings understand the usual min, h, day, ...
60 * There's a new system.conf setting DefaultTasksMax= to
61 control the default TasksMax= setting for services and
62 scopes running on the system. (TasksMax= is the primary
63 setting that exposes the "pids" cgroup controller on systemd
64 and was introduced in the previous systemd release.) The
65 setting now defaults to 512, which means services that are
66 not explicitly configured otherwise will only be able to
67 create 512 processes or threads at maximum, from this
68 version on. Note that this means that thread- or
69 process-heavy services might need to be reconfigured to set
70 TasksMax= to a higher value. It is sufficient to set
71 TasksMax= in these specific unit files to a higher value, or
72 even "infinity". Similar, there's now a logind.conf setting
73 UserTasksMax= that defaults to 4096 and limits the total
74 number of processes or tasks each user may own
75 concurrently. nspawn containers also have the TasksMax=
76 value set by default now, to 8192. Note that all of this
77 only has an effect if the "pids" cgroup controller is
78 enabled in the kernel. The general benefit of these changes
79 should be a more robust and safer system, that provides a
80 certain amount of per-service fork() bomb protection.
82 * systemd-nspawn gained the new --network-veth-extra= switch
83 to define additional and arbitrarily-named virtual Ethernet
84 links between the host and the container.
86 * A new service execution setting PassEnvironment= has been
87 added that allows importing select environment variables
88 from PID1's environment block into the environment block of
91 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
92 default now (the kernel default is 16). This is beneficial
93 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
94 allows substantially larger numbers of queued
95 datagrams. This should increase the capability of systemd to
96 parallelize boot-up, as logging and sd_notify() are unlikely
97 to stall execution anymore. If you need to change the value
98 from the new defaults, use the usual sysctl.d/ snippets.
100 * The compression framing format used by the journal or
101 coredump processing has changed to be in line with what the
102 official LZ4 tools generate. LZ4 compression support in
103 systemd was considered unsupported previously, as the format
104 was not compatible with the normal tools. With this release
105 this has changed now, and it is hence safe for downstream
106 distributions to turn it on. While not compressing as well
107 as the XZ, LZ4 is substantially faster, which makes
108 it a good default choice for the compression logic in the
109 journal and in coredump handling.
111 * Any reference to /etc/mtab has been dropped from
112 systemd. The file has been obsolete since a while, but
113 systemd refused to work on systems where it was incorrectly
114 set up (it should be a symlink or non-existent). Please make
115 sure to update to util-linux 2.27.1 or newer in conjunction
116 with this systemd release, which also drops any reference to
117 /etc/mtab. If you maintain a distribution make sure that no
118 software you package still references it, as this is a
119 likely source of bugs. There's also a glibc bug pending,
120 asking for removal of any reference to this obsolete file:
122 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
124 * Support for the ".snapshot" unit type has been removed. This
125 feature turned out to be little useful and little used, and
126 has now been removed from the core and from systemctl.
128 * The dependency types RequiresOverridable= and
129 RequisiteOverridable= have been removed from systemd. They
130 have been used only very sparingly to our knowledge and
131 other options that provide a similar effect (such as
132 systemctl --mode=ignore-dependencies) are much more useful
133 and commonly used. Moreover, they were only half-way
134 implemented as the option to control behaviour regarding
135 these dependencies was never added to systemctl. By removing
136 these dependency types the execution engine becomes a bit
137 simpler. Unit files that use these dependencies should be
138 changed to use the non-Overridable dependency types
139 instead. In fact, when parsing unit files with these
140 options, that's what systemd will automatically convert them
141 too, but it will also warn, asking users to fix the unit
142 files accordingly. Removal of these dependency types should
143 only affect a negligible number of unit files in the wild.
145 * Behaviour of networkd's IPForward= option changed
146 (again). It will no longer maintain a per-interface setting,
147 but propagate one way from interfaces where this is enabled
148 to the global kernel setting. The global setting will be
149 enabled when requested by a network that is set up, but
150 never be disabled again. This change was made to make sure
151 IPv4 and IPv6 behaviour regarding packet forwarding is
152 similar (as the Linux IPv6 stack does not support
153 per-interface control of this setting) and to minimize
156 * In unit files the behaviour of %u, %U, %h, %s has
157 changed. These specifiers will now unconditionally resolve
158 to the various user database fields of the user that the
159 systemd instance is running as, instead of the user
160 configured in the specific unit via User=. Note that this
161 effectively doesn't change much, as resolving of these
162 specifiers was already turned off in the --system instance
163 of systemd, as we cannot do NSS lookups from PID 1. In the
164 --user instance of systemd these specifiers where correctly
165 resolved, but hardly made any sense, since the user instance
166 lacks privileges to do user switches anyway, and User= is
167 hence useless. Morever, even in the --user instance of
168 systemd behaviour was awkward as it would only take settings
169 from User= assignment placed before the specifier into
170 account. In order to unify and simplify the logic around
171 this the specifiers will now always resolve to the
172 credentials of the user invoking the manager (which in case
173 of PID 1 is the root user).
175 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
176 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
177 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
178 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
179 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
180 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
181 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
182 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
183 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
184 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
185 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
186 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
187 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
188 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
191 -- Berlin, 2015-11-18
195 * systemd now depends on util-linux v2.27. More specifically,
196 the newly added mount monitor feature in libmount now
197 replaces systemd's former own implementation.
199 * libmount mandates /etc/mtab not to be regular file, and
200 systemd now enforces this condition at early boot.
201 /etc/mtab has been deprecated and warned about for a very
202 long time, so systems running systemd should already have
203 stopped having this file around as anything else than a
204 symlink to /proc/self/mounts.
206 * Support for the "pids" cgroup controller has been added. It
207 allows accounting the number of tasks in a cgroup and
208 enforcing limits on it. This adds two new setting
209 TasksAccounting= and TasksMax= to each unit, as well as a
210 global option DefaultTasksAccounting=.
212 * Support for the "net_cls" cgroup controller has been added.
213 It allows assigning a net class ID to each task in the
214 cgroup, which can then be used in firewall rules and traffic
215 shaping configurations. Note that the kernel netfilter net
216 class code does not currently work reliably for ingress
217 packets on unestablished sockets.
219 This adds a new config directive called NetClass= to CGroup
220 enabled units. Allowed values are positive numbers for fixed
221 assignments and "auto" for picking a free value
224 * 'systemctl is-system-running' now returns 'offline' if the
225 system is not booted with systemd. This command can now be
226 used as a substitute for 'systemd-notify --booted'.
228 * Watchdog timeouts have been increased to 3 minutes for all
229 in-tree service files. Apparently, disk IO issues are more
230 frequent than we hoped, and user reported >1 minute waiting
233 * 'machine-id-commit' functionality has been merged into
234 'machine-id-setup --commit'. The separate binary has been
237 * The WorkingDirectory= directive in unit files may now be set
238 to the special value '~'. In this case, the working
239 directory is set to the home directory of the user
242 * "machinectl shell" will now open the shell in the home
243 directory of the selected user by default.
245 * The CrashChVT= configuration file setting is renamed to
246 CrashChangeVT=, following our usual logic of not
247 abbreviating unnecessarily. The old directive is still
248 supported for compat reasons. Also, this directive now takes
249 an integer value between 1 and 63, or a boolean value. The
250 formerly supported '-1' value for disabling stays around for
253 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
254 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
255 RootDirectory= properties can now be set for transient
258 * The systemd-analyze tool gained a new "set-log-target" verb
259 to change the logging target the system manager logs to
260 dynamically during runtime. This is similar to how
261 "systemd-analyze set-log-level" already changes the log
264 * In nspawn /sys is now mounted as tmpfs, with only a selected
265 set of subdirectories mounted in from the real sysfs. This
266 enhances security slightly, and is useful for ensuring user
267 namespaces work correctly.
269 * Support for USB FunctionFS activation has been added. This
270 allows implementation of USB gadget services that are
271 activated as soon as they are requested, so that they don't
272 have to run continously, similar to classic socket
275 * The "systemctl exit" command now optionally takes an
276 additional parameter that sets the exit code to return from
277 the systemd manager when exiting. This is only relevant when
278 running the systemd user instance, or when running the
279 system instance in a container.
281 * sd-bus gained the new API calls sd_bus_path_encode_many()
282 and sd_bus_path_decode_many() that allow easy encoding and
283 decoding of multiple identifier strings inside a D-Bus
284 object path. Another new call sd_bus_default_flush_close()
285 has been added to flush and close per-thread default
288 * systemd-cgtop gained support for a -M/--machine= switch to
289 show the control groups within a certain container only.
291 * "systemctl kill" gained support for an optional --fail
292 switch. If specified the requested operation will fail of no
293 processes have been killed, because the unit had no
294 processes attached, or similar.
296 * A new systemd.crash_reboot=1 kernel command line option has
297 been added that triggers a reboot after crashing. This can
298 also be set through CrashReboot= in systemd.conf.
300 * The RuntimeDirectory= setting now understands unit
301 specifiers like %i or %f.
303 * A new (still internal) libary API sd-ipv4acd has been added,
304 that implements address conflict detection for IPv4. It's
305 based on code from sd-ipv4ll, and will be useful for
306 detecting DHCP address conflicts.
308 * File descriptors passed during socket activation may now be
309 named. A new API sd_listen_fds_with_names() is added to
310 access the names. The default names may be overridden,
311 either in the .socket file using the FileDescriptorName=
312 parameter, or by passing FDNAME= when storing the file
313 descriptors using sd_notify().
315 * systemd-networkd gained support for:
317 - Setting the IPv6 Router Advertisment settings via
318 IPv6AcceptRouterAdvertisements= in .network files.
320 - Configuring the HelloTimeSec=, MaxAgeSec= and
321 ForwardDelaySec= bridge parameters in .netdev files.
323 - Configuring PreferredSource= for static routes in
326 * The "ask-password" framework used to query for LUKS harddisk
327 passwords or SSL passwords during boot gained support for
328 caching passwords in the kernel keyring, if it is
329 available. This makes sure that the user only has to type in
330 a passphrase once if there are multiple objects to unlock
331 with the same one. Previously, such password caching was
332 available only when Plymouth was used; this moves the
333 caching logic into the systemd codebase itself. The
334 "systemd-ask-password" utility gained a new --keyname=
335 switch to control which kernel keyring key to use for
336 caching a password in. This functionality is also useful for
337 enabling display managers such as gdm to automatically
338 unlock the user's GNOME keyring if its passphrase, the
339 user's password and the harddisk password are the same, if
340 gdm-autologin is used.
342 * When downloading tar or raw images using "machinectl
343 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
344 file is now also downloaded, if it is available and stored
345 next to the image file.
347 * Units of type ".socket" gained a new boolean setting
348 Writable= which is only useful in conjunction with
349 ListenSpecial=. If true, enables opening the specified
350 special file in O_RDWR mode rather than O_RDONLY mode.
352 * systemd-rfkill has been reworked to become a singleton
353 service that is activated through /dev/rfkill on each rfkill
354 state change and saves the settings to disk. This way,
355 systemd-rfkill is now compatible with devices that exist
356 only intermittendly, and even restores state if the previous
357 system shutdown was abrupt rather than clean.
359 * The journal daemon gained support for vacuuming old journal
360 files controlled by the number of files that shall remain,
361 in addition to the already existing control by size and by
362 date. This is useful as journal interleaving performance
363 degrades with too many seperate journal files, and allows
364 putting an effective limit on them. The new setting defaults
365 to 100, but this may be changed by setting SystemMaxFiles=
366 and RuntimeMaxFiles= in journald.conf. Also, the
367 "journalctl" tool gained the new --vacuum-files= switch to
368 manually vacuum journal files to leave only the specified
369 number of files in place.
371 * udev will now create /dev/disk/by-path links for ATA devices
372 on kernels where that is supported.
374 * Galician, Serbian, Turkish and Korean translations were added.
376 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
377 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
378 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
379 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
380 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
381 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
382 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
383 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
384 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
385 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
386 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
387 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
388 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
389 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
390 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
391 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
392 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
393 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
395 -- Berlin, 2015-10-07
399 * The DHCP implementation of systemd-networkd gained a set of
402 - The DHCP server now supports emitting DNS and NTP
403 information. It may be enabled and configured via
404 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
405 and NTP information is enabled, but no servers are
406 configured, the corresponding uplink information (if there
407 is any) is propagated.
409 - Server and client now support transmission and reception
410 of timezone information. It can be configured via the
411 newly introduced network options UseTimezone=,
412 EmitTimezone=, and Timezone=. Transmission of timezone
413 information is enabled between host and containers by
414 default now: the container will change its local timezone
415 to what the host has set.
417 - Lease timeouts can now be configured via
418 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
420 - The DHCP server improved on the stability of
421 leases. Clients are more likely to get the same lease
422 information back, even if the server loses state.
424 - The DHCP server supports two new configuration options to
425 control the lease address pool metrics, PoolOffset= and
428 * The encapsulation limit of tunnels in systemd-networkd may
429 now be configured via 'EncapsulationLimit='. It allows
430 modifying the maximum additional levels of encapsulation
431 that are permitted to be prepended to a packet.
433 * systemd now supports the concept of user buses replacing
434 session buses, if used with dbus-1.10 (and enabled via dbus
435 --enable-user-session). It previously only supported this on
436 kdbus-enabled systems, and this release expands this to
437 'dbus-daemon' systems.
439 * systemd-networkd now supports predictable interface names
442 * systemd now optionally supports the new Linux kernel
443 "unified" control group hierarchy. If enabled via the kernel
444 command-line option 'systemd.unified_cgroup_hierarchy=1',
445 systemd will try to mount the unified cgroup hierarchy
446 directly on /sys/fs/cgroup. If not enabled, or not
447 available, systemd will fall back to the legacy cgroup
448 hierarchy setup, as before. Host system and containers can
449 mix and match legacy and unified hierarchies as they
450 wish. nspawn understands the $UNIFIED_CROUP_HIERARCHY
451 environment variable to individually select the hierarchy to
452 use for executed containers. By default, nspawn will use the
453 unified hierarchy for the containers if the host uses the
454 unified hierarchy, and the legacy hierarchy otherwise.
455 Please note that at this point the unified hierarchy is an
456 experimental kernel feature and is likely to change in one
457 of the next kernel releases. Therefore, it should not be
458 enabled by default in downstream distributions yet. The
459 minimum required kernel version for the unified hierarchy to
460 work is 4.2. Note that when the unified hierarchy is used
461 for the first time delegated access to controllers is
462 safe. Because of this systemd-nspawn containers will get
463 access to controllers now, as will systemd user
464 sessions. This means containers and user sessions may now
465 manage their own resources, partitioning up what the system
468 * A new special scope unit "init.scope" has been introduced
469 that encapsulates PID 1 of the system. It may be used to
470 determine resource usage and enforce resource limits on PID
471 1 itself. PID 1 hence moved out of the root of the control
474 * The cgtop tool gained support for filtering out kernel
475 threads when counting tasks in a control group. Also, the
476 count of processes is now recursively summed up by
477 default. Two options -k and --recursive= have been added to
478 revert to old behaviour. The tool has also been updated to
479 work correctly in containers now.
481 * systemd-nspawn's --bind= and --bind-ro= options have been
482 extended to allow creation of non-recursive bind mounts.
484 * libsystemd gained two new calls sd_pid_get_cgroup() and
485 sd_peer_get_cgroup() which return the control group path of
486 a process or peer of a connected AF_UNIX socket. This
487 function call is particularly useful when implementing
488 delegated subtrees support in the control group hierarchy.
490 * The "sd-event" event loop API of libsystemd now supports
491 correct dequeuing of real-time signals, without losing
494 * When systemd requests a PolicyKit decision when managing
495 units it will now add additional fields to the request,
496 including unit name and desired operation. This enables more
497 powerful PolicyKit policies, that make decisions depending
500 * nspawn learnt support for .nspawn settings files, that may
501 accompany the image files or directories of containers, and
502 may contain additional settings for the container. This is
503 an alternative to configuring container parameters via the
506 Contributions from: Cristian Rodríguez, Daniel Mack, David
507 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
508 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
509 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
510 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
511 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
512 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
513 Andersen, Tom Gundersen, Torstein Husebø
515 -- Berlin, 2015-09-08
519 * machinectl gained a new verb 'shell' which opens a fresh
520 shell on the target container or the host. It is similar to
521 the existing 'login' command of machinectl, but spawns the
522 shell directly without prompting for username or
523 password. The pseudo machine '.host' now refers to the local
524 host and is used by default. Hence, 'machinectl shell' can
525 be used as replacement for 'su -' which spawns a session as
526 a fresh systemd unit in a way that is fully isolated from
527 the originating session.
529 * systemd-networkd learned to cope with private-zone DHCP
530 options and allows other programs to query the values.
532 * SELinux access control when enabling/disabling units is no
533 longer enforced with this release. The previous
534 implementation was incorrect, and a new corrected
535 implementation is not yet available. As unit file operations
536 are still protected via PolicyKit and D-Bus policy this is
537 not a security problem. Yet, distributions which care about
538 optimal SELinux support should probably not stabilize on
541 * sd-bus gained support for matches of type "arg0has=", that
542 test for membership of strings in string arrays sent in bus
545 * systemd-resolved now dumps the contents of its DNS and LLMNR
546 caches to the logs on reception of the SIGUSR1 signal. This
547 is useful to debug DNS behaviour.
549 * The coredumpctl tool gained a new --directory= option to
550 operate on journal files in a specific directory.
552 * "systemctl reboot" and related commands gained a new
553 "--message=" option which may be used to set a free-text
554 wall message when shutting down or rebooting the
555 system. This message is also logged, which is useful for
556 figuring out the reason for a reboot or shutdown a
559 * The "systemd-resolve-host" tool's -i switch now takes
560 network interface numbers as alternative to interface names.
562 * A new unit file setting for services has been introduced:
563 UtmpMode= allows configuration of how precisely systemd
564 handles utmp and wtmp entries for the service if this is
565 enabled. This allows writing services that appear similar to
566 user sessions in the output of the "w", "who", "last" and
569 * systemd-resolved will now locally synthesize DNS resource
570 records for the "localhost" and "gateway" domains as well as
571 the local hostname. This should ensure that clients querying
572 RRs via resolved will get similar results as those going via
573 NSS, if nss-myhostname is enabled.
575 Contributions from: Alastair Hughes, Alex Crawford, Daniel
576 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
577 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
578 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
579 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
580 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
581 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
582 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
583 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
584 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
585 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
586 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
588 -- Berlin, 2015-08-27
592 * The systemd-efi-boot-generator functionality was merged into
593 systemd-gpt-auto-generator.
595 * systemd-networkd now supports Group Policy for vxlan
596 devices. It can be enabled via the new boolean configuration
597 option called 'GroupPolicyExtension='.
599 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
600 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
601 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
603 -- Berlin, 2015-07-31
607 * The python-systemd code has been removed from the systemd repository.
608 A new repository has been created which accommodates the code from
609 now on, and we kindly ask distributions to create a separate package
610 for this: https://github.com/systemd/python-systemd
612 * The systemd daemon will now reload its main configuration
613 (/etc/systemd/system.conf) on daemon-reload.
615 * sd-dhcp now exposes vendor specific extensions via
616 sd_dhcp_lease_get_vendor_specific().
618 * systemd-networkd gained a number of new configuration options.
620 - A new boolean configuration option for TAP devices called
621 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
622 device, thus allowing to send and receive GSO packets.
624 - A new tunnel configuration option called 'CopyDSCP='.
625 If enabled, the DSCP field of ip6 tunnels is copied into the
628 - A set of boolean bridge configuration options were added.
629 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
630 and 'UnicastFlood=' are now parsed by networkd and applied to the
631 respective bridge link device via the respective IFLA_BRPORT_*
634 - A new string configuration option to override the hostname sent
635 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
636 is true, networkd will use the configured hostname instead of the
637 system hostname when sending DHCP requests.
639 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
640 networkd will configure the IPv6 flow-label of the tunnel device
641 according to RFC2460.
643 - The 'macvtap' virtual network devices are now supported, similar to
644 the already supported 'macvlan' devices.
646 * systemd-resolved now implements RFC5452 to improve resilience against
647 cache poisoning. Additionally, source port randomization is enabled
648 by default to further protect against DNS spoofing attacks.
650 * nss-mymachines now supports translating UIDs and GIDs of running
651 containers with user-namespaces enabled. If a container 'foo'
652 translates a host uid 'UID' to the container uid 'TUID', then
653 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
654 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
655 mapped as 'vg-foo-TGID'.
657 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
658 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
659 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
660 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
661 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
662 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
663 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
664 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
665 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
666 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
668 -- Berlin, 2015-07-29
672 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
673 There are no known issues with current sysfs, and udev does not need
674 or should be used to work around such bugs.
676 * udev does no longer enable USB HID power management. Several reports
677 indicate, that some devices cannot handle that setting.
679 * The udev accelerometer helper was removed. The functionality
680 is now fully included in iio-sensor-proxy. But this means,
681 older iio-sensor-proxy versions will no longer provide
682 accelerometer/orientation data with this systemd version.
683 Please upgrade iio-sensor-proxy to version 1.0.
685 * networkd gained a new configuration option IPv6PrivacyExtensions=
686 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
687 for Stateless Address") on selected networks.
689 * For the sake of fewer build-time dependencies and less code in the
690 main repository, the python bindings are about to be removed in the
691 next release. A new repository has been created which accommodates
692 the code from now on, and we kindly ask distributions to create a
693 separate package for this. The removal will take place in v223.
695 https://github.com/systemd/python-systemd
697 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
698 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
699 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
700 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
701 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
702 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
703 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
704 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
705 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
706 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
708 -- Berlin, 2015-07-07
712 * The sd-bus.h and sd-event.h APIs have now been declared
713 stable and have been added to the official interface of
714 libsystemd.so. sd-bus implements an alternative D-Bus client
715 library, that is relatively easy to use, very efficient and
716 supports both classic D-Bus as well as kdbus as transport
717 backend. sd-event is a generic event loop abstraction that
718 is built around Linux epoll, but adds features such as event
719 prioritization or efficient timer handling. Both APIs are good
720 choices for C programs looking for a bus and/or event loop
721 implementation that is minimal and does not have to be
722 portable to other kernels.
724 * kdbus support is no longer compile-time optional. It is now
725 always built-in. However, it can still be disabled at
726 runtime using the kdbus=0 kernel command line setting, and
727 that setting may be changed to default to off, by specifying
728 --disable-kdbus at build-time. Note though that the kernel
729 command line setting has no effect if the kdbus.ko kernel
730 module is not installed, in which case kdbus is (obviously)
731 also disabled. We encourage all downstream distributions to
732 begin testing kdbus by adding it to the kernel images in the
733 development distributions, and leaving kdbus support in
736 * The minimal required util-linux version has been bumped to
739 * Support for chkconfig (--enable-chkconfig) was removed in
740 favor of calling an abstraction tool
741 /lib/systemd/systemd-sysv-install. This needs to be
742 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
743 in README for details.
745 * If there's a systemd unit and a SysV init script for the
746 same service name, and the user executes "systemctl enable"
747 for it (or a related call), then this will now enable both
748 (or execute the related operation on both), not just the
751 * The libudev API documentation has been converted from gtkdoc
754 * gudev has been removed from the systemd tree, it is now an
757 * The systemd-cgtop tool learnt a new --raw switch to generate
758 "raw" (machine parsable) output.
760 * networkd's IPForwarding= .network file setting learnt the
761 new setting "kernel", which ensures that networkd does not
762 change the IP forwarding sysctl from the default kernel
765 * The systemd-logind bus API now exposes a new boolean
766 property "Docked" that reports whether logind considers the
767 system "docked", i.e. connected to a docking station or not.
769 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
770 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
771 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
772 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
773 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
774 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
775 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
776 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
777 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
778 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
779 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
780 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
781 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
782 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
783 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
784 Fink, Zbigniew Jędrzejewski-Szmek
786 -- Berlin, 2015-06-19
790 * The gudev library has been extracted into a separate repository
791 available at: https://git.gnome.org/browse/libgudev/
792 It is now managed as part of the Gnome project. Distributions
793 are recommended to pass --disable-gudev to systemd and use
794 gudev from the Gnome project instead. gudev is still included
795 in systemd, for now. It will be removed soon, though. Please
796 also see the announcement-thread on systemd-devel:
797 http://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
799 * systemd now exposes a CPUUsageNSec= property for each
800 service unit on the bus, that contains the overall consumed
801 CPU time of a service (the sum of what each process of the
802 service consumed). This value is only available if
803 CPUAccounting= is turned on for a service, and is then shown
804 in the "systemctl status" output.
806 * Support for configuring alternative mappings of the old SysV
807 runlevels to systemd targets has been removed. They are now
808 hardcoded in a way that runlevels 2, 3, 4 all map to
809 multi-user.target and 5 to graphical.target (which
810 previously was already the default behaviour).
812 * The auto-mounter logic gained support for mount point
813 expiry, using a new TimeoutIdleSec= setting in .automount
814 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
816 * The EFI System Partition (ESP) as mounted to /boot by
817 systemd-efi-boot-generator will now be unmounted
818 automatically after 2 minutes of not being used. This should
819 minimize the risk of ESP corruptions.
821 * New /etc/fstab options x-systemd.requires= and
822 x-systemd.requires-mounts-for= are now supported to express
823 additional dependencies for mounts. This is useful for
824 journalling file systems that support external journal
825 devices or overlay file systems that require underlying file
826 systems to be mounted.
828 * systemd does not support direct live-upgrades (via systemctl
829 daemon-reexec) from versions older than v44 anymore. As no
830 distribution we are aware of shipped such old versions in a
831 stable release this should not be problematic.
833 * When systemd forks off a new per-connection service instance
834 it will now set the $REMOTE_ADDR environment variable to the
835 remote IP address, and $REMOTE_PORT environment variable to
836 the remote IP port. This behaviour is similar to the
837 corresponding environment variables defined by CGI.
839 * systemd-networkd gained support for uplink failure
840 detection. The BindCarrier= option allows binding interface
841 configuration dynamically to the link sense of other
842 interfaces. This is useful to achieve behaviour like in
845 * systemd-networkd gained support for configuring the DHCP
846 client identifier to use when requesting leases.
848 * systemd-networkd now has a per-network UseNTP= option to
849 configure whether NTP server information acquired via DHCP
850 is passed on to services like systemd-timesyncd.
852 * systemd-networkd gained support for vti6 tunnels.
854 * Note that systemd-networkd manages the sysctl variable
855 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
856 it is configured for since v219. The variable controls IP
857 forwarding, and is a per-interface alternative to the global
858 /proc/sys/net/ipv[46]/ip_forward. This setting is
859 configurable in the IPForward= option, which defaults to
860 "no". This means if networkd is used for an interface it is
861 no longer sufficient to set the global sysctl option to turn
862 on IP forwarding! Instead, the .network file option
863 IPForward= needs to be turned on! Note that the
864 implementation of this behaviour was broken in v219 and has
867 * Many bonding and vxlan options are now configurable in
870 * systemd-nspawn gained a new --property= setting to set unit
871 properties for the container scope. This is useful for
872 setting resource parameters (e.g "CPUShares=500") on
873 containers started from the command line.
875 * systemd-nspawn gained a new --private-users= switch to make
876 use of user namespacing available on recent Linux kernels.
878 * systemd-nspawn may now be called as part of a shell pipeline
879 in which case the pipes used for stdin and stdout are passed
880 directly to the process invoked in the container, without
881 indirection via a pseudo tty.
883 * systemd-nspawn gained a new switch to control the UNIX
884 signal to use when killing the init process of the container
887 * systemd-nspawn gained a new --overlay= switch for mounting
888 overlay file systems into the container using the new kernel
891 * When a container image is imported via systemd-importd and
892 the host file system is not btrfs, a loopback block device
893 file is created in /var/lib/machines.raw with a btrfs file
894 system inside. It is then mounted to /var/lib/machines to
895 enable btrfs features for container management. The loopback
896 file and btrfs file system is grown as needed when container
897 images are imported via systemd-importd.
899 * systemd-machined/systemd-importd gained support for btrfs
900 quota, to enforce container disk space limits on disk. This
901 is exposed in "machinectl set-limit".
903 * systemd-importd now can import containers from local .tar,
904 .raw and .qcow2 images, and export them to .tar and .raw. It
905 can also import dkr v2 images now from the network (on top
908 * systemd-importd gained support for verifying downloaded
909 images with gpg2 (previously only gpg1 was supported).
911 * systemd-machined, systemd-logind, systemd: most bus calls
912 are now accessible to unprivileged processes via
913 PolicyKit. Also, systemd-logind will now allow users to kill
914 their own sessions without further privileges or
917 * systemd-shutdownd has been removed. This service was
918 previously responsible for implementing scheduled shutdowns
919 as exposed in /usr/bin/shutdown's time parameter. This
920 functionality has now been moved into systemd-logind and is
921 accessible via a bus interface.
923 * "systemctl reboot" gained a new switch --firmware-setup that
924 can be used to reboot into the EFI firmware setup, if that
925 is available. systemd-logind now exposes an API on the bus
926 to trigger such reboots, in case graphical desktop UIs want
927 to cover this functionality.
929 * "systemctl enable", "systemctl disable" and "systemctl mask"
930 now support a new "--now" switch. If specified the units
931 that are enabled will also be started, and the ones
932 disabled/masked also stopped.
934 * The Gummiboot EFI boot loader tool has been merged into
935 systemd, and renamed to "systemd-boot". The bootctl tool has been
936 updated to support systemd-boot.
938 * An EFI kernel stub has been added that may be used to create
939 kernel EFI binaries that contain not only the actual kernel,
940 but also an initrd, boot splash, command line and OS release
941 information. This combined binary can then be signed as a
942 single image, so that the firmware can verify it all in one
943 step. systemd-boot has special support for EFI binaries created
944 like this and can extract OS release information from them
945 and show them in the boot menu. This functionality is useful
946 to implement cryptographically verified boot schemes.
948 * Optional support has been added to systemd-fsck to pass
949 fsck's progress report to an AF_UNIX socket in the file
952 * udev will no longer create device symlinks for all block
953 devices by default. A blacklist for excluding special block
954 devices from this logic has been turned into a whitelist
955 that requires picking block devices explicitly that require
958 * A new (currently still internal) API sd-device.h has been
959 added to libsystemd. This modernized API is supposed to
960 replace libudev eventually. In fact, already much of libudev
961 is now just a wrapper around sd-device.h.
963 * A new hwdb database for storing metadata about pointing
964 stick devices has been added.
966 * systemd-tmpfiles gained support for setting file attributes
967 similar to the "chattr" tool with new 'h' and 'H' lines.
969 * systemd-journald will no longer unconditionally set the
970 btrfs NOCOW flag on new journal files. This is instead done
971 with tmpfiles snippet using the new 'h' line type. This
972 allows easy disabling of this logic, by masking the
973 journal-nocow.conf tmpfiles file.
975 * systemd-journald will now translate audit message types to
976 human readable identifiers when writing them to the
977 journal. This should improve readability of audit messages.
979 * The LUKS logic gained support for the offset= and skip=
980 options in /etc/crypttab, as previously implemented by
983 * /usr/lib/os-release gained a new optional field VARIANT= for
984 distributions that support multiple variants (such as a
985 desktop edition, a server edition, ...)
987 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
988 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
989 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
990 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
991 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
992 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
993 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
994 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
995 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
996 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
997 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
998 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
999 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
1000 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
1001 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
1002 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
1003 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
1004 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
1005 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
1006 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
1007 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
1008 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
1009 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
1010 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
1011 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
1012 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
1013 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
1015 -- Berlin, 2015-05-22
1019 * Introduce a new API "sd-hwdb.h" for querying the hardware
1020 metadata database. With this minimal interface one can query
1021 and enumerate the udev hwdb, decoupled from the old libudev
1022 library. libudev's interface for this is now only a wrapper
1023 around sd-hwdb. A new tool systemd-hwdb has been added to
1024 interface with and update the database.
1026 * When any of systemd's tools copies files (for example due to
1027 tmpfiles' C lines) a btrfs reflink will attempted first,
1028 before bytewise copying is done.
1030 * systemd-nspawn gained a new --ephemeral switch. When
1031 specified a btrfs snapshot is taken of the container's root
1032 directory, and immediately removed when the container
1033 terminates again. Thus, a container can be started whose
1034 changes never alter the container's root directory, and are
1035 lost on container termination. This switch can also be used
1036 for starting a container off the root file system of the
1037 host without affecting the host OS. This switch is only
1038 available on btrfs file systems.
1040 * systemd-nspawn gained a new --template= switch. It takes the
1041 path to a container tree to use as template for the tree
1042 specified via --directory=, should that directory be
1043 missing. This allows instantiating containers dynamically,
1044 on first run. This switch is only available on btrfs file
1047 * When a .mount unit refers to a mount point on which multiple
1048 mounts are stacked, and the .mount unit is stopped all of
1049 the stacked mount points will now be unmounted until no
1050 mount point remains.
1052 * systemd now has an explicit notion of supported and
1053 unsupported unit types. Jobs enqueued for unsupported unit
1054 types will now fail with an "unsupported" error code. More
1055 specifically .swap, .automount and .device units are not
1056 supported in containers, .busname units are not supported on
1057 non-kdbus systems. .swap and .automount are also not
1058 supported if their respective kernel compile time options
1061 * machinectl gained support for two new "copy-from" and
1062 "copy-to" commands for copying files from a running
1063 container to the host or vice versa.
1065 * machinectl gained support for a new "bind" command to bind
1066 mount host directories into local containers. This is
1067 currently only supported for nspawn containers.
1069 * networkd gained support for configuring bridge forwarding
1070 database entries (fdb) from .network files.
1072 * A new tiny daemon "systemd-importd" has been added that can
1073 download container images in tar, raw, qcow2 or dkr formats,
1074 and make them available locally in /var/lib/machines, so
1075 that they can run as nspawn containers. The daemon can GPG
1076 verify the downloads (not supported for dkr, since it has no
1077 provisions for verifying downloads). It will transparently
1078 decompress bz2, xz, gzip compressed downloads if necessary,
1079 and restore sparse files on disk. The daemon uses privilege
1080 separation to ensure the actual download logic runs with
1081 fewer privileges than the daemon itself. machinectl has
1082 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
1083 make the functionality of importd available to the
1084 user. With this in place the Fedora and Ubuntu "Cloud"
1085 images can be downloaded and booted as containers unmodified
1086 (the Fedora images lack the appropriate GPG signature files
1087 currently, so they cannot be verified, but this will change
1088 soon, hopefully). Note that downloading images is currently
1089 only fully supported on btrfs.
1091 * machinectl is now able to list container images found in
1092 /var/lib/machines, along with some metadata about sizes of
1093 disk and similar. If the directory is located on btrfs and
1094 quota is enabled, this includes quota display. A new command
1095 "image-status" has been added that shows additional
1096 information about images.
1098 * machinectl is now able to clone container images
1099 efficiently, if the underlying file system (btrfs) supports
1100 it, with the new "machinectl list-images" command. It also
1101 gained commands for renaming and removing images, as well as
1102 marking them read-only or read-write (supported also on
1103 legacy file systems).
1105 * networkd gained support for collecting LLDP network
1106 announcements, from hardware that supports this. This is
1107 shown in networkctl output.
1109 * systemd-run gained support for a new -t (--pty) switch for
1110 invoking a binary on a pty whose input and output is
1111 connected to the invoking terminal. This allows executing
1112 processes as system services while interactively
1113 communicating with them via the terminal. Most interestingly
1114 this is supported across container boundaries. Invoking
1115 "systemd-run -t /bin/bash" is an alternative to running a
1116 full login session, the difference being that the former
1117 will not register a session, nor go through the PAM session
1120 * tmpfiles gained support for a new "v" line type for creating
1121 btrfs subvolumes. If the underlying file system is a legacy
1122 file system, this automatically degrades to creating a
1123 normal directory. Among others /var/lib/machines is now
1124 created like this at boot, should it be missing.
1126 * The directory /var/lib/containers/ has been deprecated and
1127 been replaced by /var/lib/machines. The term "machines" has
1128 been used in the systemd context as generic term for both
1129 VMs and containers, and hence appears more appropriate for
1130 this, as the directory can also contain raw images bootable
1133 * systemd-nspawn when invoked with -M but without --directory=
1134 or --image= is now capable of searching for the container
1135 root directory, subvolume or disk image automatically, in
1136 /var/lib/machines. systemd-nspawn@.service has been updated
1137 to make use of this, thus allowing it to be used for raw
1140 * A new machines.target unit has been introduced that is
1141 supposed to group all containers/VMs invoked as services on
1142 the system. systemd-nspawn@.service has been updated to
1143 integrate with that.
1145 * machinectl gained a new "start" command, for invoking a
1146 container as a service. "machinectl start foo" is mostly
1147 equivalent to "systemctl start systemd-nspawn@foo.service",
1148 but handles escaping in a nicer way.
1150 * systemd-nspawn will now mount most of the cgroupfs tree
1151 read-only into each container, with the exception of the
1152 container's own subtree in the name=systemd hierarchy.
1154 * journald now sets the special FS_NOCOW file flag for its
1155 journal files. This should improve performance on btrfs, by
1156 avoiding heavy fragmentation when journald's write-pattern
1157 is used on COW file systems. It degrades btrfs' data
1158 integrity guarantees for the files to the same levels as for
1159 ext3/ext4 however. This should be OK though as journald does
1160 its own data integrity checks and all its objects are
1161 checksummed on disk. Also, journald should handle btrfs disk
1162 full events a lot more gracefully now, by processing SIGBUS
1163 errors, and not relying on fallocate() anymore.
1165 * When journald detects that journal files it is writing to
1166 have been deleted it will immediately start new journal
1169 * systemd now provides a way to store file descriptors
1170 per-service in PID 1.This is useful for daemons to ensure
1171 that fds they require are not lost during a daemon
1172 restart. The fds are passed to the daemon on the next
1173 invocation in the same way socket activation fds are
1174 passed. This is now used by journald to ensure that the
1175 various sockets connected to all the system's stdout/stderr
1176 are not lost when journald is restarted. File descriptors
1177 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
1178 an extension to sd_notify(). Note that a limit is enforced
1179 on the number of fds a service can store in PID 1, and it
1180 defaults to 0, so that no fds may be stored, unless this is
1181 explicitly turned on.
1183 * The default TERM variable to use for units connected to a
1184 terminal, when no other value is explicitly is set is now
1185 vt220 rather than vt102. This should be fairly safe still,
1186 but allows PgUp/PgDn work.
1188 * The /etc/crypttab option header= as known from Debian is now
1191 * "loginctl user-status" and "loginctl session-status" will
1192 now show the last 10 lines of log messages of the
1193 user/session following the status output. Similar,
1194 "machinectl status" will show the last 10 log lines
1195 associated with a virtual machine or container
1196 service. (Note that this is usually not the log messages
1197 done in the VM/container itself, but simply what the
1198 container manager logs. For nspawn this includes all console
1201 * "loginctl session-status" without further argument will now
1202 show the status of the session of the caller. Similar,
1203 "lock-session", "unlock-session", "activate",
1204 "enable-linger", "disable-linger" may now be called without
1205 session/user parameter in which case they apply to the
1206 caller's session/user.
1208 * An X11 session scriptlet is now shipped that uploads
1209 $DISPLAY and $XAUTHORITY into the environment of the systemd
1210 --user daemon if a session begins. This should improve
1211 compatibility with X11 enabled applications run as systemd
1214 * Generators are now subject to masking via /etc and /run, the
1215 same way as unit files.
1217 * networkd .network files gained support for configuring
1218 per-link IPv4/IPv6 packet forwarding as well as IPv4
1219 masquerading. This is by default turned on for veth links to
1220 containers, as registered by systemd-nspawn. This means that
1221 nspawn containers run with --network-veth will now get
1222 automatic routed access to the host's networks without any
1223 further configuration or setup, as long as networkd runs on
1226 * systemd-nspawn gained the --port= (-p) switch to expose TCP
1227 or UDP posts of a container on the host. With this in place
1228 it is possible to run containers with private veth links
1229 (--network-veth), and have their functionality exposed on
1230 the host as if their services were running directly on the
1233 * systemd-nspawn's --network-veth switch now gained a short
1234 version "-n", since with the changes above it is now truly
1235 useful out-of-the-box. The systemd-nspawn@.service has been
1236 updated to make use of it too by default.
1238 * systemd-nspawn will now maintain a per-image R/W lock, to
1239 ensure that the same image is not started more than once
1240 writable. (It's OK to run an image multiple times
1241 simultaneously in read-only mode.)
1243 * systemd-nspawn's --image= option is now capable of
1244 dissecting and booting MBR and GPT disk images that contain
1245 only a single active Linux partition. Previously it
1246 supported only GPT disk images with proper GPT type
1247 IDs. This allows running cloud images from major
1248 distributions directly with systemd-nspawn, without
1251 * In addition to collecting mouse dpi data in the udev
1252 hardware database, there's now support for collecting angle
1253 information for mouse scroll wheels. The database is
1254 supposed to guarantee similar scrolling behavior on mice
1255 that it knows about. There's also support for collecting
1256 information about Touchpad types.
1258 * udev's input_id built-in will now also collect touch screen
1259 dimension data and attach it to probed devices.
1261 * /etc/os-release gained support for a Distribution Privacy
1264 * networkd gained support for creating "ipvlan", "gretap",
1265 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
1267 * systemd-tmpfiles gained support for "a" lines for setting
1270 * systemd-nspawn will now mount /tmp in the container to
1271 tmpfs, automatically.
1273 * systemd now exposes the memory.usage_in_bytes cgroup
1274 attribute and shows it for each service in the "systemctl
1275 status" output, if available.
1277 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
1278 immediate reboot is triggered. This useful if shutdown is
1279 hung and is unable to complete, to expedite the
1280 operation. Note that this kind of reboot will still unmount
1281 all file systems, and hence should not result in fsck being
1284 * A .device unit for an optical block device will now be
1285 considered active only when a medium is in the drive. Also,
1286 mount units are now bound to their backing devices thus
1287 triggering automatic unmounting when devices become
1288 unavailable. With this in place systemd will now
1289 automatically unmount left-over mounts when a CD-ROM is
1290 ejected or an USB stick is yanked from the system.
1292 * networkd-wait-online now has support for waiting for
1293 specific interfaces only (with globbing), and for giving up
1294 after a configurable timeout.
1296 * networkd now exits when idle. It will be automatically
1297 restarted as soon as interfaces show up, are removed or
1298 change state. networkd will stay around as long as there is
1299 at least one DHCP state machine or similar around, that keep
1302 * networkd may now configure IPv6 link-local addressing in
1303 addition to IPv4 link-local addressing.
1305 * The IPv6 "token" for use in SLAAC may now be configured for
1306 each .network interface in networkd.
1308 * Routes configured with networkd may now be assigned a scope
1311 * networkd's [Match] sections now support globbing and lists
1312 of multiple space-separated matches per item.
1314 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
1315 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
1316 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
1317 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
1318 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
1319 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
1320 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
1321 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
1322 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
1323 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
1324 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
1325 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
1326 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
1327 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
1328 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
1329 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
1330 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
1331 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
1332 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
1333 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
1334 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
1335 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
1336 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
1337 Hoffmann, Zbigniew Jędrzejewski-Szmek
1339 -- Berlin, 2015-02-16
1343 * When querying unit file enablement status (for example via
1344 "systemctl is-enabled"), a new state "indirect" is now known
1345 which indicates that a unit might not be enabled itself, but
1346 another unit listed in its Also= setting might be.
1348 * Similar to the various existing ConditionXYZ= settings for
1349 units, there are now matching AssertXYZ= settings. While
1350 failing conditions cause a unit to be skipped, but its job
1351 to succeed, failing assertions declared like this will cause
1352 a unit start operation and its job to fail.
1354 * hostnamed now knows a new chassis type "embedded".
1356 * systemctl gained a new "edit" command. When used on a unit
1357 file, this allows extending unit files with .d/ drop-in
1358 configuration snippets or editing the full file (after
1359 copying it from /usr/lib to /etc). This will invoke the
1360 user's editor (as configured with $EDITOR), and reload the
1361 modified configuration after editing.
1363 * "systemctl status" now shows the suggested enablement state
1364 for a unit, as declared in the (usually vendor-supplied)
1365 system preset files.
1367 * nss-myhostname will now resolve the single-label host name
1368 "gateway" to the locally configured default IP routing
1369 gateways, ordered by their metrics. This assigns a stable
1370 name to the used gateways, regardless which ones are
1371 currently configured. Note that the name will only be
1372 resolved after all other name sources (if nss-myhostname is
1373 configured properly) and should hence not negatively impact
1374 systems that use the single-label host name "gateway" in
1377 * systemd-inhibit now allows filtering by mode when listing
1380 * Scope and service units gained a new "Delegate" boolean
1381 property, which, when set, allows processes running inside the
1382 unit to further partition resources. This is primarily
1383 useful for systemd user instances as well as container
1386 * journald will now pick up audit messages directly from
1387 the kernel, and log them like any other log message. The
1388 audit fields are split up and fully indexed. This means that
1389 journalctl in many ways is now a (nicer!) alternative to
1390 ausearch, the traditional audit client. Note that this
1391 implements only a minimal audit client. If you want the
1392 special audit modes like reboot-on-log-overflow, please use
1393 the traditional auditd instead, which can be used in
1394 parallel to journald.
1396 * The ConditionSecurity= unit file option now understands the
1397 special string "audit" to check whether auditing is
1400 * journalctl gained two new commands --vacuum-size= and
1401 --vacuum-time= to delete old journal files until the
1402 remaining ones take up no more than the specified size on disk,
1403 or are not older than the specified time.
1405 * A new, native PPPoE library has been added to sd-network,
1406 systemd's library of light-weight networking protocols. This
1407 library will be used in a future version of networkd to
1408 enable PPPoE communication without an external pppd daemon.
1410 * The busctl tool now understands a new "capture" verb that
1411 works similar to "monitor", but writes a packet capture
1412 trace to STDOUT that can be redirected to a file which is
1413 compatible with libcap's capture file format. This can then
1414 be loaded in Wireshark and similar tools to inspect bus
1417 * The busctl tool now understands a new "tree" verb that shows
1418 the object trees of a specific service on the bus, or of all
1421 * The busctl tool now understands a new "introspect" verb that
1422 shows all interfaces and members of objects on the bus,
1423 including their signature and values. This is particularly
1424 useful to get more information about bus objects shown by
1425 the new "busctl tree" command.
1427 * The busctl tool now understands new verbs "call",
1428 "set-property" and "get-property" for invoking bus method
1429 calls, setting and getting bus object properties in a
1432 * busctl gained a new --augment-creds= argument that controls
1433 whether the tool shall augment credential information it
1434 gets from the bus with data from /proc, in a possibly
1437 * nspawn's --link-journal= switch gained two new values
1438 "try-guest" and "try-host" that work like "guest" and
1439 "host", but do not fail if the host has no persistent
1440 journalling enabled. -j is now equivalent to
1441 --link-journal=try-guest.
1443 * macvlan network devices created by nspawn will now have
1444 stable MAC addresses.
1446 * A new SmackProcessLabel= unit setting has been added, which
1447 controls the SMACK security label processes forked off by
1448 the respective unit shall use.
1450 * If compiled with --enable-xkbcommon, systemd-localed will
1451 verify x11 keymap settings by compiling the given keymap. It
1452 will spew out warnings if the compilation fails. This
1453 requires libxkbcommon to be installed.
1455 * When a coredump is collected, a larger number of metadata
1456 fields is now collected and included in the journal records
1457 created for it. More specifically, control group membership,
1458 environment variables, memory maps, working directory,
1459 chroot directory, /proc/$PID/status, and a list of open file
1460 descriptors is now stored in the log entry.
1462 * The udev hwdb now contains DPI information for mice. For
1465 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
1467 * All systemd programs that read standalone configuration
1468 files in /etc now also support a corresponding series of
1469 .conf.d configuration directories in /etc/, /run/,
1470 /usr/local/lib/, /usr/lib/, and (if configured with
1471 --enable-split-usr) /lib/. In particular, the following
1472 configuration files now have corresponding configuration
1473 directories: system.conf user.conf, logind.conf,
1474 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
1475 resolved.conf, timesyncd.conf, journal-remote.conf, and
1476 journal-upload.conf. Note that distributions should use the
1477 configuration directories in /usr/lib/; the directories in
1478 /etc/ are reserved for the system administrator.
1480 * systemd-rfkill will no longer take the rfkill device name
1481 into account when storing rfkill state on disk, as the name
1482 might be dynamically assigned and not stable. Instead, the
1483 ID_PATH udev variable combined with the rfkill type (wlan,
1484 bluetooth, ...) is used.
1486 * A new service systemd-machine-id-commit.service has been
1487 added. When used on systems where /etc is read-only during
1488 boot, and /etc/machine-id is not initialized (but an empty
1489 file), this service will copy the temporary machine ID
1490 created as replacement into /etc after the system is fully
1491 booted up. This is useful for systems that are freshly
1492 installed with a non-initialized machine ID, but should get
1493 a fixed machine ID for subsequent boots.
1495 * networkd's .netdev files now provide a large set of
1496 configuration parameters for VXLAN devices. Similarly, the
1497 bridge port cost parameter is now configurable in .network
1498 files. There's also new support for configuring IP source
1499 routing. networkd .link files gained support for a new
1500 OriginalName= match that is useful to match against the
1501 original interface name the kernel assigned. .network files
1502 may include MTU= and MACAddress= fields for altering the MTU
1503 and MAC address while being connected to a specific network
1506 * The LUKS logic gained supported for configuring
1507 UUID-specific key files. There's also new support for naming
1508 LUKS device from the kernel command line, using the new
1509 luks.name= argument.
1511 * Timer units may now be transiently created via the bus API
1512 (this was previously already available for scope and service
1513 units). In addition it is now possible to create multiple
1514 transient units at the same time with a single bus call. The
1515 "systemd-run" tool has been updated to make use of this for
1516 running commands on a specified time, in at(1)-style.
1518 * tmpfiles gained support for "t" lines, for assigning
1519 extended attributes to files. Among other uses this may be
1520 used to assign SMACK labels to files.
1522 Contributions from: Alin Rauta, Alison Chaiken, Andrej
1523 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
1524 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
1525 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
1526 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
1527 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
1528 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
1529 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
1530 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
1531 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
1532 Michael Biebl, Michael Chapman, Michael Marineau, Michal
1533 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
1534 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
1535 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
1536 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
1537 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
1538 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
1539 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
1541 -- Berlin, 2014-12-10
1545 * journalctl gained the new options -t/--identifier= to match
1546 on the syslog identifier (aka "tag"), as well as --utc to
1547 show log timestamps in the UTC timezone. journalctl now also
1548 accepts -n/--lines=all to disable line capping in a pager.
1550 * journalctl gained a new switch, --flush, that synchronously
1551 flushes logs from /run/log/journal to /var/log/journal if
1552 persistent storage is enabled. systemd-journal-flush.service
1553 now waits until the operation is complete.
1555 * Services can notify the manager before they start a reload
1556 (by sending RELOADING=1) or shutdown (by sending
1557 STOPPING=1). This allows the manager to track and show the
1558 internal state of daemons and closes a race condition when
1559 the process is still running but has closed its D-Bus
1562 * Services with Type=oneshot do not have to have any ExecStart
1565 * User units are now loaded also from
1566 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
1567 /run/systemd/user directory that was already previously
1568 supported, but is under the control of the user.
1570 * Job timeouts (i.e. time-outs on the time a job that is
1571 queued stays in the run queue) can now optionally result in
1572 immediate reboot or power-off actions (JobTimeoutAction= and
1573 JobTimeoutRebootArgument=). This is useful on ".target"
1574 units, to limit the maximum time a target remains
1575 undispatched in the run queue, and to trigger an emergency
1576 operation in such a case. This is now used by default to
1577 turn off the system if boot-up (as defined by everything in
1578 basic.target) hangs and does not complete for at least
1579 15min. Also, if power-off or reboot hang for at least 30min
1580 an immediate power-off/reboot operation is triggered. This
1581 functionality is particularly useful to increase reliability
1582 on embedded devices, but also on laptops which might
1583 accidentally get powered on when carried in a backpack and
1584 whose boot stays stuck in a hard disk encryption passphrase
1587 * systemd-logind can be configured to also handle lid switch
1588 events even when the machine is docked or multiple displays
1589 are attached (HandleLidSwitchDocked= option).
1591 * A helper binary and a service have been added which can be
1592 used to resume from hibernation in the initramfs. A
1593 generator will parse the resume= option on the kernel
1594 command line to trigger resume.
1596 * A user console daemon systemd-consoled has been
1597 added. Currently, it is a preview, and will so far open a
1598 single terminal on each session of the user marked as
1599 Desktop=systemd-console.
1601 * Route metrics can be specified for DHCP routes added by
1604 * The SELinux context of socket-activated services can be set
1605 from the information provided by the networking stack
1606 (SELinuxContextFromNet= option).
1608 * Userspace firmware loading support has been removed and
1609 the minimum supported kernel version is thus bumped to 3.7.
1611 * Timeout for udev workers has been increased from 1 to 3
1612 minutes, but a warning will be printed after 1 minute to
1613 help diagnose kernel modules that take a long time to load.
1615 * Udev rules can now remove tags on devices with TAG-="foobar".
1617 * systemd's readahead implementation has been removed. In many
1618 circumstances it didn't give expected benefits even for
1619 rotational disk drives and was becoming less relevant in the
1620 age of SSDs. As none of the developers has been using
1621 rotating media anymore, and nobody stepped up to actively
1622 maintain this component of systemd it has now been removed.
1624 * Swap units can use Options= to specify discard options.
1625 Discard options specified for swaps in /etc/fstab are now
1628 * Docker containers are now detected as a separate type of
1631 * The Password Agent protocol gained support for queries where
1632 the user input is shown, useful e.g. for user names.
1633 systemd-ask-password gained a new --echo option to turn that
1636 * The default sysctl.d/ snippets will now set:
1638 net.core.default_qdisc = fq_codel
1640 This selects Fair Queuing Controlled Delay as the default
1641 queuing discipline for network interfaces. fq_codel helps
1642 fight the network bufferbloat problem. It is believed to be
1643 a good default with no tuning required for most workloads.
1644 Downstream distributions may override this choice. On 10Gbit
1645 servers that do not do forwarding, "fq" may perform better.
1646 Systems without a good clocksource should use "pfifo_fast".
1648 * If kdbus is enabled during build a new option BusPolicy= is
1649 available for service units, that allows locking all service
1650 processes into a stricter bus policy, in order to limit
1651 access to various bus services, or even hide most of them
1652 from the service's view entirely.
1654 * networkctl will now show the .network and .link file
1655 networkd has applied to a specific interface.
1657 * sd-login gained a new API call sd_session_get_desktop() to
1658 query which desktop environment has been selected for a
1661 * UNIX utmp support is now compile-time optional to support
1662 legacy-free systems.
1664 * systemctl gained two new commands "add-wants" and
1665 "add-requires" for pulling in units from specific targets
1668 * If the word "rescue" is specified on the kernel command line
1669 the system will now boot into rescue mode (aka
1670 rescue.target), which was previously available only by
1671 specifying "1" or "systemd.unit=rescue.target" on the kernel
1672 command line. This new kernel command line option nicely
1673 mirrors the already existing "emergency" kernel command line
1676 * New kernel command line options mount.usr=, mount.usrflags=,
1677 mount.usrfstype= have been added that match root=, rootflags=,
1678 rootfstype= but allow mounting a specific file system to
1681 * The $NOTIFY_SOCKET is now also passed to control processes of
1682 services, not only the main process.
1684 * This version reenables support for fsck's -l switch. This
1685 means at least version v2.25 of util-linux is required for
1686 operation, otherwise dead-locks on device nodes may
1687 occur. Again: you need to update util-linux to at least
1688 v2.25 when updating systemd to v217.
1690 * The "multi-seat-x" tool has been removed from systemd, as
1691 its functionality has been integrated into X servers 1.16,
1692 and the tool is hence redundant. It is recommended to update
1693 display managers invoking this tool to simply invoke X
1694 directly from now on, again.
1696 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
1697 message flag has been added for all of systemd's PolicyKit
1698 authenticated method calls has been added. In particular
1699 this now allows optional interactive authorization via
1700 PolicyKit for many of PID1's privileged operations such as
1701 unit file enabling and disabling.
1703 * "udevadm hwdb --update" learnt a new switch "--usr" for
1704 placing the rebuilt hardware database in /usr instead of
1705 /etc. When used only hardware database entries stored in
1706 /usr will be used, and any user database entries in /etc are
1707 ignored. This functionality is useful for vendors to ship a
1708 pre-built database on systems where local configuration is
1709 unnecessary or unlikely.
1711 * Calendar time specifications in .timer units now also
1712 understand the strings "semi-annually", "quarterly" and
1713 "minutely" as shortcuts (in addition to the preexisting
1714 "anually", "hourly", ...).
1716 * systemd-tmpfiles will now correctly create files in /dev
1717 at boot which are marked for creation only at boot. It is
1718 recommended to always create static device nodes with 'c!'
1719 and 'b!', so that they are created only at boot and not
1720 overwritten at runtime.
1722 * When the watchdog logic is used for a service (WatchdogSec=)
1723 and the watchdog timeout is hit the service will now be
1724 terminated with SIGABRT (instead of just SIGTERM), in order
1725 to make sure a proper coredump and backtrace is
1726 generated. This ensures that hanging services will result in
1727 similar coredump/backtrace behaviour as services that hit a
1730 Contributions from: Andreas Henriksson, Andrei Borzenkov,
1731 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
1732 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
1733 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
1734 Herrmann, David Sommerseth, David Strauss, Emil Renner
1735 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
1736 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
1737 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
1738 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
1739 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
1740 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
1741 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
1742 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
1743 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
1744 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
1745 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
1746 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
1747 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
1748 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
1749 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
1750 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
1753 -- Berlin, 2014-10-28
1757 * timedated no longer reads NTP implementation unit names from
1758 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
1759 implementations should add a
1761 Conflicts=systemd-timesyncd.service
1763 to their unit files to take over and replace systemd's NTP
1764 default functionality.
1766 * systemd-sysusers gained a new line type "r" for configuring
1767 which UID/GID ranges to allocate system users/groups
1768 from. Lines of type "u" may now add an additional column
1769 that specifies the home directory for the system user to be
1770 created. Also, systemd-sysusers may now optionally read user
1771 information from STDIN instead of a file. This is useful for
1772 invoking it from RPM preinst scriptlets that need to create
1773 users before the first RPM file is installed since these
1774 files might need to be owned by them. A new
1775 %sysusers_create_inline RPM macro has been introduced to do
1776 just that. systemd-sysusers now updates the shadow files as
1777 well as the user/group databases, which should enhance
1778 compatibility with certain tools like grpck.
1780 * A number of bus APIs of PID 1 now optionally consult
1781 PolicyKit to permit access for otherwise unprivileged
1782 clients under certain conditions. Note that this currently
1783 doesn't support interactive authentication yet, but this is
1784 expected to be added eventually, too.
1786 * /etc/machine-info now has new fields for configuring the
1787 deployment environment of the machine, as well as the
1788 location of the machine. hostnamectl has been updated with
1789 new command to update these fields.
1791 * systemd-timesyncd has been updated to automatically acquire
1792 NTP server information from systemd-networkd, which might
1793 have been discovered via DHCP.
1795 * systemd-resolved now includes a caching DNS stub resolver
1796 and a complete LLMNR name resolution implementation. A new
1797 NSS module "nss-resolve" has been added which can be used
1798 instead of glibc's own "nss-dns" to resolve hostnames via
1799 systemd-resolved. Hostnames, addresses and arbitrary RRs may
1800 be resolved via systemd-resolved D-Bus APIs. In contrast to
1801 the glibc internal resolver systemd-resolved is aware of
1802 multi-homed system, and keeps DNS server and caches separate
1803 and per-interface. Queries are sent simultaneously on all
1804 interfaces that have DNS servers configured, in order to
1805 properly handle VPNs and local LANs which might resolve
1806 separate sets of domain names. systemd-resolved may acquire
1807 DNS server information from systemd-networkd automatically,
1808 which in turn might have discovered them via DHCP. A tool
1809 "systemd-resolve-host" has been added that may be used to
1810 query the DNS logic in resolved. systemd-resolved implements
1811 IDNA and automatically uses IDNA or UTF-8 encoding depending
1812 on whether classic DNS or LLMNR is used as transport. In the
1813 next releases we intend to add a DNSSEC and mDNS/DNS-SD
1814 implementation to systemd-resolved.
1816 * A new NSS module nss-mymachines has been added, that
1817 automatically resolves the names of all local registered
1818 containers to their respective IP addresses.
1820 * A new client tool "networkctl" for systemd-networkd has been
1821 added. It currently is entirely passive and will query
1822 networking configuration from udev, rtnetlink and networkd,
1823 and present it to the user in a very friendly
1824 way. Eventually, we hope to extend it to become a full
1825 control utility for networkd.
1827 * .socket units gained a new DeferAcceptSec= setting that
1828 controls the kernels' TCP_DEFER_ACCEPT sockopt for
1829 TCP. Similarly, support for controlling TCP keep-alive
1830 settings has been added (KeepAliveTimeSec=,
1831 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
1832 turning off Nagle's algorithm on TCP has been added
1835 * logind learned a new session type "web", for use in projects
1836 like Cockpit which register web clients as PAM sessions.
1838 * timer units with at least one OnCalendar= setting will now
1839 be started only after timer-sync.target has been
1840 reached. This way they will not elapse before the system
1841 clock has been corrected by a local NTP client or
1842 similar. This is particular useful on RTC-less embedded
1843 machines, that come up with an invalid system clock.
1845 * systemd-nspawn's --network-veth= switch should now result in
1846 stable MAC addresses for both the outer and the inner side
1849 * systemd-nspawn gained a new --volatile= switch for running
1850 container instances with /etc or /var unpopulated.
1852 * The kdbus client code has been updated to use the new Linux
1853 3.17 memfd subsystem instead of the old kdbus-specific one.
1855 * systemd-networkd's DHCP client and server now support
1856 FORCERENEW. There are also new configuration options to
1857 configure the vendor client identifier and broadcast mode
1860 * systemd will no longer inform the kernel about the current
1861 timezone, as this is necessarily incorrect and racy as the
1862 kernel has no understanding of DST and similar
1863 concepts. This hence means FAT timestamps will be always
1864 considered UTC, similar to what Android is already
1865 doing. Also, when the RTC is configured to the local time
1866 (rather than UTC) systemd will never synchronize back to it,
1867 as this might confuse Windows at a later boot.
1869 * systemd-analyze gained a new command "verify" for offline
1870 validation of unit files.
1872 * systemd-networkd gained support for a couple of additional
1873 settings for bonding networking setups. Also, the metric for
1874 statically configured routes may now be configured. For
1875 network interfaces where this is appropriate the peer IP
1876 address may now be configured.
1878 * systemd-networkd's DHCP client will no longer request
1879 broadcasting by default, as this tripped up some networks.
1880 For hardware where broadcast is required the feature should
1881 be switched back on using RequestBroadcast=yes.
1883 * systemd-networkd will now set up IPv4LL addresses (when
1884 enabled) even if DHCP is configured successfully.
1886 * udev will now default to respect network device names given
1887 by the kernel when the kernel indicates that these are
1888 predictable. This behavior can be tweaked by changing
1889 NamePolicy= in the relevant .link file.
1891 * A new library systemd-terminal has been added that
1892 implements full TTY stream parsing and rendering. This
1893 library is supposed to be used later on for implementing a
1894 full userspace VT subsystem, replacing the current kernel
1897 * A new tool systemd-journal-upload has been added to push
1898 journal data to a remote system running
1899 systemd-journal-remote.
1901 * journald will no longer forward all local data to another
1902 running syslog daemon. This change has been made because
1903 rsyslog (which appears to be the most commonly used syslog
1904 implementation these days) no longer makes use of this, and
1905 instead pulls the data out of the journal on its own. Since
1906 forwarding the messages to a non-existent syslog server is
1907 more expensive than we assumed we have now turned this
1908 off. If you run a syslog server that is not a recent rsyslog
1909 version, you have to turn this option on again
1910 (ForwardToSyslog= in journald.conf).
1912 * journald now optionally supports the LZ4 compressor for
1913 larger journal fields. This compressor should perform much
1914 better than XZ which was the previous default.
1916 * machinectl now shows the IP addresses of local containers,
1917 if it knows them, plus the interface name of the container.
1919 * A new tool "systemd-escape" has been added that makes it
1920 easy to escape strings to build unit names and similar.
1922 * sd_notify() messages may now include a new ERRNO= field
1923 which is parsed and collected by systemd and shown among the
1924 "systemctl status" output for a service.
1926 * A new component "systemd-firstboot" has been added that
1927 queries the most basic systemd information (timezone,
1928 hostname, root password) interactively on first
1929 boot. Alternatively it may also be used to provision these
1930 things offline on OS images installed into directories.
1932 * The default sysctl.d/ snippets will now set
1934 net.ipv4.conf.default.promote_secondaries=1
1936 This has the benefit of no flushing secondary IP addresses
1937 when primary addresses are removed.
1939 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
1940 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
1941 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
1942 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
1943 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
1944 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
1945 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
1946 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
1947 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
1948 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
1949 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
1950 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
1951 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
1952 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
1953 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
1955 -- Berlin, 2014-08-19
1959 * A new tool systemd-sysusers has been added. This tool
1960 creates system users and groups in /etc/passwd and
1961 /etc/group, based on static declarative system user/group
1962 definitions in /usr/lib/sysusers.d/. This is useful to
1963 enable factory resets and volatile systems that boot up with
1964 an empty /etc directory, and thus need system users and
1965 groups created during early boot. systemd now also ships
1966 with two default sysusers.d/ files for the most basic
1967 users and groups systemd and the core operating system
1970 * A new tmpfiles snippet has been added that rebuilds the
1971 essential files in /etc on boot, should they be missing.
1973 * A directive for ensuring automatic clean-up of
1974 /var/cache/man/ has been removed from the default
1975 configuration. This line should now be shipped by the man
1976 implementation. The necessary change has been made to the
1977 man-db implementation. Note that you need to update your man
1978 implementation to one that ships this line, otherwise no
1979 automatic clean-up of /var/cache/man will take place.
1981 * A new condition ConditionNeedsUpdate= has been added that
1982 may conditionalize services to only run when /etc or /var
1983 are "older" than the vendor operating system resources in
1984 /usr. This is useful for reconstructing or updating /etc
1985 after an offline update of /usr or a factory reset, on the
1986 next reboot. Services that want to run once after such an
1987 update or reset should use this condition and order
1988 themselves before the new systemd-update-done.service, which
1989 will mark the two directories as fully updated. A number of
1990 service files have been added making use of this, to rebuild
1991 the udev hardware database, the journald message catalog and
1992 dynamic loader cache (ldconfig). The systemd-sysusers tool
1993 described above also makes use of this now. With this in
1994 place it is now possible to start up a minimal operating
1995 system with /etc empty cleanly. For more information on the
1996 concepts involved see this recent blog story:
1998 http://0pointer.de/blog/projects/stateless.html
2000 * A new system group "input" has been introduced, and all
2001 input device nodes get this group assigned. This is useful
2002 for system-level software to get access to input devices. It
2003 complements what is already done for "audio" and "video".
2005 * systemd-networkd learnt minimal DHCPv4 server support in
2006 addition to the existing DHCPv4 client support. It also
2007 learnt DHCPv6 client and IPv6 Router Solicitation client
2008 support. The DHCPv4 client gained support for static routes
2009 passed in from the server. Note that the [DHCPv4] section
2010 known in older systemd-networkd versions has been renamed to
2011 [DHCP] and is now also used by the DHCPv6 client. Existing
2012 .network files using settings of this section should be
2013 updated, though compatibility is maintained. Optionally, the
2014 client hostname may now be sent to the DHCP server.
2016 * networkd gained support for vxlan virtual networks as well
2017 as tun/tap and dummy devices.
2019 * networkd gained support for automatic allocation of address
2020 ranges for interfaces from a system-wide pool of
2021 addresses. This is useful for dynamically managing a large
2022 number of interfaces with a single network configuration
2023 file. In particular this is useful to easily assign
2024 appropriate IP addresses to the veth links of a large number
2025 of nspawn instances.
2027 * RPM macros for processing sysusers, sysctl and binfmt
2028 drop-in snippets at package installation time have been
2031 * The /etc/os-release file should now be placed in
2032 /usr/lib/os-release. The old location is automatically
2033 created as symlink. /usr/lib is the more appropriate
2034 location of this file, since it shall actually describe the
2035 vendor operating system shipped in /usr, and not the
2036 configuration stored in /etc.
2038 * .mount units gained a new boolean SloppyOptions= setting
2039 that maps to mount(8)'s -s option which enables permissive
2040 parsing of unknown mount options.
2042 * tmpfiles learnt a new "L+" directive which creates a symlink
2043 but (unlike "L") deletes a pre-existing file first, should
2044 it already exist and not already be the correct
2045 symlink. Similarly, "b+", "c+" and "p+" directives have been
2046 added as well, which create block and character devices, as
2047 well as fifos in the filesystem, possibly removing any
2048 pre-existing files of different types.
2050 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
2051 'argument' field (which so far specified the source to
2052 symlink/copy the files from) is now optional. If omitted the
2053 same file os copied from /usr/share/factory/ suffixed by the
2054 full destination path. This is useful for populating /etc
2055 with essential files, by copying them from vendor defaults
2056 shipped in /usr/share/factory/etc.
2058 * A new command "systemctl preset-all" has been added that
2059 applies the service preset settings to all installed unit
2060 files. A new switch --preset-mode= has been added that
2061 controls whether only enable or only disable operations
2064 * A new command "systemctl is-system-running" has been added
2065 that allows checking the overall state of the system, for
2066 example whether it is fully up and running.
2068 * When the system boots up with an empty /etc, the equivalent
2069 to "systemctl preset-all" is executed during early boot, to
2070 make sure all default services are enabled after a factory
2073 * systemd now contains a minimal preset file that enables the
2074 most basic services systemd ships by default.
2076 * Unit files' [Install] section gained a new DefaultInstance=
2077 field for defining the default instance to create if a
2078 template unit is enabled with no instance specified.
2080 * A new passive target cryptsetup-pre.target has been added
2081 that may be used by services that need to make they run and
2082 finish before the first LUKS cryptographic device is set up.
2084 * The /dev/loop-control and /dev/btrfs-control device nodes
2085 are now owned by the "disk" group by default, opening up
2086 access to this group.
2088 * systemd-coredump will now automatically generate a
2089 stack trace of all core dumps taking place on the system,
2090 based on elfutils' libdw library. This stack trace is logged
2093 * systemd-coredump may now optionally store coredumps directly
2094 on disk (in /var/lib/systemd/coredump, possibly compressed),
2095 instead of storing them unconditionally in the journal. This
2096 mode is the new default. A new configuration file
2097 /etc/systemd/coredump.conf has been added to configure this
2098 and other parameters of systemd-coredump.
2100 * coredumpctl gained a new "info" verb to show details about a
2101 specific coredump. A new switch "-1" has also been added
2102 that makes sure to only show information about the most
2103 recent entry instead of all entries. Also, as the tool is
2104 generally useful now the "systemd-" prefix of the binary
2105 name has been removed. Distributions that want to maintain
2106 compatibility with the old name should add a symlink from
2107 the old name to the new name.
2109 * journald's SplitMode= now defaults to "uid". This makes sure
2110 that unprivileged users can access their own coredumps with
2111 coredumpctl without restrictions.
2113 * New kernel command line options "systemd.wants=" (for
2114 pulling an additional unit during boot), "systemd.mask="
2115 (for masking a specific unit for the boot), and
2116 "systemd.debug-shell" (for enabling the debug shell on tty9)
2117 have been added. This is implemented in the new generator
2118 "systemd-debug-generator".
2120 * systemd-nspawn will now by default filter a couple of
2121 syscalls for containers, among them those required for
2122 kernel module loading, direct x86 IO port access, swap
2123 management, and kexec. Most importantly though
2124 open_by_handle_at() is now prohibited for containers,
2125 closing a hole similar to a recently discussed vulnerability
2126 in docker regarding access to files on file hierarchies the
2127 container should normally not have access to. Note that, for
2128 nspawn, we generally make no security claims anyway (and
2129 this is explicitly documented in the man page), so this is
2130 just a fix for one of the most obvious problems.
2132 * A new man page file-hierarchy(7) has been added that
2133 contains a minimized, modernized version of the file system
2134 layout systemd expects, similar in style to the FHS
2135 specification or hier(5). A new tool systemd-path(1) has
2136 been added to query many of these paths for the local
2139 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
2140 longer done. Since the directory now has a per-user size
2141 limit, and is cleaned on logout this appears unnecessary,
2142 in particular since this now brings the lifecycle of this
2143 directory closer in line with how IPC objects are handled.
2145 * systemd.pc now exports a number of additional directories,
2146 including $libdir (which is useful to identify the library
2147 path for the primary architecture of the system), and a
2148 couple of drop-in directories.
2150 * udev's predictable network interface names now use the dev_port
2151 sysfs attribute, introduced in linux 3.15 instead of dev_id to
2152 distinguish between ports of the same PCI function. dev_id should
2153 only be used for ports using the same HW address, hence the need
2156 * machined has been updated to export the OS version of a
2157 container (read from /etc/os-release and
2158 /usr/lib/os-release) on the bus. This is now shown in
2159 "machinectl status" for a machine.
2161 * A new service setting RestartForceExitStatus= has been
2162 added. If configured to a set of exit signals or process
2163 return values, the service will be restarted when the main
2164 daemon process exits with any of them, regardless of the
2167 * systemctl's -H switch for connecting to remote systemd
2168 machines has been extended so that it may be used to
2169 directly connect to a specific container on the
2170 host. "systemctl -H root@foobar:waldi" will now connect as
2171 user "root" to host "foobar", and then proceed directly to
2172 the container named "waldi". Note that currently you have to
2173 authenticate as user "root" for this to work, as entering
2174 containers is a privileged operation.
2176 Contributions from: Andreas Henriksson, Benjamin Steinwender,
2177 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
2178 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
2179 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
2180 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
2181 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
2182 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
2183 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
2184 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
2185 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
2186 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
2187 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
2189 -- Berlin, 2014-07-03
2193 * As an experimental feature, udev now tries to lock the
2194 disk device node (flock(LOCK_SH|LOCK_NB)) while it
2195 executes events for the disk or any of its partitions.
2196 Applications like partitioning programs can lock the
2197 disk device node (flock(LOCK_EX)) and claim temporary
2198 device ownership that way; udev will entirely skip all event
2199 handling for this disk and its partitions. If the disk
2200 was opened for writing, the close will trigger a partition
2201 table rescan in udev's "watch" facility, and if needed
2202 synthesize "change" events for the disk and all its partitions.
2203 This is now unconditionally enabled, and if it turns out to
2204 cause major problems, we might turn it on only for specific
2205 devices, or might need to disable it entirely. Device Mapper
2206 devices are excluded from this logic.
2208 * We temporarily dropped the "-l" switch for fsck invocations,
2209 since they collide with the flock() logic above. util-linux
2210 upstream has been changed already to avoid this conflict,
2211 and we will readd "-l" as soon as util-linux with this
2212 change has been released.
2214 * The dependency on libattr has been removed. Since a long
2215 time, the extended attribute calls have moved to glibc, and
2216 libattr is thus unnecessary.
2218 * Virtualization detection works without priviliges now. This
2219 means the systemd-detect-virt binary no longer requires
2220 CAP_SYS_PTRACE file capabilities, and our daemons can run
2221 with fewer privileges.
2223 * systemd-networkd now runs under its own "systemd-network"
2224 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
2225 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
2226 loses the ability to write to files owned by root this way.
2228 * Similarly, systemd-resolved now runs under its own
2229 "systemd-resolve" user with no capabilities remaining.
2231 * Similarly, systemd-bus-proxyd now runs under its own
2232 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
2234 * systemd-networkd gained support for setting up "veth"
2235 virtual Ethernet devices for container connectivity, as well
2236 as GRE and VTI tunnels.
2238 * systemd-networkd will no longer automatically attempt to
2239 manually load kernel modules necessary for certain tunnel
2240 transports. Instead, it is assumed the kernel loads them
2241 automatically when required. This only works correctly on
2242 very new kernels. On older kernels, please consider adding
2243 the kernel modules to /etc/modules-load.d/ as a work-around.
2245 * The resolv.conf file systemd-resolved generates has been
2246 moved to /run/systemd/resolve/. If you have a symlink from
2247 /etc/resolv.conf, it might be necessary to correct it.
2249 * Two new service settings, ProtectHome= and ProtectSystem=,
2250 have been added. When enabled, they will make the user data
2251 (such as /home) inaccessible or read-only and the system
2252 (such as /usr) read-only, for specific services. This allows
2253 very light-weight per-service sandboxing to avoid
2254 modifications of user data or system files from
2255 services. These two new switches have been enabled for all
2256 of systemd's long-running services, where appropriate.
2258 * Socket units gained new SocketUser= and SocketGroup=
2259 settings to set the owner user and group of AF_UNIX sockets
2260 and FIFOs in the file system.
2262 * Socket units gained a new RemoveOnStop= setting. If enabled,
2263 all FIFOS and sockets in the file system will be removed
2264 when the specific socket unit is stopped.
2266 * Socket units gained a new Symlinks= setting. It takes a list
2267 of symlinks to create to file system sockets or FIFOs
2268 created by the specific Unix sockets. This is useful to
2269 manage symlinks to socket nodes with the same life-cycle as
2272 * The /dev/log socket and /dev/initctl FIFO have been moved to
2273 /run, and have been replaced by symlinks. This allows
2274 connecting to these facilities even if PrivateDevices=yes is
2275 used for a service (which makes /dev/log itself unavailable,
2276 but /run is left). This also has the benefit of ensuring
2277 that /dev only contains device nodes, directories and
2278 symlinks, and nothing else.
2280 * sd-daemon gained two new calls sd_pid_notify() and
2281 sd_pid_notifyf(). They are similar to sd_notify() and
2282 sd_notifyf(), but allow overriding of the source PID of
2283 notification messages if permissions permit this. This is
2284 useful to send notify messages on behalf of a different
2285 process (for example, the parent process). The
2286 systemd-notify tool has been updated to make use of this
2287 when sending messages (so that notification messages now
2288 originate from the shell script invoking systemd-notify and
2289 not the systemd-notify process itself. This should minimize
2290 a race where systemd fails to associate notification
2291 messages to services when the originating process already
2294 * A new "on-abnormal" setting for Restart= has been added. If
2295 set, it will result in automatic restarts on all "abnormal"
2296 reasons for a process to exit, which includes unclean
2297 signals, core dumps, timeouts and watchdog timeouts, but
2298 does not include clean and unclean exit codes or clean
2299 signals. Restart=on-abnormal is an alternative for
2300 Restart=on-failure for services that shall be able to
2301 terminate and avoid restarts on certain errors, by
2302 indicating so with an unclean exit code. Restart=on-failure
2303 or Restart=on-abnormal is now the recommended setting for
2304 all long-running services.
2306 * If the InaccessibleDirectories= service setting points to a
2307 mount point (or if there are any submounts contained within
2308 it), it is now attempted to completely unmount it, to make
2309 the file systems truly unavailable for the respective
2312 * The ReadOnlyDirectories= service setting and
2313 systemd-nspawn's --read-only parameter are now recursively
2314 applied to all submounts, too.
2316 * Mount units may now be created transiently via the bus APIs.
2318 * The support for SysV and LSB init scripts has been removed
2319 from the systemd daemon itself. Instead, it is now
2320 implemented as a generator that creates native systemd units
2321 from these scripts when needed. This enables us to remove a
2322 substantial amount of legacy code from PID 1, following the
2323 fact that many distributions only ship a very small number
2324 of LSB/SysV init scripts nowadays.
2326 * Privileged Xen (dom0) domains are not considered
2327 virtualization anymore by the virtualization detection
2328 logic. After all, they generally have unrestricted access to
2329 the hardware and usually are used to manage the unprivileged
2332 * systemd-tmpfiles gained a new "C" line type, for copying
2333 files or entire directories.
2335 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
2336 lines. So far, they have been non-globbing versions of the
2337 latter, and have thus been redundant. In future, it is
2338 recommended to only use "z". "m" has hence been removed
2339 from the documentation, even though it stays supported.
2341 * A tmpfiles snippet to recreate the most basic structure in
2342 /var has been added. This is enough to create the /var/run →
2343 /run symlink and create a couple of structural
2344 directories. This allows systems to boot up with an empty or
2345 volatile /var. Of course, while with this change, the core OS
2346 now is capable with dealing with a volatile /var, not all
2347 user services are ready for it. However, we hope that sooner
2348 or later, many service daemons will be changed upstream so
2349 that they are able to automatically create their necessary
2350 directories in /var at boot, should they be missing. This is
2351 the first step to allow state-less systems that only require
2352 the vendor image for /usr to boot.
2354 * systemd-nspawn has gained a new --tmpfs= switch to mount an
2355 empty tmpfs instance to a specific directory. This is
2356 particularly useful for making use of the automatic
2357 reconstruction of /var (see above), by passing --tmpfs=/var.
2359 * Access modes specified in tmpfiles snippets may now be
2360 prefixed with "~", which indicates that they shall be masked
2361 by whether the existing file or directory is currently
2362 writable, readable or executable at all. Also, if specified,
2363 the sgid/suid/sticky bits will be masked for all
2366 * A new passive target unit "network-pre.target" has been
2367 added which is useful for services that shall run before any
2368 network is configured, for example firewall scripts.
2370 * The "floppy" group that previously owned the /dev/fd*
2371 devices is no longer used. The "disk" group is now used
2372 instead. Distributions should probably deprecate usage of
2375 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
2376 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
2377 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
2378 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
2379 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
2380 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
2383 -- Berlin, 2014-06-11
2387 * A new "systemd-timesyncd" daemon has been added for
2388 synchronizing the system clock across the network. It
2389 implements an SNTP client. In contrast to NTP
2390 implementations such as chrony or the NTP reference server,
2391 this only implements a client side, and does not bother with
2392 the full NTP complexity, focusing only on querying time from
2393 one remote server and synchronizing the local clock to
2394 it. Unless you intend to serve NTP to networked clients or
2395 want to connect to local hardware clocks, this simple NTP
2396 client should be more than appropriate for most
2397 installations. The daemon runs with minimal privileges, and
2398 has been hooked up with networkd to only operate when
2399 network connectivity is available. The daemon saves the
2400 current clock to disk every time a new NTP sync has been
2401 acquired, and uses this to possibly correct the system clock
2402 early at bootup, in order to accommodate for systems that
2403 lack an RTC such as the Raspberry Pi and embedded devices,
2404 and to make sure that time monotonically progresses on these
2405 systems, even if it is not always correct. To make use of
2406 this daemon, a new system user and group "systemd-timesync"
2407 needs to be created on installation of systemd.
2409 * The queue "seqnum" interface of libudev has been disabled, as
2410 it was generally incompatible with device namespacing as
2411 sequence numbers of devices go "missing" if the devices are
2412 part of a different namespace.
2414 * "systemctl list-timers" and "systemctl list-sockets" gained
2415 a --recursive switch for showing units of these types also
2416 for all local containers, similar in style to the already
2417 supported --recursive switch for "systemctl list-units".
2419 * A new RebootArgument= setting has been added for service
2420 units, which may be used to specify a kernel reboot argument
2421 to use when triggering reboots with StartLimitAction=.
2423 * A new FailureAction= setting has been added for service
2424 units which may be used to specify an operation to trigger
2425 when a service fails. This works similarly to
2426 StartLimitAction=, but unlike it, controls what is done
2427 immediately rather than only after several attempts to
2428 restart the service in question.
2430 * hostnamed got updated to also expose the kernel name,
2431 release, and version on the bus. This is useful for
2432 executing commands like hostnamectl with the -H switch.
2433 systemd-analyze makes use of this to properly display
2434 details when running non-locally.
2436 * The bootchart tool can now show cgroup information in the
2437 graphs it generates.
2439 * The CFS CPU quota cgroup attribute is now exposed for
2440 services. The new CPUQuota= switch has been added for this
2441 which takes a percentage value. Setting this will have the
2442 result that a service may never get more CPU time than the
2443 specified percentage, even if the machine is otherwise idle.
2445 * systemd-networkd learned IPIP and SIT tunnel support.
2447 * LSB init scripts exposing a dependency on $network will now
2448 get a dependency on network-online.target rather than simply
2449 network.target. This should bring LSB handling closer to
2450 what it was on SysV systems.
2452 * A new fsck.repair= kernel option has been added to control
2453 how fsck shall deal with unclean file systems at boot.
2455 * The (.ini) configuration file parser will now silently
2456 ignore sections whose name begins with "X-". This may be
2457 used to maintain application-specific extension sections in unit
2460 * machined gained a new API to query the IP addresses of
2461 registered containers. "machinectl status" has been updated
2462 to show these addresses in its output.
2464 * A new call sd_uid_get_display() has been added to the
2465 sd-login APIs for querying the "primary" session of a
2466 user. The "primary" session of the user is elected from the
2467 user's sessions and generally a graphical session is
2468 preferred over a text one.
2470 * A minimal systemd-resolved daemon has been added. It
2471 currently simply acts as a companion to systemd-networkd and
2472 manages resolv.conf based on per-interface DNS
2473 configuration, possibly supplied via DHCP. In the long run
2474 we hope to extend this into a local DNSSEC enabled DNS and
2477 * The systemd-networkd-wait-online tool is now enabled by
2478 default. It will delay network-online.target until a network
2479 connection has been configured. The tool primarily integrates
2480 with networkd, but will also make a best effort to make sense
2481 of network configuration performed in some other way.
2483 * Two new service options StartupCPUShares= and
2484 StartupBlockIOWeight= have been added that work similarly to
2485 CPUShares= and BlockIOWeight= however only apply during
2486 system startup. This is useful to prioritize certain services
2487 differently during bootup than during normal runtime.
2489 * hostnamed has been changed to prefer the statically
2490 configured hostname in /etc/hostname (unless set to
2491 'localhost' or empty) over any dynamic one supplied by
2492 dhcp. With this change, the rules for picking the hostname
2493 match more closely the rules of other configuration settings
2494 where the local administrator's configuration in /etc always
2495 overrides any other settings.
2497 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
2498 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
2499 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
2500 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
2501 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
2502 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
2503 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
2504 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
2505 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
2506 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
2507 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
2508 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
2509 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
2510 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
2511 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
2512 Lindskog, WaLyong Cho, Will Woods, Zbigniew
2515 -- Beijing, 2014-05-28
2519 * When restoring the screen brightness at boot, stay away from
2520 the darkest setting or from the lowest 5% of the available
2521 range, depending on which is the larger value of both. This
2522 should effectively protect the user from rebooting into a
2523 black screen, should the brightness have been set to minimum
2526 * sd-login gained a new sd_machine_get_class() call to
2527 determine the class ("vm" or "container") of a machine
2528 registered with machined.
2530 * sd-login gained new calls
2531 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
2532 to query the identity of the peer of a local AF_UNIX
2533 connection. They operate similarly to their sd_pid_get_xyz()
2536 * PID 1 will now maintain a system-wide system state engine
2537 with the states "starting", "running", "degraded",
2538 "maintenance", "stopping". These states are bound to system
2539 startup, normal runtime, runtime with at least one failed
2540 service, rescue/emergency mode and system shutdown. This
2541 state is shown in the "systemctl status" output when no unit
2542 name is passed. It is useful to determine system state, in
2543 particularly when doing so for many systems or containers at
2546 * A new command "list-machines" has been added to "systemctl"
2547 that lists all local OS containers and shows their system
2548 state (see above), if systemd runs inside of them.
2550 * systemctl gained a new "-r" switch to recursively enumerate
2551 units on all local containers, when used with the
2552 "list-unit" command (which is the default one that is
2553 executed when no parameters are specified).
2555 * The GPT automatic partition discovery logic will now honour
2556 two GPT partition flags: one may be set on a partition to
2557 cause it to be mounted read-only, and the other may be set
2558 on a partition to ignore it during automatic discovery.
2560 * Two new GPT type UUIDs have been added for automatic root
2561 partition discovery, for 32-bit and 64-bit ARM. This is not
2562 particularly useful for discovering the root directory on
2563 these architectures during bare-metal boots (since UEFI is
2564 not common there), but still very useful to allow booting of
2565 ARM disk images in nspawn with the -i option.
2567 * MAC addresses of interfaces created with nspawn's
2568 --network-interface= switch will now be generated from the
2569 machine name, and thus be stable between multiple invocations
2572 * logind will now automatically remove all IPC objects owned
2573 by a user if she or he fully logs out. This makes sure that
2574 users who are logged out cannot continue to consume IPC
2575 resources. This covers SysV memory, semaphores and message
2576 queues as well as POSIX shared memory and message
2577 queues. Traditionally, SysV and POSIX IPC had no life-cycle
2578 limits. With this functionality, that is corrected. This may
2579 be turned off by using the RemoveIPC= switch of logind.conf.
2581 * The systemd-machine-id-setup and tmpfiles tools gained a
2582 --root= switch to operate on a specific root directory,
2585 * journald can now forward logged messages to the TTYs of all
2586 logged in users ("wall"). This is the default for all
2587 emergency messages now.
2589 * A new tool systemd-journal-remote has been added to stream
2590 journal log messages across the network.
2592 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
2593 controller trees are mounted into it. Note that the
2594 directories mounted beneath it are not read-only. This is a
2595 security measure and is particularly useful because glibc
2596 actually includes a search logic to pick any tmpfs it can
2597 find to implement shm_open() if /dev/shm is not available
2598 (which it might very well be in namespaced setups).
2600 * machinectl gained a new "poweroff" command to cleanly power
2601 down a local OS container.
2603 * The PrivateDevices= unit file setting will now also drop the
2604 CAP_MKNOD capability from the capability bound set, and
2605 imply DevicePolicy=closed.
2607 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
2608 comprehensively on all long-running systemd services where
2609 this is appropriate.
2611 * systemd-udevd will now run in a disassociated mount
2612 namespace. To mount directories from udev rules, make sure to
2613 pull in mount units via SYSTEMD_WANTS properties.
2615 * The kdbus support gained support for uploading policy into
2616 the kernel. sd-bus gained support for creating "monitoring"
2617 connections that can eavesdrop into all bus communication
2618 for debugging purposes.
2620 * Timestamps may now be specified in seconds since the UNIX
2621 epoch Jan 1st, 1970 by specifying "@" followed by the value
2624 * Native tcpwrap support in systemd has been removed. tcpwrap
2625 is old code, not really maintained anymore and has serious
2626 shortcomings, and better options such as firewalls
2627 exist. For setups that require tcpwrap usage, please
2628 consider invoking your socket-activated service via tcpd,
2629 like on traditional inetd.
2631 * A new system.conf configuration option
2632 DefaultTimerAccuracySec= has been added that controls the
2633 default AccuracySec= setting of .timer units.
2635 * Timer units gained a new WakeSystem= switch. If enabled,
2636 timers configured this way will cause the system to resume
2637 from system suspend (if the system supports that, which most
2640 * Timer units gained a new Persistent= switch. If enabled,
2641 timers configured this way will save to disk when they have
2642 been last triggered. This information is then used on next
2643 reboot to possible execute overdue timer events, that
2644 could not take place because the system was powered off.
2645 This enables simple anacron-like behaviour for timer units.
2647 * systemctl's "list-timers" will now also list the time a
2648 timer unit was last triggered in addition to the next time
2649 it will be triggered.
2651 * systemd-networkd will now assign predictable IPv4LL
2652 addresses to its local interfaces.
2654 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
2655 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
2656 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
2657 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
2658 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
2659 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
2660 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
2661 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
2664 -- Berlin, 2014-03-25
2668 * A new unit file setting RestrictAddressFamilies= has been
2669 added to restrict which socket address families unit
2670 processes gain access to. This takes address family names
2671 like "AF_INET" or "AF_UNIX", and is useful to minimize the
2672 attack surface of services via exotic protocol stacks. This
2673 is built on seccomp system call filters.
2675 * Two new unit file settings RuntimeDirectory= and
2676 RuntimeDirectoryMode= have been added that may be used to
2677 manage a per-daemon runtime directories below /run. This is
2678 an alternative for setting up directory permissions with
2679 tmpfiles snippets, and has the advantage that the runtime
2680 directory's lifetime is bound to the daemon runtime and that
2681 the daemon starts up with an empty directory each time. This
2682 is particularly useful when writing services that drop
2683 privileges using the User= or Group= setting.
2685 * The DeviceAllow= unit setting now supports globbing for
2686 matching against device group names.
2688 * The systemd configuration file system.conf gained new
2689 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
2690 DefaultMemoryAccounting= to globally turn on/off accounting
2691 for specific resources (cgroups) for all units. These
2692 settings may still be overridden individually in each unit
2695 * systemd-gpt-auto-generator is now able to discover /srv and
2696 root partitions in addition to /home and swap partitions. It
2697 also supports LUKS-encrypted partitions now. With this in
2698 place, automatic discovery of partitions to mount following
2699 the Discoverable Partitions Specification
2700 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
2701 is now a lot more complete. This allows booting without
2702 /etc/fstab and without root= on the kernel command line on
2703 systems prepared appropriately.
2705 * systemd-nspawn gained a new --image= switch which allows
2706 booting up disk images and Linux installations on any block
2707 device that follow the Discoverable Partitions Specification
2708 (see above). This means that installations made with
2709 appropriately updated installers may now be started and
2710 deployed using container managers, completely
2711 unmodified. (We hope that libvirt-lxc will add support for
2712 this feature soon, too.)
2714 * systemd-nspawn gained a new --network-macvlan= setting to
2715 set up a private macvlan interface for the
2716 container. Similarly, systemd-networkd gained a new
2717 Kind=macvlan setting in .netdev files.
2719 * systemd-networkd now supports configuring local addresses
2722 * A new tool systemd-network-wait-online has been added to
2723 synchronously wait for network connectivity using
2726 * The sd-bus.h bus API gained a new sd_bus_track object for
2727 tracking the life-cycle of bus peers. Note that sd-bus.h is
2728 still not a public API though (unless you specify
2729 --enable-kdbus on the configure command line, which however
2730 voids your warranty and you get no API stability guarantee).
2732 * The $XDG_RUNTIME_DIR runtime directories for each user are
2733 now individual tmpfs instances, which has the benefit of
2734 introducing separate pools for each user, with individual
2735 size limits, and thus making sure that unprivileged clients
2736 can no longer negatively impact the system or other users by
2737 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
2738 RuntimeDirectorySize= has been introduced that allows
2739 controlling the default size limit for all users. It
2740 defaults to 10% of the available physical memory. This is no
2741 replacement for quotas on tmpfs though (which the kernel
2742 still does not support), as /dev/shm and /tmp are still
2743 shared resources used by both the system and unprivileged
2746 * logind will now automatically turn off automatic suspending
2747 on laptop lid close when more than one display is
2748 connected. This was previously expected to be implemented
2749 individually in desktop environments (such as GNOME),
2750 however has been added to logind now, in order to fix a
2751 boot-time race where a desktop environment might not have
2752 been started yet and thus not been able to take an inhibitor
2753 lock at the time where logind already suspends the system
2754 due to a closed lid.
2756 * logind will now wait at least 30s after each system
2757 suspend/resume cycle, and 3min after system boot before
2758 suspending the system due to a closed laptop lid. This
2759 should give USB docking stations and similar enough time to
2760 be probed and configured after system resume and boot in
2761 order to then act as suspend blocker.
2763 * systemd-run gained a new --property= setting which allows
2764 initialization of resource control properties (and others)
2765 for the created scope or service unit. Example: "systemd-run
2766 --property=BlockIOWeight=10 updatedb" may be used to run
2767 updatedb at a low block IO scheduling weight.
2769 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
2770 now also work in --scope mode.
2772 * When systemd is compiled with kdbus support, basic support
2773 for enforced policies is now in place. (Note that enabling
2774 kdbus still voids your warranty and no API compatibility
2777 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
2778 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
2779 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
2780 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
2781 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
2782 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
2783 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
2784 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
2785 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
2786 Zbigniew Jędrzejewski-Szmek
2788 -- Berlin, 2014-03-12
2792 * systemd will now relabel /dev after loading the SMACK policy
2793 according to SMACK rules.
2795 * A new unit file option AppArmorProfile= has been added to
2796 set the AppArmor profile for the processes of a unit.
2798 * A new condition check ConditionArchitecture= has been added
2799 to conditionalize units based on the system architecture, as
2800 reported by uname()'s "machine" field.
2802 * systemd-networkd now supports matching on the system
2803 virtualization, architecture, kernel command line, host name
2806 * logind is now a lot more aggressive when suspending the
2807 machine due to a closed laptop lid. Instead of acting only
2808 on the lid close action, it will continuously watch the lid
2809 status and act on it. This is useful for laptops where the
2810 power button is on the outside of the chassis so that it can
2811 be reached without opening the lid (such as the Lenovo
2812 Yoga). On those machines, logind will now immediately
2813 re-suspend the machine if the power button has been
2814 accidentally pressed while the laptop was suspended and in a
2815 backpack or similar.
2817 * logind will now watch SW_DOCK switches and inhibit reaction
2818 to the lid switch if it is pressed. This means that logind
2819 will not suspend the machine anymore if the lid is closed
2820 and the system is docked, if the laptop supports SW_DOCK
2821 notifications via the input layer. Note that ACPI docking
2822 stations do not generate this currently. Also note that this
2823 logic is usually not fully sufficient and Desktop
2824 Environments should take a lid switch inhibitor lock when an
2825 external display is connected, as systemd will not watch
2828 * nspawn will now make use of the devices cgroup controller by
2829 default, and only permit creation of and access to the usual
2830 API device nodes like /dev/null or /dev/random, as well as
2831 access to (but not creation of) the pty devices.
2833 * We will now ship a default .network file for
2834 systemd-networkd that automatically configures DHCP for
2835 network interfaces created by nspawn's --network-veth or
2836 --network-bridge= switches.
2838 * systemd will now understand the usual M, K, G, T suffixes
2839 according to SI conventions (i.e. to the base 1000) when
2840 referring to throughput and hardware metrics. It will stay
2841 with IEC conventions (i.e. to the base 1024) for software
2842 metrics, according to what is customary according to
2843 Wikipedia. We explicitly document which base applies for
2844 each configuration option.
2846 * The DeviceAllow= setting in unit files now supports a syntax
2847 to whitelist an entire group of devices node majors at once,
2848 based on the /proc/devices listing. For example, with the
2849 string "char-pts", it is now possible to whitelist all
2850 current and future pseudo-TTYs at once.
2852 * sd-event learned a new "post" event source. Event sources of
2853 this type are triggered by the dispatching of any event
2854 source of a type that is not "post". This is useful for
2855 implementing clean-up and check event sources that are
2856 triggered by other work being done in the program.
2858 * systemd-networkd is no longer statically enabled, but uses
2859 the usual [Install] sections so that it can be
2860 enabled/disabled using systemctl. It still is enabled by
2863 * When creating a veth interface pair with systemd-nspawn, the
2864 host side will now be prefixed with "vb-" if
2865 --network-bridge= is used, and with "ve-" if --network-veth
2866 is used. This way, it is easy to distinguish these cases on
2867 the host, for example to apply different configuration to
2868 them with systemd-networkd.
2870 * The compatibility libraries for libsystemd-journal.so,
2871 libsystem-id128.so, libsystemd-login.so and
2872 libsystemd-daemon.so do not make use of IFUNC
2873 anymore. Instead, we now build libsystemd.so multiple times
2874 under these alternative names. This means that the footprint
2875 is drastically increased, but given that these are
2876 transitional compatibility libraries, this should not matter
2877 much. This change has been made necessary to support the ARM
2878 platform for these compatibility libraries, as the ARM
2879 toolchain is not really at the same level as the toolchain
2880 for other architectures like x86 and does not support
2881 IFUNC. Please make sure to use --enable-compat-libs only
2882 during a transitional period!
2884 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
2885 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
2886 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
2887 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
2888 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
2889 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
2890 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
2891 Zbigniew Jędrzejewski-Szmek
2893 -- Berlin, 2014-02-24
2897 * A new component "systemd-networkd" has been added that can
2898 be used to configure local network interfaces statically or
2899 via DHCP. It is capable of bringing up bridges, VLANs, and
2900 bonding. Currently, no hook-ups for interactive network
2901 configuration are provided. Use this for your initrd,
2902 container, embedded, or server setup if you need a simple,
2903 yet powerful, network configuration solution. This
2904 configuration subsystem is quite nifty, as it allows wildcard
2905 hotplug matching in interfaces. For example, with a single
2906 configuration snippet, you can configure that all Ethernet
2907 interfaces showing up are automatically added to a bridge,
2908 or similar. It supports link-sensing and more.
2910 * A new tool "systemd-socket-proxyd" has been added which can
2911 act as a bidirectional proxy for TCP sockets. This is
2912 useful for adding socket activation support to services that
2913 do not actually support socket activation, including virtual
2914 machines and the like.
2916 * Add a new tool to save/restore rfkill state on
2919 * Save/restore state of keyboard backlights in addition to
2920 display backlights on shutdown/boot.
2922 * udev learned a new SECLABEL{} construct to label device
2923 nodes with a specific security label when they appear. For
2924 now, only SECLABEL{selinux} is supported, but the syntax is
2925 prepared for additional security frameworks.
2927 * udev gained a new scheme to configure link-level attributes
2928 from files in /etc/systemd/network/*.link. These files can
2929 match against MAC address, device path, driver name and type,
2930 and will apply attributes like the naming policy, link speed,
2931 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
2932 address assignment policy (randomized, ...).
2934 * The configuration of network interface naming rules for
2935 "permanent interface names" has changed: a new NamePolicy=
2936 setting in the [Link] section of .link files determines the
2937 priority of possible naming schemes (onboard, slot, MAC,
2938 path). The default value of this setting is determined by
2939 /usr/lib/net/links/99-default.link. Old
2940 80-net-name-slot.rules udev configuration file has been
2941 removed, so local configuration overriding this file should
2942 be adapated to override 99-default.link instead.
2944 * When the User= switch is used in a unit file, also
2945 initialize $SHELL= based on the user database entry.
2947 * systemd no longer depends on libdbus. All communication is
2948 now done with sd-bus, systemd's low-level bus library
2951 * kdbus support has been added to PID 1 itself. When kdbus is
2952 enabled, this causes PID 1 to set up the system bus and
2953 enable support for a new ".busname" unit type that
2954 encapsulates bus name activation on kdbus. It works a little
2955 bit like ".socket" units, except for bus names. A new
2956 generator has been added that converts classic dbus1 service
2957 activation files automatically into native systemd .busname
2960 * sd-bus: add a light-weight vtable implementation that allows
2961 defining objects on the bus with a simple static const
2962 vtable array of its methods, signals and properties.
2964 * systemd will not generate or install static dbus
2965 introspection data anymore to /usr/share/dbus-1/interfaces,
2966 as the precise format of these files is unclear, and
2967 nothing makes use of it.
2969 * A proxy daemon is now provided to proxy clients connecting
2970 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
2971 compatibility with classic D-Bus.
2973 * A bus driver implementation has been added that supports the
2974 classic D-Bus bus driver calls on kdbus, also for
2975 compatibility purposes.
2977 * A new API "sd-event.h" has been added that implements a
2978 minimal event loop API built around epoll. It provides a
2979 couple of features that direct epoll usage is lacking:
2980 prioritization of events, scales to large numbers of timer
2981 events, per-event timer slack (accuracy), system-wide
2982 coalescing of timer events, exit handlers, watchdog
2983 supervision support using systemd's sd_notify() API, child
2986 * A new API "sd-rntl.h" has been added that provides an API
2987 around the route netlink interface of the kernel, similar in
2988 style to "sd-bus.h".
2990 * A new API "sd-dhcp-client.h" has been added that provides a
2991 small DHCPv4 client-side implementation. This is used by
2994 * There is a new kernel command line option
2995 "systemd.restore_state=0|1". When set to "0", none of the
2996 systemd tools will restore saved runtime state to hardware
2997 devices. More specifically, the rfkill and backlight states
3000 * The FsckPassNo= compatibility option in mount/service units
3001 has been removed. The fstab generator will now add the
3002 necessary dependencies automatically, and does not require
3003 PID1's support for that anymore.
3005 * journalctl gained a new switch, --list-boots, that lists
3006 recent boots with their times and boot IDs.
3008 * The various tools like systemctl, loginctl, timedatectl,
3009 busctl, systemd-run, ... have gained a new switch "-M" to
3010 connect to a specific, local OS container (as direct
3011 connection, without requiring SSH). This works on any
3012 container that is registered with machined, such as those
3013 created by libvirt-lxc or nspawn.
3015 * systemd-run and systemd-analyze also gained support for "-H"
3016 to connect to remote hosts via SSH. This is particularly
3017 useful for systemd-run because it enables queuing of jobs
3018 onto remote systems.
3020 * machinectl gained a new command "login" to open a getty
3021 login in any local container. This works with any container
3022 that is registered with machined (such as those created by
3023 libvirt-lxc or nspawn), and which runs systemd inside.
3025 * machinectl gained a new "reboot" command that may be used to
3026 trigger a reboot on a specific container that is registered
3027 with machined. This works on any container that runs an init
3028 system of some kind.
3030 * systemctl gained a new "list-timers" command to print a nice
3031 listing of installed timer units with the times they elapse
3034 * Alternative reboot() parameters may now be specified on the
3035 "systemctl reboot" command line and are passed to the
3036 reboot() system call.
3038 * systemctl gained a new --job-mode= switch to configure the
3039 mode to queue a job with. This is a more generic version of
3040 --fail, --irreversible, and --ignore-dependencies, which are
3041 still available but not advertised anymore.
3043 * /etc/systemd/system.conf gained new settings to configure
3044 various default timeouts of units, as well as the default
3045 start limit interval and burst. These may still be overridden
3048 * PID1 will now export on the bus profile data of the security
3049 policy upload process (such as the SELinux policy upload to
3052 * journald: when forwarding logs to the console, include
3053 timestamps (following the setting in
3054 /sys/module/printk/parameters/time).
3056 * OnCalendar= in timer units now understands the special
3057 strings "yearly" and "annually". (Both are equivalent)
3059 * The accuracy of timer units is now configurable with the new
3060 AccuracySec= setting. It defaults to 1min.
3062 * A new dependency type JoinsNamespaceOf= has been added that
3063 allows running two services within the same /tmp and network
3064 namespace, if PrivateNetwork= or PrivateTmp= are used.
3066 * A new command "cat" has been added to systemctl. It outputs
3067 the original unit file of a unit, and concatenates the
3068 contents of additional "drop-in" unit file snippets, so that
3069 the full configuration is shown.
3071 * systemctl now supports globbing on the various "list-xyz"
3072 commands, like "list-units" or "list-sockets", as well as on
3073 those commands which take multiple unit names.
3075 * journalctl's --unit= switch gained support for globbing.
3077 * All systemd daemons now make use of the watchdog logic so
3078 that systemd automatically notices when they hang.
3080 * If the $container_ttys environment variable is set,
3081 getty-generator will automatically spawn a getty for each
3082 listed tty. This is useful for container managers to request
3083 login gettys to be spawned on as many ttys as needed.
3085 * %h, %s, %U specifier support is not available anymore when
3086 used in unit files for PID 1. This is because NSS calls are
3087 not safe from PID 1. They stay available for --user
3088 instances of systemd, and as special case for the root user.
3090 * loginctl gained a new "--no-legend" switch to turn off output
3093 * The "sd-login.h" API gained three new calls:
3094 sd_session_is_remote(), sd_session_get_remote_user(),
3095 sd_session_get_remote_host() to query information about
3098 * The udev hardware database now also carries vendor/product
3099 information of SDIO devices.
3101 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
3102 determine whether watchdog notifications are requested by
3105 * Socket-activated per-connection services now include a
3106 short description of the connection parameters in the
3109 * tmpfiles gained a new "--boot" option. When this is not used,
3110 only lines where the command character is not suffixed with
3111 "!" are executed. When this option is specified, those
3112 options are executed too. This partitions tmpfiles
3113 directives into those that can be safely executed at any
3114 time, and those which should be run only at boot (for
3115 example, a line that creates /run/nologin).
3117 * A new API "sd-resolve.h" has been added which provides a simple
3118 asynchronous wrapper around glibc NSS host name resolution
3119 calls, such as getaddrinfo(). In contrast to glibc's
3120 getaddrinfo_a(), it does not use signals. In contrast to most
3121 other asynchronous name resolution libraries, this one does
3122 not reimplement DNS, but reuses NSS, so that alternate
3123 host name resolution systems continue to work, such as mDNS,
3124 LDAP, etc. This API is based on libasyncns, but it has been
3125 cleaned up for inclusion in systemd.
3127 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
3128 "sd-daemon.h" are no longer found in individual libraries
3129 libsystemd-journal.so, libsystemd-login.so,
3130 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
3131 merged them into a single library, libsystemd.so, which
3132 provides all symbols. The reason for this is cyclic
3133 dependencies, as these libraries tend to use each other's
3134 symbols. So far, we have managed to workaround that by linking
3135 a copy of a good part of our code into each of these
3136 libraries again and again, which, however, makes certain
3137 things hard to do, like sharing static variables. Also, it
3138 substantially increases footprint. With this change, there
3139 is only one library for the basic APIs systemd
3140 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
3141 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
3142 library as well, however are subject to the --enable-kdbus
3143 switch (see below). Note that "sd-dhcp-client.h" is not part
3144 of this library (this is because it only consumes, never
3145 provides, services of/to other APIs). To make the transition
3146 easy from the separate libraries to the unified one, we
3147 provide the --enable-compat-libs compile-time switch which
3148 will generate stub libraries that are compatible with the
3149 old ones but redirect all calls to the new one.
3151 * All of the kdbus logic and the new APIs "sd-bus.h",
3152 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
3153 and "sd-utf8.h" are compile-time optional via the
3154 "--enable-kdbus" switch, and they are not compiled in by
3155 default. To make use of kdbus, you have to explicitly enable
3156 the switch. Note however, that neither the kernel nor the
3157 userspace API for all of this is considered stable yet. We
3158 want to maintain the freedom to still change the APIs for
3159 now. By specifying this build-time switch, you acknowledge
3160 that you are aware of the instability of the current
3163 * Also, note that while kdbus is pretty much complete,
3164 it lacks one thing: proper policy support. This means you
3165 can build a fully working system with all features; however,
3166 it will be highly insecure. Policy support will be added in
3167 one of the next releases, at the same time that we will
3168 declare the APIs stable.
3170 * When the kernel command line argument "kdbus" is specified,
3171 systemd will automatically load the kdbus.ko kernel module. At
3172 this stage of development, it is only useful for testing kdbus
3173 and should not be used in production. Note: if "--enable-kdbus"
3174 is specified, and the kdbus.ko kernel module is available, and
3175 "kdbus" is added to the kernel command line, the entire system
3176 runs with kdbus instead of dbus-daemon, with the above mentioned
3177 problem of missing the system policy enforcement. Also a future
3178 version of kdbus.ko or a newer systemd will not be compatible with
3179 each other, and will unlikely be able to boot the machine if only
3180 one of them is updated.
3182 * systemctl gained a new "import-environment" command which
3183 uploads the caller's environment (or parts thereof) into the
3184 service manager so that it is inherited by services started
3185 by the manager. This is useful to upload variables like
3186 $DISPLAY into the user service manager.
3188 * A new PrivateDevices= switch has been added to service units
3189 which allows running a service with a namespaced /dev
3190 directory that does not contain any device nodes for
3191 physical devices. More specifically, it only includes devices
3192 such as /dev/null, /dev/urandom, and /dev/zero which are API
3195 * logind has been extended to support behaviour like VT
3196 switching on seats that do not support a VT. This makes
3197 multi-session available on seats that are not the first seat
3198 (seat0), and on systems where kernel support for VTs has
3199 been disabled at compile-time.
3201 * If a process holds a delay lock for system sleep or shutdown
3202 and fails to release it in time, we will now log its
3203 identity. This makes it easier to identify processes that
3204 cause slow suspends or power-offs.
3206 * When parsing /etc/crypttab, support for a new key-slot=
3207 option as supported by Debian is added. It allows indicating
3208 which LUKS slot to use on disk, speeding up key loading.
3210 * The sd_journald_sendv() API call has been checked and
3211 officially declared to be async-signal-safe so that it may
3212 be invoked from signal handlers for logging purposes.
3214 * Boot-time status output is now enabled automatically after a
3215 short timeout if boot does not progress, in order to give
3216 the user an indication what she or he is waiting for.
3218 * The boot-time output has been improved to show how much time
3219 remains until jobs expire.
3221 * The KillMode= switch in service units gained a new possible
3222 value "mixed". If set, and the unit is shut down, then the
3223 initial SIGTERM signal is sent only to the main daemon
3224 process, while the following SIGKILL signal is sent to
3225 all remaining processes of the service.
3227 * When a scope unit is registered, a new property "Controller"
3228 may be set. If set to a valid bus name, systemd will send a
3229 RequestStop() signal to this name when it would like to shut
3230 down the scope. This may be used to hook manager logic into
3231 the shutdown logic of scope units. Also, scope units may now
3232 be put in a special "abandoned" state, in which case the
3233 manager process which created them takes no further
3234 responsibilities for it.
3236 * When reading unit files, systemd will now verify
3237 the access mode of these files, and warn about certain
3238 suspicious combinations. This has been added to make it
3239 easier to track down packaging bugs where unit files are
3240 marked executable or world-writable.
3242 * systemd-nspawn gained a new "--setenv=" switch to set
3243 container-wide environment variables. The similar option in
3244 systemd-activate was renamed from "--environment=" to
3245 "--setenv=" for consistency.
3247 * systemd-nspawn has been updated to create a new kdbus domain
3248 for each container that is invoked, thus allowing each
3249 container to have its own set of system and user buses,
3250 independent of the host.
3252 * systemd-nspawn gained a new --drop-capability= switch to run
3253 the container with less capabilities than the default. Both
3254 --drop-capability= and --capability= now take the special
3255 string "all" for dropping or keeping all capabilities.
3257 * systemd-nspawn gained new switches for executing containers
3258 with specific SELinux labels set.
3260 * systemd-nspawn gained a new --quiet switch to not generate
3261 any additional output but the container's own console
3264 * systemd-nspawn gained a new --share-system switch to run a
3265 container without PID namespacing enabled.
3267 * systemd-nspawn gained a new --register= switch to control
3268 whether the container is registered with systemd-machined or
3269 not. This is useful for containers that do not run full
3270 OS images, but only specific apps.
3272 * systemd-nspawn gained a new --keep-unit which may be used
3273 when invoked as the only program from a service unit, and
3274 results in registration of the unit service itself in
3275 systemd-machined, instead of a newly opened scope unit.
3277 * systemd-nspawn gained a new --network-interface= switch for
3278 moving arbitrary interfaces to the container. The new
3279 --network-veth switch creates a virtual Ethernet connection
3280 between host and container. The new --network-bridge=
3281 switch then allows assigning the host side of this virtual
3282 Ethernet connection to a bridge device.
3284 * systemd-nspawn gained a new --personality= switch for
3285 setting the kernel personality for the container. This is
3286 useful when running a 32-bit container on a 64-bit host. A
3287 similar option Personality= is now also available for service
3290 * logind will now also track a "Desktop" identifier for each
3291 session which encodes the desktop environment of it. This is
3292 useful for desktop environments that want to identify
3293 multiple running sessions of itself easily.
3295 * A new SELinuxContext= setting for service units has been
3296 added that allows setting a specific SELinux execution
3297 context for a service.
3299 * Most systemd client tools will now honour $SYSTEMD_LESS for
3300 settings of the "less" pager. By default, these tools will
3301 override $LESS to allow certain operations to work, such as
3302 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
3303 influence this logic.
3305 * systemd's "seccomp" hook-up has been changed to make use of
3306 the libseccomp library instead of using its own
3307 implementation. This has benefits for portability among
3310 * For usage together with SystemCallFilter=, a new
3311 SystemCallErrorNumber= setting has been introduced that
3312 allows configuration of a system error number to be returned
3313 on filtered system calls, instead of immediately killing the
3314 process. Also, SystemCallArchitectures= has been added to
3315 limit access to system calls of a particular architecture
3316 (in order to turn off support for unused secondary
3317 architectures). There is also a global
3318 SystemCallArchitectures= setting in system.conf now to turn
3319 off support for non-native system calls system-wide.
3321 * systemd requires a kernel with a working name_to_handle_at(),
3322 please see the kernel config requirements in the README file.
3324 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
3325 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
3326 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
3327 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
3328 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
3329 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
3330 Elia Pinto, Florian Weimer, George McCollister, Goffredo
3331 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
3332 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
3333 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
3334 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
3335 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
3336 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
3337 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
3338 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
3339 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
3340 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
3341 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
3342 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
3343 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
3344 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
3345 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
3346 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
3347 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
3349 -- Berlin, 2014-02-20
3353 * logind has gained support for facilitating privileged input
3354 and drm device access for unprivileged clients. This work is
3355 useful to allow Wayland display servers (and similar
3356 programs, such as kmscon) to run under the user's ID and
3357 access input and drm devices which are normally
3358 protected. When this is used (and the kernel is new enough)
3359 logind will "mute" IO on the file descriptors passed to
3360 Wayland as long as it is in the background and "unmute" it
3361 if it returns into the foreground. This allows secure
3362 session switching without allowing background sessions to
3363 eavesdrop on input and display data. This also introduces
3364 session switching support if VT support is turned off in the
3365 kernel, and on seats that are not seat0.
3367 * A new kernel command line option luks.options= is understood
3368 now which allows specifying LUKS options for usage for LUKS
3369 encrypted partitions specified with luks.uuid=.
3371 * tmpfiles.d(5) snippets may now use specifier expansion in
3372 path names. More specifically %m, %b, %H, %v, are now
3373 replaced by the local machine id, boot id, hostname, and
3374 kernel version number.
3376 * A new tmpfiles.d(5) command "m" has been introduced which
3377 may be used to change the owner/group/access mode of a file
3378 or directory if it exists, but do nothing if it does not.
3380 * This release removes high-level support for the
3381 MemorySoftLimit= cgroup setting. The underlying kernel
3382 cgroup attribute memory.soft_limit= is currently badly
3383 designed and likely to be removed from the kernel API in its
3384 current form, hence we should not expose it for now.
3386 * The memory.use_hierarchy cgroup attribute is now enabled for
3387 all cgroups systemd creates in the memory cgroup
3388 hierarchy. This option is likely to be come the built-in
3389 default in the kernel anyway, and the non-hierarchical mode
3390 never made much sense in the intrinsically hierarchical
3393 * A new field _SYSTEMD_SLICE= is logged along with all journal
3394 messages containing the slice a message was generated
3395 from. This is useful to allow easy per-customer filtering of
3396 logs among other things.
3398 * systemd-journald will no longer adjust the group of journal
3399 files it creates to the "systemd-journal" group. Instead we
3400 rely on the journal directory to be owned by the
3401 "systemd-journal" group, and its setgid bit set, so that the
3402 kernel file system layer will automatically enforce that
3403 journal files inherit this group assignment. The reason for
3404 this change is that we cannot allow NSS look-ups from
3405 journald which would be necessary to resolve
3406 "systemd-journal" to a numeric GID, because this might
3407 create deadlocks if NSS involves synchronous queries to
3408 other daemons (such as nscd, or sssd) which in turn are
3409 logging clients of journald and might block on it, which
3410 would then dead lock. A tmpfiles.d(5) snippet included in
3411 systemd will make sure the setgid bit and group are
3412 properly set on the journal directory if it exists on every
3413 boot. However, we recommend adjusting it manually after
3414 upgrades too (or from RPM scriptlets), so that the change is
3415 not delayed until next reboot.
3417 * Backlight and random seed files in /var/lib/ have moved into
3418 the /var/lib/systemd/ directory, in order to centralize all
3419 systemd generated files in one directory.
3421 * Boot time performance measurements (as displayed by
3422 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
3423 performance information if that's available to determine how
3424 much time BIOS and boot loader initialization required. With
3425 a sufficiently new BIOS you hence no longer need to boot
3426 with Gummiboot to get access to such information.
3428 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
3429 Cristian Rodríguez, Dave Reisner, David Herrmann, David
3430 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
3431 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
3432 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
3433 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
3434 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3436 -- Berlin, 2013-10-02
3440 * The Restart= option for services now understands a new
3441 on-watchdog setting, which will restart the service
3442 automatically if the service stops sending out watchdog keep
3443 alive messages (as configured with WatchdogSec=).
3445 * The getty generator (which is responsible for bringing up a
3446 getty on configured serial consoles) will no longer only
3447 start a getty on the primary kernel console but on all
3448 others, too. This makes the order in which console= is
3449 specified on the kernel command line less important.
3451 * libsystemd-logind gained a new sd_session_get_vt() call to
3452 retrieve the VT number of a session.
3454 * If the option "tries=0" is set for an entry of /etc/crypttab
3455 its passphrase is queried indefinitely instead of any
3456 maximum number of tries.
3458 * If a service with a configure PID file terminates its PID
3459 file will now be removed automatically if it still exists
3460 afterwards. This should put an end to stale PID files.
3462 * systemd-run will now also take relative binary path names
3463 for execution and no longer insists on absolute paths.
3465 * InaccessibleDirectories= and ReadOnlyDirectories= now take
3466 paths that are optionally prefixed with "-" to indicate that
3467 it should not be considered a failure if they do not exist.
3469 * journalctl -o (and similar commands) now understands a new
3470 output mode "short-precise", it is similar to "short" but
3471 shows timestamps with usec accuracy.
3473 * The option "discard" (as known from Debian) is now
3474 synonymous to "allow-discards" in /etc/crypttab. In fact,
3475 "discard" is preferred now (since it is easier to remember
3478 * Some licensing clean-ups were made, so that more code is now
3479 LGPL-2.1 licensed than before.
3481 * A minimal tool to save/restore the display backlight
3482 brightness across reboots has been added. It will store the
3483 backlight setting as late as possible at shutdown, and
3484 restore it as early as possible during reboot.
3486 * A logic to automatically discover and enable home and swap
3487 partitions on GPT disks has been added. With this in place
3488 /etc/fstab becomes optional for many setups as systemd can
3489 discover certain partitions located on the root disk
3490 automatically. Home partitions are recognized under their
3491 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
3492 partitions are recognized under their GPT type ID
3493 0657fd6da4ab43c484e50933c84b4f4f.
3495 * systemd will no longer pass any environment from the kernel
3496 or initrd to system services. If you want to set an
3497 environment for all services, do so via the kernel command
3498 line systemd.setenv= assignment.
3500 * The systemd-sysctl tool no longer natively reads the file
3501 /etc/sysctl.conf. If desired, the file should be symlinked
3502 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
3503 legacy support by a symlink rather than built-in code, it
3504 also makes the otherwise hidden order of application of the
3505 different files visible. (Note that this partly reverts to a
3506 pre-198 application order of sysctl knobs!)
3508 * The "systemctl set-log-level" and "systemctl dump" commands
3509 have been moved to systemd-analyze.
3511 * systemd-run learned the new --remain-after-exit switch,
3512 which causes the scope unit not to be cleaned up
3513 automatically after the process terminated.
3515 * tmpfiles learned a new --exclude-prefix= switch to exclude
3516 certain paths from operation.
3518 * journald will now automatically flush all messages to disk
3519 as soon as a message at the log level CRIT, ALERT or EMERG
3522 Contributions from: Andrew Cook, Brandon Philips, Christian
3523 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
3524 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
3525 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
3526 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
3527 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
3528 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
3529 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
3530 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
3531 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
3532 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
3533 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
3534 William Giokas, Zbigniew Jędrzejewski-Szmek
3536 -- Berlin, 2013-09-13
3540 * The documentation has been updated to cover the various new
3541 concepts introduced with 205.
3543 * Unit files now understand the new %v specifier which
3544 resolves to the kernel version string as returned by "uname
3547 * systemctl now supports filtering the unit list output by
3548 load state, active state and sub state, using the new
3551 * "systemctl status" will now show the results of the
3552 condition checks (like ConditionPathExists= and similar) of
3553 the last start attempts of the unit. They are also logged to
3556 * "journalctl -b" may now be used to look for boot output of a
3557 specific boot. Try "journalctl -b -1" for the previous boot,
3558 but the syntax is substantially more powerful.
3560 * "journalctl --show-cursor" has been added which prints the
3561 cursor string the last shown log line. This may then be used
3562 with the new "journalctl --after-cursor=" switch to continue
3563 browsing logs from that point on.
3565 * "journalctl --force" may now be used to force regeneration
3568 * Creation of "dead" device nodes has been moved from udev
3569 into kmod and tmpfiles. Previously, udev would read the kmod
3570 databases to pre-generate dead device nodes based on meta
3571 information contained in kernel modules, so that these would
3572 be auto-loaded on access rather then at boot. As this
3573 does not really have much to do with the exposing actual
3574 kernel devices to userspace this has always been slightly
3575 alien in the udev codebase. Following the new scheme kmod
3576 will now generate a runtime snippet for tmpfiles from the
3577 module meta information and it now is tmpfiles' job to the
3578 create the nodes. This also allows overriding access and
3579 other parameters for the nodes using the usual tmpfiles
3580 facilities. As side effect this allows us to remove the
3581 CAP_SYS_MKNOD capability bit from udevd entirely.
3583 * logind's device ACLs may now be applied to these "dead"
3584 devices nodes too, thus finally allowing managed access to
3585 devices such as /dev/snd/sequencer whithout loading the
3586 backing module right-away.
3588 * A new RPM macro has been added that may be used to apply
3589 tmpfiles configuration during package installation.
3591 * systemd-detect-virt and ConditionVirtualization= now can
3592 detect User-Mode-Linux machines (UML).
3594 * journald will now implicitly log the effective capabilities
3595 set of processes in the message metadata.
3597 * systemd-cryptsetup has gained support for TrueCrypt volumes.
3599 * The initrd interface has been simplified (more specifically,
3600 support for passing performance data via environment
3601 variables and fsck results via files in /run has been
3602 removed). These features were non-essential, and are
3603 nowadays available in a much nicer way by having systemd in
3604 the initrd serialize its state and have the hosts systemd
3605 deserialize it again.
3607 * The udev "keymap" data files and tools to apply keyboard
3608 specific mappings of scan to key codes, and force-release
3609 scan code lists have been entirely replaced by a udev
3610 "keyboard" builtin and a hwdb data file.
3612 * systemd will now honour the kernel's "quiet" command line
3613 argument also during late shutdown, resulting in a
3614 completely silent shutdown when used.
3616 * There's now an option to control the SO_REUSEPORT socket
3617 option in .socket units.
3619 * Instance units will now automatically get a per-template
3620 subslice of system.slice unless something else is explicitly
3621 configured. For example, instances of sshd@.service will now
3622 implicitly be placed in system-sshd.slice rather than
3623 system.slice as before.
3625 * Test coverage support may now be enabled at build time.
3627 Contributions from: Dave Reisner, Frederic Crozat, Harald
3628 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
3629 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
3630 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
3631 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
3632 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
3633 Giokas, Zbigniew Jędrzejewski-Szmek
3635 -- Berlin, 2013-07-23
3639 * Two new unit types have been introduced:
3641 Scope units are very similar to service units, however, are
3642 created out of pre-existing processes -- instead of PID 1
3643 forking off the processes. By using scope units it is
3644 possible for system services and applications to group their
3645 own child processes (worker processes) in a powerful way
3646 which then maybe used to organize them, or kill them
3647 together, or apply resource limits on them.
3649 Slice units may be used to partition system resources in an
3650 hierarchical fashion and then assign other units to them. By
3651 default there are now three slices: system.slice (for all
3652 system services), user.slice (for all user sessions),
3653 machine.slice (for VMs and containers).
3655 Slices and scopes have been introduced primarily in
3656 context of the work to move cgroup handling to a
3657 single-writer scheme, where only PID 1
3658 creates/removes/manages cgroups.
3660 * There's a new concept of "transient" units. In contrast to
3661 normal units these units are created via an API at runtime,
3662 not from configuration from disk. More specifically this
3663 means it is now possible to run arbitrary programs as
3664 independent services, with all execution parameters passed
3665 in via bus APIs rather than read from disk. Transient units
3666 make systemd substantially more dynamic then it ever was,
3667 and useful as a general batch manager.
3669 * logind has been updated to make use of scope and slice units
3670 for managing user sessions. As a user logs in he will get
3671 his own private slice unit, to which all sessions are added
3672 as scope units. We also added support for automatically
3673 adding an instance of user@.service for the user into the
3674 slice. Effectively logind will no longer create cgroup
3675 hierarchies on its own now, it will defer entirely to PID 1
3676 for this by means of scope, service and slice units. Since
3677 user sessions this way become entities managed by PID 1
3678 the output of "systemctl" is now a lot more comprehensive.
3680 * A new mini-daemon "systemd-machined" has been added which
3681 may be used by virtualization managers to register local
3682 VMs/containers. nspawn has been updated accordingly, and
3683 libvirt will be updated shortly. machined will collect a bit
3684 of meta information about the VMs/containers, and assign
3685 them their own scope unit (see above). The collected
3686 meta-data is then made available via the "machinectl" tool,
3687 and exposed in "ps" and similar tools. machined/machinectl
3688 is compile-time optional.
3690 * As discussed earlier, the low-level cgroup configuration
3691 options ControlGroup=, ControlGroupModify=,
3692 ControlGroupPersistent=, ControlGroupAttribute= have been
3693 removed. Please use high-level attribute settings instead as
3694 well as slice units.
3696 * A new bus call SetUnitProperties() has been added to alter
3697 various runtime parameters of a unit. This is primarily
3698 useful to alter cgroup parameters dynamically in a nice way,
3699 but will be extended later on to make more properties
3700 modifiable at runtime. systemctl gained a new set-properties
3701 command that wraps this call.
3703 * A new tool "systemd-run" has been added which can be used to
3704 run arbitrary command lines as transient services or scopes,
3705 while configuring a number of settings via the command
3706 line. This tool is currently very basic, however already
3707 very useful. We plan to extend this tool to even allow
3708 queuing of execution jobs with time triggers from the
3709 command line, similar in fashion to "at".
3711 * nspawn will now inform the user explicitly that kernels with
3712 audit enabled break containers, and suggest the user to turn
3715 * Support for detecting the IMA and AppArmor security
3716 frameworks with ConditionSecurity= has been added.
3718 * journalctl gained a new "-k" switch for showing only kernel
3719 messages, mimicking dmesg output; in addition to "--user"
3720 and "--system" switches for showing only user's own logs
3723 * systemd-delta can now show information about drop-in
3724 snippets extending unit files.
3726 * libsystemd-bus has been substantially updated but is still
3727 not available as public API.
3729 * systemd will now look for the "debug" argument on the kernel
3730 command line and enable debug logging, similar to what
3731 "systemd.log_level=debug" already did before.
3733 * "systemctl set-default", "systemctl get-default" has been
3734 added to configure the default.target symlink, which
3735 controls what to boot into by default.
3737 * "systemctl set-log-level" has been added as a convenient
3738 way to raise and lower systemd logging threshold.
3740 * "systemd-analyze plot" will now show the time the various
3741 generators needed for execution, as well as information
3742 about the unit file loading.
3744 * libsystemd-journal gained a new sd_journal_open_files() call
3745 for opening specific journal files. journactl also gained a
3746 new switch to expose this new functionality. Previously we
3747 only supported opening all files from a directory, or all
3748 files from the system, as opening individual files only is
3749 racy due to journal file rotation.
3751 * systemd gained the new DefaultEnvironment= setting in
3752 /etc/systemd/system.conf to set environment variables for
3755 * If a privileged process logs a journal message with the
3756 OBJECT_PID= field set, then journald will automatically
3757 augment this with additional OBJECT_UID=, OBJECT_GID=,
3758 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
3759 system services want to log events about specific client
3760 processes. journactl/systemctl has been updated to make use
3761 of this information if all log messages regarding a specific
3764 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
3765 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
3766 Reisner, David Coppa, David King, David Strauss, Eelco
3767 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
3768 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
3769 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
3770 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
3771 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
3772 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
3773 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
3774 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
3775 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
3776 Łukasz Stelmach, 장동준
3780 * The Python bindings gained some minimal support for the APIs
3781 exposed by libsystemd-logind.
3783 * ConditionSecurity= gained support for detecting SMACK. Since
3784 this condition already supports SELinux and AppArmor we only
3785 miss IMA for this. Patches welcome!
3787 Contributions from: Karol Lewandowski, Lennart Poettering,
3788 Zbigniew Jędrzejewski-Szmek
3792 * systemd-nspawn will now create /etc/resolv.conf if
3793 necessary, before bind-mounting the host's file onto it.
3795 * systemd-nspawn will now store meta information about a
3796 container on the container's cgroup as extended attribute
3797 fields, including the root directory.
3799 * The cgroup hierarchy has been reworked in many ways. All
3800 objects any of the components systemd creates in the cgroup
3801 tree are now suffixed. More specifically, user sessions are
3802 now placed in cgroups suffixed with ".session", users in
3803 cgroups suffixed with ".user", and nspawn containers in
3804 cgroups suffixed with ".nspawn". Furthermore, all cgroup
3805 names are now escaped in a simple scheme to avoid collision
3806 of userspace object names with kernel filenames. This work
3807 is preparation for making these objects relocatable in the
3808 cgroup tree, in order to allow easy resource partitioning of
3809 these objects without causing naming conflicts.
3811 * systemctl list-dependencies gained the new switches
3812 --plain, --reverse, --after and --before.
3814 * systemd-inhibit now shows the process name of processes that
3815 have taken an inhibitor lock.
3817 * nss-myhostname will now also resolve "localhost"
3818 implicitly. This makes /etc/hosts an optional file and
3819 nicely handles that on IPv6 ::1 maps to both "localhost" and
3822 * libsystemd-logind.so gained a new call
3823 sd_get_machine_names() to enumerate running containers and
3824 VMs (currently only supported by very new libvirt and
3825 nspawn). sd_login_monitor can now be used to watch
3826 VMs/containers coming and going.
3828 * .include is not allowed recursively anymore, and only in
3829 unit files. Usually it is better to use drop-in snippets in
3830 .d/*.conf anyway, as introduced with systemd 198.
3832 * systemd-analyze gained a new "critical-chain" command that
3833 determines the slowest chain of units run during system
3834 boot-up. It is very useful for tracking down where
3835 optimizing boot time is the most beneficial.
3837 * systemd will no longer allow manipulating service paths in
3838 the name=systemd:/system cgroup tree using ControlGroup= in
3839 units. (But is still fine with it in all other dirs.)
3841 * There's a new systemd-nspawn@.service service file that may
3842 be used to easily run nspawn containers as system
3843 services. With the container's root directory in
3844 /var/lib/container/foobar it is now sufficient to run
3845 "systemctl start systemd-nspawn@foobar.service" to boot it.
3847 * systemd-cgls gained a new parameter "--machine" to list only
3848 the processes within a certain container.
3850 * ConditionSecurity= now can check for "apparmor". We still
3851 are lacking checks for SMACK and IMA for this condition
3852 check though. Patches welcome!
3854 * A new configuration file /etc/systemd/sleep.conf has been
3855 added that may be used to configure which kernel operation
3856 systemd is supposed to execute when "suspend", "hibernate"
3857 or "hybrid-sleep" is requested. This makes the new kernel
3858 "freeze" state accessible to the user.
3860 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
3861 the passed argument if applicable.
3863 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
3864 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
3865 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
3866 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
3867 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
3868 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
3869 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
3874 * The output of 'systemctl list-jobs' got some polishing. The
3875 '--type=' argument may now be passed more than once. A new
3876 command 'systemctl list-sockets' has been added which shows
3877 a list of kernel sockets systemd is listening on with the
3878 socket units they belong to, plus the units these socket
3881 * The experimental libsystemd-bus library got substantial
3882 updates to work in conjunction with the (also experimental)
3883 kdbus kernel project. It works well enough to exchange
3884 messages with some sophistication. Note that kdbus is not
3885 ready yet, and the library is mostly an elaborate test case
3886 for now, and not installable.
3888 * systemd gained a new unit 'systemd-static-nodes.service'
3889 that generates static device nodes earlier during boot, and
3890 can run in conjunction with udev.
3892 * libsystemd-login gained a new call sd_pid_get_user_unit()
3893 to retrieve the user systemd unit a process is running
3894 in. This is useful for systems where systemd is used as
3897 * systemd-nspawn now places all containers in the new /machine
3898 top-level cgroup directory in the name=systemd
3899 hierarchy. libvirt will soon do the same, so that we get a
3900 uniform separation of /system, /user and /machine for system
3901 services, user processes and containers/virtual
3902 machines. This new cgroup hierarchy is also useful to stick
3903 stable names to specific container instances, which can be
3904 recognized later this way (this name may be controlled
3905 via systemd-nspawn's new -M switch). libsystemd-login also
3906 gained a new call sd_pid_get_machine_name() to retrieve the
3907 name of the container/VM a specific process belongs to.
3909 * bootchart can now store its data in the journal.
3911 * libsystemd-journal gained a new call
3912 sd_journal_add_conjunction() for AND expressions to the
3913 matching logic. This can be used to express more complex
3914 logical expressions.
3916 * journactl can now take multiple --unit= and --user-unit=
3919 * The cryptsetup logic now understands the "luks.key=" kernel
3920 command line switch for specifying a file to read the
3921 decryption key from. Also, if a configured key file is not
3922 found the tool will now automatically fall back to prompting
3925 * Python systemd.journal module was updated to wrap recently
3926 added functions from libsystemd-journal. The interface was
3927 changed to bring the low level interface in s.j._Reader
3928 closer to the C API, and the high level interface in
3929 s.j.Reader was updated to wrap and convert all data about
3932 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
3933 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
3934 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
3935 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
3936 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
3937 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3941 * journalctl --update-catalog now understands a new --root=
3942 option to operate on catalogs found in a different root
3945 * During shutdown after systemd has terminated all running
3946 services a final killing loop kills all remaining left-over
3947 processes. We will now print the name of these processes
3948 when we send SIGKILL to them, since this usually indicates a
3951 * If /etc/crypttab refers to password files stored on
3952 configured mount points automatic dependencies will now be
3953 generated to ensure the specific mount is established first
3954 before the key file is attempted to be read.
3956 * 'systemctl status' will now show information about the
3957 network sockets a socket unit is listening on.
3959 * 'systemctl status' will also shown information about any
3960 drop-in configuration file for units. (Drop-In configuration
3961 files in this context are files such as
3962 /etc/systemd/systemd/foobar.service.d/*.conf)
3964 * systemd-cgtop now optionally shows summed up CPU times of
3965 cgroups. Press '%' while running cgtop to switch between
3966 percentage and absolute mode. This is useful to determine
3967 which cgroups use up the most CPU time over the entire
3968 runtime of the system. systemd-cgtop has also been updated
3969 to be 'pipeable' for processing with further shell tools.
3971 * 'hostnamectl set-hostname' will now allow setting of FQDN
3974 * The formatting and parsing of time span values has been
3975 changed. The parser now understands fractional expressions
3976 such as "5.5h". The formatter will now output fractional
3977 expressions for all time spans under 1min, i.e. "5.123456s"
3978 rather than "5s 123ms 456us". For time spans under 1s
3979 millisecond values are shown, for those under 1ms
3980 microsecond values are shown. This should greatly improve
3981 all time-related output of systemd.
3983 * libsystemd-login and libsystemd-journal gained new
3984 functions for querying the poll() events mask and poll()
3985 timeout value for integration into arbitrary event
3988 * localectl gained the ability to list available X11 keymaps
3989 (models, layouts, variants, options).
3991 * 'systemd-analyze dot' gained the ability to filter for
3992 specific units via shell-style globs, to create smaller,
3993 more useful graphs. I.e. it is now possible to create simple
3994 graphs of all the dependencies between only target units, or
3995 of all units that Avahi has dependencies with.
3997 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
3998 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
3999 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
4000 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
4001 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
4002 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
4003 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
4007 * The boot-time readahead implementation for rotating media
4008 will now read the read-ahead data in multiple passes which
4009 consist of all read requests made in equidistant time
4010 intervals. This means instead of strictly reading read-ahead
4011 data in its physical order on disk we now try to find a
4012 middle ground between physical and access time order.
4014 * /etc/os-release files gained a new BUILD_ID= field for usage
4015 on operating systems that provide continuous builds of OS
4018 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
4019 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
4020 William Douglas, Zbigniew Jędrzejewski-Szmek
4024 * systemd-python gained an API exposing libsystemd-daemon.
4026 * The SMACK setup logic gained support for uploading CIPSO
4029 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
4030 ReadOnlyDirectories= and InaccessibleDirectories= has
4031 changed. The private /tmp and /var/tmp directories are now
4032 shared by all processes of a service (which means
4033 ExecStartPre= may now leave data in /tmp that ExecStart= of
4034 the same service can still access). When a service is
4035 stopped its temporary directories are immediately deleted
4036 (normal clean-up with tmpfiles is still done in addition to
4039 * By default, systemd will now set a couple of sysctl
4040 variables in the kernel: the safe sysrq options are turned
4041 on, IP route verification is turned on, and source routing
4042 disabled. The recently added hardlink and softlink
4043 protection of the kernel is turned on. These settings should
4044 be reasonably safe, and good defaults for all new systems.
4046 * The predictable network naming logic may now be turned off
4047 with a new kernel command line switch: net.ifnames=0.
4049 * A new libsystemd-bus module has been added that implements a
4050 pretty complete D-Bus client library. For details see:
4052 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
4054 * journald will now explicitly flush the journal files to disk
4055 at the latest 5min after each write. The file will then also
4056 be marked offline until the next write. This should increase
4057 reliability in case of a crash. The synchronization delay
4058 can be configured via SyncIntervalSec= in journald.conf.
4060 * There's a new remote-fs-setup.target unit that can be used
4061 to pull in specific services when at least one remote file
4062 system is to be mounted.
4064 * There are new targets timers.target and paths.target as
4065 canonical targets to pull user timer and path units in
4066 from. This complements sockets.target with a similar
4067 purpose for socket units.
4069 * libudev gained a new call udev_device_set_attribute_value()
4070 to set sysfs attributes of a device.
4072 * The udev daemon now sets the default number of worker
4073 processes executed in parallel based on the number of available
4074 CPUs instead of the amount of available RAM. This is supposed
4075 to provide a more reliable default and limit a too aggressive
4076 paralellism for setups with 1000s of devices connected.
4078 Contributions from: Auke Kok, Colin Walters, Cristian
4079 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
4080 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
4081 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
4082 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
4083 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
4084 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
4085 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
4086 Zbigniew Jędrzejewski-Szmek
4090 * Configuration of unit files may now be extended via drop-in
4091 files without having to edit/override the unit files
4092 themselves. More specifically, if the administrator wants to
4093 change one value for a service file foobar.service he can
4094 now do so by dropping in a configuration snippet into
4095 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
4096 will load all these snippets and apply them on top of the
4097 main unit configuration file, possibly extending or
4098 overriding its settings. Using these drop-in snippets is
4099 generally nicer than the two earlier options for changing
4100 unit files locally: copying the files from
4101 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
4102 them there; or creating a new file in /etc/systemd/system/
4103 that incorporates the original one via ".include". Drop-in
4104 snippets into these .d/ directories can be placed in any
4105 directory systemd looks for units in, and the usual
4106 overriding semantics between /usr/lib, /etc and /run apply
4109 * Most unit file settings which take lists of items can now be
4110 reset by assigning the empty string to them. For example,
4111 normally, settings such as Environment=FOO=BAR append a new
4112 environment variable assignment to the environment block,
4113 each time they are used. By assigning Environment= the empty
4114 string the environment block can be reset to empty. This is
4115 particularly useful with the .d/*.conf drop-in snippets
4116 mentioned above, since this adds the ability to reset list
4117 settings from vendor unit files via these drop-ins.
4119 * systemctl gained a new "list-dependencies" command for
4120 listing the dependencies of a unit recursively.
4122 * Inhibitors are now honored and listed by "systemctl
4123 suspend", "systemctl poweroff" (and similar) too, not only
4124 GNOME. These commands will also list active sessions by
4127 * Resource limits (as exposed by the various control group
4128 controllers) can now be controlled dynamically at runtime
4129 for all units. More specifically, you can now use a command
4130 like "systemctl set-cgroup-attr foobar.service cpu.shares
4131 2000" to alter the CPU shares a specific service gets. These
4132 settings are stored persistently on disk, and thus allow the
4133 administrator to easily adjust the resource usage of
4134 services with a few simple commands. This dynamic resource
4135 management logic is also available to other programs via the
4136 bus. Almost any kernel cgroup attribute and controller is
4139 * systemd-vconsole-setup will now copy all font settings to
4140 all allocated VTs, where it previously applied them only to
4143 * libsystemd-login gained the new sd_session_get_tty() API
4146 * This release drops support for a few legacy or
4147 distribution-specific LSB facility names when parsing init
4148 scripts: $x-display-manager, $mail-transfer-agent,
4149 $mail-transport-agent, $mail-transfer-agent, $smtp,
4150 $null. Also, the mail-transfer-agent.target unit backing
4151 this has been removed. Distributions which want to retain
4152 compatibility with this should carry the burden for
4153 supporting this themselves and patch support for these back
4154 in, if they really need to. Also, the facilities $syslog and
4155 $local_fs are now ignored, since systemd does not support
4156 early-boot LSB init scripts anymore, and these facilities
4157 are implied anyway for normal services. syslog.target has
4160 * There are new bus calls on PID1's Manager object for
4161 cancelling jobs, and removing snapshot units. Previously,
4162 both calls were only available on the Job and Snapshot
4165 * systemd-journal-gatewayd gained SSL support.
4167 * The various "environment" files, such as /etc/locale.conf
4168 now support continuation lines with a backslash ("\") as
4169 last character in the line, similarly in style (but different)
4170 to how this is supported in shells.
4172 * For normal user processes the _SYSTEMD_USER_UNIT= field is
4173 now implicitly appended to every log entry logged. systemctl
4174 has been updated to filter by this field when operating on a
4175 user systemd instance.
4177 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
4178 CAP_AUDIT_CONTROL capabilities to the capabilities set for
4179 the container. This makes it easier to boot unmodified
4180 Fedora systems in a container, which however still requires
4181 audit=0 to be passed on the kernel command line. Auditing in
4182 kernel and userspace is unfortunately still too broken in
4183 context of containers, hence we recommend compiling it out
4184 of the kernel or using audit=0. Hopefully this will be fixed
4185 one day for good in the kernel.
4187 * nspawn gained the new --bind= and --bind-ro= parameters to
4188 bind mount specific directories from the host into the
4191 * nspawn will now mount its own devpts file system instance
4192 into the container, in order not to leak pty devices from
4193 the host into the container.
4195 * systemd will now read the firmware boot time performance
4196 information from the EFI variables, if the used boot loader
4197 supports this, and takes it into account for boot performance
4198 analysis via "systemd-analyze". This is currently supported
4199 only in conjunction with Gummiboot, but could be supported
4200 by other boot loaders too. For details see:
4202 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
4204 * A new generator has been added that automatically mounts the
4205 EFI System Partition (ESP) to /boot, if that directory
4206 exists, is empty, and no other file system has been
4207 configured to be mounted there.
4209 * logind will now send out PrepareForSleep(false) out
4210 unconditionally, after coming back from suspend. This may be
4211 used by applications as asynchronous notification for
4212 system resume events.
4214 * "systemctl unlock-sessions" has been added, that allows
4215 unlocking the screens of all user sessions at once, similar
4216 to how "systemctl lock-sessions" already locked all users
4217 sessions. This is backed by a new D-Bus call UnlockSessions().
4219 * "loginctl seat-status" will now show the master device of a
4220 seat. (i.e. the device of a seat that needs to be around for
4221 the seat to be considered available, usually the graphics
4224 * tmpfiles gained a new "X" line type, that allows
4225 configuration of files and directories (with wildcards) that
4226 shall be excluded from automatic cleanup ("aging").
4228 * udev default rules set the device node permissions now only
4229 at "add" events, and do not change them any longer with a
4230 later "change" event.
4232 * The log messages for lid events and power/sleep keypresses
4233 now carry a message ID.
4235 * We now have a substantially larger unit test suite, but this
4236 continues to be work in progress.
4238 * udevadm hwdb gained a new --root= parameter to change the
4239 root directory to operate relative to.
4241 * logind will now issue a background sync() request to the kernel
4242 early at shutdown, so that dirty buffers are flushed to disk early
4243 instead of at the last moment, in order to optimize shutdown
4246 * A new bootctl tool has been added that is an interface for
4247 certain boot loader operations. This is currently a preview
4248 and is likely to be extended into a small mechanism daemon
4249 like timedated, localed, hostnamed, and can be used by
4250 graphical UIs to enumerate available boot options, and
4251 request boot into firmware operations.
4253 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
4254 the rest of the package. It also has been updated to work
4255 correctly in initrds.
4257 * Policykit previously has been runtime optional, and is now
4258 also compile time optional via a configure switch.
4260 * systemd-analyze has been reimplemented in C. Also "systemctl
4261 dot" has moved into systemd-analyze.
4263 * "systemctl status" with no further parameters will now print
4264 the status of all active or failed units.
4266 * Operations such as "systemctl start" can now be executed
4267 with a new mode "--irreversible" which may be used to queue
4268 operations that cannot accidentally be reversed by a later
4269 job queuing. This is by default used to make shutdown
4270 requests more robust.
4272 * The Python API of systemd now gained a new module for
4273 reading journal files.
4275 * A new tool kernel-install has been added that can install
4276 kernel images according to the Boot Loader Specification:
4278 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
4280 * Boot time console output has been improved to provide
4281 animated boot time output for hanging jobs.
4283 * A new tool systemd-activate has been added which can be used
4284 to test socket activation with, directly from the command
4285 line. This should make it much easier to test and debug
4286 socket activation in daemons.
4288 * journalctl gained a new "--reverse" (or -r) option to show
4289 journal output in reverse order (i.e. newest line first).
4291 * journalctl gained a new "--pager-end" (or -e) option to jump
4292 to immediately jump to the end of the journal in the
4293 pager. This is only supported in conjunction with "less".
4295 * journalctl gained a new "--user-unit=" option, that works
4296 similarly to "--unit=" but filters for user units rather than
4299 * A number of unit files to ease adoption of systemd in
4300 initrds has been added. This moves some minimal logic from
4301 the various initrd implementations into systemd proper.
4303 * The journal files are now owned by a new group
4304 "systemd-journal", which exists specifically to allow access
4305 to the journal, and nothing else. Previously, we used the
4306 "adm" group for that, which however possibly covers more
4307 than just journal/log file access. This new group is now
4308 already used by systemd-journal-gatewayd to ensure this
4309 daemon gets access to the journal files and as little else
4310 as possible. Note that "make install" will also set FS ACLs
4311 up for /var/log/journal to give "adm" and "wheel" read
4312 access to it, in addition to "systemd-journal" which owns
4313 the journal files. We recommend that packaging scripts also
4314 add read access to "adm" + "wheel" to /var/log/journal, and
4315 all existing/future journal files. To normal users and
4316 administrators little changes, however packagers need to
4317 ensure to create the "systemd-journal" system group at
4318 package installation time.
4320 * The systemd-journal-gatewayd now runs as unprivileged user
4321 systemd-journal-gateway:systemd-journal-gateway. Packaging
4322 scripts need to create these system user/group at
4325 * timedated now exposes a new boolean property CanNTP that
4326 indicates whether a local NTP service is available or not.
4328 * systemd-detect-virt will now also detect xen PVs
4330 * The pstore file system is now mounted by default, if it is
4333 * In addition to the SELinux and IMA policies we will now also
4334 load SMACK policies at early boot.
4336 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
4337 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
4338 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
4339 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
4340 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
4341 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
4342 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
4343 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
4344 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
4345 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
4346 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
4347 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
4348 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
4349 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
4353 * Timer units now support calendar time events in addition to
4354 monotonic time events. That means you can now trigger a unit
4355 based on a calendar time specification such as "Thu,Fri
4356 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
4357 or fifth day of any month of the year 2013, given that it is
4358 a thursday or friday. This brings timer event support
4359 considerably closer to cron's capabilities. For details on
4360 the supported calendar time specification language see
4363 * udev now supports a number of different naming policies for
4364 network interfaces for predictable names, and a combination
4365 of these policies is now the default. Please see this wiki
4366 document for details:
4368 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
4370 * Auke Kok's bootchart implementation has been added to the
4371 systemd tree. It is an optional component that can graph the
4372 boot in quite some detail. It is one of the best bootchart
4373 implementations around and minimal in its code and
4376 * nss-myhostname has been integrated into the systemd source
4377 tree. nss-myhostname guarantees that the local hostname
4378 always stays resolvable via NSS. It has been a weak
4379 requirement of systemd-hostnamed since a long time, and
4380 since its code is actually trivial we decided to just
4381 include it in systemd's source tree. It can be turned off
4382 with a configure switch.
4384 * The read-ahead logic is now capable of properly detecting
4385 whether a btrfs file system is on SSD or rotating media, in
4386 order to optimize the read-ahead scheme. Previously, it was
4387 only capable of detecting this on traditional file systems
4390 * In udev, additional device properties are now read from the
4391 IAB in addition to the OUI database. Also, Bluetooth company
4392 identities are attached to the devices as well.
4394 * In service files %U may be used as specifier that is
4395 replaced by the configured user name of the service.
4397 * nspawn may now be invoked without a controlling TTY. This
4398 makes it suitable for invocation as its own service. This
4399 may be used to set up a simple containerized server system
4400 using only core OS tools.
4402 * systemd and nspawn can now accept socket file descriptors
4403 when they are started for socket activation. This enables
4404 implementation of socket activated nspawn
4405 containers. i.e. think about autospawning an entire OS image
4406 when the first SSH or HTTP connection is received. We expect
4407 that similar functionality will also be added to libvirt-lxc
4410 * journalctl will now suppress ANSI color codes when
4411 presenting log data.
4413 * systemctl will no longer show control group information for
4414 a unit if a the control group is empty anyway.
4416 * logind can now automatically suspend/hibernate/shutdown the
4419 * /etc/machine-info and hostnamed now also expose the chassis
4420 type of the system. This can be used to determine whether
4421 the local system is a laptop, desktop, handset or
4422 tablet. This information may either be configured by the
4423 user/vendor or is automatically determined from ACPI and DMI
4424 information if possible.
4426 * A number of PolicyKit actions are now bound together with
4427 "imply" rules. This should simplify creating UIs because
4428 many actions will now authenticate similar ones as well.
4430 * Unit files learnt a new condition ConditionACPower= which
4431 may be used to conditionalize a unit depending on whether an
4432 AC power source is connected or not, of whether the system
4433 is running on battery power.
4435 * systemctl gained a new "is-failed" verb that may be used in
4436 shell scripts and suchlike to check whether a specific unit
4437 is in the "failed" state.
4439 * The EnvironmentFile= setting in unit files now supports file
4440 globbing, and can hence be used to easily read a number of
4441 environment files at once.
4443 * systemd will no longer detect and recognize specific
4444 distributions. All distribution-specific #ifdeffery has been
4445 removed, systemd is now fully generic and
4446 distribution-agnostic. Effectively, not too much is lost as
4447 a lot of the code is still accessible via explicit configure
4448 switches. However, support for some distribution specific
4449 legacy configuration file formats has been dropped. We
4450 recommend distributions to simply adopt the configuration
4451 files everybody else uses now and convert the old
4452 configuration from packaging scripts. Most distributions
4453 already did that. If that's not possible or desirable,
4454 distributions are welcome to forward port the specific
4455 pieces of code locally from the git history.
4457 * When logging a message about a unit systemd will now always
4458 log the unit name in the message meta data.
4460 * localectl will now also discover system locale data that is
4461 not stored in locale archives, but directly unpacked.
4463 * logind will no longer unconditionally use framebuffer
4464 devices as seat masters, i.e. as devices that are required
4465 to be existing before a seat is considered preset. Instead,
4466 it will now look for all devices that are tagged as
4467 "seat-master" in udev. By default, framebuffer devices will
4468 be marked as such, but depending on local systems, other
4469 devices might be marked as well. This may be used to
4470 integrate graphics cards using closed source drivers (such
4471 as NVidia ones) more nicely into logind. Note however, that
4472 we recommend using the open source NVidia drivers instead,
4473 and no udev rules for the closed-source drivers will be
4474 shipped from us upstream.
4476 Contributions from: Adam Williamson, Alessandro Crismani, Auke
4477 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
4478 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
4479 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
4480 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
4481 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
4482 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
4483 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
4484 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
4485 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
4486 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
4487 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
4492 * udev gained support for loading additional device properties
4493 from an indexed database that is keyed by vendor/product IDs
4494 and similar device identifiers. For the beginning this
4495 "hwdb" is populated with data from the well-known PCI and
4496 USB database, but also includes PNP, ACPI and OID data. In
4497 the longer run this indexed database shall grow into
4498 becoming the one central database for non-essential
4499 userspace device metadata. Previously, data from the PCI/USB
4500 database was only attached to select devices, since the
4501 lookup was a relatively expensive operation due to O(n) time
4502 complexity (with n being the number of entries in the
4503 database). Since this is now O(1), we decided to add in this
4504 data for all devices where this is available, by
4505 default. Note that the indexed database needs to be rebuilt
4506 when new data files are installed. To achieve this you need
4507 to update your packaging scripts to invoke "udevadm hwdb
4508 --update" after installation of hwdb data files. For
4509 RPM-based distributions we introduced the new
4510 %udev_hwdb_update macro for this purpose.
4512 * The Journal gained support for the "Message Catalog", an
4513 indexed database to link up additional information with
4514 journal entries. For further details please check:
4516 http://www.freedesktop.org/wiki/Software/systemd/catalog
4518 The indexed message catalog database also needs to be
4519 rebuilt after installation of message catalog files. Use
4520 "journalctl --update-catalog" for this. For RPM-based
4521 distributions we introduced the %journal_catalog_update
4522 macro for this purpose.
4524 * The Python Journal bindings gained support for the standard
4525 Python logging framework.
4527 * The Journal API gained new functions for checking whether
4528 the underlying file system of a journal file is capable of
4529 properly reporting file change notifications, or whether
4530 applications that want to reflect journal changes "live"
4531 need to recheck journal files continuously in appropriate
4534 * It is now possible to set the "age" field for tmpfiles
4535 entries to 0, indicating that files matching this entry
4536 shall always be removed when the directories are cleaned up.
4538 * coredumpctl gained a new "gdb" verb which invokes gdb
4539 right-away on the selected coredump.
4541 * There's now support for "hybrid sleep" on kernels that
4542 support this, in addition to "suspend" and "hibernate". Use
4543 "systemctl hybrid-sleep" to make use of this.
4545 * logind's HandleSuspendKey= setting (and related settings)
4546 now gained support for a new "lock" setting to simply
4547 request the screen lock on all local sessions, instead of
4548 actually executing a suspend or hibernation.
4550 * systemd will now mount the EFI variables file system by
4553 * Socket units now gained support for configuration of the
4554 SMACK security label.
4556 * timedatectl will now output the time of the last and next
4557 daylight saving change.
4559 * We dropped support for various legacy and distro-specific
4560 concepts, such as insserv, early-boot SysV services
4561 (i.e. those for non-standard runlevels such as 'b' or 'S')
4562 or ArchLinux /etc/rc.conf support. We recommend the
4563 distributions who still need support this to either continue
4564 to maintain the necessary patches downstream, or find a
4565 different solution. (Talk to us if you have questions!)
4567 * Various systemd components will now bypass PolicyKit checks
4568 for root and otherwise handle properly if PolicyKit is not
4569 found to be around. This should fix most issues for
4570 PolicyKit-less systems. Quite frankly this should have been
4571 this way since day one. It is absolutely our intention to
4572 make systemd work fine on PolicyKit-less systems, and we
4573 consider it a bug if something does not work as it should if
4574 PolicyKit is not around.
4576 * For embedded systems it is now possible to build udev and
4577 systemd without blkid and/or kmod support.
4579 * "systemctl switch-root" is now capable of switching root
4580 more than once. I.e. in addition to transitions from the
4581 initrd to the host OS it is now possible to transition to
4582 further OS images from the host. This is useful to implement
4583 offline updating tools.
4585 * Various other additions have been made to the RPM macros
4586 shipped with systemd. Use %udev_rules_update() after
4587 installing new udev rules files. %_udevhwdbdir,
4588 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
4589 %_sysctldir are now available which resolve to the right
4590 directories for packages to place various data files in.
4592 * journalctl gained the new --full switch (in addition to
4593 --all, to disable ellipsation for long messages.
4595 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
4596 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
4597 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
4598 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
4599 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
4600 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
4601 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
4602 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
4603 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
4607 * journalctl gained new --since= and --until= switches to
4608 filter by time. It also now supports nice filtering for
4609 units via --unit=/-u.
4611 * Type=oneshot services may use ExecReload= and do the
4614 * The journal daemon now supports time-based rotation and
4615 vacuuming, in addition to the usual disk-space based
4618 * The journal will now index the available field values for
4619 each field name. This enables clients to show pretty drop
4620 downs of available match values when filtering. The bash
4621 completion of journalctl has been updated
4622 accordingly. journalctl gained a new switch -F to list all
4623 values a certain field takes in the journal database.
4625 * More service events are now written as structured messages
4626 to the journal, and made recognizable via message IDs.
4628 * The timedated, localed and hostnamed mini-services which
4629 previously only provided support for changing time, locale
4630 and hostname settings from graphical DEs such as GNOME now
4631 also have a minimal (but very useful) text-based client
4632 utility each. This is probably the nicest way to changing
4633 these settings from the command line now, especially since
4634 it lists available options and is fully integrated with bash
4637 * There's now a new tool "systemd-coredumpctl" to list and
4638 extract coredumps from the journal.
4640 * We now install a README each in /var/log/ and
4641 /etc/rc.d/init.d explaining where the system logs and init
4642 scripts went. This hopefully should help folks who go to
4643 that dirs and look into the otherwise now empty void and
4644 scratch their heads.
4646 * When user-services are invoked (by systemd --user) the
4647 $MANAGERPID env var is set to the PID of systemd.
4649 * SIGRTMIN+24 when sent to a --user instance will now result
4650 in immediate termination of systemd.
4652 * gatewayd received numerous feature additions such as a
4653 "follow" mode, for live syncing and filtering.
4655 * browse.html now allows filtering and showing detailed
4656 information on specific entries. Keyboard navigation and
4657 mouse screen support has been added.
4659 * gatewayd/journalctl now supports HTML5/JSON
4660 Server-Sent-Events as output.
4662 * The SysV init script compatibility logic will now
4663 heuristically determine whether a script supports the
4664 "reload" verb, and only then make this available as
4667 * "systemctl status --follow" has been removed, use "journalctl
4670 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
4671 have been removed since they are hardly useful to be
4674 * And I'd like to take the opportunity to specifically mention
4675 Zbigniew for his great contributions. Zbigniew, you rock!
4677 Contributions from: Andrew Eikum, Christian Hesse, Colin
4678 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
4679 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
4680 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
4681 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
4682 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
4683 Jędrzejewski-Szmek, Сковорода Никита Андреевич
4687 * If /etc/vconsole.conf is non-existent or empty we will no
4688 longer load any console font or key map at boot by
4689 default. Instead the kernel defaults will be left
4690 intact. This is definitely the right thing to do, as no
4691 configuration should mean no configuration, and hard-coding
4692 font names that are different on all archs is probably a bad
4693 idea. Also, the kernel default key map and font should be
4694 good enough for most cases anyway, and mostly identical to
4695 the userspace fonts/key maps we previously overloaded them
4696 with. If distributions want to continue to default to a
4697 non-kernel font or key map they should ship a default
4698 /etc/vconsole.conf with the appropriate contents.
4700 Contributions from: Colin Walters, Daniel J Walsh, Dave
4701 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
4702 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
4706 * journalctl gained a new --cursor= switch to show entries
4707 starting from the specified location in the journal.
4709 * We now enforce a size limit on journal entry fields exported
4710 with "-o json" in journalctl. Fields larger than 4K will be
4711 assigned null. This can be turned off with --all.
4713 * An (optional) journal gateway daemon is now available as
4714 "systemd-journal-gatewayd.service". This service provides
4715 access to the journal via HTTP and JSON. This functionality
4716 will be used to implement live log synchronization in both
4717 pull and push modes, but has various other users too, such
4718 as easy log access for debugging of embedded devices. Right
4719 now it is already useful to retrieve the journal via HTTP:
4721 # systemctl start systemd-journal-gatewayd.service
4722 # wget http://localhost:19531/entries
4724 This will download the journal contents in a
4725 /var/log/messages compatible format. The same as JSON:
4727 # curl -H"Accept: application/json" http://localhost:19531/entries
4729 This service is also accessible via a web browser where a
4730 single static HTML5 app is served that uses the JSON logic
4731 to enable the user to do some basic browsing of the
4732 journal. This will be extended later on. Here's an example
4733 screenshot of this app in its current state:
4735 http://0pointer.de/public/journal-gatewayd
4737 Contributions from: Kay Sievers, Lennart Poettering, Robert
4738 Milasan, Tom Gundersen
4742 * The bash completion logic is now available for journalctl
4745 * We do not mount the "cpuset" controller anymore together with
4746 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
4747 started if no parameters are assigned to it. "cpuset" hence
4748 broke code that assumed it it could create "cpu" groups and
4751 * journalctl -f will now subscribe to terminal size changes,
4752 and line break accordingly.
4754 Contributions from: Dave Reisner, Kay Sievers, Lennart
4755 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
4759 * nspawn will now create a symlink /etc/localtime in the
4760 container environment, copying the host's timezone
4761 setting. Previously this has been done via a bind mount, but
4762 since symlinks cannot be bind mounted this has now been
4763 changed to create/update the appropriate symlink.
4765 * journalctl -n's line number argument is now optional, and
4766 will default to 10 if omitted.
4768 * journald will now log the maximum size the journal files may
4769 take up on disk. This is particularly useful if the default
4770 built-in logic of determining this parameter from the file
4771 system size is used. Use "systemctl status
4772 systemd-journald.service" to see this information.
4774 * The multi-seat X wrapper tool has been stripped down. As X
4775 is now capable of enumerating graphics devices via udev in a
4776 seat-aware way the wrapper is not strictly necessary
4777 anymore. A stripped down temporary stop-gap is still shipped
4778 until the upstream display managers have been updated to
4779 fully support the new X logic. Expect this wrapper to be
4780 removed entirely in one of the next releases.
4782 * HandleSleepKey= in logind.conf has been split up into
4783 HandleSuspendKey= and HandleHibernateKey=. The old setting
4784 is not available anymore. X11 and the kernel are
4785 distinguishing between these keys and we should too. This
4786 also means the inhibition lock for these keys has been split
4789 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
4790 Poettering, Lukas Nykryn, Václav Pavlín
4794 * Whenever a unit changes state we will now log this to the
4795 journal and show along the unit's own log output in
4798 * ConditionPathIsMountPoint= can now properly detect bind
4799 mount points too. (Previously, a bind mount of one file
4800 system to another place in the same file system could not be
4801 detected as mount, since they shared struct stat's st_dev
4804 * We will now mount the cgroup controllers cpu, cpuacct,
4805 cpuset and the controllers net_cls, net_prio together by
4808 * nspawn containers will now have a virtualized boot
4809 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
4810 over with a randomized ID at container initialization). This
4811 has the effect of making "journalctl -b" do the right thing
4814 * The JSON output journal serialization has been updated not
4815 to generate "endless" list objects anymore, but rather one
4816 JSON object per line. This is more in line how most JSON
4817 parsers expect JSON objects. The new output mode
4818 "json-pretty" has been added to provide similar output, but
4819 neatly aligned for readability by humans.
4821 * We dropped all explicit sync() invocations in the shutdown
4822 code. The kernel does this implicitly anyway in the kernel
4823 reboot() syscall. halt(8)'s -n option is now a compatibility
4826 * We now support virtualized reboot() in containers, as
4827 supported by newer kernels. We will fall back to exit() if
4828 CAP_SYS_REBOOT is not available to the container. Also,
4829 nspawn makes use of this now and will actually reboot the
4830 container if the containerized OS asks for that.
4832 * journalctl will only show local log output by default
4833 now. Use --merge (-m) to show remote log output, too.
4835 * libsystemd-journal gained the new sd_journal_get_usage()
4836 call to determine the current disk usage of all journal
4837 files. This is exposed in the new "journalctl --disk-usage"
4840 * journald gained a new configuration setting SplitMode= in
4841 journald.conf which may be used to control how user journals
4842 are split off. See journald.conf(5) for details.
4844 * A new condition type ConditionFileNotEmpty= has been added.
4846 * tmpfiles' "w" lines now support file globbing, to write
4847 multiple files at once.
4849 * We added Python bindings for the journal submission
4850 APIs. More Python APIs for a number of selected APIs will
4851 likely follow. Note that we intend to add native bindings
4852 only for the Python language, as we consider it common
4853 enough to deserve bindings shipped within systemd. There are
4854 various projects outside of systemd that provide bindings
4855 for languages such as PHP or Lua.
4857 * Many conditions will now resolve specifiers such as %i. In
4858 addition, PathChanged= and related directives of .path units
4859 now support specifiers as well.
4861 * There's now a new RPM macro definition for the system preset
4864 * journald will now warn if it ca not forward a message to the
4865 syslog daemon because its socket is full.
4867 * timedated will no longer write or process /etc/timezone,
4868 except on Debian. As we do not support late mounted /usr
4869 anymore /etc/localtime always being a symlink is now safe,
4870 and hence the information in /etc/timezone is not necessary
4873 * logind will now always reserve one VT for a text getty (VT6
4874 by default). Previously if more than 6 X sessions where
4875 started they took up all the VTs with auto-spawned gettys,
4876 so that no text gettys were available anymore.
4878 * udev will now automatically inform the btrfs kernel logic
4879 about btrfs RAID components showing up. This should make
4880 simple hotplug based btrfs RAID assembly work.
4882 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
4883 (but not for its children which will stay at the kernel
4884 default). This should allow setups with a lot more listening
4887 * systemd will now always pass the configured timezone to the
4888 kernel at boot. timedated will do the same when the timezone
4891 * logind's inhibition logic has been updated. By default,
4892 logind will now handle the lid switch, the power and sleep
4893 keys all the time, even in graphical sessions. If DEs want
4894 to handle these events on their own they should take the new
4895 handle-power-key, handle-sleep-key and handle-lid-switch
4896 inhibitors during their runtime. A simple way to achieve
4897 that is to invoke the DE wrapped in an invocation of:
4899 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
4901 * Access to unit operations is now checked via SELinux taking
4902 the unit file label and client process label into account.
4904 * systemd will now notify the administrator in the journal
4905 when he over-mounts a non-empty directory.
4907 * There are new specifiers that are resolved in unit files,
4908 for the host name (%H), the machine ID (%m) and the boot ID
4911 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
4912 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
4913 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
4914 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
4915 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
4916 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
4917 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
4921 * Support for reading structured kernel messages from
4922 /dev/kmsg has now been added and is enabled by default.
4924 * Support for reading kernel messages from /proc/kmsg has now
4925 been removed. If you want kernel messages in the journal
4926 make sure to run a recent kernel (>= 3.5) that supports
4927 reading structured messages from /dev/kmsg (see
4928 above). /proc/kmsg is now exclusive property of classic
4929 syslog daemons again.
4931 * The libudev API gained the new
4932 udev_device_new_from_device_id() call.
4934 * The logic for file system namespace (ReadOnlyDirectory=,
4935 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
4936 require pivot_root() anymore. This means fewer temporary
4937 directories are created below /tmp for this feature.
4939 * nspawn containers will now see and receive all submounts
4940 made on the host OS below the root file system of the
4943 * Forward Secure Sealing is now supported for Journal files,
4944 which provide cryptographical sealing of journal files so
4945 that attackers cannot alter log history anymore without this
4946 being detectable. Lennart will soon post a blog story about
4947 this explaining it in more detail.
4949 * There are two new service settings RestartPreventExitStatus=
4950 and SuccessExitStatus= which allow configuration of exit
4951 status (exit code or signal) which will be excepted from the
4952 restart logic, resp. consider successful.
4954 * journalctl gained the new --verify switch that can be used
4955 to check the integrity of the structure of journal files and
4956 (if Forward Secure Sealing is enabled) the contents of
4959 * nspawn containers will now be run with /dev/stdin, /dev/fd/
4960 and similar symlinks pre-created. This makes running shells
4961 as container init process a lot more fun.
4963 * The fstab support can now handle PARTUUID= and PARTLABEL=
4966 * A new ConditionHost= condition has been added to match
4967 against the hostname (with globs) and machine ID. This is
4968 useful for clusters where a single OS image is used to
4969 provision a large number of hosts which shall run slightly
4970 different sets of services.
4972 * Services which hit the restart limit will now be placed in a
4975 Contributions from: Bertram Poettering, Dave Reisner, Huang
4976 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
4977 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
4981 * When running in --user mode systemd will now become a
4982 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
4983 tree a lot more organized.
4985 * A new PartOf= unit dependency type has been introduced that
4986 may be used to group services in a natural way.
4988 * "systemctl enable" may now be used to enable instances of
4991 * journalctl now prints error log levels in red, and
4992 warning/notice log levels in bright white. It also supports
4993 filtering by log level now.
4995 * cgtop gained a new -n switch (similar to top), to configure
4996 the maximum number of iterations to run for. It also gained
4997 -b, to run in batch mode (accepting no input).
4999 * The suffix ".service" may now be omitted on most systemctl
5000 command lines involving service unit names.
5002 * There's a new bus call in logind to lock all sessions, as
5003 well as a loginctl verb for it "lock-sessions".
5005 * libsystemd-logind.so gained a new call sd_journal_perror()
5006 that works similar to libc perror() but logs to the journal
5007 and encodes structured information about the error number.
5009 * /etc/crypttab entries now understand the new keyfile-size=
5012 * shutdown(8) now can send a (configurable) wall message when
5013 a shutdown is cancelled.
5015 * The mount propagation mode for the root file system will now
5016 default to "shared", which is useful to make containers work
5017 nicely out-of-the-box so that they receive new mounts from
5018 the host. This can be undone locally by running "mount
5019 --make-rprivate /" if needed.
5021 * The prefdm.service file has been removed. Distributions
5022 should maintain this unit downstream if they intend to keep
5023 it around. However, we recommend writing normal unit files
5024 for display managers instead.
5026 * Since systemd is a crucial part of the OS we will now
5027 default to a number of compiler switches that improve
5028 security (hardening) such as read-only relocations, stack
5029 protection, and suchlike.
5031 * The TimeoutSec= setting for services is now split into
5032 TimeoutStartSec= and TimeoutStopSec= to allow configuration
5033 of individual time outs for the start and the stop phase of
5036 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
5037 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
5038 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
5039 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
5040 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
5041 Gundersen, Zbigniew Jędrzejewski-Szmek
5045 * The journal and id128 C APIs are now fully documented as man
5048 * Extra safety checks have been added when transitioning from
5049 the initial RAM disk to the main system to avoid accidental
5052 * /etc/crypttab entries now understand the new keyfile-offset=
5055 * systemctl -t can now be used to filter by unit load state.
5057 * The journal C API gained the new sd_journal_wait() call to
5058 make writing synchronous journal clients easier.
5060 * journalctl gained the new -D switch to show journals from a
5063 * journalctl now displays a special marker between log
5064 messages of two different boots.
5066 * The journal is now explicitly flushed to /var via a service
5067 systemd-journal-flush.service, rather than implicitly simply
5068 by seeing /var/log/journal to be writable.
5070 * journalctl (and the journal C APIs) can now match for much
5071 more complex expressions, with alternatives and
5074 * When transitioning from the initial RAM disk to the main
5075 system we will now kill all processes in a killing spree to
5076 ensure no processes stay around by accident.
5078 * Three new specifiers may be used in unit files: %u, %h, %s
5079 resolve to the user name, user home directory resp. user
5080 shell. This is useful for running systemd user instances.
5082 * We now automatically rotate journal files if their data
5083 object hash table gets a fill level > 75%. We also size the
5084 hash table based on the configured maximum file size. This
5085 together should lower hash collisions drastically and thus
5086 speed things up a bit.
5088 * journalctl gained the new "--header" switch to introspect
5089 header data of journal files.
5091 * A new setting SystemCallFilters= has been added to services
5092 which may be used to apply blacklists or whitelists to
5093 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
5095 * nspawn gained a new --link-journal= switch (and quicker: -j)
5096 to link the container journal with the host. This makes it
5097 very easy to centralize log viewing on the host for all
5098 guests while still keeping the journal files separated.
5100 * Many bugfixes and optimizations
5102 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
5103 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
5104 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
5109 * Several tools now understand kernel command line arguments,
5110 which are only read when run in an initial RAM disk. They
5111 usually follow closely their normal counterparts, but are
5114 * There's a new tool to analyze the readahead files that are
5115 automatically generated at boot. Use:
5117 /usr/lib/systemd/systemd-readahead analyze /.readahead
5119 * We now provide an early debug shell on tty9 if this enabled. Use:
5121 systemctl enable debug-shell.service
5123 * All plymouth related units have been moved into the Plymouth
5124 package. Please make sure to upgrade your Plymouth version
5127 * systemd-tmpfiles now supports getting passed the basename of
5128 a configuration file only, in which case it will look for it
5129 in all appropriate directories automatically.
5131 * udevadm info now takes a /dev or /sys path as argument, and
5132 does the right thing. Example:
5134 udevadm info /dev/sda
5135 udevadm info /sys/class/block/sda
5137 * systemctl now prints a warning if a unit is stopped but a
5138 unit that might trigger it continues to run. Example: a
5139 service is stopped but the socket that activates it is left
5142 * "systemctl status" will now mention if the log output was
5143 shortened due to rotation since a service has been started.
5145 * The journal API now exposes functions to determine the
5146 "cutoff" times due to rotation.
5148 * journald now understands SIGUSR1 and SIGUSR2 for triggering
5149 immediately flushing of runtime logs to /var if possible,
5150 resp. for triggering immediate rotation of the journal
5153 * It is now considered an error if a service is attempted to
5154 be stopped that is not loaded.
5156 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
5158 * systemd-analyze now supports Python 3
5160 * tmpfiles now supports cleaning up directories via aging
5161 where the first level dirs are always kept around but
5162 directories beneath it automatically aged. This is enabled
5163 by prefixing the age field with '~'.
5165 * Seat objects now expose CanGraphical, CanTTY properties
5166 which is required to deal with very fast bootups where the
5167 display manager might be running before the graphics drivers
5168 completed initialization.
5170 * Seat objects now expose a State property.
5172 * We now include RPM macros for service enabling/disabling
5173 based on the preset logic. We recommend RPM based
5174 distributions to make use of these macros if possible. This
5175 makes it simpler to reuse RPM spec files across
5178 * We now make sure that the collected systemd unit name is
5179 always valid when services log to the journal via
5182 * There's a new man page kernel-command-line(7) detailing all
5183 command line options we understand.
5185 * The fstab generator may now be disabled at boot by passing
5186 fstab=0 on the kernel command line.
5188 * A new kernel command line option modules-load= is now understood
5189 to load a specific kernel module statically, early at boot.
5191 * Unit names specified on the systemctl command line are now
5192 automatically escaped as needed. Also, if file system or
5193 device paths are specified they are automatically turned
5194 into the appropriate mount or device unit names. Example:
5196 systemctl status /home
5197 systemctl status /dev/sda
5199 * The SysVConsole= configuration option has been removed from
5200 system.conf parsing.
5202 * The SysV search path is no longer exported on the D-Bus
5205 * The Names= option is been removed from unit file parsing.
5207 * There's a new man page bootup(7) detailing the boot process.
5209 * Every unit and every generator we ship with systemd now
5210 comes with full documentation. The self-explanatory boot is
5213 * A couple of services gained "systemd-" prefixes in their
5214 name if they wrap systemd code, rather than only external
5215 code. Among them fsck@.service which is now
5216 systemd-fsck@.service.
5218 * The HaveWatchdog property has been removed from the D-Bus
5221 * systemd.confirm_spawn= on the kernel command line should now
5224 * There's a new man page crypttab(5) which details all options
5225 we actually understand.
5227 * systemd-nspawn gained a new --capability= switch to pass
5228 additional capabilities to the container.
5230 * timedated will now read known NTP implementation unit names
5231 from /usr/lib/systemd/ntp-units.d/*.list,
5232 systemd-timedated-ntp.target has been removed.
5234 * journalctl gained a new switch "-b" that lists log data of
5235 the current boot only.
5237 * The notify socket is in the abstract namespace again, in
5238 order to support daemons which chroot() at start-up.
5240 * There is a new Storage= configuration option for journald
5241 which allows configuration of where log data should go. This
5242 also provides a way to disable journal logging entirely, so
5243 that data collected is only forwarded to the console, the
5244 kernel log buffer or another syslog implementation.
5246 * Many bugfixes and optimizations
5248 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
5249 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
5250 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
5251 Shawn Landden, Tom Gundersen
5255 * "systemctl help <unit>" now shows the man page if one is
5258 * Several new man pages have been added.
5260 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
5261 MaxLevelConsole= can now be specified in
5262 journald.conf. These options allow reducing the amount of
5263 data stored on disk or forwarded by the log level.
5265 * TimerSlackNSec= can now be specified in system.conf for
5266 PID1. This allows system-wide power savings.
5268 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
5269 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
5274 * logind is now capable of (optionally) handling power and
5275 sleep keys as well as the lid switch.
5277 * journalctl now understands the syntax "journalctl
5278 /usr/bin/avahi-daemon" to get all log output of a specific
5281 * CapabilityBoundingSet= in system.conf now also influences
5282 the capability bound set of usermode helpers of the kernel.
5284 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
5285 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
5286 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
5287 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
5291 * Note that we skipped 139 releases here in order to set the
5292 new version to something that is greater than both udev's
5293 and systemd's most recent version number.
5295 * udev: all udev sources are merged into the systemd source tree now.
5296 All future udev development will happen in the systemd tree. It
5297 is still fully supported to use the udev daemon and tools without
5298 systemd running, like in initramfs or other init systems. Building
5299 udev though, will require the *build* of the systemd tree, but
5300 udev can be properly *run* without systemd.
5302 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
5303 should be used to create dead device nodes as workarounds for broken
5306 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
5307 no longer supported. udev_monitor_new_from_netlink() needs to be
5308 used to subscribe to events.
5310 * udev: when udevd is started by systemd, processes which are left
5311 behind by forking them off of udev rules, are unconditionally cleaned
5312 up and killed now after the event handling has finished. Services or
5313 daemons must be started as systemd services. Services can be
5314 pulled-in by udev to get started, but they can no longer be directly
5315 forked by udev rules.
5317 * udev: the daemon binary is called systemd-udevd now and installed
5318 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
5319 to adapt to that, create symlink, or rename the binary after building
5322 * libudev no longer provides these symbols:
5323 udev_monitor_from_socket()
5324 udev_queue_get_failed_list_entry()
5325 udev_get_{dev,sys,run}_path()
5326 The versions number was bumped and symbol versioning introduced.
5328 * systemd-loginctl and systemd-journalctl have been renamed
5329 to loginctl and journalctl to match systemctl.
5331 * The config files: /etc/systemd/systemd-logind.conf and
5332 /etc/systemd/systemd-journald.conf have been renamed to
5333 logind.conf and journald.conf. Package updates should rename
5334 the files to the new names on upgrade.
5336 * For almost all files the license is now LGPL2.1+, changed
5337 from the previous GPL2.0+. Exceptions are some minor stuff
5338 of udev (which will be changed to LGPL2.1 eventually, too),
5339 and the MIT licensed sd-daemon.[ch] library that is suitable
5340 to be used as drop-in files.
5342 * systemd and logind now handle system sleep states, in
5343 particular suspending and hibernating.
5345 * logind now implements a sleep/shutdown/idle inhibiting logic
5346 suitable for a variety of uses. Soonishly Lennart will blog
5347 about this in more detail.
5349 * var-run.mount and var-lock.mount are no longer provided
5350 (which prevously bind mounted these directories to their new
5351 places). Distributions which have not converted these
5352 directories to symlinks should consider stealing these files
5353 from git history and add them downstream.
5355 * We introduced the Documentation= field for units and added
5356 this to all our shipped units. This is useful to make it
5357 easier to explore the boot and the purpose of the various
5360 * All smaller setup units (such as
5361 systemd-vconsole-setup.service) now detect properly if they
5362 are run in a container and are skipped when
5363 appropriate. This guarantees an entirely noise-free boot in
5364 Linux container environments such as systemd-nspawn.
5366 * A framework for implementing offline system updates is now
5367 integrated, for details see:
5368 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
5370 * A new service type Type=idle is available now which helps us
5371 avoiding ugly interleaving of getty output and boot status
5374 * There's now a system-wide CapabilityBoundingSet= option to
5375 globally reduce the set of capabilities for the
5376 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
5377 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
5378 even CAP_NET_ADMIN system-wide for secure systems.
5380 * There are now system-wide DefaultLimitXXX= options to
5381 globally change the defaults of the various resource limits
5382 for all units started by PID 1.
5384 * Harald Hoyer's systemd test suite has been integrated into
5385 systemd which allows easy testing of systemd builds in qemu
5386 and nspawn. (This is really awesome! Ask us for details!)
5388 * The fstab parser is now implemented as generator, not inside
5391 * systemctl will now warn you if .mount units generated from
5392 /etc/fstab are out of date due to changes in fstab that
5393 have not been read by systemd yet.
5395 * systemd is now suitable for usage in initrds. Dracut has
5396 already been updated to make use of this. With this in place
5397 initrds get a slight bit faster but primarily are much
5398 easier to introspect and debug since "systemctl status" in
5399 the host system can be used to introspect initrd services,
5400 and the journal from the initrd is kept around too.
5402 * systemd-delta has been added, a tool to explore differences
5403 between user/admin configuration and vendor defaults.
5405 * PrivateTmp= now affects both /tmp and /var/tmp.
5407 * Boot time status messages are now much prettier and feature
5408 proper english language. Booting up systemd has never been
5411 * Read-ahead pack files now include the inode number of all
5412 files to pre-cache. When the inode changes the pre-caching
5413 is not attempted. This should be nicer to deal with updated
5414 packages which might result in changes of read-ahead
5417 * We now temporaritly lower the kernel's read_ahead_kb variable
5418 when collecting read-ahead data to ensure the kernel's
5419 built-in read-ahead does not add noise to our measurements
5420 of necessary blocks to pre-cache.
5422 * There's now RequiresMountsFor= to add automatic dependencies
5423 for all mounts necessary for a specific file system path.
5425 * MountAuto= and SwapAuto= have been removed from
5426 system.conf. Mounting file systems at boot has to take place
5429 * nspawn now learned a new switch --uuid= to set the machine
5430 ID on the command line.
5432 * nspawn now learned the -b switch to automatically search
5435 * vt102 is now the default TERM for serial TTYs, upgraded from
5438 * systemd-logind now works on VT-less systems.
5440 * The build tree has been reorganized. The individual
5441 components now have directories of their own.
5443 * A new condition type ConditionPathIsReadWrite= is now available.
5445 * nspawn learned the new -C switch to create cgroups for the
5446 container in other hierarchies.
5448 * We now have support for hardware watchdogs, configurable in
5451 * The scheduled shutdown logic now has a public API.
5453 * We now mount /tmp as tmpfs by default, but this can be
5454 masked and /etc/fstab can override it.
5456 * Since udisks does not make use of /media anymore we are not
5457 mounting a tmpfs on it anymore.
5459 * journalctl gained a new --local switch to only interleave
5460 locally generated journal files.
5462 * We can now load the IMA policy at boot automatically.
5464 * The GTK tools have been split off into a systemd-ui.
5466 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
5467 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
5468 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
5469 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
5470 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
5471 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
5472 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
5473 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
5474 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
5479 * This is mostly a bugfix release
5481 * Support optional initialization of the machine ID from the
5482 KVM or container configured UUID.
5484 * Support immediate reboots with "systemctl reboot -ff"
5486 * Show /etc/os-release data in systemd-analyze output
5488 * Many bugfixes for the journal, including endianness fixes and
5489 ensuring that disk space enforcement works
5491 * sd-login.h is C++ comptaible again
5493 * Extend the /etc/os-release format on request of the Debian
5496 * We now refuse non-UTF8 strings used in various configuration
5497 and unit files. This is done to ensure we do not pass invalid
5498 data over D-Bus or expose it elsewhere.
5500 * Register Mimo USB Screens as suitable for automatic seat
5503 * Read SELinux client context from journal clients in a race
5506 * Reorder configuration file lookup order. /etc now always
5507 overrides /run in order to allow the administrator to always
5508 and unconditionally override vendor-supplied or
5509 automatically generated data.
5511 * The various user visible bits of the journal now have man
5512 pages. We still lack man pages for the journal API calls
5515 * We now ship all man pages in HTML format again in the
5518 Contributions from: Dave Reisner, Dirk Eibach, Frederic
5519 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
5520 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
5525 * This is mostly a bugfix release
5527 * systems lacking /etc/os-release are no longer supported.
5529 * Various functionality updates to libsystemd-login.so
5531 * Track class of PAM logins to distinguish greeters from
5534 Contributions from: Kay Sievers, Lennart Poettering, Michael
5539 * This is an important bugfix release for v41.
5541 * Building man pages is now optional which should be useful
5542 for those building systemd from git but unwilling to install
5545 * Watchdog support for supervising services is now usable. In
5546 a future release support for hardware watchdogs
5547 (i.e. /dev/watchdog) will be added building on this.
5549 * Service start rate limiting is now configurable and can be
5550 turned off per service. When a start rate limit is hit a
5551 reboot can automatically be triggered.
5553 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
5555 Contributions from: Benjamin Franzke, Bill Nottingham,
5556 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
5557 Schmidt, Michał Górny, Piotr Drąg
5561 * The systemd binary is installed /usr/lib/systemd/systemd now;
5562 An existing /sbin/init symlink needs to be adapted with the
5565 * The code that loads kernel modules has been ported to invoke
5566 libkmod directly, instead of modprobe. This means we do not
5567 support systems with module-init-tools anymore.
5569 * Watchdog support is now already useful, but still not
5572 * A new kernel command line option systemd.setenv= is
5573 understood to set system wide environment variables
5574 dynamically at boot.
5576 * We now limit the set of capabilities of systemd-journald.
5578 * We now set SIGPIPE to ignore by default, since it only is
5579 useful in shell pipelines, and has little use in general
5580 code. This can be disabled with IgnoreSIPIPE=no in unit
5583 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
5584 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
5589 * This is mostly a bugfix release
5591 * We now expose the reason why a service failed in the
5592 "Result" D-Bus property.
5594 * Rudimentary service watchdog support (will be completed over
5595 the next few releases.)
5597 * When systemd forks off in order execute some service we will
5598 now immediately changes its argv[0] to reflect which process
5599 it will execute. This is useful to minimize the time window
5600 with a generic argv[0], which makes bootcharts more useful
5602 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
5603 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
5604 Mike Kazantsev, Ray Strode
5608 * This is mostly a test release, but incorporates many
5611 * New systemd-cgtop tool to show control groups by their
5614 * Linking against libacl for ACLs is optional again. If
5615 disabled, support tracking device access for active logins
5616 goes becomes unavailable, and so does access to the user
5617 journals by the respective users.
5619 * If a group "adm" exists, journal files are automatically
5620 owned by them, thus allow members of this group full access
5621 to the system journal as well as all user journals.
5623 * The journal now stores the SELinux context of the logging
5624 client for all entries.
5626 * Add C++ inclusion guards to all public headers
5628 * New output mode "cat" in the journal to print only text
5629 messages, without any meta data like date or time.
5631 * Include tiny X server wrapper as a temporary stop-gap to
5632 teach XOrg udev display enumeration. This is used by display
5633 managers such as gdm, and will go away as soon as XOrg
5634 learned native udev hotplugging for display devices.
5636 * Add new systemd-cat tool for executing arbitrary programs
5637 with STDERR/STDOUT connected to the journal. Can also act as
5638 BSD logger replacement, and does so by default.
5640 * Optionally store all locally generated coredumps in the
5641 journal along with meta data.
5643 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
5644 writing short strings to files (for usage for /sys), and for
5645 creating symlinks, character and block device nodes.
5647 * New unit file option ControlGroupPersistent= to make cgroups
5648 persistent, following the mechanisms outlined in
5649 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
5651 * Support multiple local RTCs in a sane way
5653 * No longer monopolize IO when replaying readahead data on
5654 rotating disks, since we might starve non-file-system IO to
5655 death, since fanotify() will not see accesses done by blkid,
5658 * Do not show kernel threads in systemd-cgls anymore, unless
5659 requested with new -k switch.
5661 Contributions from: Dan Horák, Kay Sievers, Lennart
5662 Poettering, Michal Schmidt
5666 * This is mostly a test release, but incorporates many
5669 * The git repository moved to:
5670 git://anongit.freedesktop.org/systemd/systemd
5671 ssh://git.freedesktop.org/git/systemd/systemd
5673 * First release with the journal
5674 http://0pointer.de/blog/projects/the-journal.html
5676 * The journal replaces both systemd-kmsg-syslogd and
5677 systemd-stdout-bridge.
5679 * New sd_pid_get_unit() API call in libsystemd-logind
5681 * Many systemadm clean-ups
5683 * Introduce remote-fs-pre.target which is ordered before all
5684 remote mounts and may be used to start services before all
5687 * Added Mageia support
5689 * Add bash completion for systemd-loginctl
5691 * Actively monitor PID file creation for daemons which exit in
5692 the parent process before having finished writing the PID
5693 file in the daemon process. Daemons which do this need to be
5694 fixed (i.e. PID file creation must have finished before the
5695 parent exits), but we now react a bit more gracefully to them.
5697 * Add colourful boot output, mimicking the well-known output
5698 of existing distributions.
5700 * New option PassCredentials= for socket units, for
5701 compatibility with a recent kernel ABI breakage.
5703 * /etc/rc.local is now hooked in via a generator binary, and
5704 thus will no longer act as synchronization point during
5707 * systemctl list-unit-files now supports --root=.
5709 * systemd-tmpfiles now understands two new commands: z, Z for
5710 relabelling files according to the SELinux database. This is
5711 useful to apply SELinux labels to specific files in /sys,
5714 * Output of SysV services is now forwarded to both the console
5715 and the journal by default, not only just the console.
5717 * New man pages for all APIs from libsystemd-login.
5719 * The build tree got reorganized and a the build system is a
5720 lot more modular allowing embedded setups to specifically
5721 select the components of systemd they are interested in.
5723 * Support for Linux systems lacking the kernel VT subsystem is
5726 * configure's --with-rootdir= got renamed to
5727 --with-rootprefix= to follow the naming used by udev and
5730 * Unless specified otherwise we will now install to /usr instead
5731 of /usr/local by default.
5733 * Processes with '@' in argv[0][0] are now excluded from the
5734 final shut-down killing spree, following the logic explained
5736 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
5738 * All processes remaining in a service cgroup when we enter
5739 the START or START_PRE states are now killed with
5740 SIGKILL. That means it is no longer possible to spawn
5741 background processes from ExecStart= lines (which was never
5742 supported anyway, and bad style).
5744 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
5745 reloading of units together.
5747 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
5748 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
5749 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
5750 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
5751 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
~ianmdlvl / elogind.git / blob