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 * Wherever systemd expects a calendar timestamp specification
20 (like in journalctl's --since= and --until= switches) UTC
21 timestamps are now supported. Timestamps suffixed with "UTC"
22 are now considered to be in Universal Time Coordinated
23 instead of the local timezone. Also, timestamps may now
24 optionally be specified with sub-second accuracy. Both of
25 these additions also apply to recurring calendar event
26 specification, such as OnCalendar= in timer units.
28 * journalctl gained a new "--sync" switch that asks the
29 journal daemon to write all so far unwritten log messages to
30 disk and sync the files, before returning.
32 * systemd-tmpfiles learned two new line types "q" and "Q" that
33 operate like "v", but also set up a basic btrfs quota
34 hierarchy when used on a btrfs file system with quota
37 * systemd-detect-virt gained a new --chroot switch to detect
38 whether execution takes place in a chroot() environment.
40 * CPUAffinity= now takes CPU index ranges in addition to
43 * The various memory-related resource limit settings (such as
44 LimitAS=) now understand the usual K, M, G, ... suffixes to
45 the base of 1024 (IEC). Similar, the time-related resource
46 limit settings understand the usual min, h, day, ...
49 * systemd-nspawn gained the new --network-veth-extra= switch
50 to define additional and arbitrarily-named virtual Ethernet
51 links between the host and the container.
53 * A new service execution setting PassEnvironment= has been
54 added that allows importing select environment variables
55 from PID1's environment block into the environment block of
58 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
59 default now (the kernel default is 16). This is beneficial
60 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
61 allows substantially larger numbers of queued
62 datagrams. This should increase the capability of systemd to
63 parallelize boot-up, as logging and sd_notify() are unlikely
64 to stall execution anymore. If you need to change the value
65 from the new defaults, use the usual sysctl.d/ snippets.
67 * The compression framing format used by the journal or
68 coredump processing has changed to be in line with what the
69 official LZ4 tools generate. LZ4 compression support in
70 systemd was considered unsupported previously, as the format
71 was not compatible with the normal tools. With this release
72 this has changed now, and it is hence safe for downstream
73 distributions to turn it on. While not compressing as well
74 as the XZ, LZ4 is substantially faster, which makes
75 it a good default choice for the compression logic in the
76 journal and in coredump handling.
78 * Any reference to /etc/mtab has been dropped from
79 systemd. The file has been obsolete since a while, but
80 systemd refused to work on systems where it was incorrectly
81 set up (it should be a symlink or non-existent). Please make
82 sure to update to util-linux 2.27.1 or newer in conjunction
83 with this systemd release, which also drops any reference to
84 /etc/mtab. If you maintain a distribution make sure that no
85 software you package still references it, as this is a
86 likely source of bugs. There's also a glibc bug pending,
87 asking for removal of any reference to this obsolete file:
89 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
91 * Support for the ".snapshot" unit type has been removed. This
92 feature turned out to be little useful and little used, and
93 has now been removed from the core and from systemctl.
95 * The dependency types RequiresOverridable= and
96 RequisiteOverridable= have been removed from systemd. They
97 have been used only very sparingly to our knowledge and
98 other options that provide a similar effect (such as
99 systemctl --mode=ignore-dependencies) are much more useful
100 and commonly used. Moreover, they were only half-way
101 implemented as the option to control behaviour regarding
102 these dependencies was never added to systemctl. By removing
103 these dependency types the execution engine becomes a bit
104 simpler. Unit files that use these dependencies should be
105 changed to use the non-Overridable dependency types
106 instead. In fact, when parsing unit files with these
107 options, that's what systemd will automatically convert them
108 too, but it will also warn, asking users to fix the unit
109 files accordingly. Removal of these dependency types should
110 only affect a negligible number of unit files in the wild.
112 * Behaviour of networkd's IPForward= option changed
113 (again). It will no longer maintain a per-interface setting,
114 but propagate one way from interfaces where this is enabled
115 to the global kernel setting. The global setting will be
116 enabled when requested by a network that is set up, but
117 never be disabled again. This change was made to make sure
118 IPv4 and IPv6 behaviour regarding packet forwarding is
119 similar (as the Linux IPv6 stack does not support
120 per-interface control of this setting) and to minimize
123 * In unit files the behaviour of %u, %U, %h, %s has
124 changed. These specifiers will now unconditionally resolve
125 to the various user database fields of the user that the
126 systemd instance is running as, instead of the user
127 configured in the specific unit via User=. Note that this
128 effectively doesn't change much, as resolving of these
129 specifiers was already turned off in the --system instance
130 of systemd, as we cannot do NSS lookups from PID 1. In the
131 --user instance of systemd these specifiers where correctly
132 resolved, but hardly made any sense, since the user instance
133 lacks privileges to do user switches anyway, and User= is
134 hence useless. Morever, even in the --user instance of
135 systemd behaviour was awkward as it would only take settings
136 from User= assignment placed before the specifier into
137 account. In order to unify and simplify the logic around
138 this the specifiers will now always resolve to the
139 credentials of the user invoking the manager (which in case
140 of PID 1 is the root user).
142 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
143 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
144 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
145 Filipe Brandenburger, Franck Bui, Hristo Venev, Iago López
146 Galeiras, Jan Engelhardt, Jan Janssen, Jan Synacek, Jesus
147 Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers, Lennart
148 Poettering, Mantas Mikulėnas, Marcel Holtmann, Marcin Bachry,
149 Marcos Alano, Marcos Mello, Mark Theunissen, Martin Pitt,
150 Michael Marineau, Michael Olbrich, Michal Schmidt, Michal
151 Sekletar, Mirco Tischler, Nick Owens, Nicolas Cornu, Patrik
152 Flykt, Peter Hutterer, reverendhomer, Ronny Chevalier,
153 Sangjung Woo, Seong-ho Cho, Shawn Landden, Susant Sahani,
154 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
155 Torstein Husebø, Vito Caputo, Zbigniew Jędrzejewski-Szmek
157 -- Berlin, 2015-11-XX
161 * systemd now depends on util-linux v2.27. More specifically,
162 the newly added mount monitor feature in libmount now
163 replaces systemd's former own implementation.
165 * libmount mandates /etc/mtab not to be regular file, and
166 systemd now enforces this condition at early boot.
167 /etc/mtab has been deprecated and warned about for a very
168 long time, so systems running systemd should already have
169 stopped having this file around as anything else than a
170 symlink to /proc/self/mounts.
172 * Support for the "pids" cgroup controller has been added. It
173 allows accounting the number of tasks in a cgroup and
174 enforcing limits on it. This adds two new setting
175 TasksAccounting= and TasksMax= to each unit, as well as a
176 global option DefaultTasksAccounting=.
178 * Support for the "net_cls" cgroup controller has been added.
179 It allows assigning a net class ID to each task in the
180 cgroup, which can then be used in firewall rules and traffic
181 shaping configurations. Note that the kernel netfilter net
182 class code does not currently work reliably for ingress
183 packets on unestablished sockets.
185 This adds a new config directive called NetClass= to CGroup
186 enabled units. Allowed values are positive numbers for fixed
187 assignments and "auto" for picking a free value
190 * 'systemctl is-system-running' now returns 'offline' if the
191 system is not booted with systemd. This command can now be
192 used as a substitute for 'systemd-notify --booted'.
194 * Watchdog timeouts have been increased to 3 minutes for all
195 in-tree service files. Apparently, disk IO issues are more
196 frequent than we hoped, and user reported >1 minute waiting
199 * 'machine-id-commit' functionality has been merged into
200 'machine-id-setup --commit'. The separate binary has been
203 * The WorkingDirectory= directive in unit files may now be set
204 to the special value '~'. In this case, the working
205 directory is set to the home directory of the user
208 * "machinectl shell" will now open the shell in the home
209 directory of the selected user by default.
211 * The CrashChVT= configuration file setting is renamed to
212 CrashChangeVT=, following our usual logic of not
213 abbreviating unnecessarily. The old directive is still
214 supported for compat reasons. Also, this directive now takes
215 an integer value between 1 and 63, or a boolean value. The
216 formerly supported '-1' value for disabling stays around for
219 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
220 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
221 RootDirectory= properties can now be set for transient
224 * The systemd-analyze tool gained a new "set-log-target" verb
225 to change the logging target the system manager logs to
226 dynamically during runtime. This is similar to how
227 "systemd-analyze set-log-level" already changes the log
230 * In nspawn /sys is now mounted as tmpfs, with only a selected
231 set of subdirectories mounted in from the real sysfs. This
232 enhances security slightly, and is useful for ensuring user
233 namespaces work correctly.
235 * Support for USB FunctionFS activation has been added. This
236 allows implementation of USB gadget services that are
237 activated as soon as they are requested, so that they don't
238 have to run continously, similar to classic socket
241 * The "systemctl exit" command now optionally takes an
242 additional parameter that sets the exit code to return from
243 the systemd manager when exiting. This is only relevant when
244 running the systemd user instance, or when running the
245 system instance in a container.
247 * sd-bus gained the new API calls sd_bus_path_encode_many()
248 and sd_bus_path_decode_many() that allow easy encoding and
249 decoding of multiple identifier strings inside a D-Bus
250 object path. Another new call sd_bus_default_flush_close()
251 has been added to flush and close per-thread default
254 * systemd-cgtop gained support for a -M/--machine= switch to
255 show the control groups within a certain container only.
257 * "systemctl kill" gained support for an optional --fail
258 switch. If specified the requested operation will fail of no
259 processes have been killed, because the unit had no
260 processes attached, or similar.
262 * A new systemd.crash_reboot=1 kernel command line option has
263 been added that triggers a reboot after crashing. This can
264 also be set through CrashReboot= in systemd.conf.
266 * The RuntimeDirectory= setting now understands unit
267 specifiers like %i or %f.
269 * A new (still internal) libary API sd-ipv4acd has been added,
270 that implements address conflict detection for IPv4. It's
271 based on code from sd-ipv4ll, and will be useful for
272 detecting DHCP address conflicts.
274 * File descriptors passed during socket activation may now be
275 named. A new API sd_listen_fds_with_names() is added to
276 access the names. The default names may be overridden,
277 either in the .socket file using the FileDescriptorName=
278 parameter, or by passing FDNAME= when storing the file
279 descriptors using sd_notify().
281 * systemd-networkd gained support for:
283 - Setting the IPv6 Router Advertisment settings via
284 IPv6AcceptRouterAdvertisements= in .network files.
286 - Configuring the HelloTimeSec=, MaxAgeSec= and
287 ForwardDelaySec= bridge parameters in .netdev files.
289 - Configuring PreferredSource= for static routes in
292 * The "ask-password" framework used to query for LUKS harddisk
293 passwords or SSL passwords during boot gained support for
294 caching passwords in the kernel keyring, if it is
295 available. This makes sure that the user only has to type in
296 a passphrase once if there are multiple objects to unlock
297 with the same one. Previously, such password caching was
298 available only when Plymouth was used; this moves the
299 caching logic into the systemd codebase itself. The
300 "systemd-ask-password" utility gained a new --keyname=
301 switch to control which kernel keyring key to use for
302 caching a password in. This functionality is also useful for
303 enabling display managers such as gdm to automatically
304 unlock the user's GNOME keyring if its passphrase, the
305 user's password and the harddisk password are the same, if
306 gdm-autologin is used.
308 * When downloading tar or raw images using "machinectl
309 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
310 file is now also downloaded, if it is available and stored
311 next to the image file.
313 * Units of type ".socket" gained a new boolean setting
314 Writable= which is only useful in conjunction with
315 ListenSpecial=. If true, enables opening the specified
316 special file in O_RDWR mode rather than O_RDONLY mode.
318 * systemd-rfkill has been reworked to become a singleton
319 service that is activated through /dev/rfkill on each rfkill
320 state change and saves the settings to disk. This way,
321 systemd-rfkill is now compatible with devices that exist
322 only intermittendly, and even restores state if the previous
323 system shutdown was abrupt rather than clean.
325 * The journal daemon gained support for vacuuming old journal
326 files controlled by the number of files that shall remain,
327 in addition to the already existing control by size and by
328 date. This is useful as journal interleaving performance
329 degrades with too many seperate journal files, and allows
330 putting an effective limit on them. The new setting defaults
331 to 100, but this may be changed by setting SystemMaxFiles=
332 and RuntimeMaxFiles= in journald.conf. Also, the
333 "journalctl" tool gained the new --vacuum-files= switch to
334 manually vacuum journal files to leave only the specified
335 number of files in place.
337 * udev will now create /dev/disk/by-path links for ATA devices
338 on kernels where that is supported.
340 * Galician, Serbian, Turkish and Korean translations were added.
342 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
343 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
344 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
345 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
346 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
347 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
348 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
349 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
350 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
351 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
352 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
353 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
354 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
355 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
356 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
357 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
358 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
359 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
361 -- Berlin, 2015-10-07
365 * The DHCP implementation of systemd-networkd gained a set of
368 - The DHCP server now supports emitting DNS and NTP
369 information. It may be enabled and configured via
370 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
371 and NTP information is enabled, but no servers are
372 configured, the corresponding uplink information (if there
373 is any) is propagated.
375 - Server and client now support transmission and reception
376 of timezone information. It can be configured via the
377 newly introduced network options UseTimezone=,
378 EmitTimezone=, and Timezone=. Transmission of timezone
379 information is enabled between host and containers by
380 default now: the container will change its local timezone
381 to what the host has set.
383 - Lease timeouts can now be configured via
384 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
386 - The DHCP server improved on the stability of
387 leases. Clients are more likely to get the same lease
388 information back, even if the server loses state.
390 - The DHCP server supports two new configuration options to
391 control the lease address pool metrics, PoolOffset= and
394 * The encapsulation limit of tunnels in systemd-networkd may
395 now be configured via 'EncapsulationLimit='. It allows
396 modifying the maximum additional levels of encapsulation
397 that are permitted to be prepended to a packet.
399 * systemd now supports the concept of user buses replacing
400 session buses, if used with dbus-1.10 (and enabled via dbus
401 --enable-user-session). It previously only supported this on
402 kdbus-enabled systems, and this release expands this to
403 'dbus-daemon' systems.
405 * systemd-networkd now supports predictable interface names
408 * systemd now optionally supports the new Linux kernel
409 "unified" control group hierarchy. If enabled via the kernel
410 command-line option 'systemd.unified_cgroup_hierarchy=1',
411 systemd will try to mount the unified cgroup hierarchy
412 directly on /sys/fs/cgroup. If not enabled, or not
413 available, systemd will fall back to the legacy cgroup
414 hierarchy setup, as before. Host system and containers can
415 mix and match legacy and unified hierarchies as they
416 wish. nspawn understands the $UNIFIED_CROUP_HIERARCHY
417 environment variable to individually select the hierarchy to
418 use for executed containers. By default, nspawn will use the
419 unified hierarchy for the containers if the host uses the
420 unified hierarchy, and the legacy hierarchy otherwise.
421 Please note that at this point the unified hierarchy is an
422 experimental kernel feature and is likely to change in one
423 of the next kernel releases. Therefore, it should not be
424 enabled by default in downstream distributions yet. The
425 minimum required kernel version for the unified hierarchy to
426 work is 4.2. Note that when the unified hierarchy is used
427 for the first time delegated access to controllers is
428 safe. Because of this systemd-nspawn containers will get
429 access to controllers now, as will systemd user
430 sessions. This means containers and user sessions may now
431 manage their own resources, partitioning up what the system
434 * A new special scope unit "init.scope" has been introduced
435 that encapsulates PID 1 of the system. It may be used to
436 determine resource usage and enforce resource limits on PID
437 1 itself. PID 1 hence moved out of the root of the control
440 * The cgtop tool gained support for filtering out kernel
441 threads when counting tasks in a control group. Also, the
442 count of processes is now recursively summed up by
443 default. Two options -k and --recursive= have been added to
444 revert to old behaviour. The tool has also been updated to
445 work correctly in containers now.
447 * systemd-nspawn's --bind= and --bind-ro= options have been
448 extended to allow creation of non-recursive bind mounts.
450 * libsystemd gained two new calls sd_pid_get_cgroup() and
451 sd_peer_get_cgroup() which return the control group path of
452 a process or peer of a connected AF_UNIX socket. This
453 function call is particularly useful when implementing
454 delegated subtrees support in the control group hierarchy.
456 * The "sd-event" event loop API of libsystemd now supports
457 correct dequeuing of real-time signals, without losing
460 * When systemd requests a PolicyKit decision when managing
461 units it will now add additional fields to the request,
462 including unit name and desired operation. This enables more
463 powerful PolicyKit policies, that make decisions depending
466 * nspawn learnt support for .nspawn settings files, that may
467 accompany the image files or directories of containers, and
468 may contain additional settings for the container. This is
469 an alternative to configuring container parameters via the
472 Contributions from: Cristian Rodríguez, Daniel Mack, David
473 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
474 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
475 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
476 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
477 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
478 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
479 Andersen, Tom Gundersen, Torstein Husebø
481 -- Berlin, 2015-09-08
485 * machinectl gained a new verb 'shell' which opens a fresh
486 shell on the target container or the host. It is similar to
487 the existing 'login' command of machinectl, but spawns the
488 shell directly without prompting for username or
489 password. The pseudo machine '.host' now refers to the local
490 host and is used by default. Hence, 'machinectl shell' can
491 be used as replacement for 'su -' which spawns a session as
492 a fresh systemd unit in a way that is fully isolated from
493 the originating session.
495 * systemd-networkd learned to cope with private-zone DHCP
496 options and allows other programs to query the values.
498 * SELinux access control when enabling/disabling units is no
499 longer enforced with this release. The previous
500 implementation was incorrect, and a new corrected
501 implementation is not yet available. As unit file operations
502 are still protected via PolicyKit and D-Bus policy this is
503 not a security problem. Yet, distributions which care about
504 optimal SELinux support should probably not stabilize on
507 * sd-bus gained support for matches of type "arg0has=", that
508 test for membership of strings in string arrays sent in bus
511 * systemd-resolved now dumps the contents of its DNS and LLMNR
512 caches to the logs on reception of the SIGUSR1 signal. This
513 is useful to debug DNS behaviour.
515 * The coredumpctl tool gained a new --directory= option to
516 operate on journal files in a specific directory.
518 * "systemctl reboot" and related commands gained a new
519 "--message=" option which may be used to set a free-text
520 wall message when shutting down or rebooting the
521 system. This message is also logged, which is useful for
522 figuring out the reason for a reboot or shutdown a
525 * The "systemd-resolve-host" tool's -i switch now takes
526 network interface numbers as alternative to interface names.
528 * A new unit file setting for services has been introduced:
529 UtmpMode= allows configuration of how precisely systemd
530 handles utmp and wtmp entries for the service if this is
531 enabled. This allows writing services that appear similar to
532 user sessions in the output of the "w", "who", "last" and
535 * systemd-resolved will now locally synthesize DNS resource
536 records for the "localhost" and "gateway" domains as well as
537 the local hostname. This should ensure that clients querying
538 RRs via resolved will get similar results as those going via
539 NSS, if nss-myhostname is enabled.
541 Contributions from: Alastair Hughes, Alex Crawford, Daniel
542 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
543 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
544 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
545 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
546 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
547 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
548 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
549 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
550 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
551 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
552 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
554 -- Berlin, 2015-08-27
558 * The systemd-efi-boot-generator functionality was merged into
559 systemd-gpt-auto-generator.
561 * systemd-networkd now supports Group Policy for vxlan
562 devices. It can be enabled via the new boolean configuration
563 option called 'GroupPolicyExtension='.
565 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
566 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
567 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
569 -- Berlin, 2015-07-31
573 * The python-systemd code has been removed from the systemd repository.
574 A new repository has been created which accommodates the code from
575 now on, and we kindly ask distributions to create a separate package
576 for this: https://github.com/systemd/python-systemd
578 * The systemd daemon will now reload its main configuration
579 (/etc/systemd/system.conf) on daemon-reload.
581 * sd-dhcp now exposes vendor specific extensions via
582 sd_dhcp_lease_get_vendor_specific().
584 * systemd-networkd gained a number of new configuration options.
586 - A new boolean configuration option for TAP devices called
587 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
588 device, thus allowing to send and receive GSO packets.
590 - A new tunnel configuration option called 'CopyDSCP='.
591 If enabled, the DSCP field of ip6 tunnels is copied into the
594 - A set of boolean bridge configuration options were added.
595 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
596 and 'UnicastFlood=' are now parsed by networkd and applied to the
597 respective bridge link device via the respective IFLA_BRPORT_*
600 - A new string configuration option to override the hostname sent
601 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
602 is true, networkd will use the configured hostname instead of the
603 system hostname when sending DHCP requests.
605 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
606 networkd will configure the IPv6 flow-label of the tunnel device
607 according to RFC2460.
609 - The 'macvtap' virtual network devices are now supported, similar to
610 the already supported 'macvlan' devices.
612 * systemd-resolved now implements RFC5452 to improve resilience against
613 cache poisoning. Additionally, source port randomization is enabled
614 by default to further protect against DNS spoofing attacks.
616 * nss-mymachines now supports translating UIDs and GIDs of running
617 containers with user-namespaces enabled. If a container 'foo'
618 translates a host uid 'UID' to the container uid 'TUID', then
619 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
620 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
621 mapped as 'vg-foo-TGID'.
623 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
624 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
625 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
626 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
627 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
628 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
629 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
630 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
631 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
632 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
634 -- Berlin, 2015-07-29
638 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
639 There are no known issues with current sysfs, and udev does not need
640 or should be used to work around such bugs.
642 * udev does no longer enable USB HID power management. Several reports
643 indicate, that some devices cannot handle that setting.
645 * The udev accelerometer helper was removed. The functionality
646 is now fully included in iio-sensor-proxy. But this means,
647 older iio-sensor-proxy versions will no longer provide
648 accelerometer/orientation data with this systemd version.
649 Please upgrade iio-sensor-proxy to version 1.0.
651 * networkd gained a new configuration option IPv6PrivacyExtensions=
652 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
653 for Stateless Address") on selected networks.
655 * For the sake of fewer build-time dependencies and less code in the
656 main repository, the python bindings are about to be removed in the
657 next release. A new repository has been created which accommodates
658 the code from now on, and we kindly ask distributions to create a
659 separate package for this. The removal will take place in v223.
661 https://github.com/systemd/python-systemd
663 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
664 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
665 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
666 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
667 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
668 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
669 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
670 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
671 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
672 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
674 -- Berlin, 2015-07-07
678 * The sd-bus.h and sd-event.h APIs have now been declared
679 stable and have been added to the official interface of
680 libsystemd.so. sd-bus implements an alternative D-Bus client
681 library, that is relatively easy to use, very efficient and
682 supports both classic D-Bus as well as kdbus as transport
683 backend. sd-event is a generic event loop abstraction that
684 is built around Linux epoll, but adds features such as event
685 prioritization or efficient timer handling. Both APIs are good
686 choices for C programs looking for a bus and/or event loop
687 implementation that is minimal and does not have to be
688 portable to other kernels.
690 * kdbus support is no longer compile-time optional. It is now
691 always built-in. However, it can still be disabled at
692 runtime using the kdbus=0 kernel command line setting, and
693 that setting may be changed to default to off, by specifying
694 --disable-kdbus at build-time. Note though that the kernel
695 command line setting has no effect if the kdbus.ko kernel
696 module is not installed, in which case kdbus is (obviously)
697 also disabled. We encourage all downstream distributions to
698 begin testing kdbus by adding it to the kernel images in the
699 development distributions, and leaving kdbus support in
702 * The minimal required util-linux version has been bumped to
705 * Support for chkconfig (--enable-chkconfig) was removed in
706 favor of calling an abstraction tool
707 /lib/systemd/systemd-sysv-install. This needs to be
708 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
709 in README for details.
711 * If there's a systemd unit and a SysV init script for the
712 same service name, and the user executes "systemctl enable"
713 for it (or a related call), then this will now enable both
714 (or execute the related operation on both), not just the
717 * The libudev API documentation has been converted from gtkdoc
720 * gudev has been removed from the systemd tree, it is now an
723 * The systemd-cgtop tool learnt a new --raw switch to generate
724 "raw" (machine parsable) output.
726 * networkd's IPForwarding= .network file setting learnt the
727 new setting "kernel", which ensures that networkd does not
728 change the IP forwarding sysctl from the default kernel
731 * The systemd-logind bus API now exposes a new boolean
732 property "Docked" that reports whether logind considers the
733 system "docked", i.e. connected to a docking station or not.
735 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
736 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
737 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
738 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
739 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
740 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
741 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
742 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
743 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
744 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
745 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
746 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
747 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
748 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
749 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
750 Fink, Zbigniew Jędrzejewski-Szmek
752 -- Berlin, 2015-06-19
756 * The gudev library has been extracted into a separate repository
757 available at: https://git.gnome.org/browse/libgudev/
758 It is now managed as part of the Gnome project. Distributions
759 are recommended to pass --disable-gudev to systemd and use
760 gudev from the Gnome project instead. gudev is still included
761 in systemd, for now. It will be removed soon, though. Please
762 also see the announcement-thread on systemd-devel:
763 http://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
765 * systemd now exposes a CPUUsageNSec= property for each
766 service unit on the bus, that contains the overall consumed
767 CPU time of a service (the sum of what each process of the
768 service consumed). This value is only available if
769 CPUAccounting= is turned on for a service, and is then shown
770 in the "systemctl status" output.
772 * Support for configuring alternative mappings of the old SysV
773 runlevels to systemd targets has been removed. They are now
774 hardcoded in a way that runlevels 2, 3, 4 all map to
775 multi-user.target and 5 to graphical.target (which
776 previously was already the default behaviour).
778 * The auto-mounter logic gained support for mount point
779 expiry, using a new TimeoutIdleSec= setting in .automount
780 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
782 * The EFI System Partition (ESP) as mounted to /boot by
783 systemd-efi-boot-generator will now be unmounted
784 automatically after 2 minutes of not being used. This should
785 minimize the risk of ESP corruptions.
787 * New /etc/fstab options x-systemd.requires= and
788 x-systemd.requires-mounts-for= are now supported to express
789 additional dependencies for mounts. This is useful for
790 journalling file systems that support external journal
791 devices or overlay file systems that require underlying file
792 systems to be mounted.
794 * systemd does not support direct live-upgrades (via systemctl
795 daemon-reexec) from versions older than v44 anymore. As no
796 distribution we are aware of shipped such old versions in a
797 stable release this should not be problematic.
799 * When systemd forks off a new per-connection service instance
800 it will now set the $REMOTE_ADDR environment variable to the
801 remote IP address, and $REMOTE_PORT environment variable to
802 the remote IP port. This behaviour is similar to the
803 corresponding environment variables defined by CGI.
805 * systemd-networkd gained support for uplink failure
806 detection. The BindCarrier= option allows binding interface
807 configuration dynamically to the link sense of other
808 interfaces. This is useful to achieve behaviour like in
811 * systemd-networkd gained support for configuring the DHCP
812 client identifier to use when requesting leases.
814 * systemd-networkd now has a per-network UseNTP= option to
815 configure whether NTP server information acquired via DHCP
816 is passed on to services like systemd-timesyncd.
818 * systemd-networkd gained support for vti6 tunnels.
820 * Note that systemd-networkd manages the sysctl variable
821 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
822 it is configured for since v219. The variable controls IP
823 forwarding, and is a per-interface alternative to the global
824 /proc/sys/net/ipv[46]/ip_forward. This setting is
825 configurable in the IPForward= option, which defaults to
826 "no". This means if networkd is used for an interface it is
827 no longer sufficient to set the global sysctl option to turn
828 on IP forwarding! Instead, the .network file option
829 IPForward= needs to be turned on! Note that the
830 implementation of this behaviour was broken in v219 and has
833 * Many bonding and vxlan options are now configurable in
836 * systemd-nspawn gained a new --property= setting to set unit
837 properties for the container scope. This is useful for
838 setting resource parameters (e.g "CPUShares=500") on
839 containers started from the command line.
841 * systemd-nspawn gained a new --private-users= switch to make
842 use of user namespacing available on recent Linux kernels.
844 * systemd-nspawn may now be called as part of a shell pipeline
845 in which case the pipes used for stdin and stdout are passed
846 directly to the process invoked in the container, without
847 indirection via a pseudo tty.
849 * systemd-nspawn gained a new switch to control the UNIX
850 signal to use when killing the init process of the container
853 * systemd-nspawn gained a new --overlay= switch for mounting
854 overlay file systems into the container using the new kernel
857 * When a container image is imported via systemd-importd and
858 the host file system is not btrfs, a loopback block device
859 file is created in /var/lib/machines.raw with a btrfs file
860 system inside. It is then mounted to /var/lib/machines to
861 enable btrfs features for container management. The loopback
862 file and btrfs file system is grown as needed when container
863 images are imported via systemd-importd.
865 * systemd-machined/systemd-importd gained support for btrfs
866 quota, to enforce container disk space limits on disk. This
867 is exposed in "machinectl set-limit".
869 * systemd-importd now can import containers from local .tar,
870 .raw and .qcow2 images, and export them to .tar and .raw. It
871 can also import dkr v2 images now from the network (on top
874 * systemd-importd gained support for verifying downloaded
875 images with gpg2 (previously only gpg1 was supported).
877 * systemd-machined, systemd-logind, systemd: most bus calls
878 are now accessible to unprivileged processes via
879 PolicyKit. Also, systemd-logind will now allow users to kill
880 their own sessions without further privileges or
883 * systemd-shutdownd has been removed. This service was
884 previously responsible for implementing scheduled shutdowns
885 as exposed in /usr/bin/shutdown's time parameter. This
886 functionality has now been moved into systemd-logind and is
887 accessible via a bus interface.
889 * "systemctl reboot" gained a new switch --firmware-setup that
890 can be used to reboot into the EFI firmware setup, if that
891 is available. systemd-logind now exposes an API on the bus
892 to trigger such reboots, in case graphical desktop UIs want
893 to cover this functionality.
895 * "systemctl enable", "systemctl disable" and "systemctl mask"
896 now support a new "--now" switch. If specified the units
897 that are enabled will also be started, and the ones
898 disabled/masked also stopped.
900 * The Gummiboot EFI boot loader tool has been merged into
901 systemd, and renamed to "systemd-boot". The bootctl tool has been
902 updated to support systemd-boot.
904 * An EFI kernel stub has been added that may be used to create
905 kernel EFI binaries that contain not only the actual kernel,
906 but also an initrd, boot splash, command line and OS release
907 information. This combined binary can then be signed as a
908 single image, so that the firmware can verify it all in one
909 step. systemd-boot has special support for EFI binaries created
910 like this and can extract OS release information from them
911 and show them in the boot menu. This functionality is useful
912 to implement cryptographically verified boot schemes.
914 * Optional support has been added to systemd-fsck to pass
915 fsck's progress report to an AF_UNIX socket in the file
918 * udev will no longer create device symlinks for all block
919 devices by default. A blacklist for excluding special block
920 devices from this logic has been turned into a whitelist
921 that requires picking block devices explicitly that require
924 * A new (currently still internal) API sd-device.h has been
925 added to libsystemd. This modernized API is supposed to
926 replace libudev eventually. In fact, already much of libudev
927 is now just a wrapper around sd-device.h.
929 * A new hwdb database for storing metadata about pointing
930 stick devices has been added.
932 * systemd-tmpfiles gained support for setting file attributes
933 similar to the "chattr" tool with new 'h' and 'H' lines.
935 * systemd-journald will no longer unconditionally set the
936 btrfs NOCOW flag on new journal files. This is instead done
937 with tmpfiles snippet using the new 'h' line type. This
938 allows easy disabling of this logic, by masking the
939 journal-nocow.conf tmpfiles file.
941 * systemd-journald will now translate audit message types to
942 human readable identifiers when writing them to the
943 journal. This should improve readability of audit messages.
945 * The LUKS logic gained support for the offset= and skip=
946 options in /etc/crypttab, as previously implemented by
949 * /usr/lib/os-release gained a new optional field VARIANT= for
950 distributions that support multiple variants (such as a
951 desktop edition, a server edition, ...)
953 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
954 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
955 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
956 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
957 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
958 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
959 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
960 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
961 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
962 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
963 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
964 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
965 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
966 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
967 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
968 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
969 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
970 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
971 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
972 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
973 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
974 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
975 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
976 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
977 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
978 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
979 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
981 -- Berlin, 2015-05-22
985 * Introduce a new API "sd-hwdb.h" for querying the hardware
986 metadata database. With this minimal interface one can query
987 and enumerate the udev hwdb, decoupled from the old libudev
988 library. libudev's interface for this is now only a wrapper
989 around sd-hwdb. A new tool systemd-hwdb has been added to
990 interface with and update the database.
992 * When any of systemd's tools copies files (for example due to
993 tmpfiles' C lines) a btrfs reflink will attempted first,
994 before bytewise copying is done.
996 * systemd-nspawn gained a new --ephemeral switch. When
997 specified a btrfs snapshot is taken of the container's root
998 directory, and immediately removed when the container
999 terminates again. Thus, a container can be started whose
1000 changes never alter the container's root directory, and are
1001 lost on container termination. This switch can also be used
1002 for starting a container off the root file system of the
1003 host without affecting the host OS. This switch is only
1004 available on btrfs file systems.
1006 * systemd-nspawn gained a new --template= switch. It takes the
1007 path to a container tree to use as template for the tree
1008 specified via --directory=, should that directory be
1009 missing. This allows instantiating containers dynamically,
1010 on first run. This switch is only available on btrfs file
1013 * When a .mount unit refers to a mount point on which multiple
1014 mounts are stacked, and the .mount unit is stopped all of
1015 the stacked mount points will now be unmounted until no
1016 mount point remains.
1018 * systemd now has an explicit notion of supported and
1019 unsupported unit types. Jobs enqueued for unsupported unit
1020 types will now fail with an "unsupported" error code. More
1021 specifically .swap, .automount and .device units are not
1022 supported in containers, .busname units are not supported on
1023 non-kdbus systems. .swap and .automount are also not
1024 supported if their respective kernel compile time options
1027 * machinectl gained support for two new "copy-from" and
1028 "copy-to" commands for copying files from a running
1029 container to the host or vice versa.
1031 * machinectl gained support for a new "bind" command to bind
1032 mount host directories into local containers. This is
1033 currently only supported for nspawn containers.
1035 * networkd gained support for configuring bridge forwarding
1036 database entries (fdb) from .network files.
1038 * A new tiny daemon "systemd-importd" has been added that can
1039 download container images in tar, raw, qcow2 or dkr formats,
1040 and make them available locally in /var/lib/machines, so
1041 that they can run as nspawn containers. The daemon can GPG
1042 verify the downloads (not supported for dkr, since it has no
1043 provisions for verifying downloads). It will transparently
1044 decompress bz2, xz, gzip compressed downloads if necessary,
1045 and restore sparse files on disk. The daemon uses privilege
1046 separation to ensure the actual download logic runs with
1047 fewer privileges than the daemon itself. machinectl has
1048 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
1049 make the functionality of importd available to the
1050 user. With this in place the Fedora and Ubuntu "Cloud"
1051 images can be downloaded and booted as containers unmodified
1052 (the Fedora images lack the appropriate GPG signature files
1053 currently, so they cannot be verified, but this will change
1054 soon, hopefully). Note that downloading images is currently
1055 only fully supported on btrfs.
1057 * machinectl is now able to list container images found in
1058 /var/lib/machines, along with some metadata about sizes of
1059 disk and similar. If the directory is located on btrfs and
1060 quota is enabled, this includes quota display. A new command
1061 "image-status" has been added that shows additional
1062 information about images.
1064 * machinectl is now able to clone container images
1065 efficiently, if the underlying file system (btrfs) supports
1066 it, with the new "machinectl list-images" command. It also
1067 gained commands for renaming and removing images, as well as
1068 marking them read-only or read-write (supported also on
1069 legacy file systems).
1071 * networkd gained support for collecting LLDP network
1072 announcements, from hardware that supports this. This is
1073 shown in networkctl output.
1075 * systemd-run gained support for a new -t (--pty) switch for
1076 invoking a binary on a pty whose input and output is
1077 connected to the invoking terminal. This allows executing
1078 processes as system services while interactively
1079 communicating with them via the terminal. Most interestingly
1080 this is supported across container boundaries. Invoking
1081 "systemd-run -t /bin/bash" is an alternative to running a
1082 full login session, the difference being that the former
1083 will not register a session, nor go through the PAM session
1086 * tmpfiles gained support for a new "v" line type for creating
1087 btrfs subvolumes. If the underlying file system is a legacy
1088 file system, this automatically degrades to creating a
1089 normal directory. Among others /var/lib/machines is now
1090 created like this at boot, should it be missing.
1092 * The directory /var/lib/containers/ has been deprecated and
1093 been replaced by /var/lib/machines. The term "machines" has
1094 been used in the systemd context as generic term for both
1095 VMs and containers, and hence appears more appropriate for
1096 this, as the directory can also contain raw images bootable
1099 * systemd-nspawn when invoked with -M but without --directory=
1100 or --image= is now capable of searching for the container
1101 root directory, subvolume or disk image automatically, in
1102 /var/lib/machines. systemd-nspawn@.service has been updated
1103 to make use of this, thus allowing it to be used for raw
1106 * A new machines.target unit has been introduced that is
1107 supposed to group all containers/VMs invoked as services on
1108 the system. systemd-nspawn@.service has been updated to
1109 integrate with that.
1111 * machinectl gained a new "start" command, for invoking a
1112 container as a service. "machinectl start foo" is mostly
1113 equivalent to "systemctl start systemd-nspawn@foo.service",
1114 but handles escaping in a nicer way.
1116 * systemd-nspawn will now mount most of the cgroupfs tree
1117 read-only into each container, with the exception of the
1118 container's own subtree in the name=systemd hierarchy.
1120 * journald now sets the special FS_NOCOW file flag for its
1121 journal files. This should improve performance on btrfs, by
1122 avoiding heavy fragmentation when journald's write-pattern
1123 is used on COW file systems. It degrades btrfs' data
1124 integrity guarantees for the files to the same levels as for
1125 ext3/ext4 however. This should be OK though as journald does
1126 its own data integrity checks and all its objects are
1127 checksummed on disk. Also, journald should handle btrfs disk
1128 full events a lot more gracefully now, by processing SIGBUS
1129 errors, and not relying on fallocate() anymore.
1131 * When journald detects that journal files it is writing to
1132 have been deleted it will immediately start new journal
1135 * systemd now provides a way to store file descriptors
1136 per-service in PID 1.This is useful for daemons to ensure
1137 that fds they require are not lost during a daemon
1138 restart. The fds are passed to the daemon on the next
1139 invocation in the same way socket activation fds are
1140 passed. This is now used by journald to ensure that the
1141 various sockets connected to all the system's stdout/stderr
1142 are not lost when journald is restarted. File descriptors
1143 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
1144 an extension to sd_notify(). Note that a limit is enforced
1145 on the number of fds a service can store in PID 1, and it
1146 defaults to 0, so that no fds may be stored, unless this is
1147 explicitly turned on.
1149 * The default TERM variable to use for units connected to a
1150 terminal, when no other value is explicitly is set is now
1151 vt220 rather than vt102. This should be fairly safe still,
1152 but allows PgUp/PgDn work.
1154 * The /etc/crypttab option header= as known from Debian is now
1157 * "loginctl user-status" and "loginctl session-status" will
1158 now show the last 10 lines of log messages of the
1159 user/session following the status output. Similar,
1160 "machinectl status" will show the last 10 log lines
1161 associated with a virtual machine or container
1162 service. (Note that this is usually not the log messages
1163 done in the VM/container itself, but simply what the
1164 container manager logs. For nspawn this includes all console
1167 * "loginctl session-status" without further argument will now
1168 show the status of the session of the caller. Similar,
1169 "lock-session", "unlock-session", "activate",
1170 "enable-linger", "disable-linger" may now be called without
1171 session/user parameter in which case they apply to the
1172 caller's session/user.
1174 * An X11 session scriptlet is now shipped that uploads
1175 $DISPLAY and $XAUTHORITY into the environment of the systemd
1176 --user daemon if a session begins. This should improve
1177 compatibility with X11 enabled applications run as systemd
1180 * Generators are now subject to masking via /etc and /run, the
1181 same way as unit files.
1183 * networkd .network files gained support for configuring
1184 per-link IPv4/IPv6 packet forwarding as well as IPv4
1185 masquerading. This is by default turned on for veth links to
1186 containers, as registered by systemd-nspawn. This means that
1187 nspawn containers run with --network-veth will now get
1188 automatic routed access to the host's networks without any
1189 further configuration or setup, as long as networkd runs on
1192 * systemd-nspawn gained the --port= (-p) switch to expose TCP
1193 or UDP posts of a container on the host. With this in place
1194 it is possible to run containers with private veth links
1195 (--network-veth), and have their functionality exposed on
1196 the host as if their services were running directly on the
1199 * systemd-nspawn's --network-veth switch now gained a short
1200 version "-n", since with the changes above it is now truly
1201 useful out-of-the-box. The systemd-nspawn@.service has been
1202 updated to make use of it too by default.
1204 * systemd-nspawn will now maintain a per-image R/W lock, to
1205 ensure that the same image is not started more than once
1206 writable. (It's OK to run an image multiple times
1207 simultaneously in read-only mode.)
1209 * systemd-nspawn's --image= option is now capable of
1210 dissecting and booting MBR and GPT disk images that contain
1211 only a single active Linux partition. Previously it
1212 supported only GPT disk images with proper GPT type
1213 IDs. This allows running cloud images from major
1214 distributions directly with systemd-nspawn, without
1217 * In addition to collecting mouse dpi data in the udev
1218 hardware database, there's now support for collecting angle
1219 information for mouse scroll wheels. The database is
1220 supposed to guarantee similar scrolling behavior on mice
1221 that it knows about. There's also support for collecting
1222 information about Touchpad types.
1224 * udev's input_id built-in will now also collect touch screen
1225 dimension data and attach it to probed devices.
1227 * /etc/os-release gained support for a Distribution Privacy
1230 * networkd gained support for creating "ipvlan", "gretap",
1231 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
1233 * systemd-tmpfiles gained support for "a" lines for setting
1236 * systemd-nspawn will now mount /tmp in the container to
1237 tmpfs, automatically.
1239 * systemd now exposes the memory.usage_in_bytes cgroup
1240 attribute and shows it for each service in the "systemctl
1241 status" output, if available.
1243 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
1244 immediate reboot is triggered. This useful if shutdown is
1245 hung and is unable to complete, to expedite the
1246 operation. Note that this kind of reboot will still unmount
1247 all file systems, and hence should not result in fsck being
1250 * A .device unit for an optical block device will now be
1251 considered active only when a medium is in the drive. Also,
1252 mount units are now bound to their backing devices thus
1253 triggering automatic unmounting when devices become
1254 unavailable. With this in place systemd will now
1255 automatically unmount left-over mounts when a CD-ROM is
1256 ejected or an USB stick is yanked from the system.
1258 * networkd-wait-online now has support for waiting for
1259 specific interfaces only (with globbing), and for giving up
1260 after a configurable timeout.
1262 * networkd now exits when idle. It will be automatically
1263 restarted as soon as interfaces show up, are removed or
1264 change state. networkd will stay around as long as there is
1265 at least one DHCP state machine or similar around, that keep
1268 * networkd may now configure IPv6 link-local addressing in
1269 addition to IPv4 link-local addressing.
1271 * The IPv6 "token" for use in SLAAC may now be configured for
1272 each .network interface in networkd.
1274 * Routes configured with networkd may now be assigned a scope
1277 * networkd's [Match] sections now support globbing and lists
1278 of multiple space-separated matches per item.
1280 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
1281 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
1282 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
1283 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
1284 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
1285 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
1286 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
1287 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
1288 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
1289 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
1290 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
1291 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
1292 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
1293 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
1294 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
1295 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
1296 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
1297 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
1298 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
1299 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
1300 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
1301 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
1302 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
1303 Hoffmann, Zbigniew Jędrzejewski-Szmek
1305 -- Berlin, 2015-02-16
1309 * When querying unit file enablement status (for example via
1310 "systemctl is-enabled"), a new state "indirect" is now known
1311 which indicates that a unit might not be enabled itself, but
1312 another unit listed in its Also= setting might be.
1314 * Similar to the various existing ConditionXYZ= settings for
1315 units, there are now matching AssertXYZ= settings. While
1316 failing conditions cause a unit to be skipped, but its job
1317 to succeed, failing assertions declared like this will cause
1318 a unit start operation and its job to fail.
1320 * hostnamed now knows a new chassis type "embedded".
1322 * systemctl gained a new "edit" command. When used on a unit
1323 file, this allows extending unit files with .d/ drop-in
1324 configuration snippets or editing the full file (after
1325 copying it from /usr/lib to /etc). This will invoke the
1326 user's editor (as configured with $EDITOR), and reload the
1327 modified configuration after editing.
1329 * "systemctl status" now shows the suggested enablement state
1330 for a unit, as declared in the (usually vendor-supplied)
1331 system preset files.
1333 * nss-myhostname will now resolve the single-label host name
1334 "gateway" to the locally configured default IP routing
1335 gateways, ordered by their metrics. This assigns a stable
1336 name to the used gateways, regardless which ones are
1337 currently configured. Note that the name will only be
1338 resolved after all other name sources (if nss-myhostname is
1339 configured properly) and should hence not negatively impact
1340 systems that use the single-label host name "gateway" in
1343 * systemd-inhibit now allows filtering by mode when listing
1346 * Scope and service units gained a new "Delegate" boolean
1347 property, which, when set, allows processes running inside the
1348 unit to further partition resources. This is primarily
1349 useful for systemd user instances as well as container
1352 * journald will now pick up audit messages directly from
1353 the kernel, and log them like any other log message. The
1354 audit fields are split up and fully indexed. This means that
1355 journalctl in many ways is now a (nicer!) alternative to
1356 ausearch, the traditional audit client. Note that this
1357 implements only a minimal audit client. If you want the
1358 special audit modes like reboot-on-log-overflow, please use
1359 the traditional auditd instead, which can be used in
1360 parallel to journald.
1362 * The ConditionSecurity= unit file option now understands the
1363 special string "audit" to check whether auditing is
1366 * journalctl gained two new commands --vacuum-size= and
1367 --vacuum-time= to delete old journal files until the
1368 remaining ones take up no more than the specified size on disk,
1369 or are not older than the specified time.
1371 * A new, native PPPoE library has been added to sd-network,
1372 systemd's library of light-weight networking protocols. This
1373 library will be used in a future version of networkd to
1374 enable PPPoE communication without an external pppd daemon.
1376 * The busctl tool now understands a new "capture" verb that
1377 works similar to "monitor", but writes a packet capture
1378 trace to STDOUT that can be redirected to a file which is
1379 compatible with libcap's capture file format. This can then
1380 be loaded in Wireshark and similar tools to inspect bus
1383 * The busctl tool now understands a new "tree" verb that shows
1384 the object trees of a specific service on the bus, or of all
1387 * The busctl tool now understands a new "introspect" verb that
1388 shows all interfaces and members of objects on the bus,
1389 including their signature and values. This is particularly
1390 useful to get more information about bus objects shown by
1391 the new "busctl tree" command.
1393 * The busctl tool now understands new verbs "call",
1394 "set-property" and "get-property" for invoking bus method
1395 calls, setting and getting bus object properties in a
1398 * busctl gained a new --augment-creds= argument that controls
1399 whether the tool shall augment credential information it
1400 gets from the bus with data from /proc, in a possibly
1403 * nspawn's --link-journal= switch gained two new values
1404 "try-guest" and "try-host" that work like "guest" and
1405 "host", but do not fail if the host has no persistent
1406 journalling enabled. -j is now equivalent to
1407 --link-journal=try-guest.
1409 * macvlan network devices created by nspawn will now have
1410 stable MAC addresses.
1412 * A new SmackProcessLabel= unit setting has been added, which
1413 controls the SMACK security label processes forked off by
1414 the respective unit shall use.
1416 * If compiled with --enable-xkbcommon, systemd-localed will
1417 verify x11 keymap settings by compiling the given keymap. It
1418 will spew out warnings if the compilation fails. This
1419 requires libxkbcommon to be installed.
1421 * When a coredump is collected, a larger number of metadata
1422 fields is now collected and included in the journal records
1423 created for it. More specifically, control group membership,
1424 environment variables, memory maps, working directory,
1425 chroot directory, /proc/$PID/status, and a list of open file
1426 descriptors is now stored in the log entry.
1428 * The udev hwdb now contains DPI information for mice. For
1431 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
1433 * All systemd programs that read standalone configuration
1434 files in /etc now also support a corresponding series of
1435 .conf.d configuration directories in /etc/, /run/,
1436 /usr/local/lib/, /usr/lib/, and (if configured with
1437 --enable-split-usr) /lib/. In particular, the following
1438 configuration files now have corresponding configuration
1439 directories: system.conf user.conf, logind.conf,
1440 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
1441 resolved.conf, timesyncd.conf, journal-remote.conf, and
1442 journal-upload.conf. Note that distributions should use the
1443 configuration directories in /usr/lib/; the directories in
1444 /etc/ are reserved for the system administrator.
1446 * systemd-rfkill will no longer take the rfkill device name
1447 into account when storing rfkill state on disk, as the name
1448 might be dynamically assigned and not stable. Instead, the
1449 ID_PATH udev variable combined with the rfkill type (wlan,
1450 bluetooth, ...) is used.
1452 * A new service systemd-machine-id-commit.service has been
1453 added. When used on systems where /etc is read-only during
1454 boot, and /etc/machine-id is not initialized (but an empty
1455 file), this service will copy the temporary machine ID
1456 created as replacement into /etc after the system is fully
1457 booted up. This is useful for systems that are freshly
1458 installed with a non-initialized machine ID, but should get
1459 a fixed machine ID for subsequent boots.
1461 * networkd's .netdev files now provide a large set of
1462 configuration parameters for VXLAN devices. Similarly, the
1463 bridge port cost parameter is now configurable in .network
1464 files. There's also new support for configuring IP source
1465 routing. networkd .link files gained support for a new
1466 OriginalName= match that is useful to match against the
1467 original interface name the kernel assigned. .network files
1468 may include MTU= and MACAddress= fields for altering the MTU
1469 and MAC address while being connected to a specific network
1472 * The LUKS logic gained supported for configuring
1473 UUID-specific key files. There's also new support for naming
1474 LUKS device from the kernel command line, using the new
1475 luks.name= argument.
1477 * Timer units may now be transiently created via the bus API
1478 (this was previously already available for scope and service
1479 units). In addition it is now possible to create multiple
1480 transient units at the same time with a single bus call. The
1481 "systemd-run" tool has been updated to make use of this for
1482 running commands on a specified time, in at(1)-style.
1484 * tmpfiles gained support for "t" lines, for assigning
1485 extended attributes to files. Among other uses this may be
1486 used to assign SMACK labels to files.
1488 Contributions from: Alin Rauta, Alison Chaiken, Andrej
1489 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
1490 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
1491 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
1492 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
1493 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
1494 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
1495 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
1496 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
1497 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
1498 Michael Biebl, Michael Chapman, Michael Marineau, Michal
1499 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
1500 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
1501 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
1502 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
1503 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
1504 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
1505 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
1507 -- Berlin, 2014-12-10
1511 * journalctl gained the new options -t/--identifier= to match
1512 on the syslog identifier (aka "tag"), as well as --utc to
1513 show log timestamps in the UTC timezone. journalctl now also
1514 accepts -n/--lines=all to disable line capping in a pager.
1516 * journalctl gained a new switch, --flush, that synchronously
1517 flushes logs from /run/log/journal to /var/log/journal if
1518 persistent storage is enabled. systemd-journal-flush.service
1519 now waits until the operation is complete.
1521 * Services can notify the manager before they start a reload
1522 (by sending RELOADING=1) or shutdown (by sending
1523 STOPPING=1). This allows the manager to track and show the
1524 internal state of daemons and closes a race condition when
1525 the process is still running but has closed its D-Bus
1528 * Services with Type=oneshot do not have to have any ExecStart
1531 * User units are now loaded also from
1532 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
1533 /run/systemd/user directory that was already previously
1534 supported, but is under the control of the user.
1536 * Job timeouts (i.e. time-outs on the time a job that is
1537 queued stays in the run queue) can now optionally result in
1538 immediate reboot or power-off actions (JobTimeoutAction= and
1539 JobTimeoutRebootArgument=). This is useful on ".target"
1540 units, to limit the maximum time a target remains
1541 undispatched in the run queue, and to trigger an emergency
1542 operation in such a case. This is now used by default to
1543 turn off the system if boot-up (as defined by everything in
1544 basic.target) hangs and does not complete for at least
1545 15min. Also, if power-off or reboot hang for at least 30min
1546 an immediate power-off/reboot operation is triggered. This
1547 functionality is particularly useful to increase reliability
1548 on embedded devices, but also on laptops which might
1549 accidentally get powered on when carried in a backpack and
1550 whose boot stays stuck in a hard disk encryption passphrase
1553 * systemd-logind can be configured to also handle lid switch
1554 events even when the machine is docked or multiple displays
1555 are attached (HandleLidSwitchDocked= option).
1557 * A helper binary and a service have been added which can be
1558 used to resume from hibernation in the initramfs. A
1559 generator will parse the resume= option on the kernel
1560 command line to trigger resume.
1562 * A user console daemon systemd-consoled has been
1563 added. Currently, it is a preview, and will so far open a
1564 single terminal on each session of the user marked as
1565 Desktop=systemd-console.
1567 * Route metrics can be specified for DHCP routes added by
1570 * The SELinux context of socket-activated services can be set
1571 from the information provided by the networking stack
1572 (SELinuxContextFromNet= option).
1574 * Userspace firmware loading support has been removed and
1575 the minimum supported kernel version is thus bumped to 3.7.
1577 * Timeout for udev workers has been increased from 1 to 3
1578 minutes, but a warning will be printed after 1 minute to
1579 help diagnose kernel modules that take a long time to load.
1581 * Udev rules can now remove tags on devices with TAG-="foobar".
1583 * systemd's readahead implementation has been removed. In many
1584 circumstances it didn't give expected benefits even for
1585 rotational disk drives and was becoming less relevant in the
1586 age of SSDs. As none of the developers has been using
1587 rotating media anymore, and nobody stepped up to actively
1588 maintain this component of systemd it has now been removed.
1590 * Swap units can use Options= to specify discard options.
1591 Discard options specified for swaps in /etc/fstab are now
1594 * Docker containers are now detected as a separate type of
1597 * The Password Agent protocol gained support for queries where
1598 the user input is shown, useful e.g. for user names.
1599 systemd-ask-password gained a new --echo option to turn that
1602 * The default sysctl.d/ snippets will now set:
1604 net.core.default_qdisc = fq_codel
1606 This selects Fair Queuing Controlled Delay as the default
1607 queuing discipline for network interfaces. fq_codel helps
1608 fight the network bufferbloat problem. It is believed to be
1609 a good default with no tuning required for most workloads.
1610 Downstream distributions may override this choice. On 10Gbit
1611 servers that do not do forwarding, "fq" may perform better.
1612 Systems without a good clocksource should use "pfifo_fast".
1614 * If kdbus is enabled during build a new option BusPolicy= is
1615 available for service units, that allows locking all service
1616 processes into a stricter bus policy, in order to limit
1617 access to various bus services, or even hide most of them
1618 from the service's view entirely.
1620 * networkctl will now show the .network and .link file
1621 networkd has applied to a specific interface.
1623 * sd-login gained a new API call sd_session_get_desktop() to
1624 query which desktop environment has been selected for a
1627 * UNIX utmp support is now compile-time optional to support
1628 legacy-free systems.
1630 * systemctl gained two new commands "add-wants" and
1631 "add-requires" for pulling in units from specific targets
1634 * If the word "rescue" is specified on the kernel command line
1635 the system will now boot into rescue mode (aka
1636 rescue.target), which was previously available only by
1637 specifying "1" or "systemd.unit=rescue.target" on the kernel
1638 command line. This new kernel command line option nicely
1639 mirrors the already existing "emergency" kernel command line
1642 * New kernel command line options mount.usr=, mount.usrflags=,
1643 mount.usrfstype= have been added that match root=, rootflags=,
1644 rootfstype= but allow mounting a specific file system to
1647 * The $NOTIFY_SOCKET is now also passed to control processes of
1648 services, not only the main process.
1650 * This version reenables support for fsck's -l switch. This
1651 means at least version v2.25 of util-linux is required for
1652 operation, otherwise dead-locks on device nodes may
1653 occur. Again: you need to update util-linux to at least
1654 v2.25 when updating systemd to v217.
1656 * The "multi-seat-x" tool has been removed from systemd, as
1657 its functionality has been integrated into X servers 1.16,
1658 and the tool is hence redundant. It is recommended to update
1659 display managers invoking this tool to simply invoke X
1660 directly from now on, again.
1662 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
1663 message flag has been added for all of systemd's PolicyKit
1664 authenticated method calls has been added. In particular
1665 this now allows optional interactive authorization via
1666 PolicyKit for many of PID1's privileged operations such as
1667 unit file enabling and disabling.
1669 * "udevadm hwdb --update" learnt a new switch "--usr" for
1670 placing the rebuilt hardware database in /usr instead of
1671 /etc. When used only hardware database entries stored in
1672 /usr will be used, and any user database entries in /etc are
1673 ignored. This functionality is useful for vendors to ship a
1674 pre-built database on systems where local configuration is
1675 unnecessary or unlikely.
1677 * Calendar time specifications in .timer units now also
1678 understand the strings "semi-annually", "quarterly" and
1679 "minutely" as shortcuts (in addition to the preexisting
1680 "anually", "hourly", ...).
1682 * systemd-tmpfiles will now correctly create files in /dev
1683 at boot which are marked for creation only at boot. It is
1684 recommended to always create static device nodes with 'c!'
1685 and 'b!', so that they are created only at boot and not
1686 overwritten at runtime.
1688 * When the watchdog logic is used for a service (WatchdogSec=)
1689 and the watchdog timeout is hit the service will now be
1690 terminated with SIGABRT (instead of just SIGTERM), in order
1691 to make sure a proper coredump and backtrace is
1692 generated. This ensures that hanging services will result in
1693 similar coredump/backtrace behaviour as services that hit a
1696 Contributions from: Andreas Henriksson, Andrei Borzenkov,
1697 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
1698 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
1699 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
1700 Herrmann, David Sommerseth, David Strauss, Emil Renner
1701 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
1702 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
1703 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
1704 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
1705 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
1706 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
1707 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
1708 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
1709 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
1710 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
1711 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
1712 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
1713 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
1714 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
1715 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
1716 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
1719 -- Berlin, 2014-10-28
1723 * timedated no longer reads NTP implementation unit names from
1724 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
1725 implementations should add a
1727 Conflicts=systemd-timesyncd.service
1729 to their unit files to take over and replace systemd's NTP
1730 default functionality.
1732 * systemd-sysusers gained a new line type "r" for configuring
1733 which UID/GID ranges to allocate system users/groups
1734 from. Lines of type "u" may now add an additional column
1735 that specifies the home directory for the system user to be
1736 created. Also, systemd-sysusers may now optionally read user
1737 information from STDIN instead of a file. This is useful for
1738 invoking it from RPM preinst scriptlets that need to create
1739 users before the first RPM file is installed since these
1740 files might need to be owned by them. A new
1741 %sysusers_create_inline RPM macro has been introduced to do
1742 just that. systemd-sysusers now updates the shadow files as
1743 well as the user/group databases, which should enhance
1744 compatibility with certain tools like grpck.
1746 * A number of bus APIs of PID 1 now optionally consult
1747 PolicyKit to permit access for otherwise unprivileged
1748 clients under certain conditions. Note that this currently
1749 doesn't support interactive authentication yet, but this is
1750 expected to be added eventually, too.
1752 * /etc/machine-info now has new fields for configuring the
1753 deployment environment of the machine, as well as the
1754 location of the machine. hostnamectl has been updated with
1755 new command to update these fields.
1757 * systemd-timesyncd has been updated to automatically acquire
1758 NTP server information from systemd-networkd, which might
1759 have been discovered via DHCP.
1761 * systemd-resolved now includes a caching DNS stub resolver
1762 and a complete LLMNR name resolution implementation. A new
1763 NSS module "nss-resolve" has been added which can be used
1764 instead of glibc's own "nss-dns" to resolve hostnames via
1765 systemd-resolved. Hostnames, addresses and arbitrary RRs may
1766 be resolved via systemd-resolved D-Bus APIs. In contrast to
1767 the glibc internal resolver systemd-resolved is aware of
1768 multi-homed system, and keeps DNS server and caches separate
1769 and per-interface. Queries are sent simultaneously on all
1770 interfaces that have DNS servers configured, in order to
1771 properly handle VPNs and local LANs which might resolve
1772 separate sets of domain names. systemd-resolved may acquire
1773 DNS server information from systemd-networkd automatically,
1774 which in turn might have discovered them via DHCP. A tool
1775 "systemd-resolve-host" has been added that may be used to
1776 query the DNS logic in resolved. systemd-resolved implements
1777 IDNA and automatically uses IDNA or UTF-8 encoding depending
1778 on whether classic DNS or LLMNR is used as transport. In the
1779 next releases we intend to add a DNSSEC and mDNS/DNS-SD
1780 implementation to systemd-resolved.
1782 * A new NSS module nss-mymachines has been added, that
1783 automatically resolves the names of all local registered
1784 containers to their respective IP addresses.
1786 * A new client tool "networkctl" for systemd-networkd has been
1787 added. It currently is entirely passive and will query
1788 networking configuration from udev, rtnetlink and networkd,
1789 and present it to the user in a very friendly
1790 way. Eventually, we hope to extend it to become a full
1791 control utility for networkd.
1793 * .socket units gained a new DeferAcceptSec= setting that
1794 controls the kernels' TCP_DEFER_ACCEPT sockopt for
1795 TCP. Similarly, support for controlling TCP keep-alive
1796 settings has been added (KeepAliveTimeSec=,
1797 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
1798 turning off Nagle's algorithm on TCP has been added
1801 * logind learned a new session type "web", for use in projects
1802 like Cockpit which register web clients as PAM sessions.
1804 * timer units with at least one OnCalendar= setting will now
1805 be started only after timer-sync.target has been
1806 reached. This way they will not elapse before the system
1807 clock has been corrected by a local NTP client or
1808 similar. This is particular useful on RTC-less embedded
1809 machines, that come up with an invalid system clock.
1811 * systemd-nspawn's --network-veth= switch should now result in
1812 stable MAC addresses for both the outer and the inner side
1815 * systemd-nspawn gained a new --volatile= switch for running
1816 container instances with /etc or /var unpopulated.
1818 * The kdbus client code has been updated to use the new Linux
1819 3.17 memfd subsystem instead of the old kdbus-specific one.
1821 * systemd-networkd's DHCP client and server now support
1822 FORCERENEW. There are also new configuration options to
1823 configure the vendor client identifier and broadcast mode
1826 * systemd will no longer inform the kernel about the current
1827 timezone, as this is necessarily incorrect and racy as the
1828 kernel has no understanding of DST and similar
1829 concepts. This hence means FAT timestamps will be always
1830 considered UTC, similar to what Android is already
1831 doing. Also, when the RTC is configured to the local time
1832 (rather than UTC) systemd will never synchronize back to it,
1833 as this might confuse Windows at a later boot.
1835 * systemd-analyze gained a new command "verify" for offline
1836 validation of unit files.
1838 * systemd-networkd gained support for a couple of additional
1839 settings for bonding networking setups. Also, the metric for
1840 statically configured routes may now be configured. For
1841 network interfaces where this is appropriate the peer IP
1842 address may now be configured.
1844 * systemd-networkd's DHCP client will no longer request
1845 broadcasting by default, as this tripped up some networks.
1846 For hardware where broadcast is required the feature should
1847 be switched back on using RequestBroadcast=yes.
1849 * systemd-networkd will now set up IPv4LL addresses (when
1850 enabled) even if DHCP is configured successfully.
1852 * udev will now default to respect network device names given
1853 by the kernel when the kernel indicates that these are
1854 predictable. This behavior can be tweaked by changing
1855 NamePolicy= in the relevant .link file.
1857 * A new library systemd-terminal has been added that
1858 implements full TTY stream parsing and rendering. This
1859 library is supposed to be used later on for implementing a
1860 full userspace VT subsystem, replacing the current kernel
1863 * A new tool systemd-journal-upload has been added to push
1864 journal data to a remote system running
1865 systemd-journal-remote.
1867 * journald will no longer forward all local data to another
1868 running syslog daemon. This change has been made because
1869 rsyslog (which appears to be the most commonly used syslog
1870 implementation these days) no longer makes use of this, and
1871 instead pulls the data out of the journal on its own. Since
1872 forwarding the messages to a non-existent syslog server is
1873 more expensive than we assumed we have now turned this
1874 off. If you run a syslog server that is not a recent rsyslog
1875 version, you have to turn this option on again
1876 (ForwardToSyslog= in journald.conf).
1878 * journald now optionally supports the LZ4 compressor for
1879 larger journal fields. This compressor should perform much
1880 better than XZ which was the previous default.
1882 * machinectl now shows the IP addresses of local containers,
1883 if it knows them, plus the interface name of the container.
1885 * A new tool "systemd-escape" has been added that makes it
1886 easy to escape strings to build unit names and similar.
1888 * sd_notify() messages may now include a new ERRNO= field
1889 which is parsed and collected by systemd and shown among the
1890 "systemctl status" output for a service.
1892 * A new component "systemd-firstboot" has been added that
1893 queries the most basic systemd information (timezone,
1894 hostname, root password) interactively on first
1895 boot. Alternatively it may also be used to provision these
1896 things offline on OS images installed into directories.
1898 * The default sysctl.d/ snippets will now set
1900 net.ipv4.conf.default.promote_secondaries=1
1902 This has the benefit of no flushing secondary IP addresses
1903 when primary addresses are removed.
1905 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
1906 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
1907 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
1908 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
1909 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
1910 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
1911 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
1912 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
1913 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
1914 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
1915 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
1916 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
1917 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
1918 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
1919 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
1921 -- Berlin, 2014-08-19
1925 * A new tool systemd-sysusers has been added. This tool
1926 creates system users and groups in /etc/passwd and
1927 /etc/group, based on static declarative system user/group
1928 definitions in /usr/lib/sysusers.d/. This is useful to
1929 enable factory resets and volatile systems that boot up with
1930 an empty /etc directory, and thus need system users and
1931 groups created during early boot. systemd now also ships
1932 with two default sysusers.d/ files for the most basic
1933 users and groups systemd and the core operating system
1936 * A new tmpfiles snippet has been added that rebuilds the
1937 essential files in /etc on boot, should they be missing.
1939 * A directive for ensuring automatic clean-up of
1940 /var/cache/man/ has been removed from the default
1941 configuration. This line should now be shipped by the man
1942 implementation. The necessary change has been made to the
1943 man-db implementation. Note that you need to update your man
1944 implementation to one that ships this line, otherwise no
1945 automatic clean-up of /var/cache/man will take place.
1947 * A new condition ConditionNeedsUpdate= has been added that
1948 may conditionalize services to only run when /etc or /var
1949 are "older" than the vendor operating system resources in
1950 /usr. This is useful for reconstructing or updating /etc
1951 after an offline update of /usr or a factory reset, on the
1952 next reboot. Services that want to run once after such an
1953 update or reset should use this condition and order
1954 themselves before the new systemd-update-done.service, which
1955 will mark the two directories as fully updated. A number of
1956 service files have been added making use of this, to rebuild
1957 the udev hardware database, the journald message catalog and
1958 dynamic loader cache (ldconfig). The systemd-sysusers tool
1959 described above also makes use of this now. With this in
1960 place it is now possible to start up a minimal operating
1961 system with /etc empty cleanly. For more information on the
1962 concepts involved see this recent blog story:
1964 http://0pointer.de/blog/projects/stateless.html
1966 * A new system group "input" has been introduced, and all
1967 input device nodes get this group assigned. This is useful
1968 for system-level software to get access to input devices. It
1969 complements what is already done for "audio" and "video".
1971 * systemd-networkd learnt minimal DHCPv4 server support in
1972 addition to the existing DHCPv4 client support. It also
1973 learnt DHCPv6 client and IPv6 Router Solicitation client
1974 support. The DHCPv4 client gained support for static routes
1975 passed in from the server. Note that the [DHCPv4] section
1976 known in older systemd-networkd versions has been renamed to
1977 [DHCP] and is now also used by the DHCPv6 client. Existing
1978 .network files using settings of this section should be
1979 updated, though compatibility is maintained. Optionally, the
1980 client hostname may now be sent to the DHCP server.
1982 * networkd gained support for vxlan virtual networks as well
1983 as tun/tap and dummy devices.
1985 * networkd gained support for automatic allocation of address
1986 ranges for interfaces from a system-wide pool of
1987 addresses. This is useful for dynamically managing a large
1988 number of interfaces with a single network configuration
1989 file. In particular this is useful to easily assign
1990 appropriate IP addresses to the veth links of a large number
1991 of nspawn instances.
1993 * RPM macros for processing sysusers, sysctl and binfmt
1994 drop-in snippets at package installation time have been
1997 * The /etc/os-release file should now be placed in
1998 /usr/lib/os-release. The old location is automatically
1999 created as symlink. /usr/lib is the more appropriate
2000 location of this file, since it shall actually describe the
2001 vendor operating system shipped in /usr, and not the
2002 configuration stored in /etc.
2004 * .mount units gained a new boolean SloppyOptions= setting
2005 that maps to mount(8)'s -s option which enables permissive
2006 parsing of unknown mount options.
2008 * tmpfiles learnt a new "L+" directive which creates a symlink
2009 but (unlike "L") deletes a pre-existing file first, should
2010 it already exist and not already be the correct
2011 symlink. Similarly, "b+", "c+" and "p+" directives have been
2012 added as well, which create block and character devices, as
2013 well as fifos in the filesystem, possibly removing any
2014 pre-existing files of different types.
2016 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
2017 'argument' field (which so far specified the source to
2018 symlink/copy the files from) is now optional. If omitted the
2019 same file os copied from /usr/share/factory/ suffixed by the
2020 full destination path. This is useful for populating /etc
2021 with essential files, by copying them from vendor defaults
2022 shipped in /usr/share/factory/etc.
2024 * A new command "systemctl preset-all" has been added that
2025 applies the service preset settings to all installed unit
2026 files. A new switch --preset-mode= has been added that
2027 controls whether only enable or only disable operations
2030 * A new command "systemctl is-system-running" has been added
2031 that allows checking the overall state of the system, for
2032 example whether it is fully up and running.
2034 * When the system boots up with an empty /etc, the equivalent
2035 to "systemctl preset-all" is executed during early boot, to
2036 make sure all default services are enabled after a factory
2039 * systemd now contains a minimal preset file that enables the
2040 most basic services systemd ships by default.
2042 * Unit files' [Install] section gained a new DefaultInstance=
2043 field for defining the default instance to create if a
2044 template unit is enabled with no instance specified.
2046 * A new passive target cryptsetup-pre.target has been added
2047 that may be used by services that need to make they run and
2048 finish before the first LUKS cryptographic device is set up.
2050 * The /dev/loop-control and /dev/btrfs-control device nodes
2051 are now owned by the "disk" group by default, opening up
2052 access to this group.
2054 * systemd-coredump will now automatically generate a
2055 stack trace of all core dumps taking place on the system,
2056 based on elfutils' libdw library. This stack trace is logged
2059 * systemd-coredump may now optionally store coredumps directly
2060 on disk (in /var/lib/systemd/coredump, possibly compressed),
2061 instead of storing them unconditionally in the journal. This
2062 mode is the new default. A new configuration file
2063 /etc/systemd/coredump.conf has been added to configure this
2064 and other parameters of systemd-coredump.
2066 * coredumpctl gained a new "info" verb to show details about a
2067 specific coredump. A new switch "-1" has also been added
2068 that makes sure to only show information about the most
2069 recent entry instead of all entries. Also, as the tool is
2070 generally useful now the "systemd-" prefix of the binary
2071 name has been removed. Distributions that want to maintain
2072 compatibility with the old name should add a symlink from
2073 the old name to the new name.
2075 * journald's SplitMode= now defaults to "uid". This makes sure
2076 that unprivileged users can access their own coredumps with
2077 coredumpctl without restrictions.
2079 * New kernel command line options "systemd.wants=" (for
2080 pulling an additional unit during boot), "systemd.mask="
2081 (for masking a specific unit for the boot), and
2082 "systemd.debug-shell" (for enabling the debug shell on tty9)
2083 have been added. This is implemented in the new generator
2084 "systemd-debug-generator".
2086 * systemd-nspawn will now by default filter a couple of
2087 syscalls for containers, among them those required for
2088 kernel module loading, direct x86 IO port access, swap
2089 management, and kexec. Most importantly though
2090 open_by_handle_at() is now prohibited for containers,
2091 closing a hole similar to a recently discussed vulnerability
2092 in docker regarding access to files on file hierarchies the
2093 container should normally not have access to. Note that, for
2094 nspawn, we generally make no security claims anyway (and
2095 this is explicitly documented in the man page), so this is
2096 just a fix for one of the most obvious problems.
2098 * A new man page file-hierarchy(7) has been added that
2099 contains a minimized, modernized version of the file system
2100 layout systemd expects, similar in style to the FHS
2101 specification or hier(5). A new tool systemd-path(1) has
2102 been added to query many of these paths for the local
2105 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
2106 longer done. Since the directory now has a per-user size
2107 limit, and is cleaned on logout this appears unnecessary,
2108 in particular since this now brings the lifecycle of this
2109 directory closer in line with how IPC objects are handled.
2111 * systemd.pc now exports a number of additional directories,
2112 including $libdir (which is useful to identify the library
2113 path for the primary architecture of the system), and a
2114 couple of drop-in directories.
2116 * udev's predictable network interface names now use the dev_port
2117 sysfs attribute, introduced in linux 3.15 instead of dev_id to
2118 distinguish between ports of the same PCI function. dev_id should
2119 only be used for ports using the same HW address, hence the need
2122 * machined has been updated to export the OS version of a
2123 container (read from /etc/os-release and
2124 /usr/lib/os-release) on the bus. This is now shown in
2125 "machinectl status" for a machine.
2127 * A new service setting RestartForceExitStatus= has been
2128 added. If configured to a set of exit signals or process
2129 return values, the service will be restarted when the main
2130 daemon process exits with any of them, regardless of the
2133 * systemctl's -H switch for connecting to remote systemd
2134 machines has been extended so that it may be used to
2135 directly connect to a specific container on the
2136 host. "systemctl -H root@foobar:waldi" will now connect as
2137 user "root" to host "foobar", and then proceed directly to
2138 the container named "waldi". Note that currently you have to
2139 authenticate as user "root" for this to work, as entering
2140 containers is a privileged operation.
2142 Contributions from: Andreas Henriksson, Benjamin Steinwender,
2143 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
2144 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
2145 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
2146 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
2147 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
2148 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
2149 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
2150 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
2151 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
2152 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
2153 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
2155 -- Berlin, 2014-07-03
2159 * As an experimental feature, udev now tries to lock the
2160 disk device node (flock(LOCK_SH|LOCK_NB)) while it
2161 executes events for the disk or any of its partitions.
2162 Applications like partitioning programs can lock the
2163 disk device node (flock(LOCK_EX)) and claim temporary
2164 device ownership that way; udev will entirely skip all event
2165 handling for this disk and its partitions. If the disk
2166 was opened for writing, the close will trigger a partition
2167 table rescan in udev's "watch" facility, and if needed
2168 synthesize "change" events for the disk and all its partitions.
2169 This is now unconditionally enabled, and if it turns out to
2170 cause major problems, we might turn it on only for specific
2171 devices, or might need to disable it entirely. Device Mapper
2172 devices are excluded from this logic.
2174 * We temporarily dropped the "-l" switch for fsck invocations,
2175 since they collide with the flock() logic above. util-linux
2176 upstream has been changed already to avoid this conflict,
2177 and we will readd "-l" as soon as util-linux with this
2178 change has been released.
2180 * The dependency on libattr has been removed. Since a long
2181 time, the extended attribute calls have moved to glibc, and
2182 libattr is thus unnecessary.
2184 * Virtualization detection works without priviliges now. This
2185 means the systemd-detect-virt binary no longer requires
2186 CAP_SYS_PTRACE file capabilities, and our daemons can run
2187 with fewer privileges.
2189 * systemd-networkd now runs under its own "systemd-network"
2190 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
2191 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
2192 loses the ability to write to files owned by root this way.
2194 * Similarly, systemd-resolved now runs under its own
2195 "systemd-resolve" user with no capabilities remaining.
2197 * Similarly, systemd-bus-proxyd now runs under its own
2198 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
2200 * systemd-networkd gained support for setting up "veth"
2201 virtual Ethernet devices for container connectivity, as well
2202 as GRE and VTI tunnels.
2204 * systemd-networkd will no longer automatically attempt to
2205 manually load kernel modules necessary for certain tunnel
2206 transports. Instead, it is assumed the kernel loads them
2207 automatically when required. This only works correctly on
2208 very new kernels. On older kernels, please consider adding
2209 the kernel modules to /etc/modules-load.d/ as a work-around.
2211 * The resolv.conf file systemd-resolved generates has been
2212 moved to /run/systemd/resolve/. If you have a symlink from
2213 /etc/resolv.conf, it might be necessary to correct it.
2215 * Two new service settings, ProtectHome= and ProtectSystem=,
2216 have been added. When enabled, they will make the user data
2217 (such as /home) inaccessible or read-only and the system
2218 (such as /usr) read-only, for specific services. This allows
2219 very light-weight per-service sandboxing to avoid
2220 modifications of user data or system files from
2221 services. These two new switches have been enabled for all
2222 of systemd's long-running services, where appropriate.
2224 * Socket units gained new SocketUser= and SocketGroup=
2225 settings to set the owner user and group of AF_UNIX sockets
2226 and FIFOs in the file system.
2228 * Socket units gained a new RemoveOnStop= setting. If enabled,
2229 all FIFOS and sockets in the file system will be removed
2230 when the specific socket unit is stopped.
2232 * Socket units gained a new Symlinks= setting. It takes a list
2233 of symlinks to create to file system sockets or FIFOs
2234 created by the specific Unix sockets. This is useful to
2235 manage symlinks to socket nodes with the same life-cycle as
2238 * The /dev/log socket and /dev/initctl FIFO have been moved to
2239 /run, and have been replaced by symlinks. This allows
2240 connecting to these facilities even if PrivateDevices=yes is
2241 used for a service (which makes /dev/log itself unavailable,
2242 but /run is left). This also has the benefit of ensuring
2243 that /dev only contains device nodes, directories and
2244 symlinks, and nothing else.
2246 * sd-daemon gained two new calls sd_pid_notify() and
2247 sd_pid_notifyf(). They are similar to sd_notify() and
2248 sd_notifyf(), but allow overriding of the source PID of
2249 notification messages if permissions permit this. This is
2250 useful to send notify messages on behalf of a different
2251 process (for example, the parent process). The
2252 systemd-notify tool has been updated to make use of this
2253 when sending messages (so that notification messages now
2254 originate from the shell script invoking systemd-notify and
2255 not the systemd-notify process itself. This should minimize
2256 a race where systemd fails to associate notification
2257 messages to services when the originating process already
2260 * A new "on-abnormal" setting for Restart= has been added. If
2261 set, it will result in automatic restarts on all "abnormal"
2262 reasons for a process to exit, which includes unclean
2263 signals, core dumps, timeouts and watchdog timeouts, but
2264 does not include clean and unclean exit codes or clean
2265 signals. Restart=on-abnormal is an alternative for
2266 Restart=on-failure for services that shall be able to
2267 terminate and avoid restarts on certain errors, by
2268 indicating so with an unclean exit code. Restart=on-failure
2269 or Restart=on-abnormal is now the recommended setting for
2270 all long-running services.
2272 * If the InaccessibleDirectories= service setting points to a
2273 mount point (or if there are any submounts contained within
2274 it), it is now attempted to completely unmount it, to make
2275 the file systems truly unavailable for the respective
2278 * The ReadOnlyDirectories= service setting and
2279 systemd-nspawn's --read-only parameter are now recursively
2280 applied to all submounts, too.
2282 * Mount units may now be created transiently via the bus APIs.
2284 * The support for SysV and LSB init scripts has been removed
2285 from the systemd daemon itself. Instead, it is now
2286 implemented as a generator that creates native systemd units
2287 from these scripts when needed. This enables us to remove a
2288 substantial amount of legacy code from PID 1, following the
2289 fact that many distributions only ship a very small number
2290 of LSB/SysV init scripts nowadays.
2292 * Privileged Xen (dom0) domains are not considered
2293 virtualization anymore by the virtualization detection
2294 logic. After all, they generally have unrestricted access to
2295 the hardware and usually are used to manage the unprivileged
2298 * systemd-tmpfiles gained a new "C" line type, for copying
2299 files or entire directories.
2301 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
2302 lines. So far, they have been non-globbing versions of the
2303 latter, and have thus been redundant. In future, it is
2304 recommended to only use "z". "m" has hence been removed
2305 from the documentation, even though it stays supported.
2307 * A tmpfiles snippet to recreate the most basic structure in
2308 /var has been added. This is enough to create the /var/run →
2309 /run symlink and create a couple of structural
2310 directories. This allows systems to boot up with an empty or
2311 volatile /var. Of course, while with this change, the core OS
2312 now is capable with dealing with a volatile /var, not all
2313 user services are ready for it. However, we hope that sooner
2314 or later, many service daemons will be changed upstream so
2315 that they are able to automatically create their necessary
2316 directories in /var at boot, should they be missing. This is
2317 the first step to allow state-less systems that only require
2318 the vendor image for /usr to boot.
2320 * systemd-nspawn has gained a new --tmpfs= switch to mount an
2321 empty tmpfs instance to a specific directory. This is
2322 particularly useful for making use of the automatic
2323 reconstruction of /var (see above), by passing --tmpfs=/var.
2325 * Access modes specified in tmpfiles snippets may now be
2326 prefixed with "~", which indicates that they shall be masked
2327 by whether the existing file or directory is currently
2328 writable, readable or executable at all. Also, if specified,
2329 the sgid/suid/sticky bits will be masked for all
2332 * A new passive target unit "network-pre.target" has been
2333 added which is useful for services that shall run before any
2334 network is configured, for example firewall scripts.
2336 * The "floppy" group that previously owned the /dev/fd*
2337 devices is no longer used. The "disk" group is now used
2338 instead. Distributions should probably deprecate usage of
2341 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
2342 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
2343 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
2344 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
2345 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
2346 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
2349 -- Berlin, 2014-06-11
2353 * A new "systemd-timesyncd" daemon has been added for
2354 synchronizing the system clock across the network. It
2355 implements an SNTP client. In contrast to NTP
2356 implementations such as chrony or the NTP reference server,
2357 this only implements a client side, and does not bother with
2358 the full NTP complexity, focusing only on querying time from
2359 one remote server and synchronizing the local clock to
2360 it. Unless you intend to serve NTP to networked clients or
2361 want to connect to local hardware clocks, this simple NTP
2362 client should be more than appropriate for most
2363 installations. The daemon runs with minimal privileges, and
2364 has been hooked up with networkd to only operate when
2365 network connectivity is available. The daemon saves the
2366 current clock to disk every time a new NTP sync has been
2367 acquired, and uses this to possibly correct the system clock
2368 early at bootup, in order to accommodate for systems that
2369 lack an RTC such as the Raspberry Pi and embedded devices,
2370 and to make sure that time monotonically progresses on these
2371 systems, even if it is not always correct. To make use of
2372 this daemon, a new system user and group "systemd-timesync"
2373 needs to be created on installation of systemd.
2375 * The queue "seqnum" interface of libudev has been disabled, as
2376 it was generally incompatible with device namespacing as
2377 sequence numbers of devices go "missing" if the devices are
2378 part of a different namespace.
2380 * "systemctl list-timers" and "systemctl list-sockets" gained
2381 a --recursive switch for showing units of these types also
2382 for all local containers, similar in style to the already
2383 supported --recursive switch for "systemctl list-units".
2385 * A new RebootArgument= setting has been added for service
2386 units, which may be used to specify a kernel reboot argument
2387 to use when triggering reboots with StartLimitAction=.
2389 * A new FailureAction= setting has been added for service
2390 units which may be used to specify an operation to trigger
2391 when a service fails. This works similarly to
2392 StartLimitAction=, but unlike it, controls what is done
2393 immediately rather than only after several attempts to
2394 restart the service in question.
2396 * hostnamed got updated to also expose the kernel name,
2397 release, and version on the bus. This is useful for
2398 executing commands like hostnamectl with the -H switch.
2399 systemd-analyze makes use of this to properly display
2400 details when running non-locally.
2402 * The bootchart tool can now show cgroup information in the
2403 graphs it generates.
2405 * The CFS CPU quota cgroup attribute is now exposed for
2406 services. The new CPUQuota= switch has been added for this
2407 which takes a percentage value. Setting this will have the
2408 result that a service may never get more CPU time than the
2409 specified percentage, even if the machine is otherwise idle.
2411 * systemd-networkd learned IPIP and SIT tunnel support.
2413 * LSB init scripts exposing a dependency on $network will now
2414 get a dependency on network-online.target rather than simply
2415 network.target. This should bring LSB handling closer to
2416 what it was on SysV systems.
2418 * A new fsck.repair= kernel option has been added to control
2419 how fsck shall deal with unclean file systems at boot.
2421 * The (.ini) configuration file parser will now silently
2422 ignore sections whose name begins with "X-". This may be
2423 used to maintain application-specific extension sections in unit
2426 * machined gained a new API to query the IP addresses of
2427 registered containers. "machinectl status" has been updated
2428 to show these addresses in its output.
2430 * A new call sd_uid_get_display() has been added to the
2431 sd-login APIs for querying the "primary" session of a
2432 user. The "primary" session of the user is elected from the
2433 user's sessions and generally a graphical session is
2434 preferred over a text one.
2436 * A minimal systemd-resolved daemon has been added. It
2437 currently simply acts as a companion to systemd-networkd and
2438 manages resolv.conf based on per-interface DNS
2439 configuration, possibly supplied via DHCP. In the long run
2440 we hope to extend this into a local DNSSEC enabled DNS and
2443 * The systemd-networkd-wait-online tool is now enabled by
2444 default. It will delay network-online.target until a network
2445 connection has been configured. The tool primarily integrates
2446 with networkd, but will also make a best effort to make sense
2447 of network configuration performed in some other way.
2449 * Two new service options StartupCPUShares= and
2450 StartupBlockIOWeight= have been added that work similarly to
2451 CPUShares= and BlockIOWeight= however only apply during
2452 system startup. This is useful to prioritize certain services
2453 differently during bootup than during normal runtime.
2455 * hostnamed has been changed to prefer the statically
2456 configured hostname in /etc/hostname (unless set to
2457 'localhost' or empty) over any dynamic one supplied by
2458 dhcp. With this change, the rules for picking the hostname
2459 match more closely the rules of other configuration settings
2460 where the local administrator's configuration in /etc always
2461 overrides any other settings.
2463 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
2464 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
2465 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
2466 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
2467 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
2468 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
2469 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
2470 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
2471 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
2472 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
2473 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
2474 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
2475 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
2476 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
2477 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
2478 Lindskog, WaLyong Cho, Will Woods, Zbigniew
2481 -- Beijing, 2014-05-28
2485 * When restoring the screen brightness at boot, stay away from
2486 the darkest setting or from the lowest 5% of the available
2487 range, depending on which is the larger value of both. This
2488 should effectively protect the user from rebooting into a
2489 black screen, should the brightness have been set to minimum
2492 * sd-login gained a new sd_machine_get_class() call to
2493 determine the class ("vm" or "container") of a machine
2494 registered with machined.
2496 * sd-login gained new calls
2497 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
2498 to query the identity of the peer of a local AF_UNIX
2499 connection. They operate similarly to their sd_pid_get_xyz()
2502 * PID 1 will now maintain a system-wide system state engine
2503 with the states "starting", "running", "degraded",
2504 "maintenance", "stopping". These states are bound to system
2505 startup, normal runtime, runtime with at least one failed
2506 service, rescue/emergency mode and system shutdown. This
2507 state is shown in the "systemctl status" output when no unit
2508 name is passed. It is useful to determine system state, in
2509 particularly when doing so for many systems or containers at
2512 * A new command "list-machines" has been added to "systemctl"
2513 that lists all local OS containers and shows their system
2514 state (see above), if systemd runs inside of them.
2516 * systemctl gained a new "-r" switch to recursively enumerate
2517 units on all local containers, when used with the
2518 "list-unit" command (which is the default one that is
2519 executed when no parameters are specified).
2521 * The GPT automatic partition discovery logic will now honour
2522 two GPT partition flags: one may be set on a partition to
2523 cause it to be mounted read-only, and the other may be set
2524 on a partition to ignore it during automatic discovery.
2526 * Two new GPT type UUIDs have been added for automatic root
2527 partition discovery, for 32-bit and 64-bit ARM. This is not
2528 particularly useful for discovering the root directory on
2529 these architectures during bare-metal boots (since UEFI is
2530 not common there), but still very useful to allow booting of
2531 ARM disk images in nspawn with the -i option.
2533 * MAC addresses of interfaces created with nspawn's
2534 --network-interface= switch will now be generated from the
2535 machine name, and thus be stable between multiple invocations
2538 * logind will now automatically remove all IPC objects owned
2539 by a user if she or he fully logs out. This makes sure that
2540 users who are logged out cannot continue to consume IPC
2541 resources. This covers SysV memory, semaphores and message
2542 queues as well as POSIX shared memory and message
2543 queues. Traditionally, SysV and POSIX IPC had no life-cycle
2544 limits. With this functionality, that is corrected. This may
2545 be turned off by using the RemoveIPC= switch of logind.conf.
2547 * The systemd-machine-id-setup and tmpfiles tools gained a
2548 --root= switch to operate on a specific root directory,
2551 * journald can now forward logged messages to the TTYs of all
2552 logged in users ("wall"). This is the default for all
2553 emergency messages now.
2555 * A new tool systemd-journal-remote has been added to stream
2556 journal log messages across the network.
2558 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
2559 controller trees are mounted into it. Note that the
2560 directories mounted beneath it are not read-only. This is a
2561 security measure and is particularly useful because glibc
2562 actually includes a search logic to pick any tmpfs it can
2563 find to implement shm_open() if /dev/shm is not available
2564 (which it might very well be in namespaced setups).
2566 * machinectl gained a new "poweroff" command to cleanly power
2567 down a local OS container.
2569 * The PrivateDevices= unit file setting will now also drop the
2570 CAP_MKNOD capability from the capability bound set, and
2571 imply DevicePolicy=closed.
2573 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
2574 comprehensively on all long-running systemd services where
2575 this is appropriate.
2577 * systemd-udevd will now run in a disassociated mount
2578 namespace. To mount directories from udev rules, make sure to
2579 pull in mount units via SYSTEMD_WANTS properties.
2581 * The kdbus support gained support for uploading policy into
2582 the kernel. sd-bus gained support for creating "monitoring"
2583 connections that can eavesdrop into all bus communication
2584 for debugging purposes.
2586 * Timestamps may now be specified in seconds since the UNIX
2587 epoch Jan 1st, 1970 by specifying "@" followed by the value
2590 * Native tcpwrap support in systemd has been removed. tcpwrap
2591 is old code, not really maintained anymore and has serious
2592 shortcomings, and better options such as firewalls
2593 exist. For setups that require tcpwrap usage, please
2594 consider invoking your socket-activated service via tcpd,
2595 like on traditional inetd.
2597 * A new system.conf configuration option
2598 DefaultTimerAccuracySec= has been added that controls the
2599 default AccuracySec= setting of .timer units.
2601 * Timer units gained a new WakeSystem= switch. If enabled,
2602 timers configured this way will cause the system to resume
2603 from system suspend (if the system supports that, which most
2606 * Timer units gained a new Persistent= switch. If enabled,
2607 timers configured this way will save to disk when they have
2608 been last triggered. This information is then used on next
2609 reboot to possible execute overdue timer events, that
2610 could not take place because the system was powered off.
2611 This enables simple anacron-like behaviour for timer units.
2613 * systemctl's "list-timers" will now also list the time a
2614 timer unit was last triggered in addition to the next time
2615 it will be triggered.
2617 * systemd-networkd will now assign predictable IPv4LL
2618 addresses to its local interfaces.
2620 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
2621 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
2622 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
2623 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
2624 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
2625 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
2626 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
2627 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
2630 -- Berlin, 2014-03-25
2634 * A new unit file setting RestrictAddressFamilies= has been
2635 added to restrict which socket address families unit
2636 processes gain access to. This takes address family names
2637 like "AF_INET" or "AF_UNIX", and is useful to minimize the
2638 attack surface of services via exotic protocol stacks. This
2639 is built on seccomp system call filters.
2641 * Two new unit file settings RuntimeDirectory= and
2642 RuntimeDirectoryMode= have been added that may be used to
2643 manage a per-daemon runtime directories below /run. This is
2644 an alternative for setting up directory permissions with
2645 tmpfiles snippets, and has the advantage that the runtime
2646 directory's lifetime is bound to the daemon runtime and that
2647 the daemon starts up with an empty directory each time. This
2648 is particularly useful when writing services that drop
2649 privileges using the User= or Group= setting.
2651 * The DeviceAllow= unit setting now supports globbing for
2652 matching against device group names.
2654 * The systemd configuration file system.conf gained new
2655 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
2656 DefaultMemoryAccounting= to globally turn on/off accounting
2657 for specific resources (cgroups) for all units. These
2658 settings may still be overridden individually in each unit
2661 * systemd-gpt-auto-generator is now able to discover /srv and
2662 root partitions in addition to /home and swap partitions. It
2663 also supports LUKS-encrypted partitions now. With this in
2664 place, automatic discovery of partitions to mount following
2665 the Discoverable Partitions Specification
2666 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
2667 is now a lot more complete. This allows booting without
2668 /etc/fstab and without root= on the kernel command line on
2669 systems prepared appropriately.
2671 * systemd-nspawn gained a new --image= switch which allows
2672 booting up disk images and Linux installations on any block
2673 device that follow the Discoverable Partitions Specification
2674 (see above). This means that installations made with
2675 appropriately updated installers may now be started and
2676 deployed using container managers, completely
2677 unmodified. (We hope that libvirt-lxc will add support for
2678 this feature soon, too.)
2680 * systemd-nspawn gained a new --network-macvlan= setting to
2681 set up a private macvlan interface for the
2682 container. Similarly, systemd-networkd gained a new
2683 Kind=macvlan setting in .netdev files.
2685 * systemd-networkd now supports configuring local addresses
2688 * A new tool systemd-network-wait-online has been added to
2689 synchronously wait for network connectivity using
2692 * The sd-bus.h bus API gained a new sd_bus_track object for
2693 tracking the life-cycle of bus peers. Note that sd-bus.h is
2694 still not a public API though (unless you specify
2695 --enable-kdbus on the configure command line, which however
2696 voids your warranty and you get no API stability guarantee).
2698 * The $XDG_RUNTIME_DIR runtime directories for each user are
2699 now individual tmpfs instances, which has the benefit of
2700 introducing separate pools for each user, with individual
2701 size limits, and thus making sure that unprivileged clients
2702 can no longer negatively impact the system or other users by
2703 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
2704 RuntimeDirectorySize= has been introduced that allows
2705 controlling the default size limit for all users. It
2706 defaults to 10% of the available physical memory. This is no
2707 replacement for quotas on tmpfs though (which the kernel
2708 still does not support), as /dev/shm and /tmp are still
2709 shared resources used by both the system and unprivileged
2712 * logind will now automatically turn off automatic suspending
2713 on laptop lid close when more than one display is
2714 connected. This was previously expected to be implemented
2715 individually in desktop environments (such as GNOME),
2716 however has been added to logind now, in order to fix a
2717 boot-time race where a desktop environment might not have
2718 been started yet and thus not been able to take an inhibitor
2719 lock at the time where logind already suspends the system
2720 due to a closed lid.
2722 * logind will now wait at least 30s after each system
2723 suspend/resume cycle, and 3min after system boot before
2724 suspending the system due to a closed laptop lid. This
2725 should give USB docking stations and similar enough time to
2726 be probed and configured after system resume and boot in
2727 order to then act as suspend blocker.
2729 * systemd-run gained a new --property= setting which allows
2730 initialization of resource control properties (and others)
2731 for the created scope or service unit. Example: "systemd-run
2732 --property=BlockIOWeight=10 updatedb" may be used to run
2733 updatedb at a low block IO scheduling weight.
2735 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
2736 now also work in --scope mode.
2738 * When systemd is compiled with kdbus support, basic support
2739 for enforced policies is now in place. (Note that enabling
2740 kdbus still voids your warranty and no API compatibility
2743 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
2744 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
2745 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
2746 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
2747 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
2748 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
2749 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
2750 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
2751 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
2752 Zbigniew Jędrzejewski-Szmek
2754 -- Berlin, 2014-03-12
2758 * systemd will now relabel /dev after loading the SMACK policy
2759 according to SMACK rules.
2761 * A new unit file option AppArmorProfile= has been added to
2762 set the AppArmor profile for the processes of a unit.
2764 * A new condition check ConditionArchitecture= has been added
2765 to conditionalize units based on the system architecture, as
2766 reported by uname()'s "machine" field.
2768 * systemd-networkd now supports matching on the system
2769 virtualization, architecture, kernel command line, host name
2772 * logind is now a lot more aggressive when suspending the
2773 machine due to a closed laptop lid. Instead of acting only
2774 on the lid close action, it will continuously watch the lid
2775 status and act on it. This is useful for laptops where the
2776 power button is on the outside of the chassis so that it can
2777 be reached without opening the lid (such as the Lenovo
2778 Yoga). On those machines, logind will now immediately
2779 re-suspend the machine if the power button has been
2780 accidentally pressed while the laptop was suspended and in a
2781 backpack or similar.
2783 * logind will now watch SW_DOCK switches and inhibit reaction
2784 to the lid switch if it is pressed. This means that logind
2785 will not suspend the machine anymore if the lid is closed
2786 and the system is docked, if the laptop supports SW_DOCK
2787 notifications via the input layer. Note that ACPI docking
2788 stations do not generate this currently. Also note that this
2789 logic is usually not fully sufficient and Desktop
2790 Environments should take a lid switch inhibitor lock when an
2791 external display is connected, as systemd will not watch
2794 * nspawn will now make use of the devices cgroup controller by
2795 default, and only permit creation of and access to the usual
2796 API device nodes like /dev/null or /dev/random, as well as
2797 access to (but not creation of) the pty devices.
2799 * We will now ship a default .network file for
2800 systemd-networkd that automatically configures DHCP for
2801 network interfaces created by nspawn's --network-veth or
2802 --network-bridge= switches.
2804 * systemd will now understand the usual M, K, G, T suffixes
2805 according to SI conventions (i.e. to the base 1000) when
2806 referring to throughput and hardware metrics. It will stay
2807 with IEC conventions (i.e. to the base 1024) for software
2808 metrics, according to what is customary according to
2809 Wikipedia. We explicitly document which base applies for
2810 each configuration option.
2812 * The DeviceAllow= setting in unit files now supports a syntax
2813 to whitelist an entire group of devices node majors at once,
2814 based on the /proc/devices listing. For example, with the
2815 string "char-pts", it is now possible to whitelist all
2816 current and future pseudo-TTYs at once.
2818 * sd-event learned a new "post" event source. Event sources of
2819 this type are triggered by the dispatching of any event
2820 source of a type that is not "post". This is useful for
2821 implementing clean-up and check event sources that are
2822 triggered by other work being done in the program.
2824 * systemd-networkd is no longer statically enabled, but uses
2825 the usual [Install] sections so that it can be
2826 enabled/disabled using systemctl. It still is enabled by
2829 * When creating a veth interface pair with systemd-nspawn, the
2830 host side will now be prefixed with "vb-" if
2831 --network-bridge= is used, and with "ve-" if --network-veth
2832 is used. This way, it is easy to distinguish these cases on
2833 the host, for example to apply different configuration to
2834 them with systemd-networkd.
2836 * The compatibility libraries for libsystemd-journal.so,
2837 libsystem-id128.so, libsystemd-login.so and
2838 libsystemd-daemon.so do not make use of IFUNC
2839 anymore. Instead, we now build libsystemd.so multiple times
2840 under these alternative names. This means that the footprint
2841 is drastically increased, but given that these are
2842 transitional compatibility libraries, this should not matter
2843 much. This change has been made necessary to support the ARM
2844 platform for these compatibility libraries, as the ARM
2845 toolchain is not really at the same level as the toolchain
2846 for other architectures like x86 and does not support
2847 IFUNC. Please make sure to use --enable-compat-libs only
2848 during a transitional period!
2850 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
2851 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
2852 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
2853 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
2854 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
2855 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
2856 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
2857 Zbigniew Jędrzejewski-Szmek
2859 -- Berlin, 2014-02-24
2863 * A new component "systemd-networkd" has been added that can
2864 be used to configure local network interfaces statically or
2865 via DHCP. It is capable of bringing up bridges, VLANs, and
2866 bonding. Currently, no hook-ups for interactive network
2867 configuration are provided. Use this for your initrd,
2868 container, embedded, or server setup if you need a simple,
2869 yet powerful, network configuration solution. This
2870 configuration subsystem is quite nifty, as it allows wildcard
2871 hotplug matching in interfaces. For example, with a single
2872 configuration snippet, you can configure that all Ethernet
2873 interfaces showing up are automatically added to a bridge,
2874 or similar. It supports link-sensing and more.
2876 * A new tool "systemd-socket-proxyd" has been added which can
2877 act as a bidirectional proxy for TCP sockets. This is
2878 useful for adding socket activation support to services that
2879 do not actually support socket activation, including virtual
2880 machines and the like.
2882 * Add a new tool to save/restore rfkill state on
2885 * Save/restore state of keyboard backlights in addition to
2886 display backlights on shutdown/boot.
2888 * udev learned a new SECLABEL{} construct to label device
2889 nodes with a specific security label when they appear. For
2890 now, only SECLABEL{selinux} is supported, but the syntax is
2891 prepared for additional security frameworks.
2893 * udev gained a new scheme to configure link-level attributes
2894 from files in /etc/systemd/network/*.link. These files can
2895 match against MAC address, device path, driver name and type,
2896 and will apply attributes like the naming policy, link speed,
2897 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
2898 address assignment policy (randomized, ...).
2900 * The configuration of network interface naming rules for
2901 "permanent interface names" has changed: a new NamePolicy=
2902 setting in the [Link] section of .link files determines the
2903 priority of possible naming schemes (onboard, slot, MAC,
2904 path). The default value of this setting is determined by
2905 /usr/lib/net/links/99-default.link. Old
2906 80-net-name-slot.rules udev configuration file has been
2907 removed, so local configuration overriding this file should
2908 be adapated to override 99-default.link instead.
2910 * When the User= switch is used in a unit file, also
2911 initialize $SHELL= based on the user database entry.
2913 * systemd no longer depends on libdbus. All communication is
2914 now done with sd-bus, systemd's low-level bus library
2917 * kdbus support has been added to PID 1 itself. When kdbus is
2918 enabled, this causes PID 1 to set up the system bus and
2919 enable support for a new ".busname" unit type that
2920 encapsulates bus name activation on kdbus. It works a little
2921 bit like ".socket" units, except for bus names. A new
2922 generator has been added that converts classic dbus1 service
2923 activation files automatically into native systemd .busname
2926 * sd-bus: add a light-weight vtable implementation that allows
2927 defining objects on the bus with a simple static const
2928 vtable array of its methods, signals and properties.
2930 * systemd will not generate or install static dbus
2931 introspection data anymore to /usr/share/dbus-1/interfaces,
2932 as the precise format of these files is unclear, and
2933 nothing makes use of it.
2935 * A proxy daemon is now provided to proxy clients connecting
2936 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
2937 compatibility with classic D-Bus.
2939 * A bus driver implementation has been added that supports the
2940 classic D-Bus bus driver calls on kdbus, also for
2941 compatibility purposes.
2943 * A new API "sd-event.h" has been added that implements a
2944 minimal event loop API built around epoll. It provides a
2945 couple of features that direct epoll usage is lacking:
2946 prioritization of events, scales to large numbers of timer
2947 events, per-event timer slack (accuracy), system-wide
2948 coalescing of timer events, exit handlers, watchdog
2949 supervision support using systemd's sd_notify() API, child
2952 * A new API "sd-rntl.h" has been added that provides an API
2953 around the route netlink interface of the kernel, similar in
2954 style to "sd-bus.h".
2956 * A new API "sd-dhcp-client.h" has been added that provides a
2957 small DHCPv4 client-side implementation. This is used by
2960 * There is a new kernel command line option
2961 "systemd.restore_state=0|1". When set to "0", none of the
2962 systemd tools will restore saved runtime state to hardware
2963 devices. More specifically, the rfkill and backlight states
2966 * The FsckPassNo= compatibility option in mount/service units
2967 has been removed. The fstab generator will now add the
2968 necessary dependencies automatically, and does not require
2969 PID1's support for that anymore.
2971 * journalctl gained a new switch, --list-boots, that lists
2972 recent boots with their times and boot IDs.
2974 * The various tools like systemctl, loginctl, timedatectl,
2975 busctl, systemd-run, ... have gained a new switch "-M" to
2976 connect to a specific, local OS container (as direct
2977 connection, without requiring SSH). This works on any
2978 container that is registered with machined, such as those
2979 created by libvirt-lxc or nspawn.
2981 * systemd-run and systemd-analyze also gained support for "-H"
2982 to connect to remote hosts via SSH. This is particularly
2983 useful for systemd-run because it enables queuing of jobs
2984 onto remote systems.
2986 * machinectl gained a new command "login" to open a getty
2987 login in any local container. This works with any container
2988 that is registered with machined (such as those created by
2989 libvirt-lxc or nspawn), and which runs systemd inside.
2991 * machinectl gained a new "reboot" command that may be used to
2992 trigger a reboot on a specific container that is registered
2993 with machined. This works on any container that runs an init
2994 system of some kind.
2996 * systemctl gained a new "list-timers" command to print a nice
2997 listing of installed timer units with the times they elapse
3000 * Alternative reboot() parameters may now be specified on the
3001 "systemctl reboot" command line and are passed to the
3002 reboot() system call.
3004 * systemctl gained a new --job-mode= switch to configure the
3005 mode to queue a job with. This is a more generic version of
3006 --fail, --irreversible, and --ignore-dependencies, which are
3007 still available but not advertised anymore.
3009 * /etc/systemd/system.conf gained new settings to configure
3010 various default timeouts of units, as well as the default
3011 start limit interval and burst. These may still be overridden
3014 * PID1 will now export on the bus profile data of the security
3015 policy upload process (such as the SELinux policy upload to
3018 * journald: when forwarding logs to the console, include
3019 timestamps (following the setting in
3020 /sys/module/printk/parameters/time).
3022 * OnCalendar= in timer units now understands the special
3023 strings "yearly" and "annually". (Both are equivalent)
3025 * The accuracy of timer units is now configurable with the new
3026 AccuracySec= setting. It defaults to 1min.
3028 * A new dependency type JoinsNamespaceOf= has been added that
3029 allows running two services within the same /tmp and network
3030 namespace, if PrivateNetwork= or PrivateTmp= are used.
3032 * A new command "cat" has been added to systemctl. It outputs
3033 the original unit file of a unit, and concatenates the
3034 contents of additional "drop-in" unit file snippets, so that
3035 the full configuration is shown.
3037 * systemctl now supports globbing on the various "list-xyz"
3038 commands, like "list-units" or "list-sockets", as well as on
3039 those commands which take multiple unit names.
3041 * journalctl's --unit= switch gained support for globbing.
3043 * All systemd daemons now make use of the watchdog logic so
3044 that systemd automatically notices when they hang.
3046 * If the $container_ttys environment variable is set,
3047 getty-generator will automatically spawn a getty for each
3048 listed tty. This is useful for container managers to request
3049 login gettys to be spawned on as many ttys as needed.
3051 * %h, %s, %U specifier support is not available anymore when
3052 used in unit files for PID 1. This is because NSS calls are
3053 not safe from PID 1. They stay available for --user
3054 instances of systemd, and as special case for the root user.
3056 * loginctl gained a new "--no-legend" switch to turn off output
3059 * The "sd-login.h" API gained three new calls:
3060 sd_session_is_remote(), sd_session_get_remote_user(),
3061 sd_session_get_remote_host() to query information about
3064 * The udev hardware database now also carries vendor/product
3065 information of SDIO devices.
3067 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
3068 determine whether watchdog notifications are requested by
3071 * Socket-activated per-connection services now include a
3072 short description of the connection parameters in the
3075 * tmpfiles gained a new "--boot" option. When this is not used,
3076 only lines where the command character is not suffixed with
3077 "!" are executed. When this option is specified, those
3078 options are executed too. This partitions tmpfiles
3079 directives into those that can be safely executed at any
3080 time, and those which should be run only at boot (for
3081 example, a line that creates /run/nologin).
3083 * A new API "sd-resolve.h" has been added which provides a simple
3084 asynchronous wrapper around glibc NSS host name resolution
3085 calls, such as getaddrinfo(). In contrast to glibc's
3086 getaddrinfo_a(), it does not use signals. In contrast to most
3087 other asynchronous name resolution libraries, this one does
3088 not reimplement DNS, but reuses NSS, so that alternate
3089 host name resolution systems continue to work, such as mDNS,
3090 LDAP, etc. This API is based on libasyncns, but it has been
3091 cleaned up for inclusion in systemd.
3093 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
3094 "sd-daemon.h" are no longer found in individual libraries
3095 libsystemd-journal.so, libsystemd-login.so,
3096 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
3097 merged them into a single library, libsystemd.so, which
3098 provides all symbols. The reason for this is cyclic
3099 dependencies, as these libraries tend to use each other's
3100 symbols. So far, we have managed to workaround that by linking
3101 a copy of a good part of our code into each of these
3102 libraries again and again, which, however, makes certain
3103 things hard to do, like sharing static variables. Also, it
3104 substantially increases footprint. With this change, there
3105 is only one library for the basic APIs systemd
3106 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
3107 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
3108 library as well, however are subject to the --enable-kdbus
3109 switch (see below). Note that "sd-dhcp-client.h" is not part
3110 of this library (this is because it only consumes, never
3111 provides, services of/to other APIs). To make the transition
3112 easy from the separate libraries to the unified one, we
3113 provide the --enable-compat-libs compile-time switch which
3114 will generate stub libraries that are compatible with the
3115 old ones but redirect all calls to the new one.
3117 * All of the kdbus logic and the new APIs "sd-bus.h",
3118 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
3119 and "sd-utf8.h" are compile-time optional via the
3120 "--enable-kdbus" switch, and they are not compiled in by
3121 default. To make use of kdbus, you have to explicitly enable
3122 the switch. Note however, that neither the kernel nor the
3123 userspace API for all of this is considered stable yet. We
3124 want to maintain the freedom to still change the APIs for
3125 now. By specifying this build-time switch, you acknowledge
3126 that you are aware of the instability of the current
3129 * Also, note that while kdbus is pretty much complete,
3130 it lacks one thing: proper policy support. This means you
3131 can build a fully working system with all features; however,
3132 it will be highly insecure. Policy support will be added in
3133 one of the next releases, at the same time that we will
3134 declare the APIs stable.
3136 * When the kernel command line argument "kdbus" is specified,
3137 systemd will automatically load the kdbus.ko kernel module. At
3138 this stage of development, it is only useful for testing kdbus
3139 and should not be used in production. Note: if "--enable-kdbus"
3140 is specified, and the kdbus.ko kernel module is available, and
3141 "kdbus" is added to the kernel command line, the entire system
3142 runs with kdbus instead of dbus-daemon, with the above mentioned
3143 problem of missing the system policy enforcement. Also a future
3144 version of kdbus.ko or a newer systemd will not be compatible with
3145 each other, and will unlikely be able to boot the machine if only
3146 one of them is updated.
3148 * systemctl gained a new "import-environment" command which
3149 uploads the caller's environment (or parts thereof) into the
3150 service manager so that it is inherited by services started
3151 by the manager. This is useful to upload variables like
3152 $DISPLAY into the user service manager.
3154 * A new PrivateDevices= switch has been added to service units
3155 which allows running a service with a namespaced /dev
3156 directory that does not contain any device nodes for
3157 physical devices. More specifically, it only includes devices
3158 such as /dev/null, /dev/urandom, and /dev/zero which are API
3161 * logind has been extended to support behaviour like VT
3162 switching on seats that do not support a VT. This makes
3163 multi-session available on seats that are not the first seat
3164 (seat0), and on systems where kernel support for VTs has
3165 been disabled at compile-time.
3167 * If a process holds a delay lock for system sleep or shutdown
3168 and fails to release it in time, we will now log its
3169 identity. This makes it easier to identify processes that
3170 cause slow suspends or power-offs.
3172 * When parsing /etc/crypttab, support for a new key-slot=
3173 option as supported by Debian is added. It allows indicating
3174 which LUKS slot to use on disk, speeding up key loading.
3176 * The sd_journald_sendv() API call has been checked and
3177 officially declared to be async-signal-safe so that it may
3178 be invoked from signal handlers for logging purposes.
3180 * Boot-time status output is now enabled automatically after a
3181 short timeout if boot does not progress, in order to give
3182 the user an indication what she or he is waiting for.
3184 * The boot-time output has been improved to show how much time
3185 remains until jobs expire.
3187 * The KillMode= switch in service units gained a new possible
3188 value "mixed". If set, and the unit is shut down, then the
3189 initial SIGTERM signal is sent only to the main daemon
3190 process, while the following SIGKILL signal is sent to
3191 all remaining processes of the service.
3193 * When a scope unit is registered, a new property "Controller"
3194 may be set. If set to a valid bus name, systemd will send a
3195 RequestStop() signal to this name when it would like to shut
3196 down the scope. This may be used to hook manager logic into
3197 the shutdown logic of scope units. Also, scope units may now
3198 be put in a special "abandoned" state, in which case the
3199 manager process which created them takes no further
3200 responsibilities for it.
3202 * When reading unit files, systemd will now verify
3203 the access mode of these files, and warn about certain
3204 suspicious combinations. This has been added to make it
3205 easier to track down packaging bugs where unit files are
3206 marked executable or world-writable.
3208 * systemd-nspawn gained a new "--setenv=" switch to set
3209 container-wide environment variables. The similar option in
3210 systemd-activate was renamed from "--environment=" to
3211 "--setenv=" for consistency.
3213 * systemd-nspawn has been updated to create a new kdbus domain
3214 for each container that is invoked, thus allowing each
3215 container to have its own set of system and user buses,
3216 independent of the host.
3218 * systemd-nspawn gained a new --drop-capability= switch to run
3219 the container with less capabilities than the default. Both
3220 --drop-capability= and --capability= now take the special
3221 string "all" for dropping or keeping all capabilities.
3223 * systemd-nspawn gained new switches for executing containers
3224 with specific SELinux labels set.
3226 * systemd-nspawn gained a new --quiet switch to not generate
3227 any additional output but the container's own console
3230 * systemd-nspawn gained a new --share-system switch to run a
3231 container without PID namespacing enabled.
3233 * systemd-nspawn gained a new --register= switch to control
3234 whether the container is registered with systemd-machined or
3235 not. This is useful for containers that do not run full
3236 OS images, but only specific apps.
3238 * systemd-nspawn gained a new --keep-unit which may be used
3239 when invoked as the only program from a service unit, and
3240 results in registration of the unit service itself in
3241 systemd-machined, instead of a newly opened scope unit.
3243 * systemd-nspawn gained a new --network-interface= switch for
3244 moving arbitrary interfaces to the container. The new
3245 --network-veth switch creates a virtual Ethernet connection
3246 between host and container. The new --network-bridge=
3247 switch then allows assigning the host side of this virtual
3248 Ethernet connection to a bridge device.
3250 * systemd-nspawn gained a new --personality= switch for
3251 setting the kernel personality for the container. This is
3252 useful when running a 32-bit container on a 64-bit host. A
3253 similar option Personality= is now also available for service
3256 * logind will now also track a "Desktop" identifier for each
3257 session which encodes the desktop environment of it. This is
3258 useful for desktop environments that want to identify
3259 multiple running sessions of itself easily.
3261 * A new SELinuxContext= setting for service units has been
3262 added that allows setting a specific SELinux execution
3263 context for a service.
3265 * Most systemd client tools will now honour $SYSTEMD_LESS for
3266 settings of the "less" pager. By default, these tools will
3267 override $LESS to allow certain operations to work, such as
3268 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
3269 influence this logic.
3271 * systemd's "seccomp" hook-up has been changed to make use of
3272 the libseccomp library instead of using its own
3273 implementation. This has benefits for portability among
3276 * For usage together with SystemCallFilter=, a new
3277 SystemCallErrorNumber= setting has been introduced that
3278 allows configuration of a system error number to be returned
3279 on filtered system calls, instead of immediately killing the
3280 process. Also, SystemCallArchitectures= has been added to
3281 limit access to system calls of a particular architecture
3282 (in order to turn off support for unused secondary
3283 architectures). There is also a global
3284 SystemCallArchitectures= setting in system.conf now to turn
3285 off support for non-native system calls system-wide.
3287 * systemd requires a kernel with a working name_to_handle_at(),
3288 please see the kernel config requirements in the README file.
3290 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
3291 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
3292 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
3293 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
3294 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
3295 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
3296 Elia Pinto, Florian Weimer, George McCollister, Goffredo
3297 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
3298 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
3299 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
3300 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
3301 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
3302 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
3303 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
3304 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
3305 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
3306 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
3307 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
3308 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
3309 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
3310 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
3311 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
3312 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
3313 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
3315 -- Berlin, 2014-02-20
3319 * logind has gained support for facilitating privileged input
3320 and drm device access for unprivileged clients. This work is
3321 useful to allow Wayland display servers (and similar
3322 programs, such as kmscon) to run under the user's ID and
3323 access input and drm devices which are normally
3324 protected. When this is used (and the kernel is new enough)
3325 logind will "mute" IO on the file descriptors passed to
3326 Wayland as long as it is in the background and "unmute" it
3327 if it returns into the foreground. This allows secure
3328 session switching without allowing background sessions to
3329 eavesdrop on input and display data. This also introduces
3330 session switching support if VT support is turned off in the
3331 kernel, and on seats that are not seat0.
3333 * A new kernel command line option luks.options= is understood
3334 now which allows specifying LUKS options for usage for LUKS
3335 encrypted partitions specified with luks.uuid=.
3337 * tmpfiles.d(5) snippets may now use specifier expansion in
3338 path names. More specifically %m, %b, %H, %v, are now
3339 replaced by the local machine id, boot id, hostname, and
3340 kernel version number.
3342 * A new tmpfiles.d(5) command "m" has been introduced which
3343 may be used to change the owner/group/access mode of a file
3344 or directory if it exists, but do nothing if it does not.
3346 * This release removes high-level support for the
3347 MemorySoftLimit= cgroup setting. The underlying kernel
3348 cgroup attribute memory.soft_limit= is currently badly
3349 designed and likely to be removed from the kernel API in its
3350 current form, hence we should not expose it for now.
3352 * The memory.use_hierarchy cgroup attribute is now enabled for
3353 all cgroups systemd creates in the memory cgroup
3354 hierarchy. This option is likely to be come the built-in
3355 default in the kernel anyway, and the non-hierarchical mode
3356 never made much sense in the intrinsically hierarchical
3359 * A new field _SYSTEMD_SLICE= is logged along with all journal
3360 messages containing the slice a message was generated
3361 from. This is useful to allow easy per-customer filtering of
3362 logs among other things.
3364 * systemd-journald will no longer adjust the group of journal
3365 files it creates to the "systemd-journal" group. Instead we
3366 rely on the journal directory to be owned by the
3367 "systemd-journal" group, and its setgid bit set, so that the
3368 kernel file system layer will automatically enforce that
3369 journal files inherit this group assignment. The reason for
3370 this change is that we cannot allow NSS look-ups from
3371 journald which would be necessary to resolve
3372 "systemd-journal" to a numeric GID, because this might
3373 create deadlocks if NSS involves synchronous queries to
3374 other daemons (such as nscd, or sssd) which in turn are
3375 logging clients of journald and might block on it, which
3376 would then dead lock. A tmpfiles.d(5) snippet included in
3377 systemd will make sure the setgid bit and group are
3378 properly set on the journal directory if it exists on every
3379 boot. However, we recommend adjusting it manually after
3380 upgrades too (or from RPM scriptlets), so that the change is
3381 not delayed until next reboot.
3383 * Backlight and random seed files in /var/lib/ have moved into
3384 the /var/lib/systemd/ directory, in order to centralize all
3385 systemd generated files in one directory.
3387 * Boot time performance measurements (as displayed by
3388 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
3389 performance information if that's available to determine how
3390 much time BIOS and boot loader initialization required. With
3391 a sufficiently new BIOS you hence no longer need to boot
3392 with Gummiboot to get access to such information.
3394 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
3395 Cristian Rodríguez, Dave Reisner, David Herrmann, David
3396 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
3397 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
3398 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
3399 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
3400 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3402 -- Berlin, 2013-10-02
3406 * The Restart= option for services now understands a new
3407 on-watchdog setting, which will restart the service
3408 automatically if the service stops sending out watchdog keep
3409 alive messages (as configured with WatchdogSec=).
3411 * The getty generator (which is responsible for bringing up a
3412 getty on configured serial consoles) will no longer only
3413 start a getty on the primary kernel console but on all
3414 others, too. This makes the order in which console= is
3415 specified on the kernel command line less important.
3417 * libsystemd-logind gained a new sd_session_get_vt() call to
3418 retrieve the VT number of a session.
3420 * If the option "tries=0" is set for an entry of /etc/crypttab
3421 its passphrase is queried indefinitely instead of any
3422 maximum number of tries.
3424 * If a service with a configure PID file terminates its PID
3425 file will now be removed automatically if it still exists
3426 afterwards. This should put an end to stale PID files.
3428 * systemd-run will now also take relative binary path names
3429 for execution and no longer insists on absolute paths.
3431 * InaccessibleDirectories= and ReadOnlyDirectories= now take
3432 paths that are optionally prefixed with "-" to indicate that
3433 it should not be considered a failure if they do not exist.
3435 * journalctl -o (and similar commands) now understands a new
3436 output mode "short-precise", it is similar to "short" but
3437 shows timestamps with usec accuracy.
3439 * The option "discard" (as known from Debian) is now
3440 synonymous to "allow-discards" in /etc/crypttab. In fact,
3441 "discard" is preferred now (since it is easier to remember
3444 * Some licensing clean-ups were made, so that more code is now
3445 LGPL-2.1 licensed than before.
3447 * A minimal tool to save/restore the display backlight
3448 brightness across reboots has been added. It will store the
3449 backlight setting as late as possible at shutdown, and
3450 restore it as early as possible during reboot.
3452 * A logic to automatically discover and enable home and swap
3453 partitions on GPT disks has been added. With this in place
3454 /etc/fstab becomes optional for many setups as systemd can
3455 discover certain partitions located on the root disk
3456 automatically. Home partitions are recognized under their
3457 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
3458 partitions are recognized under their GPT type ID
3459 0657fd6da4ab43c484e50933c84b4f4f.
3461 * systemd will no longer pass any environment from the kernel
3462 or initrd to system services. If you want to set an
3463 environment for all services, do so via the kernel command
3464 line systemd.setenv= assignment.
3466 * The systemd-sysctl tool no longer natively reads the file
3467 /etc/sysctl.conf. If desired, the file should be symlinked
3468 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
3469 legacy support by a symlink rather than built-in code, it
3470 also makes the otherwise hidden order of application of the
3471 different files visible. (Note that this partly reverts to a
3472 pre-198 application order of sysctl knobs!)
3474 * The "systemctl set-log-level" and "systemctl dump" commands
3475 have been moved to systemd-analyze.
3477 * systemd-run learned the new --remain-after-exit switch,
3478 which causes the scope unit not to be cleaned up
3479 automatically after the process terminated.
3481 * tmpfiles learned a new --exclude-prefix= switch to exclude
3482 certain paths from operation.
3484 * journald will now automatically flush all messages to disk
3485 as soon as a message at the log level CRIT, ALERT or EMERG
3488 Contributions from: Andrew Cook, Brandon Philips, Christian
3489 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
3490 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
3491 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
3492 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
3493 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
3494 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
3495 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
3496 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
3497 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
3498 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
3499 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
3500 William Giokas, Zbigniew Jędrzejewski-Szmek
3502 -- Berlin, 2013-09-13
3506 * The documentation has been updated to cover the various new
3507 concepts introduced with 205.
3509 * Unit files now understand the new %v specifier which
3510 resolves to the kernel version string as returned by "uname
3513 * systemctl now supports filtering the unit list output by
3514 load state, active state and sub state, using the new
3517 * "systemctl status" will now show the results of the
3518 condition checks (like ConditionPathExists= and similar) of
3519 the last start attempts of the unit. They are also logged to
3522 * "journalctl -b" may now be used to look for boot output of a
3523 specific boot. Try "journalctl -b -1" for the previous boot,
3524 but the syntax is substantially more powerful.
3526 * "journalctl --show-cursor" has been added which prints the
3527 cursor string the last shown log line. This may then be used
3528 with the new "journalctl --after-cursor=" switch to continue
3529 browsing logs from that point on.
3531 * "journalctl --force" may now be used to force regeneration
3534 * Creation of "dead" device nodes has been moved from udev
3535 into kmod and tmpfiles. Previously, udev would read the kmod
3536 databases to pre-generate dead device nodes based on meta
3537 information contained in kernel modules, so that these would
3538 be auto-loaded on access rather then at boot. As this
3539 does not really have much to do with the exposing actual
3540 kernel devices to userspace this has always been slightly
3541 alien in the udev codebase. Following the new scheme kmod
3542 will now generate a runtime snippet for tmpfiles from the
3543 module meta information and it now is tmpfiles' job to the
3544 create the nodes. This also allows overriding access and
3545 other parameters for the nodes using the usual tmpfiles
3546 facilities. As side effect this allows us to remove the
3547 CAP_SYS_MKNOD capability bit from udevd entirely.
3549 * logind's device ACLs may now be applied to these "dead"
3550 devices nodes too, thus finally allowing managed access to
3551 devices such as /dev/snd/sequencer whithout loading the
3552 backing module right-away.
3554 * A new RPM macro has been added that may be used to apply
3555 tmpfiles configuration during package installation.
3557 * systemd-detect-virt and ConditionVirtualization= now can
3558 detect User-Mode-Linux machines (UML).
3560 * journald will now implicitly log the effective capabilities
3561 set of processes in the message metadata.
3563 * systemd-cryptsetup has gained support for TrueCrypt volumes.
3565 * The initrd interface has been simplified (more specifically,
3566 support for passing performance data via environment
3567 variables and fsck results via files in /run has been
3568 removed). These features were non-essential, and are
3569 nowadays available in a much nicer way by having systemd in
3570 the initrd serialize its state and have the hosts systemd
3571 deserialize it again.
3573 * The udev "keymap" data files and tools to apply keyboard
3574 specific mappings of scan to key codes, and force-release
3575 scan code lists have been entirely replaced by a udev
3576 "keyboard" builtin and a hwdb data file.
3578 * systemd will now honour the kernel's "quiet" command line
3579 argument also during late shutdown, resulting in a
3580 completely silent shutdown when used.
3582 * There's now an option to control the SO_REUSEPORT socket
3583 option in .socket units.
3585 * Instance units will now automatically get a per-template
3586 subslice of system.slice unless something else is explicitly
3587 configured. For example, instances of sshd@.service will now
3588 implicitly be placed in system-sshd.slice rather than
3589 system.slice as before.
3591 * Test coverage support may now be enabled at build time.
3593 Contributions from: Dave Reisner, Frederic Crozat, Harald
3594 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
3595 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
3596 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
3597 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
3598 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
3599 Giokas, Zbigniew Jędrzejewski-Szmek
3601 -- Berlin, 2013-07-23
3605 * Two new unit types have been introduced:
3607 Scope units are very similar to service units, however, are
3608 created out of pre-existing processes -- instead of PID 1
3609 forking off the processes. By using scope units it is
3610 possible for system services and applications to group their
3611 own child processes (worker processes) in a powerful way
3612 which then maybe used to organize them, or kill them
3613 together, or apply resource limits on them.
3615 Slice units may be used to partition system resources in an
3616 hierarchical fashion and then assign other units to them. By
3617 default there are now three slices: system.slice (for all
3618 system services), user.slice (for all user sessions),
3619 machine.slice (for VMs and containers).
3621 Slices and scopes have been introduced primarily in
3622 context of the work to move cgroup handling to a
3623 single-writer scheme, where only PID 1
3624 creates/removes/manages cgroups.
3626 * There's a new concept of "transient" units. In contrast to
3627 normal units these units are created via an API at runtime,
3628 not from configuration from disk. More specifically this
3629 means it is now possible to run arbitrary programs as
3630 independent services, with all execution parameters passed
3631 in via bus APIs rather than read from disk. Transient units
3632 make systemd substantially more dynamic then it ever was,
3633 and useful as a general batch manager.
3635 * logind has been updated to make use of scope and slice units
3636 for managing user sessions. As a user logs in he will get
3637 his own private slice unit, to which all sessions are added
3638 as scope units. We also added support for automatically
3639 adding an instance of user@.service for the user into the
3640 slice. Effectively logind will no longer create cgroup
3641 hierarchies on its own now, it will defer entirely to PID 1
3642 for this by means of scope, service and slice units. Since
3643 user sessions this way become entities managed by PID 1
3644 the output of "systemctl" is now a lot more comprehensive.
3646 * A new mini-daemon "systemd-machined" has been added which
3647 may be used by virtualization managers to register local
3648 VMs/containers. nspawn has been updated accordingly, and
3649 libvirt will be updated shortly. machined will collect a bit
3650 of meta information about the VMs/containers, and assign
3651 them their own scope unit (see above). The collected
3652 meta-data is then made available via the "machinectl" tool,
3653 and exposed in "ps" and similar tools. machined/machinectl
3654 is compile-time optional.
3656 * As discussed earlier, the low-level cgroup configuration
3657 options ControlGroup=, ControlGroupModify=,
3658 ControlGroupPersistent=, ControlGroupAttribute= have been
3659 removed. Please use high-level attribute settings instead as
3660 well as slice units.
3662 * A new bus call SetUnitProperties() has been added to alter
3663 various runtime parameters of a unit. This is primarily
3664 useful to alter cgroup parameters dynamically in a nice way,
3665 but will be extended later on to make more properties
3666 modifiable at runtime. systemctl gained a new set-properties
3667 command that wraps this call.
3669 * A new tool "systemd-run" has been added which can be used to
3670 run arbitrary command lines as transient services or scopes,
3671 while configuring a number of settings via the command
3672 line. This tool is currently very basic, however already
3673 very useful. We plan to extend this tool to even allow
3674 queuing of execution jobs with time triggers from the
3675 command line, similar in fashion to "at".
3677 * nspawn will now inform the user explicitly that kernels with
3678 audit enabled break containers, and suggest the user to turn
3681 * Support for detecting the IMA and AppArmor security
3682 frameworks with ConditionSecurity= has been added.
3684 * journalctl gained a new "-k" switch for showing only kernel
3685 messages, mimicking dmesg output; in addition to "--user"
3686 and "--system" switches for showing only user's own logs
3689 * systemd-delta can now show information about drop-in
3690 snippets extending unit files.
3692 * libsystemd-bus has been substantially updated but is still
3693 not available as public API.
3695 * systemd will now look for the "debug" argument on the kernel
3696 command line and enable debug logging, similar to what
3697 "systemd.log_level=debug" already did before.
3699 * "systemctl set-default", "systemctl get-default" has been
3700 added to configure the default.target symlink, which
3701 controls what to boot into by default.
3703 * "systemctl set-log-level" has been added as a convenient
3704 way to raise and lower systemd logging threshold.
3706 * "systemd-analyze plot" will now show the time the various
3707 generators needed for execution, as well as information
3708 about the unit file loading.
3710 * libsystemd-journal gained a new sd_journal_open_files() call
3711 for opening specific journal files. journactl also gained a
3712 new switch to expose this new functionality. Previously we
3713 only supported opening all files from a directory, or all
3714 files from the system, as opening individual files only is
3715 racy due to journal file rotation.
3717 * systemd gained the new DefaultEnvironment= setting in
3718 /etc/systemd/system.conf to set environment variables for
3721 * If a privileged process logs a journal message with the
3722 OBJECT_PID= field set, then journald will automatically
3723 augment this with additional OBJECT_UID=, OBJECT_GID=,
3724 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
3725 system services want to log events about specific client
3726 processes. journactl/systemctl has been updated to make use
3727 of this information if all log messages regarding a specific
3730 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
3731 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
3732 Reisner, David Coppa, David King, David Strauss, Eelco
3733 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
3734 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
3735 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
3736 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
3737 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
3738 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
3739 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
3740 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
3741 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
3742 Łukasz Stelmach, 장동준
3746 * The Python bindings gained some minimal support for the APIs
3747 exposed by libsystemd-logind.
3749 * ConditionSecurity= gained support for detecting SMACK. Since
3750 this condition already supports SELinux and AppArmor we only
3751 miss IMA for this. Patches welcome!
3753 Contributions from: Karol Lewandowski, Lennart Poettering,
3754 Zbigniew Jędrzejewski-Szmek
3758 * systemd-nspawn will now create /etc/resolv.conf if
3759 necessary, before bind-mounting the host's file onto it.
3761 * systemd-nspawn will now store meta information about a
3762 container on the container's cgroup as extended attribute
3763 fields, including the root directory.
3765 * The cgroup hierarchy has been reworked in many ways. All
3766 objects any of the components systemd creates in the cgroup
3767 tree are now suffixed. More specifically, user sessions are
3768 now placed in cgroups suffixed with ".session", users in
3769 cgroups suffixed with ".user", and nspawn containers in
3770 cgroups suffixed with ".nspawn". Furthermore, all cgroup
3771 names are now escaped in a simple scheme to avoid collision
3772 of userspace object names with kernel filenames. This work
3773 is preparation for making these objects relocatable in the
3774 cgroup tree, in order to allow easy resource partitioning of
3775 these objects without causing naming conflicts.
3777 * systemctl list-dependencies gained the new switches
3778 --plain, --reverse, --after and --before.
3780 * systemd-inhibit now shows the process name of processes that
3781 have taken an inhibitor lock.
3783 * nss-myhostname will now also resolve "localhost"
3784 implicitly. This makes /etc/hosts an optional file and
3785 nicely handles that on IPv6 ::1 maps to both "localhost" and
3788 * libsystemd-logind.so gained a new call
3789 sd_get_machine_names() to enumerate running containers and
3790 VMs (currently only supported by very new libvirt and
3791 nspawn). sd_login_monitor can now be used to watch
3792 VMs/containers coming and going.
3794 * .include is not allowed recursively anymore, and only in
3795 unit files. Usually it is better to use drop-in snippets in
3796 .d/*.conf anyway, as introduced with systemd 198.
3798 * systemd-analyze gained a new "critical-chain" command that
3799 determines the slowest chain of units run during system
3800 boot-up. It is very useful for tracking down where
3801 optimizing boot time is the most beneficial.
3803 * systemd will no longer allow manipulating service paths in
3804 the name=systemd:/system cgroup tree using ControlGroup= in
3805 units. (But is still fine with it in all other dirs.)
3807 * There's a new systemd-nspawn@.service service file that may
3808 be used to easily run nspawn containers as system
3809 services. With the container's root directory in
3810 /var/lib/container/foobar it is now sufficient to run
3811 "systemctl start systemd-nspawn@foobar.service" to boot it.
3813 * systemd-cgls gained a new parameter "--machine" to list only
3814 the processes within a certain container.
3816 * ConditionSecurity= now can check for "apparmor". We still
3817 are lacking checks for SMACK and IMA for this condition
3818 check though. Patches welcome!
3820 * A new configuration file /etc/systemd/sleep.conf has been
3821 added that may be used to configure which kernel operation
3822 systemd is supposed to execute when "suspend", "hibernate"
3823 or "hybrid-sleep" is requested. This makes the new kernel
3824 "freeze" state accessible to the user.
3826 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
3827 the passed argument if applicable.
3829 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
3830 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
3831 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
3832 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
3833 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
3834 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
3835 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
3840 * The output of 'systemctl list-jobs' got some polishing. The
3841 '--type=' argument may now be passed more than once. A new
3842 command 'systemctl list-sockets' has been added which shows
3843 a list of kernel sockets systemd is listening on with the
3844 socket units they belong to, plus the units these socket
3847 * The experimental libsystemd-bus library got substantial
3848 updates to work in conjunction with the (also experimental)
3849 kdbus kernel project. It works well enough to exchange
3850 messages with some sophistication. Note that kdbus is not
3851 ready yet, and the library is mostly an elaborate test case
3852 for now, and not installable.
3854 * systemd gained a new unit 'systemd-static-nodes.service'
3855 that generates static device nodes earlier during boot, and
3856 can run in conjunction with udev.
3858 * libsystemd-login gained a new call sd_pid_get_user_unit()
3859 to retrieve the user systemd unit a process is running
3860 in. This is useful for systems where systemd is used as
3863 * systemd-nspawn now places all containers in the new /machine
3864 top-level cgroup directory in the name=systemd
3865 hierarchy. libvirt will soon do the same, so that we get a
3866 uniform separation of /system, /user and /machine for system
3867 services, user processes and containers/virtual
3868 machines. This new cgroup hierarchy is also useful to stick
3869 stable names to specific container instances, which can be
3870 recognized later this way (this name may be controlled
3871 via systemd-nspawn's new -M switch). libsystemd-login also
3872 gained a new call sd_pid_get_machine_name() to retrieve the
3873 name of the container/VM a specific process belongs to.
3875 * bootchart can now store its data in the journal.
3877 * libsystemd-journal gained a new call
3878 sd_journal_add_conjunction() for AND expressions to the
3879 matching logic. This can be used to express more complex
3880 logical expressions.
3882 * journactl can now take multiple --unit= and --user-unit=
3885 * The cryptsetup logic now understands the "luks.key=" kernel
3886 command line switch for specifying a file to read the
3887 decryption key from. Also, if a configured key file is not
3888 found the tool will now automatically fall back to prompting
3891 * Python systemd.journal module was updated to wrap recently
3892 added functions from libsystemd-journal. The interface was
3893 changed to bring the low level interface in s.j._Reader
3894 closer to the C API, and the high level interface in
3895 s.j.Reader was updated to wrap and convert all data about
3898 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
3899 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
3900 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
3901 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
3902 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
3903 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3907 * journalctl --update-catalog now understands a new --root=
3908 option to operate on catalogs found in a different root
3911 * During shutdown after systemd has terminated all running
3912 services a final killing loop kills all remaining left-over
3913 processes. We will now print the name of these processes
3914 when we send SIGKILL to them, since this usually indicates a
3917 * If /etc/crypttab refers to password files stored on
3918 configured mount points automatic dependencies will now be
3919 generated to ensure the specific mount is established first
3920 before the key file is attempted to be read.
3922 * 'systemctl status' will now show information about the
3923 network sockets a socket unit is listening on.
3925 * 'systemctl status' will also shown information about any
3926 drop-in configuration file for units. (Drop-In configuration
3927 files in this context are files such as
3928 /etc/systemd/systemd/foobar.service.d/*.conf)
3930 * systemd-cgtop now optionally shows summed up CPU times of
3931 cgroups. Press '%' while running cgtop to switch between
3932 percentage and absolute mode. This is useful to determine
3933 which cgroups use up the most CPU time over the entire
3934 runtime of the system. systemd-cgtop has also been updated
3935 to be 'pipeable' for processing with further shell tools.
3937 * 'hostnamectl set-hostname' will now allow setting of FQDN
3940 * The formatting and parsing of time span values has been
3941 changed. The parser now understands fractional expressions
3942 such as "5.5h". The formatter will now output fractional
3943 expressions for all time spans under 1min, i.e. "5.123456s"
3944 rather than "5s 123ms 456us". For time spans under 1s
3945 millisecond values are shown, for those under 1ms
3946 microsecond values are shown. This should greatly improve
3947 all time-related output of systemd.
3949 * libsystemd-login and libsystemd-journal gained new
3950 functions for querying the poll() events mask and poll()
3951 timeout value for integration into arbitrary event
3954 * localectl gained the ability to list available X11 keymaps
3955 (models, layouts, variants, options).
3957 * 'systemd-analyze dot' gained the ability to filter for
3958 specific units via shell-style globs, to create smaller,
3959 more useful graphs. I.e. it is now possible to create simple
3960 graphs of all the dependencies between only target units, or
3961 of all units that Avahi has dependencies with.
3963 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
3964 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
3965 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
3966 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
3967 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
3968 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
3969 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
3973 * The boot-time readahead implementation for rotating media
3974 will now read the read-ahead data in multiple passes which
3975 consist of all read requests made in equidistant time
3976 intervals. This means instead of strictly reading read-ahead
3977 data in its physical order on disk we now try to find a
3978 middle ground between physical and access time order.
3980 * /etc/os-release files gained a new BUILD_ID= field for usage
3981 on operating systems that provide continuous builds of OS
3984 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
3985 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
3986 William Douglas, Zbigniew Jędrzejewski-Szmek
3990 * systemd-python gained an API exposing libsystemd-daemon.
3992 * The SMACK setup logic gained support for uploading CIPSO
3995 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
3996 ReadOnlyDirectories= and InaccessibleDirectories= has
3997 changed. The private /tmp and /var/tmp directories are now
3998 shared by all processes of a service (which means
3999 ExecStartPre= may now leave data in /tmp that ExecStart= of
4000 the same service can still access). When a service is
4001 stopped its temporary directories are immediately deleted
4002 (normal clean-up with tmpfiles is still done in addition to
4005 * By default, systemd will now set a couple of sysctl
4006 variables in the kernel: the safe sysrq options are turned
4007 on, IP route verification is turned on, and source routing
4008 disabled. The recently added hardlink and softlink
4009 protection of the kernel is turned on. These settings should
4010 be reasonably safe, and good defaults for all new systems.
4012 * The predictable network naming logic may now be turned off
4013 with a new kernel command line switch: net.ifnames=0.
4015 * A new libsystemd-bus module has been added that implements a
4016 pretty complete D-Bus client library. For details see:
4018 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
4020 * journald will now explicitly flush the journal files to disk
4021 at the latest 5min after each write. The file will then also
4022 be marked offline until the next write. This should increase
4023 reliability in case of a crash. The synchronization delay
4024 can be configured via SyncIntervalSec= in journald.conf.
4026 * There's a new remote-fs-setup.target unit that can be used
4027 to pull in specific services when at least one remote file
4028 system is to be mounted.
4030 * There are new targets timers.target and paths.target as
4031 canonical targets to pull user timer and path units in
4032 from. This complements sockets.target with a similar
4033 purpose for socket units.
4035 * libudev gained a new call udev_device_set_attribute_value()
4036 to set sysfs attributes of a device.
4038 * The udev daemon now sets the default number of worker
4039 processes executed in parallel based on the number of available
4040 CPUs instead of the amount of available RAM. This is supposed
4041 to provide a more reliable default and limit a too aggressive
4042 paralellism for setups with 1000s of devices connected.
4044 Contributions from: Auke Kok, Colin Walters, Cristian
4045 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
4046 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
4047 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
4048 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
4049 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
4050 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
4051 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
4052 Zbigniew Jędrzejewski-Szmek
4056 * Configuration of unit files may now be extended via drop-in
4057 files without having to edit/override the unit files
4058 themselves. More specifically, if the administrator wants to
4059 change one value for a service file foobar.service he can
4060 now do so by dropping in a configuration snippet into
4061 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
4062 will load all these snippets and apply them on top of the
4063 main unit configuration file, possibly extending or
4064 overriding its settings. Using these drop-in snippets is
4065 generally nicer than the two earlier options for changing
4066 unit files locally: copying the files from
4067 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
4068 them there; or creating a new file in /etc/systemd/system/
4069 that incorporates the original one via ".include". Drop-in
4070 snippets into these .d/ directories can be placed in any
4071 directory systemd looks for units in, and the usual
4072 overriding semantics between /usr/lib, /etc and /run apply
4075 * Most unit file settings which take lists of items can now be
4076 reset by assigning the empty string to them. For example,
4077 normally, settings such as Environment=FOO=BAR append a new
4078 environment variable assignment to the environment block,
4079 each time they are used. By assigning Environment= the empty
4080 string the environment block can be reset to empty. This is
4081 particularly useful with the .d/*.conf drop-in snippets
4082 mentioned above, since this adds the ability to reset list
4083 settings from vendor unit files via these drop-ins.
4085 * systemctl gained a new "list-dependencies" command for
4086 listing the dependencies of a unit recursively.
4088 * Inhibitors are now honored and listed by "systemctl
4089 suspend", "systemctl poweroff" (and similar) too, not only
4090 GNOME. These commands will also list active sessions by
4093 * Resource limits (as exposed by the various control group
4094 controllers) can now be controlled dynamically at runtime
4095 for all units. More specifically, you can now use a command
4096 like "systemctl set-cgroup-attr foobar.service cpu.shares
4097 2000" to alter the CPU shares a specific service gets. These
4098 settings are stored persistently on disk, and thus allow the
4099 administrator to easily adjust the resource usage of
4100 services with a few simple commands. This dynamic resource
4101 management logic is also available to other programs via the
4102 bus. Almost any kernel cgroup attribute and controller is
4105 * systemd-vconsole-setup will now copy all font settings to
4106 all allocated VTs, where it previously applied them only to
4109 * libsystemd-login gained the new sd_session_get_tty() API
4112 * This release drops support for a few legacy or
4113 distribution-specific LSB facility names when parsing init
4114 scripts: $x-display-manager, $mail-transfer-agent,
4115 $mail-transport-agent, $mail-transfer-agent, $smtp,
4116 $null. Also, the mail-transfer-agent.target unit backing
4117 this has been removed. Distributions which want to retain
4118 compatibility with this should carry the burden for
4119 supporting this themselves and patch support for these back
4120 in, if they really need to. Also, the facilities $syslog and
4121 $local_fs are now ignored, since systemd does not support
4122 early-boot LSB init scripts anymore, and these facilities
4123 are implied anyway for normal services. syslog.target has
4126 * There are new bus calls on PID1's Manager object for
4127 cancelling jobs, and removing snapshot units. Previously,
4128 both calls were only available on the Job and Snapshot
4131 * systemd-journal-gatewayd gained SSL support.
4133 * The various "environment" files, such as /etc/locale.conf
4134 now support continuation lines with a backslash ("\") as
4135 last character in the line, similarly in style (but different)
4136 to how this is supported in shells.
4138 * For normal user processes the _SYSTEMD_USER_UNIT= field is
4139 now implicitly appended to every log entry logged. systemctl
4140 has been updated to filter by this field when operating on a
4141 user systemd instance.
4143 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
4144 CAP_AUDIT_CONTROL capabilities to the capabilities set for
4145 the container. This makes it easier to boot unmodified
4146 Fedora systems in a container, which however still requires
4147 audit=0 to be passed on the kernel command line. Auditing in
4148 kernel and userspace is unfortunately still too broken in
4149 context of containers, hence we recommend compiling it out
4150 of the kernel or using audit=0. Hopefully this will be fixed
4151 one day for good in the kernel.
4153 * nspawn gained the new --bind= and --bind-ro= parameters to
4154 bind mount specific directories from the host into the
4157 * nspawn will now mount its own devpts file system instance
4158 into the container, in order not to leak pty devices from
4159 the host into the container.
4161 * systemd will now read the firmware boot time performance
4162 information from the EFI variables, if the used boot loader
4163 supports this, and takes it into account for boot performance
4164 analysis via "systemd-analyze". This is currently supported
4165 only in conjunction with Gummiboot, but could be supported
4166 by other boot loaders too. For details see:
4168 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
4170 * A new generator has been added that automatically mounts the
4171 EFI System Partition (ESP) to /boot, if that directory
4172 exists, is empty, and no other file system has been
4173 configured to be mounted there.
4175 * logind will now send out PrepareForSleep(false) out
4176 unconditionally, after coming back from suspend. This may be
4177 used by applications as asynchronous notification for
4178 system resume events.
4180 * "systemctl unlock-sessions" has been added, that allows
4181 unlocking the screens of all user sessions at once, similar
4182 to how "systemctl lock-sessions" already locked all users
4183 sessions. This is backed by a new D-Bus call UnlockSessions().
4185 * "loginctl seat-status" will now show the master device of a
4186 seat. (i.e. the device of a seat that needs to be around for
4187 the seat to be considered available, usually the graphics
4190 * tmpfiles gained a new "X" line type, that allows
4191 configuration of files and directories (with wildcards) that
4192 shall be excluded from automatic cleanup ("aging").
4194 * udev default rules set the device node permissions now only
4195 at "add" events, and do not change them any longer with a
4196 later "change" event.
4198 * The log messages for lid events and power/sleep keypresses
4199 now carry a message ID.
4201 * We now have a substantially larger unit test suite, but this
4202 continues to be work in progress.
4204 * udevadm hwdb gained a new --root= parameter to change the
4205 root directory to operate relative to.
4207 * logind will now issue a background sync() request to the kernel
4208 early at shutdown, so that dirty buffers are flushed to disk early
4209 instead of at the last moment, in order to optimize shutdown
4212 * A new bootctl tool has been added that is an interface for
4213 certain boot loader operations. This is currently a preview
4214 and is likely to be extended into a small mechanism daemon
4215 like timedated, localed, hostnamed, and can be used by
4216 graphical UIs to enumerate available boot options, and
4217 request boot into firmware operations.
4219 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
4220 the rest of the package. It also has been updated to work
4221 correctly in initrds.
4223 * Policykit previously has been runtime optional, and is now
4224 also compile time optional via a configure switch.
4226 * systemd-analyze has been reimplemented in C. Also "systemctl
4227 dot" has moved into systemd-analyze.
4229 * "systemctl status" with no further parameters will now print
4230 the status of all active or failed units.
4232 * Operations such as "systemctl start" can now be executed
4233 with a new mode "--irreversible" which may be used to queue
4234 operations that cannot accidentally be reversed by a later
4235 job queuing. This is by default used to make shutdown
4236 requests more robust.
4238 * The Python API of systemd now gained a new module for
4239 reading journal files.
4241 * A new tool kernel-install has been added that can install
4242 kernel images according to the Boot Loader Specification:
4244 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
4246 * Boot time console output has been improved to provide
4247 animated boot time output for hanging jobs.
4249 * A new tool systemd-activate has been added which can be used
4250 to test socket activation with, directly from the command
4251 line. This should make it much easier to test and debug
4252 socket activation in daemons.
4254 * journalctl gained a new "--reverse" (or -r) option to show
4255 journal output in reverse order (i.e. newest line first).
4257 * journalctl gained a new "--pager-end" (or -e) option to jump
4258 to immediately jump to the end of the journal in the
4259 pager. This is only supported in conjunction with "less".
4261 * journalctl gained a new "--user-unit=" option, that works
4262 similarly to "--unit=" but filters for user units rather than
4265 * A number of unit files to ease adoption of systemd in
4266 initrds has been added. This moves some minimal logic from
4267 the various initrd implementations into systemd proper.
4269 * The journal files are now owned by a new group
4270 "systemd-journal", which exists specifically to allow access
4271 to the journal, and nothing else. Previously, we used the
4272 "adm" group for that, which however possibly covers more
4273 than just journal/log file access. This new group is now
4274 already used by systemd-journal-gatewayd to ensure this
4275 daemon gets access to the journal files and as little else
4276 as possible. Note that "make install" will also set FS ACLs
4277 up for /var/log/journal to give "adm" and "wheel" read
4278 access to it, in addition to "systemd-journal" which owns
4279 the journal files. We recommend that packaging scripts also
4280 add read access to "adm" + "wheel" to /var/log/journal, and
4281 all existing/future journal files. To normal users and
4282 administrators little changes, however packagers need to
4283 ensure to create the "systemd-journal" system group at
4284 package installation time.
4286 * The systemd-journal-gatewayd now runs as unprivileged user
4287 systemd-journal-gateway:systemd-journal-gateway. Packaging
4288 scripts need to create these system user/group at
4291 * timedated now exposes a new boolean property CanNTP that
4292 indicates whether a local NTP service is available or not.
4294 * systemd-detect-virt will now also detect xen PVs
4296 * The pstore file system is now mounted by default, if it is
4299 * In addition to the SELinux and IMA policies we will now also
4300 load SMACK policies at early boot.
4302 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
4303 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
4304 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
4305 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
4306 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
4307 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
4308 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
4309 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
4310 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
4311 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
4312 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
4313 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
4314 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
4315 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
4319 * Timer units now support calendar time events in addition to
4320 monotonic time events. That means you can now trigger a unit
4321 based on a calendar time specification such as "Thu,Fri
4322 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
4323 or fifth day of any month of the year 2013, given that it is
4324 a thursday or friday. This brings timer event support
4325 considerably closer to cron's capabilities. For details on
4326 the supported calendar time specification language see
4329 * udev now supports a number of different naming policies for
4330 network interfaces for predictable names, and a combination
4331 of these policies is now the default. Please see this wiki
4332 document for details:
4334 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
4336 * Auke Kok's bootchart implementation has been added to the
4337 systemd tree. It is an optional component that can graph the
4338 boot in quite some detail. It is one of the best bootchart
4339 implementations around and minimal in its code and
4342 * nss-myhostname has been integrated into the systemd source
4343 tree. nss-myhostname guarantees that the local hostname
4344 always stays resolvable via NSS. It has been a weak
4345 requirement of systemd-hostnamed since a long time, and
4346 since its code is actually trivial we decided to just
4347 include it in systemd's source tree. It can be turned off
4348 with a configure switch.
4350 * The read-ahead logic is now capable of properly detecting
4351 whether a btrfs file system is on SSD or rotating media, in
4352 order to optimize the read-ahead scheme. Previously, it was
4353 only capable of detecting this on traditional file systems
4356 * In udev, additional device properties are now read from the
4357 IAB in addition to the OUI database. Also, Bluetooth company
4358 identities are attached to the devices as well.
4360 * In service files %U may be used as specifier that is
4361 replaced by the configured user name of the service.
4363 * nspawn may now be invoked without a controlling TTY. This
4364 makes it suitable for invocation as its own service. This
4365 may be used to set up a simple containerized server system
4366 using only core OS tools.
4368 * systemd and nspawn can now accept socket file descriptors
4369 when they are started for socket activation. This enables
4370 implementation of socket activated nspawn
4371 containers. i.e. think about autospawning an entire OS image
4372 when the first SSH or HTTP connection is received. We expect
4373 that similar functionality will also be added to libvirt-lxc
4376 * journalctl will now suppress ANSI color codes when
4377 presenting log data.
4379 * systemctl will no longer show control group information for
4380 a unit if a the control group is empty anyway.
4382 * logind can now automatically suspend/hibernate/shutdown the
4385 * /etc/machine-info and hostnamed now also expose the chassis
4386 type of the system. This can be used to determine whether
4387 the local system is a laptop, desktop, handset or
4388 tablet. This information may either be configured by the
4389 user/vendor or is automatically determined from ACPI and DMI
4390 information if possible.
4392 * A number of PolicyKit actions are now bound together with
4393 "imply" rules. This should simplify creating UIs because
4394 many actions will now authenticate similar ones as well.
4396 * Unit files learnt a new condition ConditionACPower= which
4397 may be used to conditionalize a unit depending on whether an
4398 AC power source is connected or not, of whether the system
4399 is running on battery power.
4401 * systemctl gained a new "is-failed" verb that may be used in
4402 shell scripts and suchlike to check whether a specific unit
4403 is in the "failed" state.
4405 * The EnvironmentFile= setting in unit files now supports file
4406 globbing, and can hence be used to easily read a number of
4407 environment files at once.
4409 * systemd will no longer detect and recognize specific
4410 distributions. All distribution-specific #ifdeffery has been
4411 removed, systemd is now fully generic and
4412 distribution-agnostic. Effectively, not too much is lost as
4413 a lot of the code is still accessible via explicit configure
4414 switches. However, support for some distribution specific
4415 legacy configuration file formats has been dropped. We
4416 recommend distributions to simply adopt the configuration
4417 files everybody else uses now and convert the old
4418 configuration from packaging scripts. Most distributions
4419 already did that. If that's not possible or desirable,
4420 distributions are welcome to forward port the specific
4421 pieces of code locally from the git history.
4423 * When logging a message about a unit systemd will now always
4424 log the unit name in the message meta data.
4426 * localectl will now also discover system locale data that is
4427 not stored in locale archives, but directly unpacked.
4429 * logind will no longer unconditionally use framebuffer
4430 devices as seat masters, i.e. as devices that are required
4431 to be existing before a seat is considered preset. Instead,
4432 it will now look for all devices that are tagged as
4433 "seat-master" in udev. By default, framebuffer devices will
4434 be marked as such, but depending on local systems, other
4435 devices might be marked as well. This may be used to
4436 integrate graphics cards using closed source drivers (such
4437 as NVidia ones) more nicely into logind. Note however, that
4438 we recommend using the open source NVidia drivers instead,
4439 and no udev rules for the closed-source drivers will be
4440 shipped from us upstream.
4442 Contributions from: Adam Williamson, Alessandro Crismani, Auke
4443 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
4444 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
4445 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
4446 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
4447 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
4448 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
4449 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
4450 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
4451 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
4452 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
4453 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
4458 * udev gained support for loading additional device properties
4459 from an indexed database that is keyed by vendor/product IDs
4460 and similar device identifiers. For the beginning this
4461 "hwdb" is populated with data from the well-known PCI and
4462 USB database, but also includes PNP, ACPI and OID data. In
4463 the longer run this indexed database shall grow into
4464 becoming the one central database for non-essential
4465 userspace device metadata. Previously, data from the PCI/USB
4466 database was only attached to select devices, since the
4467 lookup was a relatively expensive operation due to O(n) time
4468 complexity (with n being the number of entries in the
4469 database). Since this is now O(1), we decided to add in this
4470 data for all devices where this is available, by
4471 default. Note that the indexed database needs to be rebuilt
4472 when new data files are installed. To achieve this you need
4473 to update your packaging scripts to invoke "udevadm hwdb
4474 --update" after installation of hwdb data files. For
4475 RPM-based distributions we introduced the new
4476 %udev_hwdb_update macro for this purpose.
4478 * The Journal gained support for the "Message Catalog", an
4479 indexed database to link up additional information with
4480 journal entries. For further details please check:
4482 http://www.freedesktop.org/wiki/Software/systemd/catalog
4484 The indexed message catalog database also needs to be
4485 rebuilt after installation of message catalog files. Use
4486 "journalctl --update-catalog" for this. For RPM-based
4487 distributions we introduced the %journal_catalog_update
4488 macro for this purpose.
4490 * The Python Journal bindings gained support for the standard
4491 Python logging framework.
4493 * The Journal API gained new functions for checking whether
4494 the underlying file system of a journal file is capable of
4495 properly reporting file change notifications, or whether
4496 applications that want to reflect journal changes "live"
4497 need to recheck journal files continuously in appropriate
4500 * It is now possible to set the "age" field for tmpfiles
4501 entries to 0, indicating that files matching this entry
4502 shall always be removed when the directories are cleaned up.
4504 * coredumpctl gained a new "gdb" verb which invokes gdb
4505 right-away on the selected coredump.
4507 * There's now support for "hybrid sleep" on kernels that
4508 support this, in addition to "suspend" and "hibernate". Use
4509 "systemctl hybrid-sleep" to make use of this.
4511 * logind's HandleSuspendKey= setting (and related settings)
4512 now gained support for a new "lock" setting to simply
4513 request the screen lock on all local sessions, instead of
4514 actually executing a suspend or hibernation.
4516 * systemd will now mount the EFI variables file system by
4519 * Socket units now gained support for configuration of the
4520 SMACK security label.
4522 * timedatectl will now output the time of the last and next
4523 daylight saving change.
4525 * We dropped support for various legacy and distro-specific
4526 concepts, such as insserv, early-boot SysV services
4527 (i.e. those for non-standard runlevels such as 'b' or 'S')
4528 or ArchLinux /etc/rc.conf support. We recommend the
4529 distributions who still need support this to either continue
4530 to maintain the necessary patches downstream, or find a
4531 different solution. (Talk to us if you have questions!)
4533 * Various systemd components will now bypass PolicyKit checks
4534 for root and otherwise handle properly if PolicyKit is not
4535 found to be around. This should fix most issues for
4536 PolicyKit-less systems. Quite frankly this should have been
4537 this way since day one. It is absolutely our intention to
4538 make systemd work fine on PolicyKit-less systems, and we
4539 consider it a bug if something does not work as it should if
4540 PolicyKit is not around.
4542 * For embedded systems it is now possible to build udev and
4543 systemd without blkid and/or kmod support.
4545 * "systemctl switch-root" is now capable of switching root
4546 more than once. I.e. in addition to transitions from the
4547 initrd to the host OS it is now possible to transition to
4548 further OS images from the host. This is useful to implement
4549 offline updating tools.
4551 * Various other additions have been made to the RPM macros
4552 shipped with systemd. Use %udev_rules_update() after
4553 installing new udev rules files. %_udevhwdbdir,
4554 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
4555 %_sysctldir are now available which resolve to the right
4556 directories for packages to place various data files in.
4558 * journalctl gained the new --full switch (in addition to
4559 --all, to disable ellipsation for long messages.
4561 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
4562 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
4563 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
4564 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
4565 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
4566 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
4567 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
4568 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
4569 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
4573 * journalctl gained new --since= and --until= switches to
4574 filter by time. It also now supports nice filtering for
4575 units via --unit=/-u.
4577 * Type=oneshot services may use ExecReload= and do the
4580 * The journal daemon now supports time-based rotation and
4581 vacuuming, in addition to the usual disk-space based
4584 * The journal will now index the available field values for
4585 each field name. This enables clients to show pretty drop
4586 downs of available match values when filtering. The bash
4587 completion of journalctl has been updated
4588 accordingly. journalctl gained a new switch -F to list all
4589 values a certain field takes in the journal database.
4591 * More service events are now written as structured messages
4592 to the journal, and made recognizable via message IDs.
4594 * The timedated, localed and hostnamed mini-services which
4595 previously only provided support for changing time, locale
4596 and hostname settings from graphical DEs such as GNOME now
4597 also have a minimal (but very useful) text-based client
4598 utility each. This is probably the nicest way to changing
4599 these settings from the command line now, especially since
4600 it lists available options and is fully integrated with bash
4603 * There's now a new tool "systemd-coredumpctl" to list and
4604 extract coredumps from the journal.
4606 * We now install a README each in /var/log/ and
4607 /etc/rc.d/init.d explaining where the system logs and init
4608 scripts went. This hopefully should help folks who go to
4609 that dirs and look into the otherwise now empty void and
4610 scratch their heads.
4612 * When user-services are invoked (by systemd --user) the
4613 $MANAGERPID env var is set to the PID of systemd.
4615 * SIGRTMIN+24 when sent to a --user instance will now result
4616 in immediate termination of systemd.
4618 * gatewayd received numerous feature additions such as a
4619 "follow" mode, for live syncing and filtering.
4621 * browse.html now allows filtering and showing detailed
4622 information on specific entries. Keyboard navigation and
4623 mouse screen support has been added.
4625 * gatewayd/journalctl now supports HTML5/JSON
4626 Server-Sent-Events as output.
4628 * The SysV init script compatibility logic will now
4629 heuristically determine whether a script supports the
4630 "reload" verb, and only then make this available as
4633 * "systemctl status --follow" has been removed, use "journalctl
4636 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
4637 have been removed since they are hardly useful to be
4640 * And I'd like to take the opportunity to specifically mention
4641 Zbigniew for his great contributions. Zbigniew, you rock!
4643 Contributions from: Andrew Eikum, Christian Hesse, Colin
4644 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
4645 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
4646 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
4647 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
4648 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
4649 Jędrzejewski-Szmek, Сковорода Никита Андреевич
4653 * If /etc/vconsole.conf is non-existent or empty we will no
4654 longer load any console font or key map at boot by
4655 default. Instead the kernel defaults will be left
4656 intact. This is definitely the right thing to do, as no
4657 configuration should mean no configuration, and hard-coding
4658 font names that are different on all archs is probably a bad
4659 idea. Also, the kernel default key map and font should be
4660 good enough for most cases anyway, and mostly identical to
4661 the userspace fonts/key maps we previously overloaded them
4662 with. If distributions want to continue to default to a
4663 non-kernel font or key map they should ship a default
4664 /etc/vconsole.conf with the appropriate contents.
4666 Contributions from: Colin Walters, Daniel J Walsh, Dave
4667 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
4668 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
4672 * journalctl gained a new --cursor= switch to show entries
4673 starting from the specified location in the journal.
4675 * We now enforce a size limit on journal entry fields exported
4676 with "-o json" in journalctl. Fields larger than 4K will be
4677 assigned null. This can be turned off with --all.
4679 * An (optional) journal gateway daemon is now available as
4680 "systemd-journal-gatewayd.service". This service provides
4681 access to the journal via HTTP and JSON. This functionality
4682 will be used to implement live log synchronization in both
4683 pull and push modes, but has various other users too, such
4684 as easy log access for debugging of embedded devices. Right
4685 now it is already useful to retrieve the journal via HTTP:
4687 # systemctl start systemd-journal-gatewayd.service
4688 # wget http://localhost:19531/entries
4690 This will download the journal contents in a
4691 /var/log/messages compatible format. The same as JSON:
4693 # curl -H"Accept: application/json" http://localhost:19531/entries
4695 This service is also accessible via a web browser where a
4696 single static HTML5 app is served that uses the JSON logic
4697 to enable the user to do some basic browsing of the
4698 journal. This will be extended later on. Here's an example
4699 screenshot of this app in its current state:
4701 http://0pointer.de/public/journal-gatewayd
4703 Contributions from: Kay Sievers, Lennart Poettering, Robert
4704 Milasan, Tom Gundersen
4708 * The bash completion logic is now available for journalctl
4711 * We do not mount the "cpuset" controller anymore together with
4712 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
4713 started if no parameters are assigned to it. "cpuset" hence
4714 broke code that assumed it it could create "cpu" groups and
4717 * journalctl -f will now subscribe to terminal size changes,
4718 and line break accordingly.
4720 Contributions from: Dave Reisner, Kay Sievers, Lennart
4721 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
4725 * nspawn will now create a symlink /etc/localtime in the
4726 container environment, copying the host's timezone
4727 setting. Previously this has been done via a bind mount, but
4728 since symlinks cannot be bind mounted this has now been
4729 changed to create/update the appropriate symlink.
4731 * journalctl -n's line number argument is now optional, and
4732 will default to 10 if omitted.
4734 * journald will now log the maximum size the journal files may
4735 take up on disk. This is particularly useful if the default
4736 built-in logic of determining this parameter from the file
4737 system size is used. Use "systemctl status
4738 systemd-journald.service" to see this information.
4740 * The multi-seat X wrapper tool has been stripped down. As X
4741 is now capable of enumerating graphics devices via udev in a
4742 seat-aware way the wrapper is not strictly necessary
4743 anymore. A stripped down temporary stop-gap is still shipped
4744 until the upstream display managers have been updated to
4745 fully support the new X logic. Expect this wrapper to be
4746 removed entirely in one of the next releases.
4748 * HandleSleepKey= in logind.conf has been split up into
4749 HandleSuspendKey= and HandleHibernateKey=. The old setting
4750 is not available anymore. X11 and the kernel are
4751 distinguishing between these keys and we should too. This
4752 also means the inhibition lock for these keys has been split
4755 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
4756 Poettering, Lukas Nykryn, Václav Pavlín
4760 * Whenever a unit changes state we will now log this to the
4761 journal and show along the unit's own log output in
4764 * ConditionPathIsMountPoint= can now properly detect bind
4765 mount points too. (Previously, a bind mount of one file
4766 system to another place in the same file system could not be
4767 detected as mount, since they shared struct stat's st_dev
4770 * We will now mount the cgroup controllers cpu, cpuacct,
4771 cpuset and the controllers net_cls, net_prio together by
4774 * nspawn containers will now have a virtualized boot
4775 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
4776 over with a randomized ID at container initialization). This
4777 has the effect of making "journalctl -b" do the right thing
4780 * The JSON output journal serialization has been updated not
4781 to generate "endless" list objects anymore, but rather one
4782 JSON object per line. This is more in line how most JSON
4783 parsers expect JSON objects. The new output mode
4784 "json-pretty" has been added to provide similar output, but
4785 neatly aligned for readability by humans.
4787 * We dropped all explicit sync() invocations in the shutdown
4788 code. The kernel does this implicitly anyway in the kernel
4789 reboot() syscall. halt(8)'s -n option is now a compatibility
4792 * We now support virtualized reboot() in containers, as
4793 supported by newer kernels. We will fall back to exit() if
4794 CAP_SYS_REBOOT is not available to the container. Also,
4795 nspawn makes use of this now and will actually reboot the
4796 container if the containerized OS asks for that.
4798 * journalctl will only show local log output by default
4799 now. Use --merge (-m) to show remote log output, too.
4801 * libsystemd-journal gained the new sd_journal_get_usage()
4802 call to determine the current disk usage of all journal
4803 files. This is exposed in the new "journalctl --disk-usage"
4806 * journald gained a new configuration setting SplitMode= in
4807 journald.conf which may be used to control how user journals
4808 are split off. See journald.conf(5) for details.
4810 * A new condition type ConditionFileNotEmpty= has been added.
4812 * tmpfiles' "w" lines now support file globbing, to write
4813 multiple files at once.
4815 * We added Python bindings for the journal submission
4816 APIs. More Python APIs for a number of selected APIs will
4817 likely follow. Note that we intend to add native bindings
4818 only for the Python language, as we consider it common
4819 enough to deserve bindings shipped within systemd. There are
4820 various projects outside of systemd that provide bindings
4821 for languages such as PHP or Lua.
4823 * Many conditions will now resolve specifiers such as %i. In
4824 addition, PathChanged= and related directives of .path units
4825 now support specifiers as well.
4827 * There's now a new RPM macro definition for the system preset
4830 * journald will now warn if it ca not forward a message to the
4831 syslog daemon because its socket is full.
4833 * timedated will no longer write or process /etc/timezone,
4834 except on Debian. As we do not support late mounted /usr
4835 anymore /etc/localtime always being a symlink is now safe,
4836 and hence the information in /etc/timezone is not necessary
4839 * logind will now always reserve one VT for a text getty (VT6
4840 by default). Previously if more than 6 X sessions where
4841 started they took up all the VTs with auto-spawned gettys,
4842 so that no text gettys were available anymore.
4844 * udev will now automatically inform the btrfs kernel logic
4845 about btrfs RAID components showing up. This should make
4846 simple hotplug based btrfs RAID assembly work.
4848 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
4849 (but not for its children which will stay at the kernel
4850 default). This should allow setups with a lot more listening
4853 * systemd will now always pass the configured timezone to the
4854 kernel at boot. timedated will do the same when the timezone
4857 * logind's inhibition logic has been updated. By default,
4858 logind will now handle the lid switch, the power and sleep
4859 keys all the time, even in graphical sessions. If DEs want
4860 to handle these events on their own they should take the new
4861 handle-power-key, handle-sleep-key and handle-lid-switch
4862 inhibitors during their runtime. A simple way to achieve
4863 that is to invoke the DE wrapped in an invocation of:
4865 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
4867 * Access to unit operations is now checked via SELinux taking
4868 the unit file label and client process label into account.
4870 * systemd will now notify the administrator in the journal
4871 when he over-mounts a non-empty directory.
4873 * There are new specifiers that are resolved in unit files,
4874 for the host name (%H), the machine ID (%m) and the boot ID
4877 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
4878 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
4879 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
4880 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
4881 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
4882 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
4883 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
4887 * Support for reading structured kernel messages from
4888 /dev/kmsg has now been added and is enabled by default.
4890 * Support for reading kernel messages from /proc/kmsg has now
4891 been removed. If you want kernel messages in the journal
4892 make sure to run a recent kernel (>= 3.5) that supports
4893 reading structured messages from /dev/kmsg (see
4894 above). /proc/kmsg is now exclusive property of classic
4895 syslog daemons again.
4897 * The libudev API gained the new
4898 udev_device_new_from_device_id() call.
4900 * The logic for file system namespace (ReadOnlyDirectory=,
4901 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
4902 require pivot_root() anymore. This means fewer temporary
4903 directories are created below /tmp for this feature.
4905 * nspawn containers will now see and receive all submounts
4906 made on the host OS below the root file system of the
4909 * Forward Secure Sealing is now supported for Journal files,
4910 which provide cryptographical sealing of journal files so
4911 that attackers cannot alter log history anymore without this
4912 being detectable. Lennart will soon post a blog story about
4913 this explaining it in more detail.
4915 * There are two new service settings RestartPreventExitStatus=
4916 and SuccessExitStatus= which allow configuration of exit
4917 status (exit code or signal) which will be excepted from the
4918 restart logic, resp. consider successful.
4920 * journalctl gained the new --verify switch that can be used
4921 to check the integrity of the structure of journal files and
4922 (if Forward Secure Sealing is enabled) the contents of
4925 * nspawn containers will now be run with /dev/stdin, /dev/fd/
4926 and similar symlinks pre-created. This makes running shells
4927 as container init process a lot more fun.
4929 * The fstab support can now handle PARTUUID= and PARTLABEL=
4932 * A new ConditionHost= condition has been added to match
4933 against the hostname (with globs) and machine ID. This is
4934 useful for clusters where a single OS image is used to
4935 provision a large number of hosts which shall run slightly
4936 different sets of services.
4938 * Services which hit the restart limit will now be placed in a
4941 Contributions from: Bertram Poettering, Dave Reisner, Huang
4942 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
4943 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
4947 * When running in --user mode systemd will now become a
4948 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
4949 tree a lot more organized.
4951 * A new PartOf= unit dependency type has been introduced that
4952 may be used to group services in a natural way.
4954 * "systemctl enable" may now be used to enable instances of
4957 * journalctl now prints error log levels in red, and
4958 warning/notice log levels in bright white. It also supports
4959 filtering by log level now.
4961 * cgtop gained a new -n switch (similar to top), to configure
4962 the maximum number of iterations to run for. It also gained
4963 -b, to run in batch mode (accepting no input).
4965 * The suffix ".service" may now be omitted on most systemctl
4966 command lines involving service unit names.
4968 * There's a new bus call in logind to lock all sessions, as
4969 well as a loginctl verb for it "lock-sessions".
4971 * libsystemd-logind.so gained a new call sd_journal_perror()
4972 that works similar to libc perror() but logs to the journal
4973 and encodes structured information about the error number.
4975 * /etc/crypttab entries now understand the new keyfile-size=
4978 * shutdown(8) now can send a (configurable) wall message when
4979 a shutdown is cancelled.
4981 * The mount propagation mode for the root file system will now
4982 default to "shared", which is useful to make containers work
4983 nicely out-of-the-box so that they receive new mounts from
4984 the host. This can be undone locally by running "mount
4985 --make-rprivate /" if needed.
4987 * The prefdm.service file has been removed. Distributions
4988 should maintain this unit downstream if they intend to keep
4989 it around. However, we recommend writing normal unit files
4990 for display managers instead.
4992 * Since systemd is a crucial part of the OS we will now
4993 default to a number of compiler switches that improve
4994 security (hardening) such as read-only relocations, stack
4995 protection, and suchlike.
4997 * The TimeoutSec= setting for services is now split into
4998 TimeoutStartSec= and TimeoutStopSec= to allow configuration
4999 of individual time outs for the start and the stop phase of
5002 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
5003 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
5004 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
5005 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
5006 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
5007 Gundersen, Zbigniew Jędrzejewski-Szmek
5011 * The journal and id128 C APIs are now fully documented as man
5014 * Extra safety checks have been added when transitioning from
5015 the initial RAM disk to the main system to avoid accidental
5018 * /etc/crypttab entries now understand the new keyfile-offset=
5021 * systemctl -t can now be used to filter by unit load state.
5023 * The journal C API gained the new sd_journal_wait() call to
5024 make writing synchronous journal clients easier.
5026 * journalctl gained the new -D switch to show journals from a
5029 * journalctl now displays a special marker between log
5030 messages of two different boots.
5032 * The journal is now explicitly flushed to /var via a service
5033 systemd-journal-flush.service, rather than implicitly simply
5034 by seeing /var/log/journal to be writable.
5036 * journalctl (and the journal C APIs) can now match for much
5037 more complex expressions, with alternatives and
5040 * When transitioning from the initial RAM disk to the main
5041 system we will now kill all processes in a killing spree to
5042 ensure no processes stay around by accident.
5044 * Three new specifiers may be used in unit files: %u, %h, %s
5045 resolve to the user name, user home directory resp. user
5046 shell. This is useful for running systemd user instances.
5048 * We now automatically rotate journal files if their data
5049 object hash table gets a fill level > 75%. We also size the
5050 hash table based on the configured maximum file size. This
5051 together should lower hash collisions drastically and thus
5052 speed things up a bit.
5054 * journalctl gained the new "--header" switch to introspect
5055 header data of journal files.
5057 * A new setting SystemCallFilters= has been added to services
5058 which may be used to apply blacklists or whitelists to
5059 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
5061 * nspawn gained a new --link-journal= switch (and quicker: -j)
5062 to link the container journal with the host. This makes it
5063 very easy to centralize log viewing on the host for all
5064 guests while still keeping the journal files separated.
5066 * Many bugfixes and optimizations
5068 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
5069 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
5070 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
5075 * Several tools now understand kernel command line arguments,
5076 which are only read when run in an initial RAM disk. They
5077 usually follow closely their normal counterparts, but are
5080 * There's a new tool to analyze the readahead files that are
5081 automatically generated at boot. Use:
5083 /usr/lib/systemd/systemd-readahead analyze /.readahead
5085 * We now provide an early debug shell on tty9 if this enabled. Use:
5087 systemctl enable debug-shell.service
5089 * All plymouth related units have been moved into the Plymouth
5090 package. Please make sure to upgrade your Plymouth version
5093 * systemd-tmpfiles now supports getting passed the basename of
5094 a configuration file only, in which case it will look for it
5095 in all appropriate directories automatically.
5097 * udevadm info now takes a /dev or /sys path as argument, and
5098 does the right thing. Example:
5100 udevadm info /dev/sda
5101 udevadm info /sys/class/block/sda
5103 * systemctl now prints a warning if a unit is stopped but a
5104 unit that might trigger it continues to run. Example: a
5105 service is stopped but the socket that activates it is left
5108 * "systemctl status" will now mention if the log output was
5109 shortened due to rotation since a service has been started.
5111 * The journal API now exposes functions to determine the
5112 "cutoff" times due to rotation.
5114 * journald now understands SIGUSR1 and SIGUSR2 for triggering
5115 immediately flushing of runtime logs to /var if possible,
5116 resp. for triggering immediate rotation of the journal
5119 * It is now considered an error if a service is attempted to
5120 be stopped that is not loaded.
5122 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
5124 * systemd-analyze now supports Python 3
5126 * tmpfiles now supports cleaning up directories via aging
5127 where the first level dirs are always kept around but
5128 directories beneath it automatically aged. This is enabled
5129 by prefixing the age field with '~'.
5131 * Seat objects now expose CanGraphical, CanTTY properties
5132 which is required to deal with very fast bootups where the
5133 display manager might be running before the graphics drivers
5134 completed initialization.
5136 * Seat objects now expose a State property.
5138 * We now include RPM macros for service enabling/disabling
5139 based on the preset logic. We recommend RPM based
5140 distributions to make use of these macros if possible. This
5141 makes it simpler to reuse RPM spec files across
5144 * We now make sure that the collected systemd unit name is
5145 always valid when services log to the journal via
5148 * There's a new man page kernel-command-line(7) detailing all
5149 command line options we understand.
5151 * The fstab generator may now be disabled at boot by passing
5152 fstab=0 on the kernel command line.
5154 * A new kernel command line option modules-load= is now understood
5155 to load a specific kernel module statically, early at boot.
5157 * Unit names specified on the systemctl command line are now
5158 automatically escaped as needed. Also, if file system or
5159 device paths are specified they are automatically turned
5160 into the appropriate mount or device unit names. Example:
5162 systemctl status /home
5163 systemctl status /dev/sda
5165 * The SysVConsole= configuration option has been removed from
5166 system.conf parsing.
5168 * The SysV search path is no longer exported on the D-Bus
5171 * The Names= option is been removed from unit file parsing.
5173 * There's a new man page bootup(7) detailing the boot process.
5175 * Every unit and every generator we ship with systemd now
5176 comes with full documentation. The self-explanatory boot is
5179 * A couple of services gained "systemd-" prefixes in their
5180 name if they wrap systemd code, rather than only external
5181 code. Among them fsck@.service which is now
5182 systemd-fsck@.service.
5184 * The HaveWatchdog property has been removed from the D-Bus
5187 * systemd.confirm_spawn= on the kernel command line should now
5190 * There's a new man page crypttab(5) which details all options
5191 we actually understand.
5193 * systemd-nspawn gained a new --capability= switch to pass
5194 additional capabilities to the container.
5196 * timedated will now read known NTP implementation unit names
5197 from /usr/lib/systemd/ntp-units.d/*.list,
5198 systemd-timedated-ntp.target has been removed.
5200 * journalctl gained a new switch "-b" that lists log data of
5201 the current boot only.
5203 * The notify socket is in the abstract namespace again, in
5204 order to support daemons which chroot() at start-up.
5206 * There is a new Storage= configuration option for journald
5207 which allows configuration of where log data should go. This
5208 also provides a way to disable journal logging entirely, so
5209 that data collected is only forwarded to the console, the
5210 kernel log buffer or another syslog implementation.
5212 * Many bugfixes and optimizations
5214 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
5215 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
5216 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
5217 Shawn Landden, Tom Gundersen
5221 * "systemctl help <unit>" now shows the man page if one is
5224 * Several new man pages have been added.
5226 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
5227 MaxLevelConsole= can now be specified in
5228 journald.conf. These options allow reducing the amount of
5229 data stored on disk or forwarded by the log level.
5231 * TimerSlackNSec= can now be specified in system.conf for
5232 PID1. This allows system-wide power savings.
5234 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
5235 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
5240 * logind is now capable of (optionally) handling power and
5241 sleep keys as well as the lid switch.
5243 * journalctl now understands the syntax "journalctl
5244 /usr/bin/avahi-daemon" to get all log output of a specific
5247 * CapabilityBoundingSet= in system.conf now also influences
5248 the capability bound set of usermode helpers of the kernel.
5250 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
5251 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
5252 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
5253 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
5257 * Note that we skipped 139 releases here in order to set the
5258 new version to something that is greater than both udev's
5259 and systemd's most recent version number.
5261 * udev: all udev sources are merged into the systemd source tree now.
5262 All future udev development will happen in the systemd tree. It
5263 is still fully supported to use the udev daemon and tools without
5264 systemd running, like in initramfs or other init systems. Building
5265 udev though, will require the *build* of the systemd tree, but
5266 udev can be properly *run* without systemd.
5268 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
5269 should be used to create dead device nodes as workarounds for broken
5272 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
5273 no longer supported. udev_monitor_new_from_netlink() needs to be
5274 used to subscribe to events.
5276 * udev: when udevd is started by systemd, processes which are left
5277 behind by forking them off of udev rules, are unconditionally cleaned
5278 up and killed now after the event handling has finished. Services or
5279 daemons must be started as systemd services. Services can be
5280 pulled-in by udev to get started, but they can no longer be directly
5281 forked by udev rules.
5283 * udev: the daemon binary is called systemd-udevd now and installed
5284 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
5285 to adapt to that, create symlink, or rename the binary after building
5288 * libudev no longer provides these symbols:
5289 udev_monitor_from_socket()
5290 udev_queue_get_failed_list_entry()
5291 udev_get_{dev,sys,run}_path()
5292 The versions number was bumped and symbol versioning introduced.
5294 * systemd-loginctl and systemd-journalctl have been renamed
5295 to loginctl and journalctl to match systemctl.
5297 * The config files: /etc/systemd/systemd-logind.conf and
5298 /etc/systemd/systemd-journald.conf have been renamed to
5299 logind.conf and journald.conf. Package updates should rename
5300 the files to the new names on upgrade.
5302 * For almost all files the license is now LGPL2.1+, changed
5303 from the previous GPL2.0+. Exceptions are some minor stuff
5304 of udev (which will be changed to LGPL2.1 eventually, too),
5305 and the MIT licensed sd-daemon.[ch] library that is suitable
5306 to be used as drop-in files.
5308 * systemd and logind now handle system sleep states, in
5309 particular suspending and hibernating.
5311 * logind now implements a sleep/shutdown/idle inhibiting logic
5312 suitable for a variety of uses. Soonishly Lennart will blog
5313 about this in more detail.
5315 * var-run.mount and var-lock.mount are no longer provided
5316 (which prevously bind mounted these directories to their new
5317 places). Distributions which have not converted these
5318 directories to symlinks should consider stealing these files
5319 from git history and add them downstream.
5321 * We introduced the Documentation= field for units and added
5322 this to all our shipped units. This is useful to make it
5323 easier to explore the boot and the purpose of the various
5326 * All smaller setup units (such as
5327 systemd-vconsole-setup.service) now detect properly if they
5328 are run in a container and are skipped when
5329 appropriate. This guarantees an entirely noise-free boot in
5330 Linux container environments such as systemd-nspawn.
5332 * A framework for implementing offline system updates is now
5333 integrated, for details see:
5334 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
5336 * A new service type Type=idle is available now which helps us
5337 avoiding ugly interleaving of getty output and boot status
5340 * There's now a system-wide CapabilityBoundingSet= option to
5341 globally reduce the set of capabilities for the
5342 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
5343 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
5344 even CAP_NET_ADMIN system-wide for secure systems.
5346 * There are now system-wide DefaultLimitXXX= options to
5347 globally change the defaults of the various resource limits
5348 for all units started by PID 1.
5350 * Harald Hoyer's systemd test suite has been integrated into
5351 systemd which allows easy testing of systemd builds in qemu
5352 and nspawn. (This is really awesome! Ask us for details!)
5354 * The fstab parser is now implemented as generator, not inside
5357 * systemctl will now warn you if .mount units generated from
5358 /etc/fstab are out of date due to changes in fstab that
5359 have not been read by systemd yet.
5361 * systemd is now suitable for usage in initrds. Dracut has
5362 already been updated to make use of this. With this in place
5363 initrds get a slight bit faster but primarily are much
5364 easier to introspect and debug since "systemctl status" in
5365 the host system can be used to introspect initrd services,
5366 and the journal from the initrd is kept around too.
5368 * systemd-delta has been added, a tool to explore differences
5369 between user/admin configuration and vendor defaults.
5371 * PrivateTmp= now affects both /tmp and /var/tmp.
5373 * Boot time status messages are now much prettier and feature
5374 proper english language. Booting up systemd has never been
5377 * Read-ahead pack files now include the inode number of all
5378 files to pre-cache. When the inode changes the pre-caching
5379 is not attempted. This should be nicer to deal with updated
5380 packages which might result in changes of read-ahead
5383 * We now temporaritly lower the kernel's read_ahead_kb variable
5384 when collecting read-ahead data to ensure the kernel's
5385 built-in read-ahead does not add noise to our measurements
5386 of necessary blocks to pre-cache.
5388 * There's now RequiresMountsFor= to add automatic dependencies
5389 for all mounts necessary for a specific file system path.
5391 * MountAuto= and SwapAuto= have been removed from
5392 system.conf. Mounting file systems at boot has to take place
5395 * nspawn now learned a new switch --uuid= to set the machine
5396 ID on the command line.
5398 * nspawn now learned the -b switch to automatically search
5401 * vt102 is now the default TERM for serial TTYs, upgraded from
5404 * systemd-logind now works on VT-less systems.
5406 * The build tree has been reorganized. The individual
5407 components now have directories of their own.
5409 * A new condition type ConditionPathIsReadWrite= is now available.
5411 * nspawn learned the new -C switch to create cgroups for the
5412 container in other hierarchies.
5414 * We now have support for hardware watchdogs, configurable in
5417 * The scheduled shutdown logic now has a public API.
5419 * We now mount /tmp as tmpfs by default, but this can be
5420 masked and /etc/fstab can override it.
5422 * Since udisks does not make use of /media anymore we are not
5423 mounting a tmpfs on it anymore.
5425 * journalctl gained a new --local switch to only interleave
5426 locally generated journal files.
5428 * We can now load the IMA policy at boot automatically.
5430 * The GTK tools have been split off into a systemd-ui.
5432 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
5433 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
5434 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
5435 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
5436 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
5437 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
5438 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
5439 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
5440 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
5445 * This is mostly a bugfix release
5447 * Support optional initialization of the machine ID from the
5448 KVM or container configured UUID.
5450 * Support immediate reboots with "systemctl reboot -ff"
5452 * Show /etc/os-release data in systemd-analyze output
5454 * Many bugfixes for the journal, including endianness fixes and
5455 ensuring that disk space enforcement works
5457 * sd-login.h is C++ comptaible again
5459 * Extend the /etc/os-release format on request of the Debian
5462 * We now refuse non-UTF8 strings used in various configuration
5463 and unit files. This is done to ensure we do not pass invalid
5464 data over D-Bus or expose it elsewhere.
5466 * Register Mimo USB Screens as suitable for automatic seat
5469 * Read SELinux client context from journal clients in a race
5472 * Reorder configuration file lookup order. /etc now always
5473 overrides /run in order to allow the administrator to always
5474 and unconditionally override vendor-supplied or
5475 automatically generated data.
5477 * The various user visible bits of the journal now have man
5478 pages. We still lack man pages for the journal API calls
5481 * We now ship all man pages in HTML format again in the
5484 Contributions from: Dave Reisner, Dirk Eibach, Frederic
5485 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
5486 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
5491 * This is mostly a bugfix release
5493 * systems lacking /etc/os-release are no longer supported.
5495 * Various functionality updates to libsystemd-login.so
5497 * Track class of PAM logins to distinguish greeters from
5500 Contributions from: Kay Sievers, Lennart Poettering, Michael
5505 * This is an important bugfix release for v41.
5507 * Building man pages is now optional which should be useful
5508 for those building systemd from git but unwilling to install
5511 * Watchdog support for supervising services is now usable. In
5512 a future release support for hardware watchdogs
5513 (i.e. /dev/watchdog) will be added building on this.
5515 * Service start rate limiting is now configurable and can be
5516 turned off per service. When a start rate limit is hit a
5517 reboot can automatically be triggered.
5519 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
5521 Contributions from: Benjamin Franzke, Bill Nottingham,
5522 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
5523 Schmidt, Michał Górny, Piotr Drąg
5527 * The systemd binary is installed /usr/lib/systemd/systemd now;
5528 An existing /sbin/init symlink needs to be adapted with the
5531 * The code that loads kernel modules has been ported to invoke
5532 libkmod directly, instead of modprobe. This means we do not
5533 support systems with module-init-tools anymore.
5535 * Watchdog support is now already useful, but still not
5538 * A new kernel command line option systemd.setenv= is
5539 understood to set system wide environment variables
5540 dynamically at boot.
5542 * We now limit the set of capabilities of systemd-journald.
5544 * We now set SIGPIPE to ignore by default, since it only is
5545 useful in shell pipelines, and has little use in general
5546 code. This can be disabled with IgnoreSIPIPE=no in unit
5549 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
5550 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
5555 * This is mostly a bugfix release
5557 * We now expose the reason why a service failed in the
5558 "Result" D-Bus property.
5560 * Rudimentary service watchdog support (will be completed over
5561 the next few releases.)
5563 * When systemd forks off in order execute some service we will
5564 now immediately changes its argv[0] to reflect which process
5565 it will execute. This is useful to minimize the time window
5566 with a generic argv[0], which makes bootcharts more useful
5568 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
5569 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
5570 Mike Kazantsev, Ray Strode
5574 * This is mostly a test release, but incorporates many
5577 * New systemd-cgtop tool to show control groups by their
5580 * Linking against libacl for ACLs is optional again. If
5581 disabled, support tracking device access for active logins
5582 goes becomes unavailable, and so does access to the user
5583 journals by the respective users.
5585 * If a group "adm" exists, journal files are automatically
5586 owned by them, thus allow members of this group full access
5587 to the system journal as well as all user journals.
5589 * The journal now stores the SELinux context of the logging
5590 client for all entries.
5592 * Add C++ inclusion guards to all public headers
5594 * New output mode "cat" in the journal to print only text
5595 messages, without any meta data like date or time.
5597 * Include tiny X server wrapper as a temporary stop-gap to
5598 teach XOrg udev display enumeration. This is used by display
5599 managers such as gdm, and will go away as soon as XOrg
5600 learned native udev hotplugging for display devices.
5602 * Add new systemd-cat tool for executing arbitrary programs
5603 with STDERR/STDOUT connected to the journal. Can also act as
5604 BSD logger replacement, and does so by default.
5606 * Optionally store all locally generated coredumps in the
5607 journal along with meta data.
5609 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
5610 writing short strings to files (for usage for /sys), and for
5611 creating symlinks, character and block device nodes.
5613 * New unit file option ControlGroupPersistent= to make cgroups
5614 persistent, following the mechanisms outlined in
5615 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
5617 * Support multiple local RTCs in a sane way
5619 * No longer monopolize IO when replaying readahead data on
5620 rotating disks, since we might starve non-file-system IO to
5621 death, since fanotify() will not see accesses done by blkid,
5624 * Do not show kernel threads in systemd-cgls anymore, unless
5625 requested with new -k switch.
5627 Contributions from: Dan Horák, Kay Sievers, Lennart
5628 Poettering, Michal Schmidt
5632 * This is mostly a test release, but incorporates many
5635 * The git repository moved to:
5636 git://anongit.freedesktop.org/systemd/systemd
5637 ssh://git.freedesktop.org/git/systemd/systemd
5639 * First release with the journal
5640 http://0pointer.de/blog/projects/the-journal.html
5642 * The journal replaces both systemd-kmsg-syslogd and
5643 systemd-stdout-bridge.
5645 * New sd_pid_get_unit() API call in libsystemd-logind
5647 * Many systemadm clean-ups
5649 * Introduce remote-fs-pre.target which is ordered before all
5650 remote mounts and may be used to start services before all
5653 * Added Mageia support
5655 * Add bash completion for systemd-loginctl
5657 * Actively monitor PID file creation for daemons which exit in
5658 the parent process before having finished writing the PID
5659 file in the daemon process. Daemons which do this need to be
5660 fixed (i.e. PID file creation must have finished before the
5661 parent exits), but we now react a bit more gracefully to them.
5663 * Add colourful boot output, mimicking the well-known output
5664 of existing distributions.
5666 * New option PassCredentials= for socket units, for
5667 compatibility with a recent kernel ABI breakage.
5669 * /etc/rc.local is now hooked in via a generator binary, and
5670 thus will no longer act as synchronization point during
5673 * systemctl list-unit-files now supports --root=.
5675 * systemd-tmpfiles now understands two new commands: z, Z for
5676 relabelling files according to the SELinux database. This is
5677 useful to apply SELinux labels to specific files in /sys,
5680 * Output of SysV services is now forwarded to both the console
5681 and the journal by default, not only just the console.
5683 * New man pages for all APIs from libsystemd-login.
5685 * The build tree got reorganized and a the build system is a
5686 lot more modular allowing embedded setups to specifically
5687 select the components of systemd they are interested in.
5689 * Support for Linux systems lacking the kernel VT subsystem is
5692 * configure's --with-rootdir= got renamed to
5693 --with-rootprefix= to follow the naming used by udev and
5696 * Unless specified otherwise we will now install to /usr instead
5697 of /usr/local by default.
5699 * Processes with '@' in argv[0][0] are now excluded from the
5700 final shut-down killing spree, following the logic explained
5702 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
5704 * All processes remaining in a service cgroup when we enter
5705 the START or START_PRE states are now killed with
5706 SIGKILL. That means it is no longer possible to spawn
5707 background processes from ExecStart= lines (which was never
5708 supported anyway, and bad style).
5710 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
5711 reloading of units together.
5713 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
5714 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
5715 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
5716 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
5717 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek