1 systemd System and Service Manager
5 * A number of properties previously only settable in unit
6 files are now also available as properties to set when
7 creating transient units programmatically via the bus, as it
8 is exposed with systemd-run's --property=
9 setting. Specifically, these are: SyslogIdentifier=,
10 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
11 EnvironmentFile=, ReadWriteDirectories=,
12 ReadOnlyDirectories=, InaccessibleDirectories=,
13 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
15 * When creating transient services via the bus API it is now
16 possible to pass in a set of file descriptors to use as
17 STDIN/STDOUT/STDERR for the invoked process.
19 * Slice units may now be created transiently via the bus APIs,
20 similar to the way service and scope units may already be
23 * Wherever systemd expects a calendar timestamp specification
24 (like in journalctl's --since= and --until= switches) UTC
25 timestamps are now supported. Timestamps suffixed with "UTC"
26 are now considered to be in Universal Time Coordinated
27 instead of the local timezone. Also, timestamps may now
28 optionally be specified with sub-second accuracy. Both of
29 these additions also apply to recurring calendar event
30 specification, such as OnCalendar= in timer units.
32 * journalctl gained a new "--sync" switch that asks the
33 journal daemon to write all so far unwritten log messages to
34 disk and sync the files, before returning.
36 * systemd-tmpfiles learned two new line types "q" and "Q" that
37 operate like "v", but also set up a basic btrfs quota
38 hierarchy when used on a btrfs file system with quota
41 * tmpfiles' "v", "q" and "Q" will now create a plain directory
42 instead of a subvolume (even on a btrfs file system) if the
43 root directory is a plain directory, and not a
44 subvolume. This should simplify things with certain chroot()
45 environments which are not aware of the concept of btrfs
48 * systemd-detect-virt gained a new --chroot switch to detect
49 whether execution takes place in a chroot() environment.
51 * CPUAffinity= now takes CPU index ranges in addition to
54 * The various memory-related resource limit settings (such as
55 LimitAS=) now understand the usual K, M, G, ... suffixes to
56 the base of 1024 (IEC). Similar, the time-related resource
57 limit settings understand the usual min, h, day, ...
60 * There's a new system.conf setting DefaultTasksMax= to
61 control the default TasksMax= setting for services and
62 scopes running on the system. (TasksMax= is the primary
63 setting that exposes the "pids" cgroup controller on systemd
64 and was introduced in the previous systemd release.) The
65 setting now defaults to 512, which means services that are
66 not explicitly configured otherwise will only be able to
67 create 512 processes or threads at maximum, from this
68 version on. Note that this means that thread- or
69 process-heavy services might need to be reconfigured to set
70 TasksMax= to a higher value. It is sufficient to set
71 TasksMax= in these specific unit files to a higher value, or
72 even "infinity". Similar, there's now a logind.conf setting
73 UserTasksMax= that defaults to 4096 and limits the total
74 number of processes or tasks each user may own
75 concurrently. nspawn containers also have the TasksMax=
76 value set by default now, to 8192. Note that all of this
77 only has an effect if the "pids" cgroup controller is
78 enabled in the kernel. The general benefit of these changes
79 should be a more robust and safer system, that provides a
80 certain amount of per-service fork() bomb protection.
82 * systemd-nspawn gained the new --network-veth-extra= switch
83 to define additional and arbitrarily-named virtual Ethernet
84 links between the host and the container.
86 * A new service execution setting PassEnvironment= has been
87 added that allows importing select environment variables
88 from PID1's environment block into the environment block of
91 * Timer units gained support for a new RemainAfterElapse=
92 setting which takes a boolean argument. It defaults on on,
93 exposing behaviour unchanged to previous releases. If set to
94 off, timer units are unloaded after they elapsed if they
95 cannot elapse again. This is particularly useful for
96 transient timer units, which shall not stay around longer
97 than until they first elapse.
99 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
100 default now (the kernel default is 16). This is beneficial
101 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
102 allows substantially larger numbers of queued
103 datagrams. This should increase the capability of systemd to
104 parallelize boot-up, as logging and sd_notify() are unlikely
105 to stall execution anymore. If you need to change the value
106 from the new defaults, use the usual sysctl.d/ snippets.
108 * The compression framing format used by the journal or
109 coredump processing has changed to be in line with what the
110 official LZ4 tools generate. LZ4 compression support in
111 systemd was considered unsupported previously, as the format
112 was not compatible with the normal tools. With this release
113 this has changed now, and it is hence safe for downstream
114 distributions to turn it on. While not compressing as well
115 as the XZ, LZ4 is substantially faster, which makes
116 it a good default choice for the compression logic in the
117 journal and in coredump handling.
119 * Any reference to /etc/mtab has been dropped from
120 systemd. The file has been obsolete since a while, but
121 systemd refused to work on systems where it was incorrectly
122 set up (it should be a symlink or non-existent). Please make
123 sure to update to util-linux 2.27.1 or newer in conjunction
124 with this systemd release, which also drops any reference to
125 /etc/mtab. If you maintain a distribution make sure that no
126 software you package still references it, as this is a
127 likely source of bugs. There's also a glibc bug pending,
128 asking for removal of any reference to this obsolete file:
130 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
132 * Support for the ".snapshot" unit type has been removed. This
133 feature turned out to be little useful and little used, and
134 has now been removed from the core and from systemctl.
136 * The dependency types RequiresOverridable= and
137 RequisiteOverridable= have been removed from systemd. They
138 have been used only very sparingly to our knowledge and
139 other options that provide a similar effect (such as
140 systemctl --mode=ignore-dependencies) are much more useful
141 and commonly used. Moreover, they were only half-way
142 implemented as the option to control behaviour regarding
143 these dependencies was never added to systemctl. By removing
144 these dependency types the execution engine becomes a bit
145 simpler. Unit files that use these dependencies should be
146 changed to use the non-Overridable dependency types
147 instead. In fact, when parsing unit files with these
148 options, that's what systemd will automatically convert them
149 too, but it will also warn, asking users to fix the unit
150 files accordingly. Removal of these dependency types should
151 only affect a negligible number of unit files in the wild.
153 * Behaviour of networkd's IPForward= option changed
154 (again). It will no longer maintain a per-interface setting,
155 but propagate one way from interfaces where this is enabled
156 to the global kernel setting. The global setting will be
157 enabled when requested by a network that is set up, but
158 never be disabled again. This change was made to make sure
159 IPv4 and IPv6 behaviour regarding packet forwarding is
160 similar (as the Linux IPv6 stack does not support
161 per-interface control of this setting) and to minimize
164 * In unit files the behaviour of %u, %U, %h, %s has
165 changed. These specifiers will now unconditionally resolve
166 to the various user database fields of the user that the
167 systemd instance is running as, instead of the user
168 configured in the specific unit via User=. Note that this
169 effectively doesn't change much, as resolving of these
170 specifiers was already turned off in the --system instance
171 of systemd, as we cannot do NSS lookups from PID 1. In the
172 --user instance of systemd these specifiers where correctly
173 resolved, but hardly made any sense, since the user instance
174 lacks privileges to do user switches anyway, and User= is
175 hence useless. Morever, even in the --user instance of
176 systemd behaviour was awkward as it would only take settings
177 from User= assignment placed before the specifier into
178 account. In order to unify and simplify the logic around
179 this the specifiers will now always resolve to the
180 credentials of the user invoking the manager (which in case
181 of PID 1 is the root user).
183 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
184 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
185 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
186 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
187 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
188 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
189 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
190 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
191 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
192 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
193 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
194 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
195 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
196 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
199 -- Berlin, 2015-11-18
203 * systemd now depends on util-linux v2.27. More specifically,
204 the newly added mount monitor feature in libmount now
205 replaces systemd's former own implementation.
207 * libmount mandates /etc/mtab not to be regular file, and
208 systemd now enforces this condition at early boot.
209 /etc/mtab has been deprecated and warned about for a very
210 long time, so systems running systemd should already have
211 stopped having this file around as anything else than a
212 symlink to /proc/self/mounts.
214 * Support for the "pids" cgroup controller has been added. It
215 allows accounting the number of tasks in a cgroup and
216 enforcing limits on it. This adds two new setting
217 TasksAccounting= and TasksMax= to each unit, as well as a
218 global option DefaultTasksAccounting=.
220 * Support for the "net_cls" cgroup controller has been added.
221 It allows assigning a net class ID to each task in the
222 cgroup, which can then be used in firewall rules and traffic
223 shaping configurations. Note that the kernel netfilter net
224 class code does not currently work reliably for ingress
225 packets on unestablished sockets.
227 This adds a new config directive called NetClass= to CGroup
228 enabled units. Allowed values are positive numbers for fixed
229 assignments and "auto" for picking a free value
232 * 'systemctl is-system-running' now returns 'offline' if the
233 system is not booted with systemd. This command can now be
234 used as a substitute for 'systemd-notify --booted'.
236 * Watchdog timeouts have been increased to 3 minutes for all
237 in-tree service files. Apparently, disk IO issues are more
238 frequent than we hoped, and user reported >1 minute waiting
241 * 'machine-id-commit' functionality has been merged into
242 'machine-id-setup --commit'. The separate binary has been
245 * The WorkingDirectory= directive in unit files may now be set
246 to the special value '~'. In this case, the working
247 directory is set to the home directory of the user
250 * "machinectl shell" will now open the shell in the home
251 directory of the selected user by default.
253 * The CrashChVT= configuration file setting is renamed to
254 CrashChangeVT=, following our usual logic of not
255 abbreviating unnecessarily. The old directive is still
256 supported for compat reasons. Also, this directive now takes
257 an integer value between 1 and 63, or a boolean value. The
258 formerly supported '-1' value for disabling stays around for
261 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
262 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
263 RootDirectory= properties can now be set for transient
266 * The systemd-analyze tool gained a new "set-log-target" verb
267 to change the logging target the system manager logs to
268 dynamically during runtime. This is similar to how
269 "systemd-analyze set-log-level" already changes the log
272 * In nspawn /sys is now mounted as tmpfs, with only a selected
273 set of subdirectories mounted in from the real sysfs. This
274 enhances security slightly, and is useful for ensuring user
275 namespaces work correctly.
277 * Support for USB FunctionFS activation has been added. This
278 allows implementation of USB gadget services that are
279 activated as soon as they are requested, so that they don't
280 have to run continously, similar to classic socket
283 * The "systemctl exit" command now optionally takes an
284 additional parameter that sets the exit code to return from
285 the systemd manager when exiting. This is only relevant when
286 running the systemd user instance, or when running the
287 system instance in a container.
289 * sd-bus gained the new API calls sd_bus_path_encode_many()
290 and sd_bus_path_decode_many() that allow easy encoding and
291 decoding of multiple identifier strings inside a D-Bus
292 object path. Another new call sd_bus_default_flush_close()
293 has been added to flush and close per-thread default
296 * systemd-cgtop gained support for a -M/--machine= switch to
297 show the control groups within a certain container only.
299 * "systemctl kill" gained support for an optional --fail
300 switch. If specified the requested operation will fail of no
301 processes have been killed, because the unit had no
302 processes attached, or similar.
304 * A new systemd.crash_reboot=1 kernel command line option has
305 been added that triggers a reboot after crashing. This can
306 also be set through CrashReboot= in systemd.conf.
308 * The RuntimeDirectory= setting now understands unit
309 specifiers like %i or %f.
311 * A new (still internal) libary API sd-ipv4acd has been added,
312 that implements address conflict detection for IPv4. It's
313 based on code from sd-ipv4ll, and will be useful for
314 detecting DHCP address conflicts.
316 * File descriptors passed during socket activation may now be
317 named. A new API sd_listen_fds_with_names() is added to
318 access the names. The default names may be overridden,
319 either in the .socket file using the FileDescriptorName=
320 parameter, or by passing FDNAME= when storing the file
321 descriptors using sd_notify().
323 * systemd-networkd gained support for:
325 - Setting the IPv6 Router Advertisment settings via
326 IPv6AcceptRouterAdvertisements= in .network files.
328 - Configuring the HelloTimeSec=, MaxAgeSec= and
329 ForwardDelaySec= bridge parameters in .netdev files.
331 - Configuring PreferredSource= for static routes in
334 * The "ask-password" framework used to query for LUKS harddisk
335 passwords or SSL passwords during boot gained support for
336 caching passwords in the kernel keyring, if it is
337 available. This makes sure that the user only has to type in
338 a passphrase once if there are multiple objects to unlock
339 with the same one. Previously, such password caching was
340 available only when Plymouth was used; this moves the
341 caching logic into the systemd codebase itself. The
342 "systemd-ask-password" utility gained a new --keyname=
343 switch to control which kernel keyring key to use for
344 caching a password in. This functionality is also useful for
345 enabling display managers such as gdm to automatically
346 unlock the user's GNOME keyring if its passphrase, the
347 user's password and the harddisk password are the same, if
348 gdm-autologin is used.
350 * When downloading tar or raw images using "machinectl
351 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
352 file is now also downloaded, if it is available and stored
353 next to the image file.
355 * Units of type ".socket" gained a new boolean setting
356 Writable= which is only useful in conjunction with
357 ListenSpecial=. If true, enables opening the specified
358 special file in O_RDWR mode rather than O_RDONLY mode.
360 * systemd-rfkill has been reworked to become a singleton
361 service that is activated through /dev/rfkill on each rfkill
362 state change and saves the settings to disk. This way,
363 systemd-rfkill is now compatible with devices that exist
364 only intermittendly, and even restores state if the previous
365 system shutdown was abrupt rather than clean.
367 * The journal daemon gained support for vacuuming old journal
368 files controlled by the number of files that shall remain,
369 in addition to the already existing control by size and by
370 date. This is useful as journal interleaving performance
371 degrades with too many seperate journal files, and allows
372 putting an effective limit on them. The new setting defaults
373 to 100, but this may be changed by setting SystemMaxFiles=
374 and RuntimeMaxFiles= in journald.conf. Also, the
375 "journalctl" tool gained the new --vacuum-files= switch to
376 manually vacuum journal files to leave only the specified
377 number of files in place.
379 * udev will now create /dev/disk/by-path links for ATA devices
380 on kernels where that is supported.
382 * Galician, Serbian, Turkish and Korean translations were added.
384 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
385 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
386 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
387 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
388 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
389 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
390 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
391 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
392 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
393 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
394 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
395 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
396 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
397 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
398 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
399 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
400 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
401 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
403 -- Berlin, 2015-10-07
407 * The DHCP implementation of systemd-networkd gained a set of
410 - The DHCP server now supports emitting DNS and NTP
411 information. It may be enabled and configured via
412 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
413 and NTP information is enabled, but no servers are
414 configured, the corresponding uplink information (if there
415 is any) is propagated.
417 - Server and client now support transmission and reception
418 of timezone information. It can be configured via the
419 newly introduced network options UseTimezone=,
420 EmitTimezone=, and Timezone=. Transmission of timezone
421 information is enabled between host and containers by
422 default now: the container will change its local timezone
423 to what the host has set.
425 - Lease timeouts can now be configured via
426 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
428 - The DHCP server improved on the stability of
429 leases. Clients are more likely to get the same lease
430 information back, even if the server loses state.
432 - The DHCP server supports two new configuration options to
433 control the lease address pool metrics, PoolOffset= and
436 * The encapsulation limit of tunnels in systemd-networkd may
437 now be configured via 'EncapsulationLimit='. It allows
438 modifying the maximum additional levels of encapsulation
439 that are permitted to be prepended to a packet.
441 * systemd now supports the concept of user buses replacing
442 session buses, if used with dbus-1.10 (and enabled via dbus
443 --enable-user-session). It previously only supported this on
444 kdbus-enabled systems, and this release expands this to
445 'dbus-daemon' systems.
447 * systemd-networkd now supports predictable interface names
450 * systemd now optionally supports the new Linux kernel
451 "unified" control group hierarchy. If enabled via the kernel
452 command-line option 'systemd.unified_cgroup_hierarchy=1',
453 systemd will try to mount the unified cgroup hierarchy
454 directly on /sys/fs/cgroup. If not enabled, or not
455 available, systemd will fall back to the legacy cgroup
456 hierarchy setup, as before. Host system and containers can
457 mix and match legacy and unified hierarchies as they
458 wish. nspawn understands the $UNIFIED_CROUP_HIERARCHY
459 environment variable to individually select the hierarchy to
460 use for executed containers. By default, nspawn will use the
461 unified hierarchy for the containers if the host uses the
462 unified hierarchy, and the legacy hierarchy otherwise.
463 Please note that at this point the unified hierarchy is an
464 experimental kernel feature and is likely to change in one
465 of the next kernel releases. Therefore, it should not be
466 enabled by default in downstream distributions yet. The
467 minimum required kernel version for the unified hierarchy to
468 work is 4.2. Note that when the unified hierarchy is used
469 for the first time delegated access to controllers is
470 safe. Because of this systemd-nspawn containers will get
471 access to controllers now, as will systemd user
472 sessions. This means containers and user sessions may now
473 manage their own resources, partitioning up what the system
476 * A new special scope unit "init.scope" has been introduced
477 that encapsulates PID 1 of the system. It may be used to
478 determine resource usage and enforce resource limits on PID
479 1 itself. PID 1 hence moved out of the root of the control
482 * The cgtop tool gained support for filtering out kernel
483 threads when counting tasks in a control group. Also, the
484 count of processes is now recursively summed up by
485 default. Two options -k and --recursive= have been added to
486 revert to old behaviour. The tool has also been updated to
487 work correctly in containers now.
489 * systemd-nspawn's --bind= and --bind-ro= options have been
490 extended to allow creation of non-recursive bind mounts.
492 * libsystemd gained two new calls sd_pid_get_cgroup() and
493 sd_peer_get_cgroup() which return the control group path of
494 a process or peer of a connected AF_UNIX socket. This
495 function call is particularly useful when implementing
496 delegated subtrees support in the control group hierarchy.
498 * The "sd-event" event loop API of libsystemd now supports
499 correct dequeuing of real-time signals, without losing
502 * When systemd requests a PolicyKit decision when managing
503 units it will now add additional fields to the request,
504 including unit name and desired operation. This enables more
505 powerful PolicyKit policies, that make decisions depending
508 * nspawn learnt support for .nspawn settings files, that may
509 accompany the image files or directories of containers, and
510 may contain additional settings for the container. This is
511 an alternative to configuring container parameters via the
514 Contributions from: Cristian Rodríguez, Daniel Mack, David
515 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
516 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
517 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
518 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
519 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
520 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
521 Andersen, Tom Gundersen, Torstein Husebø
523 -- Berlin, 2015-09-08
527 * machinectl gained a new verb 'shell' which opens a fresh
528 shell on the target container or the host. It is similar to
529 the existing 'login' command of machinectl, but spawns the
530 shell directly without prompting for username or
531 password. The pseudo machine '.host' now refers to the local
532 host and is used by default. Hence, 'machinectl shell' can
533 be used as replacement for 'su -' which spawns a session as
534 a fresh systemd unit in a way that is fully isolated from
535 the originating session.
537 * systemd-networkd learned to cope with private-zone DHCP
538 options and allows other programs to query the values.
540 * SELinux access control when enabling/disabling units is no
541 longer enforced with this release. The previous
542 implementation was incorrect, and a new corrected
543 implementation is not yet available. As unit file operations
544 are still protected via PolicyKit and D-Bus policy this is
545 not a security problem. Yet, distributions which care about
546 optimal SELinux support should probably not stabilize on
549 * sd-bus gained support for matches of type "arg0has=", that
550 test for membership of strings in string arrays sent in bus
553 * systemd-resolved now dumps the contents of its DNS and LLMNR
554 caches to the logs on reception of the SIGUSR1 signal. This
555 is useful to debug DNS behaviour.
557 * The coredumpctl tool gained a new --directory= option to
558 operate on journal files in a specific directory.
560 * "systemctl reboot" and related commands gained a new
561 "--message=" option which may be used to set a free-text
562 wall message when shutting down or rebooting the
563 system. This message is also logged, which is useful for
564 figuring out the reason for a reboot or shutdown a
567 * The "systemd-resolve-host" tool's -i switch now takes
568 network interface numbers as alternative to interface names.
570 * A new unit file setting for services has been introduced:
571 UtmpMode= allows configuration of how precisely systemd
572 handles utmp and wtmp entries for the service if this is
573 enabled. This allows writing services that appear similar to
574 user sessions in the output of the "w", "who", "last" and
577 * systemd-resolved will now locally synthesize DNS resource
578 records for the "localhost" and "gateway" domains as well as
579 the local hostname. This should ensure that clients querying
580 RRs via resolved will get similar results as those going via
581 NSS, if nss-myhostname is enabled.
583 Contributions from: Alastair Hughes, Alex Crawford, Daniel
584 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
585 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
586 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
587 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
588 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
589 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
590 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
591 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
592 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
593 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
594 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
596 -- Berlin, 2015-08-27
600 * The systemd-efi-boot-generator functionality was merged into
601 systemd-gpt-auto-generator.
603 * systemd-networkd now supports Group Policy for vxlan
604 devices. It can be enabled via the new boolean configuration
605 option called 'GroupPolicyExtension='.
607 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
608 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
609 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
611 -- Berlin, 2015-07-31
615 * The python-systemd code has been removed from the systemd repository.
616 A new repository has been created which accommodates the code from
617 now on, and we kindly ask distributions to create a separate package
618 for this: https://github.com/systemd/python-systemd
620 * The systemd daemon will now reload its main configuration
621 (/etc/systemd/system.conf) on daemon-reload.
623 * sd-dhcp now exposes vendor specific extensions via
624 sd_dhcp_lease_get_vendor_specific().
626 * systemd-networkd gained a number of new configuration options.
628 - A new boolean configuration option for TAP devices called
629 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
630 device, thus allowing to send and receive GSO packets.
632 - A new tunnel configuration option called 'CopyDSCP='.
633 If enabled, the DSCP field of ip6 tunnels is copied into the
636 - A set of boolean bridge configuration options were added.
637 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
638 and 'UnicastFlood=' are now parsed by networkd and applied to the
639 respective bridge link device via the respective IFLA_BRPORT_*
642 - A new string configuration option to override the hostname sent
643 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
644 is true, networkd will use the configured hostname instead of the
645 system hostname when sending DHCP requests.
647 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
648 networkd will configure the IPv6 flow-label of the tunnel device
649 according to RFC2460.
651 - The 'macvtap' virtual network devices are now supported, similar to
652 the already supported 'macvlan' devices.
654 * systemd-resolved now implements RFC5452 to improve resilience against
655 cache poisoning. Additionally, source port randomization is enabled
656 by default to further protect against DNS spoofing attacks.
658 * nss-mymachines now supports translating UIDs and GIDs of running
659 containers with user-namespaces enabled. If a container 'foo'
660 translates a host uid 'UID' to the container uid 'TUID', then
661 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
662 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
663 mapped as 'vg-foo-TGID'.
665 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
666 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
667 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
668 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
669 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
670 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
671 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
672 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
673 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
674 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
676 -- Berlin, 2015-07-29
680 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
681 There are no known issues with current sysfs, and udev does not need
682 or should be used to work around such bugs.
684 * udev does no longer enable USB HID power management. Several reports
685 indicate, that some devices cannot handle that setting.
687 * The udev accelerometer helper was removed. The functionality
688 is now fully included in iio-sensor-proxy. But this means,
689 older iio-sensor-proxy versions will no longer provide
690 accelerometer/orientation data with this systemd version.
691 Please upgrade iio-sensor-proxy to version 1.0.
693 * networkd gained a new configuration option IPv6PrivacyExtensions=
694 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
695 for Stateless Address") on selected networks.
697 * For the sake of fewer build-time dependencies and less code in the
698 main repository, the python bindings are about to be removed in the
699 next release. A new repository has been created which accommodates
700 the code from now on, and we kindly ask distributions to create a
701 separate package for this. The removal will take place in v223.
703 https://github.com/systemd/python-systemd
705 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
706 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
707 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
708 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
709 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
710 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
711 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
712 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
713 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
714 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
716 -- Berlin, 2015-07-07
720 * The sd-bus.h and sd-event.h APIs have now been declared
721 stable and have been added to the official interface of
722 libsystemd.so. sd-bus implements an alternative D-Bus client
723 library, that is relatively easy to use, very efficient and
724 supports both classic D-Bus as well as kdbus as transport
725 backend. sd-event is a generic event loop abstraction that
726 is built around Linux epoll, but adds features such as event
727 prioritization or efficient timer handling. Both APIs are good
728 choices for C programs looking for a bus and/or event loop
729 implementation that is minimal and does not have to be
730 portable to other kernels.
732 * kdbus support is no longer compile-time optional. It is now
733 always built-in. However, it can still be disabled at
734 runtime using the kdbus=0 kernel command line setting, and
735 that setting may be changed to default to off, by specifying
736 --disable-kdbus at build-time. Note though that the kernel
737 command line setting has no effect if the kdbus.ko kernel
738 module is not installed, in which case kdbus is (obviously)
739 also disabled. We encourage all downstream distributions to
740 begin testing kdbus by adding it to the kernel images in the
741 development distributions, and leaving kdbus support in
744 * The minimal required util-linux version has been bumped to
747 * Support for chkconfig (--enable-chkconfig) was removed in
748 favor of calling an abstraction tool
749 /lib/systemd/systemd-sysv-install. This needs to be
750 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
751 in README for details.
753 * If there's a systemd unit and a SysV init script for the
754 same service name, and the user executes "systemctl enable"
755 for it (or a related call), then this will now enable both
756 (or execute the related operation on both), not just the
759 * The libudev API documentation has been converted from gtkdoc
762 * gudev has been removed from the systemd tree, it is now an
765 * The systemd-cgtop tool learnt a new --raw switch to generate
766 "raw" (machine parsable) output.
768 * networkd's IPForwarding= .network file setting learnt the
769 new setting "kernel", which ensures that networkd does not
770 change the IP forwarding sysctl from the default kernel
773 * The systemd-logind bus API now exposes a new boolean
774 property "Docked" that reports whether logind considers the
775 system "docked", i.e. connected to a docking station or not.
777 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
778 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
779 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
780 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
781 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
782 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
783 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
784 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
785 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
786 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
787 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
788 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
789 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
790 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
791 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
792 Fink, Zbigniew Jędrzejewski-Szmek
794 -- Berlin, 2015-06-19
798 * The gudev library has been extracted into a separate repository
799 available at: https://git.gnome.org/browse/libgudev/
800 It is now managed as part of the Gnome project. Distributions
801 are recommended to pass --disable-gudev to systemd and use
802 gudev from the Gnome project instead. gudev is still included
803 in systemd, for now. It will be removed soon, though. Please
804 also see the announcement-thread on systemd-devel:
805 http://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
807 * systemd now exposes a CPUUsageNSec= property for each
808 service unit on the bus, that contains the overall consumed
809 CPU time of a service (the sum of what each process of the
810 service consumed). This value is only available if
811 CPUAccounting= is turned on for a service, and is then shown
812 in the "systemctl status" output.
814 * Support for configuring alternative mappings of the old SysV
815 runlevels to systemd targets has been removed. They are now
816 hardcoded in a way that runlevels 2, 3, 4 all map to
817 multi-user.target and 5 to graphical.target (which
818 previously was already the default behaviour).
820 * The auto-mounter logic gained support for mount point
821 expiry, using a new TimeoutIdleSec= setting in .automount
822 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
824 * The EFI System Partition (ESP) as mounted to /boot by
825 systemd-efi-boot-generator will now be unmounted
826 automatically after 2 minutes of not being used. This should
827 minimize the risk of ESP corruptions.
829 * New /etc/fstab options x-systemd.requires= and
830 x-systemd.requires-mounts-for= are now supported to express
831 additional dependencies for mounts. This is useful for
832 journalling file systems that support external journal
833 devices or overlay file systems that require underlying file
834 systems to be mounted.
836 * systemd does not support direct live-upgrades (via systemctl
837 daemon-reexec) from versions older than v44 anymore. As no
838 distribution we are aware of shipped such old versions in a
839 stable release this should not be problematic.
841 * When systemd forks off a new per-connection service instance
842 it will now set the $REMOTE_ADDR environment variable to the
843 remote IP address, and $REMOTE_PORT environment variable to
844 the remote IP port. This behaviour is similar to the
845 corresponding environment variables defined by CGI.
847 * systemd-networkd gained support for uplink failure
848 detection. The BindCarrier= option allows binding interface
849 configuration dynamically to the link sense of other
850 interfaces. This is useful to achieve behaviour like in
853 * systemd-networkd gained support for configuring the DHCP
854 client identifier to use when requesting leases.
856 * systemd-networkd now has a per-network UseNTP= option to
857 configure whether NTP server information acquired via DHCP
858 is passed on to services like systemd-timesyncd.
860 * systemd-networkd gained support for vti6 tunnels.
862 * Note that systemd-networkd manages the sysctl variable
863 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
864 it is configured for since v219. The variable controls IP
865 forwarding, and is a per-interface alternative to the global
866 /proc/sys/net/ipv[46]/ip_forward. This setting is
867 configurable in the IPForward= option, which defaults to
868 "no". This means if networkd is used for an interface it is
869 no longer sufficient to set the global sysctl option to turn
870 on IP forwarding! Instead, the .network file option
871 IPForward= needs to be turned on! Note that the
872 implementation of this behaviour was broken in v219 and has
875 * Many bonding and vxlan options are now configurable in
878 * systemd-nspawn gained a new --property= setting to set unit
879 properties for the container scope. This is useful for
880 setting resource parameters (e.g "CPUShares=500") on
881 containers started from the command line.
883 * systemd-nspawn gained a new --private-users= switch to make
884 use of user namespacing available on recent Linux kernels.
886 * systemd-nspawn may now be called as part of a shell pipeline
887 in which case the pipes used for stdin and stdout are passed
888 directly to the process invoked in the container, without
889 indirection via a pseudo tty.
891 * systemd-nspawn gained a new switch to control the UNIX
892 signal to use when killing the init process of the container
895 * systemd-nspawn gained a new --overlay= switch for mounting
896 overlay file systems into the container using the new kernel
899 * When a container image is imported via systemd-importd and
900 the host file system is not btrfs, a loopback block device
901 file is created in /var/lib/machines.raw with a btrfs file
902 system inside. It is then mounted to /var/lib/machines to
903 enable btrfs features for container management. The loopback
904 file and btrfs file system is grown as needed when container
905 images are imported via systemd-importd.
907 * systemd-machined/systemd-importd gained support for btrfs
908 quota, to enforce container disk space limits on disk. This
909 is exposed in "machinectl set-limit".
911 * systemd-importd now can import containers from local .tar,
912 .raw and .qcow2 images, and export them to .tar and .raw. It
913 can also import dkr v2 images now from the network (on top
916 * systemd-importd gained support for verifying downloaded
917 images with gpg2 (previously only gpg1 was supported).
919 * systemd-machined, systemd-logind, systemd: most bus calls
920 are now accessible to unprivileged processes via
921 PolicyKit. Also, systemd-logind will now allow users to kill
922 their own sessions without further privileges or
925 * systemd-shutdownd has been removed. This service was
926 previously responsible for implementing scheduled shutdowns
927 as exposed in /usr/bin/shutdown's time parameter. This
928 functionality has now been moved into systemd-logind and is
929 accessible via a bus interface.
931 * "systemctl reboot" gained a new switch --firmware-setup that
932 can be used to reboot into the EFI firmware setup, if that
933 is available. systemd-logind now exposes an API on the bus
934 to trigger such reboots, in case graphical desktop UIs want
935 to cover this functionality.
937 * "systemctl enable", "systemctl disable" and "systemctl mask"
938 now support a new "--now" switch. If specified the units
939 that are enabled will also be started, and the ones
940 disabled/masked also stopped.
942 * The Gummiboot EFI boot loader tool has been merged into
943 systemd, and renamed to "systemd-boot". The bootctl tool has been
944 updated to support systemd-boot.
946 * An EFI kernel stub has been added that may be used to create
947 kernel EFI binaries that contain not only the actual kernel,
948 but also an initrd, boot splash, command line and OS release
949 information. This combined binary can then be signed as a
950 single image, so that the firmware can verify it all in one
951 step. systemd-boot has special support for EFI binaries created
952 like this and can extract OS release information from them
953 and show them in the boot menu. This functionality is useful
954 to implement cryptographically verified boot schemes.
956 * Optional support has been added to systemd-fsck to pass
957 fsck's progress report to an AF_UNIX socket in the file
960 * udev will no longer create device symlinks for all block
961 devices by default. A blacklist for excluding special block
962 devices from this logic has been turned into a whitelist
963 that requires picking block devices explicitly that require
966 * A new (currently still internal) API sd-device.h has been
967 added to libsystemd. This modernized API is supposed to
968 replace libudev eventually. In fact, already much of libudev
969 is now just a wrapper around sd-device.h.
971 * A new hwdb database for storing metadata about pointing
972 stick devices has been added.
974 * systemd-tmpfiles gained support for setting file attributes
975 similar to the "chattr" tool with new 'h' and 'H' lines.
977 * systemd-journald will no longer unconditionally set the
978 btrfs NOCOW flag on new journal files. This is instead done
979 with tmpfiles snippet using the new 'h' line type. This
980 allows easy disabling of this logic, by masking the
981 journal-nocow.conf tmpfiles file.
983 * systemd-journald will now translate audit message types to
984 human readable identifiers when writing them to the
985 journal. This should improve readability of audit messages.
987 * The LUKS logic gained support for the offset= and skip=
988 options in /etc/crypttab, as previously implemented by
991 * /usr/lib/os-release gained a new optional field VARIANT= for
992 distributions that support multiple variants (such as a
993 desktop edition, a server edition, ...)
995 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
996 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
997 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
998 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
999 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
1000 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
1001 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
1002 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
1003 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
1004 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
1005 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
1006 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
1007 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
1008 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
1009 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
1010 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
1011 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
1012 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
1013 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
1014 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
1015 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
1016 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
1017 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
1018 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
1019 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
1020 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
1021 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
1023 -- Berlin, 2015-05-22
1027 * Introduce a new API "sd-hwdb.h" for querying the hardware
1028 metadata database. With this minimal interface one can query
1029 and enumerate the udev hwdb, decoupled from the old libudev
1030 library. libudev's interface for this is now only a wrapper
1031 around sd-hwdb. A new tool systemd-hwdb has been added to
1032 interface with and update the database.
1034 * When any of systemd's tools copies files (for example due to
1035 tmpfiles' C lines) a btrfs reflink will attempted first,
1036 before bytewise copying is done.
1038 * systemd-nspawn gained a new --ephemeral switch. When
1039 specified a btrfs snapshot is taken of the container's root
1040 directory, and immediately removed when the container
1041 terminates again. Thus, a container can be started whose
1042 changes never alter the container's root directory, and are
1043 lost on container termination. This switch can also be used
1044 for starting a container off the root file system of the
1045 host without affecting the host OS. This switch is only
1046 available on btrfs file systems.
1048 * systemd-nspawn gained a new --template= switch. It takes the
1049 path to a container tree to use as template for the tree
1050 specified via --directory=, should that directory be
1051 missing. This allows instantiating containers dynamically,
1052 on first run. This switch is only available on btrfs file
1055 * When a .mount unit refers to a mount point on which multiple
1056 mounts are stacked, and the .mount unit is stopped all of
1057 the stacked mount points will now be unmounted until no
1058 mount point remains.
1060 * systemd now has an explicit notion of supported and
1061 unsupported unit types. Jobs enqueued for unsupported unit
1062 types will now fail with an "unsupported" error code. More
1063 specifically .swap, .automount and .device units are not
1064 supported in containers, .busname units are not supported on
1065 non-kdbus systems. .swap and .automount are also not
1066 supported if their respective kernel compile time options
1069 * machinectl gained support for two new "copy-from" and
1070 "copy-to" commands for copying files from a running
1071 container to the host or vice versa.
1073 * machinectl gained support for a new "bind" command to bind
1074 mount host directories into local containers. This is
1075 currently only supported for nspawn containers.
1077 * networkd gained support for configuring bridge forwarding
1078 database entries (fdb) from .network files.
1080 * A new tiny daemon "systemd-importd" has been added that can
1081 download container images in tar, raw, qcow2 or dkr formats,
1082 and make them available locally in /var/lib/machines, so
1083 that they can run as nspawn containers. The daemon can GPG
1084 verify the downloads (not supported for dkr, since it has no
1085 provisions for verifying downloads). It will transparently
1086 decompress bz2, xz, gzip compressed downloads if necessary,
1087 and restore sparse files on disk. The daemon uses privilege
1088 separation to ensure the actual download logic runs with
1089 fewer privileges than the daemon itself. machinectl has
1090 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
1091 make the functionality of importd available to the
1092 user. With this in place the Fedora and Ubuntu "Cloud"
1093 images can be downloaded and booted as containers unmodified
1094 (the Fedora images lack the appropriate GPG signature files
1095 currently, so they cannot be verified, but this will change
1096 soon, hopefully). Note that downloading images is currently
1097 only fully supported on btrfs.
1099 * machinectl is now able to list container images found in
1100 /var/lib/machines, along with some metadata about sizes of
1101 disk and similar. If the directory is located on btrfs and
1102 quota is enabled, this includes quota display. A new command
1103 "image-status" has been added that shows additional
1104 information about images.
1106 * machinectl is now able to clone container images
1107 efficiently, if the underlying file system (btrfs) supports
1108 it, with the new "machinectl list-images" command. It also
1109 gained commands for renaming and removing images, as well as
1110 marking them read-only or read-write (supported also on
1111 legacy file systems).
1113 * networkd gained support for collecting LLDP network
1114 announcements, from hardware that supports this. This is
1115 shown in networkctl output.
1117 * systemd-run gained support for a new -t (--pty) switch for
1118 invoking a binary on a pty whose input and output is
1119 connected to the invoking terminal. This allows executing
1120 processes as system services while interactively
1121 communicating with them via the terminal. Most interestingly
1122 this is supported across container boundaries. Invoking
1123 "systemd-run -t /bin/bash" is an alternative to running a
1124 full login session, the difference being that the former
1125 will not register a session, nor go through the PAM session
1128 * tmpfiles gained support for a new "v" line type for creating
1129 btrfs subvolumes. If the underlying file system is a legacy
1130 file system, this automatically degrades to creating a
1131 normal directory. Among others /var/lib/machines is now
1132 created like this at boot, should it be missing.
1134 * The directory /var/lib/containers/ has been deprecated and
1135 been replaced by /var/lib/machines. The term "machines" has
1136 been used in the systemd context as generic term for both
1137 VMs and containers, and hence appears more appropriate for
1138 this, as the directory can also contain raw images bootable
1141 * systemd-nspawn when invoked with -M but without --directory=
1142 or --image= is now capable of searching for the container
1143 root directory, subvolume or disk image automatically, in
1144 /var/lib/machines. systemd-nspawn@.service has been updated
1145 to make use of this, thus allowing it to be used for raw
1148 * A new machines.target unit has been introduced that is
1149 supposed to group all containers/VMs invoked as services on
1150 the system. systemd-nspawn@.service has been updated to
1151 integrate with that.
1153 * machinectl gained a new "start" command, for invoking a
1154 container as a service. "machinectl start foo" is mostly
1155 equivalent to "systemctl start systemd-nspawn@foo.service",
1156 but handles escaping in a nicer way.
1158 * systemd-nspawn will now mount most of the cgroupfs tree
1159 read-only into each container, with the exception of the
1160 container's own subtree in the name=systemd hierarchy.
1162 * journald now sets the special FS_NOCOW file flag for its
1163 journal files. This should improve performance on btrfs, by
1164 avoiding heavy fragmentation when journald's write-pattern
1165 is used on COW file systems. It degrades btrfs' data
1166 integrity guarantees for the files to the same levels as for
1167 ext3/ext4 however. This should be OK though as journald does
1168 its own data integrity checks and all its objects are
1169 checksummed on disk. Also, journald should handle btrfs disk
1170 full events a lot more gracefully now, by processing SIGBUS
1171 errors, and not relying on fallocate() anymore.
1173 * When journald detects that journal files it is writing to
1174 have been deleted it will immediately start new journal
1177 * systemd now provides a way to store file descriptors
1178 per-service in PID 1.This is useful for daemons to ensure
1179 that fds they require are not lost during a daemon
1180 restart. The fds are passed to the daemon on the next
1181 invocation in the same way socket activation fds are
1182 passed. This is now used by journald to ensure that the
1183 various sockets connected to all the system's stdout/stderr
1184 are not lost when journald is restarted. File descriptors
1185 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
1186 an extension to sd_notify(). Note that a limit is enforced
1187 on the number of fds a service can store in PID 1, and it
1188 defaults to 0, so that no fds may be stored, unless this is
1189 explicitly turned on.
1191 * The default TERM variable to use for units connected to a
1192 terminal, when no other value is explicitly is set is now
1193 vt220 rather than vt102. This should be fairly safe still,
1194 but allows PgUp/PgDn work.
1196 * The /etc/crypttab option header= as known from Debian is now
1199 * "loginctl user-status" and "loginctl session-status" will
1200 now show the last 10 lines of log messages of the
1201 user/session following the status output. Similar,
1202 "machinectl status" will show the last 10 log lines
1203 associated with a virtual machine or container
1204 service. (Note that this is usually not the log messages
1205 done in the VM/container itself, but simply what the
1206 container manager logs. For nspawn this includes all console
1209 * "loginctl session-status" without further argument will now
1210 show the status of the session of the caller. Similar,
1211 "lock-session", "unlock-session", "activate",
1212 "enable-linger", "disable-linger" may now be called without
1213 session/user parameter in which case they apply to the
1214 caller's session/user.
1216 * An X11 session scriptlet is now shipped that uploads
1217 $DISPLAY and $XAUTHORITY into the environment of the systemd
1218 --user daemon if a session begins. This should improve
1219 compatibility with X11 enabled applications run as systemd
1222 * Generators are now subject to masking via /etc and /run, the
1223 same way as unit files.
1225 * networkd .network files gained support for configuring
1226 per-link IPv4/IPv6 packet forwarding as well as IPv4
1227 masquerading. This is by default turned on for veth links to
1228 containers, as registered by systemd-nspawn. This means that
1229 nspawn containers run with --network-veth will now get
1230 automatic routed access to the host's networks without any
1231 further configuration or setup, as long as networkd runs on
1234 * systemd-nspawn gained the --port= (-p) switch to expose TCP
1235 or UDP posts of a container on the host. With this in place
1236 it is possible to run containers with private veth links
1237 (--network-veth), and have their functionality exposed on
1238 the host as if their services were running directly on the
1241 * systemd-nspawn's --network-veth switch now gained a short
1242 version "-n", since with the changes above it is now truly
1243 useful out-of-the-box. The systemd-nspawn@.service has been
1244 updated to make use of it too by default.
1246 * systemd-nspawn will now maintain a per-image R/W lock, to
1247 ensure that the same image is not started more than once
1248 writable. (It's OK to run an image multiple times
1249 simultaneously in read-only mode.)
1251 * systemd-nspawn's --image= option is now capable of
1252 dissecting and booting MBR and GPT disk images that contain
1253 only a single active Linux partition. Previously it
1254 supported only GPT disk images with proper GPT type
1255 IDs. This allows running cloud images from major
1256 distributions directly with systemd-nspawn, without
1259 * In addition to collecting mouse dpi data in the udev
1260 hardware database, there's now support for collecting angle
1261 information for mouse scroll wheels. The database is
1262 supposed to guarantee similar scrolling behavior on mice
1263 that it knows about. There's also support for collecting
1264 information about Touchpad types.
1266 * udev's input_id built-in will now also collect touch screen
1267 dimension data and attach it to probed devices.
1269 * /etc/os-release gained support for a Distribution Privacy
1272 * networkd gained support for creating "ipvlan", "gretap",
1273 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
1275 * systemd-tmpfiles gained support for "a" lines for setting
1278 * systemd-nspawn will now mount /tmp in the container to
1279 tmpfs, automatically.
1281 * systemd now exposes the memory.usage_in_bytes cgroup
1282 attribute and shows it for each service in the "systemctl
1283 status" output, if available.
1285 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
1286 immediate reboot is triggered. This useful if shutdown is
1287 hung and is unable to complete, to expedite the
1288 operation. Note that this kind of reboot will still unmount
1289 all file systems, and hence should not result in fsck being
1292 * A .device unit for an optical block device will now be
1293 considered active only when a medium is in the drive. Also,
1294 mount units are now bound to their backing devices thus
1295 triggering automatic unmounting when devices become
1296 unavailable. With this in place systemd will now
1297 automatically unmount left-over mounts when a CD-ROM is
1298 ejected or an USB stick is yanked from the system.
1300 * networkd-wait-online now has support for waiting for
1301 specific interfaces only (with globbing), and for giving up
1302 after a configurable timeout.
1304 * networkd now exits when idle. It will be automatically
1305 restarted as soon as interfaces show up, are removed or
1306 change state. networkd will stay around as long as there is
1307 at least one DHCP state machine or similar around, that keep
1310 * networkd may now configure IPv6 link-local addressing in
1311 addition to IPv4 link-local addressing.
1313 * The IPv6 "token" for use in SLAAC may now be configured for
1314 each .network interface in networkd.
1316 * Routes configured with networkd may now be assigned a scope
1319 * networkd's [Match] sections now support globbing and lists
1320 of multiple space-separated matches per item.
1322 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
1323 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
1324 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
1325 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
1326 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
1327 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
1328 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
1329 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
1330 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
1331 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
1332 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
1333 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
1334 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
1335 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
1336 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
1337 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
1338 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
1339 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
1340 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
1341 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
1342 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
1343 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
1344 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
1345 Hoffmann, Zbigniew Jędrzejewski-Szmek
1347 -- Berlin, 2015-02-16
1351 * When querying unit file enablement status (for example via
1352 "systemctl is-enabled"), a new state "indirect" is now known
1353 which indicates that a unit might not be enabled itself, but
1354 another unit listed in its Also= setting might be.
1356 * Similar to the various existing ConditionXYZ= settings for
1357 units, there are now matching AssertXYZ= settings. While
1358 failing conditions cause a unit to be skipped, but its job
1359 to succeed, failing assertions declared like this will cause
1360 a unit start operation and its job to fail.
1362 * hostnamed now knows a new chassis type "embedded".
1364 * systemctl gained a new "edit" command. When used on a unit
1365 file, this allows extending unit files with .d/ drop-in
1366 configuration snippets or editing the full file (after
1367 copying it from /usr/lib to /etc). This will invoke the
1368 user's editor (as configured with $EDITOR), and reload the
1369 modified configuration after editing.
1371 * "systemctl status" now shows the suggested enablement state
1372 for a unit, as declared in the (usually vendor-supplied)
1373 system preset files.
1375 * nss-myhostname will now resolve the single-label host name
1376 "gateway" to the locally configured default IP routing
1377 gateways, ordered by their metrics. This assigns a stable
1378 name to the used gateways, regardless which ones are
1379 currently configured. Note that the name will only be
1380 resolved after all other name sources (if nss-myhostname is
1381 configured properly) and should hence not negatively impact
1382 systems that use the single-label host name "gateway" in
1385 * systemd-inhibit now allows filtering by mode when listing
1388 * Scope and service units gained a new "Delegate" boolean
1389 property, which, when set, allows processes running inside the
1390 unit to further partition resources. This is primarily
1391 useful for systemd user instances as well as container
1394 * journald will now pick up audit messages directly from
1395 the kernel, and log them like any other log message. The
1396 audit fields are split up and fully indexed. This means that
1397 journalctl in many ways is now a (nicer!) alternative to
1398 ausearch, the traditional audit client. Note that this
1399 implements only a minimal audit client. If you want the
1400 special audit modes like reboot-on-log-overflow, please use
1401 the traditional auditd instead, which can be used in
1402 parallel to journald.
1404 * The ConditionSecurity= unit file option now understands the
1405 special string "audit" to check whether auditing is
1408 * journalctl gained two new commands --vacuum-size= and
1409 --vacuum-time= to delete old journal files until the
1410 remaining ones take up no more than the specified size on disk,
1411 or are not older than the specified time.
1413 * A new, native PPPoE library has been added to sd-network,
1414 systemd's library of light-weight networking protocols. This
1415 library will be used in a future version of networkd to
1416 enable PPPoE communication without an external pppd daemon.
1418 * The busctl tool now understands a new "capture" verb that
1419 works similar to "monitor", but writes a packet capture
1420 trace to STDOUT that can be redirected to a file which is
1421 compatible with libcap's capture file format. This can then
1422 be loaded in Wireshark and similar tools to inspect bus
1425 * The busctl tool now understands a new "tree" verb that shows
1426 the object trees of a specific service on the bus, or of all
1429 * The busctl tool now understands a new "introspect" verb that
1430 shows all interfaces and members of objects on the bus,
1431 including their signature and values. This is particularly
1432 useful to get more information about bus objects shown by
1433 the new "busctl tree" command.
1435 * The busctl tool now understands new verbs "call",
1436 "set-property" and "get-property" for invoking bus method
1437 calls, setting and getting bus object properties in a
1440 * busctl gained a new --augment-creds= argument that controls
1441 whether the tool shall augment credential information it
1442 gets from the bus with data from /proc, in a possibly
1445 * nspawn's --link-journal= switch gained two new values
1446 "try-guest" and "try-host" that work like "guest" and
1447 "host", but do not fail if the host has no persistent
1448 journalling enabled. -j is now equivalent to
1449 --link-journal=try-guest.
1451 * macvlan network devices created by nspawn will now have
1452 stable MAC addresses.
1454 * A new SmackProcessLabel= unit setting has been added, which
1455 controls the SMACK security label processes forked off by
1456 the respective unit shall use.
1458 * If compiled with --enable-xkbcommon, systemd-localed will
1459 verify x11 keymap settings by compiling the given keymap. It
1460 will spew out warnings if the compilation fails. This
1461 requires libxkbcommon to be installed.
1463 * When a coredump is collected, a larger number of metadata
1464 fields is now collected and included in the journal records
1465 created for it. More specifically, control group membership,
1466 environment variables, memory maps, working directory,
1467 chroot directory, /proc/$PID/status, and a list of open file
1468 descriptors is now stored in the log entry.
1470 * The udev hwdb now contains DPI information for mice. For
1473 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
1475 * All systemd programs that read standalone configuration
1476 files in /etc now also support a corresponding series of
1477 .conf.d configuration directories in /etc/, /run/,
1478 /usr/local/lib/, /usr/lib/, and (if configured with
1479 --enable-split-usr) /lib/. In particular, the following
1480 configuration files now have corresponding configuration
1481 directories: system.conf user.conf, logind.conf,
1482 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
1483 resolved.conf, timesyncd.conf, journal-remote.conf, and
1484 journal-upload.conf. Note that distributions should use the
1485 configuration directories in /usr/lib/; the directories in
1486 /etc/ are reserved for the system administrator.
1488 * systemd-rfkill will no longer take the rfkill device name
1489 into account when storing rfkill state on disk, as the name
1490 might be dynamically assigned and not stable. Instead, the
1491 ID_PATH udev variable combined with the rfkill type (wlan,
1492 bluetooth, ...) is used.
1494 * A new service systemd-machine-id-commit.service has been
1495 added. When used on systems where /etc is read-only during
1496 boot, and /etc/machine-id is not initialized (but an empty
1497 file), this service will copy the temporary machine ID
1498 created as replacement into /etc after the system is fully
1499 booted up. This is useful for systems that are freshly
1500 installed with a non-initialized machine ID, but should get
1501 a fixed machine ID for subsequent boots.
1503 * networkd's .netdev files now provide a large set of
1504 configuration parameters for VXLAN devices. Similarly, the
1505 bridge port cost parameter is now configurable in .network
1506 files. There's also new support for configuring IP source
1507 routing. networkd .link files gained support for a new
1508 OriginalName= match that is useful to match against the
1509 original interface name the kernel assigned. .network files
1510 may include MTU= and MACAddress= fields for altering the MTU
1511 and MAC address while being connected to a specific network
1514 * The LUKS logic gained supported for configuring
1515 UUID-specific key files. There's also new support for naming
1516 LUKS device from the kernel command line, using the new
1517 luks.name= argument.
1519 * Timer units may now be transiently created via the bus API
1520 (this was previously already available for scope and service
1521 units). In addition it is now possible to create multiple
1522 transient units at the same time with a single bus call. The
1523 "systemd-run" tool has been updated to make use of this for
1524 running commands on a specified time, in at(1)-style.
1526 * tmpfiles gained support for "t" lines, for assigning
1527 extended attributes to files. Among other uses this may be
1528 used to assign SMACK labels to files.
1530 Contributions from: Alin Rauta, Alison Chaiken, Andrej
1531 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
1532 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
1533 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
1534 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
1535 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
1536 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
1537 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
1538 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
1539 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
1540 Michael Biebl, Michael Chapman, Michael Marineau, Michal
1541 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
1542 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
1543 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
1544 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
1545 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
1546 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
1547 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
1549 -- Berlin, 2014-12-10
1553 * journalctl gained the new options -t/--identifier= to match
1554 on the syslog identifier (aka "tag"), as well as --utc to
1555 show log timestamps in the UTC timezone. journalctl now also
1556 accepts -n/--lines=all to disable line capping in a pager.
1558 * journalctl gained a new switch, --flush, that synchronously
1559 flushes logs from /run/log/journal to /var/log/journal if
1560 persistent storage is enabled. systemd-journal-flush.service
1561 now waits until the operation is complete.
1563 * Services can notify the manager before they start a reload
1564 (by sending RELOADING=1) or shutdown (by sending
1565 STOPPING=1). This allows the manager to track and show the
1566 internal state of daemons and closes a race condition when
1567 the process is still running but has closed its D-Bus
1570 * Services with Type=oneshot do not have to have any ExecStart
1573 * User units are now loaded also from
1574 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
1575 /run/systemd/user directory that was already previously
1576 supported, but is under the control of the user.
1578 * Job timeouts (i.e. time-outs on the time a job that is
1579 queued stays in the run queue) can now optionally result in
1580 immediate reboot or power-off actions (JobTimeoutAction= and
1581 JobTimeoutRebootArgument=). This is useful on ".target"
1582 units, to limit the maximum time a target remains
1583 undispatched in the run queue, and to trigger an emergency
1584 operation in such a case. This is now used by default to
1585 turn off the system if boot-up (as defined by everything in
1586 basic.target) hangs and does not complete for at least
1587 15min. Also, if power-off or reboot hang for at least 30min
1588 an immediate power-off/reboot operation is triggered. This
1589 functionality is particularly useful to increase reliability
1590 on embedded devices, but also on laptops which might
1591 accidentally get powered on when carried in a backpack and
1592 whose boot stays stuck in a hard disk encryption passphrase
1595 * systemd-logind can be configured to also handle lid switch
1596 events even when the machine is docked or multiple displays
1597 are attached (HandleLidSwitchDocked= option).
1599 * A helper binary and a service have been added which can be
1600 used to resume from hibernation in the initramfs. A
1601 generator will parse the resume= option on the kernel
1602 command line to trigger resume.
1604 * A user console daemon systemd-consoled has been
1605 added. Currently, it is a preview, and will so far open a
1606 single terminal on each session of the user marked as
1607 Desktop=systemd-console.
1609 * Route metrics can be specified for DHCP routes added by
1612 * The SELinux context of socket-activated services can be set
1613 from the information provided by the networking stack
1614 (SELinuxContextFromNet= option).
1616 * Userspace firmware loading support has been removed and
1617 the minimum supported kernel version is thus bumped to 3.7.
1619 * Timeout for udev workers has been increased from 1 to 3
1620 minutes, but a warning will be printed after 1 minute to
1621 help diagnose kernel modules that take a long time to load.
1623 * Udev rules can now remove tags on devices with TAG-="foobar".
1625 * systemd's readahead implementation has been removed. In many
1626 circumstances it didn't give expected benefits even for
1627 rotational disk drives and was becoming less relevant in the
1628 age of SSDs. As none of the developers has been using
1629 rotating media anymore, and nobody stepped up to actively
1630 maintain this component of systemd it has now been removed.
1632 * Swap units can use Options= to specify discard options.
1633 Discard options specified for swaps in /etc/fstab are now
1636 * Docker containers are now detected as a separate type of
1639 * The Password Agent protocol gained support for queries where
1640 the user input is shown, useful e.g. for user names.
1641 systemd-ask-password gained a new --echo option to turn that
1644 * The default sysctl.d/ snippets will now set:
1646 net.core.default_qdisc = fq_codel
1648 This selects Fair Queuing Controlled Delay as the default
1649 queuing discipline for network interfaces. fq_codel helps
1650 fight the network bufferbloat problem. It is believed to be
1651 a good default with no tuning required for most workloads.
1652 Downstream distributions may override this choice. On 10Gbit
1653 servers that do not do forwarding, "fq" may perform better.
1654 Systems without a good clocksource should use "pfifo_fast".
1656 * If kdbus is enabled during build a new option BusPolicy= is
1657 available for service units, that allows locking all service
1658 processes into a stricter bus policy, in order to limit
1659 access to various bus services, or even hide most of them
1660 from the service's view entirely.
1662 * networkctl will now show the .network and .link file
1663 networkd has applied to a specific interface.
1665 * sd-login gained a new API call sd_session_get_desktop() to
1666 query which desktop environment has been selected for a
1669 * UNIX utmp support is now compile-time optional to support
1670 legacy-free systems.
1672 * systemctl gained two new commands "add-wants" and
1673 "add-requires" for pulling in units from specific targets
1676 * If the word "rescue" is specified on the kernel command line
1677 the system will now boot into rescue mode (aka
1678 rescue.target), which was previously available only by
1679 specifying "1" or "systemd.unit=rescue.target" on the kernel
1680 command line. This new kernel command line option nicely
1681 mirrors the already existing "emergency" kernel command line
1684 * New kernel command line options mount.usr=, mount.usrflags=,
1685 mount.usrfstype= have been added that match root=, rootflags=,
1686 rootfstype= but allow mounting a specific file system to
1689 * The $NOTIFY_SOCKET is now also passed to control processes of
1690 services, not only the main process.
1692 * This version reenables support for fsck's -l switch. This
1693 means at least version v2.25 of util-linux is required for
1694 operation, otherwise dead-locks on device nodes may
1695 occur. Again: you need to update util-linux to at least
1696 v2.25 when updating systemd to v217.
1698 * The "multi-seat-x" tool has been removed from systemd, as
1699 its functionality has been integrated into X servers 1.16,
1700 and the tool is hence redundant. It is recommended to update
1701 display managers invoking this tool to simply invoke X
1702 directly from now on, again.
1704 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
1705 message flag has been added for all of systemd's PolicyKit
1706 authenticated method calls has been added. In particular
1707 this now allows optional interactive authorization via
1708 PolicyKit for many of PID1's privileged operations such as
1709 unit file enabling and disabling.
1711 * "udevadm hwdb --update" learnt a new switch "--usr" for
1712 placing the rebuilt hardware database in /usr instead of
1713 /etc. When used only hardware database entries stored in
1714 /usr will be used, and any user database entries in /etc are
1715 ignored. This functionality is useful for vendors to ship a
1716 pre-built database on systems where local configuration is
1717 unnecessary or unlikely.
1719 * Calendar time specifications in .timer units now also
1720 understand the strings "semi-annually", "quarterly" and
1721 "minutely" as shortcuts (in addition to the preexisting
1722 "anually", "hourly", ...).
1724 * systemd-tmpfiles will now correctly create files in /dev
1725 at boot which are marked for creation only at boot. It is
1726 recommended to always create static device nodes with 'c!'
1727 and 'b!', so that they are created only at boot and not
1728 overwritten at runtime.
1730 * When the watchdog logic is used for a service (WatchdogSec=)
1731 and the watchdog timeout is hit the service will now be
1732 terminated with SIGABRT (instead of just SIGTERM), in order
1733 to make sure a proper coredump and backtrace is
1734 generated. This ensures that hanging services will result in
1735 similar coredump/backtrace behaviour as services that hit a
1738 Contributions from: Andreas Henriksson, Andrei Borzenkov,
1739 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
1740 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
1741 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
1742 Herrmann, David Sommerseth, David Strauss, Emil Renner
1743 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
1744 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
1745 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
1746 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
1747 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
1748 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
1749 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
1750 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
1751 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
1752 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
1753 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
1754 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
1755 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
1756 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
1757 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
1758 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
1761 -- Berlin, 2014-10-28
1765 * timedated no longer reads NTP implementation unit names from
1766 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
1767 implementations should add a
1769 Conflicts=systemd-timesyncd.service
1771 to their unit files to take over and replace systemd's NTP
1772 default functionality.
1774 * systemd-sysusers gained a new line type "r" for configuring
1775 which UID/GID ranges to allocate system users/groups
1776 from. Lines of type "u" may now add an additional column
1777 that specifies the home directory for the system user to be
1778 created. Also, systemd-sysusers may now optionally read user
1779 information from STDIN instead of a file. This is useful for
1780 invoking it from RPM preinst scriptlets that need to create
1781 users before the first RPM file is installed since these
1782 files might need to be owned by them. A new
1783 %sysusers_create_inline RPM macro has been introduced to do
1784 just that. systemd-sysusers now updates the shadow files as
1785 well as the user/group databases, which should enhance
1786 compatibility with certain tools like grpck.
1788 * A number of bus APIs of PID 1 now optionally consult
1789 PolicyKit to permit access for otherwise unprivileged
1790 clients under certain conditions. Note that this currently
1791 doesn't support interactive authentication yet, but this is
1792 expected to be added eventually, too.
1794 * /etc/machine-info now has new fields for configuring the
1795 deployment environment of the machine, as well as the
1796 location of the machine. hostnamectl has been updated with
1797 new command to update these fields.
1799 * systemd-timesyncd has been updated to automatically acquire
1800 NTP server information from systemd-networkd, which might
1801 have been discovered via DHCP.
1803 * systemd-resolved now includes a caching DNS stub resolver
1804 and a complete LLMNR name resolution implementation. A new
1805 NSS module "nss-resolve" has been added which can be used
1806 instead of glibc's own "nss-dns" to resolve hostnames via
1807 systemd-resolved. Hostnames, addresses and arbitrary RRs may
1808 be resolved via systemd-resolved D-Bus APIs. In contrast to
1809 the glibc internal resolver systemd-resolved is aware of
1810 multi-homed system, and keeps DNS server and caches separate
1811 and per-interface. Queries are sent simultaneously on all
1812 interfaces that have DNS servers configured, in order to
1813 properly handle VPNs and local LANs which might resolve
1814 separate sets of domain names. systemd-resolved may acquire
1815 DNS server information from systemd-networkd automatically,
1816 which in turn might have discovered them via DHCP. A tool
1817 "systemd-resolve-host" has been added that may be used to
1818 query the DNS logic in resolved. systemd-resolved implements
1819 IDNA and automatically uses IDNA or UTF-8 encoding depending
1820 on whether classic DNS or LLMNR is used as transport. In the
1821 next releases we intend to add a DNSSEC and mDNS/DNS-SD
1822 implementation to systemd-resolved.
1824 * A new NSS module nss-mymachines has been added, that
1825 automatically resolves the names of all local registered
1826 containers to their respective IP addresses.
1828 * A new client tool "networkctl" for systemd-networkd has been
1829 added. It currently is entirely passive and will query
1830 networking configuration from udev, rtnetlink and networkd,
1831 and present it to the user in a very friendly
1832 way. Eventually, we hope to extend it to become a full
1833 control utility for networkd.
1835 * .socket units gained a new DeferAcceptSec= setting that
1836 controls the kernels' TCP_DEFER_ACCEPT sockopt for
1837 TCP. Similarly, support for controlling TCP keep-alive
1838 settings has been added (KeepAliveTimeSec=,
1839 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
1840 turning off Nagle's algorithm on TCP has been added
1843 * logind learned a new session type "web", for use in projects
1844 like Cockpit which register web clients as PAM sessions.
1846 * timer units with at least one OnCalendar= setting will now
1847 be started only after timer-sync.target has been
1848 reached. This way they will not elapse before the system
1849 clock has been corrected by a local NTP client or
1850 similar. This is particular useful on RTC-less embedded
1851 machines, that come up with an invalid system clock.
1853 * systemd-nspawn's --network-veth= switch should now result in
1854 stable MAC addresses for both the outer and the inner side
1857 * systemd-nspawn gained a new --volatile= switch for running
1858 container instances with /etc or /var unpopulated.
1860 * The kdbus client code has been updated to use the new Linux
1861 3.17 memfd subsystem instead of the old kdbus-specific one.
1863 * systemd-networkd's DHCP client and server now support
1864 FORCERENEW. There are also new configuration options to
1865 configure the vendor client identifier and broadcast mode
1868 * systemd will no longer inform the kernel about the current
1869 timezone, as this is necessarily incorrect and racy as the
1870 kernel has no understanding of DST and similar
1871 concepts. This hence means FAT timestamps will be always
1872 considered UTC, similar to what Android is already
1873 doing. Also, when the RTC is configured to the local time
1874 (rather than UTC) systemd will never synchronize back to it,
1875 as this might confuse Windows at a later boot.
1877 * systemd-analyze gained a new command "verify" for offline
1878 validation of unit files.
1880 * systemd-networkd gained support for a couple of additional
1881 settings for bonding networking setups. Also, the metric for
1882 statically configured routes may now be configured. For
1883 network interfaces where this is appropriate the peer IP
1884 address may now be configured.
1886 * systemd-networkd's DHCP client will no longer request
1887 broadcasting by default, as this tripped up some networks.
1888 For hardware where broadcast is required the feature should
1889 be switched back on using RequestBroadcast=yes.
1891 * systemd-networkd will now set up IPv4LL addresses (when
1892 enabled) even if DHCP is configured successfully.
1894 * udev will now default to respect network device names given
1895 by the kernel when the kernel indicates that these are
1896 predictable. This behavior can be tweaked by changing
1897 NamePolicy= in the relevant .link file.
1899 * A new library systemd-terminal has been added that
1900 implements full TTY stream parsing and rendering. This
1901 library is supposed to be used later on for implementing a
1902 full userspace VT subsystem, replacing the current kernel
1905 * A new tool systemd-journal-upload has been added to push
1906 journal data to a remote system running
1907 systemd-journal-remote.
1909 * journald will no longer forward all local data to another
1910 running syslog daemon. This change has been made because
1911 rsyslog (which appears to be the most commonly used syslog
1912 implementation these days) no longer makes use of this, and
1913 instead pulls the data out of the journal on its own. Since
1914 forwarding the messages to a non-existent syslog server is
1915 more expensive than we assumed we have now turned this
1916 off. If you run a syslog server that is not a recent rsyslog
1917 version, you have to turn this option on again
1918 (ForwardToSyslog= in journald.conf).
1920 * journald now optionally supports the LZ4 compressor for
1921 larger journal fields. This compressor should perform much
1922 better than XZ which was the previous default.
1924 * machinectl now shows the IP addresses of local containers,
1925 if it knows them, plus the interface name of the container.
1927 * A new tool "systemd-escape" has been added that makes it
1928 easy to escape strings to build unit names and similar.
1930 * sd_notify() messages may now include a new ERRNO= field
1931 which is parsed and collected by systemd and shown among the
1932 "systemctl status" output for a service.
1934 * A new component "systemd-firstboot" has been added that
1935 queries the most basic systemd information (timezone,
1936 hostname, root password) interactively on first
1937 boot. Alternatively it may also be used to provision these
1938 things offline on OS images installed into directories.
1940 * The default sysctl.d/ snippets will now set
1942 net.ipv4.conf.default.promote_secondaries=1
1944 This has the benefit of no flushing secondary IP addresses
1945 when primary addresses are removed.
1947 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
1948 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
1949 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
1950 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
1951 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
1952 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
1953 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
1954 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
1955 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
1956 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
1957 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
1958 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
1959 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
1960 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
1961 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
1963 -- Berlin, 2014-08-19
1967 * A new tool systemd-sysusers has been added. This tool
1968 creates system users and groups in /etc/passwd and
1969 /etc/group, based on static declarative system user/group
1970 definitions in /usr/lib/sysusers.d/. This is useful to
1971 enable factory resets and volatile systems that boot up with
1972 an empty /etc directory, and thus need system users and
1973 groups created during early boot. systemd now also ships
1974 with two default sysusers.d/ files for the most basic
1975 users and groups systemd and the core operating system
1978 * A new tmpfiles snippet has been added that rebuilds the
1979 essential files in /etc on boot, should they be missing.
1981 * A directive for ensuring automatic clean-up of
1982 /var/cache/man/ has been removed from the default
1983 configuration. This line should now be shipped by the man
1984 implementation. The necessary change has been made to the
1985 man-db implementation. Note that you need to update your man
1986 implementation to one that ships this line, otherwise no
1987 automatic clean-up of /var/cache/man will take place.
1989 * A new condition ConditionNeedsUpdate= has been added that
1990 may conditionalize services to only run when /etc or /var
1991 are "older" than the vendor operating system resources in
1992 /usr. This is useful for reconstructing or updating /etc
1993 after an offline update of /usr or a factory reset, on the
1994 next reboot. Services that want to run once after such an
1995 update or reset should use this condition and order
1996 themselves before the new systemd-update-done.service, which
1997 will mark the two directories as fully updated. A number of
1998 service files have been added making use of this, to rebuild
1999 the udev hardware database, the journald message catalog and
2000 dynamic loader cache (ldconfig). The systemd-sysusers tool
2001 described above also makes use of this now. With this in
2002 place it is now possible to start up a minimal operating
2003 system with /etc empty cleanly. For more information on the
2004 concepts involved see this recent blog story:
2006 http://0pointer.de/blog/projects/stateless.html
2008 * A new system group "input" has been introduced, and all
2009 input device nodes get this group assigned. This is useful
2010 for system-level software to get access to input devices. It
2011 complements what is already done for "audio" and "video".
2013 * systemd-networkd learnt minimal DHCPv4 server support in
2014 addition to the existing DHCPv4 client support. It also
2015 learnt DHCPv6 client and IPv6 Router Solicitation client
2016 support. The DHCPv4 client gained support for static routes
2017 passed in from the server. Note that the [DHCPv4] section
2018 known in older systemd-networkd versions has been renamed to
2019 [DHCP] and is now also used by the DHCPv6 client. Existing
2020 .network files using settings of this section should be
2021 updated, though compatibility is maintained. Optionally, the
2022 client hostname may now be sent to the DHCP server.
2024 * networkd gained support for vxlan virtual networks as well
2025 as tun/tap and dummy devices.
2027 * networkd gained support for automatic allocation of address
2028 ranges for interfaces from a system-wide pool of
2029 addresses. This is useful for dynamically managing a large
2030 number of interfaces with a single network configuration
2031 file. In particular this is useful to easily assign
2032 appropriate IP addresses to the veth links of a large number
2033 of nspawn instances.
2035 * RPM macros for processing sysusers, sysctl and binfmt
2036 drop-in snippets at package installation time have been
2039 * The /etc/os-release file should now be placed in
2040 /usr/lib/os-release. The old location is automatically
2041 created as symlink. /usr/lib is the more appropriate
2042 location of this file, since it shall actually describe the
2043 vendor operating system shipped in /usr, and not the
2044 configuration stored in /etc.
2046 * .mount units gained a new boolean SloppyOptions= setting
2047 that maps to mount(8)'s -s option which enables permissive
2048 parsing of unknown mount options.
2050 * tmpfiles learnt a new "L+" directive which creates a symlink
2051 but (unlike "L") deletes a pre-existing file first, should
2052 it already exist and not already be the correct
2053 symlink. Similarly, "b+", "c+" and "p+" directives have been
2054 added as well, which create block and character devices, as
2055 well as fifos in the filesystem, possibly removing any
2056 pre-existing files of different types.
2058 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
2059 'argument' field (which so far specified the source to
2060 symlink/copy the files from) is now optional. If omitted the
2061 same file os copied from /usr/share/factory/ suffixed by the
2062 full destination path. This is useful for populating /etc
2063 with essential files, by copying them from vendor defaults
2064 shipped in /usr/share/factory/etc.
2066 * A new command "systemctl preset-all" has been added that
2067 applies the service preset settings to all installed unit
2068 files. A new switch --preset-mode= has been added that
2069 controls whether only enable or only disable operations
2072 * A new command "systemctl is-system-running" has been added
2073 that allows checking the overall state of the system, for
2074 example whether it is fully up and running.
2076 * When the system boots up with an empty /etc, the equivalent
2077 to "systemctl preset-all" is executed during early boot, to
2078 make sure all default services are enabled after a factory
2081 * systemd now contains a minimal preset file that enables the
2082 most basic services systemd ships by default.
2084 * Unit files' [Install] section gained a new DefaultInstance=
2085 field for defining the default instance to create if a
2086 template unit is enabled with no instance specified.
2088 * A new passive target cryptsetup-pre.target has been added
2089 that may be used by services that need to make they run and
2090 finish before the first LUKS cryptographic device is set up.
2092 * The /dev/loop-control and /dev/btrfs-control device nodes
2093 are now owned by the "disk" group by default, opening up
2094 access to this group.
2096 * systemd-coredump will now automatically generate a
2097 stack trace of all core dumps taking place on the system,
2098 based on elfutils' libdw library. This stack trace is logged
2101 * systemd-coredump may now optionally store coredumps directly
2102 on disk (in /var/lib/systemd/coredump, possibly compressed),
2103 instead of storing them unconditionally in the journal. This
2104 mode is the new default. A new configuration file
2105 /etc/systemd/coredump.conf has been added to configure this
2106 and other parameters of systemd-coredump.
2108 * coredumpctl gained a new "info" verb to show details about a
2109 specific coredump. A new switch "-1" has also been added
2110 that makes sure to only show information about the most
2111 recent entry instead of all entries. Also, as the tool is
2112 generally useful now the "systemd-" prefix of the binary
2113 name has been removed. Distributions that want to maintain
2114 compatibility with the old name should add a symlink from
2115 the old name to the new name.
2117 * journald's SplitMode= now defaults to "uid". This makes sure
2118 that unprivileged users can access their own coredumps with
2119 coredumpctl without restrictions.
2121 * New kernel command line options "systemd.wants=" (for
2122 pulling an additional unit during boot), "systemd.mask="
2123 (for masking a specific unit for the boot), and
2124 "systemd.debug-shell" (for enabling the debug shell on tty9)
2125 have been added. This is implemented in the new generator
2126 "systemd-debug-generator".
2128 * systemd-nspawn will now by default filter a couple of
2129 syscalls for containers, among them those required for
2130 kernel module loading, direct x86 IO port access, swap
2131 management, and kexec. Most importantly though
2132 open_by_handle_at() is now prohibited for containers,
2133 closing a hole similar to a recently discussed vulnerability
2134 in docker regarding access to files on file hierarchies the
2135 container should normally not have access to. Note that, for
2136 nspawn, we generally make no security claims anyway (and
2137 this is explicitly documented in the man page), so this is
2138 just a fix for one of the most obvious problems.
2140 * A new man page file-hierarchy(7) has been added that
2141 contains a minimized, modernized version of the file system
2142 layout systemd expects, similar in style to the FHS
2143 specification or hier(5). A new tool systemd-path(1) has
2144 been added to query many of these paths for the local
2147 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
2148 longer done. Since the directory now has a per-user size
2149 limit, and is cleaned on logout this appears unnecessary,
2150 in particular since this now brings the lifecycle of this
2151 directory closer in line with how IPC objects are handled.
2153 * systemd.pc now exports a number of additional directories,
2154 including $libdir (which is useful to identify the library
2155 path for the primary architecture of the system), and a
2156 couple of drop-in directories.
2158 * udev's predictable network interface names now use the dev_port
2159 sysfs attribute, introduced in linux 3.15 instead of dev_id to
2160 distinguish between ports of the same PCI function. dev_id should
2161 only be used for ports using the same HW address, hence the need
2164 * machined has been updated to export the OS version of a
2165 container (read from /etc/os-release and
2166 /usr/lib/os-release) on the bus. This is now shown in
2167 "machinectl status" for a machine.
2169 * A new service setting RestartForceExitStatus= has been
2170 added. If configured to a set of exit signals or process
2171 return values, the service will be restarted when the main
2172 daemon process exits with any of them, regardless of the
2175 * systemctl's -H switch for connecting to remote systemd
2176 machines has been extended so that it may be used to
2177 directly connect to a specific container on the
2178 host. "systemctl -H root@foobar:waldi" will now connect as
2179 user "root" to host "foobar", and then proceed directly to
2180 the container named "waldi". Note that currently you have to
2181 authenticate as user "root" for this to work, as entering
2182 containers is a privileged operation.
2184 Contributions from: Andreas Henriksson, Benjamin Steinwender,
2185 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
2186 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
2187 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
2188 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
2189 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
2190 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
2191 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
2192 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
2193 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
2194 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
2195 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
2197 -- Berlin, 2014-07-03
2201 * As an experimental feature, udev now tries to lock the
2202 disk device node (flock(LOCK_SH|LOCK_NB)) while it
2203 executes events for the disk or any of its partitions.
2204 Applications like partitioning programs can lock the
2205 disk device node (flock(LOCK_EX)) and claim temporary
2206 device ownership that way; udev will entirely skip all event
2207 handling for this disk and its partitions. If the disk
2208 was opened for writing, the close will trigger a partition
2209 table rescan in udev's "watch" facility, and if needed
2210 synthesize "change" events for the disk and all its partitions.
2211 This is now unconditionally enabled, and if it turns out to
2212 cause major problems, we might turn it on only for specific
2213 devices, or might need to disable it entirely. Device Mapper
2214 devices are excluded from this logic.
2216 * We temporarily dropped the "-l" switch for fsck invocations,
2217 since they collide with the flock() logic above. util-linux
2218 upstream has been changed already to avoid this conflict,
2219 and we will readd "-l" as soon as util-linux with this
2220 change has been released.
2222 * The dependency on libattr has been removed. Since a long
2223 time, the extended attribute calls have moved to glibc, and
2224 libattr is thus unnecessary.
2226 * Virtualization detection works without priviliges now. This
2227 means the systemd-detect-virt binary no longer requires
2228 CAP_SYS_PTRACE file capabilities, and our daemons can run
2229 with fewer privileges.
2231 * systemd-networkd now runs under its own "systemd-network"
2232 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
2233 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
2234 loses the ability to write to files owned by root this way.
2236 * Similarly, systemd-resolved now runs under its own
2237 "systemd-resolve" user with no capabilities remaining.
2239 * Similarly, systemd-bus-proxyd now runs under its own
2240 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
2242 * systemd-networkd gained support for setting up "veth"
2243 virtual Ethernet devices for container connectivity, as well
2244 as GRE and VTI tunnels.
2246 * systemd-networkd will no longer automatically attempt to
2247 manually load kernel modules necessary for certain tunnel
2248 transports. Instead, it is assumed the kernel loads them
2249 automatically when required. This only works correctly on
2250 very new kernels. On older kernels, please consider adding
2251 the kernel modules to /etc/modules-load.d/ as a work-around.
2253 * The resolv.conf file systemd-resolved generates has been
2254 moved to /run/systemd/resolve/. If you have a symlink from
2255 /etc/resolv.conf, it might be necessary to correct it.
2257 * Two new service settings, ProtectHome= and ProtectSystem=,
2258 have been added. When enabled, they will make the user data
2259 (such as /home) inaccessible or read-only and the system
2260 (such as /usr) read-only, for specific services. This allows
2261 very light-weight per-service sandboxing to avoid
2262 modifications of user data or system files from
2263 services. These two new switches have been enabled for all
2264 of systemd's long-running services, where appropriate.
2266 * Socket units gained new SocketUser= and SocketGroup=
2267 settings to set the owner user and group of AF_UNIX sockets
2268 and FIFOs in the file system.
2270 * Socket units gained a new RemoveOnStop= setting. If enabled,
2271 all FIFOS and sockets in the file system will be removed
2272 when the specific socket unit is stopped.
2274 * Socket units gained a new Symlinks= setting. It takes a list
2275 of symlinks to create to file system sockets or FIFOs
2276 created by the specific Unix sockets. This is useful to
2277 manage symlinks to socket nodes with the same life-cycle as
2280 * The /dev/log socket and /dev/initctl FIFO have been moved to
2281 /run, and have been replaced by symlinks. This allows
2282 connecting to these facilities even if PrivateDevices=yes is
2283 used for a service (which makes /dev/log itself unavailable,
2284 but /run is left). This also has the benefit of ensuring
2285 that /dev only contains device nodes, directories and
2286 symlinks, and nothing else.
2288 * sd-daemon gained two new calls sd_pid_notify() and
2289 sd_pid_notifyf(). They are similar to sd_notify() and
2290 sd_notifyf(), but allow overriding of the source PID of
2291 notification messages if permissions permit this. This is
2292 useful to send notify messages on behalf of a different
2293 process (for example, the parent process). The
2294 systemd-notify tool has been updated to make use of this
2295 when sending messages (so that notification messages now
2296 originate from the shell script invoking systemd-notify and
2297 not the systemd-notify process itself. This should minimize
2298 a race where systemd fails to associate notification
2299 messages to services when the originating process already
2302 * A new "on-abnormal" setting for Restart= has been added. If
2303 set, it will result in automatic restarts on all "abnormal"
2304 reasons for a process to exit, which includes unclean
2305 signals, core dumps, timeouts and watchdog timeouts, but
2306 does not include clean and unclean exit codes or clean
2307 signals. Restart=on-abnormal is an alternative for
2308 Restart=on-failure for services that shall be able to
2309 terminate and avoid restarts on certain errors, by
2310 indicating so with an unclean exit code. Restart=on-failure
2311 or Restart=on-abnormal is now the recommended setting for
2312 all long-running services.
2314 * If the InaccessibleDirectories= service setting points to a
2315 mount point (or if there are any submounts contained within
2316 it), it is now attempted to completely unmount it, to make
2317 the file systems truly unavailable for the respective
2320 * The ReadOnlyDirectories= service setting and
2321 systemd-nspawn's --read-only parameter are now recursively
2322 applied to all submounts, too.
2324 * Mount units may now be created transiently via the bus APIs.
2326 * The support for SysV and LSB init scripts has been removed
2327 from the systemd daemon itself. Instead, it is now
2328 implemented as a generator that creates native systemd units
2329 from these scripts when needed. This enables us to remove a
2330 substantial amount of legacy code from PID 1, following the
2331 fact that many distributions only ship a very small number
2332 of LSB/SysV init scripts nowadays.
2334 * Privileged Xen (dom0) domains are not considered
2335 virtualization anymore by the virtualization detection
2336 logic. After all, they generally have unrestricted access to
2337 the hardware and usually are used to manage the unprivileged
2340 * systemd-tmpfiles gained a new "C" line type, for copying
2341 files or entire directories.
2343 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
2344 lines. So far, they have been non-globbing versions of the
2345 latter, and have thus been redundant. In future, it is
2346 recommended to only use "z". "m" has hence been removed
2347 from the documentation, even though it stays supported.
2349 * A tmpfiles snippet to recreate the most basic structure in
2350 /var has been added. This is enough to create the /var/run →
2351 /run symlink and create a couple of structural
2352 directories. This allows systems to boot up with an empty or
2353 volatile /var. Of course, while with this change, the core OS
2354 now is capable with dealing with a volatile /var, not all
2355 user services are ready for it. However, we hope that sooner
2356 or later, many service daemons will be changed upstream so
2357 that they are able to automatically create their necessary
2358 directories in /var at boot, should they be missing. This is
2359 the first step to allow state-less systems that only require
2360 the vendor image for /usr to boot.
2362 * systemd-nspawn has gained a new --tmpfs= switch to mount an
2363 empty tmpfs instance to a specific directory. This is
2364 particularly useful for making use of the automatic
2365 reconstruction of /var (see above), by passing --tmpfs=/var.
2367 * Access modes specified in tmpfiles snippets may now be
2368 prefixed with "~", which indicates that they shall be masked
2369 by whether the existing file or directory is currently
2370 writable, readable or executable at all. Also, if specified,
2371 the sgid/suid/sticky bits will be masked for all
2374 * A new passive target unit "network-pre.target" has been
2375 added which is useful for services that shall run before any
2376 network is configured, for example firewall scripts.
2378 * The "floppy" group that previously owned the /dev/fd*
2379 devices is no longer used. The "disk" group is now used
2380 instead. Distributions should probably deprecate usage of
2383 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
2384 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
2385 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
2386 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
2387 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
2388 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
2391 -- Berlin, 2014-06-11
2395 * A new "systemd-timesyncd" daemon has been added for
2396 synchronizing the system clock across the network. It
2397 implements an SNTP client. In contrast to NTP
2398 implementations such as chrony or the NTP reference server,
2399 this only implements a client side, and does not bother with
2400 the full NTP complexity, focusing only on querying time from
2401 one remote server and synchronizing the local clock to
2402 it. Unless you intend to serve NTP to networked clients or
2403 want to connect to local hardware clocks, this simple NTP
2404 client should be more than appropriate for most
2405 installations. The daemon runs with minimal privileges, and
2406 has been hooked up with networkd to only operate when
2407 network connectivity is available. The daemon saves the
2408 current clock to disk every time a new NTP sync has been
2409 acquired, and uses this to possibly correct the system clock
2410 early at bootup, in order to accommodate for systems that
2411 lack an RTC such as the Raspberry Pi and embedded devices,
2412 and to make sure that time monotonically progresses on these
2413 systems, even if it is not always correct. To make use of
2414 this daemon, a new system user and group "systemd-timesync"
2415 needs to be created on installation of systemd.
2417 * The queue "seqnum" interface of libudev has been disabled, as
2418 it was generally incompatible with device namespacing as
2419 sequence numbers of devices go "missing" if the devices are
2420 part of a different namespace.
2422 * "systemctl list-timers" and "systemctl list-sockets" gained
2423 a --recursive switch for showing units of these types also
2424 for all local containers, similar in style to the already
2425 supported --recursive switch for "systemctl list-units".
2427 * A new RebootArgument= setting has been added for service
2428 units, which may be used to specify a kernel reboot argument
2429 to use when triggering reboots with StartLimitAction=.
2431 * A new FailureAction= setting has been added for service
2432 units which may be used to specify an operation to trigger
2433 when a service fails. This works similarly to
2434 StartLimitAction=, but unlike it, controls what is done
2435 immediately rather than only after several attempts to
2436 restart the service in question.
2438 * hostnamed got updated to also expose the kernel name,
2439 release, and version on the bus. This is useful for
2440 executing commands like hostnamectl with the -H switch.
2441 systemd-analyze makes use of this to properly display
2442 details when running non-locally.
2444 * The bootchart tool can now show cgroup information in the
2445 graphs it generates.
2447 * The CFS CPU quota cgroup attribute is now exposed for
2448 services. The new CPUQuota= switch has been added for this
2449 which takes a percentage value. Setting this will have the
2450 result that a service may never get more CPU time than the
2451 specified percentage, even if the machine is otherwise idle.
2453 * systemd-networkd learned IPIP and SIT tunnel support.
2455 * LSB init scripts exposing a dependency on $network will now
2456 get a dependency on network-online.target rather than simply
2457 network.target. This should bring LSB handling closer to
2458 what it was on SysV systems.
2460 * A new fsck.repair= kernel option has been added to control
2461 how fsck shall deal with unclean file systems at boot.
2463 * The (.ini) configuration file parser will now silently
2464 ignore sections whose name begins with "X-". This may be
2465 used to maintain application-specific extension sections in unit
2468 * machined gained a new API to query the IP addresses of
2469 registered containers. "machinectl status" has been updated
2470 to show these addresses in its output.
2472 * A new call sd_uid_get_display() has been added to the
2473 sd-login APIs for querying the "primary" session of a
2474 user. The "primary" session of the user is elected from the
2475 user's sessions and generally a graphical session is
2476 preferred over a text one.
2478 * A minimal systemd-resolved daemon has been added. It
2479 currently simply acts as a companion to systemd-networkd and
2480 manages resolv.conf based on per-interface DNS
2481 configuration, possibly supplied via DHCP. In the long run
2482 we hope to extend this into a local DNSSEC enabled DNS and
2485 * The systemd-networkd-wait-online tool is now enabled by
2486 default. It will delay network-online.target until a network
2487 connection has been configured. The tool primarily integrates
2488 with networkd, but will also make a best effort to make sense
2489 of network configuration performed in some other way.
2491 * Two new service options StartupCPUShares= and
2492 StartupBlockIOWeight= have been added that work similarly to
2493 CPUShares= and BlockIOWeight= however only apply during
2494 system startup. This is useful to prioritize certain services
2495 differently during bootup than during normal runtime.
2497 * hostnamed has been changed to prefer the statically
2498 configured hostname in /etc/hostname (unless set to
2499 'localhost' or empty) over any dynamic one supplied by
2500 dhcp. With this change, the rules for picking the hostname
2501 match more closely the rules of other configuration settings
2502 where the local administrator's configuration in /etc always
2503 overrides any other settings.
2505 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
2506 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
2507 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
2508 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
2509 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
2510 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
2511 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
2512 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
2513 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
2514 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
2515 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
2516 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
2517 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
2518 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
2519 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
2520 Lindskog, WaLyong Cho, Will Woods, Zbigniew
2523 -- Beijing, 2014-05-28
2527 * When restoring the screen brightness at boot, stay away from
2528 the darkest setting or from the lowest 5% of the available
2529 range, depending on which is the larger value of both. This
2530 should effectively protect the user from rebooting into a
2531 black screen, should the brightness have been set to minimum
2534 * sd-login gained a new sd_machine_get_class() call to
2535 determine the class ("vm" or "container") of a machine
2536 registered with machined.
2538 * sd-login gained new calls
2539 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
2540 to query the identity of the peer of a local AF_UNIX
2541 connection. They operate similarly to their sd_pid_get_xyz()
2544 * PID 1 will now maintain a system-wide system state engine
2545 with the states "starting", "running", "degraded",
2546 "maintenance", "stopping". These states are bound to system
2547 startup, normal runtime, runtime with at least one failed
2548 service, rescue/emergency mode and system shutdown. This
2549 state is shown in the "systemctl status" output when no unit
2550 name is passed. It is useful to determine system state, in
2551 particularly when doing so for many systems or containers at
2554 * A new command "list-machines" has been added to "systemctl"
2555 that lists all local OS containers and shows their system
2556 state (see above), if systemd runs inside of them.
2558 * systemctl gained a new "-r" switch to recursively enumerate
2559 units on all local containers, when used with the
2560 "list-unit" command (which is the default one that is
2561 executed when no parameters are specified).
2563 * The GPT automatic partition discovery logic will now honour
2564 two GPT partition flags: one may be set on a partition to
2565 cause it to be mounted read-only, and the other may be set
2566 on a partition to ignore it during automatic discovery.
2568 * Two new GPT type UUIDs have been added for automatic root
2569 partition discovery, for 32-bit and 64-bit ARM. This is not
2570 particularly useful for discovering the root directory on
2571 these architectures during bare-metal boots (since UEFI is
2572 not common there), but still very useful to allow booting of
2573 ARM disk images in nspawn with the -i option.
2575 * MAC addresses of interfaces created with nspawn's
2576 --network-interface= switch will now be generated from the
2577 machine name, and thus be stable between multiple invocations
2580 * logind will now automatically remove all IPC objects owned
2581 by a user if she or he fully logs out. This makes sure that
2582 users who are logged out cannot continue to consume IPC
2583 resources. This covers SysV memory, semaphores and message
2584 queues as well as POSIX shared memory and message
2585 queues. Traditionally, SysV and POSIX IPC had no life-cycle
2586 limits. With this functionality, that is corrected. This may
2587 be turned off by using the RemoveIPC= switch of logind.conf.
2589 * The systemd-machine-id-setup and tmpfiles tools gained a
2590 --root= switch to operate on a specific root directory,
2593 * journald can now forward logged messages to the TTYs of all
2594 logged in users ("wall"). This is the default for all
2595 emergency messages now.
2597 * A new tool systemd-journal-remote has been added to stream
2598 journal log messages across the network.
2600 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
2601 controller trees are mounted into it. Note that the
2602 directories mounted beneath it are not read-only. This is a
2603 security measure and is particularly useful because glibc
2604 actually includes a search logic to pick any tmpfs it can
2605 find to implement shm_open() if /dev/shm is not available
2606 (which it might very well be in namespaced setups).
2608 * machinectl gained a new "poweroff" command to cleanly power
2609 down a local OS container.
2611 * The PrivateDevices= unit file setting will now also drop the
2612 CAP_MKNOD capability from the capability bound set, and
2613 imply DevicePolicy=closed.
2615 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
2616 comprehensively on all long-running systemd services where
2617 this is appropriate.
2619 * systemd-udevd will now run in a disassociated mount
2620 namespace. To mount directories from udev rules, make sure to
2621 pull in mount units via SYSTEMD_WANTS properties.
2623 * The kdbus support gained support for uploading policy into
2624 the kernel. sd-bus gained support for creating "monitoring"
2625 connections that can eavesdrop into all bus communication
2626 for debugging purposes.
2628 * Timestamps may now be specified in seconds since the UNIX
2629 epoch Jan 1st, 1970 by specifying "@" followed by the value
2632 * Native tcpwrap support in systemd has been removed. tcpwrap
2633 is old code, not really maintained anymore and has serious
2634 shortcomings, and better options such as firewalls
2635 exist. For setups that require tcpwrap usage, please
2636 consider invoking your socket-activated service via tcpd,
2637 like on traditional inetd.
2639 * A new system.conf configuration option
2640 DefaultTimerAccuracySec= has been added that controls the
2641 default AccuracySec= setting of .timer units.
2643 * Timer units gained a new WakeSystem= switch. If enabled,
2644 timers configured this way will cause the system to resume
2645 from system suspend (if the system supports that, which most
2648 * Timer units gained a new Persistent= switch. If enabled,
2649 timers configured this way will save to disk when they have
2650 been last triggered. This information is then used on next
2651 reboot to possible execute overdue timer events, that
2652 could not take place because the system was powered off.
2653 This enables simple anacron-like behaviour for timer units.
2655 * systemctl's "list-timers" will now also list the time a
2656 timer unit was last triggered in addition to the next time
2657 it will be triggered.
2659 * systemd-networkd will now assign predictable IPv4LL
2660 addresses to its local interfaces.
2662 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
2663 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
2664 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
2665 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
2666 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
2667 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
2668 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
2669 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
2672 -- Berlin, 2014-03-25
2676 * A new unit file setting RestrictAddressFamilies= has been
2677 added to restrict which socket address families unit
2678 processes gain access to. This takes address family names
2679 like "AF_INET" or "AF_UNIX", and is useful to minimize the
2680 attack surface of services via exotic protocol stacks. This
2681 is built on seccomp system call filters.
2683 * Two new unit file settings RuntimeDirectory= and
2684 RuntimeDirectoryMode= have been added that may be used to
2685 manage a per-daemon runtime directories below /run. This is
2686 an alternative for setting up directory permissions with
2687 tmpfiles snippets, and has the advantage that the runtime
2688 directory's lifetime is bound to the daemon runtime and that
2689 the daemon starts up with an empty directory each time. This
2690 is particularly useful when writing services that drop
2691 privileges using the User= or Group= setting.
2693 * The DeviceAllow= unit setting now supports globbing for
2694 matching against device group names.
2696 * The systemd configuration file system.conf gained new
2697 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
2698 DefaultMemoryAccounting= to globally turn on/off accounting
2699 for specific resources (cgroups) for all units. These
2700 settings may still be overridden individually in each unit
2703 * systemd-gpt-auto-generator is now able to discover /srv and
2704 root partitions in addition to /home and swap partitions. It
2705 also supports LUKS-encrypted partitions now. With this in
2706 place, automatic discovery of partitions to mount following
2707 the Discoverable Partitions Specification
2708 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
2709 is now a lot more complete. This allows booting without
2710 /etc/fstab and without root= on the kernel command line on
2711 systems prepared appropriately.
2713 * systemd-nspawn gained a new --image= switch which allows
2714 booting up disk images and Linux installations on any block
2715 device that follow the Discoverable Partitions Specification
2716 (see above). This means that installations made with
2717 appropriately updated installers may now be started and
2718 deployed using container managers, completely
2719 unmodified. (We hope that libvirt-lxc will add support for
2720 this feature soon, too.)
2722 * systemd-nspawn gained a new --network-macvlan= setting to
2723 set up a private macvlan interface for the
2724 container. Similarly, systemd-networkd gained a new
2725 Kind=macvlan setting in .netdev files.
2727 * systemd-networkd now supports configuring local addresses
2730 * A new tool systemd-network-wait-online has been added to
2731 synchronously wait for network connectivity using
2734 * The sd-bus.h bus API gained a new sd_bus_track object for
2735 tracking the life-cycle of bus peers. Note that sd-bus.h is
2736 still not a public API though (unless you specify
2737 --enable-kdbus on the configure command line, which however
2738 voids your warranty and you get no API stability guarantee).
2740 * The $XDG_RUNTIME_DIR runtime directories for each user are
2741 now individual tmpfs instances, which has the benefit of
2742 introducing separate pools for each user, with individual
2743 size limits, and thus making sure that unprivileged clients
2744 can no longer negatively impact the system or other users by
2745 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
2746 RuntimeDirectorySize= has been introduced that allows
2747 controlling the default size limit for all users. It
2748 defaults to 10% of the available physical memory. This is no
2749 replacement for quotas on tmpfs though (which the kernel
2750 still does not support), as /dev/shm and /tmp are still
2751 shared resources used by both the system and unprivileged
2754 * logind will now automatically turn off automatic suspending
2755 on laptop lid close when more than one display is
2756 connected. This was previously expected to be implemented
2757 individually in desktop environments (such as GNOME),
2758 however has been added to logind now, in order to fix a
2759 boot-time race where a desktop environment might not have
2760 been started yet and thus not been able to take an inhibitor
2761 lock at the time where logind already suspends the system
2762 due to a closed lid.
2764 * logind will now wait at least 30s after each system
2765 suspend/resume cycle, and 3min after system boot before
2766 suspending the system due to a closed laptop lid. This
2767 should give USB docking stations and similar enough time to
2768 be probed and configured after system resume and boot in
2769 order to then act as suspend blocker.
2771 * systemd-run gained a new --property= setting which allows
2772 initialization of resource control properties (and others)
2773 for the created scope or service unit. Example: "systemd-run
2774 --property=BlockIOWeight=10 updatedb" may be used to run
2775 updatedb at a low block IO scheduling weight.
2777 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
2778 now also work in --scope mode.
2780 * When systemd is compiled with kdbus support, basic support
2781 for enforced policies is now in place. (Note that enabling
2782 kdbus still voids your warranty and no API compatibility
2785 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
2786 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
2787 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
2788 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
2789 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
2790 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
2791 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
2792 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
2793 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
2794 Zbigniew Jędrzejewski-Szmek
2796 -- Berlin, 2014-03-12
2800 * systemd will now relabel /dev after loading the SMACK policy
2801 according to SMACK rules.
2803 * A new unit file option AppArmorProfile= has been added to
2804 set the AppArmor profile for the processes of a unit.
2806 * A new condition check ConditionArchitecture= has been added
2807 to conditionalize units based on the system architecture, as
2808 reported by uname()'s "machine" field.
2810 * systemd-networkd now supports matching on the system
2811 virtualization, architecture, kernel command line, host name
2814 * logind is now a lot more aggressive when suspending the
2815 machine due to a closed laptop lid. Instead of acting only
2816 on the lid close action, it will continuously watch the lid
2817 status and act on it. This is useful for laptops where the
2818 power button is on the outside of the chassis so that it can
2819 be reached without opening the lid (such as the Lenovo
2820 Yoga). On those machines, logind will now immediately
2821 re-suspend the machine if the power button has been
2822 accidentally pressed while the laptop was suspended and in a
2823 backpack or similar.
2825 * logind will now watch SW_DOCK switches and inhibit reaction
2826 to the lid switch if it is pressed. This means that logind
2827 will not suspend the machine anymore if the lid is closed
2828 and the system is docked, if the laptop supports SW_DOCK
2829 notifications via the input layer. Note that ACPI docking
2830 stations do not generate this currently. Also note that this
2831 logic is usually not fully sufficient and Desktop
2832 Environments should take a lid switch inhibitor lock when an
2833 external display is connected, as systemd will not watch
2836 * nspawn will now make use of the devices cgroup controller by
2837 default, and only permit creation of and access to the usual
2838 API device nodes like /dev/null or /dev/random, as well as
2839 access to (but not creation of) the pty devices.
2841 * We will now ship a default .network file for
2842 systemd-networkd that automatically configures DHCP for
2843 network interfaces created by nspawn's --network-veth or
2844 --network-bridge= switches.
2846 * systemd will now understand the usual M, K, G, T suffixes
2847 according to SI conventions (i.e. to the base 1000) when
2848 referring to throughput and hardware metrics. It will stay
2849 with IEC conventions (i.e. to the base 1024) for software
2850 metrics, according to what is customary according to
2851 Wikipedia. We explicitly document which base applies for
2852 each configuration option.
2854 * The DeviceAllow= setting in unit files now supports a syntax
2855 to whitelist an entire group of devices node majors at once,
2856 based on the /proc/devices listing. For example, with the
2857 string "char-pts", it is now possible to whitelist all
2858 current and future pseudo-TTYs at once.
2860 * sd-event learned a new "post" event source. Event sources of
2861 this type are triggered by the dispatching of any event
2862 source of a type that is not "post". This is useful for
2863 implementing clean-up and check event sources that are
2864 triggered by other work being done in the program.
2866 * systemd-networkd is no longer statically enabled, but uses
2867 the usual [Install] sections so that it can be
2868 enabled/disabled using systemctl. It still is enabled by
2871 * When creating a veth interface pair with systemd-nspawn, the
2872 host side will now be prefixed with "vb-" if
2873 --network-bridge= is used, and with "ve-" if --network-veth
2874 is used. This way, it is easy to distinguish these cases on
2875 the host, for example to apply different configuration to
2876 them with systemd-networkd.
2878 * The compatibility libraries for libsystemd-journal.so,
2879 libsystem-id128.so, libsystemd-login.so and
2880 libsystemd-daemon.so do not make use of IFUNC
2881 anymore. Instead, we now build libsystemd.so multiple times
2882 under these alternative names. This means that the footprint
2883 is drastically increased, but given that these are
2884 transitional compatibility libraries, this should not matter
2885 much. This change has been made necessary to support the ARM
2886 platform for these compatibility libraries, as the ARM
2887 toolchain is not really at the same level as the toolchain
2888 for other architectures like x86 and does not support
2889 IFUNC. Please make sure to use --enable-compat-libs only
2890 during a transitional period!
2892 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
2893 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
2894 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
2895 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
2896 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
2897 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
2898 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
2899 Zbigniew Jędrzejewski-Szmek
2901 -- Berlin, 2014-02-24
2905 * A new component "systemd-networkd" has been added that can
2906 be used to configure local network interfaces statically or
2907 via DHCP. It is capable of bringing up bridges, VLANs, and
2908 bonding. Currently, no hook-ups for interactive network
2909 configuration are provided. Use this for your initrd,
2910 container, embedded, or server setup if you need a simple,
2911 yet powerful, network configuration solution. This
2912 configuration subsystem is quite nifty, as it allows wildcard
2913 hotplug matching in interfaces. For example, with a single
2914 configuration snippet, you can configure that all Ethernet
2915 interfaces showing up are automatically added to a bridge,
2916 or similar. It supports link-sensing and more.
2918 * A new tool "systemd-socket-proxyd" has been added which can
2919 act as a bidirectional proxy for TCP sockets. This is
2920 useful for adding socket activation support to services that
2921 do not actually support socket activation, including virtual
2922 machines and the like.
2924 * Add a new tool to save/restore rfkill state on
2927 * Save/restore state of keyboard backlights in addition to
2928 display backlights on shutdown/boot.
2930 * udev learned a new SECLABEL{} construct to label device
2931 nodes with a specific security label when they appear. For
2932 now, only SECLABEL{selinux} is supported, but the syntax is
2933 prepared for additional security frameworks.
2935 * udev gained a new scheme to configure link-level attributes
2936 from files in /etc/systemd/network/*.link. These files can
2937 match against MAC address, device path, driver name and type,
2938 and will apply attributes like the naming policy, link speed,
2939 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
2940 address assignment policy (randomized, ...).
2942 * The configuration of network interface naming rules for
2943 "permanent interface names" has changed: a new NamePolicy=
2944 setting in the [Link] section of .link files determines the
2945 priority of possible naming schemes (onboard, slot, MAC,
2946 path). The default value of this setting is determined by
2947 /usr/lib/net/links/99-default.link. Old
2948 80-net-name-slot.rules udev configuration file has been
2949 removed, so local configuration overriding this file should
2950 be adapated to override 99-default.link instead.
2952 * When the User= switch is used in a unit file, also
2953 initialize $SHELL= based on the user database entry.
2955 * systemd no longer depends on libdbus. All communication is
2956 now done with sd-bus, systemd's low-level bus library
2959 * kdbus support has been added to PID 1 itself. When kdbus is
2960 enabled, this causes PID 1 to set up the system bus and
2961 enable support for a new ".busname" unit type that
2962 encapsulates bus name activation on kdbus. It works a little
2963 bit like ".socket" units, except for bus names. A new
2964 generator has been added that converts classic dbus1 service
2965 activation files automatically into native systemd .busname
2968 * sd-bus: add a light-weight vtable implementation that allows
2969 defining objects on the bus with a simple static const
2970 vtable array of its methods, signals and properties.
2972 * systemd will not generate or install static dbus
2973 introspection data anymore to /usr/share/dbus-1/interfaces,
2974 as the precise format of these files is unclear, and
2975 nothing makes use of it.
2977 * A proxy daemon is now provided to proxy clients connecting
2978 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
2979 compatibility with classic D-Bus.
2981 * A bus driver implementation has been added that supports the
2982 classic D-Bus bus driver calls on kdbus, also for
2983 compatibility purposes.
2985 * A new API "sd-event.h" has been added that implements a
2986 minimal event loop API built around epoll. It provides a
2987 couple of features that direct epoll usage is lacking:
2988 prioritization of events, scales to large numbers of timer
2989 events, per-event timer slack (accuracy), system-wide
2990 coalescing of timer events, exit handlers, watchdog
2991 supervision support using systemd's sd_notify() API, child
2994 * A new API "sd-rntl.h" has been added that provides an API
2995 around the route netlink interface of the kernel, similar in
2996 style to "sd-bus.h".
2998 * A new API "sd-dhcp-client.h" has been added that provides a
2999 small DHCPv4 client-side implementation. This is used by
3002 * There is a new kernel command line option
3003 "systemd.restore_state=0|1". When set to "0", none of the
3004 systemd tools will restore saved runtime state to hardware
3005 devices. More specifically, the rfkill and backlight states
3008 * The FsckPassNo= compatibility option in mount/service units
3009 has been removed. The fstab generator will now add the
3010 necessary dependencies automatically, and does not require
3011 PID1's support for that anymore.
3013 * journalctl gained a new switch, --list-boots, that lists
3014 recent boots with their times and boot IDs.
3016 * The various tools like systemctl, loginctl, timedatectl,
3017 busctl, systemd-run, ... have gained a new switch "-M" to
3018 connect to a specific, local OS container (as direct
3019 connection, without requiring SSH). This works on any
3020 container that is registered with machined, such as those
3021 created by libvirt-lxc or nspawn.
3023 * systemd-run and systemd-analyze also gained support for "-H"
3024 to connect to remote hosts via SSH. This is particularly
3025 useful for systemd-run because it enables queuing of jobs
3026 onto remote systems.
3028 * machinectl gained a new command "login" to open a getty
3029 login in any local container. This works with any container
3030 that is registered with machined (such as those created by
3031 libvirt-lxc or nspawn), and which runs systemd inside.
3033 * machinectl gained a new "reboot" command that may be used to
3034 trigger a reboot on a specific container that is registered
3035 with machined. This works on any container that runs an init
3036 system of some kind.
3038 * systemctl gained a new "list-timers" command to print a nice
3039 listing of installed timer units with the times they elapse
3042 * Alternative reboot() parameters may now be specified on the
3043 "systemctl reboot" command line and are passed to the
3044 reboot() system call.
3046 * systemctl gained a new --job-mode= switch to configure the
3047 mode to queue a job with. This is a more generic version of
3048 --fail, --irreversible, and --ignore-dependencies, which are
3049 still available but not advertised anymore.
3051 * /etc/systemd/system.conf gained new settings to configure
3052 various default timeouts of units, as well as the default
3053 start limit interval and burst. These may still be overridden
3056 * PID1 will now export on the bus profile data of the security
3057 policy upload process (such as the SELinux policy upload to
3060 * journald: when forwarding logs to the console, include
3061 timestamps (following the setting in
3062 /sys/module/printk/parameters/time).
3064 * OnCalendar= in timer units now understands the special
3065 strings "yearly" and "annually". (Both are equivalent)
3067 * The accuracy of timer units is now configurable with the new
3068 AccuracySec= setting. It defaults to 1min.
3070 * A new dependency type JoinsNamespaceOf= has been added that
3071 allows running two services within the same /tmp and network
3072 namespace, if PrivateNetwork= or PrivateTmp= are used.
3074 * A new command "cat" has been added to systemctl. It outputs
3075 the original unit file of a unit, and concatenates the
3076 contents of additional "drop-in" unit file snippets, so that
3077 the full configuration is shown.
3079 * systemctl now supports globbing on the various "list-xyz"
3080 commands, like "list-units" or "list-sockets", as well as on
3081 those commands which take multiple unit names.
3083 * journalctl's --unit= switch gained support for globbing.
3085 * All systemd daemons now make use of the watchdog logic so
3086 that systemd automatically notices when they hang.
3088 * If the $container_ttys environment variable is set,
3089 getty-generator will automatically spawn a getty for each
3090 listed tty. This is useful for container managers to request
3091 login gettys to be spawned on as many ttys as needed.
3093 * %h, %s, %U specifier support is not available anymore when
3094 used in unit files for PID 1. This is because NSS calls are
3095 not safe from PID 1. They stay available for --user
3096 instances of systemd, and as special case for the root user.
3098 * loginctl gained a new "--no-legend" switch to turn off output
3101 * The "sd-login.h" API gained three new calls:
3102 sd_session_is_remote(), sd_session_get_remote_user(),
3103 sd_session_get_remote_host() to query information about
3106 * The udev hardware database now also carries vendor/product
3107 information of SDIO devices.
3109 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
3110 determine whether watchdog notifications are requested by
3113 * Socket-activated per-connection services now include a
3114 short description of the connection parameters in the
3117 * tmpfiles gained a new "--boot" option. When this is not used,
3118 only lines where the command character is not suffixed with
3119 "!" are executed. When this option is specified, those
3120 options are executed too. This partitions tmpfiles
3121 directives into those that can be safely executed at any
3122 time, and those which should be run only at boot (for
3123 example, a line that creates /run/nologin).
3125 * A new API "sd-resolve.h" has been added which provides a simple
3126 asynchronous wrapper around glibc NSS host name resolution
3127 calls, such as getaddrinfo(). In contrast to glibc's
3128 getaddrinfo_a(), it does not use signals. In contrast to most
3129 other asynchronous name resolution libraries, this one does
3130 not reimplement DNS, but reuses NSS, so that alternate
3131 host name resolution systems continue to work, such as mDNS,
3132 LDAP, etc. This API is based on libasyncns, but it has been
3133 cleaned up for inclusion in systemd.
3135 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
3136 "sd-daemon.h" are no longer found in individual libraries
3137 libsystemd-journal.so, libsystemd-login.so,
3138 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
3139 merged them into a single library, libsystemd.so, which
3140 provides all symbols. The reason for this is cyclic
3141 dependencies, as these libraries tend to use each other's
3142 symbols. So far, we have managed to workaround that by linking
3143 a copy of a good part of our code into each of these
3144 libraries again and again, which, however, makes certain
3145 things hard to do, like sharing static variables. Also, it
3146 substantially increases footprint. With this change, there
3147 is only one library for the basic APIs systemd
3148 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
3149 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
3150 library as well, however are subject to the --enable-kdbus
3151 switch (see below). Note that "sd-dhcp-client.h" is not part
3152 of this library (this is because it only consumes, never
3153 provides, services of/to other APIs). To make the transition
3154 easy from the separate libraries to the unified one, we
3155 provide the --enable-compat-libs compile-time switch which
3156 will generate stub libraries that are compatible with the
3157 old ones but redirect all calls to the new one.
3159 * All of the kdbus logic and the new APIs "sd-bus.h",
3160 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
3161 and "sd-utf8.h" are compile-time optional via the
3162 "--enable-kdbus" switch, and they are not compiled in by
3163 default. To make use of kdbus, you have to explicitly enable
3164 the switch. Note however, that neither the kernel nor the
3165 userspace API for all of this is considered stable yet. We
3166 want to maintain the freedom to still change the APIs for
3167 now. By specifying this build-time switch, you acknowledge
3168 that you are aware of the instability of the current
3171 * Also, note that while kdbus is pretty much complete,
3172 it lacks one thing: proper policy support. This means you
3173 can build a fully working system with all features; however,
3174 it will be highly insecure. Policy support will be added in
3175 one of the next releases, at the same time that we will
3176 declare the APIs stable.
3178 * When the kernel command line argument "kdbus" is specified,
3179 systemd will automatically load the kdbus.ko kernel module. At
3180 this stage of development, it is only useful for testing kdbus
3181 and should not be used in production. Note: if "--enable-kdbus"
3182 is specified, and the kdbus.ko kernel module is available, and
3183 "kdbus" is added to the kernel command line, the entire system
3184 runs with kdbus instead of dbus-daemon, with the above mentioned
3185 problem of missing the system policy enforcement. Also a future
3186 version of kdbus.ko or a newer systemd will not be compatible with
3187 each other, and will unlikely be able to boot the machine if only
3188 one of them is updated.
3190 * systemctl gained a new "import-environment" command which
3191 uploads the caller's environment (or parts thereof) into the
3192 service manager so that it is inherited by services started
3193 by the manager. This is useful to upload variables like
3194 $DISPLAY into the user service manager.
3196 * A new PrivateDevices= switch has been added to service units
3197 which allows running a service with a namespaced /dev
3198 directory that does not contain any device nodes for
3199 physical devices. More specifically, it only includes devices
3200 such as /dev/null, /dev/urandom, and /dev/zero which are API
3203 * logind has been extended to support behaviour like VT
3204 switching on seats that do not support a VT. This makes
3205 multi-session available on seats that are not the first seat
3206 (seat0), and on systems where kernel support for VTs has
3207 been disabled at compile-time.
3209 * If a process holds a delay lock for system sleep or shutdown
3210 and fails to release it in time, we will now log its
3211 identity. This makes it easier to identify processes that
3212 cause slow suspends or power-offs.
3214 * When parsing /etc/crypttab, support for a new key-slot=
3215 option as supported by Debian is added. It allows indicating
3216 which LUKS slot to use on disk, speeding up key loading.
3218 * The sd_journald_sendv() API call has been checked and
3219 officially declared to be async-signal-safe so that it may
3220 be invoked from signal handlers for logging purposes.
3222 * Boot-time status output is now enabled automatically after a
3223 short timeout if boot does not progress, in order to give
3224 the user an indication what she or he is waiting for.
3226 * The boot-time output has been improved to show how much time
3227 remains until jobs expire.
3229 * The KillMode= switch in service units gained a new possible
3230 value "mixed". If set, and the unit is shut down, then the
3231 initial SIGTERM signal is sent only to the main daemon
3232 process, while the following SIGKILL signal is sent to
3233 all remaining processes of the service.
3235 * When a scope unit is registered, a new property "Controller"
3236 may be set. If set to a valid bus name, systemd will send a
3237 RequestStop() signal to this name when it would like to shut
3238 down the scope. This may be used to hook manager logic into
3239 the shutdown logic of scope units. Also, scope units may now
3240 be put in a special "abandoned" state, in which case the
3241 manager process which created them takes no further
3242 responsibilities for it.
3244 * When reading unit files, systemd will now verify
3245 the access mode of these files, and warn about certain
3246 suspicious combinations. This has been added to make it
3247 easier to track down packaging bugs where unit files are
3248 marked executable or world-writable.
3250 * systemd-nspawn gained a new "--setenv=" switch to set
3251 container-wide environment variables. The similar option in
3252 systemd-activate was renamed from "--environment=" to
3253 "--setenv=" for consistency.
3255 * systemd-nspawn has been updated to create a new kdbus domain
3256 for each container that is invoked, thus allowing each
3257 container to have its own set of system and user buses,
3258 independent of the host.
3260 * systemd-nspawn gained a new --drop-capability= switch to run
3261 the container with less capabilities than the default. Both
3262 --drop-capability= and --capability= now take the special
3263 string "all" for dropping or keeping all capabilities.
3265 * systemd-nspawn gained new switches for executing containers
3266 with specific SELinux labels set.
3268 * systemd-nspawn gained a new --quiet switch to not generate
3269 any additional output but the container's own console
3272 * systemd-nspawn gained a new --share-system switch to run a
3273 container without PID namespacing enabled.
3275 * systemd-nspawn gained a new --register= switch to control
3276 whether the container is registered with systemd-machined or
3277 not. This is useful for containers that do not run full
3278 OS images, but only specific apps.
3280 * systemd-nspawn gained a new --keep-unit which may be used
3281 when invoked as the only program from a service unit, and
3282 results in registration of the unit service itself in
3283 systemd-machined, instead of a newly opened scope unit.
3285 * systemd-nspawn gained a new --network-interface= switch for
3286 moving arbitrary interfaces to the container. The new
3287 --network-veth switch creates a virtual Ethernet connection
3288 between host and container. The new --network-bridge=
3289 switch then allows assigning the host side of this virtual
3290 Ethernet connection to a bridge device.
3292 * systemd-nspawn gained a new --personality= switch for
3293 setting the kernel personality for the container. This is
3294 useful when running a 32-bit container on a 64-bit host. A
3295 similar option Personality= is now also available for service
3298 * logind will now also track a "Desktop" identifier for each
3299 session which encodes the desktop environment of it. This is
3300 useful for desktop environments that want to identify
3301 multiple running sessions of itself easily.
3303 * A new SELinuxContext= setting for service units has been
3304 added that allows setting a specific SELinux execution
3305 context for a service.
3307 * Most systemd client tools will now honour $SYSTEMD_LESS for
3308 settings of the "less" pager. By default, these tools will
3309 override $LESS to allow certain operations to work, such as
3310 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
3311 influence this logic.
3313 * systemd's "seccomp" hook-up has been changed to make use of
3314 the libseccomp library instead of using its own
3315 implementation. This has benefits for portability among
3318 * For usage together with SystemCallFilter=, a new
3319 SystemCallErrorNumber= setting has been introduced that
3320 allows configuration of a system error number to be returned
3321 on filtered system calls, instead of immediately killing the
3322 process. Also, SystemCallArchitectures= has been added to
3323 limit access to system calls of a particular architecture
3324 (in order to turn off support for unused secondary
3325 architectures). There is also a global
3326 SystemCallArchitectures= setting in system.conf now to turn
3327 off support for non-native system calls system-wide.
3329 * systemd requires a kernel with a working name_to_handle_at(),
3330 please see the kernel config requirements in the README file.
3332 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
3333 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
3334 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
3335 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
3336 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
3337 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
3338 Elia Pinto, Florian Weimer, George McCollister, Goffredo
3339 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
3340 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
3341 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
3342 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
3343 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
3344 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
3345 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
3346 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
3347 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
3348 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
3349 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
3350 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
3351 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
3352 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
3353 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
3354 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
3355 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
3357 -- Berlin, 2014-02-20
3361 * logind has gained support for facilitating privileged input
3362 and drm device access for unprivileged clients. This work is
3363 useful to allow Wayland display servers (and similar
3364 programs, such as kmscon) to run under the user's ID and
3365 access input and drm devices which are normally
3366 protected. When this is used (and the kernel is new enough)
3367 logind will "mute" IO on the file descriptors passed to
3368 Wayland as long as it is in the background and "unmute" it
3369 if it returns into the foreground. This allows secure
3370 session switching without allowing background sessions to
3371 eavesdrop on input and display data. This also introduces
3372 session switching support if VT support is turned off in the
3373 kernel, and on seats that are not seat0.
3375 * A new kernel command line option luks.options= is understood
3376 now which allows specifying LUKS options for usage for LUKS
3377 encrypted partitions specified with luks.uuid=.
3379 * tmpfiles.d(5) snippets may now use specifier expansion in
3380 path names. More specifically %m, %b, %H, %v, are now
3381 replaced by the local machine id, boot id, hostname, and
3382 kernel version number.
3384 * A new tmpfiles.d(5) command "m" has been introduced which
3385 may be used to change the owner/group/access mode of a file
3386 or directory if it exists, but do nothing if it does not.
3388 * This release removes high-level support for the
3389 MemorySoftLimit= cgroup setting. The underlying kernel
3390 cgroup attribute memory.soft_limit= is currently badly
3391 designed and likely to be removed from the kernel API in its
3392 current form, hence we should not expose it for now.
3394 * The memory.use_hierarchy cgroup attribute is now enabled for
3395 all cgroups systemd creates in the memory cgroup
3396 hierarchy. This option is likely to be come the built-in
3397 default in the kernel anyway, and the non-hierarchical mode
3398 never made much sense in the intrinsically hierarchical
3401 * A new field _SYSTEMD_SLICE= is logged along with all journal
3402 messages containing the slice a message was generated
3403 from. This is useful to allow easy per-customer filtering of
3404 logs among other things.
3406 * systemd-journald will no longer adjust the group of journal
3407 files it creates to the "systemd-journal" group. Instead we
3408 rely on the journal directory to be owned by the
3409 "systemd-journal" group, and its setgid bit set, so that the
3410 kernel file system layer will automatically enforce that
3411 journal files inherit this group assignment. The reason for
3412 this change is that we cannot allow NSS look-ups from
3413 journald which would be necessary to resolve
3414 "systemd-journal" to a numeric GID, because this might
3415 create deadlocks if NSS involves synchronous queries to
3416 other daemons (such as nscd, or sssd) which in turn are
3417 logging clients of journald and might block on it, which
3418 would then dead lock. A tmpfiles.d(5) snippet included in
3419 systemd will make sure the setgid bit and group are
3420 properly set on the journal directory if it exists on every
3421 boot. However, we recommend adjusting it manually after
3422 upgrades too (or from RPM scriptlets), so that the change is
3423 not delayed until next reboot.
3425 * Backlight and random seed files in /var/lib/ have moved into
3426 the /var/lib/systemd/ directory, in order to centralize all
3427 systemd generated files in one directory.
3429 * Boot time performance measurements (as displayed by
3430 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
3431 performance information if that's available to determine how
3432 much time BIOS and boot loader initialization required. With
3433 a sufficiently new BIOS you hence no longer need to boot
3434 with Gummiboot to get access to such information.
3436 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
3437 Cristian Rodríguez, Dave Reisner, David Herrmann, David
3438 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
3439 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
3440 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
3441 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
3442 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3444 -- Berlin, 2013-10-02
3448 * The Restart= option for services now understands a new
3449 on-watchdog setting, which will restart the service
3450 automatically if the service stops sending out watchdog keep
3451 alive messages (as configured with WatchdogSec=).
3453 * The getty generator (which is responsible for bringing up a
3454 getty on configured serial consoles) will no longer only
3455 start a getty on the primary kernel console but on all
3456 others, too. This makes the order in which console= is
3457 specified on the kernel command line less important.
3459 * libsystemd-logind gained a new sd_session_get_vt() call to
3460 retrieve the VT number of a session.
3462 * If the option "tries=0" is set for an entry of /etc/crypttab
3463 its passphrase is queried indefinitely instead of any
3464 maximum number of tries.
3466 * If a service with a configure PID file terminates its PID
3467 file will now be removed automatically if it still exists
3468 afterwards. This should put an end to stale PID files.
3470 * systemd-run will now also take relative binary path names
3471 for execution and no longer insists on absolute paths.
3473 * InaccessibleDirectories= and ReadOnlyDirectories= now take
3474 paths that are optionally prefixed with "-" to indicate that
3475 it should not be considered a failure if they do not exist.
3477 * journalctl -o (and similar commands) now understands a new
3478 output mode "short-precise", it is similar to "short" but
3479 shows timestamps with usec accuracy.
3481 * The option "discard" (as known from Debian) is now
3482 synonymous to "allow-discards" in /etc/crypttab. In fact,
3483 "discard" is preferred now (since it is easier to remember
3486 * Some licensing clean-ups were made, so that more code is now
3487 LGPL-2.1 licensed than before.
3489 * A minimal tool to save/restore the display backlight
3490 brightness across reboots has been added. It will store the
3491 backlight setting as late as possible at shutdown, and
3492 restore it as early as possible during reboot.
3494 * A logic to automatically discover and enable home and swap
3495 partitions on GPT disks has been added. With this in place
3496 /etc/fstab becomes optional for many setups as systemd can
3497 discover certain partitions located on the root disk
3498 automatically. Home partitions are recognized under their
3499 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
3500 partitions are recognized under their GPT type ID
3501 0657fd6da4ab43c484e50933c84b4f4f.
3503 * systemd will no longer pass any environment from the kernel
3504 or initrd to system services. If you want to set an
3505 environment for all services, do so via the kernel command
3506 line systemd.setenv= assignment.
3508 * The systemd-sysctl tool no longer natively reads the file
3509 /etc/sysctl.conf. If desired, the file should be symlinked
3510 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
3511 legacy support by a symlink rather than built-in code, it
3512 also makes the otherwise hidden order of application of the
3513 different files visible. (Note that this partly reverts to a
3514 pre-198 application order of sysctl knobs!)
3516 * The "systemctl set-log-level" and "systemctl dump" commands
3517 have been moved to systemd-analyze.
3519 * systemd-run learned the new --remain-after-exit switch,
3520 which causes the scope unit not to be cleaned up
3521 automatically after the process terminated.
3523 * tmpfiles learned a new --exclude-prefix= switch to exclude
3524 certain paths from operation.
3526 * journald will now automatically flush all messages to disk
3527 as soon as a message at the log level CRIT, ALERT or EMERG
3530 Contributions from: Andrew Cook, Brandon Philips, Christian
3531 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
3532 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
3533 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
3534 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
3535 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
3536 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
3537 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
3538 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
3539 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
3540 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
3541 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
3542 William Giokas, Zbigniew Jędrzejewski-Szmek
3544 -- Berlin, 2013-09-13
3548 * The documentation has been updated to cover the various new
3549 concepts introduced with 205.
3551 * Unit files now understand the new %v specifier which
3552 resolves to the kernel version string as returned by "uname
3555 * systemctl now supports filtering the unit list output by
3556 load state, active state and sub state, using the new
3559 * "systemctl status" will now show the results of the
3560 condition checks (like ConditionPathExists= and similar) of
3561 the last start attempts of the unit. They are also logged to
3564 * "journalctl -b" may now be used to look for boot output of a
3565 specific boot. Try "journalctl -b -1" for the previous boot,
3566 but the syntax is substantially more powerful.
3568 * "journalctl --show-cursor" has been added which prints the
3569 cursor string the last shown log line. This may then be used
3570 with the new "journalctl --after-cursor=" switch to continue
3571 browsing logs from that point on.
3573 * "journalctl --force" may now be used to force regeneration
3576 * Creation of "dead" device nodes has been moved from udev
3577 into kmod and tmpfiles. Previously, udev would read the kmod
3578 databases to pre-generate dead device nodes based on meta
3579 information contained in kernel modules, so that these would
3580 be auto-loaded on access rather then at boot. As this
3581 does not really have much to do with the exposing actual
3582 kernel devices to userspace this has always been slightly
3583 alien in the udev codebase. Following the new scheme kmod
3584 will now generate a runtime snippet for tmpfiles from the
3585 module meta information and it now is tmpfiles' job to the
3586 create the nodes. This also allows overriding access and
3587 other parameters for the nodes using the usual tmpfiles
3588 facilities. As side effect this allows us to remove the
3589 CAP_SYS_MKNOD capability bit from udevd entirely.
3591 * logind's device ACLs may now be applied to these "dead"
3592 devices nodes too, thus finally allowing managed access to
3593 devices such as /dev/snd/sequencer whithout loading the
3594 backing module right-away.
3596 * A new RPM macro has been added that may be used to apply
3597 tmpfiles configuration during package installation.
3599 * systemd-detect-virt and ConditionVirtualization= now can
3600 detect User-Mode-Linux machines (UML).
3602 * journald will now implicitly log the effective capabilities
3603 set of processes in the message metadata.
3605 * systemd-cryptsetup has gained support for TrueCrypt volumes.
3607 * The initrd interface has been simplified (more specifically,
3608 support for passing performance data via environment
3609 variables and fsck results via files in /run has been
3610 removed). These features were non-essential, and are
3611 nowadays available in a much nicer way by having systemd in
3612 the initrd serialize its state and have the hosts systemd
3613 deserialize it again.
3615 * The udev "keymap" data files and tools to apply keyboard
3616 specific mappings of scan to key codes, and force-release
3617 scan code lists have been entirely replaced by a udev
3618 "keyboard" builtin and a hwdb data file.
3620 * systemd will now honour the kernel's "quiet" command line
3621 argument also during late shutdown, resulting in a
3622 completely silent shutdown when used.
3624 * There's now an option to control the SO_REUSEPORT socket
3625 option in .socket units.
3627 * Instance units will now automatically get a per-template
3628 subslice of system.slice unless something else is explicitly
3629 configured. For example, instances of sshd@.service will now
3630 implicitly be placed in system-sshd.slice rather than
3631 system.slice as before.
3633 * Test coverage support may now be enabled at build time.
3635 Contributions from: Dave Reisner, Frederic Crozat, Harald
3636 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
3637 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
3638 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
3639 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
3640 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
3641 Giokas, Zbigniew Jędrzejewski-Szmek
3643 -- Berlin, 2013-07-23
3647 * Two new unit types have been introduced:
3649 Scope units are very similar to service units, however, are
3650 created out of pre-existing processes -- instead of PID 1
3651 forking off the processes. By using scope units it is
3652 possible for system services and applications to group their
3653 own child processes (worker processes) in a powerful way
3654 which then maybe used to organize them, or kill them
3655 together, or apply resource limits on them.
3657 Slice units may be used to partition system resources in an
3658 hierarchical fashion and then assign other units to them. By
3659 default there are now three slices: system.slice (for all
3660 system services), user.slice (for all user sessions),
3661 machine.slice (for VMs and containers).
3663 Slices and scopes have been introduced primarily in
3664 context of the work to move cgroup handling to a
3665 single-writer scheme, where only PID 1
3666 creates/removes/manages cgroups.
3668 * There's a new concept of "transient" units. In contrast to
3669 normal units these units are created via an API at runtime,
3670 not from configuration from disk. More specifically this
3671 means it is now possible to run arbitrary programs as
3672 independent services, with all execution parameters passed
3673 in via bus APIs rather than read from disk. Transient units
3674 make systemd substantially more dynamic then it ever was,
3675 and useful as a general batch manager.
3677 * logind has been updated to make use of scope and slice units
3678 for managing user sessions. As a user logs in he will get
3679 his own private slice unit, to which all sessions are added
3680 as scope units. We also added support for automatically
3681 adding an instance of user@.service for the user into the
3682 slice. Effectively logind will no longer create cgroup
3683 hierarchies on its own now, it will defer entirely to PID 1
3684 for this by means of scope, service and slice units. Since
3685 user sessions this way become entities managed by PID 1
3686 the output of "systemctl" is now a lot more comprehensive.
3688 * A new mini-daemon "systemd-machined" has been added which
3689 may be used by virtualization managers to register local
3690 VMs/containers. nspawn has been updated accordingly, and
3691 libvirt will be updated shortly. machined will collect a bit
3692 of meta information about the VMs/containers, and assign
3693 them their own scope unit (see above). The collected
3694 meta-data is then made available via the "machinectl" tool,
3695 and exposed in "ps" and similar tools. machined/machinectl
3696 is compile-time optional.
3698 * As discussed earlier, the low-level cgroup configuration
3699 options ControlGroup=, ControlGroupModify=,
3700 ControlGroupPersistent=, ControlGroupAttribute= have been
3701 removed. Please use high-level attribute settings instead as
3702 well as slice units.
3704 * A new bus call SetUnitProperties() has been added to alter
3705 various runtime parameters of a unit. This is primarily
3706 useful to alter cgroup parameters dynamically in a nice way,
3707 but will be extended later on to make more properties
3708 modifiable at runtime. systemctl gained a new set-properties
3709 command that wraps this call.
3711 * A new tool "systemd-run" has been added which can be used to
3712 run arbitrary command lines as transient services or scopes,
3713 while configuring a number of settings via the command
3714 line. This tool is currently very basic, however already
3715 very useful. We plan to extend this tool to even allow
3716 queuing of execution jobs with time triggers from the
3717 command line, similar in fashion to "at".
3719 * nspawn will now inform the user explicitly that kernels with
3720 audit enabled break containers, and suggest the user to turn
3723 * Support for detecting the IMA and AppArmor security
3724 frameworks with ConditionSecurity= has been added.
3726 * journalctl gained a new "-k" switch for showing only kernel
3727 messages, mimicking dmesg output; in addition to "--user"
3728 and "--system" switches for showing only user's own logs
3731 * systemd-delta can now show information about drop-in
3732 snippets extending unit files.
3734 * libsystemd-bus has been substantially updated but is still
3735 not available as public API.
3737 * systemd will now look for the "debug" argument on the kernel
3738 command line and enable debug logging, similar to what
3739 "systemd.log_level=debug" already did before.
3741 * "systemctl set-default", "systemctl get-default" has been
3742 added to configure the default.target symlink, which
3743 controls what to boot into by default.
3745 * "systemctl set-log-level" has been added as a convenient
3746 way to raise and lower systemd logging threshold.
3748 * "systemd-analyze plot" will now show the time the various
3749 generators needed for execution, as well as information
3750 about the unit file loading.
3752 * libsystemd-journal gained a new sd_journal_open_files() call
3753 for opening specific journal files. journactl also gained a
3754 new switch to expose this new functionality. Previously we
3755 only supported opening all files from a directory, or all
3756 files from the system, as opening individual files only is
3757 racy due to journal file rotation.
3759 * systemd gained the new DefaultEnvironment= setting in
3760 /etc/systemd/system.conf to set environment variables for
3763 * If a privileged process logs a journal message with the
3764 OBJECT_PID= field set, then journald will automatically
3765 augment this with additional OBJECT_UID=, OBJECT_GID=,
3766 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
3767 system services want to log events about specific client
3768 processes. journactl/systemctl has been updated to make use
3769 of this information if all log messages regarding a specific
3772 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
3773 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
3774 Reisner, David Coppa, David King, David Strauss, Eelco
3775 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
3776 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
3777 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
3778 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
3779 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
3780 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
3781 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
3782 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
3783 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
3784 Łukasz Stelmach, 장동준
3788 * The Python bindings gained some minimal support for the APIs
3789 exposed by libsystemd-logind.
3791 * ConditionSecurity= gained support for detecting SMACK. Since
3792 this condition already supports SELinux and AppArmor we only
3793 miss IMA for this. Patches welcome!
3795 Contributions from: Karol Lewandowski, Lennart Poettering,
3796 Zbigniew Jędrzejewski-Szmek
3800 * systemd-nspawn will now create /etc/resolv.conf if
3801 necessary, before bind-mounting the host's file onto it.
3803 * systemd-nspawn will now store meta information about a
3804 container on the container's cgroup as extended attribute
3805 fields, including the root directory.
3807 * The cgroup hierarchy has been reworked in many ways. All
3808 objects any of the components systemd creates in the cgroup
3809 tree are now suffixed. More specifically, user sessions are
3810 now placed in cgroups suffixed with ".session", users in
3811 cgroups suffixed with ".user", and nspawn containers in
3812 cgroups suffixed with ".nspawn". Furthermore, all cgroup
3813 names are now escaped in a simple scheme to avoid collision
3814 of userspace object names with kernel filenames. This work
3815 is preparation for making these objects relocatable in the
3816 cgroup tree, in order to allow easy resource partitioning of
3817 these objects without causing naming conflicts.
3819 * systemctl list-dependencies gained the new switches
3820 --plain, --reverse, --after and --before.
3822 * systemd-inhibit now shows the process name of processes that
3823 have taken an inhibitor lock.
3825 * nss-myhostname will now also resolve "localhost"
3826 implicitly. This makes /etc/hosts an optional file and
3827 nicely handles that on IPv6 ::1 maps to both "localhost" and
3830 * libsystemd-logind.so gained a new call
3831 sd_get_machine_names() to enumerate running containers and
3832 VMs (currently only supported by very new libvirt and
3833 nspawn). sd_login_monitor can now be used to watch
3834 VMs/containers coming and going.
3836 * .include is not allowed recursively anymore, and only in
3837 unit files. Usually it is better to use drop-in snippets in
3838 .d/*.conf anyway, as introduced with systemd 198.
3840 * systemd-analyze gained a new "critical-chain" command that
3841 determines the slowest chain of units run during system
3842 boot-up. It is very useful for tracking down where
3843 optimizing boot time is the most beneficial.
3845 * systemd will no longer allow manipulating service paths in
3846 the name=systemd:/system cgroup tree using ControlGroup= in
3847 units. (But is still fine with it in all other dirs.)
3849 * There's a new systemd-nspawn@.service service file that may
3850 be used to easily run nspawn containers as system
3851 services. With the container's root directory in
3852 /var/lib/container/foobar it is now sufficient to run
3853 "systemctl start systemd-nspawn@foobar.service" to boot it.
3855 * systemd-cgls gained a new parameter "--machine" to list only
3856 the processes within a certain container.
3858 * ConditionSecurity= now can check for "apparmor". We still
3859 are lacking checks for SMACK and IMA for this condition
3860 check though. Patches welcome!
3862 * A new configuration file /etc/systemd/sleep.conf has been
3863 added that may be used to configure which kernel operation
3864 systemd is supposed to execute when "suspend", "hibernate"
3865 or "hybrid-sleep" is requested. This makes the new kernel
3866 "freeze" state accessible to the user.
3868 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
3869 the passed argument if applicable.
3871 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
3872 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
3873 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
3874 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
3875 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
3876 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
3877 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
3882 * The output of 'systemctl list-jobs' got some polishing. The
3883 '--type=' argument may now be passed more than once. A new
3884 command 'systemctl list-sockets' has been added which shows
3885 a list of kernel sockets systemd is listening on with the
3886 socket units they belong to, plus the units these socket
3889 * The experimental libsystemd-bus library got substantial
3890 updates to work in conjunction with the (also experimental)
3891 kdbus kernel project. It works well enough to exchange
3892 messages with some sophistication. Note that kdbus is not
3893 ready yet, and the library is mostly an elaborate test case
3894 for now, and not installable.
3896 * systemd gained a new unit 'systemd-static-nodes.service'
3897 that generates static device nodes earlier during boot, and
3898 can run in conjunction with udev.
3900 * libsystemd-login gained a new call sd_pid_get_user_unit()
3901 to retrieve the user systemd unit a process is running
3902 in. This is useful for systems where systemd is used as
3905 * systemd-nspawn now places all containers in the new /machine
3906 top-level cgroup directory in the name=systemd
3907 hierarchy. libvirt will soon do the same, so that we get a
3908 uniform separation of /system, /user and /machine for system
3909 services, user processes and containers/virtual
3910 machines. This new cgroup hierarchy is also useful to stick
3911 stable names to specific container instances, which can be
3912 recognized later this way (this name may be controlled
3913 via systemd-nspawn's new -M switch). libsystemd-login also
3914 gained a new call sd_pid_get_machine_name() to retrieve the
3915 name of the container/VM a specific process belongs to.
3917 * bootchart can now store its data in the journal.
3919 * libsystemd-journal gained a new call
3920 sd_journal_add_conjunction() for AND expressions to the
3921 matching logic. This can be used to express more complex
3922 logical expressions.
3924 * journactl can now take multiple --unit= and --user-unit=
3927 * The cryptsetup logic now understands the "luks.key=" kernel
3928 command line switch for specifying a file to read the
3929 decryption key from. Also, if a configured key file is not
3930 found the tool will now automatically fall back to prompting
3933 * Python systemd.journal module was updated to wrap recently
3934 added functions from libsystemd-journal. The interface was
3935 changed to bring the low level interface in s.j._Reader
3936 closer to the C API, and the high level interface in
3937 s.j.Reader was updated to wrap and convert all data about
3940 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
3941 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
3942 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
3943 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
3944 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
3945 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3949 * journalctl --update-catalog now understands a new --root=
3950 option to operate on catalogs found in a different root
3953 * During shutdown after systemd has terminated all running
3954 services a final killing loop kills all remaining left-over
3955 processes. We will now print the name of these processes
3956 when we send SIGKILL to them, since this usually indicates a
3959 * If /etc/crypttab refers to password files stored on
3960 configured mount points automatic dependencies will now be
3961 generated to ensure the specific mount is established first
3962 before the key file is attempted to be read.
3964 * 'systemctl status' will now show information about the
3965 network sockets a socket unit is listening on.
3967 * 'systemctl status' will also shown information about any
3968 drop-in configuration file for units. (Drop-In configuration
3969 files in this context are files such as
3970 /etc/systemd/systemd/foobar.service.d/*.conf)
3972 * systemd-cgtop now optionally shows summed up CPU times of
3973 cgroups. Press '%' while running cgtop to switch between
3974 percentage and absolute mode. This is useful to determine
3975 which cgroups use up the most CPU time over the entire
3976 runtime of the system. systemd-cgtop has also been updated
3977 to be 'pipeable' for processing with further shell tools.
3979 * 'hostnamectl set-hostname' will now allow setting of FQDN
3982 * The formatting and parsing of time span values has been
3983 changed. The parser now understands fractional expressions
3984 such as "5.5h". The formatter will now output fractional
3985 expressions for all time spans under 1min, i.e. "5.123456s"
3986 rather than "5s 123ms 456us". For time spans under 1s
3987 millisecond values are shown, for those under 1ms
3988 microsecond values are shown. This should greatly improve
3989 all time-related output of systemd.
3991 * libsystemd-login and libsystemd-journal gained new
3992 functions for querying the poll() events mask and poll()
3993 timeout value for integration into arbitrary event
3996 * localectl gained the ability to list available X11 keymaps
3997 (models, layouts, variants, options).
3999 * 'systemd-analyze dot' gained the ability to filter for
4000 specific units via shell-style globs, to create smaller,
4001 more useful graphs. I.e. it is now possible to create simple
4002 graphs of all the dependencies between only target units, or
4003 of all units that Avahi has dependencies with.
4005 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
4006 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
4007 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
4008 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
4009 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
4010 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
4011 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
4015 * The boot-time readahead implementation for rotating media
4016 will now read the read-ahead data in multiple passes which
4017 consist of all read requests made in equidistant time
4018 intervals. This means instead of strictly reading read-ahead
4019 data in its physical order on disk we now try to find a
4020 middle ground between physical and access time order.
4022 * /etc/os-release files gained a new BUILD_ID= field for usage
4023 on operating systems that provide continuous builds of OS
4026 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
4027 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
4028 William Douglas, Zbigniew Jędrzejewski-Szmek
4032 * systemd-python gained an API exposing libsystemd-daemon.
4034 * The SMACK setup logic gained support for uploading CIPSO
4037 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
4038 ReadOnlyDirectories= and InaccessibleDirectories= has
4039 changed. The private /tmp and /var/tmp directories are now
4040 shared by all processes of a service (which means
4041 ExecStartPre= may now leave data in /tmp that ExecStart= of
4042 the same service can still access). When a service is
4043 stopped its temporary directories are immediately deleted
4044 (normal clean-up with tmpfiles is still done in addition to
4047 * By default, systemd will now set a couple of sysctl
4048 variables in the kernel: the safe sysrq options are turned
4049 on, IP route verification is turned on, and source routing
4050 disabled. The recently added hardlink and softlink
4051 protection of the kernel is turned on. These settings should
4052 be reasonably safe, and good defaults for all new systems.
4054 * The predictable network naming logic may now be turned off
4055 with a new kernel command line switch: net.ifnames=0.
4057 * A new libsystemd-bus module has been added that implements a
4058 pretty complete D-Bus client library. For details see:
4060 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
4062 * journald will now explicitly flush the journal files to disk
4063 at the latest 5min after each write. The file will then also
4064 be marked offline until the next write. This should increase
4065 reliability in case of a crash. The synchronization delay
4066 can be configured via SyncIntervalSec= in journald.conf.
4068 * There's a new remote-fs-setup.target unit that can be used
4069 to pull in specific services when at least one remote file
4070 system is to be mounted.
4072 * There are new targets timers.target and paths.target as
4073 canonical targets to pull user timer and path units in
4074 from. This complements sockets.target with a similar
4075 purpose for socket units.
4077 * libudev gained a new call udev_device_set_attribute_value()
4078 to set sysfs attributes of a device.
4080 * The udev daemon now sets the default number of worker
4081 processes executed in parallel based on the number of available
4082 CPUs instead of the amount of available RAM. This is supposed
4083 to provide a more reliable default and limit a too aggressive
4084 paralellism for setups with 1000s of devices connected.
4086 Contributions from: Auke Kok, Colin Walters, Cristian
4087 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
4088 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
4089 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
4090 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
4091 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
4092 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
4093 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
4094 Zbigniew Jędrzejewski-Szmek
4098 * Configuration of unit files may now be extended via drop-in
4099 files without having to edit/override the unit files
4100 themselves. More specifically, if the administrator wants to
4101 change one value for a service file foobar.service he can
4102 now do so by dropping in a configuration snippet into
4103 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
4104 will load all these snippets and apply them on top of the
4105 main unit configuration file, possibly extending or
4106 overriding its settings. Using these drop-in snippets is
4107 generally nicer than the two earlier options for changing
4108 unit files locally: copying the files from
4109 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
4110 them there; or creating a new file in /etc/systemd/system/
4111 that incorporates the original one via ".include". Drop-in
4112 snippets into these .d/ directories can be placed in any
4113 directory systemd looks for units in, and the usual
4114 overriding semantics between /usr/lib, /etc and /run apply
4117 * Most unit file settings which take lists of items can now be
4118 reset by assigning the empty string to them. For example,
4119 normally, settings such as Environment=FOO=BAR append a new
4120 environment variable assignment to the environment block,
4121 each time they are used. By assigning Environment= the empty
4122 string the environment block can be reset to empty. This is
4123 particularly useful with the .d/*.conf drop-in snippets
4124 mentioned above, since this adds the ability to reset list
4125 settings from vendor unit files via these drop-ins.
4127 * systemctl gained a new "list-dependencies" command for
4128 listing the dependencies of a unit recursively.
4130 * Inhibitors are now honored and listed by "systemctl
4131 suspend", "systemctl poweroff" (and similar) too, not only
4132 GNOME. These commands will also list active sessions by
4135 * Resource limits (as exposed by the various control group
4136 controllers) can now be controlled dynamically at runtime
4137 for all units. More specifically, you can now use a command
4138 like "systemctl set-cgroup-attr foobar.service cpu.shares
4139 2000" to alter the CPU shares a specific service gets. These
4140 settings are stored persistently on disk, and thus allow the
4141 administrator to easily adjust the resource usage of
4142 services with a few simple commands. This dynamic resource
4143 management logic is also available to other programs via the
4144 bus. Almost any kernel cgroup attribute and controller is
4147 * systemd-vconsole-setup will now copy all font settings to
4148 all allocated VTs, where it previously applied them only to
4151 * libsystemd-login gained the new sd_session_get_tty() API
4154 * This release drops support for a few legacy or
4155 distribution-specific LSB facility names when parsing init
4156 scripts: $x-display-manager, $mail-transfer-agent,
4157 $mail-transport-agent, $mail-transfer-agent, $smtp,
4158 $null. Also, the mail-transfer-agent.target unit backing
4159 this has been removed. Distributions which want to retain
4160 compatibility with this should carry the burden for
4161 supporting this themselves and patch support for these back
4162 in, if they really need to. Also, the facilities $syslog and
4163 $local_fs are now ignored, since systemd does not support
4164 early-boot LSB init scripts anymore, and these facilities
4165 are implied anyway for normal services. syslog.target has
4168 * There are new bus calls on PID1's Manager object for
4169 cancelling jobs, and removing snapshot units. Previously,
4170 both calls were only available on the Job and Snapshot
4173 * systemd-journal-gatewayd gained SSL support.
4175 * The various "environment" files, such as /etc/locale.conf
4176 now support continuation lines with a backslash ("\") as
4177 last character in the line, similarly in style (but different)
4178 to how this is supported in shells.
4180 * For normal user processes the _SYSTEMD_USER_UNIT= field is
4181 now implicitly appended to every log entry logged. systemctl
4182 has been updated to filter by this field when operating on a
4183 user systemd instance.
4185 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
4186 CAP_AUDIT_CONTROL capabilities to the capabilities set for
4187 the container. This makes it easier to boot unmodified
4188 Fedora systems in a container, which however still requires
4189 audit=0 to be passed on the kernel command line. Auditing in
4190 kernel and userspace is unfortunately still too broken in
4191 context of containers, hence we recommend compiling it out
4192 of the kernel or using audit=0. Hopefully this will be fixed
4193 one day for good in the kernel.
4195 * nspawn gained the new --bind= and --bind-ro= parameters to
4196 bind mount specific directories from the host into the
4199 * nspawn will now mount its own devpts file system instance
4200 into the container, in order not to leak pty devices from
4201 the host into the container.
4203 * systemd will now read the firmware boot time performance
4204 information from the EFI variables, if the used boot loader
4205 supports this, and takes it into account for boot performance
4206 analysis via "systemd-analyze". This is currently supported
4207 only in conjunction with Gummiboot, but could be supported
4208 by other boot loaders too. For details see:
4210 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
4212 * A new generator has been added that automatically mounts the
4213 EFI System Partition (ESP) to /boot, if that directory
4214 exists, is empty, and no other file system has been
4215 configured to be mounted there.
4217 * logind will now send out PrepareForSleep(false) out
4218 unconditionally, after coming back from suspend. This may be
4219 used by applications as asynchronous notification for
4220 system resume events.
4222 * "systemctl unlock-sessions" has been added, that allows
4223 unlocking the screens of all user sessions at once, similar
4224 to how "systemctl lock-sessions" already locked all users
4225 sessions. This is backed by a new D-Bus call UnlockSessions().
4227 * "loginctl seat-status" will now show the master device of a
4228 seat. (i.e. the device of a seat that needs to be around for
4229 the seat to be considered available, usually the graphics
4232 * tmpfiles gained a new "X" line type, that allows
4233 configuration of files and directories (with wildcards) that
4234 shall be excluded from automatic cleanup ("aging").
4236 * udev default rules set the device node permissions now only
4237 at "add" events, and do not change them any longer with a
4238 later "change" event.
4240 * The log messages for lid events and power/sleep keypresses
4241 now carry a message ID.
4243 * We now have a substantially larger unit test suite, but this
4244 continues to be work in progress.
4246 * udevadm hwdb gained a new --root= parameter to change the
4247 root directory to operate relative to.
4249 * logind will now issue a background sync() request to the kernel
4250 early at shutdown, so that dirty buffers are flushed to disk early
4251 instead of at the last moment, in order to optimize shutdown
4254 * A new bootctl tool has been added that is an interface for
4255 certain boot loader operations. This is currently a preview
4256 and is likely to be extended into a small mechanism daemon
4257 like timedated, localed, hostnamed, and can be used by
4258 graphical UIs to enumerate available boot options, and
4259 request boot into firmware operations.
4261 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
4262 the rest of the package. It also has been updated to work
4263 correctly in initrds.
4265 * Policykit previously has been runtime optional, and is now
4266 also compile time optional via a configure switch.
4268 * systemd-analyze has been reimplemented in C. Also "systemctl
4269 dot" has moved into systemd-analyze.
4271 * "systemctl status" with no further parameters will now print
4272 the status of all active or failed units.
4274 * Operations such as "systemctl start" can now be executed
4275 with a new mode "--irreversible" which may be used to queue
4276 operations that cannot accidentally be reversed by a later
4277 job queuing. This is by default used to make shutdown
4278 requests more robust.
4280 * The Python API of systemd now gained a new module for
4281 reading journal files.
4283 * A new tool kernel-install has been added that can install
4284 kernel images according to the Boot Loader Specification:
4286 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
4288 * Boot time console output has been improved to provide
4289 animated boot time output for hanging jobs.
4291 * A new tool systemd-activate has been added which can be used
4292 to test socket activation with, directly from the command
4293 line. This should make it much easier to test and debug
4294 socket activation in daemons.
4296 * journalctl gained a new "--reverse" (or -r) option to show
4297 journal output in reverse order (i.e. newest line first).
4299 * journalctl gained a new "--pager-end" (or -e) option to jump
4300 to immediately jump to the end of the journal in the
4301 pager. This is only supported in conjunction with "less".
4303 * journalctl gained a new "--user-unit=" option, that works
4304 similarly to "--unit=" but filters for user units rather than
4307 * A number of unit files to ease adoption of systemd in
4308 initrds has been added. This moves some minimal logic from
4309 the various initrd implementations into systemd proper.
4311 * The journal files are now owned by a new group
4312 "systemd-journal", which exists specifically to allow access
4313 to the journal, and nothing else. Previously, we used the
4314 "adm" group for that, which however possibly covers more
4315 than just journal/log file access. This new group is now
4316 already used by systemd-journal-gatewayd to ensure this
4317 daemon gets access to the journal files and as little else
4318 as possible. Note that "make install" will also set FS ACLs
4319 up for /var/log/journal to give "adm" and "wheel" read
4320 access to it, in addition to "systemd-journal" which owns
4321 the journal files. We recommend that packaging scripts also
4322 add read access to "adm" + "wheel" to /var/log/journal, and
4323 all existing/future journal files. To normal users and
4324 administrators little changes, however packagers need to
4325 ensure to create the "systemd-journal" system group at
4326 package installation time.
4328 * The systemd-journal-gatewayd now runs as unprivileged user
4329 systemd-journal-gateway:systemd-journal-gateway. Packaging
4330 scripts need to create these system user/group at
4333 * timedated now exposes a new boolean property CanNTP that
4334 indicates whether a local NTP service is available or not.
4336 * systemd-detect-virt will now also detect xen PVs
4338 * The pstore file system is now mounted by default, if it is
4341 * In addition to the SELinux and IMA policies we will now also
4342 load SMACK policies at early boot.
4344 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
4345 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
4346 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
4347 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
4348 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
4349 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
4350 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
4351 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
4352 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
4353 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
4354 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
4355 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
4356 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
4357 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
4361 * Timer units now support calendar time events in addition to
4362 monotonic time events. That means you can now trigger a unit
4363 based on a calendar time specification such as "Thu,Fri
4364 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
4365 or fifth day of any month of the year 2013, given that it is
4366 a thursday or friday. This brings timer event support
4367 considerably closer to cron's capabilities. For details on
4368 the supported calendar time specification language see
4371 * udev now supports a number of different naming policies for
4372 network interfaces for predictable names, and a combination
4373 of these policies is now the default. Please see this wiki
4374 document for details:
4376 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
4378 * Auke Kok's bootchart implementation has been added to the
4379 systemd tree. It is an optional component that can graph the
4380 boot in quite some detail. It is one of the best bootchart
4381 implementations around and minimal in its code and
4384 * nss-myhostname has been integrated into the systemd source
4385 tree. nss-myhostname guarantees that the local hostname
4386 always stays resolvable via NSS. It has been a weak
4387 requirement of systemd-hostnamed since a long time, and
4388 since its code is actually trivial we decided to just
4389 include it in systemd's source tree. It can be turned off
4390 with a configure switch.
4392 * The read-ahead logic is now capable of properly detecting
4393 whether a btrfs file system is on SSD or rotating media, in
4394 order to optimize the read-ahead scheme. Previously, it was
4395 only capable of detecting this on traditional file systems
4398 * In udev, additional device properties are now read from the
4399 IAB in addition to the OUI database. Also, Bluetooth company
4400 identities are attached to the devices as well.
4402 * In service files %U may be used as specifier that is
4403 replaced by the configured user name of the service.
4405 * nspawn may now be invoked without a controlling TTY. This
4406 makes it suitable for invocation as its own service. This
4407 may be used to set up a simple containerized server system
4408 using only core OS tools.
4410 * systemd and nspawn can now accept socket file descriptors
4411 when they are started for socket activation. This enables
4412 implementation of socket activated nspawn
4413 containers. i.e. think about autospawning an entire OS image
4414 when the first SSH or HTTP connection is received. We expect
4415 that similar functionality will also be added to libvirt-lxc
4418 * journalctl will now suppress ANSI color codes when
4419 presenting log data.
4421 * systemctl will no longer show control group information for
4422 a unit if a the control group is empty anyway.
4424 * logind can now automatically suspend/hibernate/shutdown the
4427 * /etc/machine-info and hostnamed now also expose the chassis
4428 type of the system. This can be used to determine whether
4429 the local system is a laptop, desktop, handset or
4430 tablet. This information may either be configured by the
4431 user/vendor or is automatically determined from ACPI and DMI
4432 information if possible.
4434 * A number of PolicyKit actions are now bound together with
4435 "imply" rules. This should simplify creating UIs because
4436 many actions will now authenticate similar ones as well.
4438 * Unit files learnt a new condition ConditionACPower= which
4439 may be used to conditionalize a unit depending on whether an
4440 AC power source is connected or not, of whether the system
4441 is running on battery power.
4443 * systemctl gained a new "is-failed" verb that may be used in
4444 shell scripts and suchlike to check whether a specific unit
4445 is in the "failed" state.
4447 * The EnvironmentFile= setting in unit files now supports file
4448 globbing, and can hence be used to easily read a number of
4449 environment files at once.
4451 * systemd will no longer detect and recognize specific
4452 distributions. All distribution-specific #ifdeffery has been
4453 removed, systemd is now fully generic and
4454 distribution-agnostic. Effectively, not too much is lost as
4455 a lot of the code is still accessible via explicit configure
4456 switches. However, support for some distribution specific
4457 legacy configuration file formats has been dropped. We
4458 recommend distributions to simply adopt the configuration
4459 files everybody else uses now and convert the old
4460 configuration from packaging scripts. Most distributions
4461 already did that. If that's not possible or desirable,
4462 distributions are welcome to forward port the specific
4463 pieces of code locally from the git history.
4465 * When logging a message about a unit systemd will now always
4466 log the unit name in the message meta data.
4468 * localectl will now also discover system locale data that is
4469 not stored in locale archives, but directly unpacked.
4471 * logind will no longer unconditionally use framebuffer
4472 devices as seat masters, i.e. as devices that are required
4473 to be existing before a seat is considered preset. Instead,
4474 it will now look for all devices that are tagged as
4475 "seat-master" in udev. By default, framebuffer devices will
4476 be marked as such, but depending on local systems, other
4477 devices might be marked as well. This may be used to
4478 integrate graphics cards using closed source drivers (such
4479 as NVidia ones) more nicely into logind. Note however, that
4480 we recommend using the open source NVidia drivers instead,
4481 and no udev rules for the closed-source drivers will be
4482 shipped from us upstream.
4484 Contributions from: Adam Williamson, Alessandro Crismani, Auke
4485 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
4486 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
4487 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
4488 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
4489 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
4490 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
4491 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
4492 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
4493 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
4494 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
4495 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
4500 * udev gained support for loading additional device properties
4501 from an indexed database that is keyed by vendor/product IDs
4502 and similar device identifiers. For the beginning this
4503 "hwdb" is populated with data from the well-known PCI and
4504 USB database, but also includes PNP, ACPI and OID data. In
4505 the longer run this indexed database shall grow into
4506 becoming the one central database for non-essential
4507 userspace device metadata. Previously, data from the PCI/USB
4508 database was only attached to select devices, since the
4509 lookup was a relatively expensive operation due to O(n) time
4510 complexity (with n being the number of entries in the
4511 database). Since this is now O(1), we decided to add in this
4512 data for all devices where this is available, by
4513 default. Note that the indexed database needs to be rebuilt
4514 when new data files are installed. To achieve this you need
4515 to update your packaging scripts to invoke "udevadm hwdb
4516 --update" after installation of hwdb data files. For
4517 RPM-based distributions we introduced the new
4518 %udev_hwdb_update macro for this purpose.
4520 * The Journal gained support for the "Message Catalog", an
4521 indexed database to link up additional information with
4522 journal entries. For further details please check:
4524 http://www.freedesktop.org/wiki/Software/systemd/catalog
4526 The indexed message catalog database also needs to be
4527 rebuilt after installation of message catalog files. Use
4528 "journalctl --update-catalog" for this. For RPM-based
4529 distributions we introduced the %journal_catalog_update
4530 macro for this purpose.
4532 * The Python Journal bindings gained support for the standard
4533 Python logging framework.
4535 * The Journal API gained new functions for checking whether
4536 the underlying file system of a journal file is capable of
4537 properly reporting file change notifications, or whether
4538 applications that want to reflect journal changes "live"
4539 need to recheck journal files continuously in appropriate
4542 * It is now possible to set the "age" field for tmpfiles
4543 entries to 0, indicating that files matching this entry
4544 shall always be removed when the directories are cleaned up.
4546 * coredumpctl gained a new "gdb" verb which invokes gdb
4547 right-away on the selected coredump.
4549 * There's now support for "hybrid sleep" on kernels that
4550 support this, in addition to "suspend" and "hibernate". Use
4551 "systemctl hybrid-sleep" to make use of this.
4553 * logind's HandleSuspendKey= setting (and related settings)
4554 now gained support for a new "lock" setting to simply
4555 request the screen lock on all local sessions, instead of
4556 actually executing a suspend or hibernation.
4558 * systemd will now mount the EFI variables file system by
4561 * Socket units now gained support for configuration of the
4562 SMACK security label.
4564 * timedatectl will now output the time of the last and next
4565 daylight saving change.
4567 * We dropped support for various legacy and distro-specific
4568 concepts, such as insserv, early-boot SysV services
4569 (i.e. those for non-standard runlevels such as 'b' or 'S')
4570 or ArchLinux /etc/rc.conf support. We recommend the
4571 distributions who still need support this to either continue
4572 to maintain the necessary patches downstream, or find a
4573 different solution. (Talk to us if you have questions!)
4575 * Various systemd components will now bypass PolicyKit checks
4576 for root and otherwise handle properly if PolicyKit is not
4577 found to be around. This should fix most issues for
4578 PolicyKit-less systems. Quite frankly this should have been
4579 this way since day one. It is absolutely our intention to
4580 make systemd work fine on PolicyKit-less systems, and we
4581 consider it a bug if something does not work as it should if
4582 PolicyKit is not around.
4584 * For embedded systems it is now possible to build udev and
4585 systemd without blkid and/or kmod support.
4587 * "systemctl switch-root" is now capable of switching root
4588 more than once. I.e. in addition to transitions from the
4589 initrd to the host OS it is now possible to transition to
4590 further OS images from the host. This is useful to implement
4591 offline updating tools.
4593 * Various other additions have been made to the RPM macros
4594 shipped with systemd. Use %udev_rules_update() after
4595 installing new udev rules files. %_udevhwdbdir,
4596 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
4597 %_sysctldir are now available which resolve to the right
4598 directories for packages to place various data files in.
4600 * journalctl gained the new --full switch (in addition to
4601 --all, to disable ellipsation for long messages.
4603 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
4604 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
4605 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
4606 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
4607 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
4608 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
4609 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
4610 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
4611 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
4615 * journalctl gained new --since= and --until= switches to
4616 filter by time. It also now supports nice filtering for
4617 units via --unit=/-u.
4619 * Type=oneshot services may use ExecReload= and do the
4622 * The journal daemon now supports time-based rotation and
4623 vacuuming, in addition to the usual disk-space based
4626 * The journal will now index the available field values for
4627 each field name. This enables clients to show pretty drop
4628 downs of available match values when filtering. The bash
4629 completion of journalctl has been updated
4630 accordingly. journalctl gained a new switch -F to list all
4631 values a certain field takes in the journal database.
4633 * More service events are now written as structured messages
4634 to the journal, and made recognizable via message IDs.
4636 * The timedated, localed and hostnamed mini-services which
4637 previously only provided support for changing time, locale
4638 and hostname settings from graphical DEs such as GNOME now
4639 also have a minimal (but very useful) text-based client
4640 utility each. This is probably the nicest way to changing
4641 these settings from the command line now, especially since
4642 it lists available options and is fully integrated with bash
4645 * There's now a new tool "systemd-coredumpctl" to list and
4646 extract coredumps from the journal.
4648 * We now install a README each in /var/log/ and
4649 /etc/rc.d/init.d explaining where the system logs and init
4650 scripts went. This hopefully should help folks who go to
4651 that dirs and look into the otherwise now empty void and
4652 scratch their heads.
4654 * When user-services are invoked (by systemd --user) the
4655 $MANAGERPID env var is set to the PID of systemd.
4657 * SIGRTMIN+24 when sent to a --user instance will now result
4658 in immediate termination of systemd.
4660 * gatewayd received numerous feature additions such as a
4661 "follow" mode, for live syncing and filtering.
4663 * browse.html now allows filtering and showing detailed
4664 information on specific entries. Keyboard navigation and
4665 mouse screen support has been added.
4667 * gatewayd/journalctl now supports HTML5/JSON
4668 Server-Sent-Events as output.
4670 * The SysV init script compatibility logic will now
4671 heuristically determine whether a script supports the
4672 "reload" verb, and only then make this available as
4675 * "systemctl status --follow" has been removed, use "journalctl
4678 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
4679 have been removed since they are hardly useful to be
4682 * And I'd like to take the opportunity to specifically mention
4683 Zbigniew for his great contributions. Zbigniew, you rock!
4685 Contributions from: Andrew Eikum, Christian Hesse, Colin
4686 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
4687 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
4688 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
4689 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
4690 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
4691 Jędrzejewski-Szmek, Сковорода Никита Андреевич
4695 * If /etc/vconsole.conf is non-existent or empty we will no
4696 longer load any console font or key map at boot by
4697 default. Instead the kernel defaults will be left
4698 intact. This is definitely the right thing to do, as no
4699 configuration should mean no configuration, and hard-coding
4700 font names that are different on all archs is probably a bad
4701 idea. Also, the kernel default key map and font should be
4702 good enough for most cases anyway, and mostly identical to
4703 the userspace fonts/key maps we previously overloaded them
4704 with. If distributions want to continue to default to a
4705 non-kernel font or key map they should ship a default
4706 /etc/vconsole.conf with the appropriate contents.
4708 Contributions from: Colin Walters, Daniel J Walsh, Dave
4709 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
4710 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
4714 * journalctl gained a new --cursor= switch to show entries
4715 starting from the specified location in the journal.
4717 * We now enforce a size limit on journal entry fields exported
4718 with "-o json" in journalctl. Fields larger than 4K will be
4719 assigned null. This can be turned off with --all.
4721 * An (optional) journal gateway daemon is now available as
4722 "systemd-journal-gatewayd.service". This service provides
4723 access to the journal via HTTP and JSON. This functionality
4724 will be used to implement live log synchronization in both
4725 pull and push modes, but has various other users too, such
4726 as easy log access for debugging of embedded devices. Right
4727 now it is already useful to retrieve the journal via HTTP:
4729 # systemctl start systemd-journal-gatewayd.service
4730 # wget http://localhost:19531/entries
4732 This will download the journal contents in a
4733 /var/log/messages compatible format. The same as JSON:
4735 # curl -H"Accept: application/json" http://localhost:19531/entries
4737 This service is also accessible via a web browser where a
4738 single static HTML5 app is served that uses the JSON logic
4739 to enable the user to do some basic browsing of the
4740 journal. This will be extended later on. Here's an example
4741 screenshot of this app in its current state:
4743 http://0pointer.de/public/journal-gatewayd
4745 Contributions from: Kay Sievers, Lennart Poettering, Robert
4746 Milasan, Tom Gundersen
4750 * The bash completion logic is now available for journalctl
4753 * We do not mount the "cpuset" controller anymore together with
4754 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
4755 started if no parameters are assigned to it. "cpuset" hence
4756 broke code that assumed it it could create "cpu" groups and
4759 * journalctl -f will now subscribe to terminal size changes,
4760 and line break accordingly.
4762 Contributions from: Dave Reisner, Kay Sievers, Lennart
4763 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
4767 * nspawn will now create a symlink /etc/localtime in the
4768 container environment, copying the host's timezone
4769 setting. Previously this has been done via a bind mount, but
4770 since symlinks cannot be bind mounted this has now been
4771 changed to create/update the appropriate symlink.
4773 * journalctl -n's line number argument is now optional, and
4774 will default to 10 if omitted.
4776 * journald will now log the maximum size the journal files may
4777 take up on disk. This is particularly useful if the default
4778 built-in logic of determining this parameter from the file
4779 system size is used. Use "systemctl status
4780 systemd-journald.service" to see this information.
4782 * The multi-seat X wrapper tool has been stripped down. As X
4783 is now capable of enumerating graphics devices via udev in a
4784 seat-aware way the wrapper is not strictly necessary
4785 anymore. A stripped down temporary stop-gap is still shipped
4786 until the upstream display managers have been updated to
4787 fully support the new X logic. Expect this wrapper to be
4788 removed entirely in one of the next releases.
4790 * HandleSleepKey= in logind.conf has been split up into
4791 HandleSuspendKey= and HandleHibernateKey=. The old setting
4792 is not available anymore. X11 and the kernel are
4793 distinguishing between these keys and we should too. This
4794 also means the inhibition lock for these keys has been split
4797 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
4798 Poettering, Lukas Nykryn, Václav Pavlín
4802 * Whenever a unit changes state we will now log this to the
4803 journal and show along the unit's own log output in
4806 * ConditionPathIsMountPoint= can now properly detect bind
4807 mount points too. (Previously, a bind mount of one file
4808 system to another place in the same file system could not be
4809 detected as mount, since they shared struct stat's st_dev
4812 * We will now mount the cgroup controllers cpu, cpuacct,
4813 cpuset and the controllers net_cls, net_prio together by
4816 * nspawn containers will now have a virtualized boot
4817 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
4818 over with a randomized ID at container initialization). This
4819 has the effect of making "journalctl -b" do the right thing
4822 * The JSON output journal serialization has been updated not
4823 to generate "endless" list objects anymore, but rather one
4824 JSON object per line. This is more in line how most JSON
4825 parsers expect JSON objects. The new output mode
4826 "json-pretty" has been added to provide similar output, but
4827 neatly aligned for readability by humans.
4829 * We dropped all explicit sync() invocations in the shutdown
4830 code. The kernel does this implicitly anyway in the kernel
4831 reboot() syscall. halt(8)'s -n option is now a compatibility
4834 * We now support virtualized reboot() in containers, as
4835 supported by newer kernels. We will fall back to exit() if
4836 CAP_SYS_REBOOT is not available to the container. Also,
4837 nspawn makes use of this now and will actually reboot the
4838 container if the containerized OS asks for that.
4840 * journalctl will only show local log output by default
4841 now. Use --merge (-m) to show remote log output, too.
4843 * libsystemd-journal gained the new sd_journal_get_usage()
4844 call to determine the current disk usage of all journal
4845 files. This is exposed in the new "journalctl --disk-usage"
4848 * journald gained a new configuration setting SplitMode= in
4849 journald.conf which may be used to control how user journals
4850 are split off. See journald.conf(5) for details.
4852 * A new condition type ConditionFileNotEmpty= has been added.
4854 * tmpfiles' "w" lines now support file globbing, to write
4855 multiple files at once.
4857 * We added Python bindings for the journal submission
4858 APIs. More Python APIs for a number of selected APIs will
4859 likely follow. Note that we intend to add native bindings
4860 only for the Python language, as we consider it common
4861 enough to deserve bindings shipped within systemd. There are
4862 various projects outside of systemd that provide bindings
4863 for languages such as PHP or Lua.
4865 * Many conditions will now resolve specifiers such as %i. In
4866 addition, PathChanged= and related directives of .path units
4867 now support specifiers as well.
4869 * There's now a new RPM macro definition for the system preset
4872 * journald will now warn if it ca not forward a message to the
4873 syslog daemon because its socket is full.
4875 * timedated will no longer write or process /etc/timezone,
4876 except on Debian. As we do not support late mounted /usr
4877 anymore /etc/localtime always being a symlink is now safe,
4878 and hence the information in /etc/timezone is not necessary
4881 * logind will now always reserve one VT for a text getty (VT6
4882 by default). Previously if more than 6 X sessions where
4883 started they took up all the VTs with auto-spawned gettys,
4884 so that no text gettys were available anymore.
4886 * udev will now automatically inform the btrfs kernel logic
4887 about btrfs RAID components showing up. This should make
4888 simple hotplug based btrfs RAID assembly work.
4890 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
4891 (but not for its children which will stay at the kernel
4892 default). This should allow setups with a lot more listening
4895 * systemd will now always pass the configured timezone to the
4896 kernel at boot. timedated will do the same when the timezone
4899 * logind's inhibition logic has been updated. By default,
4900 logind will now handle the lid switch, the power and sleep
4901 keys all the time, even in graphical sessions. If DEs want
4902 to handle these events on their own they should take the new
4903 handle-power-key, handle-sleep-key and handle-lid-switch
4904 inhibitors during their runtime. A simple way to achieve
4905 that is to invoke the DE wrapped in an invocation of:
4907 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
4909 * Access to unit operations is now checked via SELinux taking
4910 the unit file label and client process label into account.
4912 * systemd will now notify the administrator in the journal
4913 when he over-mounts a non-empty directory.
4915 * There are new specifiers that are resolved in unit files,
4916 for the host name (%H), the machine ID (%m) and the boot ID
4919 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
4920 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
4921 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
4922 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
4923 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
4924 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
4925 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
4929 * Support for reading structured kernel messages from
4930 /dev/kmsg has now been added and is enabled by default.
4932 * Support for reading kernel messages from /proc/kmsg has now
4933 been removed. If you want kernel messages in the journal
4934 make sure to run a recent kernel (>= 3.5) that supports
4935 reading structured messages from /dev/kmsg (see
4936 above). /proc/kmsg is now exclusive property of classic
4937 syslog daemons again.
4939 * The libudev API gained the new
4940 udev_device_new_from_device_id() call.
4942 * The logic for file system namespace (ReadOnlyDirectory=,
4943 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
4944 require pivot_root() anymore. This means fewer temporary
4945 directories are created below /tmp for this feature.
4947 * nspawn containers will now see and receive all submounts
4948 made on the host OS below the root file system of the
4951 * Forward Secure Sealing is now supported for Journal files,
4952 which provide cryptographical sealing of journal files so
4953 that attackers cannot alter log history anymore without this
4954 being detectable. Lennart will soon post a blog story about
4955 this explaining it in more detail.
4957 * There are two new service settings RestartPreventExitStatus=
4958 and SuccessExitStatus= which allow configuration of exit
4959 status (exit code or signal) which will be excepted from the
4960 restart logic, resp. consider successful.
4962 * journalctl gained the new --verify switch that can be used
4963 to check the integrity of the structure of journal files and
4964 (if Forward Secure Sealing is enabled) the contents of
4967 * nspawn containers will now be run with /dev/stdin, /dev/fd/
4968 and similar symlinks pre-created. This makes running shells
4969 as container init process a lot more fun.
4971 * The fstab support can now handle PARTUUID= and PARTLABEL=
4974 * A new ConditionHost= condition has been added to match
4975 against the hostname (with globs) and machine ID. This is
4976 useful for clusters where a single OS image is used to
4977 provision a large number of hosts which shall run slightly
4978 different sets of services.
4980 * Services which hit the restart limit will now be placed in a
4983 Contributions from: Bertram Poettering, Dave Reisner, Huang
4984 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
4985 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
4989 * When running in --user mode systemd will now become a
4990 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
4991 tree a lot more organized.
4993 * A new PartOf= unit dependency type has been introduced that
4994 may be used to group services in a natural way.
4996 * "systemctl enable" may now be used to enable instances of
4999 * journalctl now prints error log levels in red, and
5000 warning/notice log levels in bright white. It also supports
5001 filtering by log level now.
5003 * cgtop gained a new -n switch (similar to top), to configure
5004 the maximum number of iterations to run for. It also gained
5005 -b, to run in batch mode (accepting no input).
5007 * The suffix ".service" may now be omitted on most systemctl
5008 command lines involving service unit names.
5010 * There's a new bus call in logind to lock all sessions, as
5011 well as a loginctl verb for it "lock-sessions".
5013 * libsystemd-logind.so gained a new call sd_journal_perror()
5014 that works similar to libc perror() but logs to the journal
5015 and encodes structured information about the error number.
5017 * /etc/crypttab entries now understand the new keyfile-size=
5020 * shutdown(8) now can send a (configurable) wall message when
5021 a shutdown is cancelled.
5023 * The mount propagation mode for the root file system will now
5024 default to "shared", which is useful to make containers work
5025 nicely out-of-the-box so that they receive new mounts from
5026 the host. This can be undone locally by running "mount
5027 --make-rprivate /" if needed.
5029 * The prefdm.service file has been removed. Distributions
5030 should maintain this unit downstream if they intend to keep
5031 it around. However, we recommend writing normal unit files
5032 for display managers instead.
5034 * Since systemd is a crucial part of the OS we will now
5035 default to a number of compiler switches that improve
5036 security (hardening) such as read-only relocations, stack
5037 protection, and suchlike.
5039 * The TimeoutSec= setting for services is now split into
5040 TimeoutStartSec= and TimeoutStopSec= to allow configuration
5041 of individual time outs for the start and the stop phase of
5044 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
5045 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
5046 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
5047 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
5048 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
5049 Gundersen, Zbigniew Jędrzejewski-Szmek
5053 * The journal and id128 C APIs are now fully documented as man
5056 * Extra safety checks have been added when transitioning from
5057 the initial RAM disk to the main system to avoid accidental
5060 * /etc/crypttab entries now understand the new keyfile-offset=
5063 * systemctl -t can now be used to filter by unit load state.
5065 * The journal C API gained the new sd_journal_wait() call to
5066 make writing synchronous journal clients easier.
5068 * journalctl gained the new -D switch to show journals from a
5071 * journalctl now displays a special marker between log
5072 messages of two different boots.
5074 * The journal is now explicitly flushed to /var via a service
5075 systemd-journal-flush.service, rather than implicitly simply
5076 by seeing /var/log/journal to be writable.
5078 * journalctl (and the journal C APIs) can now match for much
5079 more complex expressions, with alternatives and
5082 * When transitioning from the initial RAM disk to the main
5083 system we will now kill all processes in a killing spree to
5084 ensure no processes stay around by accident.
5086 * Three new specifiers may be used in unit files: %u, %h, %s
5087 resolve to the user name, user home directory resp. user
5088 shell. This is useful for running systemd user instances.
5090 * We now automatically rotate journal files if their data
5091 object hash table gets a fill level > 75%. We also size the
5092 hash table based on the configured maximum file size. This
5093 together should lower hash collisions drastically and thus
5094 speed things up a bit.
5096 * journalctl gained the new "--header" switch to introspect
5097 header data of journal files.
5099 * A new setting SystemCallFilters= has been added to services
5100 which may be used to apply blacklists or whitelists to
5101 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
5103 * nspawn gained a new --link-journal= switch (and quicker: -j)
5104 to link the container journal with the host. This makes it
5105 very easy to centralize log viewing on the host for all
5106 guests while still keeping the journal files separated.
5108 * Many bugfixes and optimizations
5110 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
5111 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
5112 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
5117 * Several tools now understand kernel command line arguments,
5118 which are only read when run in an initial RAM disk. They
5119 usually follow closely their normal counterparts, but are
5122 * There's a new tool to analyze the readahead files that are
5123 automatically generated at boot. Use:
5125 /usr/lib/systemd/systemd-readahead analyze /.readahead
5127 * We now provide an early debug shell on tty9 if this enabled. Use:
5129 systemctl enable debug-shell.service
5131 * All plymouth related units have been moved into the Plymouth
5132 package. Please make sure to upgrade your Plymouth version
5135 * systemd-tmpfiles now supports getting passed the basename of
5136 a configuration file only, in which case it will look for it
5137 in all appropriate directories automatically.
5139 * udevadm info now takes a /dev or /sys path as argument, and
5140 does the right thing. Example:
5142 udevadm info /dev/sda
5143 udevadm info /sys/class/block/sda
5145 * systemctl now prints a warning if a unit is stopped but a
5146 unit that might trigger it continues to run. Example: a
5147 service is stopped but the socket that activates it is left
5150 * "systemctl status" will now mention if the log output was
5151 shortened due to rotation since a service has been started.
5153 * The journal API now exposes functions to determine the
5154 "cutoff" times due to rotation.
5156 * journald now understands SIGUSR1 and SIGUSR2 for triggering
5157 immediately flushing of runtime logs to /var if possible,
5158 resp. for triggering immediate rotation of the journal
5161 * It is now considered an error if a service is attempted to
5162 be stopped that is not loaded.
5164 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
5166 * systemd-analyze now supports Python 3
5168 * tmpfiles now supports cleaning up directories via aging
5169 where the first level dirs are always kept around but
5170 directories beneath it automatically aged. This is enabled
5171 by prefixing the age field with '~'.
5173 * Seat objects now expose CanGraphical, CanTTY properties
5174 which is required to deal with very fast bootups where the
5175 display manager might be running before the graphics drivers
5176 completed initialization.
5178 * Seat objects now expose a State property.
5180 * We now include RPM macros for service enabling/disabling
5181 based on the preset logic. We recommend RPM based
5182 distributions to make use of these macros if possible. This
5183 makes it simpler to reuse RPM spec files across
5186 * We now make sure that the collected systemd unit name is
5187 always valid when services log to the journal via
5190 * There's a new man page kernel-command-line(7) detailing all
5191 command line options we understand.
5193 * The fstab generator may now be disabled at boot by passing
5194 fstab=0 on the kernel command line.
5196 * A new kernel command line option modules-load= is now understood
5197 to load a specific kernel module statically, early at boot.
5199 * Unit names specified on the systemctl command line are now
5200 automatically escaped as needed. Also, if file system or
5201 device paths are specified they are automatically turned
5202 into the appropriate mount or device unit names. Example:
5204 systemctl status /home
5205 systemctl status /dev/sda
5207 * The SysVConsole= configuration option has been removed from
5208 system.conf parsing.
5210 * The SysV search path is no longer exported on the D-Bus
5213 * The Names= option is been removed from unit file parsing.
5215 * There's a new man page bootup(7) detailing the boot process.
5217 * Every unit and every generator we ship with systemd now
5218 comes with full documentation. The self-explanatory boot is
5221 * A couple of services gained "systemd-" prefixes in their
5222 name if they wrap systemd code, rather than only external
5223 code. Among them fsck@.service which is now
5224 systemd-fsck@.service.
5226 * The HaveWatchdog property has been removed from the D-Bus
5229 * systemd.confirm_spawn= on the kernel command line should now
5232 * There's a new man page crypttab(5) which details all options
5233 we actually understand.
5235 * systemd-nspawn gained a new --capability= switch to pass
5236 additional capabilities to the container.
5238 * timedated will now read known NTP implementation unit names
5239 from /usr/lib/systemd/ntp-units.d/*.list,
5240 systemd-timedated-ntp.target has been removed.
5242 * journalctl gained a new switch "-b" that lists log data of
5243 the current boot only.
5245 * The notify socket is in the abstract namespace again, in
5246 order to support daemons which chroot() at start-up.
5248 * There is a new Storage= configuration option for journald
5249 which allows configuration of where log data should go. This
5250 also provides a way to disable journal logging entirely, so
5251 that data collected is only forwarded to the console, the
5252 kernel log buffer or another syslog implementation.
5254 * Many bugfixes and optimizations
5256 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
5257 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
5258 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
5259 Shawn Landden, Tom Gundersen
5263 * "systemctl help <unit>" now shows the man page if one is
5266 * Several new man pages have been added.
5268 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
5269 MaxLevelConsole= can now be specified in
5270 journald.conf. These options allow reducing the amount of
5271 data stored on disk or forwarded by the log level.
5273 * TimerSlackNSec= can now be specified in system.conf for
5274 PID1. This allows system-wide power savings.
5276 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
5277 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
5282 * logind is now capable of (optionally) handling power and
5283 sleep keys as well as the lid switch.
5285 * journalctl now understands the syntax "journalctl
5286 /usr/bin/avahi-daemon" to get all log output of a specific
5289 * CapabilityBoundingSet= in system.conf now also influences
5290 the capability bound set of usermode helpers of the kernel.
5292 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
5293 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
5294 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
5295 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
5299 * Note that we skipped 139 releases here in order to set the
5300 new version to something that is greater than both udev's
5301 and systemd's most recent version number.
5303 * udev: all udev sources are merged into the systemd source tree now.
5304 All future udev development will happen in the systemd tree. It
5305 is still fully supported to use the udev daemon and tools without
5306 systemd running, like in initramfs or other init systems. Building
5307 udev though, will require the *build* of the systemd tree, but
5308 udev can be properly *run* without systemd.
5310 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
5311 should be used to create dead device nodes as workarounds for broken
5314 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
5315 no longer supported. udev_monitor_new_from_netlink() needs to be
5316 used to subscribe to events.
5318 * udev: when udevd is started by systemd, processes which are left
5319 behind by forking them off of udev rules, are unconditionally cleaned
5320 up and killed now after the event handling has finished. Services or
5321 daemons must be started as systemd services. Services can be
5322 pulled-in by udev to get started, but they can no longer be directly
5323 forked by udev rules.
5325 * udev: the daemon binary is called systemd-udevd now and installed
5326 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
5327 to adapt to that, create symlink, or rename the binary after building
5330 * libudev no longer provides these symbols:
5331 udev_monitor_from_socket()
5332 udev_queue_get_failed_list_entry()
5333 udev_get_{dev,sys,run}_path()
5334 The versions number was bumped and symbol versioning introduced.
5336 * systemd-loginctl and systemd-journalctl have been renamed
5337 to loginctl and journalctl to match systemctl.
5339 * The config files: /etc/systemd/systemd-logind.conf and
5340 /etc/systemd/systemd-journald.conf have been renamed to
5341 logind.conf and journald.conf. Package updates should rename
5342 the files to the new names on upgrade.
5344 * For almost all files the license is now LGPL2.1+, changed
5345 from the previous GPL2.0+. Exceptions are some minor stuff
5346 of udev (which will be changed to LGPL2.1 eventually, too),
5347 and the MIT licensed sd-daemon.[ch] library that is suitable
5348 to be used as drop-in files.
5350 * systemd and logind now handle system sleep states, in
5351 particular suspending and hibernating.
5353 * logind now implements a sleep/shutdown/idle inhibiting logic
5354 suitable for a variety of uses. Soonishly Lennart will blog
5355 about this in more detail.
5357 * var-run.mount and var-lock.mount are no longer provided
5358 (which prevously bind mounted these directories to their new
5359 places). Distributions which have not converted these
5360 directories to symlinks should consider stealing these files
5361 from git history and add them downstream.
5363 * We introduced the Documentation= field for units and added
5364 this to all our shipped units. This is useful to make it
5365 easier to explore the boot and the purpose of the various
5368 * All smaller setup units (such as
5369 systemd-vconsole-setup.service) now detect properly if they
5370 are run in a container and are skipped when
5371 appropriate. This guarantees an entirely noise-free boot in
5372 Linux container environments such as systemd-nspawn.
5374 * A framework for implementing offline system updates is now
5375 integrated, for details see:
5376 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
5378 * A new service type Type=idle is available now which helps us
5379 avoiding ugly interleaving of getty output and boot status
5382 * There's now a system-wide CapabilityBoundingSet= option to
5383 globally reduce the set of capabilities for the
5384 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
5385 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
5386 even CAP_NET_ADMIN system-wide for secure systems.
5388 * There are now system-wide DefaultLimitXXX= options to
5389 globally change the defaults of the various resource limits
5390 for all units started by PID 1.
5392 * Harald Hoyer's systemd test suite has been integrated into
5393 systemd which allows easy testing of systemd builds in qemu
5394 and nspawn. (This is really awesome! Ask us for details!)
5396 * The fstab parser is now implemented as generator, not inside
5399 * systemctl will now warn you if .mount units generated from
5400 /etc/fstab are out of date due to changes in fstab that
5401 have not been read by systemd yet.
5403 * systemd is now suitable for usage in initrds. Dracut has
5404 already been updated to make use of this. With this in place
5405 initrds get a slight bit faster but primarily are much
5406 easier to introspect and debug since "systemctl status" in
5407 the host system can be used to introspect initrd services,
5408 and the journal from the initrd is kept around too.
5410 * systemd-delta has been added, a tool to explore differences
5411 between user/admin configuration and vendor defaults.
5413 * PrivateTmp= now affects both /tmp and /var/tmp.
5415 * Boot time status messages are now much prettier and feature
5416 proper english language. Booting up systemd has never been
5419 * Read-ahead pack files now include the inode number of all
5420 files to pre-cache. When the inode changes the pre-caching
5421 is not attempted. This should be nicer to deal with updated
5422 packages which might result in changes of read-ahead
5425 * We now temporaritly lower the kernel's read_ahead_kb variable
5426 when collecting read-ahead data to ensure the kernel's
5427 built-in read-ahead does not add noise to our measurements
5428 of necessary blocks to pre-cache.
5430 * There's now RequiresMountsFor= to add automatic dependencies
5431 for all mounts necessary for a specific file system path.
5433 * MountAuto= and SwapAuto= have been removed from
5434 system.conf. Mounting file systems at boot has to take place
5437 * nspawn now learned a new switch --uuid= to set the machine
5438 ID on the command line.
5440 * nspawn now learned the -b switch to automatically search
5443 * vt102 is now the default TERM for serial TTYs, upgraded from
5446 * systemd-logind now works on VT-less systems.
5448 * The build tree has been reorganized. The individual
5449 components now have directories of their own.
5451 * A new condition type ConditionPathIsReadWrite= is now available.
5453 * nspawn learned the new -C switch to create cgroups for the
5454 container in other hierarchies.
5456 * We now have support for hardware watchdogs, configurable in
5459 * The scheduled shutdown logic now has a public API.
5461 * We now mount /tmp as tmpfs by default, but this can be
5462 masked and /etc/fstab can override it.
5464 * Since udisks does not make use of /media anymore we are not
5465 mounting a tmpfs on it anymore.
5467 * journalctl gained a new --local switch to only interleave
5468 locally generated journal files.
5470 * We can now load the IMA policy at boot automatically.
5472 * The GTK tools have been split off into a systemd-ui.
5474 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
5475 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
5476 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
5477 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
5478 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
5479 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
5480 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
5481 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
5482 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
5487 * This is mostly a bugfix release
5489 * Support optional initialization of the machine ID from the
5490 KVM or container configured UUID.
5492 * Support immediate reboots with "systemctl reboot -ff"
5494 * Show /etc/os-release data in systemd-analyze output
5496 * Many bugfixes for the journal, including endianness fixes and
5497 ensuring that disk space enforcement works
5499 * sd-login.h is C++ comptaible again
5501 * Extend the /etc/os-release format on request of the Debian
5504 * We now refuse non-UTF8 strings used in various configuration
5505 and unit files. This is done to ensure we do not pass invalid
5506 data over D-Bus or expose it elsewhere.
5508 * Register Mimo USB Screens as suitable for automatic seat
5511 * Read SELinux client context from journal clients in a race
5514 * Reorder configuration file lookup order. /etc now always
5515 overrides /run in order to allow the administrator to always
5516 and unconditionally override vendor-supplied or
5517 automatically generated data.
5519 * The various user visible bits of the journal now have man
5520 pages. We still lack man pages for the journal API calls
5523 * We now ship all man pages in HTML format again in the
5526 Contributions from: Dave Reisner, Dirk Eibach, Frederic
5527 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
5528 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
5533 * This is mostly a bugfix release
5535 * systems lacking /etc/os-release are no longer supported.
5537 * Various functionality updates to libsystemd-login.so
5539 * Track class of PAM logins to distinguish greeters from
5542 Contributions from: Kay Sievers, Lennart Poettering, Michael
5547 * This is an important bugfix release for v41.
5549 * Building man pages is now optional which should be useful
5550 for those building systemd from git but unwilling to install
5553 * Watchdog support for supervising services is now usable. In
5554 a future release support for hardware watchdogs
5555 (i.e. /dev/watchdog) will be added building on this.
5557 * Service start rate limiting is now configurable and can be
5558 turned off per service. When a start rate limit is hit a
5559 reboot can automatically be triggered.
5561 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
5563 Contributions from: Benjamin Franzke, Bill Nottingham,
5564 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
5565 Schmidt, Michał Górny, Piotr Drąg
5569 * The systemd binary is installed /usr/lib/systemd/systemd now;
5570 An existing /sbin/init symlink needs to be adapted with the
5573 * The code that loads kernel modules has been ported to invoke
5574 libkmod directly, instead of modprobe. This means we do not
5575 support systems with module-init-tools anymore.
5577 * Watchdog support is now already useful, but still not
5580 * A new kernel command line option systemd.setenv= is
5581 understood to set system wide environment variables
5582 dynamically at boot.
5584 * We now limit the set of capabilities of systemd-journald.
5586 * We now set SIGPIPE to ignore by default, since it only is
5587 useful in shell pipelines, and has little use in general
5588 code. This can be disabled with IgnoreSIPIPE=no in unit
5591 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
5592 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
5597 * This is mostly a bugfix release
5599 * We now expose the reason why a service failed in the
5600 "Result" D-Bus property.
5602 * Rudimentary service watchdog support (will be completed over
5603 the next few releases.)
5605 * When systemd forks off in order execute some service we will
5606 now immediately changes its argv[0] to reflect which process
5607 it will execute. This is useful to minimize the time window
5608 with a generic argv[0], which makes bootcharts more useful
5610 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
5611 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
5612 Mike Kazantsev, Ray Strode
5616 * This is mostly a test release, but incorporates many
5619 * New systemd-cgtop tool to show control groups by their
5622 * Linking against libacl for ACLs is optional again. If
5623 disabled, support tracking device access for active logins
5624 goes becomes unavailable, and so does access to the user
5625 journals by the respective users.
5627 * If a group "adm" exists, journal files are automatically
5628 owned by them, thus allow members of this group full access
5629 to the system journal as well as all user journals.
5631 * The journal now stores the SELinux context of the logging
5632 client for all entries.
5634 * Add C++ inclusion guards to all public headers
5636 * New output mode "cat" in the journal to print only text
5637 messages, without any meta data like date or time.
5639 * Include tiny X server wrapper as a temporary stop-gap to
5640 teach XOrg udev display enumeration. This is used by display
5641 managers such as gdm, and will go away as soon as XOrg
5642 learned native udev hotplugging for display devices.
5644 * Add new systemd-cat tool for executing arbitrary programs
5645 with STDERR/STDOUT connected to the journal. Can also act as
5646 BSD logger replacement, and does so by default.
5648 * Optionally store all locally generated coredumps in the
5649 journal along with meta data.
5651 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
5652 writing short strings to files (for usage for /sys), and for
5653 creating symlinks, character and block device nodes.
5655 * New unit file option ControlGroupPersistent= to make cgroups
5656 persistent, following the mechanisms outlined in
5657 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
5659 * Support multiple local RTCs in a sane way
5661 * No longer monopolize IO when replaying readahead data on
5662 rotating disks, since we might starve non-file-system IO to
5663 death, since fanotify() will not see accesses done by blkid,
5666 * Do not show kernel threads in systemd-cgls anymore, unless
5667 requested with new -k switch.
5669 Contributions from: Dan Horák, Kay Sievers, Lennart
5670 Poettering, Michal Schmidt
5674 * This is mostly a test release, but incorporates many
5677 * The git repository moved to:
5678 git://anongit.freedesktop.org/systemd/systemd
5679 ssh://git.freedesktop.org/git/systemd/systemd
5681 * First release with the journal
5682 http://0pointer.de/blog/projects/the-journal.html
5684 * The journal replaces both systemd-kmsg-syslogd and
5685 systemd-stdout-bridge.
5687 * New sd_pid_get_unit() API call in libsystemd-logind
5689 * Many systemadm clean-ups
5691 * Introduce remote-fs-pre.target which is ordered before all
5692 remote mounts and may be used to start services before all
5695 * Added Mageia support
5697 * Add bash completion for systemd-loginctl
5699 * Actively monitor PID file creation for daemons which exit in
5700 the parent process before having finished writing the PID
5701 file in the daemon process. Daemons which do this need to be
5702 fixed (i.e. PID file creation must have finished before the
5703 parent exits), but we now react a bit more gracefully to them.
5705 * Add colourful boot output, mimicking the well-known output
5706 of existing distributions.
5708 * New option PassCredentials= for socket units, for
5709 compatibility with a recent kernel ABI breakage.
5711 * /etc/rc.local is now hooked in via a generator binary, and
5712 thus will no longer act as synchronization point during
5715 * systemctl list-unit-files now supports --root=.
5717 * systemd-tmpfiles now understands two new commands: z, Z for
5718 relabelling files according to the SELinux database. This is
5719 useful to apply SELinux labels to specific files in /sys,
5722 * Output of SysV services is now forwarded to both the console
5723 and the journal by default, not only just the console.
5725 * New man pages for all APIs from libsystemd-login.
5727 * The build tree got reorganized and a the build system is a
5728 lot more modular allowing embedded setups to specifically
5729 select the components of systemd they are interested in.
5731 * Support for Linux systems lacking the kernel VT subsystem is
5734 * configure's --with-rootdir= got renamed to
5735 --with-rootprefix= to follow the naming used by udev and
5738 * Unless specified otherwise we will now install to /usr instead
5739 of /usr/local by default.
5741 * Processes with '@' in argv[0][0] are now excluded from the
5742 final shut-down killing spree, following the logic explained
5744 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
5746 * All processes remaining in a service cgroup when we enter
5747 the START or START_PRE states are now killed with
5748 SIGKILL. That means it is no longer possible to spawn
5749 background processes from ExecStart= lines (which was never
5750 supported anyway, and bad style).
5752 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
5753 reloading of units together.
5755 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
5756 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
5757 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
5758 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
5759 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek