1 systemd System and Service Manager
5 * systemd now depends on util-linux v2.27. More specifically,
6 the newly added mount monitor feature in libmount now
7 replaces systemd's former own implementation.
9 * libmount mandates /etc/mtab not to be regular file, and
10 systemd now enforces this condition at early boot.
11 /etc/mtab has been deprecated and warned about for a very
12 long time, so systems running systemd should already have
13 stopped having this file around as anything else than a
14 symlink to /proc/self/mounts.
16 * Support for the "pids" cgroup controller has been added. It
17 allows accounting the number of tasks in a cgroup and
18 enforcing limits on it. This adds two new setting
19 TasksAccounting= and TasksMax= to each unit, as well as a
20 global option DefaultTasksAccounting=.
22 * Support for the "net_cls" cgroup controller has been added.
23 It allows assigning a net class ID to each task in the
24 cgroup, which can then be used in firewall rules and traffic
25 shaping configurations. Note that the kernel netfilter net
26 class code does not currently work reliably for ingress
27 packets on unestablished sockets.
29 This adds a new config directive called NetClass= to CGroup
30 enabled units. Allowed values are positive numbers for fixed
31 assignments and "auto" for picking a free value
34 * 'systemctl is-system-running' now returns 'offline' if the
35 system is not booted with systemd. This command can now be
36 used as a substitute for 'systemd-notify --booted'.
38 * Watchdog timeouts have been increased to 3 minutes for all
39 in-tree service files. Apparently, disk IO issues are more
40 frequent than we hoped, and user reported >1 minute waiting
43 * 'machine-id-commit' functionality has been merged into
44 'machine-id-setup --commit'. The separate binary has been
47 * The WorkingDirectory= directive in unit files may now be set
48 to the special value '~'. In this case, the working
49 directory is set to the home directory of the user
52 * "machinectl shell" will now open the shell in the home
53 directory of the selected user by default.
55 * A new systemd.crash_reboot=1 kernel command line option has
56 been added that triggers a reboot after crashing. This can
57 also be set through CrashReboot= in systemd.conf.
59 * The CrashChVT= configuration file setting is renamed to
60 CrashChangeVT=, following our usual logic of not
61 abbreviating unnecessarily. The old directive is still
62 supported for compat reasons. Also, this directive now takes
63 an integer value between 1 and 63, or a boolean value. The
64 formerly supported '-1' value for disabling stays around for
67 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
68 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
69 RootDirectory= properties can now be set for transient
72 * The systemd-analyze tool gained a new "set-log-target" verb
73 to change the logging target the system manager logs to
74 dynamically during runtime. This is similar to how
75 "systemd-analyze set-log-level" already changes the log
78 * In nspawn /sys is now mounted as tmpfs, with only a selected
79 set of subdirectories mounted in from the real sysfs. This
80 enhances security slightly, and is useful for ensuring user
81 namespaces work correctly.
83 * Support for USB FunctionFS activation has been added. This
84 allows implementation of USB gadget services that are
85 activated as soon as they are requested, so that they don't
86 have to run continously, similar to classic socket
89 * The "systemctl exit" command now optionally takes an
90 additional parameter that sets the exit code to return from
91 the systemd manager when exiting. This is only relevant when
92 running the systemd user instance, or when running the
93 system instance in a container.
95 * sd-bus gained the new API calls sd_bus_path_encode_many()
96 and sd_bus_path_decode_many() that allow easy encoding and
97 decoding of multiple identifier strings inside a D-Bus
98 object path. Another new call sd_bus_default_flush_close()
99 has been added to flush and close per-thread default
102 * systemd-cgtop gained support for a -M/--machine= switch to
103 show the control groups within a certain container only.
105 * "systemctl kill" gained support for an optional --fail
106 switch. If specified the requested operation will fail of no
107 processes have been killed, because the unit had no
108 processes attached, or similar.
110 * A new (still internal) libary API sd-ipv4acd has been added,
111 that implements address conflict detection for IPv4. It's
112 based on code from sd-ipv4ll, and will be useful for
113 detecting DHCP address conflicts.
115 * The RuntimeDirectory= setting now understands unit
116 specifiers like %i or %f.
118 * systemd-networkd gained support for:
120 - Setting the IPv6 Router Advertisment settings via
121 IPv6AcceptRouterAdvertisements= in .network files.
123 - Configuring the HelloTimeSec=, MaxAgeSec= and
124 ForwardDelaySec= bridge parameters in .netdev files.
126 - Configuring PreferredSource= for static routes in
129 * udev will now create /dev/disk/by-path links for ATA devices
130 on kernels where that is supported.
132 * When downloading tar or raw images using "machinectl
133 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
134 file is now also downloaded, if it is available and stored
135 next to the image file.
137 * Units of type ".socket" gained a new boolean setting
138 Writable= which is only useful in conjunction with
139 ListenSpecial=. If true, enables opening the specified
140 special file in O_RDWR mode rather than O_RDONLY mode.
142 * systemd-rfkill has been reworked to become a singleton
143 service that is activated through /dev/rfkill on each rfkill
144 state change and saves the settings to disk. This way,
145 systemd-rfkill is now compatible with devices that exist
146 only intermittendly, and even restores state if the previous
147 system shutdown was abrupt rather than clean.
149 * File descriptors passed during socket activation may now be
150 named. A new API sd_listen_fds_with_names() is added to
151 access the names. The default names may be overriden,
152 either in the .socket file using the FileDescriptorName=
153 parameter, or by passing FDNAME= when storing the file
154 descriptors using sd_notify().
156 * The journal daemon gained support for vacuuming old journal
157 files controlled by the number of files that shall remain,
158 in addition to the already existing control by size and by
159 date. This is useful as journal interleaving performance
160 degrades with too many seperate journal files, and allows
161 putting an effective limit on them. The new setting defaults
162 to 100, but this may be changed by setting SystemMaxFiles=
163 and RuntimeMaxFiles= in journald.conf. Also, the
164 "journalctl" tool gained the new --vacuum-files= switch to
165 manually vacuum journal files to leave only the specified
166 number of files in place.
168 * Galician, Serbian, Turkish and Korean translations were added.
172 -- Berlin, 2015-09-xx
176 * The DHCP implementation of systemd-networkd gained a set of
179 - The DHCP server now supports emitting DNS and NTP
180 information. It may be enabled and configured via
181 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
182 and NTP information is enabled, but no servers are
183 configured, the corresponding uplink information (if there
184 is any) is propagated.
186 - Server and client now support transmission and reception
187 of timezone information. It can be configured via the
188 newly introduced network options UseTimezone=,
189 EmitTimezone=, and Timezone=. Transmission of timezone
190 information is enabled between host and containers by
191 default now: the container will change its local timezone
192 to what the host has set.
194 - Lease timeouts can now be configured via
195 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
197 - The DHCP server improved on the stability of
198 leases. Clients are more likely to get the same lease
199 information back, even if the server loses state.
201 - The DHCP server supports two new configuration options to
202 control the lease address pool metrics, PoolOffset= and
205 * The encapsulation limit of tunnels in systemd-networkd may
206 now be configured via 'EncapsulationLimit='. It allows
207 modifying the maximum additional levels of encapsulation
208 that are permitted to be prepended to a packet.
210 * systemd now supports the concept of user buses replacing
211 session buses, if used with dbus-1.10 (and enabled via dbus
212 --enable-user-session). It previously only supported this on
213 kdbus-enabled systems, and this release expands this to
214 'dbus-daemon' systems.
216 * systemd-networkd now supports predictable interface names
219 * systemd now optionally supports the new Linux kernel
220 "unified" control group hierarchy. If enabled via the kernel
221 command-line option 'systemd.unified_cgroup_hierarchy=1',
222 systemd will try to mount the unified cgroup hierarchy
223 directly on /sys/fs/cgroup. If not enabled, or not
224 available, systemd will fall back to the legacy cgroup
225 hierarchy setup, as before. Host system and containers can
226 mix and match legacy and unified hierarchies as they
227 wish. nspawn understands the $UNIFIED_CROUP_HIERARCHY
228 environment variable to individually select the hierarchy to
229 use for executed containers. By default, nspawn will use the
230 unified hierarchy for the containers if the host uses the
231 unified hierarchy, and the legacy hierarchy otherwise.
232 Please note that at this point the unified hierarchy is an
233 experimental kernel feature and is likely to change in one
234 of the next kernel releases. Therefore, it should not be
235 enabled by default in downstream distributions yet. The
236 minimum required kernel version for the unified hierarchy to
237 work is 4.2. Note that when the unified hierarchy is used
238 for the first time delegated access to controllers is
239 safe. Because of this systemd-nspawn containers will get
240 access to controllers now, as will systemd user
241 sessions. This means containers and user sessions may now
242 manage their own resources, partitioning up what the system
245 * A new special scope unit "init.scope" has been introduced
246 that encapsulates PID 1 of the system. It may be used to
247 determine resource usage and enforce resource limits on PID
248 1 itself. PID 1 hence moved out of the root of the control
251 * The cgtop tool gained support for filtering out kernel
252 threads when counting tasks in a control group. Also, the
253 count of processes is now recursively summed up by
254 default. Two options -k and --recursive= have been added to
255 revert to old behaviour. The tool has also been updated to
256 work correctly in containers now.
258 * systemd-nspawn's --bind= and --bind-ro= options have been
259 extended to allow creation of non-recursive bind mounts.
261 * libsystemd gained two new calls sd_pid_get_cgroup() and
262 sd_peer_get_cgroup() which return the control group path of
263 a process or peer of a connected AF_UNIX socket. This
264 function call is particularly useful when implementing
265 delegated subtrees support in the control group hierarchy.
267 * The "sd-event" event loop API of libsystemd now supports
268 correct dequeuing of real-time signals, without losing
271 * When systemd requests a PolicyKit decision when managing
272 units it will now add additional fields to the request,
273 including unit name and desired operation. This enables more
274 powerful PolicyKit policies, that make decisions depending
277 * nspawn learnt support for .nspawn settings files, that may
278 accompany the image files or directories of containers, and
279 may contain additional settings for the container. This is
280 an alternative to configuring container parameters via the
283 Contributions from: Cristian Rodríguez, Daniel Mack, David
284 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
285 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
286 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
287 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
288 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
289 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
290 Andersen, Tom Gundersen, Torstein Husebø
292 -- Berlin, 2015-09-08
296 * machinectl gained a new verb 'shell' which opens a fresh
297 shell on the target container or the host. It is similar to
298 the existing 'login' command of machinectl, but spawns the
299 shell directly without prompting for username or
300 password. The pseudo machine '.host' now refers to the local
301 host and is used by default. Hence, 'machinectl shell' can
302 be used as replacement for 'su -' which spawns a session as
303 a fresh systemd unit in a way that is fully isolated from
304 the originating session.
306 * systemd-networkd learned to cope with private-zone DHCP
307 options and allows other programs to query the values.
309 * SELinux access control when enabling/disabling units is no
310 longer enforced with this release. The previous
311 implementation was incorrect, and a new corrected
312 implementation is not yet available. As unit file operations
313 are still protected via PolicyKit and D-Bus policy this is
314 not a security problem. Yet, distributions which care about
315 optimal SELinux support should probably not stabilize on
318 * sd-bus gained support for matches of type "arg0has=", that
319 test for membership of strings in string arrays sent in bus
322 * systemd-resolved now dumps the contents of its DNS and LLMNR
323 caches to the logs on reception of the SIGUSR1 signal. This
324 is useful to debug DNS behaviour.
326 * The coredumpctl tool gained a new --directory= option to
327 operate on journal files in a specific directory.
329 * "systemctl reboot" and related commands gained a new
330 "--message=" option which may be used to set a free-text
331 wall message when shutting down or rebooting the
332 system. This message is also logged, which is useful for
333 figuring out the reason for a reboot or shutdown a
336 * The "systemd-resolve-host" tool's -i switch now takes
337 network interface numbers as alternative to interface names.
339 * A new unit file setting for services has been introduced:
340 UtmpMode= allows configuration of how precisely systemd
341 handles utmp and wtmp entries for the service if this is
342 enabled. This allows writing services that appear similar to
343 user sessions in the output of the "w", "who", "last" and
346 * systemd-resolved will now locally synthesize DNS resource
347 records for the "localhost" and "gateway" domains as well as
348 the local hostname. This should ensure that clients querying
349 RRs via resolved will get similar results as those going via
350 NSS, if nss-myhostname is enabled.
352 Contributions from: Alastair Hughes, Alex Crawford, Daniel
353 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
354 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
355 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
356 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
357 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
358 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
359 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
360 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
361 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
362 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
363 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
365 -- Berlin, 2015-08-27
369 * The systemd-efi-boot-generator functionality was merged into
370 systemd-gpt-auto-generator.
372 * systemd-networkd now supports Group Policy for vxlan
373 devices. It can be enabled via the new boolean configuration
374 option called 'GroupPolicyExtension='.
376 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
377 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
378 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
380 -- Berlin, 2015-07-31
384 * The python-systemd code has been removed from the systemd repository.
385 A new repository has been created which accommodates the code from
386 now on, and we kindly ask distributions to create a separate package
387 for this: https://github.com/systemd/python-systemd
389 * The systemd daemon will now reload its main configuration
390 (/etc/systemd/system.conf) on daemon-reload.
392 * sd-dhcp now exposes vendor specific extensions via
393 sd_dhcp_lease_get_vendor_specific().
395 * systemd-networkd gained a number of new configuration options.
397 - A new boolean configuration option for TAP devices called
398 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
399 device, thus allowing to send and receive GSO packets.
401 - A new tunnel configuration option called 'CopyDSCP='.
402 If enabled, the DSCP field of ip6 tunnels is copied into the
405 - A set of boolean bridge configuration options were added.
406 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
407 and 'UnicastFlood=' are now parsed by networkd and applied to the
408 respective bridge link device via the respective IFLA_BRPORT_*
411 - A new string configuration option to override the hostname sent
412 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
413 is true, networkd will use the configured hostname instead of the
414 system hostname when sending DHCP requests.
416 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
417 networkd will configure the IPv6 flow-label of the tunnel device
418 according to RFC2460.
420 - The 'macvtap' virtual network devices are now supported, similar to
421 the already supported 'macvlan' devices.
423 * systemd-resolved now implements RFC5452 to improve resilience against
424 cache poisoning. Additionally, source port randomization is enabled
425 by default to further protect against DNS spoofing attacks.
427 * nss-mymachines now supports translating UIDs and GIDs of running
428 containers with user-namespaces enabled. If a container 'foo'
429 translates a host uid 'UID' to the container uid 'TUID', then
430 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
431 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
432 mapped as 'vg-foo-TGID'.
434 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
435 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
436 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
437 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
438 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
439 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
440 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
441 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
442 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
443 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
445 -- Berlin, 2015-07-29
449 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
450 There are no known issues with current sysfs, and udev does not need
451 or should be used to work around such bugs.
453 * udev does no longer enable USB HID power management. Several reports
454 indicate, that some devices cannot handle that setting.
456 * The udev accelerometer helper was removed. The functionality
457 is now fully included in iio-sensor-proxy. But this means,
458 older iio-sensor-proxy versions will no longer provide
459 accelerometer/orientation data with this systemd version.
460 Please upgrade iio-sensor-proxy to version 1.0.
462 * networkd gained a new configuration option IPv6PrivacyExtensions=
463 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
464 for Stateless Address") on selected networks.
466 * For the sake of fewer build-time dependencies and less code in the
467 main repository, the python bindings are about to be removed in the
468 next release. A new repository has been created which accommodates
469 the code from now on, and we kindly ask distributions to create a
470 separate package for this. The removal will take place in v223.
472 https://github.com/systemd/python-systemd
474 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
475 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
476 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
477 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
478 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
479 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
480 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
481 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
482 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
483 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
485 -- Berlin, 2015-07-07
489 * The sd-bus.h and sd-event.h APIs have now been declared
490 stable and have been added to the official interface of
491 libsystemd.so. sd-bus implements an alternative D-Bus client
492 library, that is relatively easy to use, very efficient and
493 supports both classic D-Bus as well as kdbus as transport
494 backend. sd-event is a generic event loop abstraction that
495 is built around Linux epoll, but adds features such as event
496 prioritization or efficient timer handling. Both APIs are good
497 choices for C programs looking for a bus and/or event loop
498 implementation that is minimal and does not have to be
499 portable to other kernels.
501 * kdbus support is no longer compile-time optional. It is now
502 always built-in. However, it can still be disabled at
503 runtime using the kdbus=0 kernel command line setting, and
504 that setting may be changed to default to off, by specifying
505 --disable-kdbus at build-time. Note though that the kernel
506 command line setting has no effect if the kdbus.ko kernel
507 module is not installed, in which case kdbus is (obviously)
508 also disabled. We encourage all downstream distributions to
509 begin testing kdbus by adding it to the kernel images in the
510 development distributions, and leaving kdbus support in
513 * The minimal required util-linux version has been bumped to
516 * Support for chkconfig (--enable-chkconfig) was removed in
517 favor of calling an abstraction tool
518 /lib/systemd/systemd-sysv-install. This needs to be
519 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
520 in README for details.
522 * If there's a systemd unit and a SysV init script for the
523 same service name, and the user executes "systemctl enable"
524 for it (or a related call), then this will now enable both
525 (or execute the related operation on both), not just the
528 * The libudev API documentation has been converted from gtkdoc
531 * gudev has been removed from the systemd tree, it is now an
534 * The systemd-cgtop tool learnt a new --raw switch to generate
535 "raw" (machine parsable) output.
537 * networkd's IPForwarding= .network file setting learnt the
538 new setting "kernel", which ensures that networkd does not
539 change the IP forwarding sysctl from the default kernel
542 * The systemd-logind bus API now exposes a new boolean
543 property "Docked" that reports whether logind considers the
544 system "docked", i.e. connected to a docking station or not.
546 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
547 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
548 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
549 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
550 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
551 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
552 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
553 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
554 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
555 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
556 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
557 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
558 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
559 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
560 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
561 Fink, Zbigniew Jędrzejewski-Szmek
563 -- Berlin, 2015-06-19
567 * The gudev library has been extracted into a separate repository
568 available at: https://git.gnome.org/browse/libgudev/
569 It is now managed as part of the Gnome project. Distributions
570 are recommended to pass --disable-gudev to systemd and use
571 gudev from the Gnome project instead. gudev is still included
572 in systemd, for now. It will be removed soon, though. Please
573 also see the announcement-thread on systemd-devel:
574 http://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
576 * systemd now exposes a CPUUsageNSec= property for each
577 service unit on the bus, that contains the overall consumed
578 CPU time of a service (the sum of what each process of the
579 service consumed). This value is only available if
580 CPUAccounting= is turned on for a service, and is then shown
581 in the "systemctl status" output.
583 * Support for configuring alternative mappings of the old SysV
584 runlevels to systemd targets has been removed. They are now
585 hardcoded in a way that runlevels 2, 3, 4 all map to
586 multi-user.target and 5 to graphical.target (which
587 previously was already the default behaviour).
589 * The auto-mounter logic gained support for mount point
590 expiry, using a new TimeoutIdleSec= setting in .automount
591 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
593 * The EFI System Partition (ESP) as mounted to /boot by
594 systemd-efi-boot-generator will now be unmounted
595 automatically after 2 minutes of not being used. This should
596 minimize the risk of ESP corruptions.
598 * New /etc/fstab options x-systemd.requires= and
599 x-systemd.requires-mounts-for= are now supported to express
600 additional dependencies for mounts. This is useful for
601 journalling file systems that support external journal
602 devices or overlay file systems that require underlying file
603 systems to be mounted.
605 * systemd does not support direct live-upgrades (via systemctl
606 daemon-reexec) from versions older than v44 anymore. As no
607 distribution we are aware of shipped such old versions in a
608 stable release this should not be problematic.
610 * When systemd forks off a new per-connection service instance
611 it will now set the $REMOTE_ADDR environment variable to the
612 remote IP address, and $REMOTE_PORT environment variable to
613 the remote IP port. This behaviour is similar to the
614 corresponding environment variables defined by CGI.
616 * systemd-networkd gained support for uplink failure
617 detection. The BindCarrier= option allows binding interface
618 configuration dynamically to the link sense of other
619 interfaces. This is useful to achieve behaviour like in
622 * systemd-networkd gained support for configuring the DHCP
623 client identifier to use when requesting leases.
625 * systemd-networkd now has a per-network UseNTP= option to
626 configure whether NTP server information acquired via DHCP
627 is passed on to services like systemd-timesyncd.
629 * systemd-networkd gained support for vti6 tunnels.
631 * Note that systemd-networkd manages the sysctl variable
632 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
633 it is configured for since v219. The variable controls IP
634 forwarding, and is a per-interface alternative to the global
635 /proc/sys/net/ipv[46]/ip_forward. This setting is
636 configurable in the IPForward= option, which defaults to
637 "no". This means if networkd is used for an interface it is
638 no longer sufficient to set the global sysctl option to turn
639 on IP forwarding! Instead, the .network file option
640 IPForward= needs to be turned on! Note that the
641 implementation of this behaviour was broken in v219 and has
644 * Many bonding and vxlan options are now configurable in
647 * systemd-nspawn gained a new --property= setting to set unit
648 properties for the container scope. This is useful for
649 setting resource parameters (e.g "CPUShares=500") on
650 containers started from the command line.
652 * systemd-nspawn gained a new --private-users= switch to make
653 use of user namespacing available on recent Linux kernels.
655 * systemd-nspawn may now be called as part of a shell pipeline
656 in which case the pipes used for stdin and stdout are passed
657 directly to the process invoked in the container, without
658 indirection via a pseudo tty.
660 * systemd-nspawn gained a new switch to control the UNIX
661 signal to use when killing the init process of the container
664 * systemd-nspawn gained a new --overlay= switch for mounting
665 overlay file systems into the container using the new kernel
668 * When a container image is imported via systemd-importd and
669 the host file system is not btrfs, a loopback block device
670 file is created in /var/lib/machines.raw with a btrfs file
671 system inside. It is then mounted to /var/lib/machines to
672 enable btrfs features for container management. The loopback
673 file and btrfs file system is grown as needed when container
674 images are imported via systemd-importd.
676 * systemd-machined/systemd-importd gained support for btrfs
677 quota, to enforce container disk space limits on disk. This
678 is exposed in "machinectl set-limit".
680 * systemd-importd now can import containers from local .tar,
681 .raw and .qcow2 images, and export them to .tar and .raw. It
682 can also import dkr v2 images now from the network (on top
685 * systemd-importd gained support for verifying downloaded
686 images with gpg2 (previously only gpg1 was supported).
688 * systemd-machined, systemd-logind, systemd: most bus calls
689 are now accessible to unprivileged processes via
690 PolicyKit. Also, systemd-logind will now allow users to kill
691 their own sessions without further privileges or
694 * systemd-shutdownd has been removed. This service was
695 previously responsible for implementing scheduled shutdowns
696 as exposed in /usr/bin/shutdown's time parameter. This
697 functionality has now been moved into systemd-logind and is
698 accessible via a bus interface.
700 * "systemctl reboot" gained a new switch --firmware-setup that
701 can be used to reboot into the EFI firmware setup, if that
702 is available. systemd-logind now exposes an API on the bus
703 to trigger such reboots, in case graphical desktop UIs want
704 to cover this functionality.
706 * "systemctl enable", "systemctl disable" and "systemctl mask"
707 now support a new "--now" switch. If specified the units
708 that are enabled will also be started, and the ones
709 disabled/masked also stopped.
711 * The Gummiboot EFI boot loader tool has been merged into
712 systemd, and renamed to "systemd-boot". The bootctl tool has been
713 updated to support systemd-boot.
715 * An EFI kernel stub has been added that may be used to create
716 kernel EFI binaries that contain not only the actual kernel,
717 but also an initrd, boot splash, command line and OS release
718 information. This combined binary can then be signed as a
719 single image, so that the firmware can verify it all in one
720 step. systemd-boot has special support for EFI binaries created
721 like this and can extract OS release information from them
722 and show them in the boot menu. This functionality is useful
723 to implement cryptographically verified boot schemes.
725 * Optional support has been added to systemd-fsck to pass
726 fsck's progress report to an AF_UNIX socket in the file
729 * udev will no longer create device symlinks for all block
730 devices by default. A blacklist for excluding special block
731 devices from this logic has been turned into a whitelist
732 that requires picking block devices explicitly that require
735 * A new (currently still internal) API sd-device.h has been
736 added to libsystemd. This modernized API is supposed to
737 replace libudev eventually. In fact, already much of libudev
738 is now just a wrapper around sd-device.h.
740 * A new hwdb database for storing metadata about pointing
741 stick devices has been added.
743 * systemd-tmpfiles gained support for setting file attributes
744 similar to the "chattr" tool with new 'h' and 'H' lines.
746 * systemd-journald will no longer unconditionally set the
747 btrfs NOCOW flag on new journal files. This is instead done
748 with tmpfiles snippet using the new 'h' line type. This
749 allows easy disabling of this logic, by masking the
750 journal-nocow.conf tmpfiles file.
752 * systemd-journald will now translate audit message types to
753 human readable identifiers when writing them to the
754 journal. This should improve readability of audit messages.
756 * The LUKS logic gained support for the offset= and skip=
757 options in /etc/crypttab, as previously implemented by
760 * /usr/lib/os-release gained a new optional field VARIANT= for
761 distributions that support multiple variants (such as a
762 desktop edition, a server edition, ...)
764 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
765 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
766 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
767 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
768 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
769 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
770 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
771 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
772 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
773 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
774 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
775 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
776 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
777 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
778 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
779 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
780 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
781 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
782 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
783 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
784 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
785 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
786 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
787 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
788 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
789 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
790 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
792 -- Berlin, 2015-05-22
796 * Introduce a new API "sd-hwdb.h" for querying the hardware
797 metadata database. With this minimal interface one can query
798 and enumerate the udev hwdb, decoupled from the old libudev
799 library. libudev's interface for this is now only a wrapper
800 around sd-hwdb. A new tool systemd-hwdb has been added to
801 interface with and update the database.
803 * When any of systemd's tools copies files (for example due to
804 tmpfiles' C lines) a btrfs reflink will attempted first,
805 before bytewise copying is done.
807 * systemd-nspawn gained a new --ephemeral switch. When
808 specified a btrfs snapshot is taken of the container's root
809 directory, and immediately removed when the container
810 terminates again. Thus, a container can be started whose
811 changes never alter the container's root directory, and are
812 lost on container termination. This switch can also be used
813 for starting a container off the root file system of the
814 host without affecting the host OS. This switch is only
815 available on btrfs file systems.
817 * systemd-nspawn gained a new --template= switch. It takes the
818 path to a container tree to use as template for the tree
819 specified via --directory=, should that directory be
820 missing. This allows instantiating containers dynamically,
821 on first run. This switch is only available on btrfs file
824 * When a .mount unit refers to a mount point on which multiple
825 mounts are stacked, and the .mount unit is stopped all of
826 the stacked mount points will now be unmounted until no
829 * systemd now has an explicit notion of supported and
830 unsupported unit types. Jobs enqueued for unsupported unit
831 types will now fail with an "unsupported" error code. More
832 specifically .swap, .automount and .device units are not
833 supported in containers, .busname units are not supported on
834 non-kdbus systems. .swap and .automount are also not
835 supported if their respective kernel compile time options
838 * machinectl gained support for two new "copy-from" and
839 "copy-to" commands for copying files from a running
840 container to the host or vice versa.
842 * machinectl gained support for a new "bind" command to bind
843 mount host directories into local containers. This is
844 currently only supported for nspawn containers.
846 * networkd gained support for configuring bridge forwarding
847 database entries (fdb) from .network files.
849 * A new tiny daemon "systemd-importd" has been added that can
850 download container images in tar, raw, qcow2 or dkr formats,
851 and make them available locally in /var/lib/machines, so
852 that they can run as nspawn containers. The daemon can GPG
853 verify the downloads (not supported for dkr, since it has no
854 provisions for verifying downloads). It will transparently
855 decompress bz2, xz, gzip compressed downloads if necessary,
856 and restore sparse files on disk. The daemon uses privilege
857 separation to ensure the actual download logic runs with
858 fewer privileges than the daemon itself. machinectl has
859 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
860 make the functionality of importd available to the
861 user. With this in place the Fedora and Ubuntu "Cloud"
862 images can be downloaded and booted as containers unmodified
863 (the Fedora images lack the appropriate GPG signature files
864 currently, so they cannot be verified, but this will change
865 soon, hopefully). Note that downloading images is currently
866 only fully supported on btrfs.
868 * machinectl is now able to list container images found in
869 /var/lib/machines, along with some metadata about sizes of
870 disk and similar. If the directory is located on btrfs and
871 quota is enabled, this includes quota display. A new command
872 "image-status" has been added that shows additional
873 information about images.
875 * machinectl is now able to clone container images
876 efficiently, if the underlying file system (btrfs) supports
877 it, with the new "machinectl list-images" command. It also
878 gained commands for renaming and removing images, as well as
879 marking them read-only or read-write (supported also on
880 legacy file systems).
882 * networkd gained support for collecting LLDP network
883 announcements, from hardware that supports this. This is
884 shown in networkctl output.
886 * systemd-run gained support for a new -t (--pty) switch for
887 invoking a binary on a pty whose input and output is
888 connected to the invoking terminal. This allows executing
889 processes as system services while interactively
890 communicating with them via the terminal. Most interestingly
891 this is supported across container boundaries. Invoking
892 "systemd-run -t /bin/bash" is an alternative to running a
893 full login session, the difference being that the former
894 will not register a session, nor go through the PAM session
897 * tmpfiles gained support for a new "v" line type for creating
898 btrfs subvolumes. If the underlying file system is a legacy
899 file system, this automatically degrades to creating a
900 normal directory. Among others /var/lib/machines is now
901 created like this at boot, should it be missing.
903 * The directory /var/lib/containers/ has been deprecated and
904 been replaced by /var/lib/machines. The term "machines" has
905 been used in the systemd context as generic term for both
906 VMs and containers, and hence appears more appropriate for
907 this, as the directory can also contain raw images bootable
910 * systemd-nspawn when invoked with -M but without --directory=
911 or --image= is now capable of searching for the container
912 root directory, subvolume or disk image automatically, in
913 /var/lib/machines. systemd-nspawn@.service has been updated
914 to make use of this, thus allowing it to be used for raw
917 * A new machines.target unit has been introduced that is
918 supposed to group all containers/VMs invoked as services on
919 the system. systemd-nspawn@.service has been updated to
922 * machinectl gained a new "start" command, for invoking a
923 container as a service. "machinectl start foo" is mostly
924 equivalent to "systemctl start systemd-nspawn@foo.service",
925 but handles escaping in a nicer way.
927 * systemd-nspawn will now mount most of the cgroupfs tree
928 read-only into each container, with the exception of the
929 container's own subtree in the name=systemd hierarchy.
931 * journald now sets the special FS_NOCOW file flag for its
932 journal files. This should improve performance on btrfs, by
933 avoiding heavy fragmentation when journald's write-pattern
934 is used on COW file systems. It degrades btrfs' data
935 integrity guarantees for the files to the same levels as for
936 ext3/ext4 however. This should be OK though as journald does
937 its own data integrity checks and all its objects are
938 checksummed on disk. Also, journald should handle btrfs disk
939 full events a lot more gracefully now, by processing SIGBUS
940 errors, and not relying on fallocate() anymore.
942 * When journald detects that journal files it is writing to
943 have been deleted it will immediately start new journal
946 * systemd now provides a way to store file descriptors
947 per-service in PID 1.This is useful for daemons to ensure
948 that fds they require are not lost during a daemon
949 restart. The fds are passed to the daemon on the next
950 invocation in the same way socket activation fds are
951 passed. This is now used by journald to ensure that the
952 various sockets connected to all the system's stdout/stderr
953 are not lost when journald is restarted. File descriptors
954 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
955 an extension to sd_notify(). Note that a limit is enforced
956 on the number of fds a service can store in PID 1, and it
957 defaults to 0, so that no fds may be stored, unless this is
958 explicitly turned on.
960 * The default TERM variable to use for units connected to a
961 terminal, when no other value is explicitly is set is now
962 vt220 rather than vt102. This should be fairly safe still,
963 but allows PgUp/PgDn work.
965 * The /etc/crypttab option header= as known from Debian is now
968 * "loginctl user-status" and "loginctl session-status" will
969 now show the last 10 lines of log messages of the
970 user/session following the status output. Similar,
971 "machinectl status" will show the last 10 log lines
972 associated with a virtual machine or container
973 service. (Note that this is usually not the log messages
974 done in the VM/container itself, but simply what the
975 container manager logs. For nspawn this includes all console
978 * "loginctl session-status" without further argument will now
979 show the status of the session of the caller. Similar,
980 "lock-session", "unlock-session", "activate",
981 "enable-linger", "disable-linger" may now be called without
982 session/user parameter in which case they apply to the
983 caller's session/user.
985 * An X11 session scriptlet is now shipped that uploads
986 $DISPLAY and $XAUTHORITY into the environment of the systemd
987 --user daemon if a session begins. This should improve
988 compatibility with X11 enabled applications run as systemd
991 * Generators are now subject to masking via /etc and /run, the
992 same way as unit files.
994 * networkd .network files gained support for configuring
995 per-link IPv4/IPv6 packet forwarding as well as IPv4
996 masquerading. This is by default turned on for veth links to
997 containers, as registered by systemd-nspawn. This means that
998 nspawn containers run with --network-veth will now get
999 automatic routed access to the host's networks without any
1000 further configuration or setup, as long as networkd runs on
1003 * systemd-nspawn gained the --port= (-p) switch to expose TCP
1004 or UDP posts of a container on the host. With this in place
1005 it is possible to run containers with private veth links
1006 (--network-veth), and have their functionality exposed on
1007 the host as if their services were running directly on the
1010 * systemd-nspawn's --network-veth switch now gained a short
1011 version "-n", since with the changes above it is now truly
1012 useful out-of-the-box. The systemd-nspawn@.service has been
1013 updated to make use of it too by default.
1015 * systemd-nspawn will now maintain a per-image R/W lock, to
1016 ensure that the same image is not started more than once
1017 writable. (It's OK to run an image multiple times
1018 simultaneously in read-only mode.)
1020 * systemd-nspawn's --image= option is now capable of
1021 dissecting and booting MBR and GPT disk images that contain
1022 only a single active Linux partition. Previously it
1023 supported only GPT disk images with proper GPT type
1024 IDs. This allows running cloud images from major
1025 distributions directly with systemd-nspawn, without
1028 * In addition to collecting mouse dpi data in the udev
1029 hardware database, there's now support for collecting angle
1030 information for mouse scroll wheels. The database is
1031 supposed to guarantee similar scrolling behavior on mice
1032 that it knows about. There's also support for collecting
1033 information about Touchpad types.
1035 * udev's input_id built-in will now also collect touch screen
1036 dimension data and attach it to probed devices.
1038 * /etc/os-release gained support for a Distribution Privacy
1041 * networkd gained support for creating "ipvlan", "gretap",
1042 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
1044 * systemd-tmpfiles gained support for "a" lines for setting
1047 * systemd-nspawn will now mount /tmp in the container to
1048 tmpfs, automatically.
1050 * systemd now exposes the memory.usage_in_bytes cgroup
1051 attribute and shows it for each service in the "systemctl
1052 status" output, if available.
1054 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
1055 immediate reboot is triggered. This useful if shutdown is
1056 hung and is unable to complete, to expedite the
1057 operation. Note that this kind of reboot will still unmount
1058 all file systems, and hence should not result in fsck being
1061 * A .device unit for an optical block device will now be
1062 considered active only when a medium is in the drive. Also,
1063 mount units are now bound to their backing devices thus
1064 triggering automatic unmounting when devices become
1065 unavailable. With this in place systemd will now
1066 automatically unmount left-over mounts when a CD-ROM is
1067 ejected or an USB stick is yanked from the system.
1069 * networkd-wait-online now has support for waiting for
1070 specific interfaces only (with globbing), and for giving up
1071 after a configurable timeout.
1073 * networkd now exits when idle. It will be automatically
1074 restarted as soon as interfaces show up, are removed or
1075 change state. networkd will stay around as long as there is
1076 at least one DHCP state machine or similar around, that keep
1079 * networkd may now configure IPv6 link-local addressing in
1080 addition to IPv4 link-local addressing.
1082 * The IPv6 "token" for use in SLAAC may now be configured for
1083 each .network interface in networkd.
1085 * Routes configured with networkd may now be assigned a scope
1088 * networkd's [Match] sections now support globbing and lists
1089 of multiple space-separated matches per item.
1091 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
1092 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
1093 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
1094 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
1095 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
1096 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
1097 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
1098 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
1099 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
1100 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
1101 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
1102 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
1103 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
1104 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
1105 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
1106 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
1107 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
1108 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
1109 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
1110 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
1111 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
1112 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
1113 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
1114 Hoffmann, Zbigniew Jędrzejewski-Szmek
1116 -- Berlin, 2015-02-16
1120 * When querying unit file enablement status (for example via
1121 "systemctl is-enabled"), a new state "indirect" is now known
1122 which indicates that a unit might not be enabled itself, but
1123 another unit listed in its Also= setting might be.
1125 * Similar to the various existing ConditionXYZ= settings for
1126 units there are now matching AssertXYZ= settings. While
1127 failing conditions cause a unit to be skipped, but its job
1128 to succeed, failing assertions declared like this will cause
1129 a unit start operation and its job to fail.
1131 * hostnamed now knows a new chassis type "embedded".
1133 * systemctl gained a new "edit" command. When used on a unit
1134 file this allows extending unit files with .d/ drop-in
1135 configuration snippets or editing the full file (after
1136 copying it from /usr/lib to /etc). This will invoke the
1137 user's editor (as configured with $EDITOR), and reload the
1138 modified configuration after editing.
1140 * "systemctl status" now shows the suggested enablement state
1141 for a unit, as declared in the (usually vendor-supplied)
1142 system preset files.
1144 * nss-myhostname will now resolve the single-label host name
1145 "gateway" to the locally configured default IP routing
1146 gateways, ordered by their metrics. This assigns a stable
1147 name to the used gateways, regardless which ones are
1148 currently configured. Note that the name will only be
1149 resolved after all other name sources (if nss-myhostname is
1150 configured properly) and should hence not negatively impact
1151 systems that use the single-label host name "gateway" in
1154 * systemd-inhibit now allows filtering by mode when listing
1157 * Scope and service units gained a new "Delegate" boolean
1158 property, which when set allows processes running inside the
1159 unit to further partition resources. This is primarily
1160 useful for systemd user instances as well as container
1163 * journald will now pick up audit messages directly from
1164 the kernel, and log them like any other log message. The
1165 audit fields are split up and fully indexed. This means that
1166 journalctl in many ways is now a (nicer!) alternative to
1167 ausearch, the traditional audit client. Note that this
1168 implements only a minimal audit client, if you want the
1169 special audit modes like reboot-on-log-overflow, please use
1170 the traditional auditd instead, which can be used in
1171 parallel to journald.
1173 * The ConditionSecurity= unit file option now understands the
1174 special string "audit" to check whether auditing is
1177 * journalctl gained two new commands --vacuum-size= and
1178 --vacuum-time= to delete old journal files until the
1179 remaining ones take up no more the specified size on disk,
1180 or are not older than the specified time.
1182 * A new, native PPPoE library has been added to sd-network,
1183 systemd's library of light-weight networking protocols. This
1184 library will be used in a future version of networkd to
1185 enable PPPoE communication without an external pppd daemon.
1187 * The busctl tool now understands a new "capture" verb that
1188 works similar to "monitor", but writes a packet capture
1189 trace to STDOUT that can be redirected to a file which is
1190 compatible with libcap's capture file format. This can then
1191 be loaded in Wireshark and similar tools to inspect bus
1194 * The busctl tool now understands a new "tree" verb that shows
1195 the object trees of a specific service on the bus, or of all
1198 * The busctl tool now understands a new "introspect" verb that
1199 shows all interfaces and members of objects on the bus,
1200 including their signature and values. This is particularly
1201 useful to get more information about bus objects shown by
1202 the new "busctl tree" command.
1204 * The busctl tool now understands new verbs "call",
1205 "set-property" and "get-property" for invoking bus method
1206 calls, setting and getting bus object properties in a
1209 * busctl gained a new --augment-creds= argument that controls
1210 whether the tool shall augment credential information it
1211 gets from the bus with data from /proc, in a possibly
1214 * nspawn's --link-journal= switch gained two new values
1215 "try-guest" and "try-host" that work like "guest" and
1216 "host", but do not fail if the host has no persistent
1217 journalling enabled. -j is now equivalent to
1218 --link-journal=try-guest.
1220 * macvlan network devices created by nspawn will now have
1221 stable MAC addresses.
1223 * A new SmackProcessLabel= unit setting has been added, which
1224 controls the SMACK security label processes forked off by
1225 the respective unit shall use.
1227 * If compiled with --enable-xkbcommon, systemd-localed will
1228 verify x11 keymap settings by compiling the given keymap. It
1229 will spew out warnings if the compilation fails. This
1230 requires libxkbcommon to be installed.
1232 * When a coredump is collected a larger number of metadata
1233 fields is now collected and included in the journal records
1234 created for it. More specifically control group membership,
1235 environment variables, memory maps, working directory,
1236 chroot directory, /proc/$PID/status, and a list of open file
1237 descriptors is now stored in the log entry.
1239 * The udev hwdb now contains DPI information for mice. For
1242 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
1244 * All systemd programs that read standalone configuration
1245 files in /etc now also support a corresponding series of
1246 .conf.d configuration directories in /etc/, /run/,
1247 /usr/local/lib/, /usr/lib/, and (if configured with
1248 --enable-split-usr) /lib/. In particular, the following
1249 configuration files now have corresponding configuration
1250 directories: system.conf user.conf, logind.conf,
1251 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
1252 resolved.conf, timesyncd.conf, journal-remote.conf, and
1253 journal-upload.conf. Note that distributions should use the
1254 configuration directories in /usr/lib/; the directories in
1255 /etc/ are reserved for the system administrator.
1257 * systemd-rfkill will no longer take the rfkill device name
1258 into account when storing rfkill state on disk, as the name
1259 might be dynamically assigned and not stable. Instead, the
1260 ID_PATH udev variable combined with the rfkill type (wlan,
1261 bluetooth, ...) is used.
1263 * A new service systemd-machine-id-commit.service has been
1264 added. When used on systems where /etc is read-only during
1265 boot, and /etc/machine-id is not initialized (but an empty
1266 file), this service will copy the temporary machine ID
1267 created as replacement into /etc after the system is fully
1268 booted up. This is useful for systems that are freshly
1269 installed with a non-initialized machine ID, but should get
1270 a fixed machine ID for subsequent boots.
1272 * networkd's .netdev files now provide a large set of
1273 configuration parameters for VXLAN devices. Similar, the
1274 bridge port cost parameter is now configurable in .network
1275 files. There's also new support for configuring IP source
1276 routing. networkd .link files gained support for a new
1277 OriginalName= match that is useful to match against the
1278 original interface name the kernel assigned. .network files
1279 may include MTU= and MACAddress= fields for altering the MTU
1280 and MAC address while being connected to a specific network
1283 * The LUKS logic gained supported for configuring
1284 UUID-specific key files. There's also new support for naming
1285 LUKS device from the kernel command line, using the new
1286 luks.name= argument.
1288 * Timer units may now be transiently created via the bus API
1289 (this was previously already available for scope and service
1290 units). In addition it is now possible to create multiple
1291 transient units at the same time with a single bus call. The
1292 "systemd-run" tool has been updated to make use of this for
1293 running commands on a specified time, in at(1)-style.
1295 * tmpfiles gained support for "t" lines, for assigning
1296 extended attributes to files. Among other uses this may be
1297 used to assign SMACK labels to files.
1299 Contributions from: Alin Rauta, Alison Chaiken, Andrej
1300 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
1301 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
1302 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
1303 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
1304 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
1305 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
1306 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
1307 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
1308 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
1309 Michael Biebl, Michael Chapman, Michael Marineau, Michal
1310 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
1311 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
1312 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
1313 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
1314 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
1315 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
1316 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
1318 -- Berlin, 2014-12-10
1322 * journalctl gained the new options -t/--identifier= to match
1323 on the syslog identifier (aka "tag"), as well as --utc to
1324 show log timestamps in the UTC timezone. journalctl now also
1325 accepts -n/--lines=all to disable line capping in a pager.
1327 * journalctl gained a new switch, --flush, that synchronously
1328 flushes logs from /run/log/journal to /var/log/journal if
1329 persistent storage is enabled. systemd-journal-flush.service
1330 now waits until the operation is complete.
1332 * Services can notify the manager before they start a reload
1333 (by sending RELOADING=1) or shutdown (by sending
1334 STOPPING=1). This allows the manager to track and show the
1335 internal state of daemons and closes a race condition when
1336 the process is still running but has closed its D-Bus
1339 * Services with Type=oneshot do not have to have any ExecStart
1342 * User units are now loaded also from
1343 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
1344 /run/systemd/user directory that was already previously
1345 supported, but is under the control of the user.
1347 * Job timeouts (i.e. time-outs on the time a job that is
1348 queued stays in the run queue) can now optionally result in
1349 immediate reboot or power-off actions (JobTimeoutAction= and
1350 JobTimeoutRebootArgument=). This is useful on ".target"
1351 units, to limit the maximum time a target remains
1352 undispatched in the run queue, and to trigger an emergency
1353 operation in such a case. This is now used by default to
1354 turn off the system if boot-up (as defined by everything in
1355 basic.target) hangs and does not complete for at least
1356 15min. Also, if power-off or reboot hang for at least 30min
1357 an immediate power-off/reboot operation is triggered. This
1358 functionality is particularly useful to increase reliability
1359 on embedded devices, but also on laptops which might
1360 accidentally get powered on when carried in a backpack and
1361 whose boot stays stuck in a hard disk encryption passphrase
1364 * systemd-logind can be configured to also handle lid switch
1365 events even when the machine is docked or multiple displays
1366 are attached (HandleLidSwitchDocked= option).
1368 * A helper binary and a service have been added which can be
1369 used to resume from hibernation in the initramfs. A
1370 generator will parse the resume= option on the kernel
1371 command line to trigger resume.
1373 * A user console daemon systemd-consoled has been
1374 added. Currently, it is a preview, and will so far open a
1375 single terminal on each session of the user marked as
1376 Desktop=systemd-console.
1378 * Route metrics can be specified for DHCP routes added by
1381 * The SELinux context of socket-activated services can be set
1382 from the information provided by the networking stack
1383 (SELinuxContextFromNet= option).
1385 * Userspace firmware loading support has been removed and
1386 the minimum supported kernel version is thus bumped to 3.7.
1388 * Timeout for udev workers has been increased from 1 to 3
1389 minutes, but a warning will be printed after 1 minute to
1390 help diagnose kernel modules that take a long time to load.
1392 * Udev rules can now remove tags on devices with TAG-="foobar".
1394 * systemd's readahead implementation has been removed. In many
1395 circumstances it didn't give expected benefits even for
1396 rotational disk drives and was becoming less relevant in the
1397 age of SSDs. As none of the developers has been using
1398 rotating media anymore, and nobody stepped up to actively
1399 maintain this component of systemd it has now been removed.
1401 * Swap units can use Options= to specify discard options.
1402 Discard options specified for swaps in /etc/fstab are now
1405 * Docker containers are now detected as a separate type of
1408 * The Password Agent protocol gained support for queries where
1409 the user input is shown, useful e.g. for user names.
1410 systemd-ask-password gained a new --echo option to turn that
1413 * The default sysctl.d/ snippets will now set:
1415 net.core.default_qdisc = fq_codel
1417 This selects Fair Queuing Controlled Delay as the default
1418 queuing discipline for network interfaces. fq_codel helps
1419 fight the network bufferbloat problem. It is believed to be
1420 a good default with no tuning required for most workloads.
1421 Downstream distributions may override this choice. On 10Gbit
1422 servers that do not do forwarding, "fq" may perform better.
1423 Systems without a good clocksource should use "pfifo_fast".
1425 * If kdbus is enabled during build a new option BusPolicy= is
1426 available for service units, that allows locking all service
1427 processes into a stricter bus policy, in order to limit
1428 access to various bus services, or even hide most of them
1429 from the service's view entirely.
1431 * networkctl will now show the .network and .link file
1432 networkd has applied to a specific interface.
1434 * sd-login gained a new API call sd_session_get_desktop() to
1435 query which desktop environment has been selected for a
1438 * UNIX utmp support is now compile-time optional to support
1439 legacy-free systems.
1441 * systemctl gained two new commands "add-wants" and
1442 "add-requires" for pulling in units from specific targets
1445 * If the word "rescue" is specified on the kernel command line
1446 the system will now boot into rescue mode (aka
1447 rescue.target), which was previously available only by
1448 specifying "1" or "systemd.unit=rescue.target" on the kernel
1449 command line. This new kernel command line option nicely
1450 mirrors the already existing "emergency" kernel command line
1453 * New kernel command line options mount.usr=, mount.usrflags=,
1454 mount.usrfstype= have been added that match root=, rootflags=,
1455 rootfstype= but allow mounting a specific file system to
1458 * The $NOTIFY_SOCKET is now also passed to control processes of
1459 services, not only the main process.
1461 * This version reenables support for fsck's -l switch. This
1462 means at least version v2.25 of util-linux is required for
1463 operation, otherwise dead-locks on device nodes may
1464 occur. Again: you need to update util-linux to at least
1465 v2.25 when updating systemd to v217.
1467 * The "multi-seat-x" tool has been removed from systemd, as
1468 its functionality has been integrated into X servers 1.16,
1469 and the tool is hence redundant. It is recommended to update
1470 display managers invoking this tool to simply invoke X
1471 directly from now on, again.
1473 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
1474 message flag has been added for all of systemd's PolicyKit
1475 authenticated method calls has been added. In particular
1476 this now allows optional interactive authorization via
1477 PolicyKit for many of PID1's privileged operations such as
1478 unit file enabling and disabling.
1480 * "udevadm hwdb --update" learnt a new switch "--usr" for
1481 placing the rebuilt hardware database in /usr instead of
1482 /etc. When used only hardware database entries stored in
1483 /usr will be used, and any user database entries in /etc are
1484 ignored. This functionality is useful for vendors to ship a
1485 pre-built database on systems where local configuration is
1486 unnecessary or unlikely.
1488 * Calendar time specifications in .timer units now also
1489 understand the strings "semi-annually", "quarterly" and
1490 "minutely" as shortcuts (in addition to the preexisting
1491 "anually", "hourly", ...).
1493 * systemd-tmpfiles will now correctly create files in /dev
1494 at boot which are marked for creation only at boot. It is
1495 recommended to always create static device nodes with 'c!'
1496 and 'b!', so that they are created only at boot and not
1497 overwritten at runtime.
1499 * When the watchdog logic is used for a service (WatchdogSec=)
1500 and the watchdog timeout is hit the service will now be
1501 terminated with SIGABRT (instead of just SIGTERM), in order
1502 to make sure a proper coredump and backtrace is
1503 generated. This ensures that hanging services will result in
1504 similar coredump/backtrace behaviour as services that hit a
1507 Contributions from: Andreas Henriksson, Andrei Borzenkov,
1508 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
1509 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
1510 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
1511 Herrmann, David Sommerseth, David Strauss, Emil Renner
1512 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
1513 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
1514 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
1515 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
1516 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
1517 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
1518 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
1519 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
1520 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
1521 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
1522 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
1523 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
1524 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
1525 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
1526 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
1527 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
1530 -- Berlin, 2014-10-28
1534 * timedated no longer reads NTP implementation unit names from
1535 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
1536 implementations should add a
1538 Conflicts=systemd-timesyncd.service
1540 to their unit files to take over and replace systemd's NTP
1541 default functionality.
1543 * systemd-sysusers gained a new line type "r" for configuring
1544 which UID/GID ranges to allocate system users/groups
1545 from. Lines of type "u" may now add an additional column
1546 that specifies the home directory for the system user to be
1547 created. Also, systemd-sysusers may now optionally read user
1548 information from STDIN instead of a file. This is useful for
1549 invoking it from RPM preinst scriptlets that need to create
1550 users before the first RPM file is installed since these
1551 files might need to be owned by them. A new
1552 %sysusers_create_inline RPM macro has been introduced to do
1553 just that. systemd-sysusers now updates the shadow files as
1554 well as the user/group databases, which should enhance
1555 compatibility with certain tools like grpck.
1557 * A number of bus APIs of PID 1 now optionally consult
1558 PolicyKit to permit access for otherwise unprivileged
1559 clients under certain conditions. Note that this currently
1560 doesn't support interactive authentication yet, but this is
1561 expected to be added eventually, too.
1563 * /etc/machine-info now has new fields for configuring the
1564 deployment environment of the machine, as well as the
1565 location of the machine. hostnamectl has been updated with
1566 new command to update these fields.
1568 * systemd-timesyncd has been updated to automatically acquire
1569 NTP server information from systemd-networkd, which might
1570 have been discovered via DHCP.
1572 * systemd-resolved now includes a caching DNS stub resolver
1573 and a complete LLMNR name resolution implementation. A new
1574 NSS module "nss-resolve" has been added which can be used
1575 instead of glibc's own "nss-dns" to resolve hostnames via
1576 systemd-resolved. Hostnames, addresses and arbitrary RRs may
1577 be resolved via systemd-resolved D-Bus APIs. In contrast to
1578 the glibc internal resolver systemd-resolved is aware of
1579 multi-homed system, and keeps DNS server and caches separate
1580 and per-interface. Queries are sent simultaneously on all
1581 interfaces that have DNS servers configured, in order to
1582 properly handle VPNs and local LANs which might resolve
1583 separate sets of domain names. systemd-resolved may acquire
1584 DNS server information from systemd-networkd automatically,
1585 which in turn might have discovered them via DHCP. A tool
1586 "systemd-resolve-host" has been added that may be used to
1587 query the DNS logic in resolved. systemd-resolved implements
1588 IDNA and automatically uses IDNA or UTF-8 encoding depending
1589 on whether classic DNS or LLMNR is used as transport. In the
1590 next releases we intend to add a DNSSEC and mDNS/DNS-SD
1591 implementation to systemd-resolved.
1593 * A new NSS module nss-mymachines has been added, that
1594 automatically resolves the names of all local registered
1595 containers to their respective IP addresses.
1597 * A new client tool "networkctl" for systemd-networkd has been
1598 added. It currently is entirely passive and will query
1599 networking configuration from udev, rtnetlink and networkd,
1600 and present it to the user in a very friendly
1601 way. Eventually, we hope to extend it to become a full
1602 control utility for networkd.
1604 * .socket units gained a new DeferAcceptSec= setting that
1605 controls the kernels' TCP_DEFER_ACCEPT sockopt for
1606 TCP. Similar, support for controlling TCP keep-alive
1607 settings has been added (KeepAliveTimeSec=,
1608 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
1609 turning off Nagle's algorithm on TCP has been added
1612 * logind learned a new session type "web", for use in projects
1613 like Cockpit which register web clients as PAM sessions.
1615 * timer units with at least one OnCalendar= setting will now
1616 be started only after timer-sync.target has been
1617 reached. This way they will not elapse before the system
1618 clock has been corrected by a local NTP client or
1619 similar. This is particular useful on RTC-less embedded
1620 machines, that come up with an invalid system clock.
1622 * systemd-nspawn's --network-veth= switch should now result in
1623 stable MAC addresses for both the outer and the inner side
1626 * systemd-nspawn gained a new --volatile= switch for running
1627 container instances with /etc or /var unpopulated.
1629 * The kdbus client code has been updated to use the new Linux
1630 3.17 memfd subsystem instead of the old kdbus-specific one.
1632 * systemd-networkd's DHCP client and server now support
1633 FORCERENEW. There are also new configuration options to
1634 configure the vendor client identifier and broadcast mode
1637 * systemd will no longer inform the kernel about the current
1638 timezone, as this is necessarily incorrect and racy as the
1639 kernel has no understanding of DST and similar
1640 concepts. This hence means FAT timestamps will be always
1641 considered UTC, similar to what Android is already
1642 doing. Also, when the RTC is configured to the local time
1643 (rather than UTC) systemd will never synchronize back to it,
1644 as this might confuse Windows at a later boot.
1646 * systemd-analyze gained a new command "verify" for offline
1647 validation of unit files.
1649 * systemd-networkd gained support for a couple of additional
1650 settings for bonding networking setups. Also, the metric for
1651 statically configured routes may now be configured. For
1652 network interfaces where this is appropriate the peer IP
1653 address may now be configured.
1655 * systemd-networkd's DHCP client will no longer request
1656 broadcasting by default, as this tripped up some networks.
1657 For hardware where broadcast is required the feature should
1658 be switched back on using RequestBroadcast=yes.
1660 * systemd-networkd will now set up IPv4LL addresses (when
1661 enabled) even if DHCP is configured successfully.
1663 * udev will now default to respect network device names given
1664 by the kernel when the kernel indicates that these are
1665 predictable. This behavior can be tweaked by changing
1666 NamePolicy= in the relevant .link file.
1668 * A new library systemd-terminal has been added that
1669 implements full TTY stream parsing and rendering. This
1670 library is supposed to be used later on for implementing a
1671 full userspace VT subsystem, replacing the current kernel
1674 * A new tool systemd-journal-upload has been added to push
1675 journal data to a remote system running
1676 systemd-journal-remote.
1678 * journald will no longer forward all local data to another
1679 running syslog daemon. This change has been made because
1680 rsyslog (which appears to be the most commonly used syslog
1681 implementation these days) no longer makes use of this, and
1682 instead pulls the data out of the journal on its own. Since
1683 forwarding the messages to a non-existent syslog server is
1684 more expensive than we assumed we have now turned this
1685 off. If you run a syslog server that is not a recent rsyslog
1686 version, you have to turn this option on again
1687 (ForwardToSyslog= in journald.conf).
1689 * journald now optionally supports the LZ4 compressor for
1690 larger journal fields. This compressor should perform much
1691 better than XZ which was the previous default.
1693 * machinectl now shows the IP addresses of local containers,
1694 if it knows them, plus the interface name of the container.
1696 * A new tool "systemd-escape" has been added that makes it
1697 easy to escape strings to build unit names and similar.
1699 * sd_notify() messages may now include a new ERRNO= field
1700 which is parsed and collected by systemd and shown among the
1701 "systemctl status" output for a service.
1703 * A new component "systemd-firstboot" has been added that
1704 queries the most basic systemd information (timezone,
1705 hostname, root password) interactively on first
1706 boot. Alternatively it may also be used to provision these
1707 things offline on OS images installed into directories.
1709 * The default sysctl.d/ snippets will now set
1711 net.ipv4.conf.default.promote_secondaries=1
1713 This has the benefit of no flushing secondary IP addresses
1714 when primary addresses are removed.
1716 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
1717 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
1718 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
1719 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
1720 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
1721 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
1722 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
1723 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
1724 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
1725 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
1726 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
1727 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
1728 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
1729 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
1730 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
1732 -- Berlin, 2014-08-19
1736 * A new tool systemd-sysusers has been added. This tool
1737 creates system users and groups in /etc/passwd and
1738 /etc/group, based on static declarative system user/group
1739 definitions in /usr/lib/sysusers.d/. This is useful to
1740 enable factory resets and volatile systems that boot up with
1741 an empty /etc directory, and thus need system users and
1742 groups created during early boot. systemd now also ships
1743 with two default sysusers.d/ files for the most basic
1744 users and groups systemd and the core operating system
1747 * A new tmpfiles snippet has been added that rebuilds the
1748 essential files in /etc on boot, should they be missing.
1750 * A directive for ensuring automatic clean-up of
1751 /var/cache/man/ has been removed from the default
1752 configuration. This line should now be shipped by the man
1753 implementation. The necessary change has been made to the
1754 man-db implementation. Note that you need to update your man
1755 implementation to one that ships this line, otherwise no
1756 automatic clean-up of /var/cache/man will take place.
1758 * A new condition ConditionNeedsUpdate= has been added that
1759 may conditionalize services to only run when /etc or /var
1760 are "older" than the vendor operating system resources in
1761 /usr. This is useful for reconstructing or updating /etc
1762 after an offline update of /usr or a factory reset, on the
1763 next reboot. Services that want to run once after such an
1764 update or reset should use this condition and order
1765 themselves before the new systemd-update-done.service, which
1766 will mark the two directories as fully updated. A number of
1767 service files have been added making use of this, to rebuild
1768 the udev hardware database, the journald message catalog and
1769 dynamic loader cache (ldconfig). The systemd-sysusers tool
1770 described above also makes use of this now. With this in
1771 place it is now possible to start up a minimal operating
1772 system with /etc empty cleanly. For more information on the
1773 concepts involved see this recent blog story:
1775 http://0pointer.de/blog/projects/stateless.html
1777 * A new system group "input" has been introduced, and all
1778 input device nodes get this group assigned. This is useful
1779 for system-level software to get access to input devices. It
1780 complements what is already done for "audio" and "video".
1782 * systemd-networkd learnt minimal DHCPv4 server support in
1783 addition to the existing DHCPv4 client support. It also
1784 learnt DHCPv6 client and IPv6 Router Solicitation client
1785 support. The DHCPv4 client gained support for static routes
1786 passed in from the server. Note that the [DHCPv4] section
1787 known in older systemd-networkd versions has been renamed to
1788 [DHCP] and is now also used by the DHCPv6 client. Existing
1789 .network files using settings of this section should be
1790 updated, though compatibility is maintained. Optionally, the
1791 client hostname may now be sent to the DHCP server.
1793 * networkd gained support for vxlan virtual networks as well
1794 as tun/tap and dummy devices.
1796 * networkd gained support for automatic allocation of address
1797 ranges for interfaces from a system-wide pool of
1798 addresses. This is useful for dynamically managing a large
1799 number of interfaces with a single network configuration
1800 file. In particular this is useful to easily assign
1801 appropriate IP addresses to the veth links of a large number
1802 of nspawn instances.
1804 * RPM macros for processing sysusers, sysctl and binfmt
1805 drop-in snippets at package installation time have been
1808 * The /etc/os-release file should now be placed in
1809 /usr/lib/os-release. The old location is automatically
1810 created as symlink. /usr/lib is the more appropriate
1811 location of this file, since it shall actually describe the
1812 vendor operating system shipped in /usr, and not the
1813 configuration stored in /etc.
1815 * .mount units gained a new boolean SloppyOptions= setting
1816 that maps to mount(8)'s -s option which enables permissive
1817 parsing of unknown mount options.
1819 * tmpfiles learnt a new "L+" directive which creates a symlink
1820 but (unlike "L") deletes a pre-existing file first, should
1821 it already exist and not already be the correct
1822 symlink. Similar, "b+", "c+" and "p+" directives have been
1823 added as well, which create block and character devices, as
1824 well as fifos in the filesystem, possibly removing any
1825 pre-existing files of different types.
1827 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
1828 'argument' field (which so far specified the source to
1829 symlink/copy the files from) is now optional. If omitted the
1830 same file os copied from /usr/share/factory/ suffixed by the
1831 full destination path. This is useful for populating /etc
1832 with essential files, by copying them from vendor defaults
1833 shipped in /usr/share/factory/etc.
1835 * A new command "systemctl preset-all" has been added that
1836 applies the service preset settings to all installed unit
1837 files. A new switch --preset-mode= has been added that
1838 controls whether only enable or only disable operations
1841 * A new command "systemctl is-system-running" has been added
1842 that allows checking the overall state of the system, for
1843 example whether it is fully up and running.
1845 * When the system boots up with an empty /etc, the equivalent
1846 to "systemctl preset-all" is executed during early boot, to
1847 make sure all default services are enabled after a factory
1850 * systemd now contains a minimal preset file that enables the
1851 most basic services systemd ships by default.
1853 * Unit files' [Install] section gained a new DefaultInstance=
1854 field for defining the default instance to create if a
1855 template unit is enabled with no instance specified.
1857 * A new passive target cryptsetup-pre.target has been added
1858 that may be used by services that need to make they run and
1859 finish before the first LUKS cryptographic device is set up.
1861 * The /dev/loop-control and /dev/btrfs-control device nodes
1862 are now owned by the "disk" group by default, opening up
1863 access to this group.
1865 * systemd-coredump will now automatically generate a
1866 stack trace of all core dumps taking place on the system,
1867 based on elfutils' libdw library. This stack trace is logged
1870 * systemd-coredump may now optionally store coredumps directly
1871 on disk (in /var/lib/systemd/coredump, possibly compressed),
1872 instead of storing them unconditionally in the journal. This
1873 mode is the new default. A new configuration file
1874 /etc/systemd/coredump.conf has been added to configure this
1875 and other parameters of systemd-coredump.
1877 * coredumpctl gained a new "info" verb to show details about a
1878 specific coredump. A new switch "-1" has also been added
1879 that makes sure to only show information about the most
1880 recent entry instead of all entries. Also, as the tool is
1881 generally useful now the "systemd-" prefix of the binary
1882 name has been removed. Distributions that want to maintain
1883 compatibility with the old name should add a symlink from
1884 the old name to the new name.
1886 * journald's SplitMode= now defaults to "uid". This makes sure
1887 that unprivileged users can access their own coredumps with
1888 coredumpctl without restrictions.
1890 * New kernel command line options "systemd.wants=" (for
1891 pulling an additional unit during boot), "systemd.mask="
1892 (for masking a specific unit for the boot), and
1893 "systemd.debug-shell" (for enabling the debug shell on tty9)
1894 have been added. This is implemented in the new generator
1895 "systemd-debug-generator".
1897 * systemd-nspawn will now by default filter a couple of
1898 syscalls for containers, among them those required for
1899 kernel module loading, direct x86 IO port access, swap
1900 management, and kexec. Most importantly though
1901 open_by_handle_at() is now prohibited for containers,
1902 closing a hole similar to a recently discussed vulnerability
1903 in docker regarding access to files on file hierarchies the
1904 container should normally not have access to. Note that for
1905 nspawn we generally make no security claims anyway (and
1906 this is explicitly documented in the man page), so this is
1907 just a fix for one of the most obvious problems.
1909 * A new man page file-hierarchy(7) has been added that
1910 contains a minimized, modernized version of the file system
1911 layout systemd expects, similar in style to the FHS
1912 specification or hier(5). A new tool systemd-path(1) has
1913 been added to query many of these paths for the local
1916 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
1917 longer done. Since the directory now has a per-user size
1918 limit, and is cleaned on logout this appears unnecessary,
1919 in particular since this now brings the lifecycle of this
1920 directory closer in line with how IPC objects are handled.
1922 * systemd.pc now exports a number of additional directories,
1923 including $libdir (which is useful to identify the library
1924 path for the primary architecture of the system), and a
1925 couple of drop-in directories.
1927 * udev's predictable network interface names now use the dev_port
1928 sysfs attribute, introduced in linux 3.15 instead of dev_id to
1929 distinguish between ports of the same PCI function. dev_id should
1930 only be used for ports using the same HW address, hence the need
1933 * machined has been updated to export the OS version of a
1934 container (read from /etc/os-release and
1935 /usr/lib/os-release) on the bus. This is now shown in
1936 "machinectl status" for a machine.
1938 * A new service setting RestartForceExitStatus= has been
1939 added. If configured to a set of exit signals or process
1940 return values, the service will be restarted when the main
1941 daemon process exits with any of them, regardless of the
1944 * systemctl's -H switch for connecting to remote systemd
1945 machines has been extended so that it may be used to
1946 directly connect to a specific container on the
1947 host. "systemctl -H root@foobar:waldi" will now connect as
1948 user "root" to host "foobar", and then proceed directly to
1949 the container named "waldi". Note that currently you have to
1950 authenticate as user "root" for this to work, as entering
1951 containers is a privileged operation.
1953 Contributions from: Andreas Henriksson, Benjamin Steinwender,
1954 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
1955 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
1956 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
1957 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
1958 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
1959 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
1960 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
1961 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
1962 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
1963 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
1964 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
1966 -- Berlin, 2014-07-03
1970 * As an experimental feature, udev now tries to lock the
1971 disk device node (flock(LOCK_SH|LOCK_NB)) while it
1972 executes events for the disk or any of its partitions.
1973 Applications like partitioning programs can lock the
1974 disk device node (flock(LOCK_EX)) and claim temporary
1975 device ownership that way; udev will entirely skip all event
1976 handling for this disk and its partitions. If the disk
1977 was opened for writing, the close will trigger a partition
1978 table rescan in udev's "watch" facility, and if needed
1979 synthesize "change" events for the disk and all its partitions.
1980 This is now unconditionally enabled, and if it turns out to
1981 cause major problems, we might turn it on only for specific
1982 devices, or might need to disable it entirely. Device Mapper
1983 devices are excluded from this logic.
1985 * We temporarily dropped the "-l" switch for fsck invocations,
1986 since they collide with the flock() logic above. util-linux
1987 upstream has been changed already to avoid this conflict,
1988 and we will readd "-l" as soon as util-linux with this
1989 change has been released.
1991 * The dependency on libattr has been removed. Since a long
1992 time, the extended attribute calls have moved to glibc, and
1993 libattr is thus unnecessary.
1995 * Virtualization detection works without priviliges now. This
1996 means the systemd-detect-virt binary no longer requires
1997 CAP_SYS_PTRACE file capabilities, and our daemons can run
1998 with fewer privileges.
2000 * systemd-networkd now runs under its own "systemd-network"
2001 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
2002 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
2003 loses the ability to write to files owned by root this way.
2005 * Similar, systemd-resolved now runs under its own
2006 "systemd-resolve" user with no capabilities remaining.
2008 * Similar, systemd-bus-proxyd now runs under its own
2009 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
2011 * systemd-networkd gained support for setting up "veth"
2012 virtual ethernet devices for container connectivity, as well
2013 as GRE and VTI tunnels.
2015 * systemd-networkd will no longer automatically attempt to
2016 manually load kernel modules necessary for certain tunnel
2017 transports. Instead, it is assumed the kernel loads them
2018 automatically when required. This only works correctly on
2019 very new kernels. On older kernels, please consider adding
2020 the kernel modules to /etc/modules-load.d/ as a work-around.
2022 * The resolv.conf file systemd-resolved generates has been
2023 moved to /run/systemd/resolve/. If you have a symlink from
2024 /etc/resolv.conf, it might be necessary to correct it.
2026 * Two new service settings, ProtectHome= and ProtectSystem=,
2027 have been added. When enabled, they will make the user data
2028 (such as /home) inaccessible or read-only and the system
2029 (such as /usr) read-only, for specific services. This allows
2030 very light-weight per-service sandboxing to avoid
2031 modifications of user data or system files from
2032 services. These two new switches have been enabled for all
2033 of systemd's long-running services, where appropriate.
2035 * Socket units gained new SocketUser= and SocketGroup=
2036 settings to set the owner user and group of AF_UNIX sockets
2037 and FIFOs in the file system.
2039 * Socket units gained a new RemoveOnStop= setting. If enabled,
2040 all FIFOS and sockets in the file system will be removed
2041 when the specific socket unit is stopped.
2043 * Socket units gained a new Symlinks= setting. It takes a list
2044 of symlinks to create to file system sockets or FIFOs
2045 created by the specific Unix sockets. This is useful to
2046 manage symlinks to socket nodes with the same life-cycle as
2049 * The /dev/log socket and /dev/initctl FIFO have been moved to
2050 /run, and have been replaced by symlinks. This allows
2051 connecting to these facilities even if PrivateDevices=yes is
2052 used for a service (which makes /dev/log itself unavailable,
2053 but /run is left). This also has the benefit of ensuring
2054 that /dev only contains device nodes, directories and
2055 symlinks, and nothing else.
2057 * sd-daemon gained two new calls sd_pid_notify() and
2058 sd_pid_notifyf(). They are similar to sd_notify() and
2059 sd_notifyf(), but allow overriding of the source PID of
2060 notification messages if permissions permit this. This is
2061 useful to send notify messages on behalf of a different
2062 process (for example, the parent process). The
2063 systemd-notify tool has been updated to make use of this
2064 when sending messages (so that notification messages now
2065 originate from the shell script invoking systemd-notify and
2066 not the systemd-notify process itself. This should minimize
2067 a race where systemd fails to associate notification
2068 messages to services when the originating process already
2071 * A new "on-abnormal" setting for Restart= has been added. If
2072 set, it will result in automatic restarts on all "abnormal"
2073 reasons for a process to exit, which includes unclean
2074 signals, core dumps, timeouts and watchdog timeouts, but
2075 does not include clean and unclean exit codes or clean
2076 signals. Restart=on-abnormal is an alternative for
2077 Restart=on-failure for services that shall be able to
2078 terminate and avoid restarts on certain errors, by
2079 indicating so with an unclean exit code. Restart=on-failure
2080 or Restart=on-abnormal is now the recommended setting for
2081 all long-running services.
2083 * If the InaccessibleDirectories= service setting points to a
2084 mount point (or if there are any submounts contained within
2085 it), it is now attempted to completely unmount it, to make
2086 the file systems truly unavailable for the respective
2089 * The ReadOnlyDirectories= service setting and
2090 systemd-nspawn's --read-only parameter are now recursively
2091 applied to all submounts, too.
2093 * Mount units may now be created transiently via the bus APIs.
2095 * The support for SysV and LSB init scripts has been removed
2096 from the systemd daemon itself. Instead, it is now
2097 implemented as a generator that creates native systemd units
2098 from these scripts when needed. This enables us to remove a
2099 substantial amount of legacy code from PID 1, following the
2100 fact that many distributions only ship a very small number
2101 of LSB/SysV init scripts nowadays.
2103 * Privileged Xen (dom0) domains are not considered
2104 virtualization anymore by the virtualization detection
2105 logic. After all, they generally have unrestricted access to
2106 the hardware and usually are used to manage the unprivileged
2109 * systemd-tmpfiles gained a new "C" line type, for copying
2110 files or entire directories.
2112 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
2113 lines. So far, they have been non-globbing versions of the
2114 latter, and have thus been redundant. In future, it is
2115 recommended to only use "z". "m" has hence been removed
2116 from the documentation, even though it stays supported.
2118 * A tmpfiles snippet to recreate the most basic structure in
2119 /var has been added. This is enough to create the /var/run →
2120 /run symlink and create a couple of structural
2121 directories. This allows systems to boot up with an empty or
2122 volatile /var. Of course, while with this change, the core OS
2123 now is capable with dealing with a volatile /var, not all
2124 user services are ready for it. However, we hope that sooner
2125 or later, many service daemons will be changed upstream so
2126 that they are able to automatically create their necessary
2127 directories in /var at boot, should they be missing. This is
2128 the first step to allow state-less systems that only require
2129 the vendor image for /usr to boot.
2131 * systemd-nspawn has gained a new --tmpfs= switch to mount an
2132 empty tmpfs instance to a specific directory. This is
2133 particularly useful for making use of the automatic
2134 reconstruction of /var (see above), by passing --tmpfs=/var.
2136 * Access modes specified in tmpfiles snippets may now be
2137 prefixed with "~", which indicates that they shall be masked
2138 by whether the existing file or directory is currently
2139 writable, readable or executable at all. Also, if specified,
2140 the sgid/suid/sticky bits will be masked for all
2143 * A new passive target unit "network-pre.target" has been
2144 added which is useful for services that shall run before any
2145 network is configured, for example firewall scripts.
2147 * The "floppy" group that previously owned the /dev/fd*
2148 devices is no longer used. The "disk" group is now used
2149 instead. Distributions should probably deprecate usage of
2152 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
2153 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
2154 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
2155 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
2156 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
2157 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
2160 -- Berlin, 2014-06-11
2164 * A new "systemd-timesyncd" daemon has been added for
2165 synchronizing the system clock across the network. It
2166 implements an SNTP client. In contrast to NTP
2167 implementations such as chrony or the NTP reference server,
2168 this only implements a client side, and does not bother with
2169 the full NTP complexity, focusing only on querying time from
2170 one remote server and synchronizing the local clock to
2171 it. Unless you intend to serve NTP to networked clients or
2172 want to connect to local hardware clocks, this simple NTP
2173 client should be more than appropriate for most
2174 installations. The daemon runs with minimal privileges, and
2175 has been hooked up with networkd to only operate when
2176 network connectivity is available. The daemon saves the
2177 current clock to disk every time a new NTP sync has been
2178 acquired, and uses this to possibly correct the system clock
2179 early at bootup, in order to accommodate for systems that
2180 lack an RTC such as the Raspberry Pi and embedded devices,
2181 and to make sure that time monotonically progresses on these
2182 systems, even if it is not always correct. To make use of
2183 this daemon, a new system user and group "systemd-timesync"
2184 needs to be created on installation of systemd.
2186 * The queue "seqnum" interface of libudev has been disabled, as
2187 it was generally incompatible with device namespacing as
2188 sequence numbers of devices go "missing" if the devices are
2189 part of a different namespace.
2191 * "systemctl list-timers" and "systemctl list-sockets" gained
2192 a --recursive switch for showing units of these types also
2193 for all local containers, similar in style to the already
2194 supported --recursive switch for "systemctl list-units".
2196 * A new RebootArgument= setting has been added for service
2197 units, which may be used to specify a kernel reboot argument
2198 to use when triggering reboots with StartLimitAction=.
2200 * A new FailureAction= setting has been added for service
2201 units which may be used to specify an operation to trigger
2202 when a service fails. This works similarly to
2203 StartLimitAction=, but unlike it, controls what is done
2204 immediately rather than only after several attempts to
2205 restart the service in question.
2207 * hostnamed got updated to also expose the kernel name,
2208 release, and version on the bus. This is useful for
2209 executing commands like hostnamectl with the -H switch.
2210 systemd-analyze makes use of this to properly display
2211 details when running non-locally.
2213 * The bootchart tool can now show cgroup information in the
2214 graphs it generates.
2216 * The CFS CPU quota cgroup attribute is now exposed for
2217 services. The new CPUQuota= switch has been added for this
2218 which takes a percentage value. Setting this will have the
2219 result that a service may never get more CPU time than the
2220 specified percentage, even if the machine is otherwise idle.
2222 * systemd-networkd learned IPIP and SIT tunnel support.
2224 * LSB init scripts exposing a dependency on $network will now
2225 get a dependency on network-online.target rather than simply
2226 network.target. This should bring LSB handling closer to
2227 what it was on SysV systems.
2229 * A new fsck.repair= kernel option has been added to control
2230 how fsck shall deal with unclean file systems at boot.
2232 * The (.ini) configuration file parser will now silently
2233 ignore sections whose name begins with "X-". This may be
2234 used to maintain application-specific extension sections in unit
2237 * machined gained a new API to query the IP addresses of
2238 registered containers. "machinectl status" has been updated
2239 to show these addresses in its output.
2241 * A new call sd_uid_get_display() has been added to the
2242 sd-login APIs for querying the "primary" session of a
2243 user. The "primary" session of the user is elected from the
2244 user's sessions and generally a graphical session is
2245 preferred over a text one.
2247 * A minimal systemd-resolved daemon has been added. It
2248 currently simply acts as a companion to systemd-networkd and
2249 manages resolv.conf based on per-interface DNS
2250 configuration, possibly supplied via DHCP. In the long run
2251 we hope to extend this into a local DNSSEC enabled DNS and
2254 * The systemd-networkd-wait-online tool is now enabled by
2255 default. It will delay network-online.target until a network
2256 connection has been configured. The tool primarily integrates
2257 with networkd, but will also make a best effort to make sense
2258 of network configuration performed in some other way.
2260 * Two new service options StartupCPUShares= and
2261 StartupBlockIOWeight= have been added that work similarly to
2262 CPUShares= and BlockIOWeight= however only apply during
2263 system startup. This is useful to prioritize certain services
2264 differently during bootup than during normal runtime.
2266 * hostnamed has been changed to prefer the statically
2267 configured hostname in /etc/hostname (unless set to
2268 'localhost' or empty) over any dynamic one supplied by
2269 dhcp. With this change, the rules for picking the hostname
2270 match more closely the rules of other configuration settings
2271 where the local administrator's configuration in /etc always
2272 overrides any other settings.
2274 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
2275 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
2276 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
2277 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
2278 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
2279 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
2280 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
2281 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
2282 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
2283 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
2284 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
2285 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
2286 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
2287 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
2288 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
2289 Lindskog, WaLyong Cho, Will Woods, Zbigniew
2292 -- Beijing, 2014-05-28
2296 * When restoring the screen brightness at boot, stay away from
2297 the darkest setting or from the lowest 5% of the available
2298 range, depending on which is the larger value of both. This
2299 should effectively protect the user from rebooting into a
2300 black screen, should the brightness have been set to minimum
2303 * sd-login gained a new sd_machine_get_class() call to
2304 determine the class ("vm" or "container") of a machine
2305 registered with machined.
2307 * sd-login gained new calls
2308 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
2309 to query the identity of the peer of a local AF_UNIX
2310 connection. They operate similarly to their sd_pid_get_xyz()
2313 * PID 1 will now maintain a system-wide system state engine
2314 with the states "starting", "running", "degraded",
2315 "maintenance", "stopping". These states are bound to system
2316 startup, normal runtime, runtime with at least one failed
2317 service, rescue/emergency mode and system shutdown. This
2318 state is shown in the "systemctl status" output when no unit
2319 name is passed. It is useful to determine system state, in
2320 particularly when doing so for many systems or containers at
2323 * A new command "list-machines" has been added to "systemctl"
2324 that lists all local OS containers and shows their system
2325 state (see above), if systemd runs inside of them.
2327 * systemctl gained a new "-r" switch to recursively enumerate
2328 units on all local containers, when used with the
2329 "list-unit" command (which is the default one that is
2330 executed when no parameters are specified).
2332 * The GPT automatic partition discovery logic will now honour
2333 two GPT partition flags: one may be set on a partition to
2334 cause it to be mounted read-only, and the other may be set
2335 on a partition to ignore it during automatic discovery.
2337 * Two new GPT type UUIDs have been added for automatic root
2338 partition discovery, for 32-bit and 64-bit ARM. This is not
2339 particularly useful for discovering the root directory on
2340 these architectures during bare-metal boots (since UEFI is
2341 not common there), but still very useful to allow booting of
2342 ARM disk images in nspawn with the -i option.
2344 * MAC addresses of interfaces created with nspawn's
2345 --network-interface= switch will now be generated from the
2346 machine name, and thus be stable between multiple invocations
2349 * logind will now automatically remove all IPC objects owned
2350 by a user if she or he fully logs out. This makes sure that
2351 users who are logged out cannot continue to consume IPC
2352 resources. This covers SysV memory, semaphores and message
2353 queues as well as POSIX shared memory and message
2354 queues. Traditionally, SysV and POSIX IPC had no life-cycle
2355 limits. With this functionality, that is corrected. This may
2356 be turned off by using the RemoveIPC= switch of logind.conf.
2358 * The systemd-machine-id-setup and tmpfiles tools gained a
2359 --root= switch to operate on a specific root directory,
2362 * journald can now forward logged messages to the TTYs of all
2363 logged in users ("wall"). This is the default for all
2364 emergency messages now.
2366 * A new tool systemd-journal-remote has been added to stream
2367 journal log messages across the network.
2369 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
2370 controller trees are mounted into it. Note that the
2371 directories mounted beneath it are not read-only. This is a
2372 security measure and is particularly useful because glibc
2373 actually includes a search logic to pick any tmpfs it can
2374 find to implement shm_open() if /dev/shm is not available
2375 (which it might very well be in namespaced setups).
2377 * machinectl gained a new "poweroff" command to cleanly power
2378 down a local OS container.
2380 * The PrivateDevices= unit file setting will now also drop the
2381 CAP_MKNOD capability from the capability bound set, and
2382 imply DevicePolicy=closed.
2384 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
2385 comprehensively on all long-running systemd services where
2386 this is appropriate.
2388 * systemd-udevd will now run in a disassociated mount
2389 namespace. To mount directories from udev rules, make sure to
2390 pull in mount units via SYSTEMD_WANTS properties.
2392 * The kdbus support gained support for uploading policy into
2393 the kernel. sd-bus gained support for creating "monitoring"
2394 connections that can eavesdrop into all bus communication
2395 for debugging purposes.
2397 * Timestamps may now be specified in seconds since the UNIX
2398 epoch Jan 1st, 1970 by specifying "@" followed by the value
2401 * Native tcpwrap support in systemd has been removed. tcpwrap
2402 is old code, not really maintained anymore and has serious
2403 shortcomings, and better options such as firewalls
2404 exist. For setups that require tcpwrap usage, please
2405 consider invoking your socket-activated service via tcpd,
2406 like on traditional inetd.
2408 * A new system.conf configuration option
2409 DefaultTimerAccuracySec= has been added that controls the
2410 default AccuracySec= setting of .timer units.
2412 * Timer units gained a new WakeSystem= switch. If enabled,
2413 timers configured this way will cause the system to resume
2414 from system suspend (if the system supports that, which most
2417 * Timer units gained a new Persistent= switch. If enabled,
2418 timers configured this way will save to disk when they have
2419 been last triggered. This information is then used on next
2420 reboot to possible execute overdue timer events, that
2421 could not take place because the system was powered off.
2422 This enables simple anacron-like behaviour for timer units.
2424 * systemctl's "list-timers" will now also list the time a
2425 timer unit was last triggered in addition to the next time
2426 it will be triggered.
2428 * systemd-networkd will now assign predictable IPv4LL
2429 addresses to its local interfaces.
2431 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
2432 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
2433 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
2434 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
2435 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
2436 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
2437 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
2438 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
2441 -- Berlin, 2014-03-25
2445 * A new unit file setting RestrictAddressFamilies= has been
2446 added to restrict which socket address families unit
2447 processes gain access to. This takes address family names
2448 like "AF_INET" or "AF_UNIX", and is useful to minimize the
2449 attack surface of services via exotic protocol stacks. This
2450 is built on seccomp system call filters.
2452 * Two new unit file settings RuntimeDirectory= and
2453 RuntimeDirectoryMode= have been added that may be used to
2454 manage a per-daemon runtime directories below /run. This is
2455 an alternative for setting up directory permissions with
2456 tmpfiles snippets, and has the advantage that the runtime
2457 directory's lifetime is bound to the daemon runtime and that
2458 the daemon starts up with an empty directory each time. This
2459 is particularly useful when writing services that drop
2460 privileges using the User= or Group= setting.
2462 * The DeviceAllow= unit setting now supports globbing for
2463 matching against device group names.
2465 * The systemd configuration file system.conf gained new
2466 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
2467 DefaultMemoryAccounting= to globally turn on/off accounting
2468 for specific resources (cgroups) for all units. These
2469 settings may still be overridden individually in each unit
2472 * systemd-gpt-auto-generator is now able to discover /srv and
2473 root partitions in addition to /home and swap partitions. It
2474 also supports LUKS-encrypted partitions now. With this in
2475 place, automatic discovery of partitions to mount following
2476 the Discoverable Partitions Specification
2477 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
2478 is now a lot more complete. This allows booting without
2479 /etc/fstab and without root= on the kernel command line on
2480 systems prepared appropriately.
2482 * systemd-nspawn gained a new --image= switch which allows
2483 booting up disk images and Linux installations on any block
2484 device that follow the Discoverable Partitions Specification
2485 (see above). This means that installations made with
2486 appropriately updated installers may now be started and
2487 deployed using container managers, completely
2488 unmodified. (We hope that libvirt-lxc will add support for
2489 this feature soon, too.)
2491 * systemd-nspawn gained a new --network-macvlan= setting to
2492 set up a private macvlan interface for the
2493 container. Similarly, systemd-networkd gained a new
2494 Kind=macvlan setting in .netdev files.
2496 * systemd-networkd now supports configuring local addresses
2499 * A new tool systemd-network-wait-online has been added to
2500 synchronously wait for network connectivity using
2503 * The sd-bus.h bus API gained a new sd_bus_track object for
2504 tracking the life-cycle of bus peers. Note that sd-bus.h is
2505 still not a public API though (unless you specify
2506 --enable-kdbus on the configure command line, which however
2507 voids your warranty and you get no API stability guarantee).
2509 * The $XDG_RUNTIME_DIR runtime directories for each user are
2510 now individual tmpfs instances, which has the benefit of
2511 introducing separate pools for each user, with individual
2512 size limits, and thus making sure that unprivileged clients
2513 can no longer negatively impact the system or other users by
2514 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
2515 RuntimeDirectorySize= has been introduced that allows
2516 controlling the default size limit for all users. It
2517 defaults to 10% of the available physical memory. This is no
2518 replacement for quotas on tmpfs though (which the kernel
2519 still does not support), as /dev/shm and /tmp are still
2520 shared resources used by both the system and unprivileged
2523 * logind will now automatically turn off automatic suspending
2524 on laptop lid close when more than one display is
2525 connected. This was previously expected to be implemented
2526 individually in desktop environments (such as GNOME),
2527 however has been added to logind now, in order to fix a
2528 boot-time race where a desktop environment might not have
2529 been started yet and thus not been able to take an inhibitor
2530 lock at the time where logind already suspends the system
2531 due to a closed lid.
2533 * logind will now wait at least 30s after each system
2534 suspend/resume cycle, and 3min after system boot before
2535 suspending the system due to a closed laptop lid. This
2536 should give USB docking stations and similar enough time to
2537 be probed and configured after system resume and boot in
2538 order to then act as suspend blocker.
2540 * systemd-run gained a new --property= setting which allows
2541 initialization of resource control properties (and others)
2542 for the created scope or service unit. Example: "systemd-run
2543 --property=BlockIOWeight=10 updatedb" may be used to run
2544 updatedb at a low block IO scheduling weight.
2546 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
2547 now also work in --scope mode.
2549 * When systemd is compiled with kdbus support, basic support
2550 for enforced policies is now in place. (Note that enabling
2551 kdbus still voids your warranty and no API compatibility
2554 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
2555 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
2556 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
2557 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
2558 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
2559 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
2560 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
2561 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
2562 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
2563 Zbigniew Jędrzejewski-Szmek
2565 -- Berlin, 2014-03-12
2569 * systemd will now relabel /dev after loading the SMACK policy
2570 according to SMACK rules.
2572 * A new unit file option AppArmorProfile= has been added to
2573 set the AppArmor profile for the processes of a unit.
2575 * A new condition check ConditionArchitecture= has been added
2576 to conditionalize units based on the system architecture, as
2577 reported by uname()'s "machine" field.
2579 * systemd-networkd now supports matching on the system
2580 virtualization, architecture, kernel command line, host name
2583 * logind is now a lot more aggressive when suspending the
2584 machine due to a closed laptop lid. Instead of acting only
2585 on the lid close action, it will continuously watch the lid
2586 status and act on it. This is useful for laptops where the
2587 power button is on the outside of the chassis so that it can
2588 be reached without opening the lid (such as the Lenovo
2589 Yoga). On those machines, logind will now immediately
2590 re-suspend the machine if the power button has been
2591 accidentally pressed while the laptop was suspended and in a
2592 backpack or similar.
2594 * logind will now watch SW_DOCK switches and inhibit reaction
2595 to the lid switch if it is pressed. This means that logind
2596 will not suspend the machine anymore if the lid is closed
2597 and the system is docked, if the laptop supports SW_DOCK
2598 notifications via the input layer. Note that ACPI docking
2599 stations do not generate this currently. Also note that this
2600 logic is usually not fully sufficient and Desktop
2601 Environments should take a lid switch inhibitor lock when an
2602 external display is connected, as systemd will not watch
2605 * nspawn will now make use of the devices cgroup controller by
2606 default, and only permit creation of and access to the usual
2607 API device nodes like /dev/null or /dev/random, as well as
2608 access to (but not creation of) the pty devices.
2610 * We will now ship a default .network file for
2611 systemd-networkd that automatically configures DHCP for
2612 network interfaces created by nspawn's --network-veth or
2613 --network-bridge= switches.
2615 * systemd will now understand the usual M, K, G, T suffixes
2616 according to SI conventions (i.e. to the base 1000) when
2617 referring to throughput and hardware metrics. It will stay
2618 with IEC conventions (i.e. to the base 1024) for software
2619 metrics, according to what is customary according to
2620 Wikipedia. We explicitly document which base applies for
2621 each configuration option.
2623 * The DeviceAllow= setting in unit files now supports a syntax
2624 to whitelist an entire group of devices node majors at once,
2625 based on the /proc/devices listing. For example, with the
2626 string "char-pts", it is now possible to whitelist all
2627 current and future pseudo-TTYs at once.
2629 * sd-event learned a new "post" event source. Event sources of
2630 this type are triggered by the dispatching of any event
2631 source of a type that is not "post". This is useful for
2632 implementing clean-up and check event sources that are
2633 triggered by other work being done in the program.
2635 * systemd-networkd is no longer statically enabled, but uses
2636 the usual [Install] sections so that it can be
2637 enabled/disabled using systemctl. It still is enabled by
2640 * When creating a veth interface pair with systemd-nspawn, the
2641 host side will now be prefixed with "vb-" if
2642 --network-bridge= is used, and with "ve-" if --network-veth
2643 is used. This way, it is easy to distinguish these cases on
2644 the host, for example to apply different configuration to
2645 them with systemd-networkd.
2647 * The compatibility libraries for libsystemd-journal.so,
2648 libsystem-id128.so, libsystemd-login.so and
2649 libsystemd-daemon.so do not make use of IFUNC
2650 anymore. Instead, we now build libsystemd.so multiple times
2651 under these alternative names. This means that the footprint
2652 is drastically increased, but given that these are
2653 transitional compatibility libraries, this should not matter
2654 much. This change has been made necessary to support the ARM
2655 platform for these compatibility libraries, as the ARM
2656 toolchain is not really at the same level as the toolchain
2657 for other architectures like x86 and does not support
2658 IFUNC. Please make sure to use --enable-compat-libs only
2659 during a transitional period!
2661 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
2662 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
2663 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
2664 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
2665 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
2666 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
2667 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
2668 Zbigniew Jędrzejewski-Szmek
2670 -- Berlin, 2014-02-24
2674 * A new component "systemd-networkd" has been added that can
2675 be used to configure local network interfaces statically or
2676 via DHCP. It is capable of bringing up bridges, VLANs, and
2677 bonding. Currently, no hook-ups for interactive network
2678 configuration are provided. Use this for your initrd,
2679 container, embedded, or server setup if you need a simple,
2680 yet powerful, network configuration solution. This
2681 configuration subsystem is quite nifty, as it allows wildcard
2682 hotplug matching in interfaces. For example, with a single
2683 configuration snippet, you can configure that all Ethernet
2684 interfaces showing up are automatically added to a bridge,
2685 or similar. It supports link-sensing and more.
2687 * A new tool "systemd-socket-proxyd" has been added which can
2688 act as a bidirectional proxy for TCP sockets. This is
2689 useful for adding socket activation support to services that
2690 do not actually support socket activation, including virtual
2691 machines and the like.
2693 * Add a new tool to save/restore rfkill state on
2696 * Save/restore state of keyboard backlights in addition to
2697 display backlights on shutdown/boot.
2699 * udev learned a new SECLABEL{} construct to label device
2700 nodes with a specific security label when they appear. For
2701 now, only SECLABEL{selinux} is supported, but the syntax is
2702 prepared for additional security frameworks.
2704 * udev gained a new scheme to configure link-level attributes
2705 from files in /etc/systemd/network/*.link. These files can
2706 match against MAC address, device path, driver name and type,
2707 and will apply attributes like the naming policy, link speed,
2708 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
2709 address assignment policy (randomized, ...).
2711 * The configuration of network interface naming rules for
2712 "permanent interface names" has changed: a new NamePolicy=
2713 setting in the [Link] section of .link files determines the
2714 priority of possible naming schemes (onboard, slot, mac,
2715 path). The default value of this setting is determined by
2716 /usr/lib/net/links/99-default.link. Old
2717 80-net-name-slot.rules udev configuration file has been
2718 removed, so local configuration overriding this file should
2719 be adapated to override 99-default.link instead.
2721 * When the User= switch is used in a unit file, also
2722 initialize $SHELL= based on the user database entry.
2724 * systemd no longer depends on libdbus. All communication is
2725 now done with sd-bus, systemd's low-level bus library
2728 * kdbus support has been added to PID 1 itself. When kdbus is
2729 enabled, this causes PID 1 to set up the system bus and
2730 enable support for a new ".busname" unit type that
2731 encapsulates bus name activation on kdbus. It works a little
2732 bit like ".socket" units, except for bus names. A new
2733 generator has been added that converts classic dbus1 service
2734 activation files automatically into native systemd .busname
2737 * sd-bus: add a light-weight vtable implementation that allows
2738 defining objects on the bus with a simple static const
2739 vtable array of its methods, signals and properties.
2741 * systemd will not generate or install static dbus
2742 introspection data anymore to /usr/share/dbus-1/interfaces,
2743 as the precise format of these files is unclear, and
2744 nothing makes use of it.
2746 * A proxy daemon is now provided to proxy clients connecting
2747 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
2748 compatibility with classic D-Bus.
2750 * A bus driver implementation has been added that supports the
2751 classic D-Bus bus driver calls on kdbus, also for
2752 compatibility purposes.
2754 * A new API "sd-event.h" has been added that implements a
2755 minimal event loop API built around epoll. It provides a
2756 couple of features that direct epoll usage is lacking:
2757 prioritization of events, scales to large numbers of timer
2758 events, per-event timer slack (accuracy), system-wide
2759 coalescing of timer events, exit handlers, watchdog
2760 supervision support using systemd's sd_notify() API, child
2763 * A new API "sd-rntl.h" has been added that provides an API
2764 around the route netlink interface of the kernel, similar in
2765 style to "sd-bus.h".
2767 * A new API "sd-dhcp-client.h" has been added that provides a
2768 small DHCPv4 client-side implementation. This is used by
2771 * There is a new kernel command line option
2772 "systemd.restore_state=0|1". When set to "0", none of the
2773 systemd tools will restore saved runtime state to hardware
2774 devices. More specifically, the rfkill and backlight states
2777 * The FsckPassNo= compatibility option in mount/service units
2778 has been removed. The fstab generator will now add the
2779 necessary dependencies automatically, and does not require
2780 PID1's support for that anymore.
2782 * journalctl gained a new switch, --list-boots, that lists
2783 recent boots with their times and boot IDs.
2785 * The various tools like systemctl, loginctl, timedatectl,
2786 busctl, systemd-run, ... have gained a new switch "-M" to
2787 connect to a specific, local OS container (as direct
2788 connection, without requiring SSH). This works on any
2789 container that is registered with machined, such as those
2790 created by libvirt-lxc or nspawn.
2792 * systemd-run and systemd-analyze also gained support for "-H"
2793 to connect to remote hosts via SSH. This is particularly
2794 useful for systemd-run because it enables queuing of jobs
2795 onto remote systems.
2797 * machinectl gained a new command "login" to open a getty
2798 login in any local container. This works with any container
2799 that is registered with machined (such as those created by
2800 libvirt-lxc or nspawn), and which runs systemd inside.
2802 * machinectl gained a new "reboot" command that may be used to
2803 trigger a reboot on a specific container that is registered
2804 with machined. This works on any container that runs an init
2805 system of some kind.
2807 * systemctl gained a new "list-timers" command to print a nice
2808 listing of installed timer units with the times they elapse
2811 * Alternative reboot() parameters may now be specified on the
2812 "systemctl reboot" command line and are passed to the
2813 reboot() system call.
2815 * systemctl gained a new --job-mode= switch to configure the
2816 mode to queue a job with. This is a more generic version of
2817 --fail, --irreversible, and --ignore-dependencies, which are
2818 still available but not advertised anymore.
2820 * /etc/systemd/system.conf gained new settings to configure
2821 various default timeouts of units, as well as the default
2822 start limit interval and burst. These may still be overridden
2825 * PID1 will now export on the bus profile data of the security
2826 policy upload process (such as the SELinux policy upload to
2829 * journald: when forwarding logs to the console, include
2830 timestamps (following the setting in
2831 /sys/module/printk/parameters/time).
2833 * OnCalendar= in timer units now understands the special
2834 strings "yearly" and "annually". (Both are equivalent)
2836 * The accuracy of timer units is now configurable with the new
2837 AccuracySec= setting. It defaults to 1min.
2839 * A new dependency type JoinsNamespaceOf= has been added that
2840 allows running two services within the same /tmp and network
2841 namespace, if PrivateNetwork= or PrivateTmp= are used.
2843 * A new command "cat" has been added to systemctl. It outputs
2844 the original unit file of a unit, and concatenates the
2845 contents of additional "drop-in" unit file snippets, so that
2846 the full configuration is shown.
2848 * systemctl now supports globbing on the various "list-xyz"
2849 commands, like "list-units" or "list-sockets", as well as on
2850 those commands which take multiple unit names.
2852 * journalctl's --unit= switch gained support for globbing.
2854 * All systemd daemons now make use of the watchdog logic so
2855 that systemd automatically notices when they hang.
2857 * If the $container_ttys environment variable is set,
2858 getty-generator will automatically spawn a getty for each
2859 listed tty. This is useful for container managers to request
2860 login gettys to be spawned on as many ttys as needed.
2862 * %h, %s, %U specifier support is not available anymore when
2863 used in unit files for PID 1. This is because NSS calls are
2864 not safe from PID 1. They stay available for --user
2865 instances of systemd, and as special case for the root user.
2867 * loginctl gained a new "--no-legend" switch to turn off output
2870 * The "sd-login.h" API gained three new calls:
2871 sd_session_is_remote(), sd_session_get_remote_user(),
2872 sd_session_get_remote_host() to query information about
2875 * The udev hardware database now also carries vendor/product
2876 information of SDIO devices.
2878 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
2879 determine whether watchdog notifications are requested by
2882 * Socket-activated per-connection services now include a
2883 short description of the connection parameters in the
2886 * tmpfiles gained a new "--boot" option. When this is not used,
2887 only lines where the command character is not suffixed with
2888 "!" are executed. When this option is specified, those
2889 options are executed too. This partitions tmpfiles
2890 directives into those that can be safely executed at any
2891 time, and those which should be run only at boot (for
2892 example, a line that creates /run/nologin).
2894 * A new API "sd-resolve.h" has been added which provides a simple
2895 asynchronous wrapper around glibc NSS host name resolution
2896 calls, such as getaddrinfo(). In contrast to glibc's
2897 getaddrinfo_a(), it does not use signals. In contrast to most
2898 other asynchronous name resolution libraries, this one does
2899 not reimplement DNS, but reuses NSS, so that alternate
2900 host name resolution systems continue to work, such as mDNS,
2901 LDAP, etc. This API is based on libasyncns, but it has been
2902 cleaned up for inclusion in systemd.
2904 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
2905 "sd-daemon.h" are no longer found in individual libraries
2906 libsystemd-journal.so, libsystemd-login.so,
2907 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
2908 merged them into a single library, libsystemd.so, which
2909 provides all symbols. The reason for this is cyclic
2910 dependencies, as these libraries tend to use each other's
2911 symbols. So far, we have managed to workaround that by linking
2912 a copy of a good part of our code into each of these
2913 libraries again and again, which, however, makes certain
2914 things hard to do, like sharing static variables. Also, it
2915 substantially increases footprint. With this change, there
2916 is only one library for the basic APIs systemd
2917 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
2918 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
2919 library as well, however are subject to the --enable-kdbus
2920 switch (see below). Note that "sd-dhcp-client.h" is not part
2921 of this library (this is because it only consumes, never
2922 provides, services of/to other APIs). To make the transition
2923 easy from the separate libraries to the unified one, we
2924 provide the --enable-compat-libs compile-time switch which
2925 will generate stub libraries that are compatible with the
2926 old ones but redirect all calls to the new one.
2928 * All of the kdbus logic and the new APIs "sd-bus.h",
2929 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
2930 and "sd-utf8.h" are compile-time optional via the
2931 "--enable-kdbus" switch, and they are not compiled in by
2932 default. To make use of kdbus, you have to explicitly enable
2933 the switch. Note however, that neither the kernel nor the
2934 userspace API for all of this is considered stable yet. We
2935 want to maintain the freedom to still change the APIs for
2936 now. By specifying this build-time switch, you acknowledge
2937 that you are aware of the instability of the current
2940 * Also, note that while kdbus is pretty much complete,
2941 it lacks one thing: proper policy support. This means you
2942 can build a fully working system with all features; however,
2943 it will be highly insecure. Policy support will be added in
2944 one of the next releases, at the same time that we will
2945 declare the APIs stable.
2947 * When the kernel command line argument "kdbus" is specified,
2948 systemd will automatically load the kdbus.ko kernel module. At
2949 this stage of development, it is only useful for testing kdbus
2950 and should not be used in production. Note: if "--enable-kdbus"
2951 is specified, and the kdbus.ko kernel module is available, and
2952 "kdbus" is added to the kernel command line, the entire system
2953 runs with kdbus instead of dbus-daemon, with the above mentioned
2954 problem of missing the system policy enforcement. Also a future
2955 version of kdbus.ko or a newer systemd will not be compatible with
2956 each other, and will unlikely be able to boot the machine if only
2957 one of them is updated.
2959 * systemctl gained a new "import-environment" command which
2960 uploads the caller's environment (or parts thereof) into the
2961 service manager so that it is inherited by services started
2962 by the manager. This is useful to upload variables like
2963 $DISPLAY into the user service manager.
2965 * A new PrivateDevices= switch has been added to service units
2966 which allows running a service with a namespaced /dev
2967 directory that does not contain any device nodes for
2968 physical devices. More specifically, it only includes devices
2969 such as /dev/null, /dev/urandom, and /dev/zero which are API
2972 * logind has been extended to support behaviour like VT
2973 switching on seats that do not support a VT. This makes
2974 multi-session available on seats that are not the first seat
2975 (seat0), and on systems where kernel support for VTs has
2976 been disabled at compile-time.
2978 * If a process holds a delay lock for system sleep or shutdown
2979 and fails to release it in time, we will now log its
2980 identity. This makes it easier to identify processes that
2981 cause slow suspends or power-offs.
2983 * When parsing /etc/crypttab, support for a new key-slot=
2984 option as supported by Debian is added. It allows indicating
2985 which LUKS slot to use on disk, speeding up key loading.
2987 * The sd_journald_sendv() API call has been checked and
2988 officially declared to be async-signal-safe so that it may
2989 be invoked from signal handlers for logging purposes.
2991 * Boot-time status output is now enabled automatically after a
2992 short timeout if boot does not progress, in order to give
2993 the user an indication what she or he is waiting for.
2995 * The boot-time output has been improved to show how much time
2996 remains until jobs expire.
2998 * The KillMode= switch in service units gained a new possible
2999 value "mixed". If set, and the unit is shut down, then the
3000 initial SIGTERM signal is sent only to the main daemon
3001 process, while the following SIGKILL signal is sent to
3002 all remaining processes of the service.
3004 * When a scope unit is registered, a new property "Controller"
3005 may be set. If set to a valid bus name, systemd will send a
3006 RequestStop() signal to this name when it would like to shut
3007 down the scope. This may be used to hook manager logic into
3008 the shutdown logic of scope units. Also, scope units may now
3009 be put in a special "abandoned" state, in which case the
3010 manager process which created them takes no further
3011 responsibilities for it.
3013 * When reading unit files, systemd will now verify
3014 the access mode of these files, and warn about certain
3015 suspicious combinations. This has been added to make it
3016 easier to track down packaging bugs where unit files are
3017 marked executable or world-writable.
3019 * systemd-nspawn gained a new "--setenv=" switch to set
3020 container-wide environment variables. The similar option in
3021 systemd-activate was renamed from "--environment=" to
3022 "--setenv=" for consistency.
3024 * systemd-nspawn has been updated to create a new kdbus domain
3025 for each container that is invoked, thus allowing each
3026 container to have its own set of system and user buses,
3027 independent of the host.
3029 * systemd-nspawn gained a new --drop-capability= switch to run
3030 the container with less capabilities than the default. Both
3031 --drop-capability= and --capability= now take the special
3032 string "all" for dropping or keeping all capabilities.
3034 * systemd-nspawn gained new switches for executing containers
3035 with specific SELinux labels set.
3037 * systemd-nspawn gained a new --quiet switch to not generate
3038 any additional output but the container's own console
3041 * systemd-nspawn gained a new --share-system switch to run a
3042 container without PID namespacing enabled.
3044 * systemd-nspawn gained a new --register= switch to control
3045 whether the container is registered with systemd-machined or
3046 not. This is useful for containers that do not run full
3047 OS images, but only specific apps.
3049 * systemd-nspawn gained a new --keep-unit which may be used
3050 when invoked as the only program from a service unit, and
3051 results in registration of the unit service itself in
3052 systemd-machined, instead of a newly opened scope unit.
3054 * systemd-nspawn gained a new --network-interface= switch for
3055 moving arbitrary interfaces to the container. The new
3056 --network-veth switch creates a virtual Ethernet connection
3057 between host and container. The new --network-bridge=
3058 switch then allows assigning the host side of this virtual
3059 Ethernet connection to a bridge device.
3061 * systemd-nspawn gained a new --personality= switch for
3062 setting the kernel personality for the container. This is
3063 useful when running a 32-bit container on a 64-bit host. A
3064 similar option Personality= is now also available for service
3067 * logind will now also track a "Desktop" identifier for each
3068 session which encodes the desktop environment of it. This is
3069 useful for desktop environments that want to identify
3070 multiple running sessions of itself easily.
3072 * A new SELinuxContext= setting for service units has been
3073 added that allows setting a specific SELinux execution
3074 context for a service.
3076 * Most systemd client tools will now honour $SYSTEMD_LESS for
3077 settings of the "less" pager. By default, these tools will
3078 override $LESS to allow certain operations to work, such as
3079 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
3080 influence this logic.
3082 * systemd's "seccomp" hook-up has been changed to make use of
3083 the libseccomp library instead of using its own
3084 implementation. This has benefits for portability among
3087 * For usage together with SystemCallFilter=, a new
3088 SystemCallErrorNumber= setting has been introduced that
3089 allows configuration of a system error number to be returned
3090 on filtered system calls, instead of immediately killing the
3091 process. Also, SystemCallArchitectures= has been added to
3092 limit access to system calls of a particular architecture
3093 (in order to turn off support for unused secondary
3094 architectures). There is also a global
3095 SystemCallArchitectures= setting in system.conf now to turn
3096 off support for non-native system calls system-wide.
3098 * systemd requires a kernel with a working name_to_handle_at(),
3099 please see the kernel config requirements in the README file.
3101 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
3102 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
3103 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
3104 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
3105 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
3106 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
3107 Elia Pinto, Florian Weimer, George McCollister, Goffredo
3108 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
3109 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
3110 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
3111 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
3112 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
3113 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
3114 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
3115 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
3116 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
3117 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
3118 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
3119 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
3120 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
3121 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
3122 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
3123 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
3124 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
3126 -- Berlin, 2014-02-20
3130 * logind has gained support for facilitating privileged input
3131 and drm device access for unprivileged clients. This work is
3132 useful to allow Wayland display servers (and similar
3133 programs, such as kmscon) to run under the user's ID and
3134 access input and drm devices which are normally
3135 protected. When this is used (and the kernel is new enough)
3136 logind will "mute" IO on the file descriptors passed to
3137 Wayland as long as it is in the background and "unmute" it
3138 if it returns into the foreground. This allows secure
3139 session switching without allowing background sessions to
3140 eavesdrop on input and display data. This also introduces
3141 session switching support if VT support is turned off in the
3142 kernel, and on seats that are not seat0.
3144 * A new kernel command line option luks.options= is understood
3145 now which allows specifying LUKS options for usage for LUKS
3146 encrypted partitions specified with luks.uuid=.
3148 * tmpfiles.d(5) snippets may now use specifier expansion in
3149 path names. More specifically %m, %b, %H, %v, are now
3150 replaced by the local machine id, boot id, hostname, and
3151 kernel version number.
3153 * A new tmpfiles.d(5) command "m" has been introduced which
3154 may be used to change the owner/group/access mode of a file
3155 or directory if it exists, but do nothing if it does not.
3157 * This release removes high-level support for the
3158 MemorySoftLimit= cgroup setting. The underlying kernel
3159 cgroup attribute memory.soft_limit= is currently badly
3160 designed and likely to be removed from the kernel API in its
3161 current form, hence we should not expose it for now.
3163 * The memory.use_hierarchy cgroup attribute is now enabled for
3164 all cgroups systemd creates in the memory cgroup
3165 hierarchy. This option is likely to be come the built-in
3166 default in the kernel anyway, and the non-hierarchical mode
3167 never made much sense in the intrinsically hierarchical
3170 * A new field _SYSTEMD_SLICE= is logged along with all journal
3171 messages containing the slice a message was generated
3172 from. This is useful to allow easy per-customer filtering of
3173 logs among other things.
3175 * systemd-journald will no longer adjust the group of journal
3176 files it creates to the "systemd-journal" group. Instead we
3177 rely on the journal directory to be owned by the
3178 "systemd-journal" group, and its setgid bit set, so that the
3179 kernel file system layer will automatically enforce that
3180 journal files inherit this group assignment. The reason for
3181 this change is that we cannot allow NSS look-ups from
3182 journald which would be necessary to resolve
3183 "systemd-journal" to a numeric GID, because this might
3184 create deadlocks if NSS involves synchronous queries to
3185 other daemons (such as nscd, or sssd) which in turn are
3186 logging clients of journald and might block on it, which
3187 would then dead lock. A tmpfiles.d(5) snippet included in
3188 systemd will make sure the setgid bit and group are
3189 properly set on the journal directory if it exists on every
3190 boot. However, we recommend adjusting it manually after
3191 upgrades too (or from RPM scriptlets), so that the change is
3192 not delayed until next reboot.
3194 * Backlight and random seed files in /var/lib/ have moved into
3195 the /var/lib/systemd/ directory, in order to centralize all
3196 systemd generated files in one directory.
3198 * Boot time performance measurements (as displayed by
3199 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
3200 performance information if that's available to determine how
3201 much time BIOS and boot loader initialization required. With
3202 a sufficiently new BIOS you hence no longer need to boot
3203 with Gummiboot to get access to such information.
3205 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
3206 Cristian Rodríguez, Dave Reisner, David Herrmann, David
3207 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
3208 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
3209 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
3210 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
3211 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3213 -- Berlin, 2013-10-02
3217 * The Restart= option for services now understands a new
3218 on-watchdog setting, which will restart the service
3219 automatically if the service stops sending out watchdog keep
3220 alive messages (as configured with WatchdogSec=).
3222 * The getty generator (which is responsible for bringing up a
3223 getty on configured serial consoles) will no longer only
3224 start a getty on the primary kernel console but on all
3225 others, too. This makes the order in which console= is
3226 specified on the kernel command line less important.
3228 * libsystemd-logind gained a new sd_session_get_vt() call to
3229 retrieve the VT number of a session.
3231 * If the option "tries=0" is set for an entry of /etc/crypttab
3232 its passphrase is queried indefinitely instead of any
3233 maximum number of tries.
3235 * If a service with a configure PID file terminates its PID
3236 file will now be removed automatically if it still exists
3237 afterwards. This should put an end to stale PID files.
3239 * systemd-run will now also take relative binary path names
3240 for execution and no longer insists on absolute paths.
3242 * InaccessibleDirectories= and ReadOnlyDirectories= now take
3243 paths that are optionally prefixed with "-" to indicate that
3244 it should not be considered a failure if they do not exist.
3246 * journalctl -o (and similar commands) now understands a new
3247 output mode "short-precise", it is similar to "short" but
3248 shows timestamps with usec accuracy.
3250 * The option "discard" (as known from Debian) is now
3251 synonymous to "allow-discards" in /etc/crypttab. In fact,
3252 "discard" is preferred now (since it is easier to remember
3255 * Some licensing clean-ups were made, so that more code is now
3256 LGPL-2.1 licensed than before.
3258 * A minimal tool to save/restore the display backlight
3259 brightness across reboots has been added. It will store the
3260 backlight setting as late as possible at shutdown, and
3261 restore it as early as possible during reboot.
3263 * A logic to automatically discover and enable home and swap
3264 partitions on GPT disks has been added. With this in place
3265 /etc/fstab becomes optional for many setups as systemd can
3266 discover certain partitions located on the root disk
3267 automatically. Home partitions are recognized under their
3268 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
3269 partitions are recognized under their GPT type ID
3270 0657fd6da4ab43c484e50933c84b4f4f.
3272 * systemd will no longer pass any environment from the kernel
3273 or initrd to system services. If you want to set an
3274 environment for all services, do so via the kernel command
3275 line systemd.setenv= assignment.
3277 * The systemd-sysctl tool no longer natively reads the file
3278 /etc/sysctl.conf. If desired, the file should be symlinked
3279 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
3280 legacy support by a symlink rather than built-in code, it
3281 also makes the otherwise hidden order of application of the
3282 different files visible. (Note that this partly reverts to a
3283 pre-198 application order of sysctl knobs!)
3285 * The "systemctl set-log-level" and "systemctl dump" commands
3286 have been moved to systemd-analyze.
3288 * systemd-run learned the new --remain-after-exit switch,
3289 which causes the scope unit not to be cleaned up
3290 automatically after the process terminated.
3292 * tmpfiles learned a new --exclude-prefix= switch to exclude
3293 certain paths from operation.
3295 * journald will now automatically flush all messages to disk
3296 as soon as a message at the log level CRIT, ALERT or EMERG
3299 Contributions from: Andrew Cook, Brandon Philips, Christian
3300 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
3301 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
3302 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
3303 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
3304 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
3305 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
3306 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
3307 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
3308 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
3309 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
3310 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
3311 William Giokas, Zbigniew Jędrzejewski-Szmek
3313 -- Berlin, 2013-09-13
3317 * The documentation has been updated to cover the various new
3318 concepts introduced with 205.
3320 * Unit files now understand the new %v specifier which
3321 resolves to the kernel version string as returned by "uname
3324 * systemctl now supports filtering the unit list output by
3325 load state, active state and sub state, using the new
3328 * "systemctl status" will now show the results of the
3329 condition checks (like ConditionPathExists= and similar) of
3330 the last start attempts of the unit. They are also logged to
3333 * "journalctl -b" may now be used to look for boot output of a
3334 specific boot. Try "journalctl -b -1" for the previous boot,
3335 but the syntax is substantially more powerful.
3337 * "journalctl --show-cursor" has been added which prints the
3338 cursor string the last shown log line. This may then be used
3339 with the new "journalctl --after-cursor=" switch to continue
3340 browsing logs from that point on.
3342 * "journalctl --force" may now be used to force regeneration
3345 * Creation of "dead" device nodes has been moved from udev
3346 into kmod and tmpfiles. Previously, udev would read the kmod
3347 databases to pre-generate dead device nodes based on meta
3348 information contained in kernel modules, so that these would
3349 be auto-loaded on access rather then at boot. As this
3350 does not really have much to do with the exposing actual
3351 kernel devices to userspace this has always been slightly
3352 alien in the udev codebase. Following the new scheme kmod
3353 will now generate a runtime snippet for tmpfiles from the
3354 module meta information and it now is tmpfiles' job to the
3355 create the nodes. This also allows overriding access and
3356 other parameters for the nodes using the usual tmpfiles
3357 facilities. As side effect this allows us to remove the
3358 CAP_SYS_MKNOD capability bit from udevd entirely.
3360 * logind's device ACLs may now be applied to these "dead"
3361 devices nodes too, thus finally allowing managed access to
3362 devices such as /dev/snd/sequencer whithout loading the
3363 backing module right-away.
3365 * A new RPM macro has been added that may be used to apply
3366 tmpfiles configuration during package installation.
3368 * systemd-detect-virt and ConditionVirtualization= now can
3369 detect User-Mode-Linux machines (UML).
3371 * journald will now implicitly log the effective capabilities
3372 set of processes in the message metadata.
3374 * systemd-cryptsetup has gained support for TrueCrypt volumes.
3376 * The initrd interface has been simplified (more specifically,
3377 support for passing performance data via environment
3378 variables and fsck results via files in /run has been
3379 removed). These features were non-essential, and are
3380 nowadays available in a much nicer way by having systemd in
3381 the initrd serialize its state and have the hosts systemd
3382 deserialize it again.
3384 * The udev "keymap" data files and tools to apply keyboard
3385 specific mappings of scan to key codes, and force-release
3386 scan code lists have been entirely replaced by a udev
3387 "keyboard" builtin and a hwdb data file.
3389 * systemd will now honour the kernel's "quiet" command line
3390 argument also during late shutdown, resulting in a
3391 completely silent shutdown when used.
3393 * There's now an option to control the SO_REUSEPORT socket
3394 option in .socket units.
3396 * Instance units will now automatically get a per-template
3397 subslice of system.slice unless something else is explicitly
3398 configured. For example, instances of sshd@.service will now
3399 implicitly be placed in system-sshd.slice rather than
3400 system.slice as before.
3402 * Test coverage support may now be enabled at build time.
3404 Contributions from: Dave Reisner, Frederic Crozat, Harald
3405 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
3406 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
3407 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
3408 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
3409 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
3410 Giokas, Zbigniew Jędrzejewski-Szmek
3412 -- Berlin, 2013-07-23
3416 * Two new unit types have been introduced:
3418 Scope units are very similar to service units, however, are
3419 created out of pre-existing processes -- instead of PID 1
3420 forking off the processes. By using scope units it is
3421 possible for system services and applications to group their
3422 own child processes (worker processes) in a powerful way
3423 which then maybe used to organize them, or kill them
3424 together, or apply resource limits on them.
3426 Slice units may be used to partition system resources in an
3427 hierarchical fashion and then assign other units to them. By
3428 default there are now three slices: system.slice (for all
3429 system services), user.slice (for all user sessions),
3430 machine.slice (for VMs and containers).
3432 Slices and scopes have been introduced primarily in
3433 context of the work to move cgroup handling to a
3434 single-writer scheme, where only PID 1
3435 creates/removes/manages cgroups.
3437 * There's a new concept of "transient" units. In contrast to
3438 normal units these units are created via an API at runtime,
3439 not from configuration from disk. More specifically this
3440 means it is now possible to run arbitrary programs as
3441 independent services, with all execution parameters passed
3442 in via bus APIs rather than read from disk. Transient units
3443 make systemd substantially more dynamic then it ever was,
3444 and useful as a general batch manager.
3446 * logind has been updated to make use of scope and slice units
3447 for managing user sessions. As a user logs in he will get
3448 his own private slice unit, to which all sessions are added
3449 as scope units. We also added support for automatically
3450 adding an instance of user@.service for the user into the
3451 slice. Effectively logind will no longer create cgroup
3452 hierarchies on its own now, it will defer entirely to PID 1
3453 for this by means of scope, service and slice units. Since
3454 user sessions this way become entities managed by PID 1
3455 the output of "systemctl" is now a lot more comprehensive.
3457 * A new mini-daemon "systemd-machined" has been added which
3458 may be used by virtualization managers to register local
3459 VMs/containers. nspawn has been updated accordingly, and
3460 libvirt will be updated shortly. machined will collect a bit
3461 of meta information about the VMs/containers, and assign
3462 them their own scope unit (see above). The collected
3463 meta-data is then made available via the "machinectl" tool,
3464 and exposed in "ps" and similar tools. machined/machinectl
3465 is compile-time optional.
3467 * As discussed earlier, the low-level cgroup configuration
3468 options ControlGroup=, ControlGroupModify=,
3469 ControlGroupPersistent=, ControlGroupAttribute= have been
3470 removed. Please use high-level attribute settings instead as
3471 well as slice units.
3473 * A new bus call SetUnitProperties() has been added to alter
3474 various runtime parameters of a unit. This is primarily
3475 useful to alter cgroup parameters dynamically in a nice way,
3476 but will be extended later on to make more properties
3477 modifiable at runtime. systemctl gained a new set-properties
3478 command that wraps this call.
3480 * A new tool "systemd-run" has been added which can be used to
3481 run arbitrary command lines as transient services or scopes,
3482 while configuring a number of settings via the command
3483 line. This tool is currently very basic, however already
3484 very useful. We plan to extend this tool to even allow
3485 queuing of execution jobs with time triggers from the
3486 command line, similar in fashion to "at".
3488 * nspawn will now inform the user explicitly that kernels with
3489 audit enabled break containers, and suggest the user to turn
3492 * Support for detecting the IMA and AppArmor security
3493 frameworks with ConditionSecurity= has been added.
3495 * journalctl gained a new "-k" switch for showing only kernel
3496 messages, mimicking dmesg output; in addition to "--user"
3497 and "--system" switches for showing only user's own logs
3500 * systemd-delta can now show information about drop-in
3501 snippets extending unit files.
3503 * libsystemd-bus has been substantially updated but is still
3504 not available as public API.
3506 * systemd will now look for the "debug" argument on the kernel
3507 command line and enable debug logging, similar to what
3508 "systemd.log_level=debug" already did before.
3510 * "systemctl set-default", "systemctl get-default" has been
3511 added to configure the default.target symlink, which
3512 controls what to boot into by default.
3514 * "systemctl set-log-level" has been added as a convenient
3515 way to raise and lower systemd logging threshold.
3517 * "systemd-analyze plot" will now show the time the various
3518 generators needed for execution, as well as information
3519 about the unit file loading.
3521 * libsystemd-journal gained a new sd_journal_open_files() call
3522 for opening specific journal files. journactl also gained a
3523 new switch to expose this new functionality. Previously we
3524 only supported opening all files from a directory, or all
3525 files from the system, as opening individual files only is
3526 racy due to journal file rotation.
3528 * systemd gained the new DefaultEnvironment= setting in
3529 /etc/systemd/system.conf to set environment variables for
3532 * If a privileged process logs a journal message with the
3533 OBJECT_PID= field set, then journald will automatically
3534 augment this with additional OBJECT_UID=, OBJECT_GID=,
3535 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
3536 system services want to log events about specific client
3537 processes. journactl/systemctl has been updated to make use
3538 of this information if all log messages regarding a specific
3541 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
3542 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
3543 Reisner, David Coppa, David King, David Strauss, Eelco
3544 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
3545 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
3546 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
3547 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
3548 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
3549 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
3550 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
3551 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
3552 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
3553 Łukasz Stelmach, 장동준
3557 * The Python bindings gained some minimal support for the APIs
3558 exposed by libsystemd-logind.
3560 * ConditionSecurity= gained support for detecting SMACK. Since
3561 this condition already supports SELinux and AppArmor we only
3562 miss IMA for this. Patches welcome!
3564 Contributions from: Karol Lewandowski, Lennart Poettering,
3565 Zbigniew Jędrzejewski-Szmek
3569 * systemd-nspawn will now create /etc/resolv.conf if
3570 necessary, before bind-mounting the host's file onto it.
3572 * systemd-nspawn will now store meta information about a
3573 container on the container's cgroup as extended attribute
3574 fields, including the root directory.
3576 * The cgroup hierarchy has been reworked in many ways. All
3577 objects any of the components systemd creates in the cgroup
3578 tree are now suffixed. More specifically, user sessions are
3579 now placed in cgroups suffixed with ".session", users in
3580 cgroups suffixed with ".user", and nspawn containers in
3581 cgroups suffixed with ".nspawn". Furthermore, all cgroup
3582 names are now escaped in a simple scheme to avoid collision
3583 of userspace object names with kernel filenames. This work
3584 is preparation for making these objects relocatable in the
3585 cgroup tree, in order to allow easy resource partitioning of
3586 these objects without causing naming conflicts.
3588 * systemctl list-dependencies gained the new switches
3589 --plain, --reverse, --after and --before.
3591 * systemd-inhibit now shows the process name of processes that
3592 have taken an inhibitor lock.
3594 * nss-myhostname will now also resolve "localhost"
3595 implicitly. This makes /etc/hosts an optional file and
3596 nicely handles that on IPv6 ::1 maps to both "localhost" and
3599 * libsystemd-logind.so gained a new call
3600 sd_get_machine_names() to enumerate running containers and
3601 VMs (currently only supported by very new libvirt and
3602 nspawn). sd_login_monitor can now be used to watch
3603 VMs/containers coming and going.
3605 * .include is not allowed recursively anymore, and only in
3606 unit files. Usually it is better to use drop-in snippets in
3607 .d/*.conf anyway, as introduced with systemd 198.
3609 * systemd-analyze gained a new "critical-chain" command that
3610 determines the slowest chain of units run during system
3611 boot-up. It is very useful for tracking down where
3612 optimizing boot time is the most beneficial.
3614 * systemd will no longer allow manipulating service paths in
3615 the name=systemd:/system cgroup tree using ControlGroup= in
3616 units. (But is still fine with it in all other dirs.)
3618 * There's a new systemd-nspawn@.service service file that may
3619 be used to easily run nspawn containers as system
3620 services. With the container's root directory in
3621 /var/lib/container/foobar it is now sufficient to run
3622 "systemctl start systemd-nspawn@foobar.service" to boot it.
3624 * systemd-cgls gained a new parameter "--machine" to list only
3625 the processes within a certain container.
3627 * ConditionSecurity= now can check for "apparmor". We still
3628 are lacking checks for SMACK and IMA for this condition
3629 check though. Patches welcome!
3631 * A new configuration file /etc/systemd/sleep.conf has been
3632 added that may be used to configure which kernel operation
3633 systemd is supposed to execute when "suspend", "hibernate"
3634 or "hybrid-sleep" is requested. This makes the new kernel
3635 "freeze" state accessible to the user.
3637 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
3638 the passed argument if applicable.
3640 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
3641 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
3642 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
3643 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
3644 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
3645 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
3646 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
3651 * The output of 'systemctl list-jobs' got some polishing. The
3652 '--type=' argument may now be passed more than once. A new
3653 command 'systemctl list-sockets' has been added which shows
3654 a list of kernel sockets systemd is listening on with the
3655 socket units they belong to, plus the units these socket
3658 * The experimental libsystemd-bus library got substantial
3659 updates to work in conjunction with the (also experimental)
3660 kdbus kernel project. It works well enough to exchange
3661 messages with some sophistication. Note that kdbus is not
3662 ready yet, and the library is mostly an elaborate test case
3663 for now, and not installable.
3665 * systemd gained a new unit 'systemd-static-nodes.service'
3666 that generates static device nodes earlier during boot, and
3667 can run in conjunction with udev.
3669 * libsystemd-login gained a new call sd_pid_get_user_unit()
3670 to retrieve the user systemd unit a process is running
3671 in. This is useful for systems where systemd is used as
3674 * systemd-nspawn now places all containers in the new /machine
3675 top-level cgroup directory in the name=systemd
3676 hierarchy. libvirt will soon do the same, so that we get a
3677 uniform separation of /system, /user and /machine for system
3678 services, user processes and containers/virtual
3679 machines. This new cgroup hierarchy is also useful to stick
3680 stable names to specific container instances, which can be
3681 recognized later this way (this name may be controlled
3682 via systemd-nspawn's new -M switch). libsystemd-login also
3683 gained a new call sd_pid_get_machine_name() to retrieve the
3684 name of the container/VM a specific process belongs to.
3686 * bootchart can now store its data in the journal.
3688 * libsystemd-journal gained a new call
3689 sd_journal_add_conjunction() for AND expressions to the
3690 matching logic. This can be used to express more complex
3691 logical expressions.
3693 * journactl can now take multiple --unit= and --user-unit=
3696 * The cryptsetup logic now understands the "luks.key=" kernel
3697 command line switch for specifying a file to read the
3698 decryption key from. Also, if a configured key file is not
3699 found the tool will now automatically fall back to prompting
3702 * Python systemd.journal module was updated to wrap recently
3703 added functions from libsystemd-journal. The interface was
3704 changed to bring the low level interface in s.j._Reader
3705 closer to the C API, and the high level interface in
3706 s.j.Reader was updated to wrap and convert all data about
3709 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
3710 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
3711 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
3712 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
3713 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
3714 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3718 * journalctl --update-catalog now understands a new --root=
3719 option to operate on catalogs found in a different root
3722 * During shutdown after systemd has terminated all running
3723 services a final killing loop kills all remaining left-over
3724 processes. We will now print the name of these processes
3725 when we send SIGKILL to them, since this usually indicates a
3728 * If /etc/crypttab refers to password files stored on
3729 configured mount points automatic dependencies will now be
3730 generated to ensure the specific mount is established first
3731 before the key file is attempted to be read.
3733 * 'systemctl status' will now show information about the
3734 network sockets a socket unit is listening on.
3736 * 'systemctl status' will also shown information about any
3737 drop-in configuration file for units. (Drop-In configuration
3738 files in this context are files such as
3739 /etc/systemd/systemd/foobar.service.d/*.conf)
3741 * systemd-cgtop now optionally shows summed up CPU times of
3742 cgroups. Press '%' while running cgtop to switch between
3743 percentage and absolute mode. This is useful to determine
3744 which cgroups use up the most CPU time over the entire
3745 runtime of the system. systemd-cgtop has also been updated
3746 to be 'pipeable' for processing with further shell tools.
3748 * 'hostnamectl set-hostname' will now allow setting of FQDN
3751 * The formatting and parsing of time span values has been
3752 changed. The parser now understands fractional expressions
3753 such as "5.5h". The formatter will now output fractional
3754 expressions for all time spans under 1min, i.e. "5.123456s"
3755 rather than "5s 123ms 456us". For time spans under 1s
3756 millisecond values are shown, for those under 1ms
3757 microsecond values are shown. This should greatly improve
3758 all time-related output of systemd.
3760 * libsystemd-login and libsystemd-journal gained new
3761 functions for querying the poll() events mask and poll()
3762 timeout value for integration into arbitrary event
3765 * localectl gained the ability to list available X11 keymaps
3766 (models, layouts, variants, options).
3768 * 'systemd-analyze dot' gained the ability to filter for
3769 specific units via shell-style globs, to create smaller,
3770 more useful graphs. I.e. it is now possible to create simple
3771 graphs of all the dependencies between only target units, or
3772 of all units that Avahi has dependencies with.
3774 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
3775 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
3776 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
3777 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
3778 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
3779 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
3780 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
3784 * The boot-time readahead implementation for rotating media
3785 will now read the read-ahead data in multiple passes which
3786 consist of all read requests made in equidistant time
3787 intervals. This means instead of strictly reading read-ahead
3788 data in its physical order on disk we now try to find a
3789 middle ground between physical and access time order.
3791 * /etc/os-release files gained a new BUILD_ID= field for usage
3792 on operating systems that provide continuous builds of OS
3795 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
3796 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
3797 William Douglas, Zbigniew Jędrzejewski-Szmek
3801 * systemd-python gained an API exposing libsystemd-daemon.
3803 * The SMACK setup logic gained support for uploading CIPSO
3806 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
3807 ReadOnlyDirectories= and InaccessibleDirectories= has
3808 changed. The private /tmp and /var/tmp directories are now
3809 shared by all processes of a service (which means
3810 ExecStartPre= may now leave data in /tmp that ExecStart= of
3811 the same service can still access). When a service is
3812 stopped its temporary directories are immediately deleted
3813 (normal clean-up with tmpfiles is still done in addition to
3816 * By default, systemd will now set a couple of sysctl
3817 variables in the kernel: the safe sysrq options are turned
3818 on, IP route verification is turned on, and source routing
3819 disabled. The recently added hardlink and softlink
3820 protection of the kernel is turned on. These settings should
3821 be reasonably safe, and good defaults for all new systems.
3823 * The predictable network naming logic may now be turned off
3824 with a new kernel command line switch: net.ifnames=0.
3826 * A new libsystemd-bus module has been added that implements a
3827 pretty complete D-Bus client library. For details see:
3829 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
3831 * journald will now explicitly flush the journal files to disk
3832 at the latest 5min after each write. The file will then also
3833 be marked offline until the next write. This should increase
3834 reliability in case of a crash. The synchronization delay
3835 can be configured via SyncIntervalSec= in journald.conf.
3837 * There's a new remote-fs-setup.target unit that can be used
3838 to pull in specific services when at least one remote file
3839 system is to be mounted.
3841 * There are new targets timers.target and paths.target as
3842 canonical targets to pull user timer and path units in
3843 from. This complements sockets.target with a similar
3844 purpose for socket units.
3846 * libudev gained a new call udev_device_set_attribute_value()
3847 to set sysfs attributes of a device.
3849 * The udev daemon now sets the default number of worker
3850 processes executed in parallel based on the number of available
3851 CPUs instead of the amount of available RAM. This is supposed
3852 to provide a more reliable default and limit a too aggressive
3853 paralellism for setups with 1000s of devices connected.
3855 Contributions from: Auke Kok, Colin Walters, Cristian
3856 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
3857 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
3858 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
3859 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
3860 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
3861 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
3862 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
3863 Zbigniew Jędrzejewski-Szmek
3867 * Configuration of unit files may now be extended via drop-in
3868 files without having to edit/override the unit files
3869 themselves. More specifically, if the administrator wants to
3870 change one value for a service file foobar.service he can
3871 now do so by dropping in a configuration snippet into
3872 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
3873 will load all these snippets and apply them on top of the
3874 main unit configuration file, possibly extending or
3875 overriding its settings. Using these drop-in snippets is
3876 generally nicer than the two earlier options for changing
3877 unit files locally: copying the files from
3878 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
3879 them there; or creating a new file in /etc/systemd/system/
3880 that incorporates the original one via ".include". Drop-in
3881 snippets into these .d/ directories can be placed in any
3882 directory systemd looks for units in, and the usual
3883 overriding semantics between /usr/lib, /etc and /run apply
3886 * Most unit file settings which take lists of items can now be
3887 reset by assigning the empty string to them. For example,
3888 normally, settings such as Environment=FOO=BAR append a new
3889 environment variable assignment to the environment block,
3890 each time they are used. By assigning Environment= the empty
3891 string the environment block can be reset to empty. This is
3892 particularly useful with the .d/*.conf drop-in snippets
3893 mentioned above, since this adds the ability to reset list
3894 settings from vendor unit files via these drop-ins.
3896 * systemctl gained a new "list-dependencies" command for
3897 listing the dependencies of a unit recursively.
3899 * Inhibitors are now honored and listed by "systemctl
3900 suspend", "systemctl poweroff" (and similar) too, not only
3901 GNOME. These commands will also list active sessions by
3904 * Resource limits (as exposed by the various control group
3905 controllers) can now be controlled dynamically at runtime
3906 for all units. More specifically, you can now use a command
3907 like "systemctl set-cgroup-attr foobar.service cpu.shares
3908 2000" to alter the CPU shares a specific service gets. These
3909 settings are stored persistently on disk, and thus allow the
3910 administrator to easily adjust the resource usage of
3911 services with a few simple commands. This dynamic resource
3912 management logic is also available to other programs via the
3913 bus. Almost any kernel cgroup attribute and controller is
3916 * systemd-vconsole-setup will now copy all font settings to
3917 all allocated VTs, where it previously applied them only to
3920 * libsystemd-login gained the new sd_session_get_tty() API
3923 * This release drops support for a few legacy or
3924 distribution-specific LSB facility names when parsing init
3925 scripts: $x-display-manager, $mail-transfer-agent,
3926 $mail-transport-agent, $mail-transfer-agent, $smtp,
3927 $null. Also, the mail-transfer-agent.target unit backing
3928 this has been removed. Distributions which want to retain
3929 compatibility with this should carry the burden for
3930 supporting this themselves and patch support for these back
3931 in, if they really need to. Also, the facilities $syslog and
3932 $local_fs are now ignored, since systemd does not support
3933 early-boot LSB init scripts anymore, and these facilities
3934 are implied anyway for normal services. syslog.target has
3937 * There are new bus calls on PID1's Manager object for
3938 cancelling jobs, and removing snapshot units. Previously,
3939 both calls were only available on the Job and Snapshot
3942 * systemd-journal-gatewayd gained SSL support.
3944 * The various "environment" files, such as /etc/locale.conf
3945 now support continuation lines with a backslash ("\") as
3946 last character in the line, similarly in style (but different)
3947 to how this is supported in shells.
3949 * For normal user processes the _SYSTEMD_USER_UNIT= field is
3950 now implicitly appended to every log entry logged. systemctl
3951 has been updated to filter by this field when operating on a
3952 user systemd instance.
3954 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
3955 CAP_AUDIT_CONTROL capabilities to the capabilities set for
3956 the container. This makes it easier to boot unmodified
3957 Fedora systems in a container, which however still requires
3958 audit=0 to be passed on the kernel command line. Auditing in
3959 kernel and userspace is unfortunately still too broken in
3960 context of containers, hence we recommend compiling it out
3961 of the kernel or using audit=0. Hopefully this will be fixed
3962 one day for good in the kernel.
3964 * nspawn gained the new --bind= and --bind-ro= parameters to
3965 bind mount specific directories from the host into the
3968 * nspawn will now mount its own devpts file system instance
3969 into the container, in order not to leak pty devices from
3970 the host into the container.
3972 * systemd will now read the firmware boot time performance
3973 information from the EFI variables, if the used boot loader
3974 supports this, and takes it into account for boot performance
3975 analysis via "systemd-analyze". This is currently supported
3976 only in conjunction with Gummiboot, but could be supported
3977 by other boot loaders too. For details see:
3979 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
3981 * A new generator has been added that automatically mounts the
3982 EFI System Partition (ESP) to /boot, if that directory
3983 exists, is empty, and no other file system has been
3984 configured to be mounted there.
3986 * logind will now send out PrepareForSleep(false) out
3987 unconditionally, after coming back from suspend. This may be
3988 used by applications as asynchronous notification for
3989 system resume events.
3991 * "systemctl unlock-sessions" has been added, that allows
3992 unlocking the screens of all user sessions at once, similar
3993 to how "systemctl lock-sessions" already locked all users
3994 sessions. This is backed by a new D-Bus call UnlockSessions().
3996 * "loginctl seat-status" will now show the master device of a
3997 seat. (i.e. the device of a seat that needs to be around for
3998 the seat to be considered available, usually the graphics
4001 * tmpfiles gained a new "X" line type, that allows
4002 configuration of files and directories (with wildcards) that
4003 shall be excluded from automatic cleanup ("aging").
4005 * udev default rules set the device node permissions now only
4006 at "add" events, and do not change them any longer with a
4007 later "change" event.
4009 * The log messages for lid events and power/sleep keypresses
4010 now carry a message ID.
4012 * We now have a substantially larger unit test suite, but this
4013 continues to be work in progress.
4015 * udevadm hwdb gained a new --root= parameter to change the
4016 root directory to operate relative to.
4018 * logind will now issue a background sync() request to the kernel
4019 early at shutdown, so that dirty buffers are flushed to disk early
4020 instead of at the last moment, in order to optimize shutdown
4023 * A new bootctl tool has been added that is an interface for
4024 certain boot loader operations. This is currently a preview
4025 and is likely to be extended into a small mechanism daemon
4026 like timedated, localed, hostnamed, and can be used by
4027 graphical UIs to enumerate available boot options, and
4028 request boot into firmware operations.
4030 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
4031 the rest of the package. It also has been updated to work
4032 correctly in initrds.
4034 * Policykit previously has been runtime optional, and is now
4035 also compile time optional via a configure switch.
4037 * systemd-analyze has been reimplemented in C. Also "systemctl
4038 dot" has moved into systemd-analyze.
4040 * "systemctl status" with no further parameters will now print
4041 the status of all active or failed units.
4043 * Operations such as "systemctl start" can now be executed
4044 with a new mode "--irreversible" which may be used to queue
4045 operations that cannot accidentally be reversed by a later
4046 job queuing. This is by default used to make shutdown
4047 requests more robust.
4049 * The Python API of systemd now gained a new module for
4050 reading journal files.
4052 * A new tool kernel-install has been added that can install
4053 kernel images according to the Boot Loader Specification:
4055 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
4057 * Boot time console output has been improved to provide
4058 animated boot time output for hanging jobs.
4060 * A new tool systemd-activate has been added which can be used
4061 to test socket activation with, directly from the command
4062 line. This should make it much easier to test and debug
4063 socket activation in daemons.
4065 * journalctl gained a new "--reverse" (or -r) option to show
4066 journal output in reverse order (i.e. newest line first).
4068 * journalctl gained a new "--pager-end" (or -e) option to jump
4069 to immediately jump to the end of the journal in the
4070 pager. This is only supported in conjunction with "less".
4072 * journalctl gained a new "--user-unit=" option, that works
4073 similarly to "--unit=" but filters for user units rather than
4076 * A number of unit files to ease adoption of systemd in
4077 initrds has been added. This moves some minimal logic from
4078 the various initrd implementations into systemd proper.
4080 * The journal files are now owned by a new group
4081 "systemd-journal", which exists specifically to allow access
4082 to the journal, and nothing else. Previously, we used the
4083 "adm" group for that, which however possibly covers more
4084 than just journal/log file access. This new group is now
4085 already used by systemd-journal-gatewayd to ensure this
4086 daemon gets access to the journal files and as little else
4087 as possible. Note that "make install" will also set FS ACLs
4088 up for /var/log/journal to give "adm" and "wheel" read
4089 access to it, in addition to "systemd-journal" which owns
4090 the journal files. We recommend that packaging scripts also
4091 add read access to "adm" + "wheel" to /var/log/journal, and
4092 all existing/future journal files. To normal users and
4093 administrators little changes, however packagers need to
4094 ensure to create the "systemd-journal" system group at
4095 package installation time.
4097 * The systemd-journal-gatewayd now runs as unprivileged user
4098 systemd-journal-gateway:systemd-journal-gateway. Packaging
4099 scripts need to create these system user/group at
4102 * timedated now exposes a new boolean property CanNTP that
4103 indicates whether a local NTP service is available or not.
4105 * systemd-detect-virt will now also detect xen PVs
4107 * The pstore file system is now mounted by default, if it is
4110 * In addition to the SELinux and IMA policies we will now also
4111 load SMACK policies at early boot.
4113 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
4114 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
4115 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
4116 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
4117 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
4118 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
4119 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
4120 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
4121 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
4122 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
4123 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
4124 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
4125 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
4126 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
4130 * Timer units now support calendar time events in addition to
4131 monotonic time events. That means you can now trigger a unit
4132 based on a calendar time specification such as "Thu,Fri
4133 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
4134 or fifth day of any month of the year 2013, given that it is
4135 a thursday or friday. This brings timer event support
4136 considerably closer to cron's capabilities. For details on
4137 the supported calendar time specification language see
4140 * udev now supports a number of different naming policies for
4141 network interfaces for predictable names, and a combination
4142 of these policies is now the default. Please see this wiki
4143 document for details:
4145 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
4147 * Auke Kok's bootchart implementation has been added to the
4148 systemd tree. It is an optional component that can graph the
4149 boot in quite some detail. It is one of the best bootchart
4150 implementations around and minimal in its code and
4153 * nss-myhostname has been integrated into the systemd source
4154 tree. nss-myhostname guarantees that the local hostname
4155 always stays resolvable via NSS. It has been a weak
4156 requirement of systemd-hostnamed since a long time, and
4157 since its code is actually trivial we decided to just
4158 include it in systemd's source tree. It can be turned off
4159 with a configure switch.
4161 * The read-ahead logic is now capable of properly detecting
4162 whether a btrfs file system is on SSD or rotating media, in
4163 order to optimize the read-ahead scheme. Previously, it was
4164 only capable of detecting this on traditional file systems
4167 * In udev, additional device properties are now read from the
4168 IAB in addition to the OUI database. Also, Bluetooth company
4169 identities are attached to the devices as well.
4171 * In service files %U may be used as specifier that is
4172 replaced by the configured user name of the service.
4174 * nspawn may now be invoked without a controlling TTY. This
4175 makes it suitable for invocation as its own service. This
4176 may be used to set up a simple containerized server system
4177 using only core OS tools.
4179 * systemd and nspawn can now accept socket file descriptors
4180 when they are started for socket activation. This enables
4181 implementation of socket activated nspawn
4182 containers. i.e. think about autospawning an entire OS image
4183 when the first SSH or HTTP connection is received. We expect
4184 that similar functionality will also be added to libvirt-lxc
4187 * journalctl will now suppress ANSI color codes when
4188 presenting log data.
4190 * systemctl will no longer show control group information for
4191 a unit if a the control group is empty anyway.
4193 * logind can now automatically suspend/hibernate/shutdown the
4196 * /etc/machine-info and hostnamed now also expose the chassis
4197 type of the system. This can be used to determine whether
4198 the local system is a laptop, desktop, handset or
4199 tablet. This information may either be configured by the
4200 user/vendor or is automatically determined from ACPI and DMI
4201 information if possible.
4203 * A number of PolicyKit actions are now bound together with
4204 "imply" rules. This should simplify creating UIs because
4205 many actions will now authenticate similar ones as well.
4207 * Unit files learnt a new condition ConditionACPower= which
4208 may be used to conditionalize a unit depending on whether an
4209 AC power source is connected or not, of whether the system
4210 is running on battery power.
4212 * systemctl gained a new "is-failed" verb that may be used in
4213 shell scripts and suchlike to check whether a specific unit
4214 is in the "failed" state.
4216 * The EnvironmentFile= setting in unit files now supports file
4217 globbing, and can hence be used to easily read a number of
4218 environment files at once.
4220 * systemd will no longer detect and recognize specific
4221 distributions. All distribution-specific #ifdeffery has been
4222 removed, systemd is now fully generic and
4223 distribution-agnostic. Effectively, not too much is lost as
4224 a lot of the code is still accessible via explicit configure
4225 switches. However, support for some distribution specific
4226 legacy configuration file formats has been dropped. We
4227 recommend distributions to simply adopt the configuration
4228 files everybody else uses now and convert the old
4229 configuration from packaging scripts. Most distributions
4230 already did that. If that's not possible or desirable,
4231 distributions are welcome to forward port the specific
4232 pieces of code locally from the git history.
4234 * When logging a message about a unit systemd will now always
4235 log the unit name in the message meta data.
4237 * localectl will now also discover system locale data that is
4238 not stored in locale archives, but directly unpacked.
4240 * logind will no longer unconditionally use framebuffer
4241 devices as seat masters, i.e. as devices that are required
4242 to be existing before a seat is considered preset. Instead,
4243 it will now look for all devices that are tagged as
4244 "seat-master" in udev. By default framebuffer devices will
4245 be marked as such, but depending on local systems other
4246 devices might be marked as well. This may be used to
4247 integrate graphics cards using closed source drivers (such
4248 as NVidia ones) more nicely into logind. Note however, that
4249 we recommend using the open source NVidia drivers instead,
4250 and no udev rules for the closed-source drivers will be
4251 shipped from us upstream.
4253 Contributions from: Adam Williamson, Alessandro Crismani, Auke
4254 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
4255 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
4256 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
4257 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
4258 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
4259 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
4260 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
4261 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
4262 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
4263 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
4264 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
4269 * udev gained support for loading additional device properties
4270 from an indexed database that is keyed by vendor/product IDs
4271 and similar device identifiers. For the beginning this
4272 "hwdb" is populated with data from the well-known PCI and
4273 USB database, but also includes PNP, ACPI and OID data. In
4274 the longer run this indexed database shall grow into
4275 becoming the one central database for non-essential
4276 userspace device metadata. Previously, data from the PCI/USB
4277 database was only attached to select devices, since the
4278 lookup was a relatively expensive operation due to O(n) time
4279 complexity (with n being the number of entries in the
4280 database). Since this is now O(1), we decided to add in this
4281 data for all devices where this is available, by
4282 default. Note that the indexed database needs to be rebuilt
4283 when new data files are installed. To achieve this you need
4284 to update your packaging scripts to invoke "udevadm hwdb
4285 --update" after installation of hwdb data files. For
4286 RPM-based distributions we introduced the new
4287 %udev_hwdb_update macro for this purpose.
4289 * The Journal gained support for the "Message Catalog", an
4290 indexed database to link up additional information with
4291 journal entries. For further details please check:
4293 http://www.freedesktop.org/wiki/Software/systemd/catalog
4295 The indexed message catalog database also needs to be
4296 rebuilt after installation of message catalog files. Use
4297 "journalctl --update-catalog" for this. For RPM-based
4298 distributions we introduced the %journal_catalog_update
4299 macro for this purpose.
4301 * The Python Journal bindings gained support for the standard
4302 Python logging framework.
4304 * The Journal API gained new functions for checking whether
4305 the underlying file system of a journal file is capable of
4306 properly reporting file change notifications, or whether
4307 applications that want to reflect journal changes "live"
4308 need to recheck journal files continuously in appropriate
4311 * It is now possible to set the "age" field for tmpfiles
4312 entries to 0, indicating that files matching this entry
4313 shall always be removed when the directories are cleaned up.
4315 * coredumpctl gained a new "gdb" verb which invokes gdb
4316 right-away on the selected coredump.
4318 * There's now support for "hybrid sleep" on kernels that
4319 support this, in addition to "suspend" and "hibernate". Use
4320 "systemctl hybrid-sleep" to make use of this.
4322 * logind's HandleSuspendKey= setting (and related settings)
4323 now gained support for a new "lock" setting to simply
4324 request the screen lock on all local sessions, instead of
4325 actually executing a suspend or hibernation.
4327 * systemd will now mount the EFI variables file system by
4330 * Socket units now gained support for configuration of the
4331 SMACK security label.
4333 * timedatectl will now output the time of the last and next
4334 daylight saving change.
4336 * We dropped support for various legacy and distro-specific
4337 concepts, such as insserv, early-boot SysV services
4338 (i.e. those for non-standard runlevels such as 'b' or 'S')
4339 or ArchLinux /etc/rc.conf support. We recommend the
4340 distributions who still need support this to either continue
4341 to maintain the necessary patches downstream, or find a
4342 different solution. (Talk to us if you have questions!)
4344 * Various systemd components will now bypass PolicyKit checks
4345 for root and otherwise handle properly if PolicyKit is not
4346 found to be around. This should fix most issues for
4347 PolicyKit-less systems. Quite frankly this should have been
4348 this way since day one. It is absolutely our intention to
4349 make systemd work fine on PolicyKit-less systems, and we
4350 consider it a bug if something does not work as it should if
4351 PolicyKit is not around.
4353 * For embedded systems it is now possible to build udev and
4354 systemd without blkid and/or kmod support.
4356 * "systemctl switch-root" is now capable of switching root
4357 more than once. I.e. in addition to transitions from the
4358 initrd to the host OS it is now possible to transition to
4359 further OS images from the host. This is useful to implement
4360 offline updating tools.
4362 * Various other additions have been made to the RPM macros
4363 shipped with systemd. Use %udev_rules_update() after
4364 installing new udev rules files. %_udevhwdbdir,
4365 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
4366 %_sysctldir are now available which resolve to the right
4367 directories for packages to place various data files in.
4369 * journalctl gained the new --full switch (in addition to
4370 --all, to disable ellipsation for long messages.
4372 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
4373 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
4374 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
4375 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
4376 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
4377 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
4378 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
4379 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
4380 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
4384 * journalctl gained new --since= and --until= switches to
4385 filter by time. It also now supports nice filtering for
4386 units via --unit=/-u.
4388 * Type=oneshot services may use ExecReload= and do the
4391 * The journal daemon now supports time-based rotation and
4392 vacuuming, in addition to the usual disk-space based
4395 * The journal will now index the available field values for
4396 each field name. This enables clients to show pretty drop
4397 downs of available match values when filtering. The bash
4398 completion of journalctl has been updated
4399 accordingly. journalctl gained a new switch -F to list all
4400 values a certain field takes in the journal database.
4402 * More service events are now written as structured messages
4403 to the journal, and made recognizable via message IDs.
4405 * The timedated, localed and hostnamed mini-services which
4406 previously only provided support for changing time, locale
4407 and hostname settings from graphical DEs such as GNOME now
4408 also have a minimal (but very useful) text-based client
4409 utility each. This is probably the nicest way to changing
4410 these settings from the command line now, especially since
4411 it lists available options and is fully integrated with bash
4414 * There's now a new tool "systemd-coredumpctl" to list and
4415 extract coredumps from the journal.
4417 * We now install a README each in /var/log/ and
4418 /etc/rc.d/init.d explaining where the system logs and init
4419 scripts went. This hopefully should help folks who go to
4420 that dirs and look into the otherwise now empty void and
4421 scratch their heads.
4423 * When user-services are invoked (by systemd --user) the
4424 $MANAGERPID env var is set to the PID of systemd.
4426 * SIGRTMIN+24 when sent to a --user instance will now result
4427 in immediate termination of systemd.
4429 * gatewayd received numerous feature additions such as a
4430 "follow" mode, for live syncing and filtering.
4432 * browse.html now allows filtering and showing detailed
4433 information on specific entries. Keyboard navigation and
4434 mouse screen support has been added.
4436 * gatewayd/journalctl now supports HTML5/JSON
4437 Server-Sent-Events as output.
4439 * The SysV init script compatibility logic will now
4440 heuristically determine whether a script supports the
4441 "reload" verb, and only then make this available as
4444 * "systemctl status --follow" has been removed, use "journalctl
4447 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
4448 have been removed since they are hardly useful to be
4451 * And I'd like to take the opportunity to specifically mention
4452 Zbigniew for his great contributions. Zbigniew, you rock!
4454 Contributions from: Andrew Eikum, Christian Hesse, Colin
4455 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
4456 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
4457 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
4458 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
4459 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
4460 Jędrzejewski-Szmek, Сковорода Никита Андреевич
4464 * If /etc/vconsole.conf is non-existent or empty we will no
4465 longer load any console font or key map at boot by
4466 default. Instead the kernel defaults will be left
4467 intact. This is definitely the right thing to do, as no
4468 configuration should mean no configuration, and hard-coding
4469 font names that are different on all archs is probably a bad
4470 idea. Also, the kernel default key map and font should be
4471 good enough for most cases anyway, and mostly identical to
4472 the userspace fonts/key maps we previously overloaded them
4473 with. If distributions want to continue to default to a
4474 non-kernel font or key map they should ship a default
4475 /etc/vconsole.conf with the appropriate contents.
4477 Contributions from: Colin Walters, Daniel J Walsh, Dave
4478 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
4479 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
4483 * journalctl gained a new --cursor= switch to show entries
4484 starting from the specified location in the journal.
4486 * We now enforce a size limit on journal entry fields exported
4487 with "-o json" in journalctl. Fields larger than 4K will be
4488 assigned null. This can be turned off with --all.
4490 * An (optional) journal gateway daemon is now available as
4491 "systemd-journal-gatewayd.service". This service provides
4492 access to the journal via HTTP and JSON. This functionality
4493 will be used to implement live log synchronization in both
4494 pull and push modes, but has various other users too, such
4495 as easy log access for debugging of embedded devices. Right
4496 now it is already useful to retrieve the journal via HTTP:
4498 # systemctl start systemd-journal-gatewayd.service
4499 # wget http://localhost:19531/entries
4501 This will download the journal contents in a
4502 /var/log/messages compatible format. The same as JSON:
4504 # curl -H"Accept: application/json" http://localhost:19531/entries
4506 This service is also accessible via a web browser where a
4507 single static HTML5 app is served that uses the JSON logic
4508 to enable the user to do some basic browsing of the
4509 journal. This will be extended later on. Here's an example
4510 screenshot of this app in its current state:
4512 http://0pointer.de/public/journal-gatewayd
4514 Contributions from: Kay Sievers, Lennart Poettering, Robert
4515 Milasan, Tom Gundersen
4519 * The bash completion logic is now available for journalctl
4522 * We do not mount the "cpuset" controller anymore together with
4523 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
4524 started if no parameters are assigned to it. "cpuset" hence
4525 broke code that assumed it it could create "cpu" groups and
4528 * journalctl -f will now subscribe to terminal size changes,
4529 and line break accordingly.
4531 Contributions from: Dave Reisner, Kay Sievers, Lennart
4532 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
4536 * nspawn will now create a symlink /etc/localtime in the
4537 container environment, copying the host's timezone
4538 setting. Previously this has been done via a bind mount, but
4539 since symlinks cannot be bind mounted this has now been
4540 changed to create/update the appropriate symlink.
4542 * journalctl -n's line number argument is now optional, and
4543 will default to 10 if omitted.
4545 * journald will now log the maximum size the journal files may
4546 take up on disk. This is particularly useful if the default
4547 built-in logic of determining this parameter from the file
4548 system size is used. Use "systemctl status
4549 systemd-journald.service" to see this information.
4551 * The multi-seat X wrapper tool has been stripped down. As X
4552 is now capable of enumerating graphics devices via udev in a
4553 seat-aware way the wrapper is not strictly necessary
4554 anymore. A stripped down temporary stop-gap is still shipped
4555 until the upstream display managers have been updated to
4556 fully support the new X logic. Expect this wrapper to be
4557 removed entirely in one of the next releases.
4559 * HandleSleepKey= in logind.conf has been split up into
4560 HandleSuspendKey= and HandleHibernateKey=. The old setting
4561 is not available anymore. X11 and the kernel are
4562 distinguishing between these keys and we should too. This
4563 also means the inhibition lock for these keys has been split
4566 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
4567 Poettering, Lukas Nykryn, Václav Pavlín
4571 * Whenever a unit changes state we will now log this to the
4572 journal and show along the unit's own log output in
4575 * ConditionPathIsMountPoint= can now properly detect bind
4576 mount points too. (Previously, a bind mount of one file
4577 system to another place in the same file system could not be
4578 detected as mount, since they shared struct stat's st_dev
4581 * We will now mount the cgroup controllers cpu, cpuacct,
4582 cpuset and the controllers net_cls, net_prio together by
4585 * nspawn containers will now have a virtualized boot
4586 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
4587 over with a randomized ID at container initialization). This
4588 has the effect of making "journalctl -b" do the right thing
4591 * The JSON output journal serialization has been updated not
4592 to generate "endless" list objects anymore, but rather one
4593 JSON object per line. This is more in line how most JSON
4594 parsers expect JSON objects. The new output mode
4595 "json-pretty" has been added to provide similar output, but
4596 neatly aligned for readability by humans.
4598 * We dropped all explicit sync() invocations in the shutdown
4599 code. The kernel does this implicitly anyway in the kernel
4600 reboot() syscall. halt(8)'s -n option is now a compatibility
4603 * We now support virtualized reboot() in containers, as
4604 supported by newer kernels. We will fall back to exit() if
4605 CAP_SYS_REBOOT is not available to the container. Also,
4606 nspawn makes use of this now and will actually reboot the
4607 container if the containerized OS asks for that.
4609 * journalctl will only show local log output by default
4610 now. Use --merge (-m) to show remote log output, too.
4612 * libsystemd-journal gained the new sd_journal_get_usage()
4613 call to determine the current disk usage of all journal
4614 files. This is exposed in the new "journalctl --disk-usage"
4617 * journald gained a new configuration setting SplitMode= in
4618 journald.conf which may be used to control how user journals
4619 are split off. See journald.conf(5) for details.
4621 * A new condition type ConditionFileNotEmpty= has been added.
4623 * tmpfiles' "w" lines now support file globbing, to write
4624 multiple files at once.
4626 * We added Python bindings for the journal submission
4627 APIs. More Python APIs for a number of selected APIs will
4628 likely follow. Note that we intend to add native bindings
4629 only for the Python language, as we consider it common
4630 enough to deserve bindings shipped within systemd. There are
4631 various projects outside of systemd that provide bindings
4632 for languages such as PHP or Lua.
4634 * Many conditions will now resolve specifiers such as %i. In
4635 addition, PathChanged= and related directives of .path units
4636 now support specifiers as well.
4638 * There's now a new RPM macro definition for the system preset
4641 * journald will now warn if it ca not forward a message to the
4642 syslog daemon because its socket is full.
4644 * timedated will no longer write or process /etc/timezone,
4645 except on Debian. As we do not support late mounted /usr
4646 anymore /etc/localtime always being a symlink is now safe,
4647 and hence the information in /etc/timezone is not necessary
4650 * logind will now always reserve one VT for a text getty (VT6
4651 by default). Previously if more than 6 X sessions where
4652 started they took up all the VTs with auto-spawned gettys,
4653 so that no text gettys were available anymore.
4655 * udev will now automatically inform the btrfs kernel logic
4656 about btrfs RAID components showing up. This should make
4657 simple hotplug based btrfs RAID assembly work.
4659 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
4660 (but not for its children which will stay at the kernel
4661 default). This should allow setups with a lot more listening
4664 * systemd will now always pass the configured timezone to the
4665 kernel at boot. timedated will do the same when the timezone
4668 * logind's inhibition logic has been updated. By default,
4669 logind will now handle the lid switch, the power and sleep
4670 keys all the time, even in graphical sessions. If DEs want
4671 to handle these events on their own they should take the new
4672 handle-power-key, handle-sleep-key and handle-lid-switch
4673 inhibitors during their runtime. A simple way to achieve
4674 that is to invoke the DE wrapped in an invocation of:
4676 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
4678 * Access to unit operations is now checked via SELinux taking
4679 the unit file label and client process label into account.
4681 * systemd will now notify the administrator in the journal
4682 when he over-mounts a non-empty directory.
4684 * There are new specifiers that are resolved in unit files,
4685 for the host name (%H), the machine ID (%m) and the boot ID
4688 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
4689 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
4690 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
4691 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
4692 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
4693 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
4694 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
4698 * Support for reading structured kernel messages from
4699 /dev/kmsg has now been added and is enabled by default.
4701 * Support for reading kernel messages from /proc/kmsg has now
4702 been removed. If you want kernel messages in the journal
4703 make sure to run a recent kernel (>= 3.5) that supports
4704 reading structured messages from /dev/kmsg (see
4705 above). /proc/kmsg is now exclusive property of classic
4706 syslog daemons again.
4708 * The libudev API gained the new
4709 udev_device_new_from_device_id() call.
4711 * The logic for file system namespace (ReadOnlyDirectory=,
4712 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
4713 require pivot_root() anymore. This means fewer temporary
4714 directories are created below /tmp for this feature.
4716 * nspawn containers will now see and receive all submounts
4717 made on the host OS below the root file system of the
4720 * Forward Secure Sealing is now supported for Journal files,
4721 which provide cryptographical sealing of journal files so
4722 that attackers cannot alter log history anymore without this
4723 being detectable. Lennart will soon post a blog story about
4724 this explaining it in more detail.
4726 * There are two new service settings RestartPreventExitStatus=
4727 and SuccessExitStatus= which allow configuration of exit
4728 status (exit code or signal) which will be excepted from the
4729 restart logic, resp. consider successful.
4731 * journalctl gained the new --verify switch that can be used
4732 to check the integrity of the structure of journal files and
4733 (if Forward Secure Sealing is enabled) the contents of
4736 * nspawn containers will now be run with /dev/stdin, /dev/fd/
4737 and similar symlinks pre-created. This makes running shells
4738 as container init process a lot more fun.
4740 * The fstab support can now handle PARTUUID= and PARTLABEL=
4743 * A new ConditionHost= condition has been added to match
4744 against the hostname (with globs) and machine ID. This is
4745 useful for clusters where a single OS image is used to
4746 provision a large number of hosts which shall run slightly
4747 different sets of services.
4749 * Services which hit the restart limit will now be placed in a
4752 Contributions from: Bertram Poettering, Dave Reisner, Huang
4753 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
4754 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
4758 * When running in --user mode systemd will now become a
4759 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
4760 tree a lot more organized.
4762 * A new PartOf= unit dependency type has been introduced that
4763 may be used to group services in a natural way.
4765 * "systemctl enable" may now be used to enable instances of
4768 * journalctl now prints error log levels in red, and
4769 warning/notice log levels in bright white. It also supports
4770 filtering by log level now.
4772 * cgtop gained a new -n switch (similar to top), to configure
4773 the maximum number of iterations to run for. It also gained
4774 -b, to run in batch mode (accepting no input).
4776 * The suffix ".service" may now be omitted on most systemctl
4777 command lines involving service unit names.
4779 * There's a new bus call in logind to lock all sessions, as
4780 well as a loginctl verb for it "lock-sessions".
4782 * libsystemd-logind.so gained a new call sd_journal_perror()
4783 that works similar to libc perror() but logs to the journal
4784 and encodes structured information about the error number.
4786 * /etc/crypttab entries now understand the new keyfile-size=
4789 * shutdown(8) now can send a (configurable) wall message when
4790 a shutdown is cancelled.
4792 * The mount propagation mode for the root file system will now
4793 default to "shared", which is useful to make containers work
4794 nicely out-of-the-box so that they receive new mounts from
4795 the host. This can be undone locally by running "mount
4796 --make-rprivate /" if needed.
4798 * The prefdm.service file has been removed. Distributions
4799 should maintain this unit downstream if they intend to keep
4800 it around. However, we recommend writing normal unit files
4801 for display managers instead.
4803 * Since systemd is a crucial part of the OS we will now
4804 default to a number of compiler switches that improve
4805 security (hardening) such as read-only relocations, stack
4806 protection, and suchlike.
4808 * The TimeoutSec= setting for services is now split into
4809 TimeoutStartSec= and TimeoutStopSec= to allow configuration
4810 of individual time outs for the start and the stop phase of
4813 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
4814 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
4815 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
4816 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
4817 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
4818 Gundersen, Zbigniew Jędrzejewski-Szmek
4822 * The journal and id128 C APIs are now fully documented as man
4825 * Extra safety checks have been added when transitioning from
4826 the initial RAM disk to the main system to avoid accidental
4829 * /etc/crypttab entries now understand the new keyfile-offset=
4832 * systemctl -t can now be used to filter by unit load state.
4834 * The journal C API gained the new sd_journal_wait() call to
4835 make writing synchronous journal clients easier.
4837 * journalctl gained the new -D switch to show journals from a
4840 * journalctl now displays a special marker between log
4841 messages of two different boots.
4843 * The journal is now explicitly flushed to /var via a service
4844 systemd-journal-flush.service, rather than implicitly simply
4845 by seeing /var/log/journal to be writable.
4847 * journalctl (and the journal C APIs) can now match for much
4848 more complex expressions, with alternatives and
4851 * When transitioning from the initial RAM disk to the main
4852 system we will now kill all processes in a killing spree to
4853 ensure no processes stay around by accident.
4855 * Three new specifiers may be used in unit files: %u, %h, %s
4856 resolve to the user name, user home directory resp. user
4857 shell. This is useful for running systemd user instances.
4859 * We now automatically rotate journal files if their data
4860 object hash table gets a fill level > 75%. We also size the
4861 hash table based on the configured maximum file size. This
4862 together should lower hash collisions drastically and thus
4863 speed things up a bit.
4865 * journalctl gained the new "--header" switch to introspect
4866 header data of journal files.
4868 * A new setting SystemCallFilters= has been added to services
4869 which may be used to apply blacklists or whitelists to
4870 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
4872 * nspawn gained a new --link-journal= switch (and quicker: -j)
4873 to link the container journal with the host. This makes it
4874 very easy to centralize log viewing on the host for all
4875 guests while still keeping the journal files separated.
4877 * Many bugfixes and optimizations
4879 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
4880 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
4881 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
4886 * Several tools now understand kernel command line arguments,
4887 which are only read when run in an initial RAM disk. They
4888 usually follow closely their normal counterparts, but are
4891 * There's a new tool to analyze the readahead files that are
4892 automatically generated at boot. Use:
4894 /usr/lib/systemd/systemd-readahead analyze /.readahead
4896 * We now provide an early debug shell on tty9 if this enabled. Use:
4898 systemctl enable debug-shell.service
4900 * All plymouth related units have been moved into the Plymouth
4901 package. Please make sure to upgrade your Plymouth version
4904 * systemd-tmpfiles now supports getting passed the basename of
4905 a configuration file only, in which case it will look for it
4906 in all appropriate directories automatically.
4908 * udevadm info now takes a /dev or /sys path as argument, and
4909 does the right thing. Example:
4911 udevadm info /dev/sda
4912 udevadm info /sys/class/block/sda
4914 * systemctl now prints a warning if a unit is stopped but a
4915 unit that might trigger it continues to run. Example: a
4916 service is stopped but the socket that activates it is left
4919 * "systemctl status" will now mention if the log output was
4920 shortened due to rotation since a service has been started.
4922 * The journal API now exposes functions to determine the
4923 "cutoff" times due to rotation.
4925 * journald now understands SIGUSR1 and SIGUSR2 for triggering
4926 immediately flushing of runtime logs to /var if possible,
4927 resp. for triggering immediate rotation of the journal
4930 * It is now considered an error if a service is attempted to
4931 be stopped that is not loaded.
4933 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
4935 * systemd-analyze now supports Python 3
4937 * tmpfiles now supports cleaning up directories via aging
4938 where the first level dirs are always kept around but
4939 directories beneath it automatically aged. This is enabled
4940 by prefixing the age field with '~'.
4942 * Seat objects now expose CanGraphical, CanTTY properties
4943 which is required to deal with very fast bootups where the
4944 display manager might be running before the graphics drivers
4945 completed initialization.
4947 * Seat objects now expose a State property.
4949 * We now include RPM macros for service enabling/disabling
4950 based on the preset logic. We recommend RPM based
4951 distributions to make use of these macros if possible. This
4952 makes it simpler to reuse RPM spec files across
4955 * We now make sure that the collected systemd unit name is
4956 always valid when services log to the journal via
4959 * There's a new man page kernel-command-line(7) detailing all
4960 command line options we understand.
4962 * The fstab generator may now be disabled at boot by passing
4963 fstab=0 on the kernel command line.
4965 * A new kernel command line option modules-load= is now understood
4966 to load a specific kernel module statically, early at boot.
4968 * Unit names specified on the systemctl command line are now
4969 automatically escaped as needed. Also, if file system or
4970 device paths are specified they are automatically turned
4971 into the appropriate mount or device unit names. Example:
4973 systemctl status /home
4974 systemctl status /dev/sda
4976 * The SysVConsole= configuration option has been removed from
4977 system.conf parsing.
4979 * The SysV search path is no longer exported on the D-Bus
4982 * The Names= option is been removed from unit file parsing.
4984 * There's a new man page bootup(7) detailing the boot process.
4986 * Every unit and every generator we ship with systemd now
4987 comes with full documentation. The self-explanatory boot is
4990 * A couple of services gained "systemd-" prefixes in their
4991 name if they wrap systemd code, rather than only external
4992 code. Among them fsck@.service which is now
4993 systemd-fsck@.service.
4995 * The HaveWatchdog property has been removed from the D-Bus
4998 * systemd.confirm_spawn= on the kernel command line should now
5001 * There's a new man page crypttab(5) which details all options
5002 we actually understand.
5004 * systemd-nspawn gained a new --capability= switch to pass
5005 additional capabilities to the container.
5007 * timedated will now read known NTP implementation unit names
5008 from /usr/lib/systemd/ntp-units.d/*.list,
5009 systemd-timedated-ntp.target has been removed.
5011 * journalctl gained a new switch "-b" that lists log data of
5012 the current boot only.
5014 * The notify socket is in the abstract namespace again, in
5015 order to support daemons which chroot() at start-up.
5017 * There is a new Storage= configuration option for journald
5018 which allows configuration of where log data should go. This
5019 also provides a way to disable journal logging entirely, so
5020 that data collected is only forwarded to the console, the
5021 kernel log buffer or another syslog implementation.
5023 * Many bugfixes and optimizations
5025 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
5026 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
5027 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
5028 Shawn Landden, Tom Gundersen
5032 * "systemctl help <unit>" now shows the man page if one is
5035 * Several new man pages have been added.
5037 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
5038 MaxLevelConsole= can now be specified in
5039 journald.conf. These options allow reducing the amount of
5040 data stored on disk or forwarded by the log level.
5042 * TimerSlackNSec= can now be specified in system.conf for
5043 PID1. This allows system-wide power savings.
5045 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
5046 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
5051 * logind is now capable of (optionally) handling power and
5052 sleep keys as well as the lid switch.
5054 * journalctl now understands the syntax "journalctl
5055 /usr/bin/avahi-daemon" to get all log output of a specific
5058 * CapabilityBoundingSet= in system.conf now also influences
5059 the capability bound set of usermode helpers of the kernel.
5061 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
5062 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
5063 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
5064 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
5068 * Note that we skipped 139 releases here in order to set the
5069 new version to something that is greater than both udev's
5070 and systemd's most recent version number.
5072 * udev: all udev sources are merged into the systemd source tree now.
5073 All future udev development will happen in the systemd tree. It
5074 is still fully supported to use the udev daemon and tools without
5075 systemd running, like in initramfs or other init systems. Building
5076 udev though, will require the *build* of the systemd tree, but
5077 udev can be properly *run* without systemd.
5079 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
5080 should be used to create dead device nodes as workarounds for broken
5083 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
5084 no longer supported. udev_monitor_new_from_netlink() needs to be
5085 used to subscribe to events.
5087 * udev: when udevd is started by systemd, processes which are left
5088 behind by forking them off of udev rules, are unconditionally cleaned
5089 up and killed now after the event handling has finished. Services or
5090 daemons must be started as systemd services. Services can be
5091 pulled-in by udev to get started, but they can no longer be directly
5092 forked by udev rules.
5094 * udev: the daemon binary is called systemd-udevd now and installed
5095 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
5096 to adapt to that, create symlink, or rename the binary after building
5099 * libudev no longer provides these symbols:
5100 udev_monitor_from_socket()
5101 udev_queue_get_failed_list_entry()
5102 udev_get_{dev,sys,run}_path()
5103 The versions number was bumped and symbol versioning introduced.
5105 * systemd-loginctl and systemd-journalctl have been renamed
5106 to loginctl and journalctl to match systemctl.
5108 * The config files: /etc/systemd/systemd-logind.conf and
5109 /etc/systemd/systemd-journald.conf have been renamed to
5110 logind.conf and journald.conf. Package updates should rename
5111 the files to the new names on upgrade.
5113 * For almost all files the license is now LGPL2.1+, changed
5114 from the previous GPL2.0+. Exceptions are some minor stuff
5115 of udev (which will be changed to LGPL2.1 eventually, too),
5116 and the MIT licensed sd-daemon.[ch] library that is suitable
5117 to be used as drop-in files.
5119 * systemd and logind now handle system sleep states, in
5120 particular suspending and hibernating.
5122 * logind now implements a sleep/shutdown/idle inhibiting logic
5123 suitable for a variety of uses. Soonishly Lennart will blog
5124 about this in more detail.
5126 * var-run.mount and var-lock.mount are no longer provided
5127 (which prevously bind mounted these directories to their new
5128 places). Distributions which have not converted these
5129 directories to symlinks should consider stealing these files
5130 from git history and add them downstream.
5132 * We introduced the Documentation= field for units and added
5133 this to all our shipped units. This is useful to make it
5134 easier to explore the boot and the purpose of the various
5137 * All smaller setup units (such as
5138 systemd-vconsole-setup.service) now detect properly if they
5139 are run in a container and are skipped when
5140 appropriate. This guarantees an entirely noise-free boot in
5141 Linux container environments such as systemd-nspawn.
5143 * A framework for implementing offline system updates is now
5144 integrated, for details see:
5145 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
5147 * A new service type Type=idle is available now which helps us
5148 avoiding ugly interleaving of getty output and boot status
5151 * There's now a system-wide CapabilityBoundingSet= option to
5152 globally reduce the set of capabilities for the
5153 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
5154 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
5155 even CAP_NET_ADMIN system-wide for secure systems.
5157 * There are now system-wide DefaultLimitXXX= options to
5158 globally change the defaults of the various resource limits
5159 for all units started by PID 1.
5161 * Harald Hoyer's systemd test suite has been integrated into
5162 systemd which allows easy testing of systemd builds in qemu
5163 and nspawn. (This is really awesome! Ask us for details!)
5165 * The fstab parser is now implemented as generator, not inside
5168 * systemctl will now warn you if .mount units generated from
5169 /etc/fstab are out of date due to changes in fstab that
5170 have not been read by systemd yet.
5172 * systemd is now suitable for usage in initrds. Dracut has
5173 already been updated to make use of this. With this in place
5174 initrds get a slight bit faster but primarily are much
5175 easier to introspect and debug since "systemctl status" in
5176 the host system can be used to introspect initrd services,
5177 and the journal from the initrd is kept around too.
5179 * systemd-delta has been added, a tool to explore differences
5180 between user/admin configuration and vendor defaults.
5182 * PrivateTmp= now affects both /tmp and /var/tmp.
5184 * Boot time status messages are now much prettier and feature
5185 proper english language. Booting up systemd has never been
5188 * Read-ahead pack files now include the inode number of all
5189 files to pre-cache. When the inode changes the pre-caching
5190 is not attempted. This should be nicer to deal with updated
5191 packages which might result in changes of read-ahead
5194 * We now temporaritly lower the kernel's read_ahead_kb variable
5195 when collecting read-ahead data to ensure the kernel's
5196 built-in read-ahead does not add noise to our measurements
5197 of necessary blocks to pre-cache.
5199 * There's now RequiresMountsFor= to add automatic dependencies
5200 for all mounts necessary for a specific file system path.
5202 * MountAuto= and SwapAuto= have been removed from
5203 system.conf. Mounting file systems at boot has to take place
5206 * nspawn now learned a new switch --uuid= to set the machine
5207 ID on the command line.
5209 * nspawn now learned the -b switch to automatically search
5212 * vt102 is now the default TERM for serial TTYs, upgraded from
5215 * systemd-logind now works on VT-less systems.
5217 * The build tree has been reorganized. The individual
5218 components now have directories of their own.
5220 * A new condition type ConditionPathIsReadWrite= is now available.
5222 * nspawn learned the new -C switch to create cgroups for the
5223 container in other hierarchies.
5225 * We now have support for hardware watchdogs, configurable in
5228 * The scheduled shutdown logic now has a public API.
5230 * We now mount /tmp as tmpfs by default, but this can be
5231 masked and /etc/fstab can override it.
5233 * Since udisks does not make use of /media anymore we are not
5234 mounting a tmpfs on it anymore.
5236 * journalctl gained a new --local switch to only interleave
5237 locally generated journal files.
5239 * We can now load the IMA policy at boot automatically.
5241 * The GTK tools have been split off into a systemd-ui.
5243 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
5244 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
5245 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
5246 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
5247 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
5248 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
5249 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
5250 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
5251 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
5256 * This is mostly a bugfix release
5258 * Support optional initialization of the machine ID from the
5259 KVM or container configured UUID.
5261 * Support immediate reboots with "systemctl reboot -ff"
5263 * Show /etc/os-release data in systemd-analyze output
5265 * Many bugfixes for the journal, including endianness fixes and
5266 ensuring that disk space enforcement works
5268 * sd-login.h is C++ comptaible again
5270 * Extend the /etc/os-release format on request of the Debian
5273 * We now refuse non-UTF8 strings used in various configuration
5274 and unit files. This is done to ensure we do not pass invalid
5275 data over D-Bus or expose it elsewhere.
5277 * Register Mimo USB Screens as suitable for automatic seat
5280 * Read SELinux client context from journal clients in a race
5283 * Reorder configuration file lookup order. /etc now always
5284 overrides /run in order to allow the administrator to always
5285 and unconditionally override vendor supplied or
5286 automatically generated data.
5288 * The various user visible bits of the journal now have man
5289 pages. We still lack man pages for the journal API calls
5292 * We now ship all man pages in HTML format again in the
5295 Contributions from: Dave Reisner, Dirk Eibach, Frederic
5296 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
5297 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
5302 * This is mostly a bugfix release
5304 * systems lacking /etc/os-release are no longer supported.
5306 * Various functionality updates to libsystemd-login.so
5308 * Track class of PAM logins to distinguish greeters from
5311 Contributions from: Kay Sievers, Lennart Poettering, Michael
5316 * This is an important bugfix release for v41.
5318 * Building man pages is now optional which should be useful
5319 for those building systemd from git but unwilling to install
5322 * Watchdog support for supervising services is now usable. In
5323 a future release support for hardware watchdogs
5324 (i.e. /dev/watchdog) will be added building on this.
5326 * Service start rate limiting is now configurable and can be
5327 turned off per service. When a start rate limit is hit a
5328 reboot can automatically be triggered.
5330 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
5332 Contributions from: Benjamin Franzke, Bill Nottingham,
5333 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
5334 Schmidt, Michał Górny, Piotr Drąg
5338 * The systemd binary is installed /usr/lib/systemd/systemd now;
5339 An existing /sbin/init symlink needs to be adapted with the
5342 * The code that loads kernel modules has been ported to invoke
5343 libkmod directly, instead of modprobe. This means we do not
5344 support systems with module-init-tools anymore.
5346 * Watchdog support is now already useful, but still not
5349 * A new kernel command line option systemd.setenv= is
5350 understood to set system wide environment variables
5351 dynamically at boot.
5353 * We now limit the set of capabilities of systemd-journald.
5355 * We now set SIGPIPE to ignore by default, since it only is
5356 useful in shell pipelines, and has little use in general
5357 code. This can be disabled with IgnoreSIPIPE=no in unit
5360 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
5361 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
5366 * This is mostly a bugfix release
5368 * We now expose the reason why a service failed in the
5369 "Result" D-Bus property.
5371 * Rudimentary service watchdog support (will be completed over
5372 the next few releases.)
5374 * When systemd forks off in order execute some service we will
5375 now immediately changes its argv[0] to reflect which process
5376 it will execute. This is useful to minimize the time window
5377 with a generic argv[0], which makes bootcharts more useful
5379 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
5380 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
5381 Mike Kazantsev, Ray Strode
5385 * This is mostly a test release, but incorporates many
5388 * New systemd-cgtop tool to show control groups by their
5391 * Linking against libacl for ACLs is optional again. If
5392 disabled, support tracking device access for active logins
5393 goes becomes unavailable, and so does access to the user
5394 journals by the respective users.
5396 * If a group "adm" exists, journal files are automatically
5397 owned by them, thus allow members of this group full access
5398 to the system journal as well as all user journals.
5400 * The journal now stores the SELinux context of the logging
5401 client for all entries.
5403 * Add C++ inclusion guards to all public headers
5405 * New output mode "cat" in the journal to print only text
5406 messages, without any meta data like date or time.
5408 * Include tiny X server wrapper as a temporary stop-gap to
5409 teach XOrg udev display enumeration. This is used by display
5410 managers such as gdm, and will go away as soon as XOrg
5411 learned native udev hotplugging for display devices.
5413 * Add new systemd-cat tool for executing arbitrary programs
5414 with STDERR/STDOUT connected to the journal. Can also act as
5415 BSD logger replacement, and does so by default.
5417 * Optionally store all locally generated coredumps in the
5418 journal along with meta data.
5420 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
5421 writing short strings to files (for usage for /sys), and for
5422 creating symlinks, character and block device nodes.
5424 * New unit file option ControlGroupPersistent= to make cgroups
5425 persistent, following the mechanisms outlined in
5426 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
5428 * Support multiple local RTCs in a sane way
5430 * No longer monopolize IO when replaying readahead data on
5431 rotating disks, since we might starve non-file-system IO to
5432 death, since fanotify() will not see accesses done by blkid,
5435 * Do not show kernel threads in systemd-cgls anymore, unless
5436 requested with new -k switch.
5438 Contributions from: Dan Horák, Kay Sievers, Lennart
5439 Poettering, Michal Schmidt
5443 * This is mostly a test release, but incorporates many
5446 * The git repository moved to:
5447 git://anongit.freedesktop.org/systemd/systemd
5448 ssh://git.freedesktop.org/git/systemd/systemd
5450 * First release with the journal
5451 http://0pointer.de/blog/projects/the-journal.html
5453 * The journal replaces both systemd-kmsg-syslogd and
5454 systemd-stdout-bridge.
5456 * New sd_pid_get_unit() API call in libsystemd-logind
5458 * Many systemadm clean-ups
5460 * Introduce remote-fs-pre.target which is ordered before all
5461 remote mounts and may be used to start services before all
5464 * Added Mageia support
5466 * Add bash completion for systemd-loginctl
5468 * Actively monitor PID file creation for daemons which exit in
5469 the parent process before having finished writing the PID
5470 file in the daemon process. Daemons which do this need to be
5471 fixed (i.e. PID file creation must have finished before the
5472 parent exits), but we now react a bit more gracefully to them.
5474 * Add colourful boot output, mimicking the well-known output
5475 of existing distributions.
5477 * New option PassCredentials= for socket units, for
5478 compatibility with a recent kernel ABI breakage.
5480 * /etc/rc.local is now hooked in via a generator binary, and
5481 thus will no longer act as synchronization point during
5484 * systemctl list-unit-files now supports --root=.
5486 * systemd-tmpfiles now understands two new commands: z, Z for
5487 relabelling files according to the SELinux database. This is
5488 useful to apply SELinux labels to specific files in /sys,
5491 * Output of SysV services is now forwarded to both the console
5492 and the journal by default, not only just the console.
5494 * New man pages for all APIs from libsystemd-login.
5496 * The build tree got reorganized and a the build system is a
5497 lot more modular allowing embedded setups to specifically
5498 select the components of systemd they are interested in.
5500 * Support for Linux systems lacking the kernel VT subsystem is
5503 * configure's --with-rootdir= got renamed to
5504 --with-rootprefix= to follow the naming used by udev and
5507 * Unless specified otherwise we will now install to /usr instead
5508 of /usr/local by default.
5510 * Processes with '@' in argv[0][0] are now excluded from the
5511 final shut-down killing spree, following the logic explained
5513 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
5515 * All processes remaining in a service cgroup when we enter
5516 the START or START_PRE states are now killed with
5517 SIGKILL. That means it is no longer possible to spawn
5518 background processes from ExecStart= lines (which was never
5519 supported anyway, and bad style).
5521 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
5522 reloading of units together.
5524 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
5525 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
5526 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
5527 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
5528 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek