1 systemd System and Service Manager
5 * As an experimental feature, udev now tries to lock the
6 disk device node (flock(LOCK_SH|LOCK_NB)) while it
7 executes events for the disk or any of its partitions.
8 Applications like partitioning programs can lock the
9 disk device node (flock(LOCK_EX)) and claim temporary
10 device ownership that way; udev will entirely skip all event
11 handling for this disk and its partitions. If the disk
12 was opened for writing, the close will trigger a partition
13 table rescan in udev's "watch" facility, and if needed
14 synthesize "change" events for the disk and all its partions.
15 This is now unconditionally enabled, if it turns out to
16 cause major problems, we might turn it on only for specific
17 devices, or might need to disable it entirely. Device-mapper
18 devices are excluded from this logic.
22 * A new "systemd-timesyncd" daemon has been added for
23 synchronizing the system clock across the network. It
24 implements an SNTP client. In contrast to NTP
25 implementations such as chrony or the NTP reference server
26 this only implements a client side, and does not bother with
27 the full NTP complexity, focusing only on querying time from
28 one remote server and synchronizing the local clock to
29 it. Unless you intend to serve NTP to networked clients or
30 want to connect to local hardware clocks this simple NTP
31 client should be more than appropriate for most
32 installations. The daemon runs with minimal privileges, and
33 has been hooked up with networkd to only operate when
34 network connectivity is available. The daemon saves the
35 current clock to disk every time a new NTP sync has been
36 acquired, and uses this to possibly correct the system clock
37 early at bootup, in order to accommodate for systems that
38 lack an RTC such as the Raspberry Pi and embedded devices,
39 and make sure that time monotonically progresses on these
40 systems, even if it is not always correct. To make use of
41 this daemon a new system user and group "systemd-timesync"
42 needs to be created on installation of systemd.
44 * The queue "seqnum" interface of libudev has been disabled, as
45 it was generally incompatible with device namespacing as
46 sequence numbers of devices go "missing" if the devices are
47 part of a different namespace.
49 * "systemctl list-timers" and "systemctl list-sockets" gained
50 a --recursive switch for showing units of these types also
51 for all local containers, similar in style to the already
52 supported --recursive switch for "systemctl list-units".
54 * A new RebootArgument= setting has been added for service
55 units, which may be used to specify a kernel reboot argument
56 to use when triggering reboots with StartLimitAction=.
58 * A new FailureAction= setting has been added for service
59 units which may be used to specify an operation to trigger
60 when a service fails. This works similarly to
61 StartLimitAction=, but unlike it controls what is done
62 immediately rather than only after several attempts to
63 restart the service in question.
65 * hostnamed got updated to also expose the kernel name,
66 release, and version on the bus. This is useful for
67 executing commands like hostnamectl with the -H switch.
68 systemd-analyze makes use of this to properly display
69 details when running non-locally.
71 * The bootchart tool can now show cgroup information in the
74 * The CFS CPU quota cgroup attribute is now exposed for
75 services. The new CPUQuota= switch has been added for this
76 which takes a percentage value. Setting this will have the
77 result that a service may never get more CPU time than the
78 specified percentage, even if the machine is otherwise idle.
80 * systemd-networkd learned IPIP and SIT tunnel support.
82 * LSB init scripts exposing a dependency on $network will now
83 get a dependency on network-online.target rather than simply
84 network.target. This should bring LSB handling closer to
85 what it was on SysV systems.
87 * A new fsck.repair= kernel option has been added to control
88 how fsck shall deal with unclean file systems at boot.
90 * The (.ini) configuration file parser will now silently
91 ignore sections whose name begins with "X-". This may be
92 used to maintain application-specific extension sections in unit
95 * machined gained a new API to query the IP addresses of
96 registered containers. "machinectl status" has been updated
97 to show these addresses in its output.
99 * A new call sd_uid_get_display() has been added to the
100 sd-login APIs for querying the "primary" session of a
101 user. The "primary" session of the user is elected from the
102 user's sessions and generally a graphical session is
103 preferred over a text one.
105 * A minimal systemd-resolved daemon has been added. It
106 currently simply acts as a companion to systemd-networkd and
107 manages resolv.conf based on per-interface DNS
108 configuration, possibly supplied via DHCP. In the long run
109 we hope to extend this into a local DNSSEC enabled DNS and
112 * The systemd-networkd-wait-online tool is now enabled by
113 default. It will delay network-online.target until a network
114 connection has been configured. The tool primarily integrates
115 with networkd, but will also make a best effort to make sense
116 of network configuration performed in some other way.
118 * Two new service options StartupCPUShares= and
119 StartupBlockIOWeight= have been added that work similarly to
120 CPUShares= and BlockIOWeight= however only apply during
121 system startup. This is useful to prioritize certain services
122 differently during bootup than during normal runtime.
124 * hostnamed has been changed to prefer the statically
125 configured hostname in /etc/hostname (unless set to
126 'localhost' or empty) over any dynamic one supplied by
127 dhcp. With this change the rules for picking the hostname
128 match more closely the rules of other configuration settings
129 where the local administrator's configuration in /etc always
130 overrides any other settings.
132 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
133 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
134 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
135 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
136 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
137 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
138 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
139 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
140 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
141 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
142 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
143 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
144 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
145 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
146 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
147 Lindskog, WaLyong Cho, Will Woods, Zbigniew
150 -- Beijing, 2014-05-28
154 * When restoring the screen brightness at boot, stay away from
155 the darkest setting or from the lowest 5% of the available
156 range, depending on which is the larger value of both. This
157 should effectively protect the user from rebooting into a
158 black screen, should the brightness have been set to minimum
161 * sd-login gained a new sd_machine_get_class() call to
162 determine the class ("vm" or "container") of a machine
163 registered with machined.
165 * sd-login gained new calls
166 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
167 to query the identity of the peer of a local AF_UNIX
168 connection. They operate similarly to their sd_pid_get_xyz()
171 * PID 1 will now maintain a system-wide system state engine
172 with the states "starting", "running", "degraded",
173 "maintenance", "stopping". These states are bound to system
174 startup, normal runtime, runtime with at least one failed
175 service, rescue/emergency mode and system shutdown. This
176 state is shown in the "systemctl status" output when no unit
177 name is passed. It is useful to determine system state, in
178 particularly when doing so for many systems or containers at
181 * A new command "list-machines" has been added to "systemctl"
182 that lists all local OS containers and shows their system
183 state (see above), if systemd runs inside of them.
185 * systemctl gained a new "-r" switch to recursively enumerate
186 units on all local containers, when used with the
187 "list-unit" command (which is the default one that is
188 executed when no parameters are specified).
190 * The GPT automatic partition discovery logic will now honour
191 two GPT partition flags: one may be set on a partition to
192 cause it to be mounted read-only, and the other may be set
193 on a partition to ignore it during automatic discovery.
195 * Two new GPT type UUIDs have been added for automatic root
196 partition discovery, for 32-bit and 64-bit ARM. This is not
197 particularly useful for discovering the root directory on
198 these architectures during bare-metal boots (since UEFI is
199 not common there), but still very useful to allow booting of
200 ARM disk images in nspawn with the -i option.
202 * MAC addresses of interfaces created with nspawn's
203 --network-interface= switch will now be generated from the
204 machine name, and thus be stable between multiple invocations
207 * logind will now automatically remove all IPC objects owned
208 by a user if she or he fully logs out. This makes sure that
209 users who are logged out cannot continue to consume IPC
210 resources. This covers SysV memory, semaphores and message
211 queues as well as POSIX shared memory and message
212 queues. Traditionally, SysV and POSIX IPC had no life-cycle
213 limits. With this functionality, that is corrected. This may
214 be turned off by using the RemoveIPC= switch of logind.conf.
216 * The systemd-machine-id-setup and tmpfiles tools gained a
217 --root= switch to operate on a specific root directory,
220 * journald can now forward logged messages to the TTYs of all
221 logged in users ("wall"). This is the default for all
222 emergency messages now.
224 * A new tool systemd-journal-remote has been added to stream
225 journal log messages across the network.
227 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
228 controller trees are mounted into it. Note that the
229 directories mounted beneath it are not read-only. This is a
230 security measure and is particularly useful because glibc
231 actually includes a search logic to pick any tmpfs it can
232 find to implement shm_open() if /dev/shm is not available
233 (which it might very well be in namespaced setups).
235 * machinectl gained a new "poweroff" command to cleanly power
236 down a local OS container.
238 * The PrivateDevices= unit file setting will now also drop the
239 CAP_MKNOD capability from the capability bound set, and
240 imply DevicePolicy=closed.
242 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
243 comprehensively on all long-running systemd services where
246 * systemd-udevd will now run in a disassociated mount
247 namespace. To mount directories from udev rules, make sure to
248 pull in mount units via SYSTEMD_WANTS properties.
250 * The kdbus support gained support for uploading policy into
251 the kernel. sd-bus gained support for creating "monitoring"
252 connections that can eavesdrop into all bus communication
253 for debugging purposes.
255 * Timestamps may now be specified in seconds since the UNIX
256 epoch Jan 1st, 1970 by specifying "@" followed by the value
259 * Native tcpwrap support in systemd has been removed. tcpwrap
260 is old code, not really maintained anymore and has serious
261 shortcomings, and better options such as firewalls
262 exist. For setups that require tcpwrap usage, please
263 consider invoking your socket-activated service via tcpd,
264 like on traditional inetd.
266 * A new system.conf configuration option
267 DefaultTimerAccuracySec= has been added that controls the
268 default AccuracySec= setting of .timer units.
270 * Timer units gained a new WakeSystem= switch. If enabled,
271 timers configured this way will cause the system to resume
272 from system suspend (if the system supports that, which most
275 * Timer units gained a new Persistent= switch. If enabled,
276 timers configured this way will save to disk when they have
277 been last triggered. This information is then used on next
278 reboot to possible execute overdue timer events, that
279 could not take place because the system was powered off.
280 This enables simple anacron-like behaviour for timer units.
282 * systemctl's "list-timers" will now also list the time a
283 timer unit was last triggered in addition to the next time
284 it will be triggered.
286 * systemd-networkd will now assign predictable IPv4LL
287 addresses to its local interfaces.
289 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
290 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
291 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
292 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
293 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
294 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
295 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
296 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
299 -- Berlin, 2014-03-25
303 * A new unit file setting RestrictAddressFamilies= has been
304 added to restrict which socket address families unit
305 processes gain access to. This takes address family names
306 like "AF_INET" or "AF_UNIX", and is useful to minimize the
307 attack surface of services via exotic protocol stacks. This
308 is built on seccomp system call filters.
310 * Two new unit file settings RuntimeDirectory= and
311 RuntimeDirectoryMode= have been added that may be used to
312 manage a per-daemon runtime directories below /run. This is
313 an alternative for setting up directory permissions with
314 tmpfiles snippets, and has the advantage that the runtime
315 directory's lifetime is bound to the daemon runtime and that
316 the daemon starts up with an empty directory each time. This
317 is particularly useful when writing services that drop
318 privileges using the User= or Group= setting.
320 * The DeviceAllow= unit setting now supports globbing for
321 matching against device group names.
323 * The systemd configuration file system.conf gained new
324 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
325 DefaultMemoryAccounting= to globally turn on/off accounting
326 for specific resources (cgroups) for all units. These
327 settings may still be overridden individually in each unit
330 * systemd-gpt-auto-generator is now able to discover /srv and
331 root partitions in addition to /home and swap partitions. It
332 also supports LUKS-encrypted partitions now. With this in
333 place, automatic discovery of partitions to mount following
334 the Discoverable Partitions Specification
335 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
336 is now a lot more complete. This allows booting without
337 /etc/fstab and without root= on the kernel command line on
338 systems prepared appropriately.
340 * systemd-nspawn gained a new --image= switch which allows
341 booting up disk images and Linux installations on any block
342 device that follow the Discoverable Partitions Specification
343 (see above). This means that installations made with
344 appropriately updated installers may now be started and
345 deployed using container managers, completely
346 unmodified. (We hope that libvirt-lxc will add support for
347 this feature soon, too.)
349 * systemd-nspawn gained a new --network-macvlan= setting to
350 set up a private macvlan interface for the
351 container. Similarly, systemd-networkd gained a new
352 Kind=macvlan setting in .netdev files.
354 * systemd-networkd now supports configuring local addresses
357 * A new tool systemd-network-wait-online has been added to
358 synchronously wait for network connectivity using
361 * The sd-bus.h bus API gained a new sd_bus_track object for
362 tracking the life-cycle of bus peers. Note that sd-bus.h is
363 still not a public API though (unless you specify
364 --enable-kdbus on the configure command line, which however
365 voids your warranty and you get no API stability guarantee).
367 * The $XDG_RUNTIME_DIR runtime directories for each user are
368 now individual tmpfs instances, which has the benefit of
369 introducing separate pools for each user, with individual
370 size limits, and thus making sure that unprivileged clients
371 can no longer negatively impact the system or other users by
372 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
373 RuntimeDirectorySize= has been introduced that allows
374 controlling the default size limit for all users. It
375 defaults to 10% of the available physical memory. This is no
376 replacement for quotas on tmpfs though (which the kernel
377 still does not support), as /dev/shm and /tmp are still
378 shared resources used by both the system and unprivileged
381 * logind will now automatically turn off automatic suspending
382 on laptop lid close when more than one display is
383 connected. This was previously expected to be implemented
384 individually in desktop environments (such as GNOME),
385 however has been added to logind now, in order to fix a
386 boot-time race where a desktop environment might not have
387 been started yet and thus not been able to take an inhibitor
388 lock at the time where logind already suspends the system
391 * logind will now wait at least 30s after each system
392 suspend/resume cycle, and 3min after system boot before
393 suspending the system due to a closed laptop lid. This
394 should give USB docking stations and similar enough time to
395 be probed and configured after system resume and boot in
396 order to then act as suspend blocker.
398 * systemd-run gained a new --property= setting which allows
399 initialization of resource control properties (and others)
400 for the created scope or service unit. Example: "systemd-run
401 --property=BlockIOWeight=10 updatedb" may be used to run
402 updatedb at a low block IO scheduling weight.
404 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
405 now also work in --scope mode.
407 * When systemd is compiled with kdbus support, basic support
408 for enforced policies is now in place. (Note that enabling
409 kdbus still voids your warranty and no API compatibility
412 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
413 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
414 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
415 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
416 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
417 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
418 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
419 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
420 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
421 Zbigniew Jędrzejewski-Szmek
423 -- Berlin, 2014-03-12
427 * systemd will now relabel /dev after loading the SMACK policy
428 according to SMACK rules.
430 * A new unit file option AppArmorProfile= has been added to
431 set the AppArmor profile for the processes of a unit.
433 * A new condition check ConditionArchitecture= has been added
434 to conditionalize units based on the system architecture, as
435 reported by uname()'s "machine" field.
437 * systemd-networkd now supports matching on the system
438 virtualization, architecture, kernel command line, host name
441 * logind is now a lot more aggressive when suspending the
442 machine due to a closed laptop lid. Instead of acting only
443 on the lid close action, it will continuously watch the lid
444 status and act on it. This is useful for laptops where the
445 power button is on the outside of the chassis so that it can
446 be reached without opening the lid (such as the Lenovo
447 Yoga). On those machines, logind will now immediately
448 re-suspend the machine if the power button has been
449 accidentally pressed while the laptop was suspended and in a
452 * logind will now watch SW_DOCK switches and inhibit reaction
453 to the lid switch if it is pressed. This means that logind
454 will not suspend the machine anymore if the lid is closed
455 and the system is docked, if the laptop supports SW_DOCK
456 notifications via the input layer. Note that ACPI docking
457 stations do not generate this currently. Also note that this
458 logic is usually not fully sufficient and Desktop
459 Environments should take a lid switch inhibitor lock when an
460 external display is connected, as systemd will not watch
463 * nspawn will now make use of the devices cgroup controller by
464 default, and only permit creation of and access to the usual
465 API device nodes like /dev/null or /dev/random, as well as
466 access to (but not creation of) the pty devices.
468 * We will now ship a default .network file for
469 systemd-networkd that automatically configures DHCP for
470 network interfaces created by nspawn's --network-veth or
471 --network-bridge= switches.
473 * systemd will now understand the usual M, K, G, T suffixes
474 according to SI conventions (i.e. to the base 1000) when
475 referring to throughput and hardware metrics. It will stay
476 with IEC conventions (i.e. to the base 1024) for software
477 metrics, according to what is customary according to
478 Wikipedia. We explicitly document which base applies for
479 each configuration option.
481 * The DeviceAllow= setting in unit files now supports a syntax
482 to whitelist an entire group of devices node majors at once,
483 based on the /proc/devices listing. For example, with the
484 string "char-pts", it is now possible to whitelist all
485 current and future pseudo-TTYs at once.
487 * sd-event learned a new "post" event source. Event sources of
488 this type are triggered by the dispatching of any event
489 source of a type that is not "post". This is useful for
490 implementing clean-up and check event sources that are
491 triggered by other work being done in the program.
493 * systemd-networkd is no longer statically enabled, but uses
494 the usual [Install] sections so that it can be
495 enabled/disabled using systemctl. It still is enabled by
498 * When creating a veth interface pair with systemd-nspawn, the
499 host side will now be prefixed with "vb-" if
500 --network-bridge= is used, and with "ve-" if --network-veth
501 is used. This way, it is easy to distinguish these cases on
502 the host, for example to apply different configuration to
503 them with systemd-networkd.
505 * The compatibility libraries for libsystemd-journal.so,
506 libsystem-id128.so, libsystemd-login.so and
507 libsystemd-daemon.so do not make use of IFUNC
508 anymore. Instead, we now build libsystemd.so multiple times
509 under these alternative names. This means that the footprint
510 is drastically increased, but given that these are
511 transitional compatibility libraries, this should not matter
512 much. This change has been made necessary to support the ARM
513 platform for these compatibility libraries, as the ARM
514 toolchain is not really at the same level as the toolchain
515 for other architectures like x86 and does not support
516 IFUNC. Please make sure to use --enable-compat-libs only
517 during a transitional period!
519 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
520 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
521 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
522 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
523 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
524 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
525 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
526 Zbigniew Jędrzejewski-Szmek
528 -- Berlin, 2014-02-24
532 * A new component "systemd-networkd" has been added that can
533 be used to configure local network interfaces statically or
534 via DHCP. It is capable of bringing up bridges, VLANs, and
535 bonding. Currently, no hook-ups for interactive network
536 configuration are provided. Use this for your initrd,
537 container, embedded, or server setup if you need a simple,
538 yet powerful, network configuration solution. This
539 configuration subsystem is quite nifty, as it allows wildcard
540 hotplug matching in interfaces. For example, with a single
541 configuration snippet, you can configure that all Ethernet
542 interfaces showing up are automatically added to a bridge,
543 or similar. It supports link-sensing and more.
545 * A new tool "systemd-socket-proxyd" has been added which can
546 act as a bidirectional proxy for TCP sockets. This is
547 useful for adding socket activation support to services that
548 do not actually support socket activation, including virtual
549 machines and the like.
551 * Add a new tool to save/restore rfkill state on
554 * Save/restore state of keyboard backlights in addition to
555 display backlights on shutdown/boot.
557 * udev learned a new SECLABEL{} construct to label device
558 nodes with a specific security label when they appear. For
559 now, only SECLABEL{selinux} is supported, but the syntax is
560 prepared for additional security frameworks.
562 * udev gained a new scheme to configure link-level attributes
563 from files in /etc/systemd/network/*.link. These files can
564 match against MAC address, device path, driver name and type,
565 and will apply attributes like the naming policy, link speed,
566 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
567 address assignment policy (randomized, ...).
569 * The configuration of network interface naming rules for
570 "permanent interface names" has changed: a new NamePolicy=
571 setting in the [Link] section of .link files determines the
572 priority of possible naming schemes (onboard, slot, mac,
573 path). The default value of this setting is determined by
574 /usr/lib/net/links/99-default.link. Old
575 80-net-name-slot.rules udev configuration file has been
576 removed, so local configuration overriding this file should
577 be adapated to override 99-default.link instead.
579 * When the User= switch is used in a unit file, also
580 initialize $SHELL= based on the user database entry.
582 * systemd no longer depends on libdbus. All communication is
583 now done with sd-bus, systemd's low-level bus library
586 * kdbus support has been added to PID 1 itself. When kdbus is
587 enabled, this causes PID 1 to set up the system bus and
588 enable support for a new ".busname" unit type that
589 encapsulates bus name activation on kdbus. It works a little
590 bit like ".socket" units, except for bus names. A new
591 generator has been added that converts classic dbus1 service
592 activation files automatically into native systemd .busname
595 * sd-bus: add a light-weight vtable implementation that allows
596 defining objects on the bus with a simple static const
597 vtable array of its methods, signals and properties.
599 * systemd will not generate or install static dbus
600 introspection data anymore to /usr/share/dbus-1/interfaces,
601 as the precise format of these files is unclear, and
602 nothing makes use of it.
604 * A proxy daemon is now provided to proxy clients connecting
605 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
606 compatibility with classic D-Bus.
608 * A bus driver implementation has been added that supports the
609 classic D-Bus bus driver calls on kdbus, also for
610 compatibility purposes.
612 * A new API "sd-event.h" has been added that implements a
613 minimal event loop API built around epoll. It provides a
614 couple of features that direct epoll usage is lacking:
615 prioritization of events, scales to large numbers of timer
616 events, per-event timer slack (accuracy), system-wide
617 coalescing of timer events, exit handlers, watchdog
618 supervision support using systemd's sd_notify() API, child
621 * A new API "sd-rntl.h" has been added that provides an API
622 around the route netlink interface of the kernel, similar in
625 * A new API "sd-dhcp-client.h" has been added that provides a
626 small DHCPv4 client-side implementation. This is used by
629 * There is a new kernel command line option
630 "systemd.restore_state=0|1". When set to "0", none of the
631 systemd tools will restore saved runtime state to hardware
632 devices. More specifically, the rfkill and backlight states
635 * The FsckPassNo= compatibility option in mount/service units
636 has been removed. The fstab generator will now add the
637 necessary dependencies automatically, and does not require
638 PID1's support for that anymore.
640 * journalctl gained a new switch, --list-boots, that lists
641 recent boots with their times and boot IDs.
643 * The various tools like systemctl, loginctl, timedatectl,
644 busctl, systemd-run, ... have gained a new switch "-M" to
645 connect to a specific, local OS container (as direct
646 connection, without requiring SSH). This works on any
647 container that is registered with machined, such as those
648 created by libvirt-lxc or nspawn.
650 * systemd-run and systemd-analyze also gained support for "-H"
651 to connect to remote hosts via SSH. This is particularly
652 useful for systemd-run because it enables queuing of jobs
655 * machinectl gained a new command "login" to open a getty
656 login in any local container. This works with any container
657 that is registered with machined (such as those created by
658 libvirt-lxc or nspawn), and which runs systemd inside.
660 * machinectl gained a new "reboot" command that may be used to
661 trigger a reboot on a specific container that is registered
662 with machined. This works on any container that runs an init
665 * systemctl gained a new "list-timers" command to print a nice
666 listing of installed timer units with the times they elapse
669 * Alternative reboot() parameters may now be specified on the
670 "systemctl reboot" command line and are passed to the
671 reboot() system call.
673 * systemctl gained a new --job-mode= switch to configure the
674 mode to queue a job with. This is a more generic version of
675 --fail, --irreversible, and --ignore-dependencies, which are
676 still available but not advertised anymore.
678 * /etc/systemd/system.conf gained new settings to configure
679 various default timeouts of units, as well as the default
680 start limit interval and burst. These may still be overridden
683 * PID1 will now export on the bus profile data of the security
684 policy upload process (such as the SELinux policy upload to
687 * journald: when forwarding logs to the console, include
688 timestamps (following the setting in
689 /sys/module/printk/parameters/time).
691 * OnCalendar= in timer units now understands the special
692 strings "yearly" and "annually". (Both are equivalent)
694 * The accuracy of timer units is now configurable with the new
695 AccuracySec= setting. It defaults to 1min.
697 * A new dependency type JoinsNamespaceOf= has been added that
698 allows running two services within the same /tmp and network
699 namespace, if PrivateNetwork= or PrivateTmp= are used.
701 * A new command "cat" has been added to systemctl. It outputs
702 the original unit file of a unit, and concatenates the
703 contents of additional "drop-in" unit file snippets, so that
704 the full configuration is shown.
706 * systemctl now supports globbing on the various "list-xyz"
707 commands, like "list-units" or "list-sockets", as well as on
708 those commands which take multiple unit names.
710 * journalctl's --unit= switch gained support for globbing.
712 * All systemd daemons now make use of the watchdog logic so
713 that systemd automatically notices when they hang.
715 * If the $container_ttys environment variable is set,
716 getty-generator will automatically spawn a getty for each
717 listed tty. This is useful for container managers to request
718 login gettys to be spawned on as many ttys as needed.
720 * %h, %s, %U specifier support is not available anymore when
721 used in unit files for PID 1. This is because NSS calls are
722 not safe from PID 1. They stay available for --user
723 instances of systemd, and as special case for the root user.
725 * loginctl gained a new "--no-legend" switch to turn off output
728 * The "sd-login.h" API gained three new calls:
729 sd_session_is_remote(), sd_session_get_remote_user(),
730 sd_session_get_remote_host() to query information about
733 * The udev hardware database now also carries vendor/product
734 information of SDIO devices.
736 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
737 determine whether watchdog notifications are requested by
740 * Socket-activated per-connection services now include a
741 short description of the connection parameters in the
744 * tmpfiles gained a new "--boot" option. When this is not used,
745 only lines where the command character is not suffixed with
746 "!" are executed. When this option is specified, those
747 options are executed too. This partitions tmpfiles
748 directives into those that can be safely executed at any
749 time, and those which should be run only at boot (for
750 example, a line that creates /run/nologin).
752 * A new API "sd-resolve.h" has been added which provides a simple
753 asynchronous wrapper around glibc NSS host name resolution
754 calls, such as getaddrinfo(). In contrast to glibc's
755 getaddrinfo_a(), it does not use signals. In contrast to most
756 other asynchronous name resolution libraries, this one does
757 not reimplement DNS, but reuses NSS, so that alternate
758 host name resolution systems continue to work, such as mDNS,
759 LDAP, etc. This API is based on libasyncns, but it has been
760 cleaned up for inclusion in systemd.
762 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
763 "sd-daemon.h" are no longer found in individual libraries
764 libsystemd-journal.so, libsystemd-login.so,
765 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
766 merged them into a single library, libsystemd.so, which
767 provides all symbols. The reason for this is cyclic
768 dependencies, as these libraries tend to use each other's
769 symbols. So far, we have managed to workaround that by linking
770 a copy of a good part of our code into each of these
771 libraries again and again, which, however, makes certain
772 things hard to do, like sharing static variables. Also, it
773 substantially increases footprint. With this change, there
774 is only one library for the basic APIs systemd
775 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
776 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
777 library as well, however are subject to the --enable-kdbus
778 switch (see below). Note that "sd-dhcp-client.h" is not part
779 of this library (this is because it only consumes, never
780 provides, services of/to other APIs). To make the transition
781 easy from the separate libraries to the unified one, we
782 provide the --enable-compat-libs compile-time switch which
783 will generate stub libraries that are compatible with the
784 old ones but redirect all calls to the new one.
786 * All of the kdbus logic and the new APIs "sd-bus.h",
787 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
788 and "sd-utf8.h" are compile-time optional via the
789 "--enable-kdbus" switch, and they are not compiled in by
790 default. To make use of kdbus, you have to explicitly enable
791 the switch. Note however, that neither the kernel nor the
792 userspace API for all of this is considered stable yet. We
793 want to maintain the freedom to still change the APIs for
794 now. By specifying this build-time switch, you acknowledge
795 that you are aware of the instability of the current
798 * Also, note that while kdbus is pretty much complete,
799 it lacks one thing: proper policy support. This means you
800 can build a fully working system with all features; however,
801 it will be highly insecure. Policy support will be added in
802 one of the next releases, at the same time that we will
803 declare the APIs stable.
805 * When the kernel command-line argument "kdbus" is specified,
806 systemd will automatically load the kdbus.ko kernel module. At
807 this stage of development, it is only useful for testing kdbus
808 and should not be used in production. Note: if "--enable-kdbus"
809 is specified, and the kdbus.ko kernel module is available, and
810 "kdbus" is added to the kernel command line, the entire system
811 runs with kdbus instead of dbus-daemon, with the above mentioned
812 problem of missing the system policy enforcement. Also a future
813 version of kdbus.ko or a newer systemd will not be compatible with
814 each other, and will unlikely be able to boot the machine if only
815 one of them is updated.
817 * systemctl gained a new "import-environment" command which
818 uploads the caller's environment (or parts thereof) into the
819 service manager so that it is inherited by services started
820 by the manager. This is useful to upload variables like
821 $DISPLAY into the user service manager.
823 * A new PrivateDevices= switch has been added to service units
824 which allows running a service with a namespaced /dev
825 directory that does not contain any device nodes for
826 physical devices. More specifically, it only includes devices
827 such as /dev/null, /dev/urandom, and /dev/zero which are API
830 * logind has been extended to support behaviour like VT
831 switching on seats that do not support a VT. This makes
832 multi-session available on seats that are not the first seat
833 (seat0), and on systems where kernel support for VTs has
834 been disabled at compile-time.
836 * If a process holds a delay lock for system sleep or shutdown
837 and fails to release it in time, we will now log its
838 identity. This makes it easier to identify processes that
839 cause slow suspends or power-offs.
841 * When parsing /etc/crypttab, support for a new key-slot=
842 option as supported by Debian is added. It allows indicating
843 which LUKS slot to use on disk, speeding up key loading.
845 * The sd_journald_sendv() API call has been checked and
846 officially declared to be async-signal-safe so that it may
847 be invoked from signal handlers for logging purposes.
849 * Boot-time status output is now enabled automatically after a
850 short timeout if boot does not progress, in order to give
851 the user an indication what she or he is waiting for.
853 * The boot-time output has been improved to show how much time
854 remains until jobs expire.
856 * The KillMode= switch in service units gained a new possible
857 value "mixed". If set, and the unit is shut down, then the
858 initial SIGTERM signal is sent only to the main daemon
859 process, while the following SIGKILL signal is sent to
860 all remaining processes of the service.
862 * When a scope unit is registered, a new property "Controller"
863 may be set. If set to a valid bus name, systemd will send a
864 RequestStop() signal to this name when it would like to shut
865 down the scope. This may be used to hook manager logic into
866 the shutdown logic of scope units. Also, scope units may now
867 be put in a special "abandoned" state, in which case the
868 manager process which created them takes no further
869 responsibilities for it.
871 * When reading unit files, systemd will now verify
872 the access mode of these files, and warn about certain
873 suspicious combinations. This has been added to make it
874 easier to track down packaging bugs where unit files are
875 marked executable or world-writable.
877 * systemd-nspawn gained a new "--setenv=" switch to set
878 container-wide environment variables. The similar option in
879 systemd-activate was renamed from "--environment=" to
880 "--setenv=" for consistency.
882 * systemd-nspawn has been updated to create a new kdbus domain
883 for each container that is invoked, thus allowing each
884 container to have its own set of system and user buses,
885 independent of the host.
887 * systemd-nspawn gained a new --drop-capability= switch to run
888 the container with less capabilities than the default. Both
889 --drop-capability= and --capability= now take the special
890 string "all" for dropping or keeping all capabilities.
892 * systemd-nspawn gained new switches for executing containers
893 with specific SELinux labels set.
895 * systemd-nspawn gained a new --quiet switch to not generate
896 any additional output but the container's own console
899 * systemd-nspawn gained a new --share-system switch to run a
900 container without PID namespacing enabled.
902 * systemd-nspawn gained a new --register= switch to control
903 whether the container is registered with systemd-machined or
904 not. This is useful for containers that do not run full
905 OS images, but only specific apps.
907 * systemd-nspawn gained a new --keep-unit which may be used
908 when invoked as the only program from a service unit, and
909 results in registration of the unit service itself in
910 systemd-machined, instead of a newly opened scope unit.
912 * systemd-nspawn gained a new --network-interface= switch for
913 moving arbitrary interfaces to the container. The new
914 --network-veth switch creates a virtual Ethernet connection
915 between host and container. The new --network-bridge=
916 switch then allows assigning the host side of this virtual
917 Ethernet connection to a bridge device.
919 * systemd-nspawn gained a new --personality= switch for
920 setting the kernel personality for the container. This is
921 useful when running a 32-bit container on a 64-bit host. A
922 similar option Personality= is now also available for service
925 * logind will now also track a "Desktop" identifier for each
926 session which encodes the desktop environment of it. This is
927 useful for desktop environments that want to identify
928 multiple running sessions of itself easily.
930 * A new SELinuxContext= setting for service units has been
931 added that allows setting a specific SELinux execution
932 context for a service.
934 * Most systemd client tools will now honour $SYSTEMD_LESS for
935 settings of the "less" pager. By default, these tools will
936 override $LESS to allow certain operations to work, such as
937 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
938 influence this logic.
940 * systemd's "seccomp" hook-up has been changed to make use of
941 the libseccomp library instead of using its own
942 implementation. This has benefits for portability among
945 * For usage together with SystemCallFilter=, a new
946 SystemCallErrorNumber= setting has been introduced that
947 allows configuration of a system error number to be returned
948 on filtered system calls, instead of immediately killing the
949 process. Also, SystemCallArchitectures= has been added to
950 limit access to system calls of a particular architecture
951 (in order to turn off support for unused secondary
952 architectures). There is also a global
953 SystemCallArchitectures= setting in system.conf now to turn
954 off support for non-native system calls system-wide.
956 * systemd requires a kernel with a working name_to_handle_at(),
957 please see the kernel config requirements in the README file.
959 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
960 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
961 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
962 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
963 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
964 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
965 Elia Pinto, Florian Weimer, George McCollister, Goffredo
966 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
967 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
968 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
969 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
970 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
971 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
972 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
973 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
974 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
975 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
976 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
977 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
978 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
979 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
980 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
981 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
982 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
984 -- Berlin, 2014-02-20
988 * logind has gained support for facilitating privileged input
989 and drm device access for unprivileged clients. This work is
990 useful to allow Wayland display servers (and similar
991 programs, such as kmscon) to run under the user's ID and
992 access input and drm devices which are normally
993 protected. When this is used (and the kernel is new enough)
994 logind will "mute" IO on the file descriptors passed to
995 Wayland as long as it is in the background and "unmute" it
996 if it returns into the foreground. This allows secure
997 session switching without allowing background sessions to
998 eavesdrop on input and display data. This also introduces
999 session switching support if VT support is turned off in the
1000 kernel, and on seats that are not seat0.
1002 * A new kernel command line option luks.options= is understood
1003 now which allows specifiying LUKS options for usage for LUKS
1004 encrypted partitions specified with luks.uuid=.
1006 * tmpfiles.d(5) snippets may now use specifier expansion in
1007 path names. More specifically %m, %b, %H, %v, are now
1008 replaced by the local machine id, boot id, hostname, and
1009 kernel version number.
1011 * A new tmpfiles.d(5) command "m" has been introduced which
1012 may be used to change the owner/group/access mode of a file
1013 or directory if it exists, but do nothing if it does not.
1015 * This release removes high-level support for the
1016 MemorySoftLimit= cgroup setting. The underlying kernel
1017 cgroup attribute memory.soft_limit= is currently badly
1018 designed and likely to be removed from the kernel API in its
1019 current form, hence we should not expose it for now.
1021 * The memory.use_hierarchy cgroup attribute is now enabled for
1022 all cgroups systemd creates in the memory cgroup
1023 hierarchy. This option is likely to be come the built-in
1024 default in the kernel anyway, and the non-hierarchial mode
1025 never made much sense in the intrinsically hierarchial
1028 * A new field _SYSTEMD_SLICE= is logged along with all journal
1029 messages containing the slice a message was generated
1030 from. This is useful to allow easy per-customer filtering of
1031 logs among other things.
1033 * systemd-journald will no longer adjust the group of journal
1034 files it creates to the "systemd-journal" group. Instead we
1035 rely on the journal directory to be owned by the
1036 "systemd-journal" group, and its setgid bit set, so that the
1037 kernel file system layer will automatically enforce that
1038 journal files inherit this group assignment. The reason for
1039 this change is that we cannot allow NSS look-ups from
1040 journald which would be necessary to resolve
1041 "systemd-journal" to a numeric GID, because this might
1042 create deadlocks if NSS involves synchronous queries to
1043 other daemons (such as nscd, or sssd) which in turn are
1044 logging clients of journald and might block on it, which
1045 would then dead lock. A tmpfiles.d(5) snippet included in
1046 systemd will make sure the setgid bit and group are
1047 properly set on the journal directory if it exists on every
1048 boot. However, we recommend adjusting it manually after
1049 upgrades too (or from RPM scriptlets), so that the change is
1050 not delayed until next reboot.
1052 * Backlight and random seed files in /var/lib/ have moved into
1053 the /var/lib/systemd/ directory, in order to centralize all
1054 systemd generated files in one directory.
1056 * Boot time performance measurements (as displayed by
1057 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
1058 performance information if that's available to determine how
1059 much time BIOS and boot loader initialization required. With
1060 a sufficiently new BIOS you hence no longer need to boot
1061 with Gummiboot to get access to such information.
1063 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
1064 Cristian Rodríguez, Dave Reisner, David Herrmann, David
1065 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
1066 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
1067 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
1068 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
1069 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1071 -- Berlin, 2013-10-02
1075 * The Restart= option for services now understands a new
1076 on-watchdog setting, which will restart the service
1077 automatically if the service stops sending out watchdog keep
1078 alive messages (as configured with WatchdogSec=).
1080 * The getty generator (which is responsible for bringing up a
1081 getty on configured serial consoles) will no longer only
1082 start a getty on the primary kernel console but on all
1083 others, too. This makes the order in which console= is
1084 specified on the kernel command line less important.
1086 * libsystemd-logind gained a new sd_session_get_vt() call to
1087 retrieve the VT number of a session.
1089 * If the option "tries=0" is set for an entry of /etc/crypttab
1090 its passphrase is queried indefinitely instead of any
1091 maximum number of tries.
1093 * If a service with a configure PID file terminates its PID
1094 file will now be removed automatically if it still exists
1095 afterwards. This should put an end to stale PID files.
1097 * systemd-run will now also take relative binary path names
1098 for execution and no longer insists on absolute paths.
1100 * InaccessibleDirectories= and ReadOnlyDirectories= now take
1101 paths that are optionally prefixed with "-" to indicate that
1102 it should not be considered a failure if they do not exist.
1104 * journalctl -o (and similar commands) now understands a new
1105 output mode "short-precise", it is similar to "short" but
1106 shows timestamps with usec accuracy.
1108 * The option "discard" (as known from Debian) is now
1109 synonymous to "allow-discards" in /etc/crypttab. In fact,
1110 "discard" is preferred now (since it is easier to remember
1113 * Some licensing clean-ups were made, so that more code is now
1114 LGPL-2.1 licensed than before.
1116 * A minimal tool to save/restore the display backlight
1117 brightness across reboots has been added. It will store the
1118 backlight setting as late as possible at shutdown, and
1119 restore it as early as possible during reboot.
1121 * A logic to automatically discover and enable home and swap
1122 partitions on GPT disks has been added. With this in place
1123 /etc/fstab becomes optional for many setups as systemd can
1124 discover certain partitions located on the root disk
1125 automatically. Home partitions are recognized under their
1126 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
1127 partitions are recognized under their GPT type ID
1128 0657fd6da4ab43c484e50933c84b4f4f.
1130 * systemd will no longer pass any environment from the kernel
1131 or initrd to system services. If you want to set an
1132 environment for all services, do so via the kernel command
1133 line systemd.setenv= assignment.
1135 * The systemd-sysctl tool no longer natively reads the file
1136 /etc/sysctl.conf. If desired, the file should be symlinked
1137 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
1138 legacy support by a symlink rather than built-in code, it
1139 also makes the otherwise hidden order of application of the
1140 different files visible. (Note that this partly reverts to a
1141 pre-198 application order of sysctl knobs!)
1143 * The "systemctl set-log-level" and "systemctl dump" commands
1144 have been moved to systemd-analyze.
1146 * systemd-run learned the new --remain-after-exit switch,
1147 which causes the scope unit not to be cleaned up
1148 automatically after the process terminated.
1150 * tmpfiles learned a new --exclude-prefix= switch to exclude
1151 certain paths from operation.
1153 * journald will now automatically flush all messages to disk
1154 as soon as a message of the log priorities CRIT, ALERT or
1157 Contributions from: Andrew Cook, Brandon Philips, Christian
1158 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
1159 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
1160 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
1161 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
1162 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
1163 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
1164 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
1165 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
1166 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
1167 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
1168 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
1169 William Giokas, Zbigniew Jędrzejewski-Szmek
1171 -- Berlin, 2013-09-13
1175 * The documentation has been updated to cover the various new
1176 concepts introduced with 205.
1178 * Unit files now understand the new %v specifier which
1179 resolves to the kernel version string as returned by "uname
1182 * systemctl now supports filtering the unit list output by
1183 load state, active state and sub state, using the new
1186 * "systemctl status" will now show the results of the
1187 condition checks (like ConditionPathExists= and similar) of
1188 the last start attempts of the unit. They are also logged to
1191 * "journalctl -b" may now be used to look for boot output of a
1192 specific boot. Try "journalctl -b -1" for the previous boot,
1193 but the syntax is substantially more powerful.
1195 * "journalctl --show-cursor" has been added which prints the
1196 cursor string the last shown log line. This may then be used
1197 with the new "journalctl --after-cursor=" switch to continue
1198 browsing logs from that point on.
1200 * "journalctl --force" may now be used to force regeneration
1203 * Creation of "dead" device nodes has been moved from udev
1204 into kmod and tmpfiles. Previously, udev would read the kmod
1205 databases to pre-generate dead device nodes based on meta
1206 information contained in kernel modules, so that these would
1207 be auto-loaded on access rather then at boot. As this
1208 does not really have much to do with the exposing actual
1209 kernel devices to userspace this has always been slightly
1210 alien in the udev codebase. Following the new scheme kmod
1211 will now generate a runtime snippet for tmpfiles from the
1212 module meta information and it now is tmpfiles' job to the
1213 create the nodes. This also allows overriding access and
1214 other parameters for the nodes using the usual tmpfiles
1215 facilities. As side effect this allows us to remove the
1216 CAP_SYS_MKNOD capability bit from udevd entirely.
1218 * logind's device ACLs may now be applied to these "dead"
1219 devices nodes too, thus finally allowing managed access to
1220 devices such as /dev/snd/sequencer whithout loading the
1221 backing module right-away.
1223 * A new RPM macro has been added that may be used to apply
1224 tmpfiles configuration during package installation.
1226 * systemd-detect-virt and ConditionVirtualization= now can
1227 detect User-Mode-Linux machines (UML).
1229 * journald will now implicitly log the effective capabilities
1230 set of processes in the message metadata.
1232 * systemd-cryptsetup has gained support for TrueCrypt volumes.
1234 * The initrd interface has been simplified (more specifically,
1235 support for passing performance data via environment
1236 variables and fsck results via files in /run has been
1237 removed). These features were non-essential, and are
1238 nowadays available in a much nicer way by having systemd in
1239 the initrd serialize its state and have the hosts systemd
1240 deserialize it again.
1242 * The udev "keymap" data files and tools to apply keyboard
1243 specific mappings of scan to key codes, and force-release
1244 scan code lists have been entirely replaced by a udev
1245 "keyboard" builtin and a hwdb data file.
1247 * systemd will now honour the kernel's "quiet" command line
1248 argument also during late shutdown, resulting in a
1249 completely silent shutdown when used.
1251 * There's now an option to control the SO_REUSEPORT socket
1252 option in .socket units.
1254 * Instance units will now automatically get a per-template
1255 subslice of system.slice unless something else is explicitly
1256 configured. For example, instances of sshd@.service will now
1257 implicitly be placed in system-sshd.slice rather than
1258 system.slice as before.
1260 * Test coverage support may now be enabled at build time.
1262 Contributions from: Dave Reisner, Frederic Crozat, Harald
1263 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
1264 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
1265 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
1266 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
1267 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
1268 Giokas, Zbigniew Jędrzejewski-Szmek
1270 -- Berlin, 2013-07-23
1274 * Two new unit types have been introduced:
1276 Scope units are very similar to service units, however, are
1277 created out of pre-existing processes -- instead of PID 1
1278 forking off the processes. By using scope units it is
1279 possible for system services and applications to group their
1280 own child processes (worker processes) in a powerful way
1281 which then maybe used to organize them, or kill them
1282 together, or apply resource limits on them.
1284 Slice units may be used to partition system resources in an
1285 hierarchial fashion and then assign other units to them. By
1286 default there are now three slices: system.slice (for all
1287 system services), user.slice (for all user sessions),
1288 machine.slice (for VMs and containers).
1290 Slices and scopes have been introduced primarily in
1291 context of the work to move cgroup handling to a
1292 single-writer scheme, where only PID 1
1293 creates/removes/manages cgroups.
1295 * There's a new concept of "transient" units. In contrast to
1296 normal units these units are created via an API at runtime,
1297 not from configuration from disk. More specifically this
1298 means it is now possible to run arbitrary programs as
1299 independent services, with all execution parameters passed
1300 in via bus APIs rather than read from disk. Transient units
1301 make systemd substantially more dynamic then it ever was,
1302 and useful as a general batch manager.
1304 * logind has been updated to make use of scope and slice units
1305 for managing user sessions. As a user logs in he will get
1306 his own private slice unit, to which all sessions are added
1307 as scope units. We also added support for automatically
1308 adding an instance of user@.service for the user into the
1309 slice. Effectively logind will no longer create cgroup
1310 hierarchies on its own now, it will defer entirely to PID 1
1311 for this by means of scope, service and slice units. Since
1312 user sessions this way become entities managed by PID 1
1313 the output of "systemctl" is now a lot more comprehensive.
1315 * A new mini-daemon "systemd-machined" has been added which
1316 may be used by virtualization managers to register local
1317 VMs/containers. nspawn has been updated accordingly, and
1318 libvirt will be updated shortly. machined will collect a bit
1319 of meta information about the VMs/containers, and assign
1320 them their own scope unit (see above). The collected
1321 meta-data is then made available via the "machinectl" tool,
1322 and exposed in "ps" and similar tools. machined/machinectl
1323 is compile-time optional.
1325 * As discussed earlier, the low-level cgroup configuration
1326 options ControlGroup=, ControlGroupModify=,
1327 ControlGroupPersistent=, ControlGroupAttribute= have been
1328 removed. Please use high-level attribute settings instead as
1329 well as slice units.
1331 * A new bus call SetUnitProperties() has been added to alter
1332 various runtime parameters of a unit. This is primarily
1333 useful to alter cgroup parameters dynamically in a nice way,
1334 but will be extended later on to make more properties
1335 modifiable at runtime. systemctl gained a new set-properties
1336 command that wraps this call.
1338 * A new tool "systemd-run" has been added which can be used to
1339 run arbitrary command lines as transient services or scopes,
1340 while configuring a number of settings via the command
1341 line. This tool is currently very basic, however already
1342 very useful. We plan to extend this tool to even allow
1343 queuing of execution jobs with time triggers from the
1344 command line, similar in fashion to "at".
1346 * nspawn will now inform the user explicitly that kernels with
1347 audit enabled break containers, and suggest the user to turn
1350 * Support for detecting the IMA and AppArmor security
1351 frameworks with ConditionSecurity= has been added.
1353 * journalctl gained a new "-k" switch for showing only kernel
1354 messages, mimicking dmesg output; in addition to "--user"
1355 and "--system" switches for showing only user's own logs
1358 * systemd-delta can now show information about drop-in
1359 snippets extending unit files.
1361 * libsystemd-bus has been substantially updated but is still
1362 not available as public API.
1364 * systemd will now look for the "debug" argument on the kernel
1365 command line and enable debug logging, similar to what
1366 "systemd.log_level=debug" already did before.
1368 * "systemctl set-default", "systemctl get-default" has been
1369 added to configure the default.target symlink, which
1370 controls what to boot into by default.
1372 * "systemctl set-log-level" has been added as a convenient
1373 way to raise and lower systemd logging threshold.
1375 * "systemd-analyze plot" will now show the time the various
1376 generators needed for execution, as well as information
1377 about the unit file loading.
1379 * libsystemd-journal gained a new sd_journal_open_files() call
1380 for opening specific journal files. journactl also gained a
1381 new switch to expose this new functionality. Previously we
1382 only supported opening all files from a directory, or all
1383 files from the system, as opening individual files only is
1384 racy due to journal file rotation.
1386 * systemd gained the new DefaultEnvironment= setting in
1387 /etc/systemd/system.conf to set environment variables for
1390 * If a privileged process logs a journal message with the
1391 OBJECT_PID= field set, then journald will automatically
1392 augment this with additional OBJECT_UID=, OBJECT_GID=,
1393 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
1394 system services want to log events about specific client
1395 processes. journactl/systemctl has been updated to make use
1396 of this information if all log messages regarding a specific
1399 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
1400 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
1401 Reisner, David Coppa, David King, David Strauss, Eelco
1402 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
1403 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
1404 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
1405 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
1406 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
1407 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
1408 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
1409 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
1410 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
1411 Łukasz Stelmach, 장동준
1415 * The Python bindings gained some minimal support for the APIs
1416 exposed by libsystemd-logind.
1418 * ConditionSecurity= gained support for detecting SMACK. Since
1419 this condition already supports SELinux and AppArmor we only
1420 miss IMA for this. Patches welcome!
1422 Contributions from: Karol Lewandowski, Lennart Poettering,
1423 Zbigniew Jędrzejewski-Szmek
1427 * systemd-nspawn will now create /etc/resolv.conf if
1428 necessary, before bind-mounting the host's file onto it.
1430 * systemd-nspawn will now store meta information about a
1431 container on the container's cgroup as extended attribute
1432 fields, including the root directory.
1434 * The cgroup hierarchy has been reworked in many ways. All
1435 objects any of the components systemd creates in the cgroup
1436 tree are now suffixed. More specifically, user sessions are
1437 now placed in cgroups suffixed with ".session", users in
1438 cgroups suffixed with ".user", and nspawn containers in
1439 cgroups suffixed with ".nspawn". Furthermore, all cgroup
1440 names are now escaped in a simple scheme to avoid collision
1441 of userspace object names with kernel filenames. This work
1442 is preparation for making these objects relocatable in the
1443 cgroup tree, in order to allow easy resource partitioning of
1444 these objects without causing naming conflicts.
1446 * systemctl list-dependencies gained the new switches
1447 --plain, --reverse, --after and --before.
1449 * systemd-inhibit now shows the process name of processes that
1450 have taken an inhibitor lock.
1452 * nss-myhostname will now also resolve "localhost"
1453 implicitly. This makes /etc/hosts an optional file and
1454 nicely handles that on IPv6 ::1 maps to both "localhost" and
1457 * libsystemd-logind.so gained a new call
1458 sd_get_machine_names() to enumerate running containers and
1459 VMs (currently only supported by very new libvirt and
1460 nspawn). sd_login_monitor can now be used to watch
1461 VMs/containers coming and going.
1463 * .include is not allowed recursively anymore, and only in
1464 unit files. Usually it is better to use drop-in snippets in
1465 .d/*.conf anyway, as introduced with systemd 198.
1467 * systemd-analyze gained a new "critical-chain" command that
1468 determines the slowest chain of units run during system
1469 boot-up. It is very useful for tracking down where
1470 optimizing boot time is the most beneficial.
1472 * systemd will no longer allow manipulating service paths in
1473 the name=systemd:/system cgroup tree using ControlGroup= in
1474 units. (But is still fine with it in all other dirs.)
1476 * There's a new systemd-nspawn@.service service file that may
1477 be used to easily run nspawn containers as system
1478 services. With the container's root directory in
1479 /var/lib/container/foobar it is now sufficient to run
1480 "systemctl start systemd-nspawn@foobar.service" to boot it.
1482 * systemd-cgls gained a new parameter "--machine" to list only
1483 the processes within a certain container.
1485 * ConditionSecurity= now can check for "apparmor". We still
1486 are lacking checks for SMACK and IMA for this condition
1487 check though. Patches welcome!
1489 * A new configuration file /etc/systemd/sleep.conf has been
1490 added that may be used to configure which kernel operation
1491 systemd is supposed to execute when "suspend", "hibernate"
1492 or "hybrid-sleep" is requested. This makes the new kernel
1493 "freeze" state accessible to the user.
1495 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
1496 the passed argument if applicable.
1498 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
1499 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
1500 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
1501 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
1502 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
1503 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
1504 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
1509 * The output of 'systemctl list-jobs' got some polishing. The
1510 '--type=' argument may now be passed more than once. A new
1511 command 'systemctl list-sockets' has been added which shows
1512 a list of kernel sockets systemd is listening on with the
1513 socket units they belong to, plus the units these socket
1516 * The experimental libsystemd-bus library got substantial
1517 updates to work in conjunction with the (also experimental)
1518 kdbus kernel project. It works well enough to exchange
1519 messages with some sophistication. Note that kdbus is not
1520 ready yet, and the library is mostly an elaborate test case
1521 for now, and not installable.
1523 * systemd gained a new unit 'systemd-static-nodes.service'
1524 that generates static device nodes earlier during boot, and
1525 can run in conjunction with udev.
1527 * libsystemd-login gained a new call sd_pid_get_user_unit()
1528 to retrieve the user systemd unit a process is running
1529 in. This is useful for systems where systemd is used as
1532 * systemd-nspawn now places all containers in the new /machine
1533 top-level cgroup directory in the name=systemd
1534 hierarchy. libvirt will soon do the same, so that we get a
1535 uniform separation of /system, /user and /machine for system
1536 services, user processes and containers/virtual
1537 machines. This new cgroup hierarchy is also useful to stick
1538 stable names to specific container instances, which can be
1539 recognized later this way (this name may be controlled
1540 via systemd-nspawn's new -M switch). libsystemd-login also
1541 gained a new call sd_pid_get_machine_name() to retrieve the
1542 name of the container/VM a specific process belongs to.
1544 * bootchart can now store its data in the journal.
1546 * libsystemd-journal gained a new call
1547 sd_journal_add_conjunction() for AND expressions to the
1548 matching logic. This can be used to express more complex
1549 logical expressions.
1551 * journactl can now take multiple --unit= and --user-unit=
1554 * The cryptsetup logic now understands the "luks.key=" kernel
1555 command line switch for specifying a file to read the
1556 decryption key from. Also, if a configured key file is not
1557 found the tool will now automatically fall back to prompting
1560 * Python systemd.journal module was updated to wrap recently
1561 added functions from libsystemd-journal. The interface was
1562 changed to bring the low level interface in s.j._Reader
1563 closer to the C API, and the high level interface in
1564 s.j.Reader was updated to wrap and convert all data about
1567 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
1568 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
1569 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
1570 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
1571 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
1572 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1576 * journalctl --update-catalog now understands a new --root=
1577 option to operate on catalogs found in a different root
1580 * During shutdown after systemd has terminated all running
1581 services a final killing loop kills all remaining left-over
1582 processes. We will now print the name of these processes
1583 when we send SIGKILL to them, since this usually indicates a
1586 * If /etc/crypttab refers to password files stored on
1587 configured mount points automatic dependencies will now be
1588 generated to ensure the specific mount is established first
1589 before the key file is attempted to be read.
1591 * 'systemctl status' will now show information about the
1592 network sockets a socket unit is listening on.
1594 * 'systemctl status' will also shown information about any
1595 drop-in configuration file for units. (Drop-In configuration
1596 files in this context are files such as
1597 /etc/systemd/systemd/foobar.service.d/*.conf)
1599 * systemd-cgtop now optionally shows summed up CPU times of
1600 cgroups. Press '%' while running cgtop to switch between
1601 percentage and absolute mode. This is useful to determine
1602 which cgroups use up the most CPU time over the entire
1603 runtime of the system. systemd-cgtop has also been updated
1604 to be 'pipeable' for processing with further shell tools.
1606 * 'hostnamectl set-hostname' will now allow setting of FQDN
1609 * The formatting and parsing of time span values has been
1610 changed. The parser now understands fractional expressions
1611 such as "5.5h". The formatter will now output fractional
1612 expressions for all time spans under 1min, i.e. "5.123456s"
1613 rather than "5s 123ms 456us". For time spans under 1s
1614 millisecond values are shown, for those under 1ms
1615 microsecond values are shown. This should greatly improve
1616 all time-related output of systemd.
1618 * libsystemd-login and libsystemd-journal gained new
1619 functions for querying the poll() events mask and poll()
1620 timeout value for integration into arbitrary event
1623 * localectl gained the ability to list available X11 keymaps
1624 (models, layouts, variants, options).
1626 * 'systemd-analyze dot' gained the ability to filter for
1627 specific units via shell-style globs, to create smaller,
1628 more useful graphs. I.e. it is now possible to create simple
1629 graphs of all the dependencies between only target units, or
1630 of all units that Avahi has dependencies with.
1632 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
1633 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
1634 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
1635 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
1636 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
1637 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
1638 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
1642 * The boot-time readahead implementation for rotating media
1643 will now read the read-ahead data in multiple passes which
1644 consist of all read requests made in equidistant time
1645 intervals. This means instead of strictly reading read-ahead
1646 data in its physical order on disk we now try to find a
1647 middle ground between physical and access time order.
1649 * /etc/os-release files gained a new BUILD_ID= field for usage
1650 on operating systems that provide continuous builds of OS
1653 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
1654 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
1655 William Douglas, Zbigniew Jędrzejewski-Szmek
1659 * systemd-python gained an API exposing libsystemd-daemon.
1661 * The SMACK setup logic gained support for uploading CIPSO
1664 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
1665 ReadOnlyDirectories= and InaccessibleDirectories= has
1666 changed. The private /tmp and /var/tmp directories are now
1667 shared by all processes of a service (which means
1668 ExecStartPre= may now leave data in /tmp that ExecStart= of
1669 the same service can still access). When a service is
1670 stopped its temporary directories are immediately deleted
1671 (normal clean-up with tmpfiles is still done in addition to
1674 * By default, systemd will now set a couple of sysctl
1675 variables in the kernel: the safe sysrq options are turned
1676 on, IP route verification is turned on, and source routing
1677 disabled. The recently added hardlink and softlink
1678 protection of the kernel is turned on. These settings should
1679 be reasonably safe, and good defaults for all new systems.
1681 * The predictable network naming logic may now be turned off
1682 with a new kernel command line switch: net.ifnames=0.
1684 * A new libsystemd-bus module has been added that implements a
1685 pretty complete D-Bus client library. For details see:
1687 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
1689 * journald will now explicitly flush the journal files to disk
1690 at the latest 5min after each write. The file will then also
1691 be marked offline until the next write. This should increase
1692 reliability in case of a crash. The synchronization delay
1693 can be configured via SyncIntervalSec= in journald.conf.
1695 * There's a new remote-fs-setup.target unit that can be used
1696 to pull in specific services when at least one remote file
1697 system is to be mounted.
1699 * There are new targets timers.target and paths.target as
1700 canonical targets to pull user timer and path units in
1701 from. This complements sockets.target with a similar
1702 purpose for socket units.
1704 * libudev gained a new call udev_device_set_attribute_value()
1705 to set sysfs attributes of a device.
1707 * The udev daemon now sets the default number of worker
1708 processes executed in parallel based on the number of available
1709 CPUs instead of the amount of available RAM. This is supposed
1710 to provide a more reliable default and limit a too aggressive
1711 paralellism for setups with 1000s of devices connected.
1713 Contributions from: Auke Kok, Colin Walters, Cristian
1714 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
1715 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
1716 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
1717 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
1718 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
1719 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
1720 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
1721 Zbigniew Jędrzejewski-Szmek
1725 * Configuration of unit files may now be extended via drop-in
1726 files without having to edit/override the unit files
1727 themselves. More specifically, if the administrator wants to
1728 change one value for a service file foobar.service he can
1729 now do so by dropping in a configuration snippet into
1730 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
1731 will load all these snippets and apply them on top of the
1732 main unit configuration file, possibly extending or
1733 overriding its settings. Using these drop-in snippets is
1734 generally nicer than the two earlier options for changing
1735 unit files locally: copying the files from
1736 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
1737 them there; or creating a new file in /etc/systemd/system/
1738 that incorporates the original one via ".include". Drop-in
1739 snippets into these .d/ directories can be placed in any
1740 directory systemd looks for units in, and the usual
1741 overriding semantics between /usr/lib, /etc and /run apply
1744 * Most unit file settings which take lists of items can now be
1745 reset by assigning the empty string to them. For example,
1746 normally, settings such as Environment=FOO=BAR append a new
1747 environment variable assignment to the environment block,
1748 each time they are used. By assigning Environment= the empty
1749 string the environment block can be reset to empty. This is
1750 particularly useful with the .d/*.conf drop-in snippets
1751 mentioned above, since this adds the ability to reset list
1752 settings from vendor unit files via these drop-ins.
1754 * systemctl gained a new "list-dependencies" command for
1755 listing the dependencies of a unit recursively.
1757 * Inhibitors are now honored and listed by "systemctl
1758 suspend", "systemctl poweroff" (and similar) too, not only
1759 GNOME. These commands will also list active sessions by
1762 * Resource limits (as exposed by the various control group
1763 controllers) can now be controlled dynamically at runtime
1764 for all units. More specifically, you can now use a command
1765 like "systemctl set-cgroup-attr foobar.service cpu.shares
1766 2000" to alter the CPU shares a specific service gets. These
1767 settings are stored persistently on disk, and thus allow the
1768 administrator to easily adjust the resource usage of
1769 services with a few simple commands. This dynamic resource
1770 management logic is also available to other programs via the
1771 bus. Almost any kernel cgroup attribute and controller is
1774 * systemd-vconsole-setup will now copy all font settings to
1775 all allocated VTs, where it previously applied them only to
1778 * libsystemd-login gained the new sd_session_get_tty() API
1781 * This release drops support for a few legacy or
1782 distribution-specific LSB facility names when parsing init
1783 scripts: $x-display-manager, $mail-transfer-agent,
1784 $mail-transport-agent, $mail-transfer-agent, $smtp,
1785 $null. Also, the mail-transfer-agent.target unit backing
1786 this has been removed. Distributions which want to retain
1787 compatibility with this should carry the burden for
1788 supporting this themselves and patch support for these back
1789 in, if they really need to. Also, the facilities $syslog and
1790 $local_fs are now ignored, since systemd does not support
1791 early-boot LSB init scripts anymore, and these facilities
1792 are implied anyway for normal services. syslog.target has
1795 * There are new bus calls on PID1's Manager object for
1796 cancelling jobs, and removing snapshot units. Previously,
1797 both calls were only available on the Job and Snapshot
1800 * systemd-journal-gatewayd gained SSL support.
1802 * The various "environment" files, such as /etc/locale.conf
1803 now support continuation lines with a backslash ("\") as
1804 last character in the line, similarly in style (but different)
1805 to how this is supported in shells.
1807 * For normal user processes the _SYSTEMD_USER_UNIT= field is
1808 now implicitly appended to every log entry logged. systemctl
1809 has been updated to filter by this field when operating on a
1810 user systemd instance.
1812 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
1813 CAP_AUDIT_CONTROL capabilities to the capabilities set for
1814 the container. This makes it easier to boot unmodified
1815 Fedora systems in a container, which however still requires
1816 audit=0 to be passed on the kernel command line. Auditing in
1817 kernel and userspace is unfortunately still too broken in
1818 context of containers, hence we recommend compiling it out
1819 of the kernel or using audit=0. Hopefully this will be fixed
1820 one day for good in the kernel.
1822 * nspawn gained the new --bind= and --bind-ro= parameters to
1823 bind mount specific directories from the host into the
1826 * nspawn will now mount its own devpts file system instance
1827 into the container, in order not to leak pty devices from
1828 the host into the container.
1830 * systemd will now read the firmware boot time performance
1831 information from the EFI variables, if the used boot loader
1832 supports this, and takes it into account for boot performance
1833 analysis via "systemd-analyze". This is currently supported
1834 only in conjunction with Gummiboot, but could be supported
1835 by other boot loaders too. For details see:
1837 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
1839 * A new generator has been added that automatically mounts the
1840 EFI System Partition (ESP) to /boot, if that directory
1841 exists, is empty, and no other file system has been
1842 configured to be mounted there.
1844 * logind will now send out PrepareForSleep(false) out
1845 unconditionally, after coming back from suspend. This may be
1846 used by applications as asynchronous notification for
1847 system resume events.
1849 * "systemctl unlock-sessions" has been added, that allows
1850 unlocking the screens of all user sessions at once, similar
1851 to how "systemctl lock-sessions" already locked all users
1852 sessions. This is backed by a new D-Bus call UnlockSessions().
1854 * "loginctl seat-status" will now show the master device of a
1855 seat. (i.e. the device of a seat that needs to be around for
1856 the seat to be considered available, usually the graphics
1859 * tmpfiles gained a new "X" line type, that allows
1860 configuration of files and directories (with wildcards) that
1861 shall be excluded from automatic cleanup ("aging").
1863 * udev default rules set the device node permissions now only
1864 at "add" events, and do not change them any longer with a
1865 later "change" event.
1867 * The log messages for lid events and power/sleep keypresses
1868 now carry a message ID.
1870 * We now have a substantially larger unit test suite, but this
1871 continues to be work in progress.
1873 * udevadm hwdb gained a new --root= parameter to change the
1874 root directory to operate relative to.
1876 * logind will now issue a background sync() request to the kernel
1877 early at shutdown, so that dirty buffers are flushed to disk early
1878 instead of at the last moment, in order to optimize shutdown
1881 * A new bootctl tool has been added that is an interface for
1882 certain boot loader operations. This is currently a preview
1883 and is likely to be extended into a small mechanism daemon
1884 like timedated, localed, hostnamed, and can be used by
1885 graphical UIs to enumerate available boot options, and
1886 request boot into firmware operations.
1888 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
1889 the rest of the package. It also has been updated to work
1890 correctly in initrds.
1892 * Policykit previously has been runtime optional, and is now
1893 also compile time optional via a configure switch.
1895 * systemd-analyze has been reimplemented in C. Also "systemctl
1896 dot" has moved into systemd-analyze.
1898 * "systemctl status" with no further parameters will now print
1899 the status of all active or failed units.
1901 * Operations such as "systemctl start" can now be executed
1902 with a new mode "--irreversible" which may be used to queue
1903 operations that cannot accidentally be reversed by a later
1904 job queuing. This is by default used to make shutdown
1905 requests more robust.
1907 * The Python API of systemd now gained a new module for
1908 reading journal files.
1910 * A new tool kernel-install has been added that can install
1911 kernel images according to the Boot Loader Specification:
1913 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
1915 * Boot time console output has been improved to provide
1916 animated boot time output for hanging jobs.
1918 * A new tool systemd-activate has been added which can be used
1919 to test socket activation with, directly from the command
1920 line. This should make it much easier to test and debug
1921 socket activation in daemons.
1923 * journalctl gained a new "--reverse" (or -r) option to show
1924 journal output in reverse order (i.e. newest line first).
1926 * journalctl gained a new "--pager-end" (or -e) option to jump
1927 to immediately jump to the end of the journal in the
1928 pager. This is only supported in conjunction with "less".
1930 * journalctl gained a new "--user-unit=" option, that works
1931 similarly to "--unit=" but filters for user units rather than
1934 * A number of unit files to ease adoption of systemd in
1935 initrds has been added. This moves some minimal logic from
1936 the various initrd implementations into systemd proper.
1938 * The journal files are now owned by a new group
1939 "systemd-journal", which exists specifically to allow access
1940 to the journal, and nothing else. Previously, we used the
1941 "adm" group for that, which however possibly covers more
1942 than just journal/log file access. This new group is now
1943 already used by systemd-journal-gatewayd to ensure this
1944 daemon gets access to the journal files and as little else
1945 as possible. Note that "make install" will also set FS ACLs
1946 up for /var/log/journal to give "adm" and "wheel" read
1947 access to it, in addition to "systemd-journal" which owns
1948 the journal files. We recommend that packaging scripts also
1949 add read access to "adm" + "wheel" to /var/log/journal, and
1950 all existing/future journal files. To normal users and
1951 administrators little changes, however packagers need to
1952 ensure to create the "systemd-journal" system group at
1953 package installation time.
1955 * The systemd-journal-gatewayd now runs as unprivileged user
1956 systemd-journal-gateway:systemd-journal-gateway. Packaging
1957 scripts need to create these system user/group at
1960 * timedated now exposes a new boolean property CanNTP that
1961 indicates whether a local NTP service is available or not.
1963 * systemd-detect-virt will now also detect xen PVs
1965 * The pstore file system is now mounted by default, if it is
1968 * In addition to the SELinux and IMA policies we will now also
1969 load SMACK policies at early boot.
1971 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
1972 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
1973 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
1974 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
1975 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
1976 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
1977 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
1978 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
1979 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
1980 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
1981 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
1982 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
1983 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
1984 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
1988 * Timer units now support calendar time events in addition to
1989 monotonic time events. That means you can now trigger a unit
1990 based on a calendar time specification such as "Thu,Fri
1991 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
1992 or fifth day of any month of the year 2013, given that it is
1993 a thursday or friday. This brings timer event support
1994 considerably closer to cron's capabilities. For details on
1995 the supported calendar time specification language see
1998 * udev now supports a number of different naming policies for
1999 network interfaces for predictable names, and a combination
2000 of these policies is now the default. Please see this wiki
2001 document for details:
2003 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
2005 * Auke Kok's bootchart implementation has been added to the
2006 systemd tree. It is an optional component that can graph the
2007 boot in quite some detail. It is one of the best bootchart
2008 implementations around and minimal in its code and
2011 * nss-myhostname has been integrated into the systemd source
2012 tree. nss-myhostname guarantees that the local hostname
2013 always stays resolvable via NSS. It has been a weak
2014 requirement of systemd-hostnamed since a long time, and
2015 since its code is actually trivial we decided to just
2016 include it in systemd's source tree. It can be turned off
2017 with a configure switch.
2019 * The read-ahead logic is now capable of properly detecting
2020 whether a btrfs file system is on SSD or rotating media, in
2021 order to optimize the read-ahead scheme. Previously, it was
2022 only capable of detecting this on traditional file systems
2025 * In udev, additional device properties are now read from the
2026 IAB in addition to the OUI database. Also, Bluetooth company
2027 identities are attached to the devices as well.
2029 * In service files %U may be used as specifier that is
2030 replaced by the configured user name of the service.
2032 * nspawn may now be invoked without a controlling TTY. This
2033 makes it suitable for invocation as its own service. This
2034 may be used to set up a simple containerized server system
2035 using only core OS tools.
2037 * systemd and nspawn can now accept socket file descriptors
2038 when they are started for socket activation. This enables
2039 implementation of socket activated nspawn
2040 containers. i.e. think about autospawning an entire OS image
2041 when the first SSH or HTTP connection is received. We expect
2042 that similar functionality will also be added to libvirt-lxc
2045 * journalctl will now suppress ANSI color codes when
2046 presenting log data.
2048 * systemctl will no longer show control group information for
2049 a unit if a the control group is empty anyway.
2051 * logind can now automatically suspend/hibernate/shutdown the
2054 * /etc/machine-info and hostnamed now also expose the chassis
2055 type of the system. This can be used to determine whether
2056 the local system is a laptop, desktop, handset or
2057 tablet. This information may either be configured by the
2058 user/vendor or is automatically determined from ACPI and DMI
2059 information if possible.
2061 * A number of PolicyKit actions are now bound together with
2062 "imply" rules. This should simplify creating UIs because
2063 many actions will now authenticate similar ones as well.
2065 * Unit files learnt a new condition ConditionACPower= which
2066 may be used to conditionalize a unit depending on whether an
2067 AC power source is connected or not, of whether the system
2068 is running on battery power.
2070 * systemctl gained a new "is-failed" verb that may be used in
2071 shell scripts and suchlike to check whether a specific unit
2072 is in the "failed" state.
2074 * The EnvironmentFile= setting in unit files now supports file
2075 globbing, and can hence be used to easily read a number of
2076 environment files at once.
2078 * systemd will no longer detect and recognize specific
2079 distributions. All distribution-specific #ifdeffery has been
2080 removed, systemd is now fully generic and
2081 distribution-agnostic. Effectively, not too much is lost as
2082 a lot of the code is still accessible via explicit configure
2083 switches. However, support for some distribution specific
2084 legacy configuration file formats has been dropped. We
2085 recommend distributions to simply adopt the configuration
2086 files everybody else uses now and convert the old
2087 configuration from packaging scripts. Most distributions
2088 already did that. If that's not possible or desirable,
2089 distributions are welcome to forward port the specific
2090 pieces of code locally from the git history.
2092 * When logging a message about a unit systemd will now always
2093 log the unit name in the message meta data.
2095 * localectl will now also discover system locale data that is
2096 not stored in locale archives, but directly unpacked.
2098 * logind will no longer unconditionally use framebuffer
2099 devices as seat masters, i.e. as devices that are required
2100 to be existing before a seat is considered preset. Instead,
2101 it will now look for all devices that are tagged as
2102 "seat-master" in udev. By default framebuffer devices will
2103 be marked as such, but depending on local systems other
2104 devices might be marked as well. This may be used to
2105 integrate graphics cards using closed source drivers (such
2106 as NVidia ones) more nicely into logind. Note however, that
2107 we recommend using the open source NVidia drivers instead,
2108 and no udev rules for the closed-source drivers will be
2109 shipped from us upstream.
2111 Contributions from: Adam Williamson, Alessandro Crismani, Auke
2112 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
2113 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
2114 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
2115 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
2116 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
2117 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
2118 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
2119 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
2120 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
2121 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
2122 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
2127 * udev gained support for loading additional device properties
2128 from an indexed database that is keyed by vendor/product IDs
2129 and similar device identifiers. For the beginning this
2130 "hwdb" is populated with data from the well-known PCI and
2131 USB database, but also includes PNP, ACPI and OID data. In
2132 the longer run this indexed database shall grow into
2133 becoming the one central database for non-essential
2134 userspace device metadata. Previously, data from the PCI/USB
2135 database was only attached to select devices, since the
2136 lookup was a relatively expensive operation due to O(n) time
2137 complexity (with n being the number of entries in the
2138 database). Since this is now O(1), we decided to add in this
2139 data for all devices where this is available, by
2140 default. Note that the indexed database needs to be rebuilt
2141 when new data files are installed. To achieve this you need
2142 to update your packaging scripts to invoke "udevadm hwdb
2143 --update" after installation of hwdb data files. For
2144 RPM-based distributions we introduced the new
2145 %udev_hwdb_update macro for this purpose.
2147 * The Journal gained support for the "Message Catalog", an
2148 indexed database to link up additional information with
2149 journal entries. For further details please check:
2151 http://www.freedesktop.org/wiki/Software/systemd/catalog
2153 The indexed message catalog database also needs to be
2154 rebuilt after installation of message catalog files. Use
2155 "journalctl --update-catalog" for this. For RPM-based
2156 distributions we introduced the %journal_catalog_update
2157 macro for this purpose.
2159 * The Python Journal bindings gained support for the standard
2160 Python logging framework.
2162 * The Journal API gained new functions for checking whether
2163 the underlying file system of a journal file is capable of
2164 properly reporting file change notifications, or whether
2165 applications that want to reflect journal changes "live"
2166 need to recheck journal files continuously in appropriate
2169 * It is now possible to set the "age" field for tmpfiles
2170 entries to 0, indicating that files matching this entry
2171 shall always be removed when the directories are cleaned up.
2173 * coredumpctl gained a new "gdb" verb which invokes gdb
2174 right-away on the selected coredump.
2176 * There's now support for "hybrid sleep" on kernels that
2177 support this, in addition to "suspend" and "hibernate". Use
2178 "systemctl hybrid-sleep" to make use of this.
2180 * logind's HandleSuspendKey= setting (and related settings)
2181 now gained support for a new "lock" setting to simply
2182 request the screen lock on all local sessions, instead of
2183 actually executing a suspend or hibernation.
2185 * systemd will now mount the EFI variables file system by
2188 * Socket units now gained support for configuration of the
2189 SMACK security label.
2191 * timedatectl will now output the time of the last and next
2192 daylight saving change.
2194 * We dropped support for various legacy and distro-specific
2195 concepts, such as insserv, early-boot SysV services
2196 (i.e. those for non-standard runlevels such as 'b' or 'S')
2197 or ArchLinux /etc/rc.conf support. We recommend the
2198 distributions who still need support this to either continue
2199 to maintain the necessary patches downstream, or find a
2200 different solution. (Talk to us if you have questions!)
2202 * Various systemd components will now bypass PolicyKit checks
2203 for root and otherwise handle properly if PolicyKit is not
2204 found to be around. This should fix most issues for
2205 PolicyKit-less systems. Quite frankly this should have been
2206 this way since day one. It is absolutely our intention to
2207 make systemd work fine on PolicyKit-less systems, and we
2208 consider it a bug if something does not work as it should if
2209 PolicyKit is not around.
2211 * For embedded systems it is now possible to build udev and
2212 systemd without blkid and/or kmod support.
2214 * "systemctl switch-root" is now capable of switching root
2215 more than once. I.e. in addition to transitions from the
2216 initrd to the host OS it is now possible to transition to
2217 further OS images from the host. This is useful to implement
2218 offline updating tools.
2220 * Various other additions have been made to the RPM macros
2221 shipped with systemd. Use %udev_rules_update() after
2222 installing new udev rules files. %_udevhwdbdir,
2223 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
2224 %_sysctldir are now available which resolve to the right
2225 directories for packages to place various data files in.
2227 * journalctl gained the new --full switch (in addition to
2228 --all, to disable ellipsation for long messages.
2230 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
2231 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
2232 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
2233 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
2234 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
2235 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
2236 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
2237 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
2238 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
2242 * journalctl gained new --since= and --until= switches to
2243 filter by time. It also now supports nice filtering for
2244 units via --unit=/-u.
2246 * Type=oneshot services may use ExecReload= and do the
2249 * The journal daemon now supports time-based rotation and
2250 vacuuming, in addition to the usual disk-space based
2253 * The journal will now index the available field values for
2254 each field name. This enables clients to show pretty drop
2255 downs of available match values when filtering. The bash
2256 completion of journalctl has been updated
2257 accordingly. journalctl gained a new switch -F to list all
2258 values a certain field takes in the journal database.
2260 * More service events are now written as structured messages
2261 to the journal, and made recognizable via message IDs.
2263 * The timedated, localed and hostnamed mini-services which
2264 previously only provided support for changing time, locale
2265 and hostname settings from graphical DEs such as GNOME now
2266 also have a minimal (but very useful) text-based client
2267 utility each. This is probably the nicest way to changing
2268 these settings from the command line now, especially since
2269 it lists available options and is fully integrated with bash
2272 * There's now a new tool "systemd-coredumpctl" to list and
2273 extract coredumps from the journal.
2275 * We now install a README each in /var/log/ and
2276 /etc/rc.d/init.d explaining where the system logs and init
2277 scripts went. This hopefully should help folks who go to
2278 that dirs and look into the otherwise now empty void and
2279 scratch their heads.
2281 * When user-services are invoked (by systemd --user) the
2282 $MANAGERPID env var is set to the PID of systemd.
2284 * SIGRTMIN+24 when sent to a --user instance will now result
2285 in immediate termination of systemd.
2287 * gatewayd received numerous feature additions such as a
2288 "follow" mode, for live syncing and filtering.
2290 * browse.html now allows filtering and showing detailed
2291 information on specific entries. Keyboard navigation and
2292 mouse screen support has been added.
2294 * gatewayd/journalctl now supports HTML5/JSON
2295 Server-Sent-Events as output.
2297 * The SysV init script compatibility logic will now
2298 heuristically determine whether a script supports the
2299 "reload" verb, and only then make this available as
2302 * "systemctl status --follow" has been removed, use "journalctl
2305 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
2306 have been removed since they are hardly useful to be
2309 * And I'd like to take the opportunity to specifically mention
2310 Zbigniew for his great contributions. Zbigniew, you rock!
2312 Contributions from: Andrew Eikum, Christian Hesse, Colin
2313 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
2314 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
2315 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
2316 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
2317 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
2318 Jędrzejewski-Szmek, Сковорода Никита Андреевич
2322 * If /etc/vconsole.conf is non-existent or empty we will no
2323 longer load any console font or key map at boot by
2324 default. Instead the kernel defaults will be left
2325 intact. This is definitely the right thing to do, as no
2326 configuration should mean no configuration, and hard-coding
2327 font names that are different on all archs is probably a bad
2328 idea. Also, the kernel default key map and font should be
2329 good enough for most cases anyway, and mostly identical to
2330 the userspace fonts/key maps we previously overloaded them
2331 with. If distributions want to continue to default to a
2332 non-kernel font or key map they should ship a default
2333 /etc/vconsole.conf with the appropriate contents.
2335 Contributions from: Colin Walters, Daniel J Walsh, Dave
2336 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
2337 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
2341 * journalctl gained a new --cursor= switch to show entries
2342 starting from the specified location in the journal.
2344 * We now enforce a size limit on journal entry fields exported
2345 with "-o json" in journalctl. Fields larger than 4K will be
2346 assigned null. This can be turned off with --all.
2348 * An (optional) journal gateway daemon is now available as
2349 "systemd-journal-gatewayd.service". This service provides
2350 access to the journal via HTTP and JSON. This functionality
2351 will be used to implement live log synchronization in both
2352 pull and push modes, but has various other users too, such
2353 as easy log access for debugging of embedded devices. Right
2354 now it is already useful to retrieve the journal via HTTP:
2356 # systemctl start systemd-journal-gatewayd.service
2357 # wget http://localhost:19531/entries
2359 This will download the journal contents in a
2360 /var/log/messages compatible format. The same as JSON:
2362 # curl -H"Accept: application/json" http://localhost:19531/entries
2364 This service is also accessible via a web browser where a
2365 single static HTML5 app is served that uses the JSON logic
2366 to enable the user to do some basic browsing of the
2367 journal. This will be extended later on. Here's an example
2368 screenshot of this app in its current state:
2370 http://0pointer.de/public/journal-gatewayd
2372 Contributions from: Kay Sievers, Lennart Poettering, Robert
2373 Milasan, Tom Gundersen
2377 * The bash completion logic is now available for journalctl
2380 * We do not mount the "cpuset" controller anymore together with
2381 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
2382 started if no parameters are assigned to it. "cpuset" hence
2383 broke code that assumed it it could create "cpu" groups and
2386 * journalctl -f will now subscribe to terminal size changes,
2387 and line break accordingly.
2389 Contributions from: Dave Reisner, Kay Sievers, Lennart
2390 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
2394 * nspawn will now create a symlink /etc/localtime in the
2395 container environment, copying the host's timezone
2396 setting. Previously this has been done via a bind mount, but
2397 since symlinks cannot be bind mounted this has now been
2398 changed to create/update the appropriate symlink.
2400 * journalctl -n's line number argument is now optional, and
2401 will default to 10 if omitted.
2403 * journald will now log the maximum size the journal files may
2404 take up on disk. This is particularly useful if the default
2405 built-in logic of determining this parameter from the file
2406 system size is used. Use "systemctl status
2407 systemd-journald.service" to see this information.
2409 * The multi-seat X wrapper tool has been stripped down. As X
2410 is now capable of enumerating graphics devices via udev in a
2411 seat-aware way the wrapper is not strictly necessary
2412 anymore. A stripped down temporary stop-gap is still shipped
2413 until the upstream display managers have been updated to
2414 fully support the new X logic. Expect this wrapper to be
2415 removed entirely in one of the next releases.
2417 * HandleSleepKey= in logind.conf has been split up into
2418 HandleSuspendKey= and HandleHibernateKey=. The old setting
2419 is not available anymore. X11 and the kernel are
2420 distuingishing between these keys and we should too. This
2421 also means the inhibition lock for these keys has been split
2424 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
2425 Poettering, Lukas Nykryn, Václav Pavlín
2429 * Whenever a unit changes state we will now log this to the
2430 journal and show along the unit's own log output in
2433 * ConditionPathIsMountPoint= can now properly detect bind
2434 mount points too. (Previously, a bind mount of one file
2435 system to another place in the same file system could not be
2436 detected as mount, since they shared struct stat's st_dev
2439 * We will now mount the cgroup controllers cpu, cpuacct,
2440 cpuset and the controllers net_cls, net_prio together by
2443 * nspawn containers will now have a virtualized boot
2444 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
2445 over with a randomized ID at container initialization). This
2446 has the effect of making "journalctl -b" do the right thing
2449 * The JSON output journal serialization has been updated not
2450 to generate "endless" list objects anymore, but rather one
2451 JSON object per line. This is more in line how most JSON
2452 parsers expect JSON objects. The new output mode
2453 "json-pretty" has been added to provide similar output, but
2454 neatly aligned for readability by humans.
2456 * We dropped all explicit sync() invocations in the shutdown
2457 code. The kernel does this implicitly anyway in the kernel
2458 reboot() syscall. halt(8)'s -n option is now a compatibility
2461 * We now support virtualized reboot() in containers, as
2462 supported by newer kernels. We will fall back to exit() if
2463 CAP_SYS_REBOOT is not available to the container. Also,
2464 nspawn makes use of this now and will actually reboot the
2465 container if the containerized OS asks for that.
2467 * journalctl will only show local log output by default
2468 now. Use --merge (-m) to show remote log output, too.
2470 * libsystemd-journal gained the new sd_journal_get_usage()
2471 call to determine the current disk usage of all journal
2472 files. This is exposed in the new "journalctl --disk-usage"
2475 * journald gained a new configuration setting SplitMode= in
2476 journald.conf which may be used to control how user journals
2477 are split off. See journald.conf(5) for details.
2479 * A new condition type ConditionFileNotEmpty= has been added.
2481 * tmpfiles' "w" lines now support file globbing, to write
2482 multiple files at once.
2484 * We added Python bindings for the journal submission
2485 APIs. More Python APIs for a number of selected APIs will
2486 likely follow. Note that we intend to add native bindings
2487 only for the Python language, as we consider it common
2488 enough to deserve bindings shipped within systemd. There are
2489 various projects outside of systemd that provide bindings
2490 for languages such as PHP or Lua.
2492 * Many conditions will now resolve specifiers such as %i. In
2493 addition, PathChanged= and related directives of .path units
2494 now support specifiers as well.
2496 * There's now a new RPM macro definition for the system preset
2499 * journald will now warn if it ca not forward a message to the
2500 syslog daemon because its socket is full.
2502 * timedated will no longer write or process /etc/timezone,
2503 except on Debian. As we do not support late mounted /usr
2504 anymore /etc/localtime always being a symlink is now safe,
2505 and hence the information in /etc/timezone is not necessary
2508 * logind will now always reserve one VT for a text getty (VT6
2509 by default). Previously if more than 6 X sessions where
2510 started they took up all the VTs with auto-spawned gettys,
2511 so that no text gettys were available anymore.
2513 * udev will now automatically inform the btrfs kernel logic
2514 about btrfs RAID components showing up. This should make
2515 simple hotplug based btrfs RAID assembly work.
2517 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
2518 (but not for its children which will stay at the kernel
2519 default). This should allow setups with a lot more listening
2522 * systemd will now always pass the configured timezone to the
2523 kernel at boot. timedated will do the same when the timezone
2526 * logind's inhibition logic has been updated. By default,
2527 logind will now handle the lid switch, the power and sleep
2528 keys all the time, even in graphical sessions. If DEs want
2529 to handle these events on their own they should take the new
2530 handle-power-key, handle-sleep-key and handle-lid-switch
2531 inhibitors during their runtime. A simple way to achiveve
2532 that is to invoke the DE wrapped in an invocation of:
2534 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
2536 * Access to unit operations is now checked via SELinux taking
2537 the unit file label and client process label into account.
2539 * systemd will now notify the administrator in the journal
2540 when he over-mounts a non-empty directory.
2542 * There are new specifiers that are resolved in unit files,
2543 for the host name (%H), the machine ID (%m) and the boot ID
2546 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
2547 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
2548 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
2549 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
2550 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
2551 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
2552 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
2556 * Support for reading structured kernel messages from
2557 /dev/kmsg has now been added and is enabled by default.
2559 * Support for reading kernel messages from /proc/kmsg has now
2560 been removed. If you want kernel messages in the journal
2561 make sure to run a recent kernel (>= 3.5) that supports
2562 reading structured messages from /dev/kmsg (see
2563 above). /proc/kmsg is now exclusive property of classic
2564 syslog daemons again.
2566 * The libudev API gained the new
2567 udev_device_new_from_device_id() call.
2569 * The logic for file system namespace (ReadOnlyDirectory=,
2570 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
2571 require pivot_root() anymore. This means fewer temporary
2572 directories are created below /tmp for this feature.
2574 * nspawn containers will now see and receive all submounts
2575 made on the host OS below the root file system of the
2578 * Forward Secure Sealing is now supported for Journal files,
2579 which provide cryptographical sealing of journal files so
2580 that attackers cannot alter log history anymore without this
2581 being detectable. Lennart will soon post a blog story about
2582 this explaining it in more detail.
2584 * There are two new service settings RestartPreventExitStatus=
2585 and SuccessExitStatus= which allow configuration of exit
2586 status (exit code or signal) which will be excepted from the
2587 restart logic, resp. consider successful.
2589 * journalctl gained the new --verify switch that can be used
2590 to check the integrity of the structure of journal files and
2591 (if Forward Secure Sealing is enabled) the contents of
2594 * nspawn containers will now be run with /dev/stdin, /dev/fd/
2595 and similar symlinks pre-created. This makes running shells
2596 as container init process a lot more fun.
2598 * The fstab support can now handle PARTUUID= and PARTLABEL=
2601 * A new ConditionHost= condition has been added to match
2602 against the hostname (with globs) and machine ID. This is
2603 useful for clusters where a single OS image is used to
2604 provision a large number of hosts which shall run slightly
2605 different sets of services.
2607 * Services which hit the restart limit will now be placed in a
2610 Contributions from: Bertram Poettering, Dave Reisner, Huang
2611 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
2612 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
2616 * When running in --user mode systemd will now become a
2617 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
2618 tree a lot more organized.
2620 * A new PartOf= unit dependency type has been introduced that
2621 may be used to group services in a natural way.
2623 * "systemctl enable" may now be used to enable instances of
2626 * journalctl now prints error log levels in red, and
2627 warning/notice log levels in bright white. It also supports
2628 filtering by log level now.
2630 * cgtop gained a new -n switch (similar to top), to configure
2631 the maximum number of iterations to run for. It also gained
2632 -b, to run in batch mode (accepting no input).
2634 * The suffix ".service" may now be omitted on most systemctl
2635 command lines involving service unit names.
2637 * There's a new bus call in logind to lock all sessions, as
2638 well as a loginctl verb for it "lock-sessions".
2640 * libsystemd-logind.so gained a new call sd_journal_perror()
2641 that works similar to libc perror() but logs to the journal
2642 and encodes structured information about the error number.
2644 * /etc/crypttab entries now understand the new keyfile-size=
2647 * shutdown(8) now can send a (configurable) wall message when
2648 a shutdown is cancelled.
2650 * The mount propagation mode for the root file system will now
2651 default to "shared", which is useful to make containers work
2652 nicely out-of-the-box so that they receive new mounts from
2653 the host. This can be undone locally by running "mount
2654 --make-rprivate /" if needed.
2656 * The prefdm.service file has been removed. Distributions
2657 should maintain this unit downstream if they intend to keep
2658 it around. However, we recommend writing normal unit files
2659 for display managers instead.
2661 * Since systemd is a crucial part of the OS we will now
2662 default to a number of compiler switches that improve
2663 security (hardening) such as read-only relocations, stack
2664 protection, and suchlike.
2666 * The TimeoutSec= setting for services is now split into
2667 TimeoutStartSec= and TimeoutStopSec= to allow configuration
2668 of individual time outs for the start and the stop phase of
2671 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
2672 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
2673 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
2674 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
2675 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
2676 Gundersen, Zbigniew Jędrzejewski-Szmek
2680 * The journal and id128 C APIs are now fully documented as man
2683 * Extra safety checks have been added when transitioning from
2684 the initial RAM disk to the main system to avoid accidental
2687 * /etc/crypttab entries now understand the new keyfile-offset=
2690 * systemctl -t can now be used to filter by unit load state.
2692 * The journal C API gained the new sd_journal_wait() call to
2693 make writing synchronous journal clients easier.
2695 * journalctl gained the new -D switch to show journals from a
2698 * journalctl now displays a special marker between log
2699 messages of two different boots.
2701 * The journal is now explicitly flushed to /var via a service
2702 systemd-journal-flush.service, rather than implicitly simply
2703 by seeing /var/log/journal to be writable.
2705 * journalctl (and the journal C APIs) can now match for much
2706 more complex expressions, with alternatives and
2709 * When transitioning from the initial RAM disk to the main
2710 system we will now kill all processes in a killing spree to
2711 ensure no processes stay around by accident.
2713 * Three new specifiers may be used in unit files: %u, %h, %s
2714 resolve to the user name, user home directory resp. user
2715 shell. This is useful for running systemd user instances.
2717 * We now automatically rotate journal files if their data
2718 object hash table gets a fill level > 75%. We also size the
2719 hash table based on the configured maximum file size. This
2720 together should lower hash collisions drastically and thus
2721 speed things up a bit.
2723 * journalctl gained the new "--header" switch to introspect
2724 header data of journal files.
2726 * A new setting SystemCallFilters= has been added to services
2727 which may be used to apply blacklists or whitelists to
2728 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
2730 * nspawn gained a new --link-journal= switch (and quicker: -j)
2731 to link the container journal with the host. This makes it
2732 very easy to centralize log viewing on the host for all
2733 guests while still keeping the journal files separated.
2735 * Many bugfixes and optimizations
2737 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
2738 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
2739 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
2744 * Several tools now understand kernel command line arguments,
2745 which are only read when run in an initial RAM disk. They
2746 usually follow closely their normal counterparts, but are
2749 * There's a new tool to analyze the readahead files that are
2750 automatically generated at boot. Use:
2752 /usr/lib/systemd/systemd-readahead analyze /.readahead
2754 * We now provide an early debug shell on tty9 if this enabled. Use:
2756 systemctl enable debug-shell.service
2758 * All plymouth related units have been moved into the Plymouth
2759 package. Please make sure to upgrade your Plymouth version
2762 * systemd-tmpfiles now supports getting passed the basename of
2763 a configuration file only, in which case it will look for it
2764 in all appropriate directories automatically.
2766 * udevadm info now takes a /dev or /sys path as argument, and
2767 does the right thing. Example:
2769 udevadm info /dev/sda
2770 udevadm info /sys/class/block/sda
2772 * systemctl now prints a warning if a unit is stopped but a
2773 unit that might trigger it continues to run. Example: a
2774 service is stopped but the socket that activates it is left
2777 * "systemctl status" will now mention if the log output was
2778 shortened due to rotation since a service has been started.
2780 * The journal API now exposes functions to determine the
2781 "cutoff" times due to rotation.
2783 * journald now understands SIGUSR1 and SIGUSR2 for triggering
2784 immediately flushing of runtime logs to /var if possible,
2785 resp. for triggering immediate rotation of the journal
2788 * It is now considered an error if a service is attempted to
2789 be stopped that is not loaded.
2791 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
2793 * systemd-analyze now supports Python 3
2795 * tmpfiles now supports cleaning up directories via aging
2796 where the first level dirs are always kept around but
2797 directories beneath it automatically aged. This is enabled
2798 by prefixing the age field with '~'.
2800 * Seat objects now expose CanGraphical, CanTTY properties
2801 which is required to deal with very fast bootups where the
2802 display manager might be running before the graphics drivers
2803 completed initialization.
2805 * Seat objects now expose a State property.
2807 * We now include RPM macros for service enabling/disabling
2808 based on the preset logic. We recommend RPM based
2809 distributions to make use of these macros if possible. This
2810 makes it simpler to reuse RPM spec files across
2813 * We now make sure that the collected systemd unit name is
2814 always valid when services log to the journal via
2817 * There's a new man page kernel-command-line(7) detailing all
2818 command line options we understand.
2820 * The fstab generator may now be disabled at boot by passing
2821 fstab=0 on the kernel command line.
2823 * A new kernel command line option modules-load= is now understood
2824 to load a specific kernel module statically, early at boot.
2826 * Unit names specified on the systemctl command line are now
2827 automatically escaped as needed. Also, if file system or
2828 device paths are specified they are automatically turned
2829 into the appropriate mount or device unit names. Example:
2831 systemctl status /home
2832 systemctl status /dev/sda
2834 * The SysVConsole= configuration option has been removed from
2835 system.conf parsing.
2837 * The SysV search path is no longer exported on the D-Bus
2840 * The Names= option is been removed from unit file parsing.
2842 * There's a new man page bootup(7) detailing the boot process.
2844 * Every unit and every generator we ship with systemd now
2845 comes with full documentation. The self-explanatory boot is
2848 * A couple of services gained "systemd-" prefixes in their
2849 name if they wrap systemd code, rather than only external
2850 code. Among them fsck@.service which is now
2851 systemd-fsck@.service.
2853 * The HaveWatchdog property has been removed from the D-Bus
2856 * systemd.confirm_spawn= on the kernel command line should now
2859 * There's a new man page crypttab(5) which details all options
2860 we actually understand.
2862 * systemd-nspawn gained a new --capability= switch to pass
2863 additional capabilities to the container.
2865 * timedated will now read known NTP implementation unit names
2866 from /usr/lib/systemd/ntp-units.d/*.list,
2867 systemd-timedated-ntp.target has been removed.
2869 * journalctl gained a new switch "-b" that lists log data of
2870 the current boot only.
2872 * The notify socket is in the abstract namespace again, in
2873 order to support daemons which chroot() at start-up.
2875 * There is a new Storage= configuration option for journald
2876 which allows configuration of where log data should go. This
2877 also provides a way to disable journal logging entirely, so
2878 that data collected is only forwarded to the console, the
2879 kernel log buffer or another syslog implementation.
2881 * Many bugfixes and optimizations
2883 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
2884 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
2885 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
2886 Shawn Landden, Tom Gundersen
2890 * "systemctl help <unit>" now shows the man page if one is
2893 * Several new man pages have been added.
2895 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
2896 MaxLevelConsole= can now be specified in
2897 journald.conf. These options allow reducing the amount of
2898 data stored on disk or forwarded by the log level.
2900 * TimerSlackNSec= can now be specified in system.conf for
2901 PID1. This allows system-wide power savings.
2903 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
2904 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
2909 * logind is now capable of (optionally) handling power and
2910 sleep keys as well as the lid switch.
2912 * journalctl now understands the syntax "journalctl
2913 /usr/bin/avahi-daemon" to get all log output of a specific
2916 * CapabilityBoundingSet= in system.conf now also influences
2917 the capability bound set of usermode helpers of the kernel.
2919 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
2920 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
2921 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
2922 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
2926 * Note that we skipped 139 releases here in order to set the
2927 new version to something that is greater than both udev's
2928 and systemd's most recent version number.
2930 * udev: all udev sources are merged into the systemd source tree now.
2931 All future udev development will happen in the systemd tree. It
2932 is still fully supported to use the udev daemon and tools without
2933 systemd running, like in initramfs or other init systems. Building
2934 udev though, will require the *build* of the systemd tree, but
2935 udev can be properly *run* without systemd.
2937 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
2938 should be used to create dead device nodes as workarounds for broken
2941 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
2942 no longer supported. udev_monitor_new_from_netlink() needs to be
2943 used to subscribe to events.
2945 * udev: when udevd is started by systemd, processes which are left
2946 behind by forking them off of udev rules, are unconditionally cleaned
2947 up and killed now after the event handling has finished. Services or
2948 daemons must be started as systemd services. Services can be
2949 pulled-in by udev to get started, but they can no longer be directly
2950 forked by udev rules.
2952 * udev: the daemon binary is called systemd-udevd now and installed
2953 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
2954 to adapt to that, create symlink, or rename the binary after building
2957 * libudev no longer provides these symbols:
2958 udev_monitor_from_socket()
2959 udev_queue_get_failed_list_entry()
2960 udev_get_{dev,sys,run}_path()
2961 The versions number was bumped and symbol versioning introduced.
2963 * systemd-loginctl and systemd-journalctl have been renamed
2964 to loginctl and journalctl to match systemctl.
2966 * The config files: /etc/systemd/systemd-logind.conf and
2967 /etc/systemd/systemd-journald.conf have been renamed to
2968 logind.conf and journald.conf. Package updates should rename
2969 the files to the new names on upgrade.
2971 * For almost all files the license is now LGPL2.1+, changed
2972 from the previous GPL2.0+. Exceptions are some minor stuff
2973 of udev (which will be changed to LGPL2.1 eventually, too),
2974 and the MIT licensed sd-daemon.[ch] library that is suitable
2975 to be used as drop-in files.
2977 * systemd and logind now handle system sleep states, in
2978 particular suspending and hibernating.
2980 * logind now implements a sleep/shutdown/idle inhibiting logic
2981 suitable for a variety of uses. Soonishly Lennart will blog
2982 about this in more detail.
2984 * var-run.mount and var-lock.mount are no longer provided
2985 (which prevously bind mounted these directories to their new
2986 places). Distributions which have not converted these
2987 directories to symlinks should consider stealing these files
2988 from git history and add them downstream.
2990 * We introduced the Documentation= field for units and added
2991 this to all our shipped units. This is useful to make it
2992 easier to explore the boot and the purpose of the various
2995 * All smaller setup units (such as
2996 systemd-vconsole-setup.service) now detect properly if they
2997 are run in a container and are skipped when
2998 appropriate. This guarantees an entirely noise-free boot in
2999 Linux container environments such as systemd-nspawn.
3001 * A framework for implementing offline system updates is now
3002 integrated, for details see:
3003 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
3005 * A new service type Type=idle is available now which helps us
3006 avoiding ugly interleaving of getty output and boot status
3009 * There's now a system-wide CapabilityBoundingSet= option to
3010 globally reduce the set of capabilities for the
3011 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
3012 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
3013 even CAP_NET_ADMIN system-wide for secure systems.
3015 * There are now system-wide DefaultLimitXXX= options to
3016 globally change the defaults of the various resource limits
3017 for all units started by PID 1.
3019 * Harald Hoyer's systemd test suite has been integrated into
3020 systemd which allows easy testing of systemd builds in qemu
3021 and nspawn. (This is really awesome! Ask us for details!)
3023 * The fstab parser is now implemented as generator, not inside
3026 * systemctl will now warn you if .mount units generated from
3027 /etc/fstab are out of date due to changes in fstab that
3028 have not been read by systemd yet.
3030 * systemd is now suitable for usage in initrds. Dracut has
3031 already been updated to make use of this. With this in place
3032 initrds get a slight bit faster but primarily are much
3033 easier to introspect and debug since "systemctl status" in
3034 the host system can be used to introspect initrd services,
3035 and the journal from the initrd is kept around too.
3037 * systemd-delta has been added, a tool to explore differences
3038 between user/admin configuration and vendor defaults.
3040 * PrivateTmp= now affects both /tmp and /var/tmp.
3042 * Boot time status messages are now much prettier and feature
3043 proper english language. Booting up systemd has never been
3046 * Read-ahead pack files now include the inode number of all
3047 files to pre-cache. When the inode changes the pre-caching
3048 is not attempted. This should be nicer to deal with updated
3049 packages which might result in changes of read-ahead
3052 * We now temporaritly lower the kernel's read_ahead_kb variable
3053 when collecting read-ahead data to ensure the kernel's
3054 built-in read-ahead does not add noise to our measurements
3055 of necessary blocks to pre-cache.
3057 * There's now RequiresMountsFor= to add automatic dependencies
3058 for all mounts necessary for a specific file system path.
3060 * MountAuto= and SwapAuto= have been removed from
3061 system.conf. Mounting file systems at boot has to take place
3064 * nspawn now learned a new switch --uuid= to set the machine
3065 ID on the command line.
3067 * nspawn now learned the -b switch to automatically search
3070 * vt102 is now the default TERM for serial TTYs, upgraded from
3073 * systemd-logind now works on VT-less systems.
3075 * The build tree has been reorganized. The individual
3076 components now have directories of their own.
3078 * A new condition type ConditionPathIsReadWrite= is now available.
3080 * nspawn learned the new -C switch to create cgroups for the
3081 container in other hierarchies.
3083 * We now have support for hardware watchdogs, configurable in
3086 * The scheduled shutdown logic now has a public API.
3088 * We now mount /tmp as tmpfs by default, but this can be
3089 masked and /etc/fstab can override it.
3091 * Since udisks does not make use of /media anymore we are not
3092 mounting a tmpfs on it anymore.
3094 * journalctl gained a new --local switch to only interleave
3095 locally generated journal files.
3097 * We can now load the IMA policy at boot automatically.
3099 * The GTK tools have been split off into a systemd-ui.
3101 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
3102 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
3103 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
3104 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
3105 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
3106 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
3107 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
3108 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
3109 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
3114 * This is mostly a bugfix release
3116 * Support optional initialization of the machine ID from the
3117 KVM or container configured UUID.
3119 * Support immediate reboots with "systemctl reboot -ff"
3121 * Show /etc/os-release data in systemd-analyze output
3123 * Many bugfixes for the journal, including endianness fixes and
3124 ensuring that disk space enforcement works
3126 * sd-login.h is C++ comptaible again
3128 * Extend the /etc/os-release format on request of the Debian
3131 * We now refuse non-UTF8 strings used in various configuration
3132 and unit files. This is done to ensure we do not pass invalid
3133 data over D-Bus or expose it elsewhere.
3135 * Register Mimo USB Screens as suitable for automatic seat
3138 * Read SELinux client context from journal clients in a race
3141 * Reorder configuration file lookup order. /etc now always
3142 overrides /run in order to allow the administrator to always
3143 and unconditionally override vendor supplied or
3144 automatically generated data.
3146 * The various user visible bits of the journal now have man
3147 pages. We still lack man pages for the journal API calls
3150 * We now ship all man pages in HTML format again in the
3153 Contributions from: Dave Reisner, Dirk Eibach, Frederic
3154 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
3155 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
3160 * This is mostly a bugfix release
3162 * systems lacking /etc/os-release are no longer supported.
3164 * Various functionality updates to libsystemd-login.so
3166 * Track class of PAM logins to distuingish greeters from
3169 Contributions from: Kay Sievers, Lennart Poettering, Michael
3174 * This is an important bugfix release for v41.
3176 * Building man pages is now optional which should be useful
3177 for those building systemd from git but unwilling to install
3180 * Watchdog support for supervising services is now usable. In
3181 a future release support for hardware watchdogs
3182 (i.e. /dev/watchdog) will be added building on this.
3184 * Service start rate limiting is now configurable and can be
3185 turned off per service. When a start rate limit is hit a
3186 reboot can automatically be triggered.
3188 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
3190 Contributions from: Benjamin Franzke, Bill Nottingham,
3191 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
3192 Schmidt, Michał Górny, Piotr Drąg
3196 * The systemd binary is installed /usr/lib/systemd/systemd now;
3197 An existing /sbin/init symlink needs to be adapted with the
3200 * The code that loads kernel modules has been ported to invoke
3201 libkmod directly, instead of modprobe. This means we do not
3202 support systems with module-init-tools anymore.
3204 * Watchdog support is now already useful, but still not
3207 * A new kernel command line option systemd.setenv= is
3208 understood to set system wide environment variables
3209 dynamically at boot.
3211 * We now limit the set of capabilities of systemd-journald.
3213 * We now set SIGPIPE to ignore by default, since it only is
3214 useful in shell pipelines, and has little use in general
3215 code. This can be disabled with IgnoreSIPIPE=no in unit
3218 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
3219 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
3224 * This is mostly a bugfix release
3226 * We now expose the reason why a service failed in the
3227 "Result" D-Bus property.
3229 * Rudimentary service watchdog support (will be completed over
3230 the next few releases.)
3232 * When systemd forks off in order execute some service we will
3233 now immediately changes its argv[0] to reflect which process
3234 it will execute. This is useful to minimize the time window
3235 with a generic argv[0], which makes bootcharts more useful
3237 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
3238 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
3239 Mike Kazantsev, Ray Strode
3243 * This is mostly a test release, but incorporates many
3246 * New systemd-cgtop tool to show control groups by their
3249 * Linking against libacl for ACLs is optional again. If
3250 disabled, support tracking device access for active logins
3251 goes becomes unavailable, and so does access to the user
3252 journals by the respective users.
3254 * If a group "adm" exists, journal files are automatically
3255 owned by them, thus allow members of this group full access
3256 to the system journal as well as all user journals.
3258 * The journal now stores the SELinux context of the logging
3259 client for all entries.
3261 * Add C++ inclusion guards to all public headers
3263 * New output mode "cat" in the journal to print only text
3264 messages, without any meta data like date or time.
3266 * Include tiny X server wrapper as a temporary stop-gap to
3267 teach XOrg udev display enumeration. This is used by display
3268 managers such as gdm, and will go away as soon as XOrg
3269 learned native udev hotplugging for display devices.
3271 * Add new systemd-cat tool for executing arbitrary programs
3272 with STDERR/STDOUT connected to the journal. Can also act as
3273 BSD logger replacement, and does so by default.
3275 * Optionally store all locally generated coredumps in the
3276 journal along with meta data.
3278 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
3279 writing short strings to files (for usage for /sys), and for
3280 creating symlinks, character and block device nodes.
3282 * New unit file option ControlGroupPersistent= to make cgroups
3283 persistent, following the mechanisms outlined in
3284 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
3286 * Support multiple local RTCs in a sane way
3288 * No longer monopolize IO when replaying readahead data on
3289 rotating disks, since we might starve non-file-system IO to
3290 death, since fanotify() will not see accesses done by blkid,
3293 * Do not show kernel threads in systemd-cgls anymore, unless
3294 requested with new -k switch.
3296 Contributions from: Dan Horák, Kay Sievers, Lennart
3297 Poettering, Michal Schmidt
3301 * This is mostly a test release, but incorporates many
3304 * The git repository moved to:
3305 git://anongit.freedesktop.org/systemd/systemd
3306 ssh://git.freedesktop.org/git/systemd/systemd
3308 * First release with the journal
3309 http://0pointer.de/blog/projects/the-journal.html
3311 * The journal replaces both systemd-kmsg-syslogd and
3312 systemd-stdout-bridge.
3314 * New sd_pid_get_unit() API call in libsystemd-logind
3316 * Many systemadm clean-ups
3318 * Introduce remote-fs-pre.target which is ordered before all
3319 remote mounts and may be used to start services before all
3322 * Added Mageia support
3324 * Add bash completion for systemd-loginctl
3326 * Actively monitor PID file creation for daemons which exit in
3327 the parent process before having finished writing the PID
3328 file in the daemon process. Daemons which do this need to be
3329 fixed (i.e. PID file creation must have finished before the
3330 parent exits), but we now react a bit more gracefully to them.
3332 * Add colourful boot output, mimicking the well-known output
3333 of existing distributions.
3335 * New option PassCredentials= for socket units, for
3336 compatibility with a recent kernel ABI breakage.
3338 * /etc/rc.local is now hooked in via a generator binary, and
3339 thus will no longer act as synchronization point during
3342 * systemctl list-unit-files now supports --root=.
3344 * systemd-tmpfiles now understands two new commands: z, Z for
3345 relabelling files according to the SELinux database. This is
3346 useful to apply SELinux labels to specific files in /sys,
3349 * Output of SysV services is now forwarded to both the console
3350 and the journal by default, not only just the console.
3352 * New man pages for all APIs from libsystemd-login.
3354 * The build tree got reorganized and a the build system is a
3355 lot more modular allowing embedded setups to specifically
3356 select the components of systemd they are interested in.
3358 * Support for Linux systems lacking the kernel VT subsystem is
3361 * configure's --with-rootdir= got renamed to
3362 --with-rootprefix= to follow the naming used by udev and
3365 * Unless specified otherwise we will now install to /usr instead
3366 of /usr/local by default.
3368 * Processes with '@' in argv[0][0] are now excluded from the
3369 final shut-down killing spree, following the logic explained
3371 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
3373 * All processes remaining in a service cgroup when we enter
3374 the START or START_PRE states are now killed with
3375 SIGKILL. That means it is no longer possible to spawn
3376 background processes from ExecStart= lines (which was never
3377 supported anyway, and bad style).
3379 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
3380 reloading of units together.
3382 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
3383 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
3384 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
3385 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
3386 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek