1 systemd System and Service Manager
5 * A new "systemd-timesyncd" daemon has been added for
6 synchronizing the system clock across the network. It
7 implements an SNTP client. In contrast to NTP
8 implementations such as chrony or the NTP reference server
9 this only implements a client side, and does not bother with
10 the full NTP complexity, focusing only on querying time from
11 one remote server and synchronizing the local clock to
12 it. Unless you intend to serve NTP to networked clients or
13 want to connect to local hardware clocks this simple NTP
14 client should be more than appropriate for most
15 installations. The daemon runs with minimal privileges, and
16 has been hooked up with networkd to only operate when
17 network connectivity is available. The daemon saves the
18 current clock to disk every time a new NTP sync has been
19 acquired, and uses this to possibly correct the system clock
20 early at bootup, in order to accommodate for systems that
21 lack an RTC such as the Raspberry Pi and embedded devices,
22 and make sure that time monotonically progresses on these
23 systems, even if it is not always correct. To make use of
24 this daemon a new system user and group "systemd-timesync"
25 needs to be created on installation of systemd.
27 * The queue "seqnum" interface of libudev has been disabled, as
28 it was generally incompatible with device namespacing as
29 sequence numbers of devices go "missing" if the devices are
30 part of a different namespace.
32 * "systemctl list-timers" and "systemctl list-sockets" gained
33 a --recursive switch for showing units of these types also
34 for all local containers, similar in style to the already
35 supported --recursive switch for "systemctl list-units".
37 * A new RebootArgument= setting has been added for service
38 units, which may be used to specify a kernel reboot argument
39 to use when triggering reboots with StartLimitAction=.
41 * A new FailureAction= setting has been added for service
42 units which may be used to specify an operation to trigger
43 when a service fails. This works similarly to
44 StartLimitAction=, but unlike it controls what is done
45 immediately rather than only after several attempts to
46 restart the service in question.
48 * hostnamed got updated to also expose the kernel name,
49 release, and version on the bus. This is useful for
50 executing commands like hostnamectl with the -H switch.
51 systemd-analyze makes use of this to properly display
52 details when running non-locally.
54 * The bootchart tool can now show cgroup information in the
57 * The CFS CPU quota cgroup attribute is now exposed for
58 services. The new CPUQuota= switch has been added for this
59 which takes a percentage value. Setting this will have the
60 result that a service may never get more CPU time than the
61 specified percentage, even if the machine is otherwise idle.
63 * systemd-networkd learned IPIP and SIT tunnel support.
65 * LSB init scripts exposing a dependency on $network will now
66 get a dependency on network-online.target rather than simply
67 network.target. This should bring LSB handling closer to
68 what it was on SysV systems.
70 * A new fsck.repair= kernel option has been added to control
71 how fsck shall deal with unclean file systems at boot.
73 * The (.ini) configuration file parser will now silently
74 ignore sections whose name begins with "X-". This may be
75 used to maintain application-specific extension sections in unit
78 * machined gained a new API to query the IP addresses of
79 registered containers. "machinectl status" has been updated
80 to show these addresses in its output.
82 * A new call sd_uid_get_display() has been added to the
83 sd-login APIs for querying the "primary" session of a
84 user. The "primary" session of the user is elected from the
85 user's sessions and generally a graphical session is
86 preferred over a text one.
88 * A minimal systemd-resolved daemon has been added. It
89 currently simply acts as a companion to systemd-networkd and
90 manages resolv.conf based on per-interface DNS
91 configuration, possibly supplied via DHCP. In the long run
92 we hope to extend this into a local DNSSEC enabled DNS and
95 * The systemd-networkd-wait-online tool is now enabled by
96 default. It will delay network-online.target until a network
97 connection has been configured. The tool primarily integrates
98 with networkd, but will also make a best effort to make sense
99 of network configuration performed in some other way.
101 * Two new service options StartupCPUShares= and
102 StartupBlockIOWeight= have been added that work similarly to
103 CPUShares= and BlockIOWeight= however only apply during
104 system startup. This is useful to prioritize certain services
105 differently during bootup than during normal runtime.
107 * hostnamed has been changed to prefer the statically
108 configured hostname in /etc/hostname (unless set to
109 'localhost' or empty) over any dynamic one supplied by
110 dhcp. With this change the rules for picking the hostname
111 match more closely the rules of other configuration settings
112 where the local administrator's configuration in /etc always
113 overrides any other settings.
115 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
116 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
117 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
118 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
119 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
120 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
121 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
122 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
123 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
124 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
125 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
126 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
127 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
128 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
129 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
130 Lindskog, WaLyong Cho, Will Woods, Zbigniew
133 -- Beijing, 2014-05-28
137 * When restoring the screen brightness at boot, stay away from
138 the darkest setting or from the lowest 5% of the available
139 range, depending on which is the larger value of both. This
140 should effectively protect the user from rebooting into a
141 black screen, should the brightness have been set to minimum
144 * sd-login gained a new sd_machine_get_class() call to
145 determine the class ("vm" or "container") of a machine
146 registered with machined.
148 * sd-login gained new calls
149 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
150 to query the identity of the peer of a local AF_UNIX
151 connection. They operate similarly to their sd_pid_get_xyz()
154 * PID 1 will now maintain a system-wide system state engine
155 with the states "starting", "running", "degraded",
156 "maintenance", "stopping". These states are bound to system
157 startup, normal runtime, runtime with at least one failed
158 service, rescue/emergency mode and system shutdown. This
159 state is shown in the "systemctl status" output when no unit
160 name is passed. It is useful to determine system state, in
161 particularly when doing so for many systems or containers at
164 * A new command "list-machines" has been added to "systemctl"
165 that lists all local OS containers and shows their system
166 state (see above), if systemd runs inside of them.
168 * systemctl gained a new "-r" switch to recursively enumerate
169 units on all local containers, when used with the
170 "list-unit" command (which is the default one that is
171 executed when no parameters are specified).
173 * The GPT automatic partition discovery logic will now honour
174 two GPT partition flags: one may be set on a partition to
175 cause it to be mounted read-only, and the other may be set
176 on a partition to ignore it during automatic discovery.
178 * Two new GPT type UUIDs have been added for automatic root
179 partition discovery, for 32-bit and 64-bit ARM. This is not
180 particularly useful for discovering the root directory on
181 these architectures during bare-metal boots (since UEFI is
182 not common there), but still very useful to allow booting of
183 ARM disk images in nspawn with the -i option.
185 * MAC addresses of interfaces created with nspawn's
186 --network-interface= switch will now be generated from the
187 machine name, and thus be stable between multiple invocations
190 * logind will now automatically remove all IPC objects owned
191 by a user if she or he fully logs out. This makes sure that
192 users who are logged out cannot continue to consume IPC
193 resources. This covers SysV memory, semaphores and message
194 queues as well as POSIX shared memory and message
195 queues. Traditionally, SysV and POSIX IPC had no life-cycle
196 limits. With this functionality, that is corrected. This may
197 be turned off by using the RemoveIPC= switch of logind.conf.
199 * The systemd-machine-id-setup and tmpfiles tools gained a
200 --root= switch to operate on a specific root directory,
203 * journald can now forward logged messages to the TTYs of all
204 logged in users ("wall"). This is the default for all
205 emergency messages now.
207 * A new tool systemd-journal-remote has been added to stream
208 journal log messages across the network.
210 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
211 controller trees are mounted into it. Note that the
212 directories mounted beneath it are not read-only. This is a
213 security measure and is particularly useful because glibc
214 actually includes a search logic to pick any tmpfs it can
215 find to implement shm_open() if /dev/shm is not available
216 (which it might very well be in namespaced setups).
218 * machinectl gained a new "poweroff" command to cleanly power
219 down a local OS container.
221 * The PrivateDevices= unit file setting will now also drop the
222 CAP_MKNOD capability from the capability bound set, and
223 imply DevicePolicy=closed.
225 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
226 comprehensively on all long-running systemd services where
229 * systemd-udevd will now run in a disassociated mount
230 namespace. To mount directories from udev rules, make sure to
231 pull in mount units via SYSTEMD_WANTS properties.
233 * The kdbus support gained support for uploading policy into
234 the kernel. sd-bus gained support for creating "monitoring"
235 connections that can eavesdrop into all bus communication
236 for debugging purposes.
238 * Timestamps may now be specified in seconds since the UNIX
239 epoch Jan 1st, 1970 by specifying "@" followed by the value
242 * Native tcpwrap support in systemd has been removed. tcpwrap
243 is old code, not really maintained anymore and has serious
244 shortcomings, and better options such as firewalls
245 exist. For setups that require tcpwrap usage, please
246 consider invoking your socket-activated service via tcpd,
247 like on traditional inetd.
249 * A new system.conf configuration option
250 DefaultTimerAccuracySec= has been added that controls the
251 default AccuracySec= setting of .timer units.
253 * Timer units gained a new WakeSystem= switch. If enabled,
254 timers configured this way will cause the system to resume
255 from system suspend (if the system supports that, which most
258 * Timer units gained a new Persistent= switch. If enabled,
259 timers configured this way will save to disk when they have
260 been last triggered. This information is then used on next
261 reboot to possible execute overdue timer events, that
262 could not take place because the system was powered off.
263 This enables simple anacron-like behaviour for timer units.
265 * systemctl's "list-timers" will now also list the time a
266 timer unit was last triggered in addition to the next time
267 it will be triggered.
269 * systemd-networkd will now assign predictable IPv4LL
270 addresses to its local interfaces.
272 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
273 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
274 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
275 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
276 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
277 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
278 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
279 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
282 -- Berlin, 2014-03-25
286 * A new unit file setting RestrictAddressFamilies= has been
287 added to restrict which socket address families unit
288 processes gain access to. This takes address family names
289 like "AF_INET" or "AF_UNIX", and is useful to minimize the
290 attack surface of services via exotic protocol stacks. This
291 is built on seccomp system call filters.
293 * Two new unit file settings RuntimeDirectory= and
294 RuntimeDirectoryMode= have been added that may be used to
295 manage a per-daemon runtime directories below /run. This is
296 an alternative for setting up directory permissions with
297 tmpfiles snippets, and has the advantage that the runtime
298 directory's lifetime is bound to the daemon runtime and that
299 the daemon starts up with an empty directory each time. This
300 is particularly useful when writing services that drop
301 privileges using the User= or Group= setting.
303 * The DeviceAllow= unit setting now supports globbing for
304 matching against device group names.
306 * The systemd configuration file system.conf gained new
307 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
308 DefaultMemoryAccounting= to globally turn on/off accounting
309 for specific resources (cgroups) for all units. These
310 settings may still be overridden individually in each unit
313 * systemd-gpt-auto-generator is now able to discover /srv and
314 root partitions in addition to /home and swap partitions. It
315 also supports LUKS-encrypted partitions now. With this in
316 place, automatic discovery of partitions to mount following
317 the Discoverable Partitions Specification
318 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
319 is now a lot more complete. This allows booting without
320 /etc/fstab and without root= on the kernel command line on
321 systems prepared appropriately.
323 * systemd-nspawn gained a new --image= switch which allows
324 booting up disk images and Linux installations on any block
325 device that follow the Discoverable Partitions Specification
326 (see above). This means that installations made with
327 appropriately updated installers may now be started and
328 deployed using container managers, completely
329 unmodified. (We hope that libvirt-lxc will add support for
330 this feature soon, too.)
332 * systemd-nspawn gained a new --network-macvlan= setting to
333 set up a private macvlan interface for the
334 container. Similarly, systemd-networkd gained a new
335 Kind=macvlan setting in .netdev files.
337 * systemd-networkd now supports configuring local addresses
340 * A new tool systemd-network-wait-online has been added to
341 synchronously wait for network connectivity using
344 * The sd-bus.h bus API gained a new sd_bus_track object for
345 tracking the life-cycle of bus peers. Note that sd-bus.h is
346 still not a public API though (unless you specify
347 --enable-kdbus on the configure command line, which however
348 voids your warranty and you get no API stability guarantee).
350 * The $XDG_RUNTIME_DIR runtime directories for each user are
351 now individual tmpfs instances, which has the benefit of
352 introducing separate pools for each user, with individual
353 size limits, and thus making sure that unprivileged clients
354 can no longer negatively impact the system or other users by
355 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
356 RuntimeDirectorySize= has been introduced that allows
357 controlling the default size limit for all users. It
358 defaults to 10% of the available physical memory. This is no
359 replacement for quotas on tmpfs though (which the kernel
360 still does not support), as /dev/shm and /tmp are still
361 shared resources used by both the system and unprivileged
364 * logind will now automatically turn off automatic suspending
365 on laptop lid close when more than one display is
366 connected. This was previously expected to be implemented
367 individually in desktop environments (such as GNOME),
368 however has been added to logind now, in order to fix a
369 boot-time race where a desktop environment might not have
370 been started yet and thus not been able to take an inhibitor
371 lock at the time where logind already suspends the system
374 * logind will now wait at least 30s after each system
375 suspend/resume cycle, and 3min after system boot before
376 suspending the system due to a closed laptop lid. This
377 should give USB docking stations and similar enough time to
378 be probed and configured after system resume and boot in
379 order to then act as suspend blocker.
381 * systemd-run gained a new --property= setting which allows
382 initialization of resource control properties (and others)
383 for the created scope or service unit. Example: "systemd-run
384 --property=BlockIOWeight=10 updatedb" may be used to run
385 updatedb at a low block IO scheduling weight.
387 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
388 now also work in --scope mode.
390 * When systemd is compiled with kdbus support, basic support
391 for enforced policies is now in place. (Note that enabling
392 kdbus still voids your warranty and no API compatibility
395 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
396 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
397 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
398 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
399 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
400 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
401 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
402 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
403 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
404 Zbigniew Jędrzejewski-Szmek
406 -- Berlin, 2014-03-12
410 * systemd will now relabel /dev after loading the SMACK policy
411 according to SMACK rules.
413 * A new unit file option AppArmorProfile= has been added to
414 set the AppArmor profile for the processes of a unit.
416 * A new condition check ConditionArchitecture= has been added
417 to conditionalize units based on the system architecture, as
418 reported by uname()'s "machine" field.
420 * systemd-networkd now supports matching on the system
421 virtualization, architecture, kernel command line, host name
424 * logind is now a lot more aggressive when suspending the
425 machine due to a closed laptop lid. Instead of acting only
426 on the lid close action, it will continuously watch the lid
427 status and act on it. This is useful for laptops where the
428 power button is on the outside of the chassis so that it can
429 be reached without opening the lid (such as the Lenovo
430 Yoga). On those machines, logind will now immediately
431 re-suspend the machine if the power button has been
432 accidentally pressed while the laptop was suspended and in a
435 * logind will now watch SW_DOCK switches and inhibit reaction
436 to the lid switch if it is pressed. This means that logind
437 will not suspend the machine anymore if the lid is closed
438 and the system is docked, if the laptop supports SW_DOCK
439 notifications via the input layer. Note that ACPI docking
440 stations do not generate this currently. Also note that this
441 logic is usually not fully sufficient and Desktop
442 Environments should take a lid switch inhibitor lock when an
443 external display is connected, as systemd will not watch
446 * nspawn will now make use of the devices cgroup controller by
447 default, and only permit creation of and access to the usual
448 API device nodes like /dev/null or /dev/random, as well as
449 access to (but not creation of) the pty devices.
451 * We will now ship a default .network file for
452 systemd-networkd that automatically configures DHCP for
453 network interfaces created by nspawn's --network-veth or
454 --network-bridge= switches.
456 * systemd will now understand the usual M, K, G, T suffixes
457 according to SI conventions (i.e. to the base 1000) when
458 referring to throughput and hardware metrics. It will stay
459 with IEC conventions (i.e. to the base 1024) for software
460 metrics, according to what is customary according to
461 Wikipedia. We explicitly document which base applies for
462 each configuration option.
464 * The DeviceAllow= setting in unit files now supports a syntax
465 to whitelist an entire group of devices node majors at once,
466 based on the /proc/devices listing. For example, with the
467 string "char-pts", it is now possible to whitelist all
468 current and future pseudo-TTYs at once.
470 * sd-event learned a new "post" event source. Event sources of
471 this type are triggered by the dispatching of any event
472 source of a type that is not "post". This is useful for
473 implementing clean-up and check event sources that are
474 triggered by other work being done in the program.
476 * systemd-networkd is no longer statically enabled, but uses
477 the usual [Install] sections so that it can be
478 enabled/disabled using systemctl. It still is enabled by
481 * When creating a veth interface pair with systemd-nspawn, the
482 host side will now be prefixed with "vb-" if
483 --network-bridge= is used, and with "ve-" if --network-veth
484 is used. This way, it is easy to distinguish these cases on
485 the host, for example to apply different configuration to
486 them with systemd-networkd.
488 * The compatibility libraries for libsystemd-journal.so,
489 libsystem-id128.so, libsystemd-login.so and
490 libsystemd-daemon.so do not make use of IFUNC
491 anymore. Instead, we now build libsystemd.so multiple times
492 under these alternative names. This means that the footprint
493 is drastically increased, but given that these are
494 transitional compatibility libraries, this should not matter
495 much. This change has been made necessary to support the ARM
496 platform for these compatibility libraries, as the ARM
497 toolchain is not really at the same level as the toolchain
498 for other architectures like x86 and does not support
499 IFUNC. Please make sure to use --enable-compat-libs only
500 during a transitional period!
502 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
503 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
504 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
505 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
506 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
507 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
508 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
509 Zbigniew Jędrzejewski-Szmek
511 -- Berlin, 2014-02-24
515 * A new component "systemd-networkd" has been added that can
516 be used to configure local network interfaces statically or
517 via DHCP. It is capable of bringing up bridges, VLANs, and
518 bonding. Currently, no hook-ups for interactive network
519 configuration are provided. Use this for your initrd,
520 container, embedded, or server setup if you need a simple,
521 yet powerful, network configuration solution. This
522 configuration subsystem is quite nifty, as it allows wildcard
523 hotplug matching in interfaces. For example, with a single
524 configuration snippet, you can configure that all Ethernet
525 interfaces showing up are automatically added to a bridge,
526 or similar. It supports link-sensing and more.
528 * A new tool "systemd-socket-proxyd" has been added which can
529 act as a bidirectional proxy for TCP sockets. This is
530 useful for adding socket activation support to services that
531 do not actually support socket activation, including virtual
532 machines and the like.
534 * Add a new tool to save/restore rfkill state on
537 * Save/restore state of keyboard backlights in addition to
538 display backlights on shutdown/boot.
540 * udev learned a new SECLABEL{} construct to label device
541 nodes with a specific security label when they appear. For
542 now, only SECLABEL{selinux} is supported, but the syntax is
543 prepared for additional security frameworks.
545 * udev gained a new scheme to configure link-level attributes
546 from files in /etc/systemd/network/*.link. These files can
547 match against MAC address, device path, driver name and type,
548 and will apply attributes like the naming policy, link speed,
549 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
550 address assignment policy (randomized, ...).
552 * The configuration of network interface naming rules for
553 "permanent interface names" has changed: a new NamePolicy=
554 setting in the [Link] section of .link files determines the
555 priority of possible naming schemes (onboard, slot, mac,
556 path). The default value of this setting is determined by
557 /usr/lib/net/links/99-default.link. Old
558 80-net-name-slot.rules udev configuration file has been
559 removed, so local configuration overriding this file should
560 be adapated to override 99-default.link instead.
562 * When the User= switch is used in a unit file, also
563 initialize $SHELL= based on the user database entry.
565 * systemd no longer depends on libdbus. All communication is
566 now done with sd-bus, systemd's low-level bus library
569 * kdbus support has been added to PID 1 itself. When kdbus is
570 enabled, this causes PID 1 to set up the system bus and
571 enable support for a new ".busname" unit type that
572 encapsulates bus name activation on kdbus. It works a little
573 bit like ".socket" units, except for bus names. A new
574 generator has been added that converts classic dbus1 service
575 activation files automatically into native systemd .busname
578 * sd-bus: add a light-weight vtable implementation that allows
579 defining objects on the bus with a simple static const
580 vtable array of its methods, signals and properties.
582 * systemd will not generate or install static dbus
583 introspection data anymore to /usr/share/dbus-1/interfaces,
584 as the precise format of these files is unclear, and
585 nothing makes use of it.
587 * A proxy daemon is now provided to proxy clients connecting
588 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
589 compatibility with classic D-Bus.
591 * A bus driver implementation has been added that supports the
592 classic D-Bus bus driver calls on kdbus, also for
593 compatibility purposes.
595 * A new API "sd-event.h" has been added that implements a
596 minimal event loop API built around epoll. It provides a
597 couple of features that direct epoll usage is lacking:
598 prioritization of events, scales to large numbers of timer
599 events, per-event timer slack (accuracy), system-wide
600 coalescing of timer events, exit handlers, watchdog
601 supervision support using systemd's sd_notify() API, child
604 * A new API "sd-rntl.h" has been added that provides an API
605 around the route netlink interface of the kernel, similar in
608 * A new API "sd-dhcp-client.h" has been added that provides a
609 small DHCPv4 client-side implementation. This is used by
612 * There is a new kernel command line option
613 "systemd.restore_state=0|1". When set to "0", none of the
614 systemd tools will restore saved runtime state to hardware
615 devices. More specifically, the rfkill and backlight states
618 * The FsckPassNo= compatibility option in mount/service units
619 has been removed. The fstab generator will now add the
620 necessary dependencies automatically, and does not require
621 PID1's support for that anymore.
623 * journalctl gained a new switch, --list-boots, that lists
624 recent boots with their times and boot IDs.
626 * The various tools like systemctl, loginctl, timedatectl,
627 busctl, systemd-run, ... have gained a new switch "-M" to
628 connect to a specific, local OS container (as direct
629 connection, without requiring SSH). This works on any
630 container that is registered with machined, such as those
631 created by libvirt-lxc or nspawn.
633 * systemd-run and systemd-analyze also gained support for "-H"
634 to connect to remote hosts via SSH. This is particularly
635 useful for systemd-run because it enables queuing of jobs
638 * machinectl gained a new command "login" to open a getty
639 login in any local container. This works with any container
640 that is registered with machined (such as those created by
641 libvirt-lxc or nspawn), and which runs systemd inside.
643 * machinectl gained a new "reboot" command that may be used to
644 trigger a reboot on a specific container that is registered
645 with machined. This works on any container that runs an init
648 * systemctl gained a new "list-timers" command to print a nice
649 listing of installed timer units with the times they elapse
652 * Alternative reboot() parameters may now be specified on the
653 "systemctl reboot" command line and are passed to the
654 reboot() system call.
656 * systemctl gained a new --job-mode= switch to configure the
657 mode to queue a job with. This is a more generic version of
658 --fail, --irreversible, and --ignore-dependencies, which are
659 still available but not advertised anymore.
661 * /etc/systemd/system.conf gained new settings to configure
662 various default timeouts of units, as well as the default
663 start limit interval and burst. These may still be overridden
666 * PID1 will now export on the bus profile data of the security
667 policy upload process (such as the SELinux policy upload to
670 * journald: when forwarding logs to the console, include
671 timestamps (following the setting in
672 /sys/module/printk/parameters/time).
674 * OnCalendar= in timer units now understands the special
675 strings "yearly" and "annually". (Both are equivalent)
677 * The accuracy of timer units is now configurable with the new
678 AccuracySec= setting. It defaults to 1min.
680 * A new dependency type JoinsNamespaceOf= has been added that
681 allows running two services within the same /tmp and network
682 namespace, if PrivateNetwork= or PrivateTmp= are used.
684 * A new command "cat" has been added to systemctl. It outputs
685 the original unit file of a unit, and concatenates the
686 contents of additional "drop-in" unit file snippets, so that
687 the full configuration is shown.
689 * systemctl now supports globbing on the various "list-xyz"
690 commands, like "list-units" or "list-sockets", as well as on
691 those commands which take multiple unit names.
693 * journalctl's --unit= switch gained support for globbing.
695 * All systemd daemons now make use of the watchdog logic so
696 that systemd automatically notices when they hang.
698 * If the $container_ttys environment variable is set,
699 getty-generator will automatically spawn a getty for each
700 listed tty. This is useful for container managers to request
701 login gettys to be spawned on as many ttys as needed.
703 * %h, %s, %U specifier support is not available anymore when
704 used in unit files for PID 1. This is because NSS calls are
705 not safe from PID 1. They stay available for --user
706 instances of systemd, and as special case for the root user.
708 * loginctl gained a new "--no-legend" switch to turn off output
711 * The "sd-login.h" API gained three new calls:
712 sd_session_is_remote(), sd_session_get_remote_user(),
713 sd_session_get_remote_host() to query information about
716 * The udev hardware database now also carries vendor/product
717 information of SDIO devices.
719 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
720 determine whether watchdog notifications are requested by
723 * Socket-activated per-connection services now include a
724 short description of the connection parameters in the
727 * tmpfiles gained a new "--boot" option. When this is not used,
728 only lines where the command character is not suffixed with
729 "!" are executed. When this option is specified, those
730 options are executed too. This partitions tmpfiles
731 directives into those that can be safely executed at any
732 time, and those which should be run only at boot (for
733 example, a line that creates /run/nologin).
735 * A new API "sd-resolve.h" has been added which provides a simple
736 asynchronous wrapper around glibc NSS host name resolution
737 calls, such as getaddrinfo(). In contrast to glibc's
738 getaddrinfo_a(), it does not use signals. In contrast to most
739 other asynchronous name resolution libraries, this one does
740 not reimplement DNS, but reuses NSS, so that alternate
741 host name resolution systems continue to work, such as mDNS,
742 LDAP, etc. This API is based on libasyncns, but it has been
743 cleaned up for inclusion in systemd.
745 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
746 "sd-daemon.h" are no longer found in individual libraries
747 libsystemd-journal.so, libsystemd-login.so,
748 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
749 merged them into a single library, libsystemd.so, which
750 provides all symbols. The reason for this is cyclic
751 dependencies, as these libraries tend to use each other's
752 symbols. So far, we have managed to workaround that by linking
753 a copy of a good part of our code into each of these
754 libraries again and again, which, however, makes certain
755 things hard to do, like sharing static variables. Also, it
756 substantially increases footprint. With this change, there
757 is only one library for the basic APIs systemd
758 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
759 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
760 library as well, however are subject to the --enable-kdbus
761 switch (see below). Note that "sd-dhcp-client.h" is not part
762 of this library (this is because it only consumes, never
763 provides, services of/to other APIs). To make the transition
764 easy from the separate libraries to the unified one, we
765 provide the --enable-compat-libs compile-time switch which
766 will generate stub libraries that are compatible with the
767 old ones but redirect all calls to the new one.
769 * All of the kdbus logic and the new APIs "sd-bus.h",
770 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
771 and "sd-utf8.h" are compile-time optional via the
772 "--enable-kdbus" switch, and they are not compiled in by
773 default. To make use of kdbus, you have to explicitly enable
774 the switch. Note however, that neither the kernel nor the
775 userspace API for all of this is considered stable yet. We
776 want to maintain the freedom to still change the APIs for
777 now. By specifying this build-time switch, you acknowledge
778 that you are aware of the instability of the current
781 * Also, note that while kdbus is pretty much complete,
782 it lacks one thing: proper policy support. This means you
783 can build a fully working system with all features; however,
784 it will be highly insecure. Policy support will be added in
785 one of the next releases, at the same time that we will
786 declare the APIs stable.
788 * When the kernel command-line argument "kdbus" is specified,
789 systemd will automatically load the kdbus.ko kernel module. At
790 this stage of development, it is only useful for testing kdbus
791 and should not be used in production. Note: if "--enable-kdbus"
792 is specified, and the kdbus.ko kernel module is available, and
793 "kdbus" is added to the kernel command line, the entire system
794 runs with kdbus instead of dbus-daemon, with the above mentioned
795 problem of missing the system policy enforcement. Also a future
796 version of kdbus.ko or a newer systemd will not be compatible with
797 each other, and will unlikely be able to boot the machine if only
798 one of them is updated.
800 * systemctl gained a new "import-environment" command which
801 uploads the caller's environment (or parts thereof) into the
802 service manager so that it is inherited by services started
803 by the manager. This is useful to upload variables like
804 $DISPLAY into the user service manager.
806 * A new PrivateDevices= switch has been added to service units
807 which allows running a service with a namespaced /dev
808 directory that does not contain any device nodes for
809 physical devices. More specifically, it only includes devices
810 such as /dev/null, /dev/urandom, and /dev/zero which are API
813 * logind has been extended to support behaviour like VT
814 switching on seats that do not support a VT. This makes
815 multi-session available on seats that are not the first seat
816 (seat0), and on systems where kernel support for VTs has
817 been disabled at compile-time.
819 * If a process holds a delay lock for system sleep or shutdown
820 and fails to release it in time, we will now log its
821 identity. This makes it easier to identify processes that
822 cause slow suspends or power-offs.
824 * When parsing /etc/crypttab, support for a new key-slot=
825 option as supported by Debian is added. It allows indicating
826 which LUKS slot to use on disk, speeding up key loading.
828 * The sd_journald_sendv() API call has been checked and
829 officially declared to be async-signal-safe so that it may
830 be invoked from signal handlers for logging purposes.
832 * Boot-time status output is now enabled automatically after a
833 short timeout if boot does not progress, in order to give
834 the user an indication what she or he is waiting for.
836 * The boot-time output has been improved to show how much time
837 remains until jobs expire.
839 * The KillMode= switch in service units gained a new possible
840 value "mixed". If set, and the unit is shut down, then the
841 initial SIGTERM signal is sent only to the main daemon
842 process, while the following SIGKILL signal is sent to
843 all remaining processes of the service.
845 * When a scope unit is registered, a new property "Controller"
846 may be set. If set to a valid bus name, systemd will send a
847 RequestStop() signal to this name when it would like to shut
848 down the scope. This may be used to hook manager logic into
849 the shutdown logic of scope units. Also, scope units may now
850 be put in a special "abandoned" state, in which case the
851 manager process which created them takes no further
852 responsibilities for it.
854 * When reading unit files, systemd will now verify
855 the access mode of these files, and warn about certain
856 suspicious combinations. This has been added to make it
857 easier to track down packaging bugs where unit files are
858 marked executable or world-writable.
860 * systemd-nspawn gained a new "--setenv=" switch to set
861 container-wide environment variables. The similar option in
862 systemd-activate was renamed from "--environment=" to
863 "--setenv=" for consistency.
865 * systemd-nspawn has been updated to create a new kdbus domain
866 for each container that is invoked, thus allowing each
867 container to have its own set of system and user buses,
868 independent of the host.
870 * systemd-nspawn gained a new --drop-capability= switch to run
871 the container with less capabilities than the default. Both
872 --drop-capability= and --capability= now take the special
873 string "all" for dropping or keeping all capabilities.
875 * systemd-nspawn gained new switches for executing containers
876 with specific SELinux labels set.
878 * systemd-nspawn gained a new --quiet switch to not generate
879 any additional output but the container's own console
882 * systemd-nspawn gained a new --share-system switch to run a
883 container without PID namespacing enabled.
885 * systemd-nspawn gained a new --register= switch to control
886 whether the container is registered with systemd-machined or
887 not. This is useful for containers that do not run full
888 OS images, but only specific apps.
890 * systemd-nspawn gained a new --keep-unit which may be used
891 when invoked as the only program from a service unit, and
892 results in registration of the unit service itself in
893 systemd-machined, instead of a newly opened scope unit.
895 * systemd-nspawn gained a new --network-interface= switch for
896 moving arbitrary interfaces to the container. The new
897 --network-veth switch creates a virtual Ethernet connection
898 between host and container. The new --network-bridge=
899 switch then allows assigning the host side of this virtual
900 Ethernet connection to a bridge device.
902 * systemd-nspawn gained a new --personality= switch for
903 setting the kernel personality for the container. This is
904 useful when running a 32-bit container on a 64-bit host. A
905 similar option Personality= is now also available for service
908 * logind will now also track a "Desktop" identifier for each
909 session which encodes the desktop environment of it. This is
910 useful for desktop environments that want to identify
911 multiple running sessions of itself easily.
913 * A new SELinuxContext= setting for service units has been
914 added that allows setting a specific SELinux execution
915 context for a service.
917 * Most systemd client tools will now honour $SYSTEMD_LESS for
918 settings of the "less" pager. By default, these tools will
919 override $LESS to allow certain operations to work, such as
920 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
921 influence this logic.
923 * systemd's "seccomp" hook-up has been changed to make use of
924 the libseccomp library instead of using its own
925 implementation. This has benefits for portability among
928 * For usage together with SystemCallFilter=, a new
929 SystemCallErrorNumber= setting has been introduced that
930 allows configuration of a system error number to be returned
931 on filtered system calls, instead of immediately killing the
932 process. Also, SystemCallArchitectures= has been added to
933 limit access to system calls of a particular architecture
934 (in order to turn off support for unused secondary
935 architectures). There is also a global
936 SystemCallArchitectures= setting in system.conf now to turn
937 off support for non-native system calls system-wide.
939 * systemd requires a kernel with a working name_to_handle_at(),
940 please see the kernel config requirements in the README file.
942 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
943 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
944 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
945 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
946 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
947 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
948 Elia Pinto, Florian Weimer, George McCollister, Goffredo
949 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
950 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
951 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
952 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
953 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
954 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
955 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
956 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
957 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
958 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
959 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
960 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
961 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
962 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
963 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
964 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
965 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
967 -- Berlin, 2014-02-20
971 * logind has gained support for facilitating privileged input
972 and drm device access for unprivileged clients. This work is
973 useful to allow Wayland display servers (and similar
974 programs, such as kmscon) to run under the user's ID and
975 access input and drm devices which are normally
976 protected. When this is used (and the kernel is new enough)
977 logind will "mute" IO on the file descriptors passed to
978 Wayland as long as it is in the background and "unmute" it
979 if it returns into the foreground. This allows secure
980 session switching without allowing background sessions to
981 eavesdrop on input and display data. This also introduces
982 session switching support if VT support is turned off in the
983 kernel, and on seats that are not seat0.
985 * A new kernel command line option luks.options= is understood
986 now which allows specifiying LUKS options for usage for LUKS
987 encrypted partitions specified with luks.uuid=.
989 * tmpfiles.d(5) snippets may now use specifier expansion in
990 path names. More specifically %m, %b, %H, %v, are now
991 replaced by the local machine id, boot id, hostname, and
992 kernel version number.
994 * A new tmpfiles.d(5) command "m" has been introduced which
995 may be used to change the owner/group/access mode of a file
996 or directory if it exists, but do nothing if it does not.
998 * This release removes high-level support for the
999 MemorySoftLimit= cgroup setting. The underlying kernel
1000 cgroup attribute memory.soft_limit= is currently badly
1001 designed and likely to be removed from the kernel API in its
1002 current form, hence we should not expose it for now.
1004 * The memory.use_hierarchy cgroup attribute is now enabled for
1005 all cgroups systemd creates in the memory cgroup
1006 hierarchy. This option is likely to be come the built-in
1007 default in the kernel anyway, and the non-hierarchial mode
1008 never made much sense in the intrinsically hierarchial
1011 * A new field _SYSTEMD_SLICE= is logged along with all journal
1012 messages containing the slice a message was generated
1013 from. This is useful to allow easy per-customer filtering of
1014 logs among other things.
1016 * systemd-journald will no longer adjust the group of journal
1017 files it creates to the "systemd-journal" group. Instead we
1018 rely on the journal directory to be owned by the
1019 "systemd-journal" group, and its setgid bit set, so that the
1020 kernel file system layer will automatically enforce that
1021 journal files inherit this group assignment. The reason for
1022 this change is that we cannot allow NSS look-ups from
1023 journald which would be necessary to resolve
1024 "systemd-journal" to a numeric GID, because this might
1025 create deadlocks if NSS involves synchronous queries to
1026 other daemons (such as nscd, or sssd) which in turn are
1027 logging clients of journald and might block on it, which
1028 would then dead lock. A tmpfiles.d(5) snippet included in
1029 systemd will make sure the setgid bit and group are
1030 properly set on the journal directory if it exists on every
1031 boot. However, we recommend adjusting it manually after
1032 upgrades too (or from RPM scriptlets), so that the change is
1033 not delayed until next reboot.
1035 * Backlight and random seed files in /var/lib/ have moved into
1036 the /var/lib/systemd/ directory, in order to centralize all
1037 systemd generated files in one directory.
1039 * Boot time performance measurements (as displayed by
1040 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
1041 performance information if that's available to determine how
1042 much time BIOS and boot loader initialization required. With
1043 a sufficiently new BIOS you hence no longer need to boot
1044 with Gummiboot to get access to such information.
1046 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
1047 Cristian Rodríguez, Dave Reisner, David Herrmann, David
1048 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
1049 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
1050 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
1051 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
1052 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1054 -- Berlin, 2013-10-02
1058 * The Restart= option for services now understands a new
1059 on-watchdog setting, which will restart the service
1060 automatically if the service stops sending out watchdog keep
1061 alive messages (as configured with WatchdogSec=).
1063 * The getty generator (which is responsible for bringing up a
1064 getty on configured serial consoles) will no longer only
1065 start a getty on the primary kernel console but on all
1066 others, too. This makes the order in which console= is
1067 specified on the kernel command line less important.
1069 * libsystemd-logind gained a new sd_session_get_vt() call to
1070 retrieve the VT number of a session.
1072 * If the option "tries=0" is set for an entry of /etc/crypttab
1073 its passphrase is queried indefinitely instead of any
1074 maximum number of tries.
1076 * If a service with a configure PID file terminates its PID
1077 file will now be removed automatically if it still exists
1078 afterwards. This should put an end to stale PID files.
1080 * systemd-run will now also take relative binary path names
1081 for execution and no longer insists on absolute paths.
1083 * InaccessibleDirectories= and ReadOnlyDirectories= now take
1084 paths that are optionally prefixed with "-" to indicate that
1085 it should not be considered a failure if they do not exist.
1087 * journalctl -o (and similar commands) now understands a new
1088 output mode "short-precise", it is similar to "short" but
1089 shows timestamps with usec accuracy.
1091 * The option "discard" (as known from Debian) is now
1092 synonymous to "allow-discards" in /etc/crypttab. In fact,
1093 "discard" is preferred now (since it is easier to remember
1096 * Some licensing clean-ups were made, so that more code is now
1097 LGPL-2.1 licensed than before.
1099 * A minimal tool to save/restore the display backlight
1100 brightness across reboots has been added. It will store the
1101 backlight setting as late as possible at shutdown, and
1102 restore it as early as possible during reboot.
1104 * A logic to automatically discover and enable home and swap
1105 partitions on GPT disks has been added. With this in place
1106 /etc/fstab becomes optional for many setups as systemd can
1107 discover certain partitions located on the root disk
1108 automatically. Home partitions are recognized under their
1109 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
1110 partitions are recognized under their GPT type ID
1111 0657fd6da4ab43c484e50933c84b4f4f.
1113 * systemd will no longer pass any environment from the kernel
1114 or initrd to system services. If you want to set an
1115 environment for all services, do so via the kernel command
1116 line systemd.setenv= assignment.
1118 * The systemd-sysctl tool no longer natively reads the file
1119 /etc/sysctl.conf. If desired, the file should be symlinked
1120 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
1121 legacy support by a symlink rather than built-in code, it
1122 also makes the otherwise hidden order of application of the
1123 different files visible. (Note that this partly reverts to a
1124 pre-198 application order of sysctl knobs!)
1126 * The "systemctl set-log-level" and "systemctl dump" commands
1127 have been moved to systemd-analyze.
1129 * systemd-run learned the new --remain-after-exit switch,
1130 which causes the scope unit not to be cleaned up
1131 automatically after the process terminated.
1133 * tmpfiles learned a new --exclude-prefix= switch to exclude
1134 certain paths from operation.
1136 * journald will now automatically flush all messages to disk
1137 as soon as a message of the log priorities CRIT, ALERT or
1140 Contributions from: Andrew Cook, Brandon Philips, Christian
1141 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
1142 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
1143 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
1144 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
1145 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
1146 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
1147 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
1148 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
1149 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
1150 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
1151 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
1152 William Giokas, Zbigniew Jędrzejewski-Szmek
1154 -- Berlin, 2013-09-13
1158 * The documentation has been updated to cover the various new
1159 concepts introduced with 205.
1161 * Unit files now understand the new %v specifier which
1162 resolves to the kernel version string as returned by "uname
1165 * systemctl now supports filtering the unit list output by
1166 load state, active state and sub state, using the new
1169 * "systemctl status" will now show the results of the
1170 condition checks (like ConditionPathExists= and similar) of
1171 the last start attempts of the unit. They are also logged to
1174 * "journalctl -b" may now be used to look for boot output of a
1175 specific boot. Try "journalctl -b -1" for the previous boot,
1176 but the syntax is substantially more powerful.
1178 * "journalctl --show-cursor" has been added which prints the
1179 cursor string the last shown log line. This may then be used
1180 with the new "journalctl --after-cursor=" switch to continue
1181 browsing logs from that point on.
1183 * "journalctl --force" may now be used to force regeneration
1186 * Creation of "dead" device nodes has been moved from udev
1187 into kmod and tmpfiles. Previously, udev would read the kmod
1188 databases to pre-generate dead device nodes based on meta
1189 information contained in kernel modules, so that these would
1190 be auto-loaded on access rather then at boot. As this
1191 does not really have much to do with the exposing actual
1192 kernel devices to userspace this has always been slightly
1193 alien in the udev codebase. Following the new scheme kmod
1194 will now generate a runtime snippet for tmpfiles from the
1195 module meta information and it now is tmpfiles' job to the
1196 create the nodes. This also allows overriding access and
1197 other parameters for the nodes using the usual tmpfiles
1198 facilities. As side effect this allows us to remove the
1199 CAP_SYS_MKNOD capability bit from udevd entirely.
1201 * logind's device ACLs may now be applied to these "dead"
1202 devices nodes too, thus finally allowing managed access to
1203 devices such as /dev/snd/sequencer whithout loading the
1204 backing module right-away.
1206 * A new RPM macro has been added that may be used to apply
1207 tmpfiles configuration during package installation.
1209 * systemd-detect-virt and ConditionVirtualization= now can
1210 detect User-Mode-Linux machines (UML).
1212 * journald will now implicitly log the effective capabilities
1213 set of processes in the message metadata.
1215 * systemd-cryptsetup has gained support for TrueCrypt volumes.
1217 * The initrd interface has been simplified (more specifically,
1218 support for passing performance data via environment
1219 variables and fsck results via files in /run has been
1220 removed). These features were non-essential, and are
1221 nowadays available in a much nicer way by having systemd in
1222 the initrd serialize its state and have the hosts systemd
1223 deserialize it again.
1225 * The udev "keymap" data files and tools to apply keyboard
1226 specific mappings of scan to key codes, and force-release
1227 scan code lists have been entirely replaced by a udev
1228 "keyboard" builtin and a hwdb data file.
1230 * systemd will now honour the kernel's "quiet" command line
1231 argument also during late shutdown, resulting in a
1232 completely silent shutdown when used.
1234 * There's now an option to control the SO_REUSEPORT socket
1235 option in .socket units.
1237 * Instance units will now automatically get a per-template
1238 subslice of system.slice unless something else is explicitly
1239 configured. For example, instances of sshd@.service will now
1240 implicitly be placed in system-sshd.slice rather than
1241 system.slice as before.
1243 * Test coverage support may now be enabled at build time.
1245 Contributions from: Dave Reisner, Frederic Crozat, Harald
1246 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
1247 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
1248 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
1249 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
1250 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
1251 Giokas, Zbigniew Jędrzejewski-Szmek
1253 -- Berlin, 2013-07-23
1257 * Two new unit types have been introduced:
1259 Scope units are very similar to service units, however, are
1260 created out of pre-existing processes -- instead of PID 1
1261 forking off the processes. By using scope units it is
1262 possible for system services and applications to group their
1263 own child processes (worker processes) in a powerful way
1264 which then maybe used to organize them, or kill them
1265 together, or apply resource limits on them.
1267 Slice units may be used to partition system resources in an
1268 hierarchial fashion and then assign other units to them. By
1269 default there are now three slices: system.slice (for all
1270 system services), user.slice (for all user sessions),
1271 machine.slice (for VMs and containers).
1273 Slices and scopes have been introduced primarily in
1274 context of the work to move cgroup handling to a
1275 single-writer scheme, where only PID 1
1276 creates/removes/manages cgroups.
1278 * There's a new concept of "transient" units. In contrast to
1279 normal units these units are created via an API at runtime,
1280 not from configuration from disk. More specifically this
1281 means it is now possible to run arbitrary programs as
1282 independent services, with all execution parameters passed
1283 in via bus APIs rather than read from disk. Transient units
1284 make systemd substantially more dynamic then it ever was,
1285 and useful as a general batch manager.
1287 * logind has been updated to make use of scope and slice units
1288 for managing user sessions. As a user logs in he will get
1289 his own private slice unit, to which all sessions are added
1290 as scope units. We also added support for automatically
1291 adding an instance of user@.service for the user into the
1292 slice. Effectively logind will no longer create cgroup
1293 hierarchies on its own now, it will defer entirely to PID 1
1294 for this by means of scope, service and slice units. Since
1295 user sessions this way become entities managed by PID 1
1296 the output of "systemctl" is now a lot more comprehensive.
1298 * A new mini-daemon "systemd-machined" has been added which
1299 may be used by virtualization managers to register local
1300 VMs/containers. nspawn has been updated accordingly, and
1301 libvirt will be updated shortly. machined will collect a bit
1302 of meta information about the VMs/containers, and assign
1303 them their own scope unit (see above). The collected
1304 meta-data is then made available via the "machinectl" tool,
1305 and exposed in "ps" and similar tools. machined/machinectl
1306 is compile-time optional.
1308 * As discussed earlier, the low-level cgroup configuration
1309 options ControlGroup=, ControlGroupModify=,
1310 ControlGroupPersistent=, ControlGroupAttribute= have been
1311 removed. Please use high-level attribute settings instead as
1312 well as slice units.
1314 * A new bus call SetUnitProperties() has been added to alter
1315 various runtime parameters of a unit. This is primarily
1316 useful to alter cgroup parameters dynamically in a nice way,
1317 but will be extended later on to make more properties
1318 modifiable at runtime. systemctl gained a new set-properties
1319 command that wraps this call.
1321 * A new tool "systemd-run" has been added which can be used to
1322 run arbitrary command lines as transient services or scopes,
1323 while configuring a number of settings via the command
1324 line. This tool is currently very basic, however already
1325 very useful. We plan to extend this tool to even allow
1326 queuing of execution jobs with time triggers from the
1327 command line, similar in fashion to "at".
1329 * nspawn will now inform the user explicitly that kernels with
1330 audit enabled break containers, and suggest the user to turn
1333 * Support for detecting the IMA and AppArmor security
1334 frameworks with ConditionSecurity= has been added.
1336 * journalctl gained a new "-k" switch for showing only kernel
1337 messages, mimicking dmesg output; in addition to "--user"
1338 and "--system" switches for showing only user's own logs
1341 * systemd-delta can now show information about drop-in
1342 snippets extending unit files.
1344 * libsystemd-bus has been substantially updated but is still
1345 not available as public API.
1347 * systemd will now look for the "debug" argument on the kernel
1348 command line and enable debug logging, similar to what
1349 "systemd.log_level=debug" already did before.
1351 * "systemctl set-default", "systemctl get-default" has been
1352 added to configure the default.target symlink, which
1353 controls what to boot into by default.
1355 * "systemctl set-log-level" has been added as a convenient
1356 way to raise and lower systemd logging threshold.
1358 * "systemd-analyze plot" will now show the time the various
1359 generators needed for execution, as well as information
1360 about the unit file loading.
1362 * libsystemd-journal gained a new sd_journal_open_files() call
1363 for opening specific journal files. journactl also gained a
1364 new switch to expose this new functionality. Previously we
1365 only supported opening all files from a directory, or all
1366 files from the system, as opening individual files only is
1367 racy due to journal file rotation.
1369 * systemd gained the new DefaultEnvironment= setting in
1370 /etc/systemd/system.conf to set environment variables for
1373 * If a privileged process logs a journal message with the
1374 OBJECT_PID= field set, then journald will automatically
1375 augment this with additional OBJECT_UID=, OBJECT_GID=,
1376 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
1377 system services want to log events about specific client
1378 processes. journactl/systemctl has been updated to make use
1379 of this information if all log messages regarding a specific
1382 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
1383 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
1384 Reisner, David Coppa, David King, David Strauss, Eelco
1385 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
1386 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
1387 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
1388 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
1389 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
1390 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
1391 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
1392 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
1393 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
1394 Łukasz Stelmach, 장동준
1398 * The Python bindings gained some minimal support for the APIs
1399 exposed by libsystemd-logind.
1401 * ConditionSecurity= gained support for detecting SMACK. Since
1402 this condition already supports SELinux and AppArmor we only
1403 miss IMA for this. Patches welcome!
1405 Contributions from: Karol Lewandowski, Lennart Poettering,
1406 Zbigniew Jędrzejewski-Szmek
1410 * systemd-nspawn will now create /etc/resolv.conf if
1411 necessary, before bind-mounting the host's file onto it.
1413 * systemd-nspawn will now store meta information about a
1414 container on the container's cgroup as extended attribute
1415 fields, including the root directory.
1417 * The cgroup hierarchy has been reworked in many ways. All
1418 objects any of the components systemd creates in the cgroup
1419 tree are now suffixed. More specifically, user sessions are
1420 now placed in cgroups suffixed with ".session", users in
1421 cgroups suffixed with ".user", and nspawn containers in
1422 cgroups suffixed with ".nspawn". Furthermore, all cgroup
1423 names are now escaped in a simple scheme to avoid collision
1424 of userspace object names with kernel filenames. This work
1425 is preparation for making these objects relocatable in the
1426 cgroup tree, in order to allow easy resource partitioning of
1427 these objects without causing naming conflicts.
1429 * systemctl list-dependencies gained the new switches
1430 --plain, --reverse, --after and --before.
1432 * systemd-inhibit now shows the process name of processes that
1433 have taken an inhibitor lock.
1435 * nss-myhostname will now also resolve "localhost"
1436 implicitly. This makes /etc/hosts an optional file and
1437 nicely handles that on IPv6 ::1 maps to both "localhost" and
1440 * libsystemd-logind.so gained a new call
1441 sd_get_machine_names() to enumerate running containers and
1442 VMs (currently only supported by very new libvirt and
1443 nspawn). sd_login_monitor can now be used to watch
1444 VMs/containers coming and going.
1446 * .include is not allowed recursively anymore, and only in
1447 unit files. Usually it is better to use drop-in snippets in
1448 .d/*.conf anyway, as introduced with systemd 198.
1450 * systemd-analyze gained a new "critical-chain" command that
1451 determines the slowest chain of units run during system
1452 boot-up. It is very useful for tracking down where
1453 optimizing boot time is the most beneficial.
1455 * systemd will no longer allow manipulating service paths in
1456 the name=systemd:/system cgroup tree using ControlGroup= in
1457 units. (But is still fine with it in all other dirs.)
1459 * There's a new systemd-nspawn@.service service file that may
1460 be used to easily run nspawn containers as system
1461 services. With the container's root directory in
1462 /var/lib/container/foobar it is now sufficient to run
1463 "systemctl start systemd-nspawn@foobar.service" to boot it.
1465 * systemd-cgls gained a new parameter "--machine" to list only
1466 the processes within a certain container.
1468 * ConditionSecurity= now can check for "apparmor". We still
1469 are lacking checks for SMACK and IMA for this condition
1470 check though. Patches welcome!
1472 * A new configuration file /etc/systemd/sleep.conf has been
1473 added that may be used to configure which kernel operation
1474 systemd is supposed to execute when "suspend", "hibernate"
1475 or "hybrid-sleep" is requested. This makes the new kernel
1476 "freeze" state accessible to the user.
1478 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
1479 the passed argument if applicable.
1481 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
1482 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
1483 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
1484 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
1485 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
1486 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
1487 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
1492 * The output of 'systemctl list-jobs' got some polishing. The
1493 '--type=' argument may now be passed more than once. A new
1494 command 'systemctl list-sockets' has been added which shows
1495 a list of kernel sockets systemd is listening on with the
1496 socket units they belong to, plus the units these socket
1499 * The experimental libsystemd-bus library got substantial
1500 updates to work in conjunction with the (also experimental)
1501 kdbus kernel project. It works well enough to exchange
1502 messages with some sophistication. Note that kdbus is not
1503 ready yet, and the library is mostly an elaborate test case
1504 for now, and not installable.
1506 * systemd gained a new unit 'systemd-static-nodes.service'
1507 that generates static device nodes earlier during boot, and
1508 can run in conjunction with udev.
1510 * libsystemd-login gained a new call sd_pid_get_user_unit()
1511 to retrieve the user systemd unit a process is running
1512 in. This is useful for systems where systemd is used as
1515 * systemd-nspawn now places all containers in the new /machine
1516 top-level cgroup directory in the name=systemd
1517 hierarchy. libvirt will soon do the same, so that we get a
1518 uniform separation of /system, /user and /machine for system
1519 services, user processes and containers/virtual
1520 machines. This new cgroup hierarchy is also useful to stick
1521 stable names to specific container instances, which can be
1522 recognized later this way (this name may be controlled
1523 via systemd-nspawn's new -M switch). libsystemd-login also
1524 gained a new call sd_pid_get_machine_name() to retrieve the
1525 name of the container/VM a specific process belongs to.
1527 * bootchart can now store its data in the journal.
1529 * libsystemd-journal gained a new call
1530 sd_journal_add_conjunction() for AND expressions to the
1531 matching logic. This can be used to express more complex
1532 logical expressions.
1534 * journactl can now take multiple --unit= and --user-unit=
1537 * The cryptsetup logic now understands the "luks.key=" kernel
1538 command line switch for specifying a file to read the
1539 decryption key from. Also, if a configured key file is not
1540 found the tool will now automatically fall back to prompting
1543 * Python systemd.journal module was updated to wrap recently
1544 added functions from libsystemd-journal. The interface was
1545 changed to bring the low level interface in s.j._Reader
1546 closer to the C API, and the high level interface in
1547 s.j.Reader was updated to wrap and convert all data about
1550 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
1551 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
1552 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
1553 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
1554 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
1555 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1559 * journalctl --update-catalog now understands a new --root=
1560 option to operate on catalogs found in a different root
1563 * During shutdown after systemd has terminated all running
1564 services a final killing loop kills all remaining left-over
1565 processes. We will now print the name of these processes
1566 when we send SIGKILL to them, since this usually indicates a
1569 * If /etc/crypttab refers to password files stored on
1570 configured mount points automatic dependencies will now be
1571 generated to ensure the specific mount is established first
1572 before the key file is attempted to be read.
1574 * 'systemctl status' will now show information about the
1575 network sockets a socket unit is listening on.
1577 * 'systemctl status' will also shown information about any
1578 drop-in configuration file for units. (Drop-In configuration
1579 files in this context are files such as
1580 /etc/systemd/systemd/foobar.service.d/*.conf)
1582 * systemd-cgtop now optionally shows summed up CPU times of
1583 cgroups. Press '%' while running cgtop to switch between
1584 percentage and absolute mode. This is useful to determine
1585 which cgroups use up the most CPU time over the entire
1586 runtime of the system. systemd-cgtop has also been updated
1587 to be 'pipeable' for processing with further shell tools.
1589 * 'hostnamectl set-hostname' will now allow setting of FQDN
1592 * The formatting and parsing of time span values has been
1593 changed. The parser now understands fractional expressions
1594 such as "5.5h". The formatter will now output fractional
1595 expressions for all time spans under 1min, i.e. "5.123456s"
1596 rather than "5s 123ms 456us". For time spans under 1s
1597 millisecond values are shown, for those under 1ms
1598 microsecond values are shown. This should greatly improve
1599 all time-related output of systemd.
1601 * libsystemd-login and libsystemd-journal gained new
1602 functions for querying the poll() events mask and poll()
1603 timeout value for integration into arbitrary event
1606 * localectl gained the ability to list available X11 keymaps
1607 (models, layouts, variants, options).
1609 * 'systemd-analyze dot' gained the ability to filter for
1610 specific units via shell-style globs, to create smaller,
1611 more useful graphs. I.e. it is now possible to create simple
1612 graphs of all the dependencies between only target units, or
1613 of all units that Avahi has dependencies with.
1615 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
1616 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
1617 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
1618 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
1619 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
1620 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
1621 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
1625 * The boot-time readahead implementation for rotating media
1626 will now read the read-ahead data in multiple passes which
1627 consist of all read requests made in equidistant time
1628 intervals. This means instead of strictly reading read-ahead
1629 data in its physical order on disk we now try to find a
1630 middle ground between physical and access time order.
1632 * /etc/os-release files gained a new BUILD_ID= field for usage
1633 on operating systems that provide continuous builds of OS
1636 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
1637 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
1638 William Douglas, Zbigniew Jędrzejewski-Szmek
1642 * systemd-python gained an API exposing libsystemd-daemon.
1644 * The SMACK setup logic gained support for uploading CIPSO
1647 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
1648 ReadOnlyDirectories= and InaccessibleDirectories= has
1649 changed. The private /tmp and /var/tmp directories are now
1650 shared by all processes of a service (which means
1651 ExecStartPre= may now leave data in /tmp that ExecStart= of
1652 the same service can still access). When a service is
1653 stopped its temporary directories are immediately deleted
1654 (normal clean-up with tmpfiles is still done in addition to
1657 * By default, systemd will now set a couple of sysctl
1658 variables in the kernel: the safe sysrq options are turned
1659 on, IP route verification is turned on, and source routing
1660 disabled. The recently added hardlink and softlink
1661 protection of the kernel is turned on. These settings should
1662 be reasonably safe, and good defaults for all new systems.
1664 * The predictable network naming logic may now be turned off
1665 with a new kernel command line switch: net.ifnames=0.
1667 * A new libsystemd-bus module has been added that implements a
1668 pretty complete D-Bus client library. For details see:
1670 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
1672 * journald will now explicitly flush the journal files to disk
1673 at the latest 5min after each write. The file will then also
1674 be marked offline until the next write. This should increase
1675 reliability in case of a crash. The synchronization delay
1676 can be configured via SyncIntervalSec= in journald.conf.
1678 * There's a new remote-fs-setup.target unit that can be used
1679 to pull in specific services when at least one remote file
1680 system is to be mounted.
1682 * There are new targets timers.target and paths.target as
1683 canonical targets to pull user timer and path units in
1684 from. This complements sockets.target with a similar
1685 purpose for socket units.
1687 * libudev gained a new call udev_device_set_attribute_value()
1688 to set sysfs attributes of a device.
1690 * The udev daemon now sets the default number of worker
1691 processes executed in parallel based on the number of available
1692 CPUs instead of the amount of available RAM. This is supposed
1693 to provide a more reliable default and limit a too aggressive
1694 paralellism for setups with 1000s of devices connected.
1696 Contributions from: Auke Kok, Colin Walters, Cristian
1697 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
1698 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
1699 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
1700 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
1701 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
1702 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
1703 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
1704 Zbigniew Jędrzejewski-Szmek
1708 * Configuration of unit files may now be extended via drop-in
1709 files without having to edit/override the unit files
1710 themselves. More specifically, if the administrator wants to
1711 change one value for a service file foobar.service he can
1712 now do so by dropping in a configuration snippet into
1713 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
1714 will load all these snippets and apply them on top of the
1715 main unit configuration file, possibly extending or
1716 overriding its settings. Using these drop-in snippets is
1717 generally nicer than the two earlier options for changing
1718 unit files locally: copying the files from
1719 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
1720 them there; or creating a new file in /etc/systemd/system/
1721 that incorporates the original one via ".include". Drop-in
1722 snippets into these .d/ directories can be placed in any
1723 directory systemd looks for units in, and the usual
1724 overriding semantics between /usr/lib, /etc and /run apply
1727 * Most unit file settings which take lists of items can now be
1728 reset by assigning the empty string to them. For example,
1729 normally, settings such as Environment=FOO=BAR append a new
1730 environment variable assignment to the environment block,
1731 each time they are used. By assigning Environment= the empty
1732 string the environment block can be reset to empty. This is
1733 particularly useful with the .d/*.conf drop-in snippets
1734 mentioned above, since this adds the ability to reset list
1735 settings from vendor unit files via these drop-ins.
1737 * systemctl gained a new "list-dependencies" command for
1738 listing the dependencies of a unit recursively.
1740 * Inhibitors are now honored and listed by "systemctl
1741 suspend", "systemctl poweroff" (and similar) too, not only
1742 GNOME. These commands will also list active sessions by
1745 * Resource limits (as exposed by the various control group
1746 controllers) can now be controlled dynamically at runtime
1747 for all units. More specifically, you can now use a command
1748 like "systemctl set-cgroup-attr foobar.service cpu.shares
1749 2000" to alter the CPU shares a specific service gets. These
1750 settings are stored persistently on disk, and thus allow the
1751 administrator to easily adjust the resource usage of
1752 services with a few simple commands. This dynamic resource
1753 management logic is also available to other programs via the
1754 bus. Almost any kernel cgroup attribute and controller is
1757 * systemd-vconsole-setup will now copy all font settings to
1758 all allocated VTs, where it previously applied them only to
1761 * libsystemd-login gained the new sd_session_get_tty() API
1764 * This release drops support for a few legacy or
1765 distribution-specific LSB facility names when parsing init
1766 scripts: $x-display-manager, $mail-transfer-agent,
1767 $mail-transport-agent, $mail-transfer-agent, $smtp,
1768 $null. Also, the mail-transfer-agent.target unit backing
1769 this has been removed. Distributions which want to retain
1770 compatibility with this should carry the burden for
1771 supporting this themselves and patch support for these back
1772 in, if they really need to. Also, the facilities $syslog and
1773 $local_fs are now ignored, since systemd does not support
1774 early-boot LSB init scripts anymore, and these facilities
1775 are implied anyway for normal services. syslog.target has
1778 * There are new bus calls on PID1's Manager object for
1779 cancelling jobs, and removing snapshot units. Previously,
1780 both calls were only available on the Job and Snapshot
1783 * systemd-journal-gatewayd gained SSL support.
1785 * The various "environment" files, such as /etc/locale.conf
1786 now support continuation lines with a backslash ("\") as
1787 last character in the line, similarly in style (but different)
1788 to how this is supported in shells.
1790 * For normal user processes the _SYSTEMD_USER_UNIT= field is
1791 now implicitly appended to every log entry logged. systemctl
1792 has been updated to filter by this field when operating on a
1793 user systemd instance.
1795 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
1796 CAP_AUDIT_CONTROL capabilities to the capabilities set for
1797 the container. This makes it easier to boot unmodified
1798 Fedora systems in a container, which however still requires
1799 audit=0 to be passed on the kernel command line. Auditing in
1800 kernel and userspace is unfortunately still too broken in
1801 context of containers, hence we recommend compiling it out
1802 of the kernel or using audit=0. Hopefully this will be fixed
1803 one day for good in the kernel.
1805 * nspawn gained the new --bind= and --bind-ro= parameters to
1806 bind mount specific directories from the host into the
1809 * nspawn will now mount its own devpts file system instance
1810 into the container, in order not to leak pty devices from
1811 the host into the container.
1813 * systemd will now read the firmware boot time performance
1814 information from the EFI variables, if the used boot loader
1815 supports this, and takes it into account for boot performance
1816 analysis via "systemd-analyze". This is currently supported
1817 only in conjunction with Gummiboot, but could be supported
1818 by other boot loaders too. For details see:
1820 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
1822 * A new generator has been added that automatically mounts the
1823 EFI System Partition (ESP) to /boot, if that directory
1824 exists, is empty, and no other file system has been
1825 configured to be mounted there.
1827 * logind will now send out PrepareForSleep(false) out
1828 unconditionally, after coming back from suspend. This may be
1829 used by applications as asynchronous notification for
1830 system resume events.
1832 * "systemctl unlock-sessions" has been added, that allows
1833 unlocking the screens of all user sessions at once, similar
1834 to how "systemctl lock-sessions" already locked all users
1835 sessions. This is backed by a new D-Bus call UnlockSessions().
1837 * "loginctl seat-status" will now show the master device of a
1838 seat. (i.e. the device of a seat that needs to be around for
1839 the seat to be considered available, usually the graphics
1842 * tmpfiles gained a new "X" line type, that allows
1843 configuration of files and directories (with wildcards) that
1844 shall be excluded from automatic cleanup ("aging").
1846 * udev default rules set the device node permissions now only
1847 at "add" events, and do not change them any longer with a
1848 later "change" event.
1850 * The log messages for lid events and power/sleep keypresses
1851 now carry a message ID.
1853 * We now have a substantially larger unit test suite, but this
1854 continues to be work in progress.
1856 * udevadm hwdb gained a new --root= parameter to change the
1857 root directory to operate relative to.
1859 * logind will now issue a background sync() request to the kernel
1860 early at shutdown, so that dirty buffers are flushed to disk early
1861 instead of at the last moment, in order to optimize shutdown
1864 * A new bootctl tool has been added that is an interface for
1865 certain boot loader operations. This is currently a preview
1866 and is likely to be extended into a small mechanism daemon
1867 like timedated, localed, hostnamed, and can be used by
1868 graphical UIs to enumerate available boot options, and
1869 request boot into firmware operations.
1871 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
1872 the rest of the package. It also has been updated to work
1873 correctly in initrds.
1875 * Policykit previously has been runtime optional, and is now
1876 also compile time optional via a configure switch.
1878 * systemd-analyze has been reimplemented in C. Also "systemctl
1879 dot" has moved into systemd-analyze.
1881 * "systemctl status" with no further parameters will now print
1882 the status of all active or failed units.
1884 * Operations such as "systemctl start" can now be executed
1885 with a new mode "--irreversible" which may be used to queue
1886 operations that cannot accidentally be reversed by a later
1887 job queuing. This is by default used to make shutdown
1888 requests more robust.
1890 * The Python API of systemd now gained a new module for
1891 reading journal files.
1893 * A new tool kernel-install has been added that can install
1894 kernel images according to the Boot Loader Specification:
1896 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
1898 * Boot time console output has been improved to provide
1899 animated boot time output for hanging jobs.
1901 * A new tool systemd-activate has been added which can be used
1902 to test socket activation with, directly from the command
1903 line. This should make it much easier to test and debug
1904 socket activation in daemons.
1906 * journalctl gained a new "--reverse" (or -r) option to show
1907 journal output in reverse order (i.e. newest line first).
1909 * journalctl gained a new "--pager-end" (or -e) option to jump
1910 to immediately jump to the end of the journal in the
1911 pager. This is only supported in conjunction with "less".
1913 * journalctl gained a new "--user-unit=" option, that works
1914 similarly to "--unit=" but filters for user units rather than
1917 * A number of unit files to ease adoption of systemd in
1918 initrds has been added. This moves some minimal logic from
1919 the various initrd implementations into systemd proper.
1921 * The journal files are now owned by a new group
1922 "systemd-journal", which exists specifically to allow access
1923 to the journal, and nothing else. Previously, we used the
1924 "adm" group for that, which however possibly covers more
1925 than just journal/log file access. This new group is now
1926 already used by systemd-journal-gatewayd to ensure this
1927 daemon gets access to the journal files and as little else
1928 as possible. Note that "make install" will also set FS ACLs
1929 up for /var/log/journal to give "adm" and "wheel" read
1930 access to it, in addition to "systemd-journal" which owns
1931 the journal files. We recommend that packaging scripts also
1932 add read access to "adm" + "wheel" to /var/log/journal, and
1933 all existing/future journal files. To normal users and
1934 administrators little changes, however packagers need to
1935 ensure to create the "systemd-journal" system group at
1936 package installation time.
1938 * The systemd-journal-gatewayd now runs as unprivileged user
1939 systemd-journal-gateway:systemd-journal-gateway. Packaging
1940 scripts need to create these system user/group at
1943 * timedated now exposes a new boolean property CanNTP that
1944 indicates whether a local NTP service is available or not.
1946 * systemd-detect-virt will now also detect xen PVs
1948 * The pstore file system is now mounted by default, if it is
1951 * In addition to the SELinux and IMA policies we will now also
1952 load SMACK policies at early boot.
1954 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
1955 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
1956 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
1957 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
1958 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
1959 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
1960 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
1961 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
1962 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
1963 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
1964 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
1965 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
1966 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
1967 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
1971 * Timer units now support calendar time events in addition to
1972 monotonic time events. That means you can now trigger a unit
1973 based on a calendar time specification such as "Thu,Fri
1974 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
1975 or fifth day of any month of the year 2013, given that it is
1976 a thursday or friday. This brings timer event support
1977 considerably closer to cron's capabilities. For details on
1978 the supported calendar time specification language see
1981 * udev now supports a number of different naming policies for
1982 network interfaces for predictable names, and a combination
1983 of these policies is now the default. Please see this wiki
1984 document for details:
1986 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
1988 * Auke Kok's bootchart implementation has been added to the
1989 systemd tree. It is an optional component that can graph the
1990 boot in quite some detail. It is one of the best bootchart
1991 implementations around and minimal in its code and
1994 * nss-myhostname has been integrated into the systemd source
1995 tree. nss-myhostname guarantees that the local hostname
1996 always stays resolvable via NSS. It has been a weak
1997 requirement of systemd-hostnamed since a long time, and
1998 since its code is actually trivial we decided to just
1999 include it in systemd's source tree. It can be turned off
2000 with a configure switch.
2002 * The read-ahead logic is now capable of properly detecting
2003 whether a btrfs file system is on SSD or rotating media, in
2004 order to optimize the read-ahead scheme. Previously, it was
2005 only capable of detecting this on traditional file systems
2008 * In udev, additional device properties are now read from the
2009 IAB in addition to the OUI database. Also, Bluetooth company
2010 identities are attached to the devices as well.
2012 * In service files %U may be used as specifier that is
2013 replaced by the configured user name of the service.
2015 * nspawn may now be invoked without a controlling TTY. This
2016 makes it suitable for invocation as its own service. This
2017 may be used to set up a simple containerized server system
2018 using only core OS tools.
2020 * systemd and nspawn can now accept socket file descriptors
2021 when they are started for socket activation. This enables
2022 implementation of socket activated nspawn
2023 containers. i.e. think about autospawning an entire OS image
2024 when the first SSH or HTTP connection is received. We expect
2025 that similar functionality will also be added to libvirt-lxc
2028 * journalctl will now suppress ANSI color codes when
2029 presenting log data.
2031 * systemctl will no longer show control group information for
2032 a unit if a the control group is empty anyway.
2034 * logind can now automatically suspend/hibernate/shutdown the
2037 * /etc/machine-info and hostnamed now also expose the chassis
2038 type of the system. This can be used to determine whether
2039 the local system is a laptop, desktop, handset or
2040 tablet. This information may either be configured by the
2041 user/vendor or is automatically determined from ACPI and DMI
2042 information if possible.
2044 * A number of PolicyKit actions are now bound together with
2045 "imply" rules. This should simplify creating UIs because
2046 many actions will now authenticate similar ones as well.
2048 * Unit files learnt a new condition ConditionACPower= which
2049 may be used to conditionalize a unit depending on whether an
2050 AC power source is connected or not, of whether the system
2051 is running on battery power.
2053 * systemctl gained a new "is-failed" verb that may be used in
2054 shell scripts and suchlike to check whether a specific unit
2055 is in the "failed" state.
2057 * The EnvironmentFile= setting in unit files now supports file
2058 globbing, and can hence be used to easily read a number of
2059 environment files at once.
2061 * systemd will no longer detect and recognize specific
2062 distributions. All distribution-specific #ifdeffery has been
2063 removed, systemd is now fully generic and
2064 distribution-agnostic. Effectively, not too much is lost as
2065 a lot of the code is still accessible via explicit configure
2066 switches. However, support for some distribution specific
2067 legacy configuration file formats has been dropped. We
2068 recommend distributions to simply adopt the configuration
2069 files everybody else uses now and convert the old
2070 configuration from packaging scripts. Most distributions
2071 already did that. If that's not possible or desirable,
2072 distributions are welcome to forward port the specific
2073 pieces of code locally from the git history.
2075 * When logging a message about a unit systemd will now always
2076 log the unit name in the message meta data.
2078 * localectl will now also discover system locale data that is
2079 not stored in locale archives, but directly unpacked.
2081 * logind will no longer unconditionally use framebuffer
2082 devices as seat masters, i.e. as devices that are required
2083 to be existing before a seat is considered preset. Instead,
2084 it will now look for all devices that are tagged as
2085 "seat-master" in udev. By default framebuffer devices will
2086 be marked as such, but depending on local systems other
2087 devices might be marked as well. This may be used to
2088 integrate graphics cards using closed source drivers (such
2089 as NVidia ones) more nicely into logind. Note however, that
2090 we recommend using the open source NVidia drivers instead,
2091 and no udev rules for the closed-source drivers will be
2092 shipped from us upstream.
2094 Contributions from: Adam Williamson, Alessandro Crismani, Auke
2095 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
2096 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
2097 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
2098 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
2099 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
2100 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
2101 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
2102 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
2103 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
2104 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
2105 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
2110 * udev gained support for loading additional device properties
2111 from an indexed database that is keyed by vendor/product IDs
2112 and similar device identifiers. For the beginning this
2113 "hwdb" is populated with data from the well-known PCI and
2114 USB database, but also includes PNP, ACPI and OID data. In
2115 the longer run this indexed database shall grow into
2116 becoming the one central database for non-essential
2117 userspace device metadata. Previously, data from the PCI/USB
2118 database was only attached to select devices, since the
2119 lookup was a relatively expensive operation due to O(n) time
2120 complexity (with n being the number of entries in the
2121 database). Since this is now O(1), we decided to add in this
2122 data for all devices where this is available, by
2123 default. Note that the indexed database needs to be rebuilt
2124 when new data files are installed. To achieve this you need
2125 to update your packaging scripts to invoke "udevadm hwdb
2126 --update" after installation of hwdb data files. For
2127 RPM-based distributions we introduced the new
2128 %udev_hwdb_update macro for this purpose.
2130 * The Journal gained support for the "Message Catalog", an
2131 indexed database to link up additional information with
2132 journal entries. For further details please check:
2134 http://www.freedesktop.org/wiki/Software/systemd/catalog
2136 The indexed message catalog database also needs to be
2137 rebuilt after installation of message catalog files. Use
2138 "journalctl --update-catalog" for this. For RPM-based
2139 distributions we introduced the %journal_catalog_update
2140 macro for this purpose.
2142 * The Python Journal bindings gained support for the standard
2143 Python logging framework.
2145 * The Journal API gained new functions for checking whether
2146 the underlying file system of a journal file is capable of
2147 properly reporting file change notifications, or whether
2148 applications that want to reflect journal changes "live"
2149 need to recheck journal files continuously in appropriate
2152 * It is now possible to set the "age" field for tmpfiles
2153 entries to 0, indicating that files matching this entry
2154 shall always be removed when the directories are cleaned up.
2156 * coredumpctl gained a new "gdb" verb which invokes gdb
2157 right-away on the selected coredump.
2159 * There's now support for "hybrid sleep" on kernels that
2160 support this, in addition to "suspend" and "hibernate". Use
2161 "systemctl hybrid-sleep" to make use of this.
2163 * logind's HandleSuspendKey= setting (and related settings)
2164 now gained support for a new "lock" setting to simply
2165 request the screen lock on all local sessions, instead of
2166 actually executing a suspend or hibernation.
2168 * systemd will now mount the EFI variables file system by
2171 * Socket units now gained support for configuration of the
2172 SMACK security label.
2174 * timedatectl will now output the time of the last and next
2175 daylight saving change.
2177 * We dropped support for various legacy and distro-specific
2178 concepts, such as insserv, early-boot SysV services
2179 (i.e. those for non-standard runlevels such as 'b' or 'S')
2180 or ArchLinux /etc/rc.conf support. We recommend the
2181 distributions who still need support this to either continue
2182 to maintain the necessary patches downstream, or find a
2183 different solution. (Talk to us if you have questions!)
2185 * Various systemd components will now bypass PolicyKit checks
2186 for root and otherwise handle properly if PolicyKit is not
2187 found to be around. This should fix most issues for
2188 PolicyKit-less systems. Quite frankly this should have been
2189 this way since day one. It is absolutely our intention to
2190 make systemd work fine on PolicyKit-less systems, and we
2191 consider it a bug if something does not work as it should if
2192 PolicyKit is not around.
2194 * For embedded systems it is now possible to build udev and
2195 systemd without blkid and/or kmod support.
2197 * "systemctl switch-root" is now capable of switching root
2198 more than once. I.e. in addition to transitions from the
2199 initrd to the host OS it is now possible to transition to
2200 further OS images from the host. This is useful to implement
2201 offline updating tools.
2203 * Various other additions have been made to the RPM macros
2204 shipped with systemd. Use %udev_rules_update() after
2205 installing new udev rules files. %_udevhwdbdir,
2206 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
2207 %_sysctldir are now available which resolve to the right
2208 directories for packages to place various data files in.
2210 * journalctl gained the new --full switch (in addition to
2211 --all, to disable ellipsation for long messages.
2213 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
2214 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
2215 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
2216 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
2217 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
2218 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
2219 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
2220 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
2221 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
2225 * journalctl gained new --since= and --until= switches to
2226 filter by time. It also now supports nice filtering for
2227 units via --unit=/-u.
2229 * Type=oneshot services may use ExecReload= and do the
2232 * The journal daemon now supports time-based rotation and
2233 vacuuming, in addition to the usual disk-space based
2236 * The journal will now index the available field values for
2237 each field name. This enables clients to show pretty drop
2238 downs of available match values when filtering. The bash
2239 completion of journalctl has been updated
2240 accordingly. journalctl gained a new switch -F to list all
2241 values a certain field takes in the journal database.
2243 * More service events are now written as structured messages
2244 to the journal, and made recognizable via message IDs.
2246 * The timedated, localed and hostnamed mini-services which
2247 previously only provided support for changing time, locale
2248 and hostname settings from graphical DEs such as GNOME now
2249 also have a minimal (but very useful) text-based client
2250 utility each. This is probably the nicest way to changing
2251 these settings from the command line now, especially since
2252 it lists available options and is fully integrated with bash
2255 * There's now a new tool "systemd-coredumpctl" to list and
2256 extract coredumps from the journal.
2258 * We now install a README each in /var/log/ and
2259 /etc/rc.d/init.d explaining where the system logs and init
2260 scripts went. This hopefully should help folks who go to
2261 that dirs and look into the otherwise now empty void and
2262 scratch their heads.
2264 * When user-services are invoked (by systemd --user) the
2265 $MANAGERPID env var is set to the PID of systemd.
2267 * SIGRTMIN+24 when sent to a --user instance will now result
2268 in immediate termination of systemd.
2270 * gatewayd received numerous feature additions such as a
2271 "follow" mode, for live syncing and filtering.
2273 * browse.html now allows filtering and showing detailed
2274 information on specific entries. Keyboard navigation and
2275 mouse screen support has been added.
2277 * gatewayd/journalctl now supports HTML5/JSON
2278 Server-Sent-Events as output.
2280 * The SysV init script compatibility logic will now
2281 heuristically determine whether a script supports the
2282 "reload" verb, and only then make this available as
2285 * "systemctl status --follow" has been removed, use "journalctl
2288 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
2289 have been removed since they are hardly useful to be
2292 * And I'd like to take the opportunity to specifically mention
2293 Zbigniew for his great contributions. Zbigniew, you rock!
2295 Contributions from: Andrew Eikum, Christian Hesse, Colin
2296 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
2297 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
2298 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
2299 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
2300 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
2301 Jędrzejewski-Szmek, Сковорода Никита Андреевич
2305 * If /etc/vconsole.conf is non-existent or empty we will no
2306 longer load any console font or key map at boot by
2307 default. Instead the kernel defaults will be left
2308 intact. This is definitely the right thing to do, as no
2309 configuration should mean no configuration, and hard-coding
2310 font names that are different on all archs is probably a bad
2311 idea. Also, the kernel default key map and font should be
2312 good enough for most cases anyway, and mostly identical to
2313 the userspace fonts/key maps we previously overloaded them
2314 with. If distributions want to continue to default to a
2315 non-kernel font or key map they should ship a default
2316 /etc/vconsole.conf with the appropriate contents.
2318 Contributions from: Colin Walters, Daniel J Walsh, Dave
2319 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
2320 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
2324 * journalctl gained a new --cursor= switch to show entries
2325 starting from the specified location in the journal.
2327 * We now enforce a size limit on journal entry fields exported
2328 with "-o json" in journalctl. Fields larger than 4K will be
2329 assigned null. This can be turned off with --all.
2331 * An (optional) journal gateway daemon is now available as
2332 "systemd-journal-gatewayd.service". This service provides
2333 access to the journal via HTTP and JSON. This functionality
2334 will be used to implement live log synchronization in both
2335 pull and push modes, but has various other users too, such
2336 as easy log access for debugging of embedded devices. Right
2337 now it is already useful to retrieve the journal via HTTP:
2339 # systemctl start systemd-journal-gatewayd.service
2340 # wget http://localhost:19531/entries
2342 This will download the journal contents in a
2343 /var/log/messages compatible format. The same as JSON:
2345 # curl -H"Accept: application/json" http://localhost:19531/entries
2347 This service is also accessible via a web browser where a
2348 single static HTML5 app is served that uses the JSON logic
2349 to enable the user to do some basic browsing of the
2350 journal. This will be extended later on. Here's an example
2351 screenshot of this app in its current state:
2353 http://0pointer.de/public/journal-gatewayd
2355 Contributions from: Kay Sievers, Lennart Poettering, Robert
2356 Milasan, Tom Gundersen
2360 * The bash completion logic is now available for journalctl
2363 * We do not mount the "cpuset" controller anymore together with
2364 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
2365 started if no parameters are assigned to it. "cpuset" hence
2366 broke code that assumed it it could create "cpu" groups and
2369 * journalctl -f will now subscribe to terminal size changes,
2370 and line break accordingly.
2372 Contributions from: Dave Reisner, Kay Sievers, Lennart
2373 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
2377 * nspawn will now create a symlink /etc/localtime in the
2378 container environment, copying the host's timezone
2379 setting. Previously this has been done via a bind mount, but
2380 since symlinks cannot be bind mounted this has now been
2381 changed to create/update the appropriate symlink.
2383 * journalctl -n's line number argument is now optional, and
2384 will default to 10 if omitted.
2386 * journald will now log the maximum size the journal files may
2387 take up on disk. This is particularly useful if the default
2388 built-in logic of determining this parameter from the file
2389 system size is used. Use "systemctl status
2390 systemd-journald.service" to see this information.
2392 * The multi-seat X wrapper tool has been stripped down. As X
2393 is now capable of enumerating graphics devices via udev in a
2394 seat-aware way the wrapper is not strictly necessary
2395 anymore. A stripped down temporary stop-gap is still shipped
2396 until the upstream display managers have been updated to
2397 fully support the new X logic. Expect this wrapper to be
2398 removed entirely in one of the next releases.
2400 * HandleSleepKey= in logind.conf has been split up into
2401 HandleSuspendKey= and HandleHibernateKey=. The old setting
2402 is not available anymore. X11 and the kernel are
2403 distuingishing between these keys and we should too. This
2404 also means the inhibition lock for these keys has been split
2407 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
2408 Poettering, Lukas Nykryn, Václav Pavlín
2412 * Whenever a unit changes state we will now log this to the
2413 journal and show along the unit's own log output in
2416 * ConditionPathIsMountPoint= can now properly detect bind
2417 mount points too. (Previously, a bind mount of one file
2418 system to another place in the same file system could not be
2419 detected as mount, since they shared struct stat's st_dev
2422 * We will now mount the cgroup controllers cpu, cpuacct,
2423 cpuset and the controllers net_cls, net_prio together by
2426 * nspawn containers will now have a virtualized boot
2427 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
2428 over with a randomized ID at container initialization). This
2429 has the effect of making "journalctl -b" do the right thing
2432 * The JSON output journal serialization has been updated not
2433 to generate "endless" list objects anymore, but rather one
2434 JSON object per line. This is more in line how most JSON
2435 parsers expect JSON objects. The new output mode
2436 "json-pretty" has been added to provide similar output, but
2437 neatly aligned for readability by humans.
2439 * We dropped all explicit sync() invocations in the shutdown
2440 code. The kernel does this implicitly anyway in the kernel
2441 reboot() syscall. halt(8)'s -n option is now a compatibility
2444 * We now support virtualized reboot() in containers, as
2445 supported by newer kernels. We will fall back to exit() if
2446 CAP_SYS_REBOOT is not available to the container. Also,
2447 nspawn makes use of this now and will actually reboot the
2448 container if the containerized OS asks for that.
2450 * journalctl will only show local log output by default
2451 now. Use --merge (-m) to show remote log output, too.
2453 * libsystemd-journal gained the new sd_journal_get_usage()
2454 call to determine the current disk usage of all journal
2455 files. This is exposed in the new "journalctl --disk-usage"
2458 * journald gained a new configuration setting SplitMode= in
2459 journald.conf which may be used to control how user journals
2460 are split off. See journald.conf(5) for details.
2462 * A new condition type ConditionFileNotEmpty= has been added.
2464 * tmpfiles' "w" lines now support file globbing, to write
2465 multiple files at once.
2467 * We added Python bindings for the journal submission
2468 APIs. More Python APIs for a number of selected APIs will
2469 likely follow. Note that we intend to add native bindings
2470 only for the Python language, as we consider it common
2471 enough to deserve bindings shipped within systemd. There are
2472 various projects outside of systemd that provide bindings
2473 for languages such as PHP or Lua.
2475 * Many conditions will now resolve specifiers such as %i. In
2476 addition, PathChanged= and related directives of .path units
2477 now support specifiers as well.
2479 * There's now a new RPM macro definition for the system preset
2482 * journald will now warn if it ca not forward a message to the
2483 syslog daemon because its socket is full.
2485 * timedated will no longer write or process /etc/timezone,
2486 except on Debian. As we do not support late mounted /usr
2487 anymore /etc/localtime always being a symlink is now safe,
2488 and hence the information in /etc/timezone is not necessary
2491 * logind will now always reserve one VT for a text getty (VT6
2492 by default). Previously if more than 6 X sessions where
2493 started they took up all the VTs with auto-spawned gettys,
2494 so that no text gettys were available anymore.
2496 * udev will now automatically inform the btrfs kernel logic
2497 about btrfs RAID components showing up. This should make
2498 simple hotplug based btrfs RAID assembly work.
2500 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
2501 (but not for its children which will stay at the kernel
2502 default). This should allow setups with a lot more listening
2505 * systemd will now always pass the configured timezone to the
2506 kernel at boot. timedated will do the same when the timezone
2509 * logind's inhibition logic has been updated. By default,
2510 logind will now handle the lid switch, the power and sleep
2511 keys all the time, even in graphical sessions. If DEs want
2512 to handle these events on their own they should take the new
2513 handle-power-key, handle-sleep-key and handle-lid-switch
2514 inhibitors during their runtime. A simple way to achiveve
2515 that is to invoke the DE wrapped in an invocation of:
2517 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
2519 * Access to unit operations is now checked via SELinux taking
2520 the unit file label and client process label into account.
2522 * systemd will now notify the administrator in the journal
2523 when he over-mounts a non-empty directory.
2525 * There are new specifiers that are resolved in unit files,
2526 for the host name (%H), the machine ID (%m) and the boot ID
2529 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
2530 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
2531 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
2532 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
2533 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
2534 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
2535 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
2539 * Support for reading structured kernel messages from
2540 /dev/kmsg has now been added and is enabled by default.
2542 * Support for reading kernel messages from /proc/kmsg has now
2543 been removed. If you want kernel messages in the journal
2544 make sure to run a recent kernel (>= 3.5) that supports
2545 reading structured messages from /dev/kmsg (see
2546 above). /proc/kmsg is now exclusive property of classic
2547 syslog daemons again.
2549 * The libudev API gained the new
2550 udev_device_new_from_device_id() call.
2552 * The logic for file system namespace (ReadOnlyDirectory=,
2553 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
2554 require pivot_root() anymore. This means fewer temporary
2555 directories are created below /tmp for this feature.
2557 * nspawn containers will now see and receive all submounts
2558 made on the host OS below the root file system of the
2561 * Forward Secure Sealing is now supported for Journal files,
2562 which provide cryptographical sealing of journal files so
2563 that attackers cannot alter log history anymore without this
2564 being detectable. Lennart will soon post a blog story about
2565 this explaining it in more detail.
2567 * There are two new service settings RestartPreventExitStatus=
2568 and SuccessExitStatus= which allow configuration of exit
2569 status (exit code or signal) which will be excepted from the
2570 restart logic, resp. consider successful.
2572 * journalctl gained the new --verify switch that can be used
2573 to check the integrity of the structure of journal files and
2574 (if Forward Secure Sealing is enabled) the contents of
2577 * nspawn containers will now be run with /dev/stdin, /dev/fd/
2578 and similar symlinks pre-created. This makes running shells
2579 as container init process a lot more fun.
2581 * The fstab support can now handle PARTUUID= and PARTLABEL=
2584 * A new ConditionHost= condition has been added to match
2585 against the hostname (with globs) and machine ID. This is
2586 useful for clusters where a single OS image is used to
2587 provision a large number of hosts which shall run slightly
2588 different sets of services.
2590 * Services which hit the restart limit will now be placed in a
2593 Contributions from: Bertram Poettering, Dave Reisner, Huang
2594 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
2595 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
2599 * When running in --user mode systemd will now become a
2600 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
2601 tree a lot more organized.
2603 * A new PartOf= unit dependency type has been introduced that
2604 may be used to group services in a natural way.
2606 * "systemctl enable" may now be used to enable instances of
2609 * journalctl now prints error log levels in red, and
2610 warning/notice log levels in bright white. It also supports
2611 filtering by log level now.
2613 * cgtop gained a new -n switch (similar to top), to configure
2614 the maximum number of iterations to run for. It also gained
2615 -b, to run in batch mode (accepting no input).
2617 * The suffix ".service" may now be omitted on most systemctl
2618 command lines involving service unit names.
2620 * There's a new bus call in logind to lock all sessions, as
2621 well as a loginctl verb for it "lock-sessions".
2623 * libsystemd-logind.so gained a new call sd_journal_perror()
2624 that works similar to libc perror() but logs to the journal
2625 and encodes structured information about the error number.
2627 * /etc/crypttab entries now understand the new keyfile-size=
2630 * shutdown(8) now can send a (configurable) wall message when
2631 a shutdown is cancelled.
2633 * The mount propagation mode for the root file system will now
2634 default to "shared", which is useful to make containers work
2635 nicely out-of-the-box so that they receive new mounts from
2636 the host. This can be undone locally by running "mount
2637 --make-rprivate /" if needed.
2639 * The prefdm.service file has been removed. Distributions
2640 should maintain this unit downstream if they intend to keep
2641 it around. However, we recommend writing normal unit files
2642 for display managers instead.
2644 * Since systemd is a crucial part of the OS we will now
2645 default to a number of compiler switches that improve
2646 security (hardening) such as read-only relocations, stack
2647 protection, and suchlike.
2649 * The TimeoutSec= setting for services is now split into
2650 TimeoutStartSec= and TimeoutStopSec= to allow configuration
2651 of individual time outs for the start and the stop phase of
2654 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
2655 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
2656 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
2657 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
2658 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
2659 Gundersen, Zbigniew Jędrzejewski-Szmek
2663 * The journal and id128 C APIs are now fully documented as man
2666 * Extra safety checks have been added when transitioning from
2667 the initial RAM disk to the main system to avoid accidental
2670 * /etc/crypttab entries now understand the new keyfile-offset=
2673 * systemctl -t can now be used to filter by unit load state.
2675 * The journal C API gained the new sd_journal_wait() call to
2676 make writing synchronous journal clients easier.
2678 * journalctl gained the new -D switch to show journals from a
2681 * journalctl now displays a special marker between log
2682 messages of two different boots.
2684 * The journal is now explicitly flushed to /var via a service
2685 systemd-journal-flush.service, rather than implicitly simply
2686 by seeing /var/log/journal to be writable.
2688 * journalctl (and the journal C APIs) can now match for much
2689 more complex expressions, with alternatives and
2692 * When transitioning from the initial RAM disk to the main
2693 system we will now kill all processes in a killing spree to
2694 ensure no processes stay around by accident.
2696 * Three new specifiers may be used in unit files: %u, %h, %s
2697 resolve to the user name, user home directory resp. user
2698 shell. This is useful for running systemd user instances.
2700 * We now automatically rotate journal files if their data
2701 object hash table gets a fill level > 75%. We also size the
2702 hash table based on the configured maximum file size. This
2703 together should lower hash collisions drastically and thus
2704 speed things up a bit.
2706 * journalctl gained the new "--header" switch to introspect
2707 header data of journal files.
2709 * A new setting SystemCallFilters= has been added to services
2710 which may be used to apply blacklists or whitelists to
2711 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
2713 * nspawn gained a new --link-journal= switch (and quicker: -j)
2714 to link the container journal with the host. This makes it
2715 very easy to centralize log viewing on the host for all
2716 guests while still keeping the journal files separated.
2718 * Many bugfixes and optimizations
2720 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
2721 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
2722 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
2727 * Several tools now understand kernel command line arguments,
2728 which are only read when run in an initial RAM disk. They
2729 usually follow closely their normal counterparts, but are
2732 * There's a new tool to analyze the readahead files that are
2733 automatically generated at boot. Use:
2735 /usr/lib/systemd/systemd-readahead analyze /.readahead
2737 * We now provide an early debug shell on tty9 if this enabled. Use:
2739 systemctl enable debug-shell.service
2741 * All plymouth related units have been moved into the Plymouth
2742 package. Please make sure to upgrade your Plymouth version
2745 * systemd-tmpfiles now supports getting passed the basename of
2746 a configuration file only, in which case it will look for it
2747 in all appropriate directories automatically.
2749 * udevadm info now takes a /dev or /sys path as argument, and
2750 does the right thing. Example:
2752 udevadm info /dev/sda
2753 udevadm info /sys/class/block/sda
2755 * systemctl now prints a warning if a unit is stopped but a
2756 unit that might trigger it continues to run. Example: a
2757 service is stopped but the socket that activates it is left
2760 * "systemctl status" will now mention if the log output was
2761 shortened due to rotation since a service has been started.
2763 * The journal API now exposes functions to determine the
2764 "cutoff" times due to rotation.
2766 * journald now understands SIGUSR1 and SIGUSR2 for triggering
2767 immediately flushing of runtime logs to /var if possible,
2768 resp. for triggering immediate rotation of the journal
2771 * It is now considered an error if a service is attempted to
2772 be stopped that is not loaded.
2774 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
2776 * systemd-analyze now supports Python 3
2778 * tmpfiles now supports cleaning up directories via aging
2779 where the first level dirs are always kept around but
2780 directories beneath it automatically aged. This is enabled
2781 by prefixing the age field with '~'.
2783 * Seat objects now expose CanGraphical, CanTTY properties
2784 which is required to deal with very fast bootups where the
2785 display manager might be running before the graphics drivers
2786 completed initialization.
2788 * Seat objects now expose a State property.
2790 * We now include RPM macros for service enabling/disabling
2791 based on the preset logic. We recommend RPM based
2792 distributions to make use of these macros if possible. This
2793 makes it simpler to reuse RPM spec files across
2796 * We now make sure that the collected systemd unit name is
2797 always valid when services log to the journal via
2800 * There's a new man page kernel-command-line(7) detailing all
2801 command line options we understand.
2803 * The fstab generator may now be disabled at boot by passing
2804 fstab=0 on the kernel command line.
2806 * A new kernel command line option modules-load= is now understood
2807 to load a specific kernel module statically, early at boot.
2809 * Unit names specified on the systemctl command line are now
2810 automatically escaped as needed. Also, if file system or
2811 device paths are specified they are automatically turned
2812 into the appropriate mount or device unit names. Example:
2814 systemctl status /home
2815 systemctl status /dev/sda
2817 * The SysVConsole= configuration option has been removed from
2818 system.conf parsing.
2820 * The SysV search path is no longer exported on the D-Bus
2823 * The Names= option is been removed from unit file parsing.
2825 * There's a new man page bootup(7) detailing the boot process.
2827 * Every unit and every generator we ship with systemd now
2828 comes with full documentation. The self-explanatory boot is
2831 * A couple of services gained "systemd-" prefixes in their
2832 name if they wrap systemd code, rather than only external
2833 code. Among them fsck@.service which is now
2834 systemd-fsck@.service.
2836 * The HaveWatchdog property has been removed from the D-Bus
2839 * systemd.confirm_spawn= on the kernel command line should now
2842 * There's a new man page crypttab(5) which details all options
2843 we actually understand.
2845 * systemd-nspawn gained a new --capability= switch to pass
2846 additional capabilities to the container.
2848 * timedated will now read known NTP implementation unit names
2849 from /usr/lib/systemd/ntp-units.d/*.list,
2850 systemd-timedated-ntp.target has been removed.
2852 * journalctl gained a new switch "-b" that lists log data of
2853 the current boot only.
2855 * The notify socket is in the abstract namespace again, in
2856 order to support daemons which chroot() at start-up.
2858 * There is a new Storage= configuration option for journald
2859 which allows configuration of where log data should go. This
2860 also provides a way to disable journal logging entirely, so
2861 that data collected is only forwarded to the console, the
2862 kernel log buffer or another syslog implementation.
2864 * Many bugfixes and optimizations
2866 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
2867 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
2868 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
2869 Shawn Landden, Tom Gundersen
2873 * "systemctl help <unit>" now shows the man page if one is
2876 * Several new man pages have been added.
2878 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
2879 MaxLevelConsole= can now be specified in
2880 journald.conf. These options allow reducing the amount of
2881 data stored on disk or forwarded by the log level.
2883 * TimerSlackNSec= can now be specified in system.conf for
2884 PID1. This allows system-wide power savings.
2886 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
2887 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
2892 * logind is now capable of (optionally) handling power and
2893 sleep keys as well as the lid switch.
2895 * journalctl now understands the syntax "journalctl
2896 /usr/bin/avahi-daemon" to get all log output of a specific
2899 * CapabilityBoundingSet= in system.conf now also influences
2900 the capability bound set of usermode helpers of the kernel.
2902 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
2903 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
2904 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
2905 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
2909 * Note that we skipped 139 releases here in order to set the
2910 new version to something that is greater than both udev's
2911 and systemd's most recent version number.
2913 * udev: all udev sources are merged into the systemd source tree now.
2914 All future udev development will happen in the systemd tree. It
2915 is still fully supported to use the udev daemon and tools without
2916 systemd running, like in initramfs or other init systems. Building
2917 udev though, will require the *build* of the systemd tree, but
2918 udev can be properly *run* without systemd.
2920 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
2921 should be used to create dead device nodes as workarounds for broken
2924 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
2925 no longer supported. udev_monitor_new_from_netlink() needs to be
2926 used to subscribe to events.
2928 * udev: when udevd is started by systemd, processes which are left
2929 behind by forking them off of udev rules, are unconditionally cleaned
2930 up and killed now after the event handling has finished. Services or
2931 daemons must be started as systemd services. Services can be
2932 pulled-in by udev to get started, but they can no longer be directly
2933 forked by udev rules.
2935 * udev: the daemon binary is called systemd-udevd now and installed
2936 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
2937 to adapt to that, create symlink, or rename the binary after building
2940 * libudev no longer provides these symbols:
2941 udev_monitor_from_socket()
2942 udev_queue_get_failed_list_entry()
2943 udev_get_{dev,sys,run}_path()
2944 The versions number was bumped and symbol versioning introduced.
2946 * systemd-loginctl and systemd-journalctl have been renamed
2947 to loginctl and journalctl to match systemctl.
2949 * The config files: /etc/systemd/systemd-logind.conf and
2950 /etc/systemd/systemd-journald.conf have been renamed to
2951 logind.conf and journald.conf. Package updates should rename
2952 the files to the new names on upgrade.
2954 * For almost all files the license is now LGPL2.1+, changed
2955 from the previous GPL2.0+. Exceptions are some minor stuff
2956 of udev (which will be changed to LGPL2.1 eventually, too),
2957 and the MIT licensed sd-daemon.[ch] library that is suitable
2958 to be used as drop-in files.
2960 * systemd and logind now handle system sleep states, in
2961 particular suspending and hibernating.
2963 * logind now implements a sleep/shutdown/idle inhibiting logic
2964 suitable for a variety of uses. Soonishly Lennart will blog
2965 about this in more detail.
2967 * var-run.mount and var-lock.mount are no longer provided
2968 (which prevously bind mounted these directories to their new
2969 places). Distributions which have not converted these
2970 directories to symlinks should consider stealing these files
2971 from git history and add them downstream.
2973 * We introduced the Documentation= field for units and added
2974 this to all our shipped units. This is useful to make it
2975 easier to explore the boot and the purpose of the various
2978 * All smaller setup units (such as
2979 systemd-vconsole-setup.service) now detect properly if they
2980 are run in a container and are skipped when
2981 appropriate. This guarantees an entirely noise-free boot in
2982 Linux container environments such as systemd-nspawn.
2984 * A framework for implementing offline system updates is now
2985 integrated, for details see:
2986 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
2988 * A new service type Type=idle is available now which helps us
2989 avoiding ugly interleaving of getty output and boot status
2992 * There's now a system-wide CapabilityBoundingSet= option to
2993 globally reduce the set of capabilities for the
2994 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
2995 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
2996 even CAP_NET_ADMIN system-wide for secure systems.
2998 * There are now system-wide DefaultLimitXXX= options to
2999 globally change the defaults of the various resource limits
3000 for all units started by PID 1.
3002 * Harald Hoyer's systemd test suite has been integrated into
3003 systemd which allows easy testing of systemd builds in qemu
3004 and nspawn. (This is really awesome! Ask us for details!)
3006 * The fstab parser is now implemented as generator, not inside
3009 * systemctl will now warn you if .mount units generated from
3010 /etc/fstab are out of date due to changes in fstab that
3011 have not been read by systemd yet.
3013 * systemd is now suitable for usage in initrds. Dracut has
3014 already been updated to make use of this. With this in place
3015 initrds get a slight bit faster but primarily are much
3016 easier to introspect and debug since "systemctl status" in
3017 the host system can be used to introspect initrd services,
3018 and the journal from the initrd is kept around too.
3020 * systemd-delta has been added, a tool to explore differences
3021 between user/admin configuration and vendor defaults.
3023 * PrivateTmp= now affects both /tmp and /var/tmp.
3025 * Boot time status messages are now much prettier and feature
3026 proper english language. Booting up systemd has never been
3029 * Read-ahead pack files now include the inode number of all
3030 files to pre-cache. When the inode changes the pre-caching
3031 is not attempted. This should be nicer to deal with updated
3032 packages which might result in changes of read-ahead
3035 * We now temporaritly lower the kernel's read_ahead_kb variable
3036 when collecting read-ahead data to ensure the kernel's
3037 built-in read-ahead does not add noise to our measurements
3038 of necessary blocks to pre-cache.
3040 * There's now RequiresMountsFor= to add automatic dependencies
3041 for all mounts necessary for a specific file system path.
3043 * MountAuto= and SwapAuto= have been removed from
3044 system.conf. Mounting file systems at boot has to take place
3047 * nspawn now learned a new switch --uuid= to set the machine
3048 ID on the command line.
3050 * nspawn now learned the -b switch to automatically search
3053 * vt102 is now the default TERM for serial TTYs, upgraded from
3056 * systemd-logind now works on VT-less systems.
3058 * The build tree has been reorganized. The individual
3059 components now have directories of their own.
3061 * A new condition type ConditionPathIsReadWrite= is now available.
3063 * nspawn learned the new -C switch to create cgroups for the
3064 container in other hierarchies.
3066 * We now have support for hardware watchdogs, configurable in
3069 * The scheduled shutdown logic now has a public API.
3071 * We now mount /tmp as tmpfs by default, but this can be
3072 masked and /etc/fstab can override it.
3074 * Since udisks does not make use of /media anymore we are not
3075 mounting a tmpfs on it anymore.
3077 * journalctl gained a new --local switch to only interleave
3078 locally generated journal files.
3080 * We can now load the IMA policy at boot automatically.
3082 * The GTK tools have been split off into a systemd-ui.
3084 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
3085 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
3086 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
3087 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
3088 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
3089 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
3090 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
3091 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
3092 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
3097 * This is mostly a bugfix release
3099 * Support optional initialization of the machine ID from the
3100 KVM or container configured UUID.
3102 * Support immediate reboots with "systemctl reboot -ff"
3104 * Show /etc/os-release data in systemd-analyze output
3106 * Many bugfixes for the journal, including endianness fixes and
3107 ensuring that disk space enforcement works
3109 * sd-login.h is C++ comptaible again
3111 * Extend the /etc/os-release format on request of the Debian
3114 * We now refuse non-UTF8 strings used in various configuration
3115 and unit files. This is done to ensure we do not pass invalid
3116 data over D-Bus or expose it elsewhere.
3118 * Register Mimo USB Screens as suitable for automatic seat
3121 * Read SELinux client context from journal clients in a race
3124 * Reorder configuration file lookup order. /etc now always
3125 overrides /run in order to allow the administrator to always
3126 and unconditionally override vendor supplied or
3127 automatically generated data.
3129 * The various user visible bits of the journal now have man
3130 pages. We still lack man pages for the journal API calls
3133 * We now ship all man pages in HTML format again in the
3136 Contributions from: Dave Reisner, Dirk Eibach, Frederic
3137 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
3138 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
3143 * This is mostly a bugfix release
3145 * systems lacking /etc/os-release are no longer supported.
3147 * Various functionality updates to libsystemd-login.so
3149 * Track class of PAM logins to distuingish greeters from
3152 Contributions from: Kay Sievers, Lennart Poettering, Michael
3157 * This is an important bugfix release for v41.
3159 * Building man pages is now optional which should be useful
3160 for those building systemd from git but unwilling to install
3163 * Watchdog support for supervising services is now usable. In
3164 a future release support for hardware watchdogs
3165 (i.e. /dev/watchdog) will be added building on this.
3167 * Service start rate limiting is now configurable and can be
3168 turned off per service. When a start rate limit is hit a
3169 reboot can automatically be triggered.
3171 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
3173 Contributions from: Benjamin Franzke, Bill Nottingham,
3174 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
3175 Schmidt, Michał Górny, Piotr Drąg
3179 * The systemd binary is installed /usr/lib/systemd/systemd now;
3180 An existing /sbin/init symlink needs to be adapted with the
3183 * The code that loads kernel modules has been ported to invoke
3184 libkmod directly, instead of modprobe. This means we do not
3185 support systems with module-init-tools anymore.
3187 * Watchdog support is now already useful, but still not
3190 * A new kernel command line option systemd.setenv= is
3191 understood to set system wide environment variables
3192 dynamically at boot.
3194 * We now limit the set of capabilities of systemd-journald.
3196 * We now set SIGPIPE to ignore by default, since it only is
3197 useful in shell pipelines, and has little use in general
3198 code. This can be disabled with IgnoreSIPIPE=no in unit
3201 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
3202 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
3207 * This is mostly a bugfix release
3209 * We now expose the reason why a service failed in the
3210 "Result" D-Bus property.
3212 * Rudimentary service watchdog support (will be completed over
3213 the next few releases.)
3215 * When systemd forks off in order execute some service we will
3216 now immediately changes its argv[0] to reflect which process
3217 it will execute. This is useful to minimize the time window
3218 with a generic argv[0], which makes bootcharts more useful
3220 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
3221 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
3222 Mike Kazantsev, Ray Strode
3226 * This is mostly a test release, but incorporates many
3229 * New systemd-cgtop tool to show control groups by their
3232 * Linking against libacl for ACLs is optional again. If
3233 disabled, support tracking device access for active logins
3234 goes becomes unavailable, and so does access to the user
3235 journals by the respective users.
3237 * If a group "adm" exists, journal files are automatically
3238 owned by them, thus allow members of this group full access
3239 to the system journal as well as all user journals.
3241 * The journal now stores the SELinux context of the logging
3242 client for all entries.
3244 * Add C++ inclusion guards to all public headers
3246 * New output mode "cat" in the journal to print only text
3247 messages, without any meta data like date or time.
3249 * Include tiny X server wrapper as a temporary stop-gap to
3250 teach XOrg udev display enumeration. This is used by display
3251 managers such as gdm, and will go away as soon as XOrg
3252 learned native udev hotplugging for display devices.
3254 * Add new systemd-cat tool for executing arbitrary programs
3255 with STDERR/STDOUT connected to the journal. Can also act as
3256 BSD logger replacement, and does so by default.
3258 * Optionally store all locally generated coredumps in the
3259 journal along with meta data.
3261 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
3262 writing short strings to files (for usage for /sys), and for
3263 creating symlinks, character and block device nodes.
3265 * New unit file option ControlGroupPersistent= to make cgroups
3266 persistent, following the mechanisms outlined in
3267 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
3269 * Support multiple local RTCs in a sane way
3271 * No longer monopolize IO when replaying readahead data on
3272 rotating disks, since we might starve non-file-system IO to
3273 death, since fanotify() will not see accesses done by blkid,
3276 * Do not show kernel threads in systemd-cgls anymore, unless
3277 requested with new -k switch.
3279 Contributions from: Dan Horák, Kay Sievers, Lennart
3280 Poettering, Michal Schmidt
3284 * This is mostly a test release, but incorporates many
3287 * The git repository moved to:
3288 git://anongit.freedesktop.org/systemd/systemd
3289 ssh://git.freedesktop.org/git/systemd/systemd
3291 * First release with the journal
3292 http://0pointer.de/blog/projects/the-journal.html
3294 * The journal replaces both systemd-kmsg-syslogd and
3295 systemd-stdout-bridge.
3297 * New sd_pid_get_unit() API call in libsystemd-logind
3299 * Many systemadm clean-ups
3301 * Introduce remote-fs-pre.target which is ordered before all
3302 remote mounts and may be used to start services before all
3305 * Added Mageia support
3307 * Add bash completion for systemd-loginctl
3309 * Actively monitor PID file creation for daemons which exit in
3310 the parent process before having finished writing the PID
3311 file in the daemon process. Daemons which do this need to be
3312 fixed (i.e. PID file creation must have finished before the
3313 parent exits), but we now react a bit more gracefully to them.
3315 * Add colourful boot output, mimicking the well-known output
3316 of existing distributions.
3318 * New option PassCredentials= for socket units, for
3319 compatibility with a recent kernel ABI breakage.
3321 * /etc/rc.local is now hooked in via a generator binary, and
3322 thus will no longer act as synchronization point during
3325 * systemctl list-unit-files now supports --root=.
3327 * systemd-tmpfiles now understands two new commands: z, Z for
3328 relabelling files according to the SELinux database. This is
3329 useful to apply SELinux labels to specific files in /sys,
3332 * Output of SysV services is now forwarded to both the console
3333 and the journal by default, not only just the console.
3335 * New man pages for all APIs from libsystemd-login.
3337 * The build tree got reorganized and a the build system is a
3338 lot more modular allowing embedded setups to specifically
3339 select the components of systemd they are interested in.
3341 * Support for Linux systems lacking the kernel VT subsystem is
3344 * configure's --with-rootdir= got renamed to
3345 --with-rootprefix= to follow the naming used by udev and
3348 * Unless specified otherwise we will now install to /usr instead
3349 of /usr/local by default.
3351 * Processes with '@' in argv[0][0] are now excluded from the
3352 final shut-down killing spree, following the logic explained
3354 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
3356 * All processes remaining in a service cgroup when we enter
3357 the START or START_PRE states are now killed with
3358 SIGKILL. That means it is no longer possible to spawn
3359 background processes from ExecStart= lines (which was never
3360 supported anyway, and bad style).
3362 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
3363 reloading of units together.
3365 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
3366 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
3367 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
3368 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
3369 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek