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
11 from 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.
25 * The queue "seqnum" interface of libudev has been disabled, as
26 it was generally incompatible with device namespacing as
27 sequence numbers of devices go "missing" if the devices are
28 part of a different namespace.
30 * "systemctl list-timers" and "systemctl list-sockets" gained
31 a --recursive switch for showing units of these types also
32 for all local containers, similar in style to the
33 --recursive switch "systemctl list-units" already supports.
35 * A new RebootArgument= setting has been added for service
36 units, which may be used to specify a kernel reboot argument
37 for usage when triggering reboots with StartLimitAction=.
39 * A new FailureAction= setting has been added for service
40 units which may be used to specify an operation to trigger
41 when a service fails. This works similar to
42 StartLimitAction=, but unlike it controls what is done
43 immediately rather than only after several attempts to
44 restart the service in question.
46 * hostnamed got updated to also expose the kernel name,
47 release and version on the bus. This is useful for executing
48 commands like hostnamectl with the -H switch.
50 * The bootchart tool can now show cgroup information in the
53 * The CFS CPU quota cgroup attribute is now exposed for
54 services. The new CPUQuota= switch has been added for this
55 which takes a percentage value. Setting this will have the
56 result that a service may never get more CPU time than the
57 specified percentage, even if the machine is otherwise idle.
59 * systemd-networkd learned IPIP and SIT tunnel support.
61 * LSB init scripts exposing a dependency on $network will now
62 get a dependency on network-online.target rather than simply
63 network.target. This should bring LSB handling closer to
64 what it was on SysV systems.
66 * A new fsck.repair= kernel option has been added to control
67 how fsck shall deal with unclean file systems at boot.
69 * The (.ini) configuration file parser will now silently
70 ignore sections whose name begins with "X-". This may be
71 used to maintain application-specific extension sections in unit
74 * machined gained a new API to query the IP addresses of
75 registered containers. "machinectl status" has been updated
76 to show these addresses in its output.
78 * A new call sd_uid_get_display() has been added to the
79 sd-login APIs for querying the "primary" session of a
80 user. The "primary" session of the user is elected from the
81 user's sessions and generally a graphical session is
82 preferred over a text one.
84 * A minimal systemd-resolved daemon has been added. It
85 currently simply acts as a companion to systemd-networkd and
86 manages resolv.conf based on per-interface DNS
87 configuration, possibly supplied via DHCP. In the long run
88 we hope to extend this into a local DNSSEC enabled DNS and
91 * The systemd-networkd-wait-online tool is now enabled by
92 default. It will delay network-online.target until a network
93 connection has been configured. The tool primarily integrates
94 with networkd, but will also make a best effort to make sense
95 of network configuration performed in some other way.
97 * Two new service options StartupCPUShares= and
98 StartupBlockIOWeight= have been added that work similar to
99 CPUShares= and BlockIOWeight= however only apply during
100 system startup. This is useful to prioritize certain services
101 differently during bootup than during normal runtime.
103 Contributions from: Ali H. Caliskan, Alison Chaiken, Bas van
104 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
105 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
106 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
107 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
108 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
109 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
110 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
111 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
112 Marcel Holtmann, Matthew Monaco, Michael Marineau, Michael
113 Olbrich, Michal Sekletar, Mike Gilbert, Nis Martensen, Patrik
114 Flykt, Philip Lorenz, poma, Ray Strode, Robert Milasan, Scott
115 Thrasher, Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas
116 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut
117 Tezduyar Lindskog, WaLyong Cho, Will Woods, Zbigniew
120 -- Beijing, 2014-05-24
124 * When restoring the screen brightness at boot, stay away from
125 the darkest setting or from the lowest 5% of the available
126 range, depending on which is the larger value of both. This
127 should effectively protect the user from rebooting into a
128 black screen, should the brightness have been set to minimum
131 * sd-login gained a new sd_machine_get_class() call to
132 determine the class ("vm" or "container") of a machine
133 registered with machined.
135 * sd-login gained new calls
136 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
137 to query the identity of the peer of a local AF_UNIX
138 connection. They operate similar to their sd_pid_get_xyz()
141 * PID 1 will now maintain a system-wide system state engine
142 with the states "starting", "running", "degraded",
143 "maintenance", "stopping". These states are bound to system
144 startup, normal runtime, runtime with at least one failed
145 service, rescue/emergency mode and system shutdown. This
146 state is shown in the "systemctl status" output when no unit
147 name is passed. It is useful to determine system state, in
148 particularly when doing so for many systems or containers at
151 * A new command "list-machines" has been added to "systemctl"
152 that lists all local OS containers and shows their system
153 state (see above), if systemd runs inside of them.
155 * systemctl gained a new "-r" switch to recursively enumerate
156 units on all local containers, when used with the
157 "list-unit" command (which is the default one that is
158 executed when no parameters are specified).
160 * The GPT automatic partition discovery logic will now honour
161 two GPT partition flags: one may be set on a partition to
162 cause it to be mounted read-only, and the other may be set
163 on a partition to ignore it during automatic discovery.
165 * Two new GPT type UUIDs have been added for automatic root
166 partition discovery, for 32-bit and 64-bit ARM. This is not
167 particularly useful for discovering the root directory on
168 these architectures during bare-metal boots (since UEFI is
169 not common there), but still very useful to allow booting of
170 ARM disk images in nspawn with the -i option.
172 * MAC addresses of interfaces created with nspawn's
173 --network-interface= switch will now be generated from the
174 machine name, and thus be stable between multiple invocations
177 * logind will now automatically remove all IPC objects owned
178 by a user if she or he fully logs out. This makes sure that
179 users who are logged out cannot continue to consume IPC
180 resources. This covers SysV memory, semaphores and message
181 queues as well as POSIX shared memory and message
182 queues. Traditionally, SysV and POSIX IPC had no life-cycle
183 limits. With this functionality, that is corrected. This may
184 be turned off by using the RemoveIPC= switch of logind.conf.
186 * The systemd-machine-id-setup and tmpfiles tools gained a
187 --root= switch to operate on a specific root directory,
190 * journald can now forward logged messages to the TTYs of all
191 logged in users ("wall"). This is the default for all
192 emergency messages now.
194 * A new tool systemd-journal-remote has been added to stream
195 journal log messages across the network.
197 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
198 controller trees are mounted into it. Note that the
199 directories mounted beneath it are not read-only. This is a
200 security measure and is particularly useful because glibc
201 actually includes a search logic to pick any tmpfs it can
202 find to implement shm_open() if /dev/shm is not available
203 (which it might very well be in namespaced setups).
205 * machinectl gained a new "poweroff" command to cleanly power
206 down a local OS container.
208 * The PrivateDevices= unit file setting will now also drop the
209 CAP_MKNOD capability from the capability bound set, and
210 imply DevicePolicy=closed.
212 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
213 comprehensively on all long-running systemd services where
216 * systemd-udevd will now run in a disassociated mount
217 namespace. To mount directories from udev rules, make sure to
218 pull in mount units via SYSTEMD_WANTS properties.
220 * The kdbus support gained support for uploading policy into
221 the kernel. sd-bus gained support for creating "monitoring"
222 connections that can eavesdrop into all bus communication
223 for debugging purposes.
225 * Timestamps may now be specified in seconds since the UNIX
226 epoch Jan 1st, 1970 by specifying "@" followed by the value
229 * Native tcpwrap support in systemd has been removed. tcpwrap
230 is old code, not really maintained anymore and has serious
231 shortcomings, and better options such as firewalls
232 exist. For setups that require tcpwrap usage, please
233 consider invoking your socket-activated service via tcpd,
234 like on traditional inetd.
236 * A new system.conf configuration option
237 DefaultTimerAccuracySec= has been added that controls the
238 default AccuracySec= setting of .timer units.
240 * Timer units gained a new WakeSystem= switch. If enabled,
241 timers configured this way will cause the system to resume
242 from system suspend (if the system supports that, which most
245 * Timer units gained a new Persistent= switch. If enabled,
246 timers configured this way will save to disk when they have
247 been last triggered. This information is then used on next
248 reboot to possible execute overdue timer events, that
249 could not take place because the system was powered off.
250 This enables simple anacron-like behaviour for timer units.
252 * systemctl's "list-timers" will now also list the time a
253 timer unit was last triggered in addition to the next time
254 it will be triggered.
256 * systemd-networkd will now assign predictable IPv4LL
257 addresses to its local interfaces.
259 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
260 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
261 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
262 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
263 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
264 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
265 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
266 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
269 -- Berlin, 2014-03-25
273 * A new unit file setting RestrictAddressFamilies= has been
274 added to restrict which socket address families unit
275 processes gain access to. This takes address family names
276 like "AF_INET" or "AF_UNIX", and is useful to minimize the
277 attack surface of services via exotic protocol stacks. This
278 is built on seccomp system call filters.
280 * Two new unit file settings RuntimeDirectory= and
281 RuntimeDirectoryMode= have been added that may be used to
282 manage a per-daemon runtime directories below /run. This is
283 an alternative for setting up directory permissions with
284 tmpfiles snippets, and has the advantage that the runtime
285 directory's lifetime is bound to the daemon runtime and that
286 the daemon starts up with an empty directory each time. This
287 is particularly useful when writing services that drop
288 privileges using the User= or Group= setting.
290 * The DeviceAllow= unit setting now supports globbing for
291 matching against device group names.
293 * The systemd configuration file system.conf gained new
294 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
295 DefaultMemoryAccounting= to globally turn on/off accounting
296 for specific resources (cgroups) for all units. These
297 settings may still be overridden individually in each unit
300 * systemd-gpt-auto-generator is now able to discover /srv and
301 root partitions in addition to /home and swap partitions. It
302 also supports LUKS-encrypted partitions now. With this in
303 place, automatic discovery of partitions to mount following
304 the Discoverable Partitions Specification
305 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
306 is now a lot more complete. This allows booting without
307 /etc/fstab and without root= on the kernel command line on
308 systems prepared appropriately.
310 * systemd-nspawn gained a new --image= switch which allows
311 booting up disk images and Linux installations on any block
312 device that follow the Discoverable Partitions Specification
313 (see above). This means that installations made with
314 appropriately updated installers may now be started and
315 deployed using container managers, completely
316 unmodified. (We hope that libvirt-lxc will add support for
317 this feature soon, too.)
319 * systemd-nspawn gained a new --network-macvlan= setting to
320 set up a private macvlan interface for the
321 container. Similar, systemd-networkd gained a new
322 Kind=macvlan setting in .netdev files.
324 * systemd-networkd now supports configuring local addresses
327 * A new tool systemd-network-wait-online has been added to
328 synchronously wait for network connectivity using
331 * The sd-bus.h bus API gained a new sd_bus_track object for
332 tracking the life-cycle of bus peers. Note that sd-bus.h is
333 still not a public API though (unless you specify
334 --enable-kdbus on the configure command line, which however
335 voids your warranty and you get no API stability guarantee).
337 * The $XDG_RUNTIME_DIR runtime directories for each user are
338 now individual tmpfs instances, which has the benefit of
339 introducing separate pools for each user, with individual
340 size limits, and thus making sure that unprivileged clients
341 can no longer negatively impact the system or other users by
342 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
343 RuntimeDirectorySize= has been introduced that allows
344 controlling the default size limit for all users. It
345 defaults to 10% of the available physical memory. This is no
346 replacement for quotas on tmpfs though (which the kernel
347 still does not support), as /dev/shm and /tmp are still
348 shared resources used by both the system and unprivileged
351 * logind will now automatically turn off automatic suspending
352 on laptop lid close when more than one display is
353 connected. This was previously expected to be implemented
354 individually in desktop environments (such as GNOME),
355 however has been added to logind now, in order to fix a
356 boot-time race where a desktop environment might not have
357 been started yet and thus not been able to take an inhibitor
358 lock at the time where logind already suspends the system
361 * logind will now wait at least 30s after each system
362 suspend/resume cycle, and 3min after system boot before
363 suspending the system due to a closed laptop lid. This
364 should give USB docking stations and similar enough time to
365 be probed and configured after system resume and boot in
366 order to then act as suspend blocker.
368 * systemd-run gained a new --property= setting which allows
369 initialization of resource control properties (and others)
370 for the created scope or service unit. Example: "systemd-run
371 --property=BlockIOWeight=10 updatedb" may be used to run
372 updatedb at a low block IO scheduling weight.
374 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
375 now also work in --scope mode.
377 * When systemd is compiled with kdbus support, basic support
378 for enforced policies is now in place. (Note that enabling
379 kdbus still voids your warranty and no API compatibility
382 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
383 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
384 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
385 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
386 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
387 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
388 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
389 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
390 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
391 Zbigniew Jędrzejewski-Szmek
393 -- Berlin, 2014-03-12
397 * systemd will now relabel /dev after loading the SMACK policy
398 according to SMACK rules.
400 * A new unit file option AppArmorProfile= has been added to
401 set the AppArmor profile for the processes of a unit.
403 * A new condition check ConditionArchitecture= has been added
404 to conditionalize units based on the system architecture, as
405 reported by uname()'s "machine" field.
407 * systemd-networkd now supports matching on the system
408 virtualization, architecture, kernel command line, host name
411 * logind is now a lot more aggressive when suspending the
412 machine due to a closed laptop lid. Instead of acting only
413 on the lid close action, it will continuously watch the lid
414 status and act on it. This is useful for laptops where the
415 power button is on the outside of the chassis so that it can
416 be reached without opening the lid (such as the Lenovo
417 Yoga). On those machines, logind will now immediately
418 re-suspend the machine if the power button has been
419 accidentally pressed while the laptop was suspended and in a
422 * logind will now watch SW_DOCK switches and inhibit reaction
423 to the lid switch if it is pressed. This means that logind
424 will not suspend the machine anymore if the lid is closed
425 and the system is docked, if the laptop supports SW_DOCK
426 notifications via the input layer. Note that ACPI docking
427 stations do not generate this currently. Also note that this
428 logic is usually not fully sufficient and Desktop
429 Environments should take a lid switch inhibitor lock when an
430 external display is connected, as systemd will not watch
433 * nspawn will now make use of the devices cgroup controller by
434 default, and only permit creation of and access to the usual
435 API device nodes like /dev/null or /dev/random, as well as
436 access to (but not creation of) the pty devices.
438 * We will now ship a default .network file for
439 systemd-networkd that automatically configures DHCP for
440 network interfaces created by nspawn's --network-veth or
441 --network-bridge= switches.
443 * systemd will now understand the usual M, K, G, T suffixes
444 according to SI conventions (i.e. to the base 1000) when
445 referring to throughput and hardware metrics. It will stay
446 with IEC conventions (i.e. to the base 1024) for software
447 metrics, according to what is customary according to
448 Wikipedia. We explicitly document which base applies for
449 each configuration option.
451 * The DeviceAllow= setting in unit files now supports a syntax
452 to whitelist an entire group of devices node majors at once,
453 based on the /proc/devices listing. For example, with the
454 string "char-pts", it is now possible to whitelist all
455 current and future pseudo-TTYs at once.
457 * sd-event learned a new "post" event source. Event sources of
458 this type are triggered by the dispatching of any event
459 source of a type that is not "post". This is useful for
460 implementing clean-up and check event sources that are
461 triggered by other work being done in the program.
463 * systemd-networkd is no longer statically enabled, but uses
464 the usual [Install] sections so that it can be
465 enabled/disabled using systemctl. It still is enabled by
468 * When creating a veth interface pair with systemd-nspawn, the
469 host side will now be prefixed with "vb-" if
470 --network-bridge= is used, and with "ve-" if --network-veth
471 is used. This way, it is easy to distinguish these cases on
472 the host, for example to apply different configuration to
473 them with systemd-networkd.
475 * The compatibility libraries for libsystemd-journal.so,
476 libsystem-id128.so, libsystemd-login.so and
477 libsystemd-daemon.so do not make use of IFUNC
478 anymore. Instead, we now build libsystemd.so multiple times
479 under these alternative names. This means that the footprint
480 is drastically increased, but given that these are
481 transitional compatibility libraries, this should not matter
482 much. This change has been made necessary to support the ARM
483 platform for these compatibility libraries, as the ARM
484 toolchain is not really at the same level as the toolchain
485 for other architectures like x86 and does not support
486 IFUNC. Please make sure to use --enable-compat-libs only
487 during a transitional period!
489 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
490 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
491 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
492 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
493 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
494 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
495 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
496 Zbigniew Jędrzejewski-Szmek
498 -- Berlin, 2014-02-24
502 * A new component "systemd-networkd" has been added that can
503 be used to configure local network interfaces statically or
504 via DHCP. It is capable of bringing up bridges, VLANs, and
505 bonding. Currently, no hook-ups for interactive network
506 configuration are provided. Use this for your initrd,
507 container, embedded, or server setup if you need a simple,
508 yet powerful, network configuration solution. This
509 configuration subsystem is quite nifty, as it allows wildcard
510 hotplug matching in interfaces. For example, with a single
511 configuration snippet, you can configure that all Ethernet
512 interfaces showing up are automatically added to a bridge,
513 or similar. It supports link-sensing and more.
515 * A new tool "systemd-socket-proxyd" has been added which can
516 act as a bidirectional proxy for TCP sockets. This is
517 useful for adding socket activation support to services that
518 do not actually support socket activation, including virtual
519 machines and the like.
521 * Add a new tool to save/restore rfkill state on
524 * Save/restore state of keyboard backlights in addition to
525 display backlights on shutdown/boot.
527 * udev learned a new SECLABEL{} construct to label device
528 nodes with a specific security label when they appear. For
529 now, only SECLABEL{selinux} is supported, but the syntax is
530 prepared for additional security frameworks.
532 * udev gained a new scheme to configure link-level attributes
533 from files in /etc/systemd/network/*.link. These files can
534 match against MAC address, device path, driver name and type,
535 and will apply attributes like the naming policy, link speed,
536 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
537 address assignment policy (randomized, ...).
539 * The configuration of network interface naming rules for
540 "permanent interface names" has changed: a new NamePolicy=
541 setting in the [Link] section of .link files determines the
542 priority of possible naming schemes (onboard, slot, mac,
543 path). The default value of this setting is determined by
544 /usr/lib/net/links/99-default.link. Old
545 80-net-name-slot.rules udev configuration file has been
546 removed, so local configuration overriding this file should
547 be adapated to override 99-default.link instead.
549 * When the User= switch is used in a unit file, also
550 initialize $SHELL= based on the user database entry.
552 * systemd no longer depends on libdbus. All communication is
553 now done with sd-bus, systemd's low-level bus library
556 * kdbus support has been added to PID 1 itself. When kdbus is
557 enabled, this causes PID 1 to set up the system bus and
558 enable support for a new ".busname" unit type that
559 encapsulates bus name activation on kdbus. It works a little
560 bit like ".socket" units, except for bus names. A new
561 generator has been added that converts classic dbus1 service
562 activation files automatically into native systemd .busname
565 * sd-bus: add a light-weight vtable implementation that allows
566 defining objects on the bus with a simple static const
567 vtable array of its methods, signals and properties.
569 * systemd will not generate or install static dbus
570 introspection data anymore to /usr/share/dbus-1/interfaces,
571 as the precise format of these files is unclear, and
572 nothing makes use of it.
574 * A proxy daemon is now provided to proxy clients connecting
575 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
576 compatibility with classic D-Bus.
578 * A bus driver implementation has been added that supports the
579 classic D-Bus bus driver calls on kdbus, also for
580 compatibility purposes.
582 * A new API "sd-event.h" has been added that implements a
583 minimal event loop API built around epoll. It provides a
584 couple of features that direct epoll usage is lacking:
585 prioritization of events, scales to large numbers of timer
586 events, per-event timer slack (accuracy), system-wide
587 coalescing of timer events, exit handlers, watchdog
588 supervision support using systemd's sd_notify() API, child
591 * A new API "sd-rntl.h" has been added that provides an API
592 around the route netlink interface of the kernel, similar in
595 * A new API "sd-dhcp-client.h" has been added that provides a
596 small DHCPv4 client-side implementation. This is used by
599 * There is a new kernel command line option
600 "systemd.restore_state=0|1". When set to "0", none of the
601 systemd tools will restore saved runtime state to hardware
602 devices. More specifically, the rfkill and backlight states
605 * The FsckPassNo= compatibility option in mount/service units
606 has been removed. The fstab generator will now add the
607 necessary dependencies automatically, and does not require
608 PID1's support for that anymore.
610 * journalctl gained a new switch, --list-boots, that lists
611 recent boots with their times and boot IDs.
613 * The various tools like systemctl, loginctl, timedatectl,
614 busctl, systemd-run, ... have gained a new switch "-M" to
615 connect to a specific, local OS container (as direct
616 connection, without requiring SSH). This works on any
617 container that is registered with machined, such as those
618 created by libvirt-lxc or nspawn.
620 * systemd-run and systemd-analyze also gained support for "-H"
621 to connect to remote hosts via SSH. This is particularly
622 useful for systemd-run because it enables queuing of jobs
625 * machinectl gained a new command "login" to open a getty
626 login in any local container. This works with any container
627 that is registered with machined (such as those created by
628 libvirt-lxc or nspawn), and which runs systemd inside.
630 * machinectl gained a new "reboot" command that may be used to
631 trigger a reboot on a specific container that is registered
632 with machined. This works on any container that runs an init
635 * systemctl gained a new "list-timers" command to print a nice
636 listing of installed timer units with the times they elapse
639 * Alternative reboot() parameters may now be specified on the
640 "systemctl reboot" command line and are passed to the
641 reboot() system call.
643 * systemctl gained a new --job-mode= switch to configure the
644 mode to queue a job with. This is a more generic version of
645 --fail, --irreversible, and --ignore-dependencies, which are
646 still available but not advertised anymore.
648 * /etc/systemd/system.conf gained new settings to configure
649 various default timeouts of units, as well as the default
650 start limit interval and burst. These may still be overridden
653 * PID1 will now export on the bus profile data of the security
654 policy upload process (such as the SELinux policy upload to
657 * journald: when forwarding logs to the console, include
658 timestamps (following the setting in
659 /sys/module/printk/parameters/time).
661 * OnCalendar= in timer units now understands the special
662 strings "yearly" and "annually". (Both are equivalent)
664 * The accuracy of timer units is now configurable with the new
665 AccuracySec= setting. It defaults to 1min.
667 * A new dependency type JoinsNamespaceOf= has been added that
668 allows running two services within the same /tmp and network
669 namespace, if PrivateNetwork= or PrivateTmp= are used.
671 * A new command "cat" has been added to systemctl. It outputs
672 the original unit file of a unit, and concatenates the
673 contents of additional "drop-in" unit file snippets, so that
674 the full configuration is shown.
676 * systemctl now supports globbing on the various "list-xyz"
677 commands, like "list-units" or "list-sockets", as well as on
678 those commands which take multiple unit names.
680 * journalctl's --unit= switch gained support for globbing.
682 * All systemd daemons now make use of the watchdog logic so
683 that systemd automatically notices when they hang.
685 * If the $container_ttys environment variable is set,
686 getty-generator will automatically spawn a getty for each
687 listed tty. This is useful for container managers to request
688 login gettys to be spawned on as many ttys as needed.
690 * %h, %s, %U specifier support is not available anymore when
691 used in unit files for PID 1. This is because NSS calls are
692 not safe from PID 1. They stay available for --user
693 instances of systemd, and as special case for the root user.
695 * loginctl gained a new "--no-legend" switch to turn off output
698 * The "sd-login.h" API gained three new calls:
699 sd_session_is_remote(), sd_session_get_remote_user(),
700 sd_session_get_remote_host() to query information about
703 * The udev hardware database now also carries vendor/product
704 information of SDIO devices.
706 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
707 determine whether watchdog notifications are requested by
710 * Socket-activated per-connection services now include a
711 short description of the connection parameters in the
714 * tmpfiles gained a new "--boot" option. When this is not used,
715 only lines where the command character is not suffixed with
716 "!" are executed. When this option is specified, those
717 options are executed too. This partitions tmpfiles
718 directives into those that can be safely executed at any
719 time, and those which should be run only at boot (for
720 example, a line that creates /run/nologin).
722 * A new API "sd-resolve.h" has been added which provides a simple
723 asynchronous wrapper around glibc NSS host name resolution
724 calls, such as getaddrinfo(). In contrast to glibc's
725 getaddrinfo_a(), it does not use signals. In contrast to most
726 other asynchronous name resolution libraries, this one does
727 not reimplement DNS, but reuses NSS, so that alternate
728 host name resolution systems continue to work, such as mDNS,
729 LDAP, etc. This API is based on libasyncns, but it has been
730 cleaned up for inclusion in systemd.
732 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
733 "sd-daemon.h" are no longer found in individual libraries
734 libsystemd-journal.so, libsystemd-login.so,
735 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
736 merged them into a single library, libsystemd.so, which
737 provides all symbols. The reason for this is cyclic
738 dependencies, as these libraries tend to use each other's
739 symbols. So far, we have managed to workaround that by linking
740 a copy of a good part of our code into each of these
741 libraries again and again, which, however, makes certain
742 things hard to do, like sharing static variables. Also, it
743 substantially increases footprint. With this change, there
744 is only one library for the basic APIs systemd
745 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
746 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
747 library as well, however are subject to the --enable-kdbus
748 switch (see below). Note that "sd-dhcp-client.h" is not part
749 of this library (this is because it only consumes, never
750 provides, services of/to other APIs). To make the transition
751 easy from the separate libraries to the unified one, we
752 provide the --enable-compat-libs compile-time switch which
753 will generate stub libraries that are compatible with the
754 old ones but redirect all calls to the new one.
756 * All of the kdbus logic and the new APIs "sd-bus.h",
757 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
758 and "sd-utf8.h" are compile-time optional via the
759 "--enable-kdbus" switch, and they are not compiled in by
760 default. To make use of kdbus, you have to explicitly enable
761 the switch. Note however, that neither the kernel nor the
762 userspace API for all of this is considered stable yet. We
763 want to maintain the freedom to still change the APIs for
764 now. By specifying this build-time switch, you acknowledge
765 that you are aware of the instability of the current
768 * Also, note that while kdbus is pretty much complete,
769 it lacks one thing: proper policy support. This means you
770 can build a fully working system with all features; however,
771 it will be highly insecure. Policy support will be added in
772 one of the next releases, at the same time that we will
773 declare the APIs stable.
775 * When the kernel command-line argument "kdbus" is specified,
776 systemd will automatically load the kdbus.ko kernel module. At
777 this stage of development, it is only useful for testing kdbus
778 and should not be used in production. Note: if "--enable-kdbus"
779 is specified, and the kdbus.ko kernel module is available, and
780 "kdbus" is added to the kernel command line, the entire system
781 runs with kdbus instead of dbus-daemon, with the above mentioned
782 problem of missing the system policy enforcement. Also a future
783 version of kdbus.ko or a newer systemd will not be compatible with
784 each other, and will unlikely be able to boot the machine if only
785 one of them is updated.
787 * systemctl gained a new "import-environment" command which
788 uploads the caller's environment (or parts thereof) into the
789 service manager so that it is inherited by services started
790 by the manager. This is useful to upload variables like
791 $DISPLAY into the user service manager.
793 * A new PrivateDevices= switch has been added to service units
794 which allows running a service with a namespaced /dev
795 directory that does not contain any device nodes for
796 physical devices. More specifically, it only includes devices
797 such as /dev/null, /dev/urandom, and /dev/zero which are API
800 * logind has been extended to support behaviour like VT
801 switching on seats that do not support a VT. This makes
802 multi-session available on seats that are not the first seat
803 (seat0), and on systems where kernel support for VTs has
804 been disabled at compile-time.
806 * If a process holds a delay lock for system sleep or shutdown
807 and fails to release it in time, we will now log its
808 identity. This makes it easier to identify processes that
809 cause slow suspends or power-offs.
811 * When parsing /etc/crypttab, support for a new key-slot=
812 option as supported by Debian is added. It allows indicating
813 which LUKS slot to use on disk, speeding up key loading.
815 * The sd_journald_sendv() API call has been checked and
816 officially declared to be async-signal-safe so that it may
817 be invoked from signal handlers for logging purposes.
819 * Boot-time status output is now enabled automatically after a
820 short timeout if boot does not progress, in order to give
821 the user an indication what she or he is waiting for.
823 * The boot-time output has been improved to show how much time
824 remains until jobs expire.
826 * The KillMode= switch in service units gained a new possible
827 value "mixed". If set, and the unit is shut down, then the
828 initial SIGTERM signal is sent only to the main daemon
829 process, while the following SIGKILL signal is sent to
830 all remaining processes of the service.
832 * When a scope unit is registered, a new property "Controller"
833 may be set. If set to a valid bus name, systemd will send a
834 RequestStop() signal to this name when it would like to shut
835 down the scope. This may be used to hook manager logic into
836 the shutdown logic of scope units. Also, scope units may now
837 be put in a special "abandoned" state, in which case the
838 manager process which created them takes no further
839 responsibilities for it.
841 * When reading unit files, systemd will now verify
842 the access mode of these files, and warn about certain
843 suspicious combinations. This has been added to make it
844 easier to track down packaging bugs where unit files are
845 marked executable or world-writable.
847 * systemd-nspawn gained a new "--setenv=" switch to set
848 container-wide environment variables. The similar option in
849 systemd-activate was renamed from "--environment=" to
850 "--setenv=" for consistency.
852 * systemd-nspawn has been updated to create a new kdbus domain
853 for each container that is invoked, thus allowing each
854 container to have its own set of system and user buses,
855 independent of the host.
857 * systemd-nspawn gained a new --drop-capability= switch to run
858 the container with less capabilities than the default. Both
859 --drop-capability= and --capability= now take the special
860 string "all" for dropping or keeping all capabilities.
862 * systemd-nspawn gained new switches for executing containers
863 with specific SELinux labels set.
865 * systemd-nspawn gained a new --quiet switch to not generate
866 any additional output but the container's own console
869 * systemd-nspawn gained a new --share-system switch to run a
870 container without PID namespacing enabled.
872 * systemd-nspawn gained a new --register= switch to control
873 whether the container is registered with systemd-machined or
874 not. This is useful for containers that do not run full
875 OS images, but only specific apps.
877 * systemd-nspawn gained a new --keep-unit which may be used
878 when invoked as the only program from a service unit, and
879 results in registration of the unit service itself in
880 systemd-machined, instead of a newly opened scope unit.
882 * systemd-nspawn gained a new --network-interface= switch for
883 moving arbitrary interfaces to the container. The new
884 --network-veth switch creates a virtual Ethernet connection
885 between host and container. The new --network-bridge=
886 switch then allows assigning the host side of this virtual
887 Ethernet connection to a bridge device.
889 * systemd-nspawn gained a new --personality= switch for
890 setting the kernel personality for the container. This is
891 useful when running a 32-bit container on a 64-bit host. A
892 similar option Personality= is now also available for service
895 * logind will now also track a "Desktop" identifier for each
896 session which encodes the desktop environment of it. This is
897 useful for desktop environments that want to identify
898 multiple running sessions of itself easily.
900 * A new SELinuxContext= setting for service units has been
901 added that allows setting a specific SELinux execution
902 context for a service.
904 * Most systemd client tools will now honour $SYSTEMD_LESS for
905 settings of the "less" pager. By default, these tools will
906 override $LESS to allow certain operations to work, such as
907 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
908 influence this logic.
910 * systemd's "seccomp" hook-up has been changed to make use of
911 the libseccomp library instead of using its own
912 implementation. This has benefits for portability among
915 * For usage together with SystemCallFilter=, a new
916 SystemCallErrorNumber= setting has been introduced that
917 allows configuration of a system error number to be returned
918 on filtered system calls, instead of immediately killing the
919 process. Also, SystemCallArchitectures= has been added to
920 limit access to system calls of a particular architecture
921 (in order to turn off support for unused secondary
922 architectures). There is also a global
923 SystemCallArchitectures= setting in system.conf now to turn
924 off support for non-native system calls system-wide.
926 * systemd requires a kernel with a working name_to_handle_at(),
927 please see the kernel config requirements in the README file.
929 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
930 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
931 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
932 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
933 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
934 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
935 Elia Pinto, Florian Weimer, George McCollister, Goffredo
936 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
937 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
938 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
939 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
940 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
941 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
942 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
943 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
944 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
945 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
946 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
947 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
948 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
949 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
950 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
951 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
952 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
954 -- Berlin, 2014-02-20
958 * logind has gained support for facilitating privileged input
959 and drm device access for unprivileged clients. This work is
960 useful to allow Wayland display servers (and similar
961 programs, such as kmscon) to run under the user's ID and
962 access input and drm devices which are normally
963 protected. When this is used (and the kernel is new enough)
964 logind will "mute" IO on the file descriptors passed to
965 Wayland as long as it is in the background and "unmute" it
966 if it returns into the foreground. This allows secure
967 session switching without allowing background sessions to
968 eavesdrop on input and display data. This also introduces
969 session switching support if VT support is turned off in the
970 kernel, and on seats that are not seat0.
972 * A new kernel command line option luks.options= is understood
973 now which allows specifiying LUKS options for usage for LUKS
974 encrypted partitions specified with luks.uuid=.
976 * tmpfiles.d(5) snippets may now use specifier expansion in
977 path names. More specifically %m, %b, %H, %v, are now
978 replaced by the local machine id, boot id, hostname, and
979 kernel version number.
981 * A new tmpfiles.d(5) command "m" has been introduced which
982 may be used to change the owner/group/access mode of a file
983 or directory if it exists, but do nothing if it does not.
985 * This release removes high-level support for the
986 MemorySoftLimit= cgroup setting. The underlying kernel
987 cgroup attribute memory.soft_limit= is currently badly
988 designed and likely to be removed from the kernel API in its
989 current form, hence we should not expose it for now.
991 * The memory.use_hierarchy cgroup attribute is now enabled for
992 all cgroups systemd creates in the memory cgroup
993 hierarchy. This option is likely to be come the built-in
994 default in the kernel anyway, and the non-hierarchial mode
995 never made much sense in the intrinsically hierarchial
998 * A new field _SYSTEMD_SLICE= is logged along with all journal
999 messages containing the slice a message was generated
1000 from. This is useful to allow easy per-customer filtering of
1001 logs among other things.
1003 * systemd-journald will no longer adjust the group of journal
1004 files it creates to the "systemd-journal" group. Instead we
1005 rely on the journal directory to be owned by the
1006 "systemd-journal" group, and its setgid bit set, so that the
1007 kernel file system layer will automatically enforce that
1008 journal files inherit this group assignment. The reason for
1009 this change is that we cannot allow NSS look-ups from
1010 journald which would be necessary to resolve
1011 "systemd-journal" to a numeric GID, because this might
1012 create deadlocks if NSS involves synchronous queries to
1013 other daemons (such as nscd, or sssd) which in turn are
1014 logging clients of journald and might block on it, which
1015 would then dead lock. A tmpfiles.d(5) snippet included in
1016 systemd will make sure the setgid bit and group are
1017 properly set on the journal directory if it exists on every
1018 boot. However, we recommend adjusting it manually after
1019 upgrades too (or from RPM scriptlets), so that the change is
1020 not delayed until next reboot.
1022 * Backlight and random seed files in /var/lib/ have moved into
1023 the /var/lib/systemd/ directory, in order to centralize all
1024 systemd generated files in one directory.
1026 * Boot time performance measurements (as displayed by
1027 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
1028 performance information if that's available to determine how
1029 much time BIOS and boot loader initialization required. With
1030 a sufficiently new BIOS you hence no longer need to boot
1031 with Gummiboot to get access to such information.
1033 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
1034 Cristian Rodríguez, Dave Reisner, David Herrmann, David
1035 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
1036 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
1037 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
1038 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
1039 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1041 -- Berlin, 2013-10-02
1045 * The Restart= option for services now understands a new
1046 on-watchdog setting, which will restart the service
1047 automatically if the service stops sending out watchdog keep
1048 alive messages (as configured with WatchdogSec=).
1050 * The getty generator (which is responsible for bringing up a
1051 getty on configured serial consoles) will no longer only
1052 start a getty on the primary kernel console but on all
1053 others, too. This makes the order in which console= is
1054 specified on the kernel command line less important.
1056 * libsystemd-logind gained a new sd_session_get_vt() call to
1057 retrieve the VT number of a session.
1059 * If the option "tries=0" is set for an entry of /etc/crypttab
1060 its passphrase is queried indefinitely instead of any
1061 maximum number of tries.
1063 * If a service with a configure PID file terminates its PID
1064 file will now be removed automatically if it still exists
1065 afterwards. This should put an end to stale PID files.
1067 * systemd-run will now also take relative binary path names
1068 for execution and no longer insists on absolute paths.
1070 * InaccessibleDirectories= and ReadOnlyDirectories= now take
1071 paths that are optionally prefixed with "-" to indicate that
1072 it should not be considered a failure if they do not exist.
1074 * journalctl -o (and similar commands) now understands a new
1075 output mode "short-precise", it is similar to "short" but
1076 shows timestamps with usec accuracy.
1078 * The option "discard" (as known from Debian) is now
1079 synonymous to "allow-discards" in /etc/crypttab. In fact,
1080 "discard" is preferred now (since it is easier to remember
1083 * Some licensing clean-ups were made, so that more code is now
1084 LGPL-2.1 licensed than before.
1086 * A minimal tool to save/restore the display backlight
1087 brightness across reboots has been added. It will store the
1088 backlight setting as late as possible at shutdown, and
1089 restore it as early as possible during reboot.
1091 * A logic to automatically discover and enable home and swap
1092 partitions on GPT disks has been added. With this in place
1093 /etc/fstab becomes optional for many setups as systemd can
1094 discover certain partitions located on the root disk
1095 automatically. Home partitions are recognized under their
1096 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
1097 partitions are recognized under their GPT type ID
1098 0657fd6da4ab43c484e50933c84b4f4f.
1100 * systemd will no longer pass any environment from the kernel
1101 or initrd to system services. If you want to set an
1102 environment for all services, do so via the kernel command
1103 line systemd.setenv= assignment.
1105 * The systemd-sysctl tool no longer natively reads the file
1106 /etc/sysctl.conf. If desired, the file should be symlinked
1107 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
1108 legacy support by a symlink rather than built-in code, it
1109 also makes the otherwise hidden order of application of the
1110 different files visible. (Note that this partly reverts to a
1111 pre-198 application order of sysctl knobs!)
1113 * The "systemctl set-log-level" and "systemctl dump" commands
1114 have been moved to systemd-analyze.
1116 * systemd-run learned the new --remain-after-exit switch,
1117 which causes the scope unit not to be cleaned up
1118 automatically after the process terminated.
1120 * tmpfiles learned a new --exclude-prefix= switch to exclude
1121 certain paths from operation.
1123 * journald will now automatically flush all messages to disk
1124 as soon as a message of the log priorities CRIT, ALERT or
1127 Contributions from: Andrew Cook, Brandon Philips, Christian
1128 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
1129 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
1130 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
1131 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
1132 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
1133 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
1134 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
1135 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
1136 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
1137 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
1138 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
1139 William Giokas, Zbigniew Jędrzejewski-Szmek
1141 -- Berlin, 2013-09-13
1145 * The documentation has been updated to cover the various new
1146 concepts introduced with 205.
1148 * Unit files now understand the new %v specifier which
1149 resolves to the kernel version string as returned by "uname
1152 * systemctl now supports filtering the unit list output by
1153 load state, active state and sub state, using the new
1156 * "systemctl status" will now show the results of the
1157 condition checks (like ConditionPathExists= and similar) of
1158 the last start attempts of the unit. They are also logged to
1161 * "journalctl -b" may now be used to look for boot output of a
1162 specific boot. Try "journalctl -b -1" for the previous boot,
1163 but the syntax is substantially more powerful.
1165 * "journalctl --show-cursor" has been added which prints the
1166 cursor string the last shown log line. This may then be used
1167 with the new "journalctl --after-cursor=" switch to continue
1168 browsing logs from that point on.
1170 * "journalctl --force" may now be used to force regeneration
1173 * Creation of "dead" device nodes has been moved from udev
1174 into kmod and tmpfiles. Previously, udev would read the kmod
1175 databases to pre-generate dead device nodes based on meta
1176 information contained in kernel modules, so that these would
1177 be auto-loaded on access rather then at boot. As this
1178 does not really have much to do with the exposing actual
1179 kernel devices to userspace this has always been slightly
1180 alien in the udev codebase. Following the new scheme kmod
1181 will now generate a runtime snippet for tmpfiles from the
1182 module meta information and it now is tmpfiles' job to the
1183 create the nodes. This also allows overriding access and
1184 other parameters for the nodes using the usual tmpfiles
1185 facilities. As side effect this allows us to remove the
1186 CAP_SYS_MKNOD capability bit from udevd entirely.
1188 * logind's device ACLs may now be applied to these "dead"
1189 devices nodes too, thus finally allowing managed access to
1190 devices such as /dev/snd/sequencer whithout loading the
1191 backing module right-away.
1193 * A new RPM macro has been added that may be used to apply
1194 tmpfiles configuration during package installation.
1196 * systemd-detect-virt and ConditionVirtualization= now can
1197 detect User-Mode-Linux machines (UML).
1199 * journald will now implicitly log the effective capabilities
1200 set of processes in the message metadata.
1202 * systemd-cryptsetup has gained support for TrueCrypt volumes.
1204 * The initrd interface has been simplified (more specifically,
1205 support for passing performance data via environment
1206 variables and fsck results via files in /run has been
1207 removed). These features were non-essential, and are
1208 nowadays available in a much nicer way by having systemd in
1209 the initrd serialize its state and have the hosts systemd
1210 deserialize it again.
1212 * The udev "keymap" data files and tools to apply keyboard
1213 specific mappings of scan to key codes, and force-release
1214 scan code lists have been entirely replaced by a udev
1215 "keyboard" builtin and a hwdb data file.
1217 * systemd will now honour the kernel's "quiet" command line
1218 argument also during late shutdown, resulting in a
1219 completely silent shutdown when used.
1221 * There's now an option to control the SO_REUSEPORT socket
1222 option in .socket units.
1224 * Instance units will now automatically get a per-template
1225 subslice of system.slice unless something else is explicitly
1226 configured. For example, instances of sshd@.service will now
1227 implicitly be placed in system-sshd.slice rather than
1228 system.slice as before.
1230 * Test coverage support may now be enabled at build time.
1232 Contributions from: Dave Reisner, Frederic Crozat, Harald
1233 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
1234 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
1235 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
1236 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
1237 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
1238 Giokas, Zbigniew Jędrzejewski-Szmek
1240 -- Berlin, 2013-07-23
1244 * Two new unit types have been introduced:
1246 Scope units are very similar to service units, however, are
1247 created out of pre-existing processes -- instead of PID 1
1248 forking off the processes. By using scope units it is
1249 possible for system services and applications to group their
1250 own child processes (worker processes) in a powerful way
1251 which then maybe used to organize them, or kill them
1252 together, or apply resource limits on them.
1254 Slice units may be used to partition system resources in an
1255 hierarchial fashion and then assign other units to them. By
1256 default there are now three slices: system.slice (for all
1257 system services), user.slice (for all user sessions),
1258 machine.slice (for VMs and containers).
1260 Slices and scopes have been introduced primarily in
1261 context of the work to move cgroup handling to a
1262 single-writer scheme, where only PID 1
1263 creates/removes/manages cgroups.
1265 * There's a new concept of "transient" units. In contrast to
1266 normal units these units are created via an API at runtime,
1267 not from configuration from disk. More specifically this
1268 means it is now possible to run arbitrary programs as
1269 independent services, with all execution parameters passed
1270 in via bus APIs rather than read from disk. Transient units
1271 make systemd substantially more dynamic then it ever was,
1272 and useful as a general batch manager.
1274 * logind has been updated to make use of scope and slice units
1275 for managing user sessions. As a user logs in he will get
1276 his own private slice unit, to which all sessions are added
1277 as scope units. We also added support for automatically
1278 adding an instance of user@.service for the user into the
1279 slice. Effectively logind will no longer create cgroup
1280 hierarchies on its own now, it will defer entirely to PID 1
1281 for this by means of scope, service and slice units. Since
1282 user sessions this way become entities managed by PID 1
1283 the output of "systemctl" is now a lot more comprehensive.
1285 * A new mini-daemon "systemd-machined" has been added which
1286 may be used by virtualization managers to register local
1287 VMs/containers. nspawn has been updated accordingly, and
1288 libvirt will be updated shortly. machined will collect a bit
1289 of meta information about the VMs/containers, and assign
1290 them their own scope unit (see above). The collected
1291 meta-data is then made available via the "machinectl" tool,
1292 and exposed in "ps" and similar tools. machined/machinectl
1293 is compile-time optional.
1295 * As discussed earlier, the low-level cgroup configuration
1296 options ControlGroup=, ControlGroupModify=,
1297 ControlGroupPersistent=, ControlGroupAttribute= have been
1298 removed. Please use high-level attribute settings instead as
1299 well as slice units.
1301 * A new bus call SetUnitProperties() has been added to alter
1302 various runtime parameters of a unit. This is primarily
1303 useful to alter cgroup parameters dynamically in a nice way,
1304 but will be extended later on to make more properties
1305 modifiable at runtime. systemctl gained a new set-properties
1306 command that wraps this call.
1308 * A new tool "systemd-run" has been added which can be used to
1309 run arbitrary command lines as transient services or scopes,
1310 while configuring a number of settings via the command
1311 line. This tool is currently very basic, however already
1312 very useful. We plan to extend this tool to even allow
1313 queuing of execution jobs with time triggers from the
1314 command line, similar in fashion to "at".
1316 * nspawn will now inform the user explicitly that kernels with
1317 audit enabled break containers, and suggest the user to turn
1320 * Support for detecting the IMA and AppArmor security
1321 frameworks with ConditionSecurity= has been added.
1323 * journalctl gained a new "-k" switch for showing only kernel
1324 messages, mimicking dmesg output; in addition to "--user"
1325 and "--system" switches for showing only user's own logs
1328 * systemd-delta can now show information about drop-in
1329 snippets extending unit files.
1331 * libsystemd-bus has been substantially updated but is still
1332 not available as public API.
1334 * systemd will now look for the "debug" argument on the kernel
1335 command line and enable debug logging, similar to
1336 "systemd.log_level=debug" already did before.
1338 * "systemctl set-default", "systemctl get-default" has been
1339 added to configure the default.target symlink, which
1340 controls what to boot into by default.
1342 * "systemctl set-log-level" has been added as a convenient
1343 way to raise and lower systemd logging threshold.
1345 * "systemd-analyze plot" will now show the time the various
1346 generators needed for execution, as well as information
1347 about the unit file loading.
1349 * libsystemd-journal gained a new sd_journal_open_files() call
1350 for opening specific journal files. journactl also gained a
1351 new switch to expose this new functionality. Previously we
1352 only supported opening all files from a directory, or all
1353 files from the system, as opening individual files only is
1354 racy due to journal file rotation.
1356 * systemd gained the new DefaultEnvironment= setting in
1357 /etc/systemd/system.conf to set environment variables for
1360 * If a privileged process logs a journal message with the
1361 OBJECT_PID= field set, then journald will automatically
1362 augment this with additional OBJECT_UID=, OBJECT_GID=,
1363 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
1364 system services want to log events about specific client
1365 processes. journactl/systemctl has been updated to make use
1366 of this information if all log messages regarding a specific
1369 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
1370 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
1371 Reisner, David Coppa, David King, David Strauss, Eelco
1372 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
1373 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
1374 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
1375 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
1376 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
1377 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
1378 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
1379 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
1380 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
1381 Łukasz Stelmach, 장동준
1385 * The Python bindings gained some minimal support for the APIs
1386 exposed by libsystemd-logind.
1388 * ConditionSecurity= gained support for detecting SMACK. Since
1389 this condition already supports SELinux and AppArmor we only
1390 miss IMA for this. Patches welcome!
1392 Contributions from: Karol Lewandowski, Lennart Poettering,
1393 Zbigniew Jędrzejewski-Szmek
1397 * systemd-nspawn will now create /etc/resolv.conf if
1398 necessary, before bind-mounting the host's file onto it.
1400 * systemd-nspawn will now store meta information about a
1401 container on the container's cgroup as extended attribute
1402 fields, including the root directory.
1404 * The cgroup hierarchy has been reworked in many ways. All
1405 objects any of the components systemd creates in the cgroup
1406 tree are now suffixed. More specifically, user sessions are
1407 now placed in cgroups suffixed with ".session", users in
1408 cgroups suffixed with ".user", and nspawn containers in
1409 cgroups suffixed with ".nspawn". Furthermore, all cgroup
1410 names are now escaped in a simple scheme to avoid collision
1411 of userspace object names with kernel filenames. This work
1412 is preparation for making these objects relocatable in the
1413 cgroup tree, in order to allow easy resource partitioning of
1414 these objects without causing naming conflicts.
1416 * systemctl list-dependencies gained the new switches
1417 --plain, --reverse, --after and --before.
1419 * systemd-inhibit now shows the process name of processes that
1420 have taken an inhibitor lock.
1422 * nss-myhostname will now also resolve "localhost"
1423 implicitly. This makes /etc/hosts an optional file and
1424 nicely handles that on IPv6 ::1 maps to both "localhost" and
1427 * libsystemd-logind.so gained a new call
1428 sd_get_machine_names() to enumerate running containers and
1429 VMs (currently only supported by very new libvirt and
1430 nspawn). sd_login_monitor can now be used to watch
1431 VMs/containers coming and going.
1433 * .include is not allowed recursively anymore, and only in
1434 unit files. Usually it is better to use drop-in snippets in
1435 .d/*.conf anyway, as introduced with systemd 198.
1437 * systemd-analyze gained a new "critical-chain" command that
1438 determines the slowest chain of units run during system
1439 boot-up. It is very useful for tracking down where
1440 optimizing boot time is the most beneficial.
1442 * systemd will no longer allow manipulating service paths in
1443 the name=systemd:/system cgroup tree using ControlGroup= in
1444 units. (But is still fine with it in all other dirs.)
1446 * There's a new systemd-nspawn@.service service file that may
1447 be used to easily run nspawn containers as system
1448 services. With the container's root directory in
1449 /var/lib/container/foobar it is now sufficient to run
1450 "systemctl start systemd-nspawn@foobar.service" to boot it.
1452 * systemd-cgls gained a new parameter "--machine" to list only
1453 the processes within a certain container.
1455 * ConditionSecurity= now can check for "apparmor". We still
1456 are lacking checks for SMACK and IMA for this condition
1457 check though. Patches welcome!
1459 * A new configuration file /etc/systemd/sleep.conf has been
1460 added that may be used to configure which kernel operation
1461 systemd is supposed to execute when "suspend", "hibernate"
1462 or "hybrid-sleep" is requested. This makes the new kernel
1463 "freeze" state accessible to the user.
1465 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
1466 the passed argument if applicable.
1468 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
1469 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
1470 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
1471 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
1472 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
1473 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
1474 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
1479 * The output of 'systemctl list-jobs' got some polishing. The
1480 '--type=' argument may now be passed more than once. A new
1481 command 'systemctl list-sockets' has been added which shows
1482 a list of kernel sockets systemd is listening on with the
1483 socket units they belong to, plus the units these socket
1486 * The experimental libsystemd-bus library got substantial
1487 updates to work in conjunction with the (also experimental)
1488 kdbus kernel project. It works well enough to exchange
1489 messages with some sophistication. Note that kdbus is not
1490 ready yet, and the library is mostly an elaborate test case
1491 for now, and not installable.
1493 * systemd gained a new unit 'systemd-static-nodes.service'
1494 that generates static device nodes earlier during boot, and
1495 can run in conjunction with udev.
1497 * libsystemd-login gained a new call sd_pid_get_user_unit()
1498 to retrieve the user systemd unit a process is running
1499 in. This is useful for systems where systemd is used as
1502 * systemd-nspawn now places all containers in the new /machine
1503 top-level cgroup directory in the name=systemd
1504 hierarchy. libvirt will soon do the same, so that we get a
1505 uniform separation of /system, /user and /machine for system
1506 services, user processes and containers/virtual
1507 machines. This new cgroup hierarchy is also useful to stick
1508 stable names to specific container instances, which can be
1509 recognized later this way (this name may be controlled
1510 via systemd-nspawn's new -M switch). libsystemd-login also
1511 gained a new call sd_pid_get_machine_name() to retrieve the
1512 name of the container/VM a specific process belongs to.
1514 * bootchart can now store its data in the journal.
1516 * libsystemd-journal gained a new call
1517 sd_journal_add_conjunction() for AND expressions to the
1518 matching logic. This can be used to express more complex
1519 logical expressions.
1521 * journactl can now take multiple --unit= and --user-unit=
1524 * The cryptsetup logic now understands the "luks.key=" kernel
1525 command line switch for specifying a file to read the
1526 decryption key from. Also, if a configured key file is not
1527 found the tool will now automatically fall back to prompting
1530 * Python systemd.journal module was updated to wrap recently
1531 added functions from libsystemd-journal. The interface was
1532 changed to bring the low level interface in s.j._Reader
1533 closer to the C API, and the high level interface in
1534 s.j.Reader was updated to wrap and convert all data about
1537 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
1538 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
1539 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
1540 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
1541 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
1542 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1546 * journalctl --update-catalog now understands a new --root=
1547 option to operate on catalogs found in a different root
1550 * During shutdown after systemd has terminated all running
1551 services a final killing loop kills all remaining left-over
1552 processes. We will now print the name of these processes
1553 when we send SIGKILL to them, since this usually indicates a
1556 * If /etc/crypttab refers to password files stored on
1557 configured mount points automatic dependencies will now be
1558 generated to ensure the specific mount is established first
1559 before the key file is attempted to be read.
1561 * 'systemctl status' will now show information about the
1562 network sockets a socket unit is listening on.
1564 * 'systemctl status' will also shown information about any
1565 drop-in configuration file for units. (Drop-In configuration
1566 files in this context are files such as
1567 /etc/systemd/systemd/foobar.service.d/*.conf)
1569 * systemd-cgtop now optionally shows summed up CPU times of
1570 cgroups. Press '%' while running cgtop to switch between
1571 percentage and absolute mode. This is useful to determine
1572 which cgroups use up the most CPU time over the entire
1573 runtime of the system. systemd-cgtop has also been updated
1574 to be 'pipeable' for processing with further shell tools.
1576 * 'hostnamectl set-hostname' will now allow setting of FQDN
1579 * The formatting and parsing of time span values has been
1580 changed. The parser now understands fractional expressions
1581 such as "5.5h". The formatter will now output fractional
1582 expressions for all time spans under 1min, i.e. "5.123456s"
1583 rather than "5s 123ms 456us". For time spans under 1s
1584 millisecond values are shown, for those under 1ms
1585 microsecond values are shown. This should greatly improve
1586 all time-related output of systemd.
1588 * libsystemd-login and libsystemd-journal gained new
1589 functions for querying the poll() events mask and poll()
1590 timeout value for integration into arbitrary event
1593 * localectl gained the ability to list available X11 keymaps
1594 (models, layouts, variants, options).
1596 * 'systemd-analyze dot' gained the ability to filter for
1597 specific units via shell-style globs, to create smaller,
1598 more useful graphs. I.e. it is now possible to create simple
1599 graphs of all the dependencies between only target units, or
1600 of all units that Avahi has dependencies with.
1602 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
1603 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
1604 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
1605 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
1606 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
1607 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
1608 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
1612 * The boot-time readahead implementation for rotating media
1613 will now read the read-ahead data in multiple passes which
1614 consist of all read requests made in equidistant time
1615 intervals. This means instead of strictly reading read-ahead
1616 data in its physical order on disk we now try to find a
1617 middle ground between physical and access time order.
1619 * /etc/os-release files gained a new BUILD_ID= field for usage
1620 on operating systems that provide continuous builds of OS
1623 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
1624 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
1625 William Douglas, Zbigniew Jędrzejewski-Szmek
1629 * systemd-python gained an API exposing libsystemd-daemon.
1631 * The SMACK setup logic gained support for uploading CIPSO
1634 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
1635 ReadOnlyDirectories= and InaccessibleDirectories= has
1636 changed. The private /tmp and /var/tmp directories are now
1637 shared by all processes of a service (which means
1638 ExecStartPre= may now leave data in /tmp that ExecStart= of
1639 the same service can still access). When a service is
1640 stopped its temporary directories are immediately deleted
1641 (normal clean-up with tmpfiles is still done in addition to
1644 * By default, systemd will now set a couple of sysctl
1645 variables in the kernel: the safe sysrq options are turned
1646 on, IP route verification is turned on, and source routing
1647 disabled. The recently added hardlink and softlink
1648 protection of the kernel is turned on. These settings should
1649 be reasonably safe, and good defaults for all new systems.
1651 * The predictable network naming logic may now be turned off
1652 with a new kernel command line switch: net.ifnames=0.
1654 * A new libsystemd-bus module has been added that implements a
1655 pretty complete D-Bus client library. For details see:
1657 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
1659 * journald will now explicitly flush the journal files to disk
1660 at the latest 5min after each write. The file will then also
1661 be marked offline until the next write. This should increase
1662 reliability in case of a crash. The synchronization delay
1663 can be configured via SyncIntervalSec= in journald.conf.
1665 * There's a new remote-fs-setup.target unit that can be used
1666 to pull in specific services when at least one remote file
1667 system is to be mounted.
1669 * There are new targets timers.target and paths.target as
1670 canonical targets to pull user timer and path units in
1671 from. This complements sockets.target with a similar
1672 purpose for socket units.
1674 * libudev gained a new call udev_device_set_attribute_value()
1675 to set sysfs attributes of a device.
1677 * The udev daemon now sets the default number of worker
1678 processes executed in parallel based on the number of available
1679 CPUs instead of the amount of available RAM. This is supposed
1680 to provide a more reliable default and limit a too aggressive
1681 paralellism for setups with 1000s of devices connected.
1683 Contributions from: Auke Kok, Colin Walters, Cristian
1684 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
1685 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
1686 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
1687 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
1688 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
1689 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
1690 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
1691 Zbigniew Jędrzejewski-Szmek
1695 * Configuration of unit files may now be extended via drop-in
1696 files without having to edit/override the unit files
1697 themselves. More specifically, if the administrator wants to
1698 change one value for a service file foobar.service he can
1699 now do so by dropping in a configuration snippet into
1700 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
1701 will load all these snippets and apply them on top of the
1702 main unit configuration file, possibly extending or
1703 overriding its settings. Using these drop-in snippets is
1704 generally nicer than the two earlier options for changing
1705 unit files locally: copying the files from
1706 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
1707 them there; or creating a new file in /etc/systemd/system/
1708 that incorporates the original one via ".include". Drop-in
1709 snippets into these .d/ directories can be placed in any
1710 directory systemd looks for units in, and the usual
1711 overriding semantics between /usr/lib, /etc and /run apply
1714 * Most unit file settings which take lists of items can now be
1715 reset by assigning the empty string to them. For example,
1716 normally, settings such as Environment=FOO=BAR append a new
1717 environment variable assignment to the environment block,
1718 each time they are used. By assigning Environment= the empty
1719 string the environment block can be reset to empty. This is
1720 particularly useful with the .d/*.conf drop-in snippets
1721 mentioned above, since this adds the ability to reset list
1722 settings from vendor unit files via these drop-ins.
1724 * systemctl gained a new "list-dependencies" command for
1725 listing the dependencies of a unit recursively.
1727 * Inhibitors are now honored and listed by "systemctl
1728 suspend", "systemctl poweroff" (and similar) too, not only
1729 GNOME. These commands will also list active sessions by
1732 * Resource limits (as exposed by the various control group
1733 controllers) can now be controlled dynamically at runtime
1734 for all units. More specifically, you can now use a command
1735 like "systemctl set-cgroup-attr foobar.service cpu.shares
1736 2000" to alter the CPU shares a specific service gets. These
1737 settings are stored persistently on disk, and thus allow the
1738 administrator to easily adjust the resource usage of
1739 services with a few simple commands. This dynamic resource
1740 management logic is also available to other programs via the
1741 bus. Almost any kernel cgroup attribute and controller is
1744 * systemd-vconsole-setup will now copy all font settings to
1745 all allocated VTs, where it previously applied them only to
1748 * libsystemd-login gained the new sd_session_get_tty() API
1751 * This release drops support for a few legacy or
1752 distribution-specific LSB facility names when parsing init
1753 scripts: $x-display-manager, $mail-transfer-agent,
1754 $mail-transport-agent, $mail-transfer-agent, $smtp,
1755 $null. Also, the mail-transfer-agent.target unit backing
1756 this has been removed. Distributions which want to retain
1757 compatibility with this should carry the burden for
1758 supporting this themselves and patch support for these back
1759 in, if they really need to. Also, the facilities $syslog and
1760 $local_fs are now ignored, since systemd does not support
1761 early-boot LSB init scripts anymore, and these facilities
1762 are implied anyway for normal services. syslog.target has
1765 * There are new bus calls on PID1's Manager object for
1766 cancelling jobs, and removing snapshot units. Previously,
1767 both calls were only available on the Job and Snapshot
1770 * systemd-journal-gatewayd gained SSL support.
1772 * The various "environment" files, such as /etc/locale.conf
1773 now support continuation lines with a backslash ("\") as
1774 last character in the line, similar in style (but different)
1775 to how this is supported in shells.
1777 * For normal user processes the _SYSTEMD_USER_UNIT= field is
1778 now implicitly appended to every log entry logged. systemctl
1779 has been updated to filter by this field when operating on a
1780 user systemd instance.
1782 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
1783 CAP_AUDIT_CONTROL capabilities to the capabilities set for
1784 the container. This makes it easier to boot unmodified
1785 Fedora systems in a container, which however still requires
1786 audit=0 to be passed on the kernel command line. Auditing in
1787 kernel and userspace is unfortunately still too broken in
1788 context of containers, hence we recommend compiling it out
1789 of the kernel or using audit=0. Hopefully this will be fixed
1790 one day for good in the kernel.
1792 * nspawn gained the new --bind= and --bind-ro= parameters to
1793 bind mount specific directories from the host into the
1796 * nspawn will now mount its own devpts file system instance
1797 into the container, in order not to leak pty devices from
1798 the host into the container.
1800 * systemd will now read the firmware boot time performance
1801 information from the EFI variables, if the used boot loader
1802 supports this, and takes it into account for boot performance
1803 analysis via "systemd-analyze". This is currently supported
1804 only in conjunction with Gummiboot, but could be supported
1805 by other boot loaders too. For details see:
1807 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
1809 * A new generator has been added that automatically mounts the
1810 EFI System Partition (ESP) to /boot, if that directory
1811 exists, is empty, and no other file system has been
1812 configured to be mounted there.
1814 * logind will now send out PrepareForSleep(false) out
1815 unconditionally, after coming back from suspend. This may be
1816 used by applications as asynchronous notification for
1817 system resume events.
1819 * "systemctl unlock-sessions" has been added, that allows
1820 unlocking the screens of all user sessions at once, similar
1821 how "systemctl lock-sessions" already locked all users
1822 sessions. This is backed by a new D-Bus call UnlockSessions().
1824 * "loginctl seat-status" will now show the master device of a
1825 seat. (i.e. the device of a seat that needs to be around for
1826 the seat to be considered available, usually the graphics
1829 * tmpfiles gained a new "X" line type, that allows
1830 configuration of files and directories (with wildcards) that
1831 shall be excluded from automatic cleanup ("aging").
1833 * udev default rules set the device node permissions now only
1834 at "add" events, and do not change them any longer with a
1835 later "change" event.
1837 * The log messages for lid events and power/sleep keypresses
1838 now carry a message ID.
1840 * We now have a substantially larger unit test suite, but this
1841 continues to be work in progress.
1843 * udevadm hwdb gained a new --root= parameter to change the
1844 root directory to operate relative to.
1846 * logind will now issue a background sync() request to the kernel
1847 early at shutdown, so that dirty buffers are flushed to disk early
1848 instead of at the last moment, in order to optimize shutdown
1851 * A new bootctl tool has been added that is an interface for
1852 certain boot loader operations. This is currently a preview
1853 and is likely to be extended into a small mechanism daemon
1854 like timedated, localed, hostnamed, and can be used by
1855 graphical UIs to enumerate available boot options, and
1856 request boot into firmware operations.
1858 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
1859 the rest of the package. It also has been updated to work
1860 correctly in initrds.
1862 * Policykit previously has been runtime optional, and is now
1863 also compile time optional via a configure switch.
1865 * systemd-analyze has been reimplemented in C. Also "systemctl
1866 dot" has moved into systemd-analyze.
1868 * "systemctl status" with no further parameters will now print
1869 the status of all active or failed units.
1871 * Operations such as "systemctl start" can now be executed
1872 with a new mode "--irreversible" which may be used to queue
1873 operations that cannot accidentally be reversed by a later
1874 job queuing. This is by default used to make shutdown
1875 requests more robust.
1877 * The Python API of systemd now gained a new module for
1878 reading journal files.
1880 * A new tool kernel-install has been added that can install
1881 kernel images according to the Boot Loader Specification:
1883 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
1885 * Boot time console output has been improved to provide
1886 animated boot time output for hanging jobs.
1888 * A new tool systemd-activate has been added which can be used
1889 to test socket activation with, directly from the command
1890 line. This should make it much easier to test and debug
1891 socket activation in daemons.
1893 * journalctl gained a new "--reverse" (or -r) option to show
1894 journal output in reverse order (i.e. newest line first).
1896 * journalctl gained a new "--pager-end" (or -e) option to jump
1897 to immediately jump to the end of the journal in the
1898 pager. This is only supported in conjunction with "less".
1900 * journalctl gained a new "--user-unit=" option, that works
1901 similar to "--unit=" but filters for user units rather than
1904 * A number of unit files to ease adoption of systemd in
1905 initrds has been added. This moves some minimal logic from
1906 the various initrd implementations into systemd proper.
1908 * The journal files are now owned by a new group
1909 "systemd-journal", which exists specifically to allow access
1910 to the journal, and nothing else. Previously, we used the
1911 "adm" group for that, which however possibly covers more
1912 than just journal/log file access. This new group is now
1913 already used by systemd-journal-gatewayd to ensure this
1914 daemon gets access to the journal files and as little else
1915 as possible. Note that "make install" will also set FS ACLs
1916 up for /var/log/journal to give "adm" and "wheel" read
1917 access to it, in addition to "systemd-journal" which owns
1918 the journal files. We recommend that packaging scripts also
1919 add read access to "adm" + "wheel" to /var/log/journal, and
1920 all existing/future journal files. To normal users and
1921 administrators little changes, however packagers need to
1922 ensure to create the "systemd-journal" system group at
1923 package installation time.
1925 * The systemd-journal-gatewayd now runs as unprivileged user
1926 systemd-journal-gateway:systemd-journal-gateway. Packaging
1927 scripts need to create these system user/group at
1930 * timedated now exposes a new boolean property CanNTP that
1931 indicates whether a local NTP service is available or not.
1933 * systemd-detect-virt will now also detect xen PVs
1935 * The pstore file system is now mounted by default, if it is
1938 * In addition to the SELinux and IMA policies we will now also
1939 load SMACK policies at early boot.
1941 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
1942 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
1943 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
1944 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
1945 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
1946 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
1947 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
1948 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
1949 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
1950 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
1951 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
1952 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
1953 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
1954 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
1958 * Timer units now support calendar time events in addition to
1959 monotonic time events. That means you can now trigger a unit
1960 based on a calendar time specification such as "Thu,Fri
1961 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
1962 or fifth day of any month of the year 2013, given that it is
1963 a thursday or friday. This brings timer event support
1964 considerably closer to cron's capabilities. For details on
1965 the supported calendar time specification language see
1968 * udev now supports a number of different naming policies for
1969 network interfaces for predictable names, and a combination
1970 of these policies is now the default. Please see this wiki
1971 document for details:
1973 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
1975 * Auke Kok's bootchart implementation has been added to the
1976 systemd tree. It is an optional component that can graph the
1977 boot in quite some detail. It is one of the best bootchart
1978 implementations around and minimal in its code and
1981 * nss-myhostname has been integrated into the systemd source
1982 tree. nss-myhostname guarantees that the local hostname
1983 always stays resolvable via NSS. It has been a weak
1984 requirement of systemd-hostnamed since a long time, and
1985 since its code is actually trivial we decided to just
1986 include it in systemd's source tree. It can be turned off
1987 with a configure switch.
1989 * The read-ahead logic is now capable of properly detecting
1990 whether a btrfs file system is on SSD or rotating media, in
1991 order to optimize the read-ahead scheme. Previously, it was
1992 only capable of detecting this on traditional file systems
1995 * In udev, additional device properties are now read from the
1996 IAB in addition to the OUI database. Also, Bluetooth company
1997 identities are attached to the devices as well.
1999 * In service files %U may be used as specifier that is
2000 replaced by the configured user name of the service.
2002 * nspawn may now be invoked without a controlling TTY. This
2003 makes it suitable for invocation as its own service. This
2004 may be used to set up a simple containerized server system
2005 using only core OS tools.
2007 * systemd and nspawn can now accept socket file descriptors
2008 when they are started for socket activation. This enables
2009 implementation of socket activated nspawn
2010 containers. i.e. think about autospawning an entire OS image
2011 when the first SSH or HTTP connection is received. We expect
2012 that similar functionality will also be added to libvirt-lxc
2015 * journalctl will now suppress ANSI color codes when
2016 presenting log data.
2018 * systemctl will no longer show control group information for
2019 a unit if a the control group is empty anyway.
2021 * logind can now automatically suspend/hibernate/shutdown the
2024 * /etc/machine-info and hostnamed now also expose the chassis
2025 type of the system. This can be used to determine whether
2026 the local system is a laptop, desktop, handset or
2027 tablet. This information may either be configured by the
2028 user/vendor or is automatically determined from ACPI and DMI
2029 information if possible.
2031 * A number of PolicyKit actions are now bound together with
2032 "imply" rules. This should simplify creating UIs because
2033 many actions will now authenticate similar ones as well.
2035 * Unit files learnt a new condition ConditionACPower= which
2036 may be used to conditionalize a unit depending on whether an
2037 AC power source is connected or not, of whether the system
2038 is running on battery power.
2040 * systemctl gained a new "is-failed" verb that may be used in
2041 shell scripts and suchlike to check whether a specific unit
2042 is in the "failed" state.
2044 * The EnvironmentFile= setting in unit files now supports file
2045 globbing, and can hence be used to easily read a number of
2046 environment files at once.
2048 * systemd will no longer detect and recognize specific
2049 distributions. All distribution-specific #ifdeffery has been
2050 removed, systemd is now fully generic and
2051 distribution-agnostic. Effectively, not too much is lost as
2052 a lot of the code is still accessible via explicit configure
2053 switches. However, support for some distribution specific
2054 legacy configuration file formats has been dropped. We
2055 recommend distributions to simply adopt the configuration
2056 files everybody else uses now and convert the old
2057 configuration from packaging scripts. Most distributions
2058 already did that. If that's not possible or desirable,
2059 distributions are welcome to forward port the specific
2060 pieces of code locally from the git history.
2062 * When logging a message about a unit systemd will now always
2063 log the unit name in the message meta data.
2065 * localectl will now also discover system locale data that is
2066 not stored in locale archives, but directly unpacked.
2068 * logind will no longer unconditionally use framebuffer
2069 devices as seat masters, i.e. as devices that are required
2070 to be existing before a seat is considered preset. Instead,
2071 it will now look for all devices that are tagged as
2072 "seat-master" in udev. By default framebuffer devices will
2073 be marked as such, but depending on local systems other
2074 devices might be marked as well. This may be used to
2075 integrate graphics cards using closed source drivers (such
2076 as NVidia ones) more nicely into logind. Note however, that
2077 we recommend using the open source NVidia drivers instead,
2078 and no udev rules for the closed-source drivers will be
2079 shipped from us upstream.
2081 Contributions from: Adam Williamson, Alessandro Crismani, Auke
2082 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
2083 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
2084 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
2085 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
2086 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
2087 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
2088 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
2089 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
2090 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
2091 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
2092 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
2097 * udev gained support for loading additional device properties
2098 from an indexed database that is keyed by vendor/product IDs
2099 and similar device identifiers. For the beginning this
2100 "hwdb" is populated with data from the well-known PCI and
2101 USB database, but also includes PNP, ACPI and OID data. In
2102 the longer run this indexed database shall grow into
2103 becoming the one central database for non-essential
2104 userspace device metadata. Previously, data from the PCI/USB
2105 database was only attached to select devices, since the
2106 lookup was a relatively expensive operation due to O(n) time
2107 complexity (with n being the number of entries in the
2108 database). Since this is now O(1), we decided to add in this
2109 data for all devices where this is available, by
2110 default. Note that the indexed database needs to be rebuilt
2111 when new data files are installed. To achieve this you need
2112 to update your packaging scripts to invoke "udevadm hwdb
2113 --update" after installation of hwdb data files. For
2114 RPM-based distributions we introduced the new
2115 %udev_hwdb_update macro for this purpose.
2117 * The Journal gained support for the "Message Catalog", an
2118 indexed database to link up additional information with
2119 journal entries. For further details please check:
2121 http://www.freedesktop.org/wiki/Software/systemd/catalog
2123 The indexed message catalog database also needs to be
2124 rebuilt after installation of message catalog files. Use
2125 "journalctl --update-catalog" for this. For RPM-based
2126 distributions we introduced the %journal_catalog_update
2127 macro for this purpose.
2129 * The Python Journal bindings gained support for the standard
2130 Python logging framework.
2132 * The Journal API gained new functions for checking whether
2133 the underlying file system of a journal file is capable of
2134 properly reporting file change notifications, or whether
2135 applications that want to reflect journal changes "live"
2136 need to recheck journal files continuously in appropriate
2139 * It is now possible to set the "age" field for tmpfiles
2140 entries to 0, indicating that files matching this entry
2141 shall always be removed when the directories are cleaned up.
2143 * coredumpctl gained a new "gdb" verb which invokes gdb
2144 right-away on the selected coredump.
2146 * There's now support for "hybrid sleep" on kernels that
2147 support this, in addition to "suspend" and "hibernate". Use
2148 "systemctl hybrid-sleep" to make use of this.
2150 * logind's HandleSuspendKey= setting (and related settings)
2151 now gained support for a new "lock" setting to simply
2152 request the screen lock on all local sessions, instead of
2153 actually executing a suspend or hibernation.
2155 * systemd will now mount the EFI variables file system by
2158 * Socket units now gained support for configuration of the
2159 SMACK security label.
2161 * timedatectl will now output the time of the last and next
2162 daylight saving change.
2164 * We dropped support for various legacy and distro-specific
2165 concepts, such as insserv, early-boot SysV services
2166 (i.e. those for non-standard runlevels such as 'b' or 'S')
2167 or ArchLinux /etc/rc.conf support. We recommend the
2168 distributions who still need support this to either continue
2169 to maintain the necessary patches downstream, or find a
2170 different solution. (Talk to us if you have questions!)
2172 * Various systemd components will now bypass PolicyKit checks
2173 for root and otherwise handle properly if PolicyKit is not
2174 found to be around. This should fix most issues for
2175 PolicyKit-less systems. Quite frankly this should have been
2176 this way since day one. It is absolutely our intention to
2177 make systemd work fine on PolicyKit-less systems, and we
2178 consider it a bug if something does not work as it should if
2179 PolicyKit is not around.
2181 * For embedded systems it is now possible to build udev and
2182 systemd without blkid and/or kmod support.
2184 * "systemctl switch-root" is now capable of switching root
2185 more than once. I.e. in addition to transitions from the
2186 initrd to the host OS it is now possible to transition to
2187 further OS images from the host. This is useful to implement
2188 offline updating tools.
2190 * Various other additions have been made to the RPM macros
2191 shipped with systemd. Use %udev_rules_update() after
2192 installing new udev rules files. %_udevhwdbdir,
2193 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
2194 %_sysctldir are now available which resolve to the right
2195 directories for packages to place various data files in.
2197 * journalctl gained the new --full switch (in addition to
2198 --all, to disable ellipsation for long messages.
2200 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
2201 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
2202 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
2203 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
2204 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
2205 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
2206 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
2207 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
2208 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
2212 * journalctl gained new --since= and --until= switches to
2213 filter by time. It also now supports nice filtering for
2214 units via --unit=/-u.
2216 * Type=oneshot services may use ExecReload= and do the
2219 * The journal daemon now supports time-based rotation and
2220 vacuuming, in addition to the usual disk-space based
2223 * The journal will now index the available field values for
2224 each field name. This enables clients to show pretty drop
2225 downs of available match values when filtering. The bash
2226 completion of journalctl has been updated
2227 accordingly. journalctl gained a new switch -F to list all
2228 values a certain field takes in the journal database.
2230 * More service events are now written as structured messages
2231 to the journal, and made recognizable via message IDs.
2233 * The timedated, localed and hostnamed mini-services which
2234 previously only provided support for changing time, locale
2235 and hostname settings from graphical DEs such as GNOME now
2236 also have a minimal (but very useful) text-based client
2237 utility each. This is probably the nicest way to changing
2238 these settings from the command line now, especially since
2239 it lists available options and is fully integrated with bash
2242 * There's now a new tool "systemd-coredumpctl" to list and
2243 extract coredumps from the journal.
2245 * We now install a README each in /var/log/ and
2246 /etc/rc.d/init.d explaining where the system logs and init
2247 scripts went. This hopefully should help folks who go to
2248 that dirs and look into the otherwise now empty void and
2249 scratch their heads.
2251 * When user-services are invoked (by systemd --user) the
2252 $MANAGERPID env var is set to the PID of systemd.
2254 * SIGRTMIN+24 when sent to a --user instance will now result
2255 in immediate termination of systemd.
2257 * gatewayd received numerous feature additions such as a
2258 "follow" mode, for live syncing and filtering.
2260 * browse.html now allows filtering and showing detailed
2261 information on specific entries. Keyboard navigation and
2262 mouse screen support has been added.
2264 * gatewayd/journalctl now supports HTML5/JSON
2265 Server-Sent-Events as output.
2267 * The SysV init script compatibility logic will now
2268 heuristically determine whether a script supports the
2269 "reload" verb, and only then make this available as
2272 * "systemctl status --follow" has been removed, use "journalctl
2275 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
2276 have been removed since they are hardly useful to be
2279 * And I'd like to take the opportunity to specifically mention
2280 Zbigniew for his great contributions. Zbigniew, you rock!
2282 Contributions from: Andrew Eikum, Christian Hesse, Colin
2283 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
2284 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
2285 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
2286 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
2287 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
2288 Jędrzejewski-Szmek, Сковорода Никита Андреевич
2292 * If /etc/vconsole.conf is non-existent or empty we will no
2293 longer load any console font or key map at boot by
2294 default. Instead the kernel defaults will be left
2295 intact. This is definitely the right thing to do, as no
2296 configuration should mean no configuration, and hard-coding
2297 font names that are different on all archs is probably a bad
2298 idea. Also, the kernel default key map and font should be
2299 good enough for most cases anyway, and mostly identical to
2300 the userspace fonts/key maps we previously overloaded them
2301 with. If distributions want to continue to default to a
2302 non-kernel font or key map they should ship a default
2303 /etc/vconsole.conf with the appropriate contents.
2305 Contributions from: Colin Walters, Daniel J Walsh, Dave
2306 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
2307 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
2311 * journalctl gained a new --cursor= switch to show entries
2312 starting from the specified location in the journal.
2314 * We now enforce a size limit on journal entry fields exported
2315 with "-o json" in journalctl. Fields larger than 4K will be
2316 assigned null. This can be turned off with --all.
2318 * An (optional) journal gateway daemon is now available as
2319 "systemd-journal-gatewayd.service". This service provides
2320 access to the journal via HTTP and JSON. This functionality
2321 will be used to implement live log synchronization in both
2322 pull and push modes, but has various other users too, such
2323 as easy log access for debugging of embedded devices. Right
2324 now it is already useful to retrieve the journal via HTTP:
2326 # systemctl start systemd-journal-gatewayd.service
2327 # wget http://localhost:19531/entries
2329 This will download the journal contents in a
2330 /var/log/messages compatible format. The same as JSON:
2332 # curl -H"Accept: application/json" http://localhost:19531/entries
2334 This service is also accessible via a web browser where a
2335 single static HTML5 app is served that uses the JSON logic
2336 to enable the user to do some basic browsing of the
2337 journal. This will be extended later on. Here's an example
2338 screenshot of this app in its current state:
2340 http://0pointer.de/public/journal-gatewayd
2342 Contributions from: Kay Sievers, Lennart Poettering, Robert
2343 Milasan, Tom Gundersen
2347 * The bash completion logic is now available for journalctl
2350 * We do not mount the "cpuset" controller anymore together with
2351 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
2352 started if no parameters are assigned to it. "cpuset" hence
2353 broke code that assumed it it could create "cpu" groups and
2356 * journalctl -f will now subscribe to terminal size changes,
2357 and line break accordingly.
2359 Contributions from: Dave Reisner, Kay Sievers, Lennart
2360 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
2364 * nspawn will now create a symlink /etc/localtime in the
2365 container environment, copying the host's timezone
2366 setting. Previously this has been done via a bind mount, but
2367 since symlinks cannot be bind mounted this has now been
2368 changed to create/update the appropriate symlink.
2370 * journalctl -n's line number argument is now optional, and
2371 will default to 10 if omitted.
2373 * journald will now log the maximum size the journal files may
2374 take up on disk. This is particularly useful if the default
2375 built-in logic of determining this parameter from the file
2376 system size is used. Use "systemctl status
2377 systemd-journald.service" to see this information.
2379 * The multi-seat X wrapper tool has been stripped down. As X
2380 is now capable of enumerating graphics devices via udev in a
2381 seat-aware way the wrapper is not strictly necessary
2382 anymore. A stripped down temporary stop-gap is still shipped
2383 until the upstream display managers have been updated to
2384 fully support the new X logic. Expect this wrapper to be
2385 removed entirely in one of the next releases.
2387 * HandleSleepKey= in logind.conf has been split up into
2388 HandleSuspendKey= and HandleHibernateKey=. The old setting
2389 is not available anymore. X11 and the kernel are
2390 distuingishing between these keys and we should too. This
2391 also means the inhibition lock for these keys has been split
2394 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
2395 Poettering, Lukas Nykryn, Václav Pavlín
2399 * Whenever a unit changes state we will now log this to the
2400 journal and show along the unit's own log output in
2403 * ConditionPathIsMountPoint= can now properly detect bind
2404 mount points too. (Previously, a bind mount of one file
2405 system to another place in the same file system could not be
2406 detected as mount, since they shared struct stat's st_dev
2409 * We will now mount the cgroup controllers cpu, cpuacct,
2410 cpuset and the controllers net_cls, net_prio together by
2413 * nspawn containers will now have a virtualized boot
2414 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
2415 over with a randomized ID at container initialization). This
2416 has the effect of making "journalctl -b" do the right thing
2419 * The JSON output journal serialization has been updated not
2420 to generate "endless" list objects anymore, but rather one
2421 JSON object per line. This is more in line how most JSON
2422 parsers expect JSON objects. The new output mode
2423 "json-pretty" has been added to provide similar output, but
2424 neatly aligned for readability by humans.
2426 * We dropped all explicit sync() invocations in the shutdown
2427 code. The kernel does this implicitly anyway in the kernel
2428 reboot() syscall. halt(8)'s -n option is now a compatibility
2431 * We now support virtualized reboot() in containers, as
2432 supported by newer kernels. We will fall back to exit() if
2433 CAP_SYS_REBOOT is not available to the container. Also,
2434 nspawn makes use of this now and will actually reboot the
2435 container if the containerized OS asks for that.
2437 * journalctl will only show local log output by default
2438 now. Use --merge (-m) to show remote log output, too.
2440 * libsystemd-journal gained the new sd_journal_get_usage()
2441 call to determine the current disk usage of all journal
2442 files. This is exposed in the new "journalctl --disk-usage"
2445 * journald gained a new configuration setting SplitMode= in
2446 journald.conf which may be used to control how user journals
2447 are split off. See journald.conf(5) for details.
2449 * A new condition type ConditionFileNotEmpty= has been added.
2451 * tmpfiles' "w" lines now support file globbing, to write
2452 multiple files at once.
2454 * We added Python bindings for the journal submission
2455 APIs. More Python APIs for a number of selected APIs will
2456 likely follow. Note that we intend to add native bindings
2457 only for the Python language, as we consider it common
2458 enough to deserve bindings shipped within systemd. There are
2459 various projects outside of systemd that provide bindings
2460 for languages such as PHP or Lua.
2462 * Many conditions will now resolve specifiers such as %i. In
2463 addition, PathChanged= and related directives of .path units
2464 now support specifiers as well.
2466 * There's now a new RPM macro definition for the system preset
2469 * journald will now warn if it ca not forward a message to the
2470 syslog daemon because its socket is full.
2472 * timedated will no longer write or process /etc/timezone,
2473 except on Debian. As we do not support late mounted /usr
2474 anymore /etc/localtime always being a symlink is now safe,
2475 and hence the information in /etc/timezone is not necessary
2478 * logind will now always reserve one VT for a text getty (VT6
2479 by default). Previously if more than 6 X sessions where
2480 started they took up all the VTs with auto-spawned gettys,
2481 so that no text gettys were available anymore.
2483 * udev will now automatically inform the btrfs kernel logic
2484 about btrfs RAID components showing up. This should make
2485 simple hotplug based btrfs RAID assembly work.
2487 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
2488 (but not for its children which will stay at the kernel
2489 default). This should allow setups with a lot more listening
2492 * systemd will now always pass the configured timezone to the
2493 kernel at boot. timedated will do the same when the timezone
2496 * logind's inhibition logic has been updated. By default,
2497 logind will now handle the lid switch, the power and sleep
2498 keys all the time, even in graphical sessions. If DEs want
2499 to handle these events on their own they should take the new
2500 handle-power-key, handle-sleep-key and handle-lid-switch
2501 inhibitors during their runtime. A simple way to achiveve
2502 that is to invoke the DE wrapped in an invocation of:
2504 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
2506 * Access to unit operations is now checked via SELinux taking
2507 the unit file label and client process label into account.
2509 * systemd will now notify the administrator in the journal
2510 when he over-mounts a non-empty directory.
2512 * There are new specifiers that are resolved in unit files,
2513 for the host name (%H), the machine ID (%m) and the boot ID
2516 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
2517 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
2518 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
2519 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
2520 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
2521 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
2522 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
2526 * Support for reading structured kernel messages from
2527 /dev/kmsg has now been added and is enabled by default.
2529 * Support for reading kernel messages from /proc/kmsg has now
2530 been removed. If you want kernel messages in the journal
2531 make sure to run a recent kernel (>= 3.5) that supports
2532 reading structured messages from /dev/kmsg (see
2533 above). /proc/kmsg is now exclusive property of classic
2534 syslog daemons again.
2536 * The libudev API gained the new
2537 udev_device_new_from_device_id() call.
2539 * The logic for file system namespace (ReadOnlyDirectory=,
2540 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
2541 require pivot_root() anymore. This means fewer temporary
2542 directories are created below /tmp for this feature.
2544 * nspawn containers will now see and receive all submounts
2545 made on the host OS below the root file system of the
2548 * Forward Secure Sealing is now supported for Journal files,
2549 which provide cryptographical sealing of journal files so
2550 that attackers cannot alter log history anymore without this
2551 being detectable. Lennart will soon post a blog story about
2552 this explaining it in more detail.
2554 * There are two new service settings RestartPreventExitStatus=
2555 and SuccessExitStatus= which allow configuration of exit
2556 status (exit code or signal) which will be excepted from the
2557 restart logic, resp. consider successful.
2559 * journalctl gained the new --verify switch that can be used
2560 to check the integrity of the structure of journal files and
2561 (if Forward Secure Sealing is enabled) the contents of
2564 * nspawn containers will now be run with /dev/stdin, /dev/fd/
2565 and similar symlinks pre-created. This makes running shells
2566 as container init process a lot more fun.
2568 * The fstab support can now handle PARTUUID= and PARTLABEL=
2571 * A new ConditionHost= condition has been added to match
2572 against the hostname (with globs) and machine ID. This is
2573 useful for clusters where a single OS image is used to
2574 provision a large number of hosts which shall run slightly
2575 different sets of services.
2577 * Services which hit the restart limit will now be placed in a
2580 Contributions from: Bertram Poettering, Dave Reisner, Huang
2581 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
2582 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
2586 * When running in --user mode systemd will now become a
2587 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
2588 tree a lot more organized.
2590 * A new PartOf= unit dependency type has been introduced that
2591 may be used to group services in a natural way.
2593 * "systemctl enable" may now be used to enable instances of
2596 * journalctl now prints error log levels in red, and
2597 warning/notice log levels in bright white. It also supports
2598 filtering by log level now.
2600 * cgtop gained a new -n switch (similar to top), to configure
2601 the maximum number of iterations to run for. It also gained
2602 -b, to run in batch mode (accepting no input).
2604 * The suffix ".service" may now be omitted on most systemctl
2605 command lines involving service unit names.
2607 * There's a new bus call in logind to lock all sessions, as
2608 well as a loginctl verb for it "lock-sessions".
2610 * libsystemd-logind.so gained a new call sd_journal_perror()
2611 that works similar to libc perror() but logs to the journal
2612 and encodes structured information about the error number.
2614 * /etc/crypttab entries now understand the new keyfile-size=
2617 * shutdown(8) now can send a (configurable) wall message when
2618 a shutdown is cancelled.
2620 * The mount propagation mode for the root file system will now
2621 default to "shared", which is useful to make containers work
2622 nicely out-of-the-box so that they receive new mounts from
2623 the host. This can be undone locally by running "mount
2624 --make-rprivate /" if needed.
2626 * The prefdm.service file has been removed. Distributions
2627 should maintain this unit downstream if they intend to keep
2628 it around. However, we recommend writing normal unit files
2629 for display managers instead.
2631 * Since systemd is a crucial part of the OS we will now
2632 default to a number of compiler switches that improve
2633 security (hardening) such as read-only relocations, stack
2634 protection, and suchlike.
2636 * The TimeoutSec= setting for services is now split into
2637 TimeoutStartSec= and TimeoutStopSec= to allow configuration
2638 of individual time outs for the start and the stop phase of
2641 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
2642 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
2643 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
2644 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
2645 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
2646 Gundersen, Zbigniew Jędrzejewski-Szmek
2650 * The journal and id128 C APIs are now fully documented as man
2653 * Extra safety checks have been added when transitioning from
2654 the initial RAM disk to the main system to avoid accidental
2657 * /etc/crypttab entries now understand the new keyfile-offset=
2660 * systemctl -t can now be used to filter by unit load state.
2662 * The journal C API gained the new sd_journal_wait() call to
2663 make writing synchronous journal clients easier.
2665 * journalctl gained the new -D switch to show journals from a
2668 * journalctl now displays a special marker between log
2669 messages of two different boots.
2671 * The journal is now explicitly flushed to /var via a service
2672 systemd-journal-flush.service, rather than implicitly simply
2673 by seeing /var/log/journal to be writable.
2675 * journalctl (and the journal C APIs) can now match for much
2676 more complex expressions, with alternatives and
2679 * When transitioning from the initial RAM disk to the main
2680 system we will now kill all processes in a killing spree to
2681 ensure no processes stay around by accident.
2683 * Three new specifiers may be used in unit files: %u, %h, %s
2684 resolve to the user name, user home directory resp. user
2685 shell. This is useful for running systemd user instances.
2687 * We now automatically rotate journal files if their data
2688 object hash table gets a fill level > 75%. We also size the
2689 hash table based on the configured maximum file size. This
2690 together should lower hash collisions drastically and thus
2691 speed things up a bit.
2693 * journalctl gained the new "--header" switch to introspect
2694 header data of journal files.
2696 * A new setting SystemCallFilters= has been added to services
2697 which may be used to apply blacklists or whitelists to
2698 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
2700 * nspawn gained a new --link-journal= switch (and quicker: -j)
2701 to link the container journal with the host. This makes it
2702 very easy to centralize log viewing on the host for all
2703 guests while still keeping the journal files separated.
2705 * Many bugfixes and optimizations
2707 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
2708 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
2709 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
2714 * Several tools now understand kernel command line arguments,
2715 which are only read when run in an initial RAM disk. They
2716 usually follow closely their normal counterparts, but are
2719 * There's a new tool to analyze the readahead files that are
2720 automatically generated at boot. Use:
2722 /usr/lib/systemd/systemd-readahead analyze /.readahead
2724 * We now provide an early debug shell on tty9 if this enabled. Use:
2726 systemctl enable debug-shell.service
2728 * All plymouth related units have been moved into the Plymouth
2729 package. Please make sure to upgrade your Plymouth version
2732 * systemd-tmpfiles now supports getting passed the basename of
2733 a configuration file only, in which case it will look for it
2734 in all appropriate directories automatically.
2736 * udevadm info now takes a /dev or /sys path as argument, and
2737 does the right thing. Example:
2739 udevadm info /dev/sda
2740 udevadm info /sys/class/block/sda
2742 * systemctl now prints a warning if a unit is stopped but a
2743 unit that might trigger it continues to run. Example: a
2744 service is stopped but the socket that activates it is left
2747 * "systemctl status" will now mention if the log output was
2748 shortened due to rotation since a service has been started.
2750 * The journal API now exposes functions to determine the
2751 "cutoff" times due to rotation.
2753 * journald now understands SIGUSR1 and SIGUSR2 for triggering
2754 immediately flushing of runtime logs to /var if possible,
2755 resp. for triggering immediate rotation of the journal
2758 * It is now considered an error if a service is attempted to
2759 be stopped that is not loaded.
2761 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
2763 * systemd-analyze now supports Python 3
2765 * tmpfiles now supports cleaning up directories via aging
2766 where the first level dirs are always kept around but
2767 directories beneath it automatically aged. This is enabled
2768 by prefixing the age field with '~'.
2770 * Seat objects now expose CanGraphical, CanTTY properties
2771 which is required to deal with very fast bootups where the
2772 display manager might be running before the graphics drivers
2773 completed initialization.
2775 * Seat objects now expose a State property.
2777 * We now include RPM macros for service enabling/disabling
2778 based on the preset logic. We recommend RPM based
2779 distributions to make use of these macros if possible. This
2780 makes it simpler to reuse RPM spec files across
2783 * We now make sure that the collected systemd unit name is
2784 always valid when services log to the journal via
2787 * There's a new man page kernel-command-line(7) detailing all
2788 command line options we understand.
2790 * The fstab generator may now be disabled at boot by passing
2791 fstab=0 on the kernel command line.
2793 * A new kernel command line option modules-load= is now understood
2794 to load a specific kernel module statically, early at boot.
2796 * Unit names specified on the systemctl command line are now
2797 automatically escaped as needed. Also, if file system or
2798 device paths are specified they are automatically turned
2799 into the appropriate mount or device unit names. Example:
2801 systemctl status /home
2802 systemctl status /dev/sda
2804 * The SysVConsole= configuration option has been removed from
2805 system.conf parsing.
2807 * The SysV search path is no longer exported on the D-Bus
2810 * The Names= option is been removed from unit file parsing.
2812 * There's a new man page bootup(7) detailing the boot process.
2814 * Every unit and every generator we ship with systemd now
2815 comes with full documentation. The self-explanatory boot is
2818 * A couple of services gained "systemd-" prefixes in their
2819 name if they wrap systemd code, rather than only external
2820 code. Among them fsck@.service which is now
2821 systemd-fsck@.service.
2823 * The HaveWatchdog property has been removed from the D-Bus
2826 * systemd.confirm_spawn= on the kernel command line should now
2829 * There's a new man page crypttab(5) which details all options
2830 we actually understand.
2832 * systemd-nspawn gained a new --capability= switch to pass
2833 additional capabilities to the container.
2835 * timedated will now read known NTP implementation unit names
2836 from /usr/lib/systemd/ntp-units.d/*.list,
2837 systemd-timedated-ntp.target has been removed.
2839 * journalctl gained a new switch "-b" that lists log data of
2840 the current boot only.
2842 * The notify socket is in the abstract namespace again, in
2843 order to support daemons which chroot() at start-up.
2845 * There is a new Storage= configuration option for journald
2846 which allows configuration of where log data should go. This
2847 also provides a way to disable journal logging entirely, so
2848 that data collected is only forwarded to the console, the
2849 kernel log buffer or another syslog implementation.
2851 * Many bugfixes and optimizations
2853 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
2854 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
2855 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
2856 Shawn Landden, Tom Gundersen
2860 * "systemctl help <unit>" now shows the man page if one is
2863 * Several new man pages have been added.
2865 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
2866 MaxLevelConsole= can now be specified in
2867 journald.conf. These options allow reducing the amount of
2868 data stored on disk or forwarded by the log level.
2870 * TimerSlackNSec= can now be specified in system.conf for
2871 PID1. This allows system-wide power savings.
2873 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
2874 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
2879 * logind is now capable of (optionally) handling power and
2880 sleep keys as well as the lid switch.
2882 * journalctl now understands the syntax "journalctl
2883 /usr/bin/avahi-daemon" to get all log output of a specific
2886 * CapabilityBoundingSet= in system.conf now also influences
2887 the capability bound set of usermode helpers of the kernel.
2889 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
2890 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
2891 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
2892 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
2896 * Note that we skipped 139 releases here in order to set the
2897 new version to something that is greater than both udev's
2898 and systemd's most recent version number.
2900 * udev: all udev sources are merged into the systemd source tree now.
2901 All future udev development will happen in the systemd tree. It
2902 is still fully supported to use the udev daemon and tools without
2903 systemd running, like in initramfs or other init systems. Building
2904 udev though, will require the *build* of the systemd tree, but
2905 udev can be properly *run* without systemd.
2907 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
2908 should be used to create dead device nodes as workarounds for broken
2911 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
2912 no longer supported. udev_monitor_new_from_netlink() needs to be
2913 used to subscribe to events.
2915 * udev: when udevd is started by systemd, processes which are left
2916 behind by forking them off of udev rules, are unconditionally cleaned
2917 up and killed now after the event handling has finished. Services or
2918 daemons must be started as systemd services. Services can be
2919 pulled-in by udev to get started, but they can no longer be directly
2920 forked by udev rules.
2922 * udev: the daemon binary is called systemd-udevd now and installed
2923 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
2924 to adapt to that, create symlink, or rename the binary after building
2927 * libudev no longer provides these symbols:
2928 udev_monitor_from_socket()
2929 udev_queue_get_failed_list_entry()
2930 udev_get_{dev,sys,run}_path()
2931 The versions number was bumped and symbol versioning introduced.
2933 * systemd-loginctl and systemd-journalctl have been renamed
2934 to loginctl and journalctl to match systemctl.
2936 * The config files: /etc/systemd/systemd-logind.conf and
2937 /etc/systemd/systemd-journald.conf have been renamed to
2938 logind.conf and journald.conf. Package updates should rename
2939 the files to the new names on upgrade.
2941 * For almost all files the license is now LGPL2.1+, changed
2942 from the previous GPL2.0+. Exceptions are some minor stuff
2943 of udev (which will be changed to LGPL2.1 eventually, too),
2944 and the MIT licensed sd-daemon.[ch] library that is suitable
2945 to be used as drop-in files.
2947 * systemd and logind now handle system sleep states, in
2948 particular suspending and hibernating.
2950 * logind now implements a sleep/shutdown/idle inhibiting logic
2951 suitable for a variety of uses. Soonishly Lennart will blog
2952 about this in more detail.
2954 * var-run.mount and var-lock.mount are no longer provided
2955 (which prevously bind mounted these directories to their new
2956 places). Distributions which have not converted these
2957 directories to symlinks should consider stealing these files
2958 from git history and add them downstream.
2960 * We introduced the Documentation= field for units and added
2961 this to all our shipped units. This is useful to make it
2962 easier to explore the boot and the purpose of the various
2965 * All smaller setup units (such as
2966 systemd-vconsole-setup.service) now detect properly if they
2967 are run in a container and are skipped when
2968 appropriate. This guarantees an entirely noise-free boot in
2969 Linux container environments such as systemd-nspawn.
2971 * A framework for implementing offline system updates is now
2972 integrated, for details see:
2973 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
2975 * A new service type Type=idle is available now which helps us
2976 avoiding ugly interleaving of getty output and boot status
2979 * There's now a system-wide CapabilityBoundingSet= option to
2980 globally reduce the set of capabilities for the
2981 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
2982 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
2983 even CAP_NET_ADMIN system-wide for secure systems.
2985 * There are now system-wide DefaultLimitXXX= options to
2986 globally change the defaults of the various resource limits
2987 for all units started by PID 1.
2989 * Harald Hoyer's systemd test suite has been integrated into
2990 systemd which allows easy testing of systemd builds in qemu
2991 and nspawn. (This is really awesome! Ask us for details!)
2993 * The fstab parser is now implemented as generator, not inside
2996 * systemctl will now warn you if .mount units generated from
2997 /etc/fstab are out of date due to changes in fstab that
2998 have not been read by systemd yet.
3000 * systemd is now suitable for usage in initrds. Dracut has
3001 already been updated to make use of this. With this in place
3002 initrds get a slight bit faster but primarily are much
3003 easier to introspect and debug since "systemctl status" in
3004 the host system can be used to introspect initrd services,
3005 and the journal from the initrd is kept around too.
3007 * systemd-delta has been added, a tool to explore differences
3008 between user/admin configuration and vendor defaults.
3010 * PrivateTmp= now affects both /tmp and /var/tmp.
3012 * Boot time status messages are now much prettier and feature
3013 proper english language. Booting up systemd has never been
3016 * Read-ahead pack files now include the inode number of all
3017 files to pre-cache. When the inode changes the pre-caching
3018 is not attempted. This should be nicer to deal with updated
3019 packages which might result in changes of read-ahead
3022 * We now temporaritly lower the kernel's read_ahead_kb variable
3023 when collecting read-ahead data to ensure the kernel's
3024 built-in read-ahead does not add noise to our measurements
3025 of necessary blocks to pre-cache.
3027 * There's now RequiresMountsFor= to add automatic dependencies
3028 for all mounts necessary for a specific file system path.
3030 * MountAuto= and SwapAuto= have been removed from
3031 system.conf. Mounting file systems at boot has to take place
3034 * nspawn now learned a new switch --uuid= to set the machine
3035 ID on the command line.
3037 * nspawn now learned the -b switch to automatically search
3040 * vt102 is now the default TERM for serial TTYs, upgraded from
3043 * systemd-logind now works on VT-less systems.
3045 * The build tree has been reorganized. The individual
3046 components now have directories of their own.
3048 * A new condition type ConditionPathIsReadWrite= is now available.
3050 * nspawn learned the new -C switch to create cgroups for the
3051 container in other hierarchies.
3053 * We now have support for hardware watchdogs, configurable in
3056 * The scheduled shutdown logic now has a public API.
3058 * We now mount /tmp as tmpfs by default, but this can be
3059 masked and /etc/fstab can override it.
3061 * Since udisks does not make use of /media anymore we are not
3062 mounting a tmpfs on it anymore.
3064 * journalctl gained a new --local switch to only interleave
3065 locally generated journal files.
3067 * We can now load the IMA policy at boot automatically.
3069 * The GTK tools have been split off into a systemd-ui.
3071 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
3072 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
3073 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
3074 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
3075 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
3076 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
3077 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
3078 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
3079 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
3084 * This is mostly a bugfix release
3086 * Support optional initialization of the machine ID from the
3087 KVM or container configured UUID.
3089 * Support immediate reboots with "systemctl reboot -ff"
3091 * Show /etc/os-release data in systemd-analyze output
3093 * Many bugfixes for the journal, including endianness fixes and
3094 ensuring that disk space enforcement works
3096 * sd-login.h is C++ comptaible again
3098 * Extend the /etc/os-release format on request of the Debian
3101 * We now refuse non-UTF8 strings used in various configuration
3102 and unit files. This is done to ensure we do not pass invalid
3103 data over D-Bus or expose it elsewhere.
3105 * Register Mimo USB Screens as suitable for automatic seat
3108 * Read SELinux client context from journal clients in a race
3111 * Reorder configuration file lookup order. /etc now always
3112 overrides /run in order to allow the administrator to always
3113 and unconditionally override vendor supplied or
3114 automatically generated data.
3116 * The various user visible bits of the journal now have man
3117 pages. We still lack man pages for the journal API calls
3120 * We now ship all man pages in HTML format again in the
3123 Contributions from: Dave Reisner, Dirk Eibach, Frederic
3124 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
3125 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
3130 * This is mostly a bugfix release
3132 * systems lacking /etc/os-release are no longer supported.
3134 * Various functionality updates to libsystemd-login.so
3136 * Track class of PAM logins to distuingish greeters from
3139 Contributions from: Kay Sievers, Lennart Poettering, Michael
3144 * This is an important bugfix release for v41.
3146 * Building man pages is now optional which should be useful
3147 for those building systemd from git but unwilling to install
3150 * Watchdog support for supervising services is now usable. In
3151 a future release support for hardware watchdogs
3152 (i.e. /dev/watchdog) will be added building on this.
3154 * Service start rate limiting is now configurable and can be
3155 turned off per service. When a start rate limit is hit a
3156 reboot can automatically be triggered.
3158 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
3160 Contributions from: Benjamin Franzke, Bill Nottingham,
3161 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
3162 Schmidt, Michał Górny, Piotr Drąg
3166 * The systemd binary is installed /usr/lib/systemd/systemd now;
3167 An existing /sbin/init symlink needs to be adapted with the
3170 * The code that loads kernel modules has been ported to invoke
3171 libkmod directly, instead of modprobe. This means we do not
3172 support systems with module-init-tools anymore.
3174 * Watchdog support is now already useful, but still not
3177 * A new kernel command line option systemd.setenv= is
3178 understood to set system wide environment variables
3179 dynamically at boot.
3181 * We now limit the set of capabilities of systemd-journald.
3183 * We now set SIGPIPE to ignore by default, since it only is
3184 useful in shell pipelines, and has little use in general
3185 code. This can be disabled with IgnoreSIPIPE=no in unit
3188 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
3189 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
3194 * This is mostly a bugfix release
3196 * We now expose the reason why a service failed in the
3197 "Result" D-Bus property.
3199 * Rudimentary service watchdog support (will be completed over
3200 the next few releases.)
3202 * When systemd forks off in order execute some service we will
3203 now immediately changes its argv[0] to reflect which process
3204 it will execute. This is useful to minimize the time window
3205 with a generic argv[0], which makes bootcharts more useful
3207 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
3208 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
3209 Mike Kazantsev, Ray Strode
3213 * This is mostly a test release, but incorporates many
3216 * New systemd-cgtop tool to show control groups by their
3219 * Linking against libacl for ACLs is optional again. If
3220 disabled, support tracking device access for active logins
3221 goes becomes unavailable, and so does access to the user
3222 journals by the respective users.
3224 * If a group "adm" exists, journal files are automatically
3225 owned by them, thus allow members of this group full access
3226 to the system journal as well as all user journals.
3228 * The journal now stores the SELinux context of the logging
3229 client for all entries.
3231 * Add C++ inclusion guards to all public headers
3233 * New output mode "cat" in the journal to print only text
3234 messages, without any meta data like date or time.
3236 * Include tiny X server wrapper as a temporary stop-gap to
3237 teach XOrg udev display enumeration. This is used by display
3238 managers such as gdm, and will go away as soon as XOrg
3239 learned native udev hotplugging for display devices.
3241 * Add new systemd-cat tool for executing arbitrary programs
3242 with STDERR/STDOUT connected to the journal. Can also act as
3243 BSD logger replacement, and does so by default.
3245 * Optionally store all locally generated coredumps in the
3246 journal along with meta data.
3248 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
3249 writing short strings to files (for usage for /sys), and for
3250 creating symlinks, character and block device nodes.
3252 * New unit file option ControlGroupPersistent= to make cgroups
3253 persistent, following the mechanisms outlined in
3254 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
3256 * Support multiple local RTCs in a sane way
3258 * No longer monopolize IO when replaying readahead data on
3259 rotating disks, since we might starve non-file-system IO to
3260 death, since fanotify() will not see accesses done by blkid,
3263 * Do not show kernel threads in systemd-cgls anymore, unless
3264 requested with new -k switch.
3266 Contributions from: Dan Horák, Kay Sievers, Lennart
3267 Poettering, Michal Schmidt
3271 * This is mostly a test release, but incorporates many
3274 * The git repository moved to:
3275 git://anongit.freedesktop.org/systemd/systemd
3276 ssh://git.freedesktop.org/git/systemd/systemd
3278 * First release with the journal
3279 http://0pointer.de/blog/projects/the-journal.html
3281 * The journal replaces both systemd-kmsg-syslogd and
3282 systemd-stdout-bridge.
3284 * New sd_pid_get_unit() API call in libsystemd-logind
3286 * Many systemadm clean-ups
3288 * Introduce remote-fs-pre.target which is ordered before all
3289 remote mounts and may be used to start services before all
3292 * Added Mageia support
3294 * Add bash completion for systemd-loginctl
3296 * Actively monitor PID file creation for daemons which exit in
3297 the parent process before having finished writing the PID
3298 file in the daemon process. Daemons which do this need to be
3299 fixed (i.e. PID file creation must have finished before the
3300 parent exits), but we now react a bit more gracefully to them.
3302 * Add colourful boot output, mimicking the well-known output
3303 of existing distributions.
3305 * New option PassCredentials= for socket units, for
3306 compatibility with a recent kernel ABI breakage.
3308 * /etc/rc.local is now hooked in via a generator binary, and
3309 thus will no longer act as synchronization point during
3312 * systemctl list-unit-files now supports --root=.
3314 * systemd-tmpfiles now understands two new commands: z, Z for
3315 relabelling files according to the SELinux database. This is
3316 useful to apply SELinux labels to specific files in /sys,
3319 * Output of SysV services is now forwarded to both the console
3320 and the journal by default, not only just the console.
3322 * New man pages for all APIs from libsystemd-login.
3324 * The build tree got reorganized and a the build system is a
3325 lot more modular allowing embedded setups to specifically
3326 select the components of systemd they are interested in.
3328 * Support for Linux systems lacking the kernel VT subsystem is
3331 * configure's --with-rootdir= got renamed to
3332 --with-rootprefix= to follow the naming used by udev and
3335 * Unless specified otherwise we will now install to /usr instead
3336 of /usr/local by default.
3338 * Processes with '@' in argv[0][0] are now excluded from the
3339 final shut-down killing spree, following the logic explained
3341 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
3343 * All processes remaining in a service cgroup when we enter
3344 the START or START_PRE states are now killed with
3345 SIGKILL. That means it is no longer possible to spawn
3346 background processes from ExecStart= lines (which was never
3347 supported anyway, and bad style).
3349 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
3350 reloading of units together.
3352 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
3353 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
3354 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
3355 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
3356 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek