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 already
33 supported --recursive switch for "systemctl list-units".
35 * A new RebootArgument= setting has been added for service
36 units, which may be used to specify a kernel reboot argument
37 to use 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 similarly 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
48 executing commands like hostnamectl with the -H switch.
49 systemd-analyze makes use of this to properly display
50 details when running non-locally.
52 * The bootchart tool can now show cgroup information in the
55 * The CFS CPU quota cgroup attribute is now exposed for
56 services. The new CPUQuota= switch has been added for this
57 which takes a percentage value. Setting this will have the
58 result that a service may never get more CPU time than the
59 specified percentage, even if the machine is otherwise idle.
61 * systemd-networkd learned IPIP and SIT tunnel support.
63 * LSB init scripts exposing a dependency on $network will now
64 get a dependency on network-online.target rather than simply
65 network.target. This should bring LSB handling closer to
66 what it was on SysV systems.
68 * A new fsck.repair= kernel option has been added to control
69 how fsck shall deal with unclean file systems at boot.
71 * The (.ini) configuration file parser will now silently
72 ignore sections whose name begins with "X-". This may be
73 used to maintain application-specific extension sections in unit
76 * machined gained a new API to query the IP addresses of
77 registered containers. "machinectl status" has been updated
78 to show these addresses in its output.
80 * A new call sd_uid_get_display() has been added to the
81 sd-login APIs for querying the "primary" session of a
82 user. The "primary" session of the user is elected from the
83 user's sessions and generally a graphical session is
84 preferred over a text one.
86 * A minimal systemd-resolved daemon has been added. It
87 currently simply acts as a companion to systemd-networkd and
88 manages resolv.conf based on per-interface DNS
89 configuration, possibly supplied via DHCP. In the long run
90 we hope to extend this into a local DNSSEC enabled DNS and
93 * The systemd-networkd-wait-online tool is now enabled by
94 default. It will delay network-online.target until a network
95 connection has been configured. The tool primarily integrates
96 with networkd, but will also make a best effort to make sense
97 of network configuration performed in some other way.
99 * Two new service options StartupCPUShares= and
100 StartupBlockIOWeight= have been added that work similarly to
101 CPUShares= and BlockIOWeight= however only apply during
102 system startup. This is useful to prioritize certain services
103 differently during bootup than during normal runtime.
105 * hostnamed has been changed to prefer the statically
106 configured hostname in /etc/hostname (unless set to
107 'localhost' or empty) over any dynamic one supplied by
108 dhcp. With this change the rules for picking the hostname
109 match more closely the rules of other configuration settings
110 where the local administrator's configuration in /etc always
111 overrides any other settings.
113 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
114 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
115 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
116 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
117 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
118 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
119 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
120 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
121 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
122 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
123 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
124 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
125 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
126 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
127 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
128 Lindskog, WaLyong Cho, Will Woods, Zbigniew
131 -- Beijing, 2014-05-28
135 * When restoring the screen brightness at boot, stay away from
136 the darkest setting or from the lowest 5% of the available
137 range, depending on which is the larger value of both. This
138 should effectively protect the user from rebooting into a
139 black screen, should the brightness have been set to minimum
142 * sd-login gained a new sd_machine_get_class() call to
143 determine the class ("vm" or "container") of a machine
144 registered with machined.
146 * sd-login gained new calls
147 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
148 to query the identity of the peer of a local AF_UNIX
149 connection. They operate similarly to their sd_pid_get_xyz()
152 * PID 1 will now maintain a system-wide system state engine
153 with the states "starting", "running", "degraded",
154 "maintenance", "stopping". These states are bound to system
155 startup, normal runtime, runtime with at least one failed
156 service, rescue/emergency mode and system shutdown. This
157 state is shown in the "systemctl status" output when no unit
158 name is passed. It is useful to determine system state, in
159 particularly when doing so for many systems or containers at
162 * A new command "list-machines" has been added to "systemctl"
163 that lists all local OS containers and shows their system
164 state (see above), if systemd runs inside of them.
166 * systemctl gained a new "-r" switch to recursively enumerate
167 units on all local containers, when used with the
168 "list-unit" command (which is the default one that is
169 executed when no parameters are specified).
171 * The GPT automatic partition discovery logic will now honour
172 two GPT partition flags: one may be set on a partition to
173 cause it to be mounted read-only, and the other may be set
174 on a partition to ignore it during automatic discovery.
176 * Two new GPT type UUIDs have been added for automatic root
177 partition discovery, for 32-bit and 64-bit ARM. This is not
178 particularly useful for discovering the root directory on
179 these architectures during bare-metal boots (since UEFI is
180 not common there), but still very useful to allow booting of
181 ARM disk images in nspawn with the -i option.
183 * MAC addresses of interfaces created with nspawn's
184 --network-interface= switch will now be generated from the
185 machine name, and thus be stable between multiple invocations
188 * logind will now automatically remove all IPC objects owned
189 by a user if she or he fully logs out. This makes sure that
190 users who are logged out cannot continue to consume IPC
191 resources. This covers SysV memory, semaphores and message
192 queues as well as POSIX shared memory and message
193 queues. Traditionally, SysV and POSIX IPC had no life-cycle
194 limits. With this functionality, that is corrected. This may
195 be turned off by using the RemoveIPC= switch of logind.conf.
197 * The systemd-machine-id-setup and tmpfiles tools gained a
198 --root= switch to operate on a specific root directory,
201 * journald can now forward logged messages to the TTYs of all
202 logged in users ("wall"). This is the default for all
203 emergency messages now.
205 * A new tool systemd-journal-remote has been added to stream
206 journal log messages across the network.
208 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
209 controller trees are mounted into it. Note that the
210 directories mounted beneath it are not read-only. This is a
211 security measure and is particularly useful because glibc
212 actually includes a search logic to pick any tmpfs it can
213 find to implement shm_open() if /dev/shm is not available
214 (which it might very well be in namespaced setups).
216 * machinectl gained a new "poweroff" command to cleanly power
217 down a local OS container.
219 * The PrivateDevices= unit file setting will now also drop the
220 CAP_MKNOD capability from the capability bound set, and
221 imply DevicePolicy=closed.
223 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
224 comprehensively on all long-running systemd services where
227 * systemd-udevd will now run in a disassociated mount
228 namespace. To mount directories from udev rules, make sure to
229 pull in mount units via SYSTEMD_WANTS properties.
231 * The kdbus support gained support for uploading policy into
232 the kernel. sd-bus gained support for creating "monitoring"
233 connections that can eavesdrop into all bus communication
234 for debugging purposes.
236 * Timestamps may now be specified in seconds since the UNIX
237 epoch Jan 1st, 1970 by specifying "@" followed by the value
240 * Native tcpwrap support in systemd has been removed. tcpwrap
241 is old code, not really maintained anymore and has serious
242 shortcomings, and better options such as firewalls
243 exist. For setups that require tcpwrap usage, please
244 consider invoking your socket-activated service via tcpd,
245 like on traditional inetd.
247 * A new system.conf configuration option
248 DefaultTimerAccuracySec= has been added that controls the
249 default AccuracySec= setting of .timer units.
251 * Timer units gained a new WakeSystem= switch. If enabled,
252 timers configured this way will cause the system to resume
253 from system suspend (if the system supports that, which most
256 * Timer units gained a new Persistent= switch. If enabled,
257 timers configured this way will save to disk when they have
258 been last triggered. This information is then used on next
259 reboot to possible execute overdue timer events, that
260 could not take place because the system was powered off.
261 This enables simple anacron-like behaviour for timer units.
263 * systemctl's "list-timers" will now also list the time a
264 timer unit was last triggered in addition to the next time
265 it will be triggered.
267 * systemd-networkd will now assign predictable IPv4LL
268 addresses to its local interfaces.
270 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
271 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
272 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
273 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
274 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
275 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
276 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
277 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
280 -- Berlin, 2014-03-25
284 * A new unit file setting RestrictAddressFamilies= has been
285 added to restrict which socket address families unit
286 processes gain access to. This takes address family names
287 like "AF_INET" or "AF_UNIX", and is useful to minimize the
288 attack surface of services via exotic protocol stacks. This
289 is built on seccomp system call filters.
291 * Two new unit file settings RuntimeDirectory= and
292 RuntimeDirectoryMode= have been added that may be used to
293 manage a per-daemon runtime directories below /run. This is
294 an alternative for setting up directory permissions with
295 tmpfiles snippets, and has the advantage that the runtime
296 directory's lifetime is bound to the daemon runtime and that
297 the daemon starts up with an empty directory each time. This
298 is particularly useful when writing services that drop
299 privileges using the User= or Group= setting.
301 * The DeviceAllow= unit setting now supports globbing for
302 matching against device group names.
304 * The systemd configuration file system.conf gained new
305 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
306 DefaultMemoryAccounting= to globally turn on/off accounting
307 for specific resources (cgroups) for all units. These
308 settings may still be overridden individually in each unit
311 * systemd-gpt-auto-generator is now able to discover /srv and
312 root partitions in addition to /home and swap partitions. It
313 also supports LUKS-encrypted partitions now. With this in
314 place, automatic discovery of partitions to mount following
315 the Discoverable Partitions Specification
316 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
317 is now a lot more complete. This allows booting without
318 /etc/fstab and without root= on the kernel command line on
319 systems prepared appropriately.
321 * systemd-nspawn gained a new --image= switch which allows
322 booting up disk images and Linux installations on any block
323 device that follow the Discoverable Partitions Specification
324 (see above). This means that installations made with
325 appropriately updated installers may now be started and
326 deployed using container managers, completely
327 unmodified. (We hope that libvirt-lxc will add support for
328 this feature soon, too.)
330 * systemd-nspawn gained a new --network-macvlan= setting to
331 set up a private macvlan interface for the
332 container. Similarly, systemd-networkd gained a new
333 Kind=macvlan setting in .netdev files.
335 * systemd-networkd now supports configuring local addresses
338 * A new tool systemd-network-wait-online has been added to
339 synchronously wait for network connectivity using
342 * The sd-bus.h bus API gained a new sd_bus_track object for
343 tracking the life-cycle of bus peers. Note that sd-bus.h is
344 still not a public API though (unless you specify
345 --enable-kdbus on the configure command line, which however
346 voids your warranty and you get no API stability guarantee).
348 * The $XDG_RUNTIME_DIR runtime directories for each user are
349 now individual tmpfs instances, which has the benefit of
350 introducing separate pools for each user, with individual
351 size limits, and thus making sure that unprivileged clients
352 can no longer negatively impact the system or other users by
353 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
354 RuntimeDirectorySize= has been introduced that allows
355 controlling the default size limit for all users. It
356 defaults to 10% of the available physical memory. This is no
357 replacement for quotas on tmpfs though (which the kernel
358 still does not support), as /dev/shm and /tmp are still
359 shared resources used by both the system and unprivileged
362 * logind will now automatically turn off automatic suspending
363 on laptop lid close when more than one display is
364 connected. This was previously expected to be implemented
365 individually in desktop environments (such as GNOME),
366 however has been added to logind now, in order to fix a
367 boot-time race where a desktop environment might not have
368 been started yet and thus not been able to take an inhibitor
369 lock at the time where logind already suspends the system
372 * logind will now wait at least 30s after each system
373 suspend/resume cycle, and 3min after system boot before
374 suspending the system due to a closed laptop lid. This
375 should give USB docking stations and similar enough time to
376 be probed and configured after system resume and boot in
377 order to then act as suspend blocker.
379 * systemd-run gained a new --property= setting which allows
380 initialization of resource control properties (and others)
381 for the created scope or service unit. Example: "systemd-run
382 --property=BlockIOWeight=10 updatedb" may be used to run
383 updatedb at a low block IO scheduling weight.
385 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
386 now also work in --scope mode.
388 * When systemd is compiled with kdbus support, basic support
389 for enforced policies is now in place. (Note that enabling
390 kdbus still voids your warranty and no API compatibility
393 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
394 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
395 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
396 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
397 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
398 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
399 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
400 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
401 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
402 Zbigniew Jędrzejewski-Szmek
404 -- Berlin, 2014-03-12
408 * systemd will now relabel /dev after loading the SMACK policy
409 according to SMACK rules.
411 * A new unit file option AppArmorProfile= has been added to
412 set the AppArmor profile for the processes of a unit.
414 * A new condition check ConditionArchitecture= has been added
415 to conditionalize units based on the system architecture, as
416 reported by uname()'s "machine" field.
418 * systemd-networkd now supports matching on the system
419 virtualization, architecture, kernel command line, host name
422 * logind is now a lot more aggressive when suspending the
423 machine due to a closed laptop lid. Instead of acting only
424 on the lid close action, it will continuously watch the lid
425 status and act on it. This is useful for laptops where the
426 power button is on the outside of the chassis so that it can
427 be reached without opening the lid (such as the Lenovo
428 Yoga). On those machines, logind will now immediately
429 re-suspend the machine if the power button has been
430 accidentally pressed while the laptop was suspended and in a
433 * logind will now watch SW_DOCK switches and inhibit reaction
434 to the lid switch if it is pressed. This means that logind
435 will not suspend the machine anymore if the lid is closed
436 and the system is docked, if the laptop supports SW_DOCK
437 notifications via the input layer. Note that ACPI docking
438 stations do not generate this currently. Also note that this
439 logic is usually not fully sufficient and Desktop
440 Environments should take a lid switch inhibitor lock when an
441 external display is connected, as systemd will not watch
444 * nspawn will now make use of the devices cgroup controller by
445 default, and only permit creation of and access to the usual
446 API device nodes like /dev/null or /dev/random, as well as
447 access to (but not creation of) the pty devices.
449 * We will now ship a default .network file for
450 systemd-networkd that automatically configures DHCP for
451 network interfaces created by nspawn's --network-veth or
452 --network-bridge= switches.
454 * systemd will now understand the usual M, K, G, T suffixes
455 according to SI conventions (i.e. to the base 1000) when
456 referring to throughput and hardware metrics. It will stay
457 with IEC conventions (i.e. to the base 1024) for software
458 metrics, according to what is customary according to
459 Wikipedia. We explicitly document which base applies for
460 each configuration option.
462 * The DeviceAllow= setting in unit files now supports a syntax
463 to whitelist an entire group of devices node majors at once,
464 based on the /proc/devices listing. For example, with the
465 string "char-pts", it is now possible to whitelist all
466 current and future pseudo-TTYs at once.
468 * sd-event learned a new "post" event source. Event sources of
469 this type are triggered by the dispatching of any event
470 source of a type that is not "post". This is useful for
471 implementing clean-up and check event sources that are
472 triggered by other work being done in the program.
474 * systemd-networkd is no longer statically enabled, but uses
475 the usual [Install] sections so that it can be
476 enabled/disabled using systemctl. It still is enabled by
479 * When creating a veth interface pair with systemd-nspawn, the
480 host side will now be prefixed with "vb-" if
481 --network-bridge= is used, and with "ve-" if --network-veth
482 is used. This way, it is easy to distinguish these cases on
483 the host, for example to apply different configuration to
484 them with systemd-networkd.
486 * The compatibility libraries for libsystemd-journal.so,
487 libsystem-id128.so, libsystemd-login.so and
488 libsystemd-daemon.so do not make use of IFUNC
489 anymore. Instead, we now build libsystemd.so multiple times
490 under these alternative names. This means that the footprint
491 is drastically increased, but given that these are
492 transitional compatibility libraries, this should not matter
493 much. This change has been made necessary to support the ARM
494 platform for these compatibility libraries, as the ARM
495 toolchain is not really at the same level as the toolchain
496 for other architectures like x86 and does not support
497 IFUNC. Please make sure to use --enable-compat-libs only
498 during a transitional period!
500 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
501 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
502 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
503 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
504 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
505 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
506 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
507 Zbigniew Jędrzejewski-Szmek
509 -- Berlin, 2014-02-24
513 * A new component "systemd-networkd" has been added that can
514 be used to configure local network interfaces statically or
515 via DHCP. It is capable of bringing up bridges, VLANs, and
516 bonding. Currently, no hook-ups for interactive network
517 configuration are provided. Use this for your initrd,
518 container, embedded, or server setup if you need a simple,
519 yet powerful, network configuration solution. This
520 configuration subsystem is quite nifty, as it allows wildcard
521 hotplug matching in interfaces. For example, with a single
522 configuration snippet, you can configure that all Ethernet
523 interfaces showing up are automatically added to a bridge,
524 or similar. It supports link-sensing and more.
526 * A new tool "systemd-socket-proxyd" has been added which can
527 act as a bidirectional proxy for TCP sockets. This is
528 useful for adding socket activation support to services that
529 do not actually support socket activation, including virtual
530 machines and the like.
532 * Add a new tool to save/restore rfkill state on
535 * Save/restore state of keyboard backlights in addition to
536 display backlights on shutdown/boot.
538 * udev learned a new SECLABEL{} construct to label device
539 nodes with a specific security label when they appear. For
540 now, only SECLABEL{selinux} is supported, but the syntax is
541 prepared for additional security frameworks.
543 * udev gained a new scheme to configure link-level attributes
544 from files in /etc/systemd/network/*.link. These files can
545 match against MAC address, device path, driver name and type,
546 and will apply attributes like the naming policy, link speed,
547 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
548 address assignment policy (randomized, ...).
550 * The configuration of network interface naming rules for
551 "permanent interface names" has changed: a new NamePolicy=
552 setting in the [Link] section of .link files determines the
553 priority of possible naming schemes (onboard, slot, mac,
554 path). The default value of this setting is determined by
555 /usr/lib/net/links/99-default.link. Old
556 80-net-name-slot.rules udev configuration file has been
557 removed, so local configuration overriding this file should
558 be adapated to override 99-default.link instead.
560 * When the User= switch is used in a unit file, also
561 initialize $SHELL= based on the user database entry.
563 * systemd no longer depends on libdbus. All communication is
564 now done with sd-bus, systemd's low-level bus library
567 * kdbus support has been added to PID 1 itself. When kdbus is
568 enabled, this causes PID 1 to set up the system bus and
569 enable support for a new ".busname" unit type that
570 encapsulates bus name activation on kdbus. It works a little
571 bit like ".socket" units, except for bus names. A new
572 generator has been added that converts classic dbus1 service
573 activation files automatically into native systemd .busname
576 * sd-bus: add a light-weight vtable implementation that allows
577 defining objects on the bus with a simple static const
578 vtable array of its methods, signals and properties.
580 * systemd will not generate or install static dbus
581 introspection data anymore to /usr/share/dbus-1/interfaces,
582 as the precise format of these files is unclear, and
583 nothing makes use of it.
585 * A proxy daemon is now provided to proxy clients connecting
586 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
587 compatibility with classic D-Bus.
589 * A bus driver implementation has been added that supports the
590 classic D-Bus bus driver calls on kdbus, also for
591 compatibility purposes.
593 * A new API "sd-event.h" has been added that implements a
594 minimal event loop API built around epoll. It provides a
595 couple of features that direct epoll usage is lacking:
596 prioritization of events, scales to large numbers of timer
597 events, per-event timer slack (accuracy), system-wide
598 coalescing of timer events, exit handlers, watchdog
599 supervision support using systemd's sd_notify() API, child
602 * A new API "sd-rntl.h" has been added that provides an API
603 around the route netlink interface of the kernel, similar in
606 * A new API "sd-dhcp-client.h" has been added that provides a
607 small DHCPv4 client-side implementation. This is used by
610 * There is a new kernel command line option
611 "systemd.restore_state=0|1". When set to "0", none of the
612 systemd tools will restore saved runtime state to hardware
613 devices. More specifically, the rfkill and backlight states
616 * The FsckPassNo= compatibility option in mount/service units
617 has been removed. The fstab generator will now add the
618 necessary dependencies automatically, and does not require
619 PID1's support for that anymore.
621 * journalctl gained a new switch, --list-boots, that lists
622 recent boots with their times and boot IDs.
624 * The various tools like systemctl, loginctl, timedatectl,
625 busctl, systemd-run, ... have gained a new switch "-M" to
626 connect to a specific, local OS container (as direct
627 connection, without requiring SSH). This works on any
628 container that is registered with machined, such as those
629 created by libvirt-lxc or nspawn.
631 * systemd-run and systemd-analyze also gained support for "-H"
632 to connect to remote hosts via SSH. This is particularly
633 useful for systemd-run because it enables queuing of jobs
636 * machinectl gained a new command "login" to open a getty
637 login in any local container. This works with any container
638 that is registered with machined (such as those created by
639 libvirt-lxc or nspawn), and which runs systemd inside.
641 * machinectl gained a new "reboot" command that may be used to
642 trigger a reboot on a specific container that is registered
643 with machined. This works on any container that runs an init
646 * systemctl gained a new "list-timers" command to print a nice
647 listing of installed timer units with the times they elapse
650 * Alternative reboot() parameters may now be specified on the
651 "systemctl reboot" command line and are passed to the
652 reboot() system call.
654 * systemctl gained a new --job-mode= switch to configure the
655 mode to queue a job with. This is a more generic version of
656 --fail, --irreversible, and --ignore-dependencies, which are
657 still available but not advertised anymore.
659 * /etc/systemd/system.conf gained new settings to configure
660 various default timeouts of units, as well as the default
661 start limit interval and burst. These may still be overridden
664 * PID1 will now export on the bus profile data of the security
665 policy upload process (such as the SELinux policy upload to
668 * journald: when forwarding logs to the console, include
669 timestamps (following the setting in
670 /sys/module/printk/parameters/time).
672 * OnCalendar= in timer units now understands the special
673 strings "yearly" and "annually". (Both are equivalent)
675 * The accuracy of timer units is now configurable with the new
676 AccuracySec= setting. It defaults to 1min.
678 * A new dependency type JoinsNamespaceOf= has been added that
679 allows running two services within the same /tmp and network
680 namespace, if PrivateNetwork= or PrivateTmp= are used.
682 * A new command "cat" has been added to systemctl. It outputs
683 the original unit file of a unit, and concatenates the
684 contents of additional "drop-in" unit file snippets, so that
685 the full configuration is shown.
687 * systemctl now supports globbing on the various "list-xyz"
688 commands, like "list-units" or "list-sockets", as well as on
689 those commands which take multiple unit names.
691 * journalctl's --unit= switch gained support for globbing.
693 * All systemd daemons now make use of the watchdog logic so
694 that systemd automatically notices when they hang.
696 * If the $container_ttys environment variable is set,
697 getty-generator will automatically spawn a getty for each
698 listed tty. This is useful for container managers to request
699 login gettys to be spawned on as many ttys as needed.
701 * %h, %s, %U specifier support is not available anymore when
702 used in unit files for PID 1. This is because NSS calls are
703 not safe from PID 1. They stay available for --user
704 instances of systemd, and as special case for the root user.
706 * loginctl gained a new "--no-legend" switch to turn off output
709 * The "sd-login.h" API gained three new calls:
710 sd_session_is_remote(), sd_session_get_remote_user(),
711 sd_session_get_remote_host() to query information about
714 * The udev hardware database now also carries vendor/product
715 information of SDIO devices.
717 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
718 determine whether watchdog notifications are requested by
721 * Socket-activated per-connection services now include a
722 short description of the connection parameters in the
725 * tmpfiles gained a new "--boot" option. When this is not used,
726 only lines where the command character is not suffixed with
727 "!" are executed. When this option is specified, those
728 options are executed too. This partitions tmpfiles
729 directives into those that can be safely executed at any
730 time, and those which should be run only at boot (for
731 example, a line that creates /run/nologin).
733 * A new API "sd-resolve.h" has been added which provides a simple
734 asynchronous wrapper around glibc NSS host name resolution
735 calls, such as getaddrinfo(). In contrast to glibc's
736 getaddrinfo_a(), it does not use signals. In contrast to most
737 other asynchronous name resolution libraries, this one does
738 not reimplement DNS, but reuses NSS, so that alternate
739 host name resolution systems continue to work, such as mDNS,
740 LDAP, etc. This API is based on libasyncns, but it has been
741 cleaned up for inclusion in systemd.
743 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
744 "sd-daemon.h" are no longer found in individual libraries
745 libsystemd-journal.so, libsystemd-login.so,
746 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
747 merged them into a single library, libsystemd.so, which
748 provides all symbols. The reason for this is cyclic
749 dependencies, as these libraries tend to use each other's
750 symbols. So far, we have managed to workaround that by linking
751 a copy of a good part of our code into each of these
752 libraries again and again, which, however, makes certain
753 things hard to do, like sharing static variables. Also, it
754 substantially increases footprint. With this change, there
755 is only one library for the basic APIs systemd
756 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
757 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
758 library as well, however are subject to the --enable-kdbus
759 switch (see below). Note that "sd-dhcp-client.h" is not part
760 of this library (this is because it only consumes, never
761 provides, services of/to other APIs). To make the transition
762 easy from the separate libraries to the unified one, we
763 provide the --enable-compat-libs compile-time switch which
764 will generate stub libraries that are compatible with the
765 old ones but redirect all calls to the new one.
767 * All of the kdbus logic and the new APIs "sd-bus.h",
768 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
769 and "sd-utf8.h" are compile-time optional via the
770 "--enable-kdbus" switch, and they are not compiled in by
771 default. To make use of kdbus, you have to explicitly enable
772 the switch. Note however, that neither the kernel nor the
773 userspace API for all of this is considered stable yet. We
774 want to maintain the freedom to still change the APIs for
775 now. By specifying this build-time switch, you acknowledge
776 that you are aware of the instability of the current
779 * Also, note that while kdbus is pretty much complete,
780 it lacks one thing: proper policy support. This means you
781 can build a fully working system with all features; however,
782 it will be highly insecure. Policy support will be added in
783 one of the next releases, at the same time that we will
784 declare the APIs stable.
786 * When the kernel command-line argument "kdbus" is specified,
787 systemd will automatically load the kdbus.ko kernel module. At
788 this stage of development, it is only useful for testing kdbus
789 and should not be used in production. Note: if "--enable-kdbus"
790 is specified, and the kdbus.ko kernel module is available, and
791 "kdbus" is added to the kernel command line, the entire system
792 runs with kdbus instead of dbus-daemon, with the above mentioned
793 problem of missing the system policy enforcement. Also a future
794 version of kdbus.ko or a newer systemd will not be compatible with
795 each other, and will unlikely be able to boot the machine if only
796 one of them is updated.
798 * systemctl gained a new "import-environment" command which
799 uploads the caller's environment (or parts thereof) into the
800 service manager so that it is inherited by services started
801 by the manager. This is useful to upload variables like
802 $DISPLAY into the user service manager.
804 * A new PrivateDevices= switch has been added to service units
805 which allows running a service with a namespaced /dev
806 directory that does not contain any device nodes for
807 physical devices. More specifically, it only includes devices
808 such as /dev/null, /dev/urandom, and /dev/zero which are API
811 * logind has been extended to support behaviour like VT
812 switching on seats that do not support a VT. This makes
813 multi-session available on seats that are not the first seat
814 (seat0), and on systems where kernel support for VTs has
815 been disabled at compile-time.
817 * If a process holds a delay lock for system sleep or shutdown
818 and fails to release it in time, we will now log its
819 identity. This makes it easier to identify processes that
820 cause slow suspends or power-offs.
822 * When parsing /etc/crypttab, support for a new key-slot=
823 option as supported by Debian is added. It allows indicating
824 which LUKS slot to use on disk, speeding up key loading.
826 * The sd_journald_sendv() API call has been checked and
827 officially declared to be async-signal-safe so that it may
828 be invoked from signal handlers for logging purposes.
830 * Boot-time status output is now enabled automatically after a
831 short timeout if boot does not progress, in order to give
832 the user an indication what she or he is waiting for.
834 * The boot-time output has been improved to show how much time
835 remains until jobs expire.
837 * The KillMode= switch in service units gained a new possible
838 value "mixed". If set, and the unit is shut down, then the
839 initial SIGTERM signal is sent only to the main daemon
840 process, while the following SIGKILL signal is sent to
841 all remaining processes of the service.
843 * When a scope unit is registered, a new property "Controller"
844 may be set. If set to a valid bus name, systemd will send a
845 RequestStop() signal to this name when it would like to shut
846 down the scope. This may be used to hook manager logic into
847 the shutdown logic of scope units. Also, scope units may now
848 be put in a special "abandoned" state, in which case the
849 manager process which created them takes no further
850 responsibilities for it.
852 * When reading unit files, systemd will now verify
853 the access mode of these files, and warn about certain
854 suspicious combinations. This has been added to make it
855 easier to track down packaging bugs where unit files are
856 marked executable or world-writable.
858 * systemd-nspawn gained a new "--setenv=" switch to set
859 container-wide environment variables. The similar option in
860 systemd-activate was renamed from "--environment=" to
861 "--setenv=" for consistency.
863 * systemd-nspawn has been updated to create a new kdbus domain
864 for each container that is invoked, thus allowing each
865 container to have its own set of system and user buses,
866 independent of the host.
868 * systemd-nspawn gained a new --drop-capability= switch to run
869 the container with less capabilities than the default. Both
870 --drop-capability= and --capability= now take the special
871 string "all" for dropping or keeping all capabilities.
873 * systemd-nspawn gained new switches for executing containers
874 with specific SELinux labels set.
876 * systemd-nspawn gained a new --quiet switch to not generate
877 any additional output but the container's own console
880 * systemd-nspawn gained a new --share-system switch to run a
881 container without PID namespacing enabled.
883 * systemd-nspawn gained a new --register= switch to control
884 whether the container is registered with systemd-machined or
885 not. This is useful for containers that do not run full
886 OS images, but only specific apps.
888 * systemd-nspawn gained a new --keep-unit which may be used
889 when invoked as the only program from a service unit, and
890 results in registration of the unit service itself in
891 systemd-machined, instead of a newly opened scope unit.
893 * systemd-nspawn gained a new --network-interface= switch for
894 moving arbitrary interfaces to the container. The new
895 --network-veth switch creates a virtual Ethernet connection
896 between host and container. The new --network-bridge=
897 switch then allows assigning the host side of this virtual
898 Ethernet connection to a bridge device.
900 * systemd-nspawn gained a new --personality= switch for
901 setting the kernel personality for the container. This is
902 useful when running a 32-bit container on a 64-bit host. A
903 similar option Personality= is now also available for service
906 * logind will now also track a "Desktop" identifier for each
907 session which encodes the desktop environment of it. This is
908 useful for desktop environments that want to identify
909 multiple running sessions of itself easily.
911 * A new SELinuxContext= setting for service units has been
912 added that allows setting a specific SELinux execution
913 context for a service.
915 * Most systemd client tools will now honour $SYSTEMD_LESS for
916 settings of the "less" pager. By default, these tools will
917 override $LESS to allow certain operations to work, such as
918 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
919 influence this logic.
921 * systemd's "seccomp" hook-up has been changed to make use of
922 the libseccomp library instead of using its own
923 implementation. This has benefits for portability among
926 * For usage together with SystemCallFilter=, a new
927 SystemCallErrorNumber= setting has been introduced that
928 allows configuration of a system error number to be returned
929 on filtered system calls, instead of immediately killing the
930 process. Also, SystemCallArchitectures= has been added to
931 limit access to system calls of a particular architecture
932 (in order to turn off support for unused secondary
933 architectures). There is also a global
934 SystemCallArchitectures= setting in system.conf now to turn
935 off support for non-native system calls system-wide.
937 * systemd requires a kernel with a working name_to_handle_at(),
938 please see the kernel config requirements in the README file.
940 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
941 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
942 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
943 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
944 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
945 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
946 Elia Pinto, Florian Weimer, George McCollister, Goffredo
947 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
948 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
949 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
950 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
951 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
952 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
953 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
954 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
955 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
956 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
957 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
958 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
959 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
960 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
961 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
962 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
963 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
965 -- Berlin, 2014-02-20
969 * logind has gained support for facilitating privileged input
970 and drm device access for unprivileged clients. This work is
971 useful to allow Wayland display servers (and similar
972 programs, such as kmscon) to run under the user's ID and
973 access input and drm devices which are normally
974 protected. When this is used (and the kernel is new enough)
975 logind will "mute" IO on the file descriptors passed to
976 Wayland as long as it is in the background and "unmute" it
977 if it returns into the foreground. This allows secure
978 session switching without allowing background sessions to
979 eavesdrop on input and display data. This also introduces
980 session switching support if VT support is turned off in the
981 kernel, and on seats that are not seat0.
983 * A new kernel command line option luks.options= is understood
984 now which allows specifiying LUKS options for usage for LUKS
985 encrypted partitions specified with luks.uuid=.
987 * tmpfiles.d(5) snippets may now use specifier expansion in
988 path names. More specifically %m, %b, %H, %v, are now
989 replaced by the local machine id, boot id, hostname, and
990 kernel version number.
992 * A new tmpfiles.d(5) command "m" has been introduced which
993 may be used to change the owner/group/access mode of a file
994 or directory if it exists, but do nothing if it does not.
996 * This release removes high-level support for the
997 MemorySoftLimit= cgroup setting. The underlying kernel
998 cgroup attribute memory.soft_limit= is currently badly
999 designed and likely to be removed from the kernel API in its
1000 current form, hence we should not expose it for now.
1002 * The memory.use_hierarchy cgroup attribute is now enabled for
1003 all cgroups systemd creates in the memory cgroup
1004 hierarchy. This option is likely to be come the built-in
1005 default in the kernel anyway, and the non-hierarchial mode
1006 never made much sense in the intrinsically hierarchial
1009 * A new field _SYSTEMD_SLICE= is logged along with all journal
1010 messages containing the slice a message was generated
1011 from. This is useful to allow easy per-customer filtering of
1012 logs among other things.
1014 * systemd-journald will no longer adjust the group of journal
1015 files it creates to the "systemd-journal" group. Instead we
1016 rely on the journal directory to be owned by the
1017 "systemd-journal" group, and its setgid bit set, so that the
1018 kernel file system layer will automatically enforce that
1019 journal files inherit this group assignment. The reason for
1020 this change is that we cannot allow NSS look-ups from
1021 journald which would be necessary to resolve
1022 "systemd-journal" to a numeric GID, because this might
1023 create deadlocks if NSS involves synchronous queries to
1024 other daemons (such as nscd, or sssd) which in turn are
1025 logging clients of journald and might block on it, which
1026 would then dead lock. A tmpfiles.d(5) snippet included in
1027 systemd will make sure the setgid bit and group are
1028 properly set on the journal directory if it exists on every
1029 boot. However, we recommend adjusting it manually after
1030 upgrades too (or from RPM scriptlets), so that the change is
1031 not delayed until next reboot.
1033 * Backlight and random seed files in /var/lib/ have moved into
1034 the /var/lib/systemd/ directory, in order to centralize all
1035 systemd generated files in one directory.
1037 * Boot time performance measurements (as displayed by
1038 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
1039 performance information if that's available to determine how
1040 much time BIOS and boot loader initialization required. With
1041 a sufficiently new BIOS you hence no longer need to boot
1042 with Gummiboot to get access to such information.
1044 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
1045 Cristian Rodríguez, Dave Reisner, David Herrmann, David
1046 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
1047 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
1048 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
1049 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
1050 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1052 -- Berlin, 2013-10-02
1056 * The Restart= option for services now understands a new
1057 on-watchdog setting, which will restart the service
1058 automatically if the service stops sending out watchdog keep
1059 alive messages (as configured with WatchdogSec=).
1061 * The getty generator (which is responsible for bringing up a
1062 getty on configured serial consoles) will no longer only
1063 start a getty on the primary kernel console but on all
1064 others, too. This makes the order in which console= is
1065 specified on the kernel command line less important.
1067 * libsystemd-logind gained a new sd_session_get_vt() call to
1068 retrieve the VT number of a session.
1070 * If the option "tries=0" is set for an entry of /etc/crypttab
1071 its passphrase is queried indefinitely instead of any
1072 maximum number of tries.
1074 * If a service with a configure PID file terminates its PID
1075 file will now be removed automatically if it still exists
1076 afterwards. This should put an end to stale PID files.
1078 * systemd-run will now also take relative binary path names
1079 for execution and no longer insists on absolute paths.
1081 * InaccessibleDirectories= and ReadOnlyDirectories= now take
1082 paths that are optionally prefixed with "-" to indicate that
1083 it should not be considered a failure if they do not exist.
1085 * journalctl -o (and similar commands) now understands a new
1086 output mode "short-precise", it is similar to "short" but
1087 shows timestamps with usec accuracy.
1089 * The option "discard" (as known from Debian) is now
1090 synonymous to "allow-discards" in /etc/crypttab. In fact,
1091 "discard" is preferred now (since it is easier to remember
1094 * Some licensing clean-ups were made, so that more code is now
1095 LGPL-2.1 licensed than before.
1097 * A minimal tool to save/restore the display backlight
1098 brightness across reboots has been added. It will store the
1099 backlight setting as late as possible at shutdown, and
1100 restore it as early as possible during reboot.
1102 * A logic to automatically discover and enable home and swap
1103 partitions on GPT disks has been added. With this in place
1104 /etc/fstab becomes optional for many setups as systemd can
1105 discover certain partitions located on the root disk
1106 automatically. Home partitions are recognized under their
1107 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
1108 partitions are recognized under their GPT type ID
1109 0657fd6da4ab43c484e50933c84b4f4f.
1111 * systemd will no longer pass any environment from the kernel
1112 or initrd to system services. If you want to set an
1113 environment for all services, do so via the kernel command
1114 line systemd.setenv= assignment.
1116 * The systemd-sysctl tool no longer natively reads the file
1117 /etc/sysctl.conf. If desired, the file should be symlinked
1118 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
1119 legacy support by a symlink rather than built-in code, it
1120 also makes the otherwise hidden order of application of the
1121 different files visible. (Note that this partly reverts to a
1122 pre-198 application order of sysctl knobs!)
1124 * The "systemctl set-log-level" and "systemctl dump" commands
1125 have been moved to systemd-analyze.
1127 * systemd-run learned the new --remain-after-exit switch,
1128 which causes the scope unit not to be cleaned up
1129 automatically after the process terminated.
1131 * tmpfiles learned a new --exclude-prefix= switch to exclude
1132 certain paths from operation.
1134 * journald will now automatically flush all messages to disk
1135 as soon as a message of the log priorities CRIT, ALERT or
1138 Contributions from: Andrew Cook, Brandon Philips, Christian
1139 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
1140 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
1141 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
1142 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
1143 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
1144 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
1145 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
1146 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
1147 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
1148 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
1149 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
1150 William Giokas, Zbigniew Jędrzejewski-Szmek
1152 -- Berlin, 2013-09-13
1156 * The documentation has been updated to cover the various new
1157 concepts introduced with 205.
1159 * Unit files now understand the new %v specifier which
1160 resolves to the kernel version string as returned by "uname
1163 * systemctl now supports filtering the unit list output by
1164 load state, active state and sub state, using the new
1167 * "systemctl status" will now show the results of the
1168 condition checks (like ConditionPathExists= and similar) of
1169 the last start attempts of the unit. They are also logged to
1172 * "journalctl -b" may now be used to look for boot output of a
1173 specific boot. Try "journalctl -b -1" for the previous boot,
1174 but the syntax is substantially more powerful.
1176 * "journalctl --show-cursor" has been added which prints the
1177 cursor string the last shown log line. This may then be used
1178 with the new "journalctl --after-cursor=" switch to continue
1179 browsing logs from that point on.
1181 * "journalctl --force" may now be used to force regeneration
1184 * Creation of "dead" device nodes has been moved from udev
1185 into kmod and tmpfiles. Previously, udev would read the kmod
1186 databases to pre-generate dead device nodes based on meta
1187 information contained in kernel modules, so that these would
1188 be auto-loaded on access rather then at boot. As this
1189 does not really have much to do with the exposing actual
1190 kernel devices to userspace this has always been slightly
1191 alien in the udev codebase. Following the new scheme kmod
1192 will now generate a runtime snippet for tmpfiles from the
1193 module meta information and it now is tmpfiles' job to the
1194 create the nodes. This also allows overriding access and
1195 other parameters for the nodes using the usual tmpfiles
1196 facilities. As side effect this allows us to remove the
1197 CAP_SYS_MKNOD capability bit from udevd entirely.
1199 * logind's device ACLs may now be applied to these "dead"
1200 devices nodes too, thus finally allowing managed access to
1201 devices such as /dev/snd/sequencer whithout loading the
1202 backing module right-away.
1204 * A new RPM macro has been added that may be used to apply
1205 tmpfiles configuration during package installation.
1207 * systemd-detect-virt and ConditionVirtualization= now can
1208 detect User-Mode-Linux machines (UML).
1210 * journald will now implicitly log the effective capabilities
1211 set of processes in the message metadata.
1213 * systemd-cryptsetup has gained support for TrueCrypt volumes.
1215 * The initrd interface has been simplified (more specifically,
1216 support for passing performance data via environment
1217 variables and fsck results via files in /run has been
1218 removed). These features were non-essential, and are
1219 nowadays available in a much nicer way by having systemd in
1220 the initrd serialize its state and have the hosts systemd
1221 deserialize it again.
1223 * The udev "keymap" data files and tools to apply keyboard
1224 specific mappings of scan to key codes, and force-release
1225 scan code lists have been entirely replaced by a udev
1226 "keyboard" builtin and a hwdb data file.
1228 * systemd will now honour the kernel's "quiet" command line
1229 argument also during late shutdown, resulting in a
1230 completely silent shutdown when used.
1232 * There's now an option to control the SO_REUSEPORT socket
1233 option in .socket units.
1235 * Instance units will now automatically get a per-template
1236 subslice of system.slice unless something else is explicitly
1237 configured. For example, instances of sshd@.service will now
1238 implicitly be placed in system-sshd.slice rather than
1239 system.slice as before.
1241 * Test coverage support may now be enabled at build time.
1243 Contributions from: Dave Reisner, Frederic Crozat, Harald
1244 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
1245 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
1246 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
1247 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
1248 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
1249 Giokas, Zbigniew Jędrzejewski-Szmek
1251 -- Berlin, 2013-07-23
1255 * Two new unit types have been introduced:
1257 Scope units are very similar to service units, however, are
1258 created out of pre-existing processes -- instead of PID 1
1259 forking off the processes. By using scope units it is
1260 possible for system services and applications to group their
1261 own child processes (worker processes) in a powerful way
1262 which then maybe used to organize them, or kill them
1263 together, or apply resource limits on them.
1265 Slice units may be used to partition system resources in an
1266 hierarchial fashion and then assign other units to them. By
1267 default there are now three slices: system.slice (for all
1268 system services), user.slice (for all user sessions),
1269 machine.slice (for VMs and containers).
1271 Slices and scopes have been introduced primarily in
1272 context of the work to move cgroup handling to a
1273 single-writer scheme, where only PID 1
1274 creates/removes/manages cgroups.
1276 * There's a new concept of "transient" units. In contrast to
1277 normal units these units are created via an API at runtime,
1278 not from configuration from disk. More specifically this
1279 means it is now possible to run arbitrary programs as
1280 independent services, with all execution parameters passed
1281 in via bus APIs rather than read from disk. Transient units
1282 make systemd substantially more dynamic then it ever was,
1283 and useful as a general batch manager.
1285 * logind has been updated to make use of scope and slice units
1286 for managing user sessions. As a user logs in he will get
1287 his own private slice unit, to which all sessions are added
1288 as scope units. We also added support for automatically
1289 adding an instance of user@.service for the user into the
1290 slice. Effectively logind will no longer create cgroup
1291 hierarchies on its own now, it will defer entirely to PID 1
1292 for this by means of scope, service and slice units. Since
1293 user sessions this way become entities managed by PID 1
1294 the output of "systemctl" is now a lot more comprehensive.
1296 * A new mini-daemon "systemd-machined" has been added which
1297 may be used by virtualization managers to register local
1298 VMs/containers. nspawn has been updated accordingly, and
1299 libvirt will be updated shortly. machined will collect a bit
1300 of meta information about the VMs/containers, and assign
1301 them their own scope unit (see above). The collected
1302 meta-data is then made available via the "machinectl" tool,
1303 and exposed in "ps" and similar tools. machined/machinectl
1304 is compile-time optional.
1306 * As discussed earlier, the low-level cgroup configuration
1307 options ControlGroup=, ControlGroupModify=,
1308 ControlGroupPersistent=, ControlGroupAttribute= have been
1309 removed. Please use high-level attribute settings instead as
1310 well as slice units.
1312 * A new bus call SetUnitProperties() has been added to alter
1313 various runtime parameters of a unit. This is primarily
1314 useful to alter cgroup parameters dynamically in a nice way,
1315 but will be extended later on to make more properties
1316 modifiable at runtime. systemctl gained a new set-properties
1317 command that wraps this call.
1319 * A new tool "systemd-run" has been added which can be used to
1320 run arbitrary command lines as transient services or scopes,
1321 while configuring a number of settings via the command
1322 line. This tool is currently very basic, however already
1323 very useful. We plan to extend this tool to even allow
1324 queuing of execution jobs with time triggers from the
1325 command line, similar in fashion to "at".
1327 * nspawn will now inform the user explicitly that kernels with
1328 audit enabled break containers, and suggest the user to turn
1331 * Support for detecting the IMA and AppArmor security
1332 frameworks with ConditionSecurity= has been added.
1334 * journalctl gained a new "-k" switch for showing only kernel
1335 messages, mimicking dmesg output; in addition to "--user"
1336 and "--system" switches for showing only user's own logs
1339 * systemd-delta can now show information about drop-in
1340 snippets extending unit files.
1342 * libsystemd-bus has been substantially updated but is still
1343 not available as public API.
1345 * systemd will now look for the "debug" argument on the kernel
1346 command line and enable debug logging, similar to what
1347 "systemd.log_level=debug" already did before.
1349 * "systemctl set-default", "systemctl get-default" has been
1350 added to configure the default.target symlink, which
1351 controls what to boot into by default.
1353 * "systemctl set-log-level" has been added as a convenient
1354 way to raise and lower systemd logging threshold.
1356 * "systemd-analyze plot" will now show the time the various
1357 generators needed for execution, as well as information
1358 about the unit file loading.
1360 * libsystemd-journal gained a new sd_journal_open_files() call
1361 for opening specific journal files. journactl also gained a
1362 new switch to expose this new functionality. Previously we
1363 only supported opening all files from a directory, or all
1364 files from the system, as opening individual files only is
1365 racy due to journal file rotation.
1367 * systemd gained the new DefaultEnvironment= setting in
1368 /etc/systemd/system.conf to set environment variables for
1371 * If a privileged process logs a journal message with the
1372 OBJECT_PID= field set, then journald will automatically
1373 augment this with additional OBJECT_UID=, OBJECT_GID=,
1374 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
1375 system services want to log events about specific client
1376 processes. journactl/systemctl has been updated to make use
1377 of this information if all log messages regarding a specific
1380 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
1381 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
1382 Reisner, David Coppa, David King, David Strauss, Eelco
1383 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
1384 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
1385 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
1386 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
1387 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
1388 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
1389 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
1390 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
1391 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
1392 Łukasz Stelmach, 장동준
1396 * The Python bindings gained some minimal support for the APIs
1397 exposed by libsystemd-logind.
1399 * ConditionSecurity= gained support for detecting SMACK. Since
1400 this condition already supports SELinux and AppArmor we only
1401 miss IMA for this. Patches welcome!
1403 Contributions from: Karol Lewandowski, Lennart Poettering,
1404 Zbigniew Jędrzejewski-Szmek
1408 * systemd-nspawn will now create /etc/resolv.conf if
1409 necessary, before bind-mounting the host's file onto it.
1411 * systemd-nspawn will now store meta information about a
1412 container on the container's cgroup as extended attribute
1413 fields, including the root directory.
1415 * The cgroup hierarchy has been reworked in many ways. All
1416 objects any of the components systemd creates in the cgroup
1417 tree are now suffixed. More specifically, user sessions are
1418 now placed in cgroups suffixed with ".session", users in
1419 cgroups suffixed with ".user", and nspawn containers in
1420 cgroups suffixed with ".nspawn". Furthermore, all cgroup
1421 names are now escaped in a simple scheme to avoid collision
1422 of userspace object names with kernel filenames. This work
1423 is preparation for making these objects relocatable in the
1424 cgroup tree, in order to allow easy resource partitioning of
1425 these objects without causing naming conflicts.
1427 * systemctl list-dependencies gained the new switches
1428 --plain, --reverse, --after and --before.
1430 * systemd-inhibit now shows the process name of processes that
1431 have taken an inhibitor lock.
1433 * nss-myhostname will now also resolve "localhost"
1434 implicitly. This makes /etc/hosts an optional file and
1435 nicely handles that on IPv6 ::1 maps to both "localhost" and
1438 * libsystemd-logind.so gained a new call
1439 sd_get_machine_names() to enumerate running containers and
1440 VMs (currently only supported by very new libvirt and
1441 nspawn). sd_login_monitor can now be used to watch
1442 VMs/containers coming and going.
1444 * .include is not allowed recursively anymore, and only in
1445 unit files. Usually it is better to use drop-in snippets in
1446 .d/*.conf anyway, as introduced with systemd 198.
1448 * systemd-analyze gained a new "critical-chain" command that
1449 determines the slowest chain of units run during system
1450 boot-up. It is very useful for tracking down where
1451 optimizing boot time is the most beneficial.
1453 * systemd will no longer allow manipulating service paths in
1454 the name=systemd:/system cgroup tree using ControlGroup= in
1455 units. (But is still fine with it in all other dirs.)
1457 * There's a new systemd-nspawn@.service service file that may
1458 be used to easily run nspawn containers as system
1459 services. With the container's root directory in
1460 /var/lib/container/foobar it is now sufficient to run
1461 "systemctl start systemd-nspawn@foobar.service" to boot it.
1463 * systemd-cgls gained a new parameter "--machine" to list only
1464 the processes within a certain container.
1466 * ConditionSecurity= now can check for "apparmor". We still
1467 are lacking checks for SMACK and IMA for this condition
1468 check though. Patches welcome!
1470 * A new configuration file /etc/systemd/sleep.conf has been
1471 added that may be used to configure which kernel operation
1472 systemd is supposed to execute when "suspend", "hibernate"
1473 or "hybrid-sleep" is requested. This makes the new kernel
1474 "freeze" state accessible to the user.
1476 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
1477 the passed argument if applicable.
1479 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
1480 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
1481 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
1482 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
1483 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
1484 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
1485 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
1490 * The output of 'systemctl list-jobs' got some polishing. The
1491 '--type=' argument may now be passed more than once. A new
1492 command 'systemctl list-sockets' has been added which shows
1493 a list of kernel sockets systemd is listening on with the
1494 socket units they belong to, plus the units these socket
1497 * The experimental libsystemd-bus library got substantial
1498 updates to work in conjunction with the (also experimental)
1499 kdbus kernel project. It works well enough to exchange
1500 messages with some sophistication. Note that kdbus is not
1501 ready yet, and the library is mostly an elaborate test case
1502 for now, and not installable.
1504 * systemd gained a new unit 'systemd-static-nodes.service'
1505 that generates static device nodes earlier during boot, and
1506 can run in conjunction with udev.
1508 * libsystemd-login gained a new call sd_pid_get_user_unit()
1509 to retrieve the user systemd unit a process is running
1510 in. This is useful for systems where systemd is used as
1513 * systemd-nspawn now places all containers in the new /machine
1514 top-level cgroup directory in the name=systemd
1515 hierarchy. libvirt will soon do the same, so that we get a
1516 uniform separation of /system, /user and /machine for system
1517 services, user processes and containers/virtual
1518 machines. This new cgroup hierarchy is also useful to stick
1519 stable names to specific container instances, which can be
1520 recognized later this way (this name may be controlled
1521 via systemd-nspawn's new -M switch). libsystemd-login also
1522 gained a new call sd_pid_get_machine_name() to retrieve the
1523 name of the container/VM a specific process belongs to.
1525 * bootchart can now store its data in the journal.
1527 * libsystemd-journal gained a new call
1528 sd_journal_add_conjunction() for AND expressions to the
1529 matching logic. This can be used to express more complex
1530 logical expressions.
1532 * journactl can now take multiple --unit= and --user-unit=
1535 * The cryptsetup logic now understands the "luks.key=" kernel
1536 command line switch for specifying a file to read the
1537 decryption key from. Also, if a configured key file is not
1538 found the tool will now automatically fall back to prompting
1541 * Python systemd.journal module was updated to wrap recently
1542 added functions from libsystemd-journal. The interface was
1543 changed to bring the low level interface in s.j._Reader
1544 closer to the C API, and the high level interface in
1545 s.j.Reader was updated to wrap and convert all data about
1548 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
1549 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
1550 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
1551 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
1552 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
1553 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1557 * journalctl --update-catalog now understands a new --root=
1558 option to operate on catalogs found in a different root
1561 * During shutdown after systemd has terminated all running
1562 services a final killing loop kills all remaining left-over
1563 processes. We will now print the name of these processes
1564 when we send SIGKILL to them, since this usually indicates a
1567 * If /etc/crypttab refers to password files stored on
1568 configured mount points automatic dependencies will now be
1569 generated to ensure the specific mount is established first
1570 before the key file is attempted to be read.
1572 * 'systemctl status' will now show information about the
1573 network sockets a socket unit is listening on.
1575 * 'systemctl status' will also shown information about any
1576 drop-in configuration file for units. (Drop-In configuration
1577 files in this context are files such as
1578 /etc/systemd/systemd/foobar.service.d/*.conf)
1580 * systemd-cgtop now optionally shows summed up CPU times of
1581 cgroups. Press '%' while running cgtop to switch between
1582 percentage and absolute mode. This is useful to determine
1583 which cgroups use up the most CPU time over the entire
1584 runtime of the system. systemd-cgtop has also been updated
1585 to be 'pipeable' for processing with further shell tools.
1587 * 'hostnamectl set-hostname' will now allow setting of FQDN
1590 * The formatting and parsing of time span values has been
1591 changed. The parser now understands fractional expressions
1592 such as "5.5h". The formatter will now output fractional
1593 expressions for all time spans under 1min, i.e. "5.123456s"
1594 rather than "5s 123ms 456us". For time spans under 1s
1595 millisecond values are shown, for those under 1ms
1596 microsecond values are shown. This should greatly improve
1597 all time-related output of systemd.
1599 * libsystemd-login and libsystemd-journal gained new
1600 functions for querying the poll() events mask and poll()
1601 timeout value for integration into arbitrary event
1604 * localectl gained the ability to list available X11 keymaps
1605 (models, layouts, variants, options).
1607 * 'systemd-analyze dot' gained the ability to filter for
1608 specific units via shell-style globs, to create smaller,
1609 more useful graphs. I.e. it is now possible to create simple
1610 graphs of all the dependencies between only target units, or
1611 of all units that Avahi has dependencies with.
1613 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
1614 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
1615 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
1616 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
1617 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
1618 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
1619 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
1623 * The boot-time readahead implementation for rotating media
1624 will now read the read-ahead data in multiple passes which
1625 consist of all read requests made in equidistant time
1626 intervals. This means instead of strictly reading read-ahead
1627 data in its physical order on disk we now try to find a
1628 middle ground between physical and access time order.
1630 * /etc/os-release files gained a new BUILD_ID= field for usage
1631 on operating systems that provide continuous builds of OS
1634 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
1635 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
1636 William Douglas, Zbigniew Jędrzejewski-Szmek
1640 * systemd-python gained an API exposing libsystemd-daemon.
1642 * The SMACK setup logic gained support for uploading CIPSO
1645 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
1646 ReadOnlyDirectories= and InaccessibleDirectories= has
1647 changed. The private /tmp and /var/tmp directories are now
1648 shared by all processes of a service (which means
1649 ExecStartPre= may now leave data in /tmp that ExecStart= of
1650 the same service can still access). When a service is
1651 stopped its temporary directories are immediately deleted
1652 (normal clean-up with tmpfiles is still done in addition to
1655 * By default, systemd will now set a couple of sysctl
1656 variables in the kernel: the safe sysrq options are turned
1657 on, IP route verification is turned on, and source routing
1658 disabled. The recently added hardlink and softlink
1659 protection of the kernel is turned on. These settings should
1660 be reasonably safe, and good defaults for all new systems.
1662 * The predictable network naming logic may now be turned off
1663 with a new kernel command line switch: net.ifnames=0.
1665 * A new libsystemd-bus module has been added that implements a
1666 pretty complete D-Bus client library. For details see:
1668 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
1670 * journald will now explicitly flush the journal files to disk
1671 at the latest 5min after each write. The file will then also
1672 be marked offline until the next write. This should increase
1673 reliability in case of a crash. The synchronization delay
1674 can be configured via SyncIntervalSec= in journald.conf.
1676 * There's a new remote-fs-setup.target unit that can be used
1677 to pull in specific services when at least one remote file
1678 system is to be mounted.
1680 * There are new targets timers.target and paths.target as
1681 canonical targets to pull user timer and path units in
1682 from. This complements sockets.target with a similar
1683 purpose for socket units.
1685 * libudev gained a new call udev_device_set_attribute_value()
1686 to set sysfs attributes of a device.
1688 * The udev daemon now sets the default number of worker
1689 processes executed in parallel based on the number of available
1690 CPUs instead of the amount of available RAM. This is supposed
1691 to provide a more reliable default and limit a too aggressive
1692 paralellism for setups with 1000s of devices connected.
1694 Contributions from: Auke Kok, Colin Walters, Cristian
1695 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
1696 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
1697 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
1698 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
1699 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
1700 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
1701 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
1702 Zbigniew Jędrzejewski-Szmek
1706 * Configuration of unit files may now be extended via drop-in
1707 files without having to edit/override the unit files
1708 themselves. More specifically, if the administrator wants to
1709 change one value for a service file foobar.service he can
1710 now do so by dropping in a configuration snippet into
1711 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
1712 will load all these snippets and apply them on top of the
1713 main unit configuration file, possibly extending or
1714 overriding its settings. Using these drop-in snippets is
1715 generally nicer than the two earlier options for changing
1716 unit files locally: copying the files from
1717 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
1718 them there; or creating a new file in /etc/systemd/system/
1719 that incorporates the original one via ".include". Drop-in
1720 snippets into these .d/ directories can be placed in any
1721 directory systemd looks for units in, and the usual
1722 overriding semantics between /usr/lib, /etc and /run apply
1725 * Most unit file settings which take lists of items can now be
1726 reset by assigning the empty string to them. For example,
1727 normally, settings such as Environment=FOO=BAR append a new
1728 environment variable assignment to the environment block,
1729 each time they are used. By assigning Environment= the empty
1730 string the environment block can be reset to empty. This is
1731 particularly useful with the .d/*.conf drop-in snippets
1732 mentioned above, since this adds the ability to reset list
1733 settings from vendor unit files via these drop-ins.
1735 * systemctl gained a new "list-dependencies" command for
1736 listing the dependencies of a unit recursively.
1738 * Inhibitors are now honored and listed by "systemctl
1739 suspend", "systemctl poweroff" (and similar) too, not only
1740 GNOME. These commands will also list active sessions by
1743 * Resource limits (as exposed by the various control group
1744 controllers) can now be controlled dynamically at runtime
1745 for all units. More specifically, you can now use a command
1746 like "systemctl set-cgroup-attr foobar.service cpu.shares
1747 2000" to alter the CPU shares a specific service gets. These
1748 settings are stored persistently on disk, and thus allow the
1749 administrator to easily adjust the resource usage of
1750 services with a few simple commands. This dynamic resource
1751 management logic is also available to other programs via the
1752 bus. Almost any kernel cgroup attribute and controller is
1755 * systemd-vconsole-setup will now copy all font settings to
1756 all allocated VTs, where it previously applied them only to
1759 * libsystemd-login gained the new sd_session_get_tty() API
1762 * This release drops support for a few legacy or
1763 distribution-specific LSB facility names when parsing init
1764 scripts: $x-display-manager, $mail-transfer-agent,
1765 $mail-transport-agent, $mail-transfer-agent, $smtp,
1766 $null. Also, the mail-transfer-agent.target unit backing
1767 this has been removed. Distributions which want to retain
1768 compatibility with this should carry the burden for
1769 supporting this themselves and patch support for these back
1770 in, if they really need to. Also, the facilities $syslog and
1771 $local_fs are now ignored, since systemd does not support
1772 early-boot LSB init scripts anymore, and these facilities
1773 are implied anyway for normal services. syslog.target has
1776 * There are new bus calls on PID1's Manager object for
1777 cancelling jobs, and removing snapshot units. Previously,
1778 both calls were only available on the Job and Snapshot
1781 * systemd-journal-gatewayd gained SSL support.
1783 * The various "environment" files, such as /etc/locale.conf
1784 now support continuation lines with a backslash ("\") as
1785 last character in the line, similarly in style (but different)
1786 to how this is supported in shells.
1788 * For normal user processes the _SYSTEMD_USER_UNIT= field is
1789 now implicitly appended to every log entry logged. systemctl
1790 has been updated to filter by this field when operating on a
1791 user systemd instance.
1793 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
1794 CAP_AUDIT_CONTROL capabilities to the capabilities set for
1795 the container. This makes it easier to boot unmodified
1796 Fedora systems in a container, which however still requires
1797 audit=0 to be passed on the kernel command line. Auditing in
1798 kernel and userspace is unfortunately still too broken in
1799 context of containers, hence we recommend compiling it out
1800 of the kernel or using audit=0. Hopefully this will be fixed
1801 one day for good in the kernel.
1803 * nspawn gained the new --bind= and --bind-ro= parameters to
1804 bind mount specific directories from the host into the
1807 * nspawn will now mount its own devpts file system instance
1808 into the container, in order not to leak pty devices from
1809 the host into the container.
1811 * systemd will now read the firmware boot time performance
1812 information from the EFI variables, if the used boot loader
1813 supports this, and takes it into account for boot performance
1814 analysis via "systemd-analyze". This is currently supported
1815 only in conjunction with Gummiboot, but could be supported
1816 by other boot loaders too. For details see:
1818 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
1820 * A new generator has been added that automatically mounts the
1821 EFI System Partition (ESP) to /boot, if that directory
1822 exists, is empty, and no other file system has been
1823 configured to be mounted there.
1825 * logind will now send out PrepareForSleep(false) out
1826 unconditionally, after coming back from suspend. This may be
1827 used by applications as asynchronous notification for
1828 system resume events.
1830 * "systemctl unlock-sessions" has been added, that allows
1831 unlocking the screens of all user sessions at once, similar
1832 to how "systemctl lock-sessions" already locked all users
1833 sessions. This is backed by a new D-Bus call UnlockSessions().
1835 * "loginctl seat-status" will now show the master device of a
1836 seat. (i.e. the device of a seat that needs to be around for
1837 the seat to be considered available, usually the graphics
1840 * tmpfiles gained a new "X" line type, that allows
1841 configuration of files and directories (with wildcards) that
1842 shall be excluded from automatic cleanup ("aging").
1844 * udev default rules set the device node permissions now only
1845 at "add" events, and do not change them any longer with a
1846 later "change" event.
1848 * The log messages for lid events and power/sleep keypresses
1849 now carry a message ID.
1851 * We now have a substantially larger unit test suite, but this
1852 continues to be work in progress.
1854 * udevadm hwdb gained a new --root= parameter to change the
1855 root directory to operate relative to.
1857 * logind will now issue a background sync() request to the kernel
1858 early at shutdown, so that dirty buffers are flushed to disk early
1859 instead of at the last moment, in order to optimize shutdown
1862 * A new bootctl tool has been added that is an interface for
1863 certain boot loader operations. This is currently a preview
1864 and is likely to be extended into a small mechanism daemon
1865 like timedated, localed, hostnamed, and can be used by
1866 graphical UIs to enumerate available boot options, and
1867 request boot into firmware operations.
1869 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
1870 the rest of the package. It also has been updated to work
1871 correctly in initrds.
1873 * Policykit previously has been runtime optional, and is now
1874 also compile time optional via a configure switch.
1876 * systemd-analyze has been reimplemented in C. Also "systemctl
1877 dot" has moved into systemd-analyze.
1879 * "systemctl status" with no further parameters will now print
1880 the status of all active or failed units.
1882 * Operations such as "systemctl start" can now be executed
1883 with a new mode "--irreversible" which may be used to queue
1884 operations that cannot accidentally be reversed by a later
1885 job queuing. This is by default used to make shutdown
1886 requests more robust.
1888 * The Python API of systemd now gained a new module for
1889 reading journal files.
1891 * A new tool kernel-install has been added that can install
1892 kernel images according to the Boot Loader Specification:
1894 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
1896 * Boot time console output has been improved to provide
1897 animated boot time output for hanging jobs.
1899 * A new tool systemd-activate has been added which can be used
1900 to test socket activation with, directly from the command
1901 line. This should make it much easier to test and debug
1902 socket activation in daemons.
1904 * journalctl gained a new "--reverse" (or -r) option to show
1905 journal output in reverse order (i.e. newest line first).
1907 * journalctl gained a new "--pager-end" (or -e) option to jump
1908 to immediately jump to the end of the journal in the
1909 pager. This is only supported in conjunction with "less".
1911 * journalctl gained a new "--user-unit=" option, that works
1912 similarly to "--unit=" but filters for user units rather than
1915 * A number of unit files to ease adoption of systemd in
1916 initrds has been added. This moves some minimal logic from
1917 the various initrd implementations into systemd proper.
1919 * The journal files are now owned by a new group
1920 "systemd-journal", which exists specifically to allow access
1921 to the journal, and nothing else. Previously, we used the
1922 "adm" group for that, which however possibly covers more
1923 than just journal/log file access. This new group is now
1924 already used by systemd-journal-gatewayd to ensure this
1925 daemon gets access to the journal files and as little else
1926 as possible. Note that "make install" will also set FS ACLs
1927 up for /var/log/journal to give "adm" and "wheel" read
1928 access to it, in addition to "systemd-journal" which owns
1929 the journal files. We recommend that packaging scripts also
1930 add read access to "adm" + "wheel" to /var/log/journal, and
1931 all existing/future journal files. To normal users and
1932 administrators little changes, however packagers need to
1933 ensure to create the "systemd-journal" system group at
1934 package installation time.
1936 * The systemd-journal-gatewayd now runs as unprivileged user
1937 systemd-journal-gateway:systemd-journal-gateway. Packaging
1938 scripts need to create these system user/group at
1941 * timedated now exposes a new boolean property CanNTP that
1942 indicates whether a local NTP service is available or not.
1944 * systemd-detect-virt will now also detect xen PVs
1946 * The pstore file system is now mounted by default, if it is
1949 * In addition to the SELinux and IMA policies we will now also
1950 load SMACK policies at early boot.
1952 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
1953 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
1954 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
1955 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
1956 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
1957 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
1958 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
1959 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
1960 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
1961 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
1962 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
1963 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
1964 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
1965 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
1969 * Timer units now support calendar time events in addition to
1970 monotonic time events. That means you can now trigger a unit
1971 based on a calendar time specification such as "Thu,Fri
1972 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
1973 or fifth day of any month of the year 2013, given that it is
1974 a thursday or friday. This brings timer event support
1975 considerably closer to cron's capabilities. For details on
1976 the supported calendar time specification language see
1979 * udev now supports a number of different naming policies for
1980 network interfaces for predictable names, and a combination
1981 of these policies is now the default. Please see this wiki
1982 document for details:
1984 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
1986 * Auke Kok's bootchart implementation has been added to the
1987 systemd tree. It is an optional component that can graph the
1988 boot in quite some detail. It is one of the best bootchart
1989 implementations around and minimal in its code and
1992 * nss-myhostname has been integrated into the systemd source
1993 tree. nss-myhostname guarantees that the local hostname
1994 always stays resolvable via NSS. It has been a weak
1995 requirement of systemd-hostnamed since a long time, and
1996 since its code is actually trivial we decided to just
1997 include it in systemd's source tree. It can be turned off
1998 with a configure switch.
2000 * The read-ahead logic is now capable of properly detecting
2001 whether a btrfs file system is on SSD or rotating media, in
2002 order to optimize the read-ahead scheme. Previously, it was
2003 only capable of detecting this on traditional file systems
2006 * In udev, additional device properties are now read from the
2007 IAB in addition to the OUI database. Also, Bluetooth company
2008 identities are attached to the devices as well.
2010 * In service files %U may be used as specifier that is
2011 replaced by the configured user name of the service.
2013 * nspawn may now be invoked without a controlling TTY. This
2014 makes it suitable for invocation as its own service. This
2015 may be used to set up a simple containerized server system
2016 using only core OS tools.
2018 * systemd and nspawn can now accept socket file descriptors
2019 when they are started for socket activation. This enables
2020 implementation of socket activated nspawn
2021 containers. i.e. think about autospawning an entire OS image
2022 when the first SSH or HTTP connection is received. We expect
2023 that similar functionality will also be added to libvirt-lxc
2026 * journalctl will now suppress ANSI color codes when
2027 presenting log data.
2029 * systemctl will no longer show control group information for
2030 a unit if a the control group is empty anyway.
2032 * logind can now automatically suspend/hibernate/shutdown the
2035 * /etc/machine-info and hostnamed now also expose the chassis
2036 type of the system. This can be used to determine whether
2037 the local system is a laptop, desktop, handset or
2038 tablet. This information may either be configured by the
2039 user/vendor or is automatically determined from ACPI and DMI
2040 information if possible.
2042 * A number of PolicyKit actions are now bound together with
2043 "imply" rules. This should simplify creating UIs because
2044 many actions will now authenticate similar ones as well.
2046 * Unit files learnt a new condition ConditionACPower= which
2047 may be used to conditionalize a unit depending on whether an
2048 AC power source is connected or not, of whether the system
2049 is running on battery power.
2051 * systemctl gained a new "is-failed" verb that may be used in
2052 shell scripts and suchlike to check whether a specific unit
2053 is in the "failed" state.
2055 * The EnvironmentFile= setting in unit files now supports file
2056 globbing, and can hence be used to easily read a number of
2057 environment files at once.
2059 * systemd will no longer detect and recognize specific
2060 distributions. All distribution-specific #ifdeffery has been
2061 removed, systemd is now fully generic and
2062 distribution-agnostic. Effectively, not too much is lost as
2063 a lot of the code is still accessible via explicit configure
2064 switches. However, support for some distribution specific
2065 legacy configuration file formats has been dropped. We
2066 recommend distributions to simply adopt the configuration
2067 files everybody else uses now and convert the old
2068 configuration from packaging scripts. Most distributions
2069 already did that. If that's not possible or desirable,
2070 distributions are welcome to forward port the specific
2071 pieces of code locally from the git history.
2073 * When logging a message about a unit systemd will now always
2074 log the unit name in the message meta data.
2076 * localectl will now also discover system locale data that is
2077 not stored in locale archives, but directly unpacked.
2079 * logind will no longer unconditionally use framebuffer
2080 devices as seat masters, i.e. as devices that are required
2081 to be existing before a seat is considered preset. Instead,
2082 it will now look for all devices that are tagged as
2083 "seat-master" in udev. By default framebuffer devices will
2084 be marked as such, but depending on local systems other
2085 devices might be marked as well. This may be used to
2086 integrate graphics cards using closed source drivers (such
2087 as NVidia ones) more nicely into logind. Note however, that
2088 we recommend using the open source NVidia drivers instead,
2089 and no udev rules for the closed-source drivers will be
2090 shipped from us upstream.
2092 Contributions from: Adam Williamson, Alessandro Crismani, Auke
2093 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
2094 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
2095 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
2096 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
2097 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
2098 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
2099 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
2100 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
2101 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
2102 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
2103 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
2108 * udev gained support for loading additional device properties
2109 from an indexed database that is keyed by vendor/product IDs
2110 and similar device identifiers. For the beginning this
2111 "hwdb" is populated with data from the well-known PCI and
2112 USB database, but also includes PNP, ACPI and OID data. In
2113 the longer run this indexed database shall grow into
2114 becoming the one central database for non-essential
2115 userspace device metadata. Previously, data from the PCI/USB
2116 database was only attached to select devices, since the
2117 lookup was a relatively expensive operation due to O(n) time
2118 complexity (with n being the number of entries in the
2119 database). Since this is now O(1), we decided to add in this
2120 data for all devices where this is available, by
2121 default. Note that the indexed database needs to be rebuilt
2122 when new data files are installed. To achieve this you need
2123 to update your packaging scripts to invoke "udevadm hwdb
2124 --update" after installation of hwdb data files. For
2125 RPM-based distributions we introduced the new
2126 %udev_hwdb_update macro for this purpose.
2128 * The Journal gained support for the "Message Catalog", an
2129 indexed database to link up additional information with
2130 journal entries. For further details please check:
2132 http://www.freedesktop.org/wiki/Software/systemd/catalog
2134 The indexed message catalog database also needs to be
2135 rebuilt after installation of message catalog files. Use
2136 "journalctl --update-catalog" for this. For RPM-based
2137 distributions we introduced the %journal_catalog_update
2138 macro for this purpose.
2140 * The Python Journal bindings gained support for the standard
2141 Python logging framework.
2143 * The Journal API gained new functions for checking whether
2144 the underlying file system of a journal file is capable of
2145 properly reporting file change notifications, or whether
2146 applications that want to reflect journal changes "live"
2147 need to recheck journal files continuously in appropriate
2150 * It is now possible to set the "age" field for tmpfiles
2151 entries to 0, indicating that files matching this entry
2152 shall always be removed when the directories are cleaned up.
2154 * coredumpctl gained a new "gdb" verb which invokes gdb
2155 right-away on the selected coredump.
2157 * There's now support for "hybrid sleep" on kernels that
2158 support this, in addition to "suspend" and "hibernate". Use
2159 "systemctl hybrid-sleep" to make use of this.
2161 * logind's HandleSuspendKey= setting (and related settings)
2162 now gained support for a new "lock" setting to simply
2163 request the screen lock on all local sessions, instead of
2164 actually executing a suspend or hibernation.
2166 * systemd will now mount the EFI variables file system by
2169 * Socket units now gained support for configuration of the
2170 SMACK security label.
2172 * timedatectl will now output the time of the last and next
2173 daylight saving change.
2175 * We dropped support for various legacy and distro-specific
2176 concepts, such as insserv, early-boot SysV services
2177 (i.e. those for non-standard runlevels such as 'b' or 'S')
2178 or ArchLinux /etc/rc.conf support. We recommend the
2179 distributions who still need support this to either continue
2180 to maintain the necessary patches downstream, or find a
2181 different solution. (Talk to us if you have questions!)
2183 * Various systemd components will now bypass PolicyKit checks
2184 for root and otherwise handle properly if PolicyKit is not
2185 found to be around. This should fix most issues for
2186 PolicyKit-less systems. Quite frankly this should have been
2187 this way since day one. It is absolutely our intention to
2188 make systemd work fine on PolicyKit-less systems, and we
2189 consider it a bug if something does not work as it should if
2190 PolicyKit is not around.
2192 * For embedded systems it is now possible to build udev and
2193 systemd without blkid and/or kmod support.
2195 * "systemctl switch-root" is now capable of switching root
2196 more than once. I.e. in addition to transitions from the
2197 initrd to the host OS it is now possible to transition to
2198 further OS images from the host. This is useful to implement
2199 offline updating tools.
2201 * Various other additions have been made to the RPM macros
2202 shipped with systemd. Use %udev_rules_update() after
2203 installing new udev rules files. %_udevhwdbdir,
2204 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
2205 %_sysctldir are now available which resolve to the right
2206 directories for packages to place various data files in.
2208 * journalctl gained the new --full switch (in addition to
2209 --all, to disable ellipsation for long messages.
2211 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
2212 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
2213 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
2214 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
2215 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
2216 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
2217 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
2218 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
2219 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
2223 * journalctl gained new --since= and --until= switches to
2224 filter by time. It also now supports nice filtering for
2225 units via --unit=/-u.
2227 * Type=oneshot services may use ExecReload= and do the
2230 * The journal daemon now supports time-based rotation and
2231 vacuuming, in addition to the usual disk-space based
2234 * The journal will now index the available field values for
2235 each field name. This enables clients to show pretty drop
2236 downs of available match values when filtering. The bash
2237 completion of journalctl has been updated
2238 accordingly. journalctl gained a new switch -F to list all
2239 values a certain field takes in the journal database.
2241 * More service events are now written as structured messages
2242 to the journal, and made recognizable via message IDs.
2244 * The timedated, localed and hostnamed mini-services which
2245 previously only provided support for changing time, locale
2246 and hostname settings from graphical DEs such as GNOME now
2247 also have a minimal (but very useful) text-based client
2248 utility each. This is probably the nicest way to changing
2249 these settings from the command line now, especially since
2250 it lists available options and is fully integrated with bash
2253 * There's now a new tool "systemd-coredumpctl" to list and
2254 extract coredumps from the journal.
2256 * We now install a README each in /var/log/ and
2257 /etc/rc.d/init.d explaining where the system logs and init
2258 scripts went. This hopefully should help folks who go to
2259 that dirs and look into the otherwise now empty void and
2260 scratch their heads.
2262 * When user-services are invoked (by systemd --user) the
2263 $MANAGERPID env var is set to the PID of systemd.
2265 * SIGRTMIN+24 when sent to a --user instance will now result
2266 in immediate termination of systemd.
2268 * gatewayd received numerous feature additions such as a
2269 "follow" mode, for live syncing and filtering.
2271 * browse.html now allows filtering and showing detailed
2272 information on specific entries. Keyboard navigation and
2273 mouse screen support has been added.
2275 * gatewayd/journalctl now supports HTML5/JSON
2276 Server-Sent-Events as output.
2278 * The SysV init script compatibility logic will now
2279 heuristically determine whether a script supports the
2280 "reload" verb, and only then make this available as
2283 * "systemctl status --follow" has been removed, use "journalctl
2286 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
2287 have been removed since they are hardly useful to be
2290 * And I'd like to take the opportunity to specifically mention
2291 Zbigniew for his great contributions. Zbigniew, you rock!
2293 Contributions from: Andrew Eikum, Christian Hesse, Colin
2294 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
2295 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
2296 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
2297 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
2298 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
2299 Jędrzejewski-Szmek, Сковорода Никита Андреевич
2303 * If /etc/vconsole.conf is non-existent or empty we will no
2304 longer load any console font or key map at boot by
2305 default. Instead the kernel defaults will be left
2306 intact. This is definitely the right thing to do, as no
2307 configuration should mean no configuration, and hard-coding
2308 font names that are different on all archs is probably a bad
2309 idea. Also, the kernel default key map and font should be
2310 good enough for most cases anyway, and mostly identical to
2311 the userspace fonts/key maps we previously overloaded them
2312 with. If distributions want to continue to default to a
2313 non-kernel font or key map they should ship a default
2314 /etc/vconsole.conf with the appropriate contents.
2316 Contributions from: Colin Walters, Daniel J Walsh, Dave
2317 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
2318 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
2322 * journalctl gained a new --cursor= switch to show entries
2323 starting from the specified location in the journal.
2325 * We now enforce a size limit on journal entry fields exported
2326 with "-o json" in journalctl. Fields larger than 4K will be
2327 assigned null. This can be turned off with --all.
2329 * An (optional) journal gateway daemon is now available as
2330 "systemd-journal-gatewayd.service". This service provides
2331 access to the journal via HTTP and JSON. This functionality
2332 will be used to implement live log synchronization in both
2333 pull and push modes, but has various other users too, such
2334 as easy log access for debugging of embedded devices. Right
2335 now it is already useful to retrieve the journal via HTTP:
2337 # systemctl start systemd-journal-gatewayd.service
2338 # wget http://localhost:19531/entries
2340 This will download the journal contents in a
2341 /var/log/messages compatible format. The same as JSON:
2343 # curl -H"Accept: application/json" http://localhost:19531/entries
2345 This service is also accessible via a web browser where a
2346 single static HTML5 app is served that uses the JSON logic
2347 to enable the user to do some basic browsing of the
2348 journal. This will be extended later on. Here's an example
2349 screenshot of this app in its current state:
2351 http://0pointer.de/public/journal-gatewayd
2353 Contributions from: Kay Sievers, Lennart Poettering, Robert
2354 Milasan, Tom Gundersen
2358 * The bash completion logic is now available for journalctl
2361 * We do not mount the "cpuset" controller anymore together with
2362 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
2363 started if no parameters are assigned to it. "cpuset" hence
2364 broke code that assumed it it could create "cpu" groups and
2367 * journalctl -f will now subscribe to terminal size changes,
2368 and line break accordingly.
2370 Contributions from: Dave Reisner, Kay Sievers, Lennart
2371 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
2375 * nspawn will now create a symlink /etc/localtime in the
2376 container environment, copying the host's timezone
2377 setting. Previously this has been done via a bind mount, but
2378 since symlinks cannot be bind mounted this has now been
2379 changed to create/update the appropriate symlink.
2381 * journalctl -n's line number argument is now optional, and
2382 will default to 10 if omitted.
2384 * journald will now log the maximum size the journal files may
2385 take up on disk. This is particularly useful if the default
2386 built-in logic of determining this parameter from the file
2387 system size is used. Use "systemctl status
2388 systemd-journald.service" to see this information.
2390 * The multi-seat X wrapper tool has been stripped down. As X
2391 is now capable of enumerating graphics devices via udev in a
2392 seat-aware way the wrapper is not strictly necessary
2393 anymore. A stripped down temporary stop-gap is still shipped
2394 until the upstream display managers have been updated to
2395 fully support the new X logic. Expect this wrapper to be
2396 removed entirely in one of the next releases.
2398 * HandleSleepKey= in logind.conf has been split up into
2399 HandleSuspendKey= and HandleHibernateKey=. The old setting
2400 is not available anymore. X11 and the kernel are
2401 distuingishing between these keys and we should too. This
2402 also means the inhibition lock for these keys has been split
2405 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
2406 Poettering, Lukas Nykryn, Václav Pavlín
2410 * Whenever a unit changes state we will now log this to the
2411 journal and show along the unit's own log output in
2414 * ConditionPathIsMountPoint= can now properly detect bind
2415 mount points too. (Previously, a bind mount of one file
2416 system to another place in the same file system could not be
2417 detected as mount, since they shared struct stat's st_dev
2420 * We will now mount the cgroup controllers cpu, cpuacct,
2421 cpuset and the controllers net_cls, net_prio together by
2424 * nspawn containers will now have a virtualized boot
2425 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
2426 over with a randomized ID at container initialization). This
2427 has the effect of making "journalctl -b" do the right thing
2430 * The JSON output journal serialization has been updated not
2431 to generate "endless" list objects anymore, but rather one
2432 JSON object per line. This is more in line how most JSON
2433 parsers expect JSON objects. The new output mode
2434 "json-pretty" has been added to provide similar output, but
2435 neatly aligned for readability by humans.
2437 * We dropped all explicit sync() invocations in the shutdown
2438 code. The kernel does this implicitly anyway in the kernel
2439 reboot() syscall. halt(8)'s -n option is now a compatibility
2442 * We now support virtualized reboot() in containers, as
2443 supported by newer kernels. We will fall back to exit() if
2444 CAP_SYS_REBOOT is not available to the container. Also,
2445 nspawn makes use of this now and will actually reboot the
2446 container if the containerized OS asks for that.
2448 * journalctl will only show local log output by default
2449 now. Use --merge (-m) to show remote log output, too.
2451 * libsystemd-journal gained the new sd_journal_get_usage()
2452 call to determine the current disk usage of all journal
2453 files. This is exposed in the new "journalctl --disk-usage"
2456 * journald gained a new configuration setting SplitMode= in
2457 journald.conf which may be used to control how user journals
2458 are split off. See journald.conf(5) for details.
2460 * A new condition type ConditionFileNotEmpty= has been added.
2462 * tmpfiles' "w" lines now support file globbing, to write
2463 multiple files at once.
2465 * We added Python bindings for the journal submission
2466 APIs. More Python APIs for a number of selected APIs will
2467 likely follow. Note that we intend to add native bindings
2468 only for the Python language, as we consider it common
2469 enough to deserve bindings shipped within systemd. There are
2470 various projects outside of systemd that provide bindings
2471 for languages such as PHP or Lua.
2473 * Many conditions will now resolve specifiers such as %i. In
2474 addition, PathChanged= and related directives of .path units
2475 now support specifiers as well.
2477 * There's now a new RPM macro definition for the system preset
2480 * journald will now warn if it ca not forward a message to the
2481 syslog daemon because its socket is full.
2483 * timedated will no longer write or process /etc/timezone,
2484 except on Debian. As we do not support late mounted /usr
2485 anymore /etc/localtime always being a symlink is now safe,
2486 and hence the information in /etc/timezone is not necessary
2489 * logind will now always reserve one VT for a text getty (VT6
2490 by default). Previously if more than 6 X sessions where
2491 started they took up all the VTs with auto-spawned gettys,
2492 so that no text gettys were available anymore.
2494 * udev will now automatically inform the btrfs kernel logic
2495 about btrfs RAID components showing up. This should make
2496 simple hotplug based btrfs RAID assembly work.
2498 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
2499 (but not for its children which will stay at the kernel
2500 default). This should allow setups with a lot more listening
2503 * systemd will now always pass the configured timezone to the
2504 kernel at boot. timedated will do the same when the timezone
2507 * logind's inhibition logic has been updated. By default,
2508 logind will now handle the lid switch, the power and sleep
2509 keys all the time, even in graphical sessions. If DEs want
2510 to handle these events on their own they should take the new
2511 handle-power-key, handle-sleep-key and handle-lid-switch
2512 inhibitors during their runtime. A simple way to achiveve
2513 that is to invoke the DE wrapped in an invocation of:
2515 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
2517 * Access to unit operations is now checked via SELinux taking
2518 the unit file label and client process label into account.
2520 * systemd will now notify the administrator in the journal
2521 when he over-mounts a non-empty directory.
2523 * There are new specifiers that are resolved in unit files,
2524 for the host name (%H), the machine ID (%m) and the boot ID
2527 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
2528 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
2529 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
2530 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
2531 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
2532 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
2533 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
2537 * Support for reading structured kernel messages from
2538 /dev/kmsg has now been added and is enabled by default.
2540 * Support for reading kernel messages from /proc/kmsg has now
2541 been removed. If you want kernel messages in the journal
2542 make sure to run a recent kernel (>= 3.5) that supports
2543 reading structured messages from /dev/kmsg (see
2544 above). /proc/kmsg is now exclusive property of classic
2545 syslog daemons again.
2547 * The libudev API gained the new
2548 udev_device_new_from_device_id() call.
2550 * The logic for file system namespace (ReadOnlyDirectory=,
2551 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
2552 require pivot_root() anymore. This means fewer temporary
2553 directories are created below /tmp for this feature.
2555 * nspawn containers will now see and receive all submounts
2556 made on the host OS below the root file system of the
2559 * Forward Secure Sealing is now supported for Journal files,
2560 which provide cryptographical sealing of journal files so
2561 that attackers cannot alter log history anymore without this
2562 being detectable. Lennart will soon post a blog story about
2563 this explaining it in more detail.
2565 * There are two new service settings RestartPreventExitStatus=
2566 and SuccessExitStatus= which allow configuration of exit
2567 status (exit code or signal) which will be excepted from the
2568 restart logic, resp. consider successful.
2570 * journalctl gained the new --verify switch that can be used
2571 to check the integrity of the structure of journal files and
2572 (if Forward Secure Sealing is enabled) the contents of
2575 * nspawn containers will now be run with /dev/stdin, /dev/fd/
2576 and similar symlinks pre-created. This makes running shells
2577 as container init process a lot more fun.
2579 * The fstab support can now handle PARTUUID= and PARTLABEL=
2582 * A new ConditionHost= condition has been added to match
2583 against the hostname (with globs) and machine ID. This is
2584 useful for clusters where a single OS image is used to
2585 provision a large number of hosts which shall run slightly
2586 different sets of services.
2588 * Services which hit the restart limit will now be placed in a
2591 Contributions from: Bertram Poettering, Dave Reisner, Huang
2592 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
2593 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
2597 * When running in --user mode systemd will now become a
2598 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
2599 tree a lot more organized.
2601 * A new PartOf= unit dependency type has been introduced that
2602 may be used to group services in a natural way.
2604 * "systemctl enable" may now be used to enable instances of
2607 * journalctl now prints error log levels in red, and
2608 warning/notice log levels in bright white. It also supports
2609 filtering by log level now.
2611 * cgtop gained a new -n switch (similar to top), to configure
2612 the maximum number of iterations to run for. It also gained
2613 -b, to run in batch mode (accepting no input).
2615 * The suffix ".service" may now be omitted on most systemctl
2616 command lines involving service unit names.
2618 * There's a new bus call in logind to lock all sessions, as
2619 well as a loginctl verb for it "lock-sessions".
2621 * libsystemd-logind.so gained a new call sd_journal_perror()
2622 that works similar to libc perror() but logs to the journal
2623 and encodes structured information about the error number.
2625 * /etc/crypttab entries now understand the new keyfile-size=
2628 * shutdown(8) now can send a (configurable) wall message when
2629 a shutdown is cancelled.
2631 * The mount propagation mode for the root file system will now
2632 default to "shared", which is useful to make containers work
2633 nicely out-of-the-box so that they receive new mounts from
2634 the host. This can be undone locally by running "mount
2635 --make-rprivate /" if needed.
2637 * The prefdm.service file has been removed. Distributions
2638 should maintain this unit downstream if they intend to keep
2639 it around. However, we recommend writing normal unit files
2640 for display managers instead.
2642 * Since systemd is a crucial part of the OS we will now
2643 default to a number of compiler switches that improve
2644 security (hardening) such as read-only relocations, stack
2645 protection, and suchlike.
2647 * The TimeoutSec= setting for services is now split into
2648 TimeoutStartSec= and TimeoutStopSec= to allow configuration
2649 of individual time outs for the start and the stop phase of
2652 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
2653 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
2654 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
2655 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
2656 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
2657 Gundersen, Zbigniew Jędrzejewski-Szmek
2661 * The journal and id128 C APIs are now fully documented as man
2664 * Extra safety checks have been added when transitioning from
2665 the initial RAM disk to the main system to avoid accidental
2668 * /etc/crypttab entries now understand the new keyfile-offset=
2671 * systemctl -t can now be used to filter by unit load state.
2673 * The journal C API gained the new sd_journal_wait() call to
2674 make writing synchronous journal clients easier.
2676 * journalctl gained the new -D switch to show journals from a
2679 * journalctl now displays a special marker between log
2680 messages of two different boots.
2682 * The journal is now explicitly flushed to /var via a service
2683 systemd-journal-flush.service, rather than implicitly simply
2684 by seeing /var/log/journal to be writable.
2686 * journalctl (and the journal C APIs) can now match for much
2687 more complex expressions, with alternatives and
2690 * When transitioning from the initial RAM disk to the main
2691 system we will now kill all processes in a killing spree to
2692 ensure no processes stay around by accident.
2694 * Three new specifiers may be used in unit files: %u, %h, %s
2695 resolve to the user name, user home directory resp. user
2696 shell. This is useful for running systemd user instances.
2698 * We now automatically rotate journal files if their data
2699 object hash table gets a fill level > 75%. We also size the
2700 hash table based on the configured maximum file size. This
2701 together should lower hash collisions drastically and thus
2702 speed things up a bit.
2704 * journalctl gained the new "--header" switch to introspect
2705 header data of journal files.
2707 * A new setting SystemCallFilters= has been added to services
2708 which may be used to apply blacklists or whitelists to
2709 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
2711 * nspawn gained a new --link-journal= switch (and quicker: -j)
2712 to link the container journal with the host. This makes it
2713 very easy to centralize log viewing on the host for all
2714 guests while still keeping the journal files separated.
2716 * Many bugfixes and optimizations
2718 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
2719 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
2720 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
2725 * Several tools now understand kernel command line arguments,
2726 which are only read when run in an initial RAM disk. They
2727 usually follow closely their normal counterparts, but are
2730 * There's a new tool to analyze the readahead files that are
2731 automatically generated at boot. Use:
2733 /usr/lib/systemd/systemd-readahead analyze /.readahead
2735 * We now provide an early debug shell on tty9 if this enabled. Use:
2737 systemctl enable debug-shell.service
2739 * All plymouth related units have been moved into the Plymouth
2740 package. Please make sure to upgrade your Plymouth version
2743 * systemd-tmpfiles now supports getting passed the basename of
2744 a configuration file only, in which case it will look for it
2745 in all appropriate directories automatically.
2747 * udevadm info now takes a /dev or /sys path as argument, and
2748 does the right thing. Example:
2750 udevadm info /dev/sda
2751 udevadm info /sys/class/block/sda
2753 * systemctl now prints a warning if a unit is stopped but a
2754 unit that might trigger it continues to run. Example: a
2755 service is stopped but the socket that activates it is left
2758 * "systemctl status" will now mention if the log output was
2759 shortened due to rotation since a service has been started.
2761 * The journal API now exposes functions to determine the
2762 "cutoff" times due to rotation.
2764 * journald now understands SIGUSR1 and SIGUSR2 for triggering
2765 immediately flushing of runtime logs to /var if possible,
2766 resp. for triggering immediate rotation of the journal
2769 * It is now considered an error if a service is attempted to
2770 be stopped that is not loaded.
2772 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
2774 * systemd-analyze now supports Python 3
2776 * tmpfiles now supports cleaning up directories via aging
2777 where the first level dirs are always kept around but
2778 directories beneath it automatically aged. This is enabled
2779 by prefixing the age field with '~'.
2781 * Seat objects now expose CanGraphical, CanTTY properties
2782 which is required to deal with very fast bootups where the
2783 display manager might be running before the graphics drivers
2784 completed initialization.
2786 * Seat objects now expose a State property.
2788 * We now include RPM macros for service enabling/disabling
2789 based on the preset logic. We recommend RPM based
2790 distributions to make use of these macros if possible. This
2791 makes it simpler to reuse RPM spec files across
2794 * We now make sure that the collected systemd unit name is
2795 always valid when services log to the journal via
2798 * There's a new man page kernel-command-line(7) detailing all
2799 command line options we understand.
2801 * The fstab generator may now be disabled at boot by passing
2802 fstab=0 on the kernel command line.
2804 * A new kernel command line option modules-load= is now understood
2805 to load a specific kernel module statically, early at boot.
2807 * Unit names specified on the systemctl command line are now
2808 automatically escaped as needed. Also, if file system or
2809 device paths are specified they are automatically turned
2810 into the appropriate mount or device unit names. Example:
2812 systemctl status /home
2813 systemctl status /dev/sda
2815 * The SysVConsole= configuration option has been removed from
2816 system.conf parsing.
2818 * The SysV search path is no longer exported on the D-Bus
2821 * The Names= option is been removed from unit file parsing.
2823 * There's a new man page bootup(7) detailing the boot process.
2825 * Every unit and every generator we ship with systemd now
2826 comes with full documentation. The self-explanatory boot is
2829 * A couple of services gained "systemd-" prefixes in their
2830 name if they wrap systemd code, rather than only external
2831 code. Among them fsck@.service which is now
2832 systemd-fsck@.service.
2834 * The HaveWatchdog property has been removed from the D-Bus
2837 * systemd.confirm_spawn= on the kernel command line should now
2840 * There's a new man page crypttab(5) which details all options
2841 we actually understand.
2843 * systemd-nspawn gained a new --capability= switch to pass
2844 additional capabilities to the container.
2846 * timedated will now read known NTP implementation unit names
2847 from /usr/lib/systemd/ntp-units.d/*.list,
2848 systemd-timedated-ntp.target has been removed.
2850 * journalctl gained a new switch "-b" that lists log data of
2851 the current boot only.
2853 * The notify socket is in the abstract namespace again, in
2854 order to support daemons which chroot() at start-up.
2856 * There is a new Storage= configuration option for journald
2857 which allows configuration of where log data should go. This
2858 also provides a way to disable journal logging entirely, so
2859 that data collected is only forwarded to the console, the
2860 kernel log buffer or another syslog implementation.
2862 * Many bugfixes and optimizations
2864 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
2865 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
2866 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
2867 Shawn Landden, Tom Gundersen
2871 * "systemctl help <unit>" now shows the man page if one is
2874 * Several new man pages have been added.
2876 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
2877 MaxLevelConsole= can now be specified in
2878 journald.conf. These options allow reducing the amount of
2879 data stored on disk or forwarded by the log level.
2881 * TimerSlackNSec= can now be specified in system.conf for
2882 PID1. This allows system-wide power savings.
2884 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
2885 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
2890 * logind is now capable of (optionally) handling power and
2891 sleep keys as well as the lid switch.
2893 * journalctl now understands the syntax "journalctl
2894 /usr/bin/avahi-daemon" to get all log output of a specific
2897 * CapabilityBoundingSet= in system.conf now also influences
2898 the capability bound set of usermode helpers of the kernel.
2900 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
2901 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
2902 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
2903 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
2907 * Note that we skipped 139 releases here in order to set the
2908 new version to something that is greater than both udev's
2909 and systemd's most recent version number.
2911 * udev: all udev sources are merged into the systemd source tree now.
2912 All future udev development will happen in the systemd tree. It
2913 is still fully supported to use the udev daemon and tools without
2914 systemd running, like in initramfs or other init systems. Building
2915 udev though, will require the *build* of the systemd tree, but
2916 udev can be properly *run* without systemd.
2918 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
2919 should be used to create dead device nodes as workarounds for broken
2922 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
2923 no longer supported. udev_monitor_new_from_netlink() needs to be
2924 used to subscribe to events.
2926 * udev: when udevd is started by systemd, processes which are left
2927 behind by forking them off of udev rules, are unconditionally cleaned
2928 up and killed now after the event handling has finished. Services or
2929 daemons must be started as systemd services. Services can be
2930 pulled-in by udev to get started, but they can no longer be directly
2931 forked by udev rules.
2933 * udev: the daemon binary is called systemd-udevd now and installed
2934 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
2935 to adapt to that, create symlink, or rename the binary after building
2938 * libudev no longer provides these symbols:
2939 udev_monitor_from_socket()
2940 udev_queue_get_failed_list_entry()
2941 udev_get_{dev,sys,run}_path()
2942 The versions number was bumped and symbol versioning introduced.
2944 * systemd-loginctl and systemd-journalctl have been renamed
2945 to loginctl and journalctl to match systemctl.
2947 * The config files: /etc/systemd/systemd-logind.conf and
2948 /etc/systemd/systemd-journald.conf have been renamed to
2949 logind.conf and journald.conf. Package updates should rename
2950 the files to the new names on upgrade.
2952 * For almost all files the license is now LGPL2.1+, changed
2953 from the previous GPL2.0+. Exceptions are some minor stuff
2954 of udev (which will be changed to LGPL2.1 eventually, too),
2955 and the MIT licensed sd-daemon.[ch] library that is suitable
2956 to be used as drop-in files.
2958 * systemd and logind now handle system sleep states, in
2959 particular suspending and hibernating.
2961 * logind now implements a sleep/shutdown/idle inhibiting logic
2962 suitable for a variety of uses. Soonishly Lennart will blog
2963 about this in more detail.
2965 * var-run.mount and var-lock.mount are no longer provided
2966 (which prevously bind mounted these directories to their new
2967 places). Distributions which have not converted these
2968 directories to symlinks should consider stealing these files
2969 from git history and add them downstream.
2971 * We introduced the Documentation= field for units and added
2972 this to all our shipped units. This is useful to make it
2973 easier to explore the boot and the purpose of the various
2976 * All smaller setup units (such as
2977 systemd-vconsole-setup.service) now detect properly if they
2978 are run in a container and are skipped when
2979 appropriate. This guarantees an entirely noise-free boot in
2980 Linux container environments such as systemd-nspawn.
2982 * A framework for implementing offline system updates is now
2983 integrated, for details see:
2984 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
2986 * A new service type Type=idle is available now which helps us
2987 avoiding ugly interleaving of getty output and boot status
2990 * There's now a system-wide CapabilityBoundingSet= option to
2991 globally reduce the set of capabilities for the
2992 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
2993 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
2994 even CAP_NET_ADMIN system-wide for secure systems.
2996 * There are now system-wide DefaultLimitXXX= options to
2997 globally change the defaults of the various resource limits
2998 for all units started by PID 1.
3000 * Harald Hoyer's systemd test suite has been integrated into
3001 systemd which allows easy testing of systemd builds in qemu
3002 and nspawn. (This is really awesome! Ask us for details!)
3004 * The fstab parser is now implemented as generator, not inside
3007 * systemctl will now warn you if .mount units generated from
3008 /etc/fstab are out of date due to changes in fstab that
3009 have not been read by systemd yet.
3011 * systemd is now suitable for usage in initrds. Dracut has
3012 already been updated to make use of this. With this in place
3013 initrds get a slight bit faster but primarily are much
3014 easier to introspect and debug since "systemctl status" in
3015 the host system can be used to introspect initrd services,
3016 and the journal from the initrd is kept around too.
3018 * systemd-delta has been added, a tool to explore differences
3019 between user/admin configuration and vendor defaults.
3021 * PrivateTmp= now affects both /tmp and /var/tmp.
3023 * Boot time status messages are now much prettier and feature
3024 proper english language. Booting up systemd has never been
3027 * Read-ahead pack files now include the inode number of all
3028 files to pre-cache. When the inode changes the pre-caching
3029 is not attempted. This should be nicer to deal with updated
3030 packages which might result in changes of read-ahead
3033 * We now temporaritly lower the kernel's read_ahead_kb variable
3034 when collecting read-ahead data to ensure the kernel's
3035 built-in read-ahead does not add noise to our measurements
3036 of necessary blocks to pre-cache.
3038 * There's now RequiresMountsFor= to add automatic dependencies
3039 for all mounts necessary for a specific file system path.
3041 * MountAuto= and SwapAuto= have been removed from
3042 system.conf. Mounting file systems at boot has to take place
3045 * nspawn now learned a new switch --uuid= to set the machine
3046 ID on the command line.
3048 * nspawn now learned the -b switch to automatically search
3051 * vt102 is now the default TERM for serial TTYs, upgraded from
3054 * systemd-logind now works on VT-less systems.
3056 * The build tree has been reorganized. The individual
3057 components now have directories of their own.
3059 * A new condition type ConditionPathIsReadWrite= is now available.
3061 * nspawn learned the new -C switch to create cgroups for the
3062 container in other hierarchies.
3064 * We now have support for hardware watchdogs, configurable in
3067 * The scheduled shutdown logic now has a public API.
3069 * We now mount /tmp as tmpfs by default, but this can be
3070 masked and /etc/fstab can override it.
3072 * Since udisks does not make use of /media anymore we are not
3073 mounting a tmpfs on it anymore.
3075 * journalctl gained a new --local switch to only interleave
3076 locally generated journal files.
3078 * We can now load the IMA policy at boot automatically.
3080 * The GTK tools have been split off into a systemd-ui.
3082 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
3083 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
3084 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
3085 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
3086 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
3087 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
3088 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
3089 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
3090 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
3095 * This is mostly a bugfix release
3097 * Support optional initialization of the machine ID from the
3098 KVM or container configured UUID.
3100 * Support immediate reboots with "systemctl reboot -ff"
3102 * Show /etc/os-release data in systemd-analyze output
3104 * Many bugfixes for the journal, including endianness fixes and
3105 ensuring that disk space enforcement works
3107 * sd-login.h is C++ comptaible again
3109 * Extend the /etc/os-release format on request of the Debian
3112 * We now refuse non-UTF8 strings used in various configuration
3113 and unit files. This is done to ensure we do not pass invalid
3114 data over D-Bus or expose it elsewhere.
3116 * Register Mimo USB Screens as suitable for automatic seat
3119 * Read SELinux client context from journal clients in a race
3122 * Reorder configuration file lookup order. /etc now always
3123 overrides /run in order to allow the administrator to always
3124 and unconditionally override vendor supplied or
3125 automatically generated data.
3127 * The various user visible bits of the journal now have man
3128 pages. We still lack man pages for the journal API calls
3131 * We now ship all man pages in HTML format again in the
3134 Contributions from: Dave Reisner, Dirk Eibach, Frederic
3135 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
3136 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
3141 * This is mostly a bugfix release
3143 * systems lacking /etc/os-release are no longer supported.
3145 * Various functionality updates to libsystemd-login.so
3147 * Track class of PAM logins to distuingish greeters from
3150 Contributions from: Kay Sievers, Lennart Poettering, Michael
3155 * This is an important bugfix release for v41.
3157 * Building man pages is now optional which should be useful
3158 for those building systemd from git but unwilling to install
3161 * Watchdog support for supervising services is now usable. In
3162 a future release support for hardware watchdogs
3163 (i.e. /dev/watchdog) will be added building on this.
3165 * Service start rate limiting is now configurable and can be
3166 turned off per service. When a start rate limit is hit a
3167 reboot can automatically be triggered.
3169 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
3171 Contributions from: Benjamin Franzke, Bill Nottingham,
3172 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
3173 Schmidt, Michał Górny, Piotr Drąg
3177 * The systemd binary is installed /usr/lib/systemd/systemd now;
3178 An existing /sbin/init symlink needs to be adapted with the
3181 * The code that loads kernel modules has been ported to invoke
3182 libkmod directly, instead of modprobe. This means we do not
3183 support systems with module-init-tools anymore.
3185 * Watchdog support is now already useful, but still not
3188 * A new kernel command line option systemd.setenv= is
3189 understood to set system wide environment variables
3190 dynamically at boot.
3192 * We now limit the set of capabilities of systemd-journald.
3194 * We now set SIGPIPE to ignore by default, since it only is
3195 useful in shell pipelines, and has little use in general
3196 code. This can be disabled with IgnoreSIPIPE=no in unit
3199 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
3200 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
3205 * This is mostly a bugfix release
3207 * We now expose the reason why a service failed in the
3208 "Result" D-Bus property.
3210 * Rudimentary service watchdog support (will be completed over
3211 the next few releases.)
3213 * When systemd forks off in order execute some service we will
3214 now immediately changes its argv[0] to reflect which process
3215 it will execute. This is useful to minimize the time window
3216 with a generic argv[0], which makes bootcharts more useful
3218 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
3219 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
3220 Mike Kazantsev, Ray Strode
3224 * This is mostly a test release, but incorporates many
3227 * New systemd-cgtop tool to show control groups by their
3230 * Linking against libacl for ACLs is optional again. If
3231 disabled, support tracking device access for active logins
3232 goes becomes unavailable, and so does access to the user
3233 journals by the respective users.
3235 * If a group "adm" exists, journal files are automatically
3236 owned by them, thus allow members of this group full access
3237 to the system journal as well as all user journals.
3239 * The journal now stores the SELinux context of the logging
3240 client for all entries.
3242 * Add C++ inclusion guards to all public headers
3244 * New output mode "cat" in the journal to print only text
3245 messages, without any meta data like date or time.
3247 * Include tiny X server wrapper as a temporary stop-gap to
3248 teach XOrg udev display enumeration. This is used by display
3249 managers such as gdm, and will go away as soon as XOrg
3250 learned native udev hotplugging for display devices.
3252 * Add new systemd-cat tool for executing arbitrary programs
3253 with STDERR/STDOUT connected to the journal. Can also act as
3254 BSD logger replacement, and does so by default.
3256 * Optionally store all locally generated coredumps in the
3257 journal along with meta data.
3259 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
3260 writing short strings to files (for usage for /sys), and for
3261 creating symlinks, character and block device nodes.
3263 * New unit file option ControlGroupPersistent= to make cgroups
3264 persistent, following the mechanisms outlined in
3265 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
3267 * Support multiple local RTCs in a sane way
3269 * No longer monopolize IO when replaying readahead data on
3270 rotating disks, since we might starve non-file-system IO to
3271 death, since fanotify() will not see accesses done by blkid,
3274 * Do not show kernel threads in systemd-cgls anymore, unless
3275 requested with new -k switch.
3277 Contributions from: Dan Horák, Kay Sievers, Lennart
3278 Poettering, Michal Schmidt
3282 * This is mostly a test release, but incorporates many
3285 * The git repository moved to:
3286 git://anongit.freedesktop.org/systemd/systemd
3287 ssh://git.freedesktop.org/git/systemd/systemd
3289 * First release with the journal
3290 http://0pointer.de/blog/projects/the-journal.html
3292 * The journal replaces both systemd-kmsg-syslogd and
3293 systemd-stdout-bridge.
3295 * New sd_pid_get_unit() API call in libsystemd-logind
3297 * Many systemadm clean-ups
3299 * Introduce remote-fs-pre.target which is ordered before all
3300 remote mounts and may be used to start services before all
3303 * Added Mageia support
3305 * Add bash completion for systemd-loginctl
3307 * Actively monitor PID file creation for daemons which exit in
3308 the parent process before having finished writing the PID
3309 file in the daemon process. Daemons which do this need to be
3310 fixed (i.e. PID file creation must have finished before the
3311 parent exits), but we now react a bit more gracefully to them.
3313 * Add colourful boot output, mimicking the well-known output
3314 of existing distributions.
3316 * New option PassCredentials= for socket units, for
3317 compatibility with a recent kernel ABI breakage.
3319 * /etc/rc.local is now hooked in via a generator binary, and
3320 thus will no longer act as synchronization point during
3323 * systemctl list-unit-files now supports --root=.
3325 * systemd-tmpfiles now understands two new commands: z, Z for
3326 relabelling files according to the SELinux database. This is
3327 useful to apply SELinux labels to specific files in /sys,
3330 * Output of SysV services is now forwarded to both the console
3331 and the journal by default, not only just the console.
3333 * New man pages for all APIs from libsystemd-login.
3335 * The build tree got reorganized and a the build system is a
3336 lot more modular allowing embedded setups to specifically
3337 select the components of systemd they are interested in.
3339 * Support for Linux systems lacking the kernel VT subsystem is
3342 * configure's --with-rootdir= got renamed to
3343 --with-rootprefix= to follow the naming used by udev and
3346 * Unless specified otherwise we will now install to /usr instead
3347 of /usr/local by default.
3349 * Processes with '@' in argv[0][0] are now excluded from the
3350 final shut-down killing spree, following the logic explained
3352 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
3354 * All processes remaining in a service cgroup when we enter
3355 the START or START_PRE states are now killed with
3356 SIGKILL. That means it is no longer possible to spawn
3357 background processes from ExecStart= lines (which was never
3358 supported anyway, and bad style).
3360 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
3361 reloading of units together.
3363 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
3364 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
3365 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
3366 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
3367 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek