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 Contributions from: Ali H. Caliskan, Alison Chaiken, Bas van
106 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
107 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
108 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
109 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
110 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
111 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
112 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
113 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
114 Marcel Holtmann, Matthew Monaco, Michael Marineau, Michael
115 Olbrich, Michal Sekletar, Mike Gilbert, Nis Martensen, Patrik
116 Flykt, Philip Lorenz, poma, Ray Strode, Robert Milasan, Scott
117 Thrasher, Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas
118 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut
119 Tezduyar Lindskog, WaLyong Cho, Will Woods, Zbigniew
122 -- Beijing, 2014-05-24
126 * When restoring the screen brightness at boot, stay away from
127 the darkest setting or from the lowest 5% of the available
128 range, depending on which is the larger value of both. This
129 should effectively protect the user from rebooting into a
130 black screen, should the brightness have been set to minimum
133 * sd-login gained a new sd_machine_get_class() call to
134 determine the class ("vm" or "container") of a machine
135 registered with machined.
137 * sd-login gained new calls
138 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
139 to query the identity of the peer of a local AF_UNIX
140 connection. They operate similarly to their sd_pid_get_xyz()
143 * PID 1 will now maintain a system-wide system state engine
144 with the states "starting", "running", "degraded",
145 "maintenance", "stopping". These states are bound to system
146 startup, normal runtime, runtime with at least one failed
147 service, rescue/emergency mode and system shutdown. This
148 state is shown in the "systemctl status" output when no unit
149 name is passed. It is useful to determine system state, in
150 particularly when doing so for many systems or containers at
153 * A new command "list-machines" has been added to "systemctl"
154 that lists all local OS containers and shows their system
155 state (see above), if systemd runs inside of them.
157 * systemctl gained a new "-r" switch to recursively enumerate
158 units on all local containers, when used with the
159 "list-unit" command (which is the default one that is
160 executed when no parameters are specified).
162 * The GPT automatic partition discovery logic will now honour
163 two GPT partition flags: one may be set on a partition to
164 cause it to be mounted read-only, and the other may be set
165 on a partition to ignore it during automatic discovery.
167 * Two new GPT type UUIDs have been added for automatic root
168 partition discovery, for 32-bit and 64-bit ARM. This is not
169 particularly useful for discovering the root directory on
170 these architectures during bare-metal boots (since UEFI is
171 not common there), but still very useful to allow booting of
172 ARM disk images in nspawn with the -i option.
174 * MAC addresses of interfaces created with nspawn's
175 --network-interface= switch will now be generated from the
176 machine name, and thus be stable between multiple invocations
179 * logind will now automatically remove all IPC objects owned
180 by a user if she or he fully logs out. This makes sure that
181 users who are logged out cannot continue to consume IPC
182 resources. This covers SysV memory, semaphores and message
183 queues as well as POSIX shared memory and message
184 queues. Traditionally, SysV and POSIX IPC had no life-cycle
185 limits. With this functionality, that is corrected. This may
186 be turned off by using the RemoveIPC= switch of logind.conf.
188 * The systemd-machine-id-setup and tmpfiles tools gained a
189 --root= switch to operate on a specific root directory,
192 * journald can now forward logged messages to the TTYs of all
193 logged in users ("wall"). This is the default for all
194 emergency messages now.
196 * A new tool systemd-journal-remote has been added to stream
197 journal log messages across the network.
199 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
200 controller trees are mounted into it. Note that the
201 directories mounted beneath it are not read-only. This is a
202 security measure and is particularly useful because glibc
203 actually includes a search logic to pick any tmpfs it can
204 find to implement shm_open() if /dev/shm is not available
205 (which it might very well be in namespaced setups).
207 * machinectl gained a new "poweroff" command to cleanly power
208 down a local OS container.
210 * The PrivateDevices= unit file setting will now also drop the
211 CAP_MKNOD capability from the capability bound set, and
212 imply DevicePolicy=closed.
214 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
215 comprehensively on all long-running systemd services where
218 * systemd-udevd will now run in a disassociated mount
219 namespace. To mount directories from udev rules, make sure to
220 pull in mount units via SYSTEMD_WANTS properties.
222 * The kdbus support gained support for uploading policy into
223 the kernel. sd-bus gained support for creating "monitoring"
224 connections that can eavesdrop into all bus communication
225 for debugging purposes.
227 * Timestamps may now be specified in seconds since the UNIX
228 epoch Jan 1st, 1970 by specifying "@" followed by the value
231 * Native tcpwrap support in systemd has been removed. tcpwrap
232 is old code, not really maintained anymore and has serious
233 shortcomings, and better options such as firewalls
234 exist. For setups that require tcpwrap usage, please
235 consider invoking your socket-activated service via tcpd,
236 like on traditional inetd.
238 * A new system.conf configuration option
239 DefaultTimerAccuracySec= has been added that controls the
240 default AccuracySec= setting of .timer units.
242 * Timer units gained a new WakeSystem= switch. If enabled,
243 timers configured this way will cause the system to resume
244 from system suspend (if the system supports that, which most
247 * Timer units gained a new Persistent= switch. If enabled,
248 timers configured this way will save to disk when they have
249 been last triggered. This information is then used on next
250 reboot to possible execute overdue timer events, that
251 could not take place because the system was powered off.
252 This enables simple anacron-like behaviour for timer units.
254 * systemctl's "list-timers" will now also list the time a
255 timer unit was last triggered in addition to the next time
256 it will be triggered.
258 * systemd-networkd will now assign predictable IPv4LL
259 addresses to its local interfaces.
261 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
262 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
263 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
264 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
265 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
266 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
267 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
268 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
271 -- Berlin, 2014-03-25
275 * A new unit file setting RestrictAddressFamilies= has been
276 added to restrict which socket address families unit
277 processes gain access to. This takes address family names
278 like "AF_INET" or "AF_UNIX", and is useful to minimize the
279 attack surface of services via exotic protocol stacks. This
280 is built on seccomp system call filters.
282 * Two new unit file settings RuntimeDirectory= and
283 RuntimeDirectoryMode= have been added that may be used to
284 manage a per-daemon runtime directories below /run. This is
285 an alternative for setting up directory permissions with
286 tmpfiles snippets, and has the advantage that the runtime
287 directory's lifetime is bound to the daemon runtime and that
288 the daemon starts up with an empty directory each time. This
289 is particularly useful when writing services that drop
290 privileges using the User= or Group= setting.
292 * The DeviceAllow= unit setting now supports globbing for
293 matching against device group names.
295 * The systemd configuration file system.conf gained new
296 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
297 DefaultMemoryAccounting= to globally turn on/off accounting
298 for specific resources (cgroups) for all units. These
299 settings may still be overridden individually in each unit
302 * systemd-gpt-auto-generator is now able to discover /srv and
303 root partitions in addition to /home and swap partitions. It
304 also supports LUKS-encrypted partitions now. With this in
305 place, automatic discovery of partitions to mount following
306 the Discoverable Partitions Specification
307 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
308 is now a lot more complete. This allows booting without
309 /etc/fstab and without root= on the kernel command line on
310 systems prepared appropriately.
312 * systemd-nspawn gained a new --image= switch which allows
313 booting up disk images and Linux installations on any block
314 device that follow the Discoverable Partitions Specification
315 (see above). This means that installations made with
316 appropriately updated installers may now be started and
317 deployed using container managers, completely
318 unmodified. (We hope that libvirt-lxc will add support for
319 this feature soon, too.)
321 * systemd-nspawn gained a new --network-macvlan= setting to
322 set up a private macvlan interface for the
323 container. Similarly, systemd-networkd gained a new
324 Kind=macvlan setting in .netdev files.
326 * systemd-networkd now supports configuring local addresses
329 * A new tool systemd-network-wait-online has been added to
330 synchronously wait for network connectivity using
333 * The sd-bus.h bus API gained a new sd_bus_track object for
334 tracking the life-cycle of bus peers. Note that sd-bus.h is
335 still not a public API though (unless you specify
336 --enable-kdbus on the configure command line, which however
337 voids your warranty and you get no API stability guarantee).
339 * The $XDG_RUNTIME_DIR runtime directories for each user are
340 now individual tmpfs instances, which has the benefit of
341 introducing separate pools for each user, with individual
342 size limits, and thus making sure that unprivileged clients
343 can no longer negatively impact the system or other users by
344 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
345 RuntimeDirectorySize= has been introduced that allows
346 controlling the default size limit for all users. It
347 defaults to 10% of the available physical memory. This is no
348 replacement for quotas on tmpfs though (which the kernel
349 still does not support), as /dev/shm and /tmp are still
350 shared resources used by both the system and unprivileged
353 * logind will now automatically turn off automatic suspending
354 on laptop lid close when more than one display is
355 connected. This was previously expected to be implemented
356 individually in desktop environments (such as GNOME),
357 however has been added to logind now, in order to fix a
358 boot-time race where a desktop environment might not have
359 been started yet and thus not been able to take an inhibitor
360 lock at the time where logind already suspends the system
363 * logind will now wait at least 30s after each system
364 suspend/resume cycle, and 3min after system boot before
365 suspending the system due to a closed laptop lid. This
366 should give USB docking stations and similar enough time to
367 be probed and configured after system resume and boot in
368 order to then act as suspend blocker.
370 * systemd-run gained a new --property= setting which allows
371 initialization of resource control properties (and others)
372 for the created scope or service unit. Example: "systemd-run
373 --property=BlockIOWeight=10 updatedb" may be used to run
374 updatedb at a low block IO scheduling weight.
376 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
377 now also work in --scope mode.
379 * When systemd is compiled with kdbus support, basic support
380 for enforced policies is now in place. (Note that enabling
381 kdbus still voids your warranty and no API compatibility
384 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
385 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
386 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
387 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
388 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
389 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
390 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
391 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
392 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
393 Zbigniew Jędrzejewski-Szmek
395 -- Berlin, 2014-03-12
399 * systemd will now relabel /dev after loading the SMACK policy
400 according to SMACK rules.
402 * A new unit file option AppArmorProfile= has been added to
403 set the AppArmor profile for the processes of a unit.
405 * A new condition check ConditionArchitecture= has been added
406 to conditionalize units based on the system architecture, as
407 reported by uname()'s "machine" field.
409 * systemd-networkd now supports matching on the system
410 virtualization, architecture, kernel command line, host name
413 * logind is now a lot more aggressive when suspending the
414 machine due to a closed laptop lid. Instead of acting only
415 on the lid close action, it will continuously watch the lid
416 status and act on it. This is useful for laptops where the
417 power button is on the outside of the chassis so that it can
418 be reached without opening the lid (such as the Lenovo
419 Yoga). On those machines, logind will now immediately
420 re-suspend the machine if the power button has been
421 accidentally pressed while the laptop was suspended and in a
424 * logind will now watch SW_DOCK switches and inhibit reaction
425 to the lid switch if it is pressed. This means that logind
426 will not suspend the machine anymore if the lid is closed
427 and the system is docked, if the laptop supports SW_DOCK
428 notifications via the input layer. Note that ACPI docking
429 stations do not generate this currently. Also note that this
430 logic is usually not fully sufficient and Desktop
431 Environments should take a lid switch inhibitor lock when an
432 external display is connected, as systemd will not watch
435 * nspawn will now make use of the devices cgroup controller by
436 default, and only permit creation of and access to the usual
437 API device nodes like /dev/null or /dev/random, as well as
438 access to (but not creation of) the pty devices.
440 * We will now ship a default .network file for
441 systemd-networkd that automatically configures DHCP for
442 network interfaces created by nspawn's --network-veth or
443 --network-bridge= switches.
445 * systemd will now understand the usual M, K, G, T suffixes
446 according to SI conventions (i.e. to the base 1000) when
447 referring to throughput and hardware metrics. It will stay
448 with IEC conventions (i.e. to the base 1024) for software
449 metrics, according to what is customary according to
450 Wikipedia. We explicitly document which base applies for
451 each configuration option.
453 * The DeviceAllow= setting in unit files now supports a syntax
454 to whitelist an entire group of devices node majors at once,
455 based on the /proc/devices listing. For example, with the
456 string "char-pts", it is now possible to whitelist all
457 current and future pseudo-TTYs at once.
459 * sd-event learned a new "post" event source. Event sources of
460 this type are triggered by the dispatching of any event
461 source of a type that is not "post". This is useful for
462 implementing clean-up and check event sources that are
463 triggered by other work being done in the program.
465 * systemd-networkd is no longer statically enabled, but uses
466 the usual [Install] sections so that it can be
467 enabled/disabled using systemctl. It still is enabled by
470 * When creating a veth interface pair with systemd-nspawn, the
471 host side will now be prefixed with "vb-" if
472 --network-bridge= is used, and with "ve-" if --network-veth
473 is used. This way, it is easy to distinguish these cases on
474 the host, for example to apply different configuration to
475 them with systemd-networkd.
477 * The compatibility libraries for libsystemd-journal.so,
478 libsystem-id128.so, libsystemd-login.so and
479 libsystemd-daemon.so do not make use of IFUNC
480 anymore. Instead, we now build libsystemd.so multiple times
481 under these alternative names. This means that the footprint
482 is drastically increased, but given that these are
483 transitional compatibility libraries, this should not matter
484 much. This change has been made necessary to support the ARM
485 platform for these compatibility libraries, as the ARM
486 toolchain is not really at the same level as the toolchain
487 for other architectures like x86 and does not support
488 IFUNC. Please make sure to use --enable-compat-libs only
489 during a transitional period!
491 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
492 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
493 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
494 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
495 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
496 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
497 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
498 Zbigniew Jędrzejewski-Szmek
500 -- Berlin, 2014-02-24
504 * A new component "systemd-networkd" has been added that can
505 be used to configure local network interfaces statically or
506 via DHCP. It is capable of bringing up bridges, VLANs, and
507 bonding. Currently, no hook-ups for interactive network
508 configuration are provided. Use this for your initrd,
509 container, embedded, or server setup if you need a simple,
510 yet powerful, network configuration solution. This
511 configuration subsystem is quite nifty, as it allows wildcard
512 hotplug matching in interfaces. For example, with a single
513 configuration snippet, you can configure that all Ethernet
514 interfaces showing up are automatically added to a bridge,
515 or similar. It supports link-sensing and more.
517 * A new tool "systemd-socket-proxyd" has been added which can
518 act as a bidirectional proxy for TCP sockets. This is
519 useful for adding socket activation support to services that
520 do not actually support socket activation, including virtual
521 machines and the like.
523 * Add a new tool to save/restore rfkill state on
526 * Save/restore state of keyboard backlights in addition to
527 display backlights on shutdown/boot.
529 * udev learned a new SECLABEL{} construct to label device
530 nodes with a specific security label when they appear. For
531 now, only SECLABEL{selinux} is supported, but the syntax is
532 prepared for additional security frameworks.
534 * udev gained a new scheme to configure link-level attributes
535 from files in /etc/systemd/network/*.link. These files can
536 match against MAC address, device path, driver name and type,
537 and will apply attributes like the naming policy, link speed,
538 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
539 address assignment policy (randomized, ...).
541 * The configuration of network interface naming rules for
542 "permanent interface names" has changed: a new NamePolicy=
543 setting in the [Link] section of .link files determines the
544 priority of possible naming schemes (onboard, slot, mac,
545 path). The default value of this setting is determined by
546 /usr/lib/net/links/99-default.link. Old
547 80-net-name-slot.rules udev configuration file has been
548 removed, so local configuration overriding this file should
549 be adapated to override 99-default.link instead.
551 * When the User= switch is used in a unit file, also
552 initialize $SHELL= based on the user database entry.
554 * systemd no longer depends on libdbus. All communication is
555 now done with sd-bus, systemd's low-level bus library
558 * kdbus support has been added to PID 1 itself. When kdbus is
559 enabled, this causes PID 1 to set up the system bus and
560 enable support for a new ".busname" unit type that
561 encapsulates bus name activation on kdbus. It works a little
562 bit like ".socket" units, except for bus names. A new
563 generator has been added that converts classic dbus1 service
564 activation files automatically into native systemd .busname
567 * sd-bus: add a light-weight vtable implementation that allows
568 defining objects on the bus with a simple static const
569 vtable array of its methods, signals and properties.
571 * systemd will not generate or install static dbus
572 introspection data anymore to /usr/share/dbus-1/interfaces,
573 as the precise format of these files is unclear, and
574 nothing makes use of it.
576 * A proxy daemon is now provided to proxy clients connecting
577 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
578 compatibility with classic D-Bus.
580 * A bus driver implementation has been added that supports the
581 classic D-Bus bus driver calls on kdbus, also for
582 compatibility purposes.
584 * A new API "sd-event.h" has been added that implements a
585 minimal event loop API built around epoll. It provides a
586 couple of features that direct epoll usage is lacking:
587 prioritization of events, scales to large numbers of timer
588 events, per-event timer slack (accuracy), system-wide
589 coalescing of timer events, exit handlers, watchdog
590 supervision support using systemd's sd_notify() API, child
593 * A new API "sd-rntl.h" has been added that provides an API
594 around the route netlink interface of the kernel, similar in
597 * A new API "sd-dhcp-client.h" has been added that provides a
598 small DHCPv4 client-side implementation. This is used by
601 * There is a new kernel command line option
602 "systemd.restore_state=0|1". When set to "0", none of the
603 systemd tools will restore saved runtime state to hardware
604 devices. More specifically, the rfkill and backlight states
607 * The FsckPassNo= compatibility option in mount/service units
608 has been removed. The fstab generator will now add the
609 necessary dependencies automatically, and does not require
610 PID1's support for that anymore.
612 * journalctl gained a new switch, --list-boots, that lists
613 recent boots with their times and boot IDs.
615 * The various tools like systemctl, loginctl, timedatectl,
616 busctl, systemd-run, ... have gained a new switch "-M" to
617 connect to a specific, local OS container (as direct
618 connection, without requiring SSH). This works on any
619 container that is registered with machined, such as those
620 created by libvirt-lxc or nspawn.
622 * systemd-run and systemd-analyze also gained support for "-H"
623 to connect to remote hosts via SSH. This is particularly
624 useful for systemd-run because it enables queuing of jobs
627 * machinectl gained a new command "login" to open a getty
628 login in any local container. This works with any container
629 that is registered with machined (such as those created by
630 libvirt-lxc or nspawn), and which runs systemd inside.
632 * machinectl gained a new "reboot" command that may be used to
633 trigger a reboot on a specific container that is registered
634 with machined. This works on any container that runs an init
637 * systemctl gained a new "list-timers" command to print a nice
638 listing of installed timer units with the times they elapse
641 * Alternative reboot() parameters may now be specified on the
642 "systemctl reboot" command line and are passed to the
643 reboot() system call.
645 * systemctl gained a new --job-mode= switch to configure the
646 mode to queue a job with. This is a more generic version of
647 --fail, --irreversible, and --ignore-dependencies, which are
648 still available but not advertised anymore.
650 * /etc/systemd/system.conf gained new settings to configure
651 various default timeouts of units, as well as the default
652 start limit interval and burst. These may still be overridden
655 * PID1 will now export on the bus profile data of the security
656 policy upload process (such as the SELinux policy upload to
659 * journald: when forwarding logs to the console, include
660 timestamps (following the setting in
661 /sys/module/printk/parameters/time).
663 * OnCalendar= in timer units now understands the special
664 strings "yearly" and "annually". (Both are equivalent)
666 * The accuracy of timer units is now configurable with the new
667 AccuracySec= setting. It defaults to 1min.
669 * A new dependency type JoinsNamespaceOf= has been added that
670 allows running two services within the same /tmp and network
671 namespace, if PrivateNetwork= or PrivateTmp= are used.
673 * A new command "cat" has been added to systemctl. It outputs
674 the original unit file of a unit, and concatenates the
675 contents of additional "drop-in" unit file snippets, so that
676 the full configuration is shown.
678 * systemctl now supports globbing on the various "list-xyz"
679 commands, like "list-units" or "list-sockets", as well as on
680 those commands which take multiple unit names.
682 * journalctl's --unit= switch gained support for globbing.
684 * All systemd daemons now make use of the watchdog logic so
685 that systemd automatically notices when they hang.
687 * If the $container_ttys environment variable is set,
688 getty-generator will automatically spawn a getty for each
689 listed tty. This is useful for container managers to request
690 login gettys to be spawned on as many ttys as needed.
692 * %h, %s, %U specifier support is not available anymore when
693 used in unit files for PID 1. This is because NSS calls are
694 not safe from PID 1. They stay available for --user
695 instances of systemd, and as special case for the root user.
697 * loginctl gained a new "--no-legend" switch to turn off output
700 * The "sd-login.h" API gained three new calls:
701 sd_session_is_remote(), sd_session_get_remote_user(),
702 sd_session_get_remote_host() to query information about
705 * The udev hardware database now also carries vendor/product
706 information of SDIO devices.
708 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
709 determine whether watchdog notifications are requested by
712 * Socket-activated per-connection services now include a
713 short description of the connection parameters in the
716 * tmpfiles gained a new "--boot" option. When this is not used,
717 only lines where the command character is not suffixed with
718 "!" are executed. When this option is specified, those
719 options are executed too. This partitions tmpfiles
720 directives into those that can be safely executed at any
721 time, and those which should be run only at boot (for
722 example, a line that creates /run/nologin).
724 * A new API "sd-resolve.h" has been added which provides a simple
725 asynchronous wrapper around glibc NSS host name resolution
726 calls, such as getaddrinfo(). In contrast to glibc's
727 getaddrinfo_a(), it does not use signals. In contrast to most
728 other asynchronous name resolution libraries, this one does
729 not reimplement DNS, but reuses NSS, so that alternate
730 host name resolution systems continue to work, such as mDNS,
731 LDAP, etc. This API is based on libasyncns, but it has been
732 cleaned up for inclusion in systemd.
734 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
735 "sd-daemon.h" are no longer found in individual libraries
736 libsystemd-journal.so, libsystemd-login.so,
737 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
738 merged them into a single library, libsystemd.so, which
739 provides all symbols. The reason for this is cyclic
740 dependencies, as these libraries tend to use each other's
741 symbols. So far, we have managed to workaround that by linking
742 a copy of a good part of our code into each of these
743 libraries again and again, which, however, makes certain
744 things hard to do, like sharing static variables. Also, it
745 substantially increases footprint. With this change, there
746 is only one library for the basic APIs systemd
747 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
748 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
749 library as well, however are subject to the --enable-kdbus
750 switch (see below). Note that "sd-dhcp-client.h" is not part
751 of this library (this is because it only consumes, never
752 provides, services of/to other APIs). To make the transition
753 easy from the separate libraries to the unified one, we
754 provide the --enable-compat-libs compile-time switch which
755 will generate stub libraries that are compatible with the
756 old ones but redirect all calls to the new one.
758 * All of the kdbus logic and the new APIs "sd-bus.h",
759 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
760 and "sd-utf8.h" are compile-time optional via the
761 "--enable-kdbus" switch, and they are not compiled in by
762 default. To make use of kdbus, you have to explicitly enable
763 the switch. Note however, that neither the kernel nor the
764 userspace API for all of this is considered stable yet. We
765 want to maintain the freedom to still change the APIs for
766 now. By specifying this build-time switch, you acknowledge
767 that you are aware of the instability of the current
770 * Also, note that while kdbus is pretty much complete,
771 it lacks one thing: proper policy support. This means you
772 can build a fully working system with all features; however,
773 it will be highly insecure. Policy support will be added in
774 one of the next releases, at the same time that we will
775 declare the APIs stable.
777 * When the kernel command-line argument "kdbus" is specified,
778 systemd will automatically load the kdbus.ko kernel module. At
779 this stage of development, it is only useful for testing kdbus
780 and should not be used in production. Note: if "--enable-kdbus"
781 is specified, and the kdbus.ko kernel module is available, and
782 "kdbus" is added to the kernel command line, the entire system
783 runs with kdbus instead of dbus-daemon, with the above mentioned
784 problem of missing the system policy enforcement. Also a future
785 version of kdbus.ko or a newer systemd will not be compatible with
786 each other, and will unlikely be able to boot the machine if only
787 one of them is updated.
789 * systemctl gained a new "import-environment" command which
790 uploads the caller's environment (or parts thereof) into the
791 service manager so that it is inherited by services started
792 by the manager. This is useful to upload variables like
793 $DISPLAY into the user service manager.
795 * A new PrivateDevices= switch has been added to service units
796 which allows running a service with a namespaced /dev
797 directory that does not contain any device nodes for
798 physical devices. More specifically, it only includes devices
799 such as /dev/null, /dev/urandom, and /dev/zero which are API
802 * logind has been extended to support behaviour like VT
803 switching on seats that do not support a VT. This makes
804 multi-session available on seats that are not the first seat
805 (seat0), and on systems where kernel support for VTs has
806 been disabled at compile-time.
808 * If a process holds a delay lock for system sleep or shutdown
809 and fails to release it in time, we will now log its
810 identity. This makes it easier to identify processes that
811 cause slow suspends or power-offs.
813 * When parsing /etc/crypttab, support for a new key-slot=
814 option as supported by Debian is added. It allows indicating
815 which LUKS slot to use on disk, speeding up key loading.
817 * The sd_journald_sendv() API call has been checked and
818 officially declared to be async-signal-safe so that it may
819 be invoked from signal handlers for logging purposes.
821 * Boot-time status output is now enabled automatically after a
822 short timeout if boot does not progress, in order to give
823 the user an indication what she or he is waiting for.
825 * The boot-time output has been improved to show how much time
826 remains until jobs expire.
828 * The KillMode= switch in service units gained a new possible
829 value "mixed". If set, and the unit is shut down, then the
830 initial SIGTERM signal is sent only to the main daemon
831 process, while the following SIGKILL signal is sent to
832 all remaining processes of the service.
834 * When a scope unit is registered, a new property "Controller"
835 may be set. If set to a valid bus name, systemd will send a
836 RequestStop() signal to this name when it would like to shut
837 down the scope. This may be used to hook manager logic into
838 the shutdown logic of scope units. Also, scope units may now
839 be put in a special "abandoned" state, in which case the
840 manager process which created them takes no further
841 responsibilities for it.
843 * When reading unit files, systemd will now verify
844 the access mode of these files, and warn about certain
845 suspicious combinations. This has been added to make it
846 easier to track down packaging bugs where unit files are
847 marked executable or world-writable.
849 * systemd-nspawn gained a new "--setenv=" switch to set
850 container-wide environment variables. The similar option in
851 systemd-activate was renamed from "--environment=" to
852 "--setenv=" for consistency.
854 * systemd-nspawn has been updated to create a new kdbus domain
855 for each container that is invoked, thus allowing each
856 container to have its own set of system and user buses,
857 independent of the host.
859 * systemd-nspawn gained a new --drop-capability= switch to run
860 the container with less capabilities than the default. Both
861 --drop-capability= and --capability= now take the special
862 string "all" for dropping or keeping all capabilities.
864 * systemd-nspawn gained new switches for executing containers
865 with specific SELinux labels set.
867 * systemd-nspawn gained a new --quiet switch to not generate
868 any additional output but the container's own console
871 * systemd-nspawn gained a new --share-system switch to run a
872 container without PID namespacing enabled.
874 * systemd-nspawn gained a new --register= switch to control
875 whether the container is registered with systemd-machined or
876 not. This is useful for containers that do not run full
877 OS images, but only specific apps.
879 * systemd-nspawn gained a new --keep-unit which may be used
880 when invoked as the only program from a service unit, and
881 results in registration of the unit service itself in
882 systemd-machined, instead of a newly opened scope unit.
884 * systemd-nspawn gained a new --network-interface= switch for
885 moving arbitrary interfaces to the container. The new
886 --network-veth switch creates a virtual Ethernet connection
887 between host and container. The new --network-bridge=
888 switch then allows assigning the host side of this virtual
889 Ethernet connection to a bridge device.
891 * systemd-nspawn gained a new --personality= switch for
892 setting the kernel personality for the container. This is
893 useful when running a 32-bit container on a 64-bit host. A
894 similar option Personality= is now also available for service
897 * logind will now also track a "Desktop" identifier for each
898 session which encodes the desktop environment of it. This is
899 useful for desktop environments that want to identify
900 multiple running sessions of itself easily.
902 * A new SELinuxContext= setting for service units has been
903 added that allows setting a specific SELinux execution
904 context for a service.
906 * Most systemd client tools will now honour $SYSTEMD_LESS for
907 settings of the "less" pager. By default, these tools will
908 override $LESS to allow certain operations to work, such as
909 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
910 influence this logic.
912 * systemd's "seccomp" hook-up has been changed to make use of
913 the libseccomp library instead of using its own
914 implementation. This has benefits for portability among
917 * For usage together with SystemCallFilter=, a new
918 SystemCallErrorNumber= setting has been introduced that
919 allows configuration of a system error number to be returned
920 on filtered system calls, instead of immediately killing the
921 process. Also, SystemCallArchitectures= has been added to
922 limit access to system calls of a particular architecture
923 (in order to turn off support for unused secondary
924 architectures). There is also a global
925 SystemCallArchitectures= setting in system.conf now to turn
926 off support for non-native system calls system-wide.
928 * systemd requires a kernel with a working name_to_handle_at(),
929 please see the kernel config requirements in the README file.
931 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
932 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
933 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
934 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
935 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
936 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
937 Elia Pinto, Florian Weimer, George McCollister, Goffredo
938 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
939 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
940 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
941 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
942 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
943 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
944 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
945 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
946 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
947 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
948 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
949 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
950 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
951 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
952 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
953 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
954 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
956 -- Berlin, 2014-02-20
960 * logind has gained support for facilitating privileged input
961 and drm device access for unprivileged clients. This work is
962 useful to allow Wayland display servers (and similar
963 programs, such as kmscon) to run under the user's ID and
964 access input and drm devices which are normally
965 protected. When this is used (and the kernel is new enough)
966 logind will "mute" IO on the file descriptors passed to
967 Wayland as long as it is in the background and "unmute" it
968 if it returns into the foreground. This allows secure
969 session switching without allowing background sessions to
970 eavesdrop on input and display data. This also introduces
971 session switching support if VT support is turned off in the
972 kernel, and on seats that are not seat0.
974 * A new kernel command line option luks.options= is understood
975 now which allows specifiying LUKS options for usage for LUKS
976 encrypted partitions specified with luks.uuid=.
978 * tmpfiles.d(5) snippets may now use specifier expansion in
979 path names. More specifically %m, %b, %H, %v, are now
980 replaced by the local machine id, boot id, hostname, and
981 kernel version number.
983 * A new tmpfiles.d(5) command "m" has been introduced which
984 may be used to change the owner/group/access mode of a file
985 or directory if it exists, but do nothing if it does not.
987 * This release removes high-level support for the
988 MemorySoftLimit= cgroup setting. The underlying kernel
989 cgroup attribute memory.soft_limit= is currently badly
990 designed and likely to be removed from the kernel API in its
991 current form, hence we should not expose it for now.
993 * The memory.use_hierarchy cgroup attribute is now enabled for
994 all cgroups systemd creates in the memory cgroup
995 hierarchy. This option is likely to be come the built-in
996 default in the kernel anyway, and the non-hierarchial mode
997 never made much sense in the intrinsically hierarchial
1000 * A new field _SYSTEMD_SLICE= is logged along with all journal
1001 messages containing the slice a message was generated
1002 from. This is useful to allow easy per-customer filtering of
1003 logs among other things.
1005 * systemd-journald will no longer adjust the group of journal
1006 files it creates to the "systemd-journal" group. Instead we
1007 rely on the journal directory to be owned by the
1008 "systemd-journal" group, and its setgid bit set, so that the
1009 kernel file system layer will automatically enforce that
1010 journal files inherit this group assignment. The reason for
1011 this change is that we cannot allow NSS look-ups from
1012 journald which would be necessary to resolve
1013 "systemd-journal" to a numeric GID, because this might
1014 create deadlocks if NSS involves synchronous queries to
1015 other daemons (such as nscd, or sssd) which in turn are
1016 logging clients of journald and might block on it, which
1017 would then dead lock. A tmpfiles.d(5) snippet included in
1018 systemd will make sure the setgid bit and group are
1019 properly set on the journal directory if it exists on every
1020 boot. However, we recommend adjusting it manually after
1021 upgrades too (or from RPM scriptlets), so that the change is
1022 not delayed until next reboot.
1024 * Backlight and random seed files in /var/lib/ have moved into
1025 the /var/lib/systemd/ directory, in order to centralize all
1026 systemd generated files in one directory.
1028 * Boot time performance measurements (as displayed by
1029 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
1030 performance information if that's available to determine how
1031 much time BIOS and boot loader initialization required. With
1032 a sufficiently new BIOS you hence no longer need to boot
1033 with Gummiboot to get access to such information.
1035 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
1036 Cristian Rodríguez, Dave Reisner, David Herrmann, David
1037 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
1038 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
1039 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
1040 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
1041 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1043 -- Berlin, 2013-10-02
1047 * The Restart= option for services now understands a new
1048 on-watchdog setting, which will restart the service
1049 automatically if the service stops sending out watchdog keep
1050 alive messages (as configured with WatchdogSec=).
1052 * The getty generator (which is responsible for bringing up a
1053 getty on configured serial consoles) will no longer only
1054 start a getty on the primary kernel console but on all
1055 others, too. This makes the order in which console= is
1056 specified on the kernel command line less important.
1058 * libsystemd-logind gained a new sd_session_get_vt() call to
1059 retrieve the VT number of a session.
1061 * If the option "tries=0" is set for an entry of /etc/crypttab
1062 its passphrase is queried indefinitely instead of any
1063 maximum number of tries.
1065 * If a service with a configure PID file terminates its PID
1066 file will now be removed automatically if it still exists
1067 afterwards. This should put an end to stale PID files.
1069 * systemd-run will now also take relative binary path names
1070 for execution and no longer insists on absolute paths.
1072 * InaccessibleDirectories= and ReadOnlyDirectories= now take
1073 paths that are optionally prefixed with "-" to indicate that
1074 it should not be considered a failure if they do not exist.
1076 * journalctl -o (and similar commands) now understands a new
1077 output mode "short-precise", it is similar to "short" but
1078 shows timestamps with usec accuracy.
1080 * The option "discard" (as known from Debian) is now
1081 synonymous to "allow-discards" in /etc/crypttab. In fact,
1082 "discard" is preferred now (since it is easier to remember
1085 * Some licensing clean-ups were made, so that more code is now
1086 LGPL-2.1 licensed than before.
1088 * A minimal tool to save/restore the display backlight
1089 brightness across reboots has been added. It will store the
1090 backlight setting as late as possible at shutdown, and
1091 restore it as early as possible during reboot.
1093 * A logic to automatically discover and enable home and swap
1094 partitions on GPT disks has been added. With this in place
1095 /etc/fstab becomes optional for many setups as systemd can
1096 discover certain partitions located on the root disk
1097 automatically. Home partitions are recognized under their
1098 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
1099 partitions are recognized under their GPT type ID
1100 0657fd6da4ab43c484e50933c84b4f4f.
1102 * systemd will no longer pass any environment from the kernel
1103 or initrd to system services. If you want to set an
1104 environment for all services, do so via the kernel command
1105 line systemd.setenv= assignment.
1107 * The systemd-sysctl tool no longer natively reads the file
1108 /etc/sysctl.conf. If desired, the file should be symlinked
1109 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
1110 legacy support by a symlink rather than built-in code, it
1111 also makes the otherwise hidden order of application of the
1112 different files visible. (Note that this partly reverts to a
1113 pre-198 application order of sysctl knobs!)
1115 * The "systemctl set-log-level" and "systemctl dump" commands
1116 have been moved to systemd-analyze.
1118 * systemd-run learned the new --remain-after-exit switch,
1119 which causes the scope unit not to be cleaned up
1120 automatically after the process terminated.
1122 * tmpfiles learned a new --exclude-prefix= switch to exclude
1123 certain paths from operation.
1125 * journald will now automatically flush all messages to disk
1126 as soon as a message of the log priorities CRIT, ALERT or
1129 Contributions from: Andrew Cook, Brandon Philips, Christian
1130 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
1131 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
1132 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
1133 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
1134 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
1135 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
1136 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
1137 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
1138 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
1139 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
1140 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
1141 William Giokas, Zbigniew Jędrzejewski-Szmek
1143 -- Berlin, 2013-09-13
1147 * The documentation has been updated to cover the various new
1148 concepts introduced with 205.
1150 * Unit files now understand the new %v specifier which
1151 resolves to the kernel version string as returned by "uname
1154 * systemctl now supports filtering the unit list output by
1155 load state, active state and sub state, using the new
1158 * "systemctl status" will now show the results of the
1159 condition checks (like ConditionPathExists= and similar) of
1160 the last start attempts of the unit. They are also logged to
1163 * "journalctl -b" may now be used to look for boot output of a
1164 specific boot. Try "journalctl -b -1" for the previous boot,
1165 but the syntax is substantially more powerful.
1167 * "journalctl --show-cursor" has been added which prints the
1168 cursor string the last shown log line. This may then be used
1169 with the new "journalctl --after-cursor=" switch to continue
1170 browsing logs from that point on.
1172 * "journalctl --force" may now be used to force regeneration
1175 * Creation of "dead" device nodes has been moved from udev
1176 into kmod and tmpfiles. Previously, udev would read the kmod
1177 databases to pre-generate dead device nodes based on meta
1178 information contained in kernel modules, so that these would
1179 be auto-loaded on access rather then at boot. As this
1180 does not really have much to do with the exposing actual
1181 kernel devices to userspace this has always been slightly
1182 alien in the udev codebase. Following the new scheme kmod
1183 will now generate a runtime snippet for tmpfiles from the
1184 module meta information and it now is tmpfiles' job to the
1185 create the nodes. This also allows overriding access and
1186 other parameters for the nodes using the usual tmpfiles
1187 facilities. As side effect this allows us to remove the
1188 CAP_SYS_MKNOD capability bit from udevd entirely.
1190 * logind's device ACLs may now be applied to these "dead"
1191 devices nodes too, thus finally allowing managed access to
1192 devices such as /dev/snd/sequencer whithout loading the
1193 backing module right-away.
1195 * A new RPM macro has been added that may be used to apply
1196 tmpfiles configuration during package installation.
1198 * systemd-detect-virt and ConditionVirtualization= now can
1199 detect User-Mode-Linux machines (UML).
1201 * journald will now implicitly log the effective capabilities
1202 set of processes in the message metadata.
1204 * systemd-cryptsetup has gained support for TrueCrypt volumes.
1206 * The initrd interface has been simplified (more specifically,
1207 support for passing performance data via environment
1208 variables and fsck results via files in /run has been
1209 removed). These features were non-essential, and are
1210 nowadays available in a much nicer way by having systemd in
1211 the initrd serialize its state and have the hosts systemd
1212 deserialize it again.
1214 * The udev "keymap" data files and tools to apply keyboard
1215 specific mappings of scan to key codes, and force-release
1216 scan code lists have been entirely replaced by a udev
1217 "keyboard" builtin and a hwdb data file.
1219 * systemd will now honour the kernel's "quiet" command line
1220 argument also during late shutdown, resulting in a
1221 completely silent shutdown when used.
1223 * There's now an option to control the SO_REUSEPORT socket
1224 option in .socket units.
1226 * Instance units will now automatically get a per-template
1227 subslice of system.slice unless something else is explicitly
1228 configured. For example, instances of sshd@.service will now
1229 implicitly be placed in system-sshd.slice rather than
1230 system.slice as before.
1232 * Test coverage support may now be enabled at build time.
1234 Contributions from: Dave Reisner, Frederic Crozat, Harald
1235 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
1236 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
1237 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
1238 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
1239 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
1240 Giokas, Zbigniew Jędrzejewski-Szmek
1242 -- Berlin, 2013-07-23
1246 * Two new unit types have been introduced:
1248 Scope units are very similar to service units, however, are
1249 created out of pre-existing processes -- instead of PID 1
1250 forking off the processes. By using scope units it is
1251 possible for system services and applications to group their
1252 own child processes (worker processes) in a powerful way
1253 which then maybe used to organize them, or kill them
1254 together, or apply resource limits on them.
1256 Slice units may be used to partition system resources in an
1257 hierarchial fashion and then assign other units to them. By
1258 default there are now three slices: system.slice (for all
1259 system services), user.slice (for all user sessions),
1260 machine.slice (for VMs and containers).
1262 Slices and scopes have been introduced primarily in
1263 context of the work to move cgroup handling to a
1264 single-writer scheme, where only PID 1
1265 creates/removes/manages cgroups.
1267 * There's a new concept of "transient" units. In contrast to
1268 normal units these units are created via an API at runtime,
1269 not from configuration from disk. More specifically this
1270 means it is now possible to run arbitrary programs as
1271 independent services, with all execution parameters passed
1272 in via bus APIs rather than read from disk. Transient units
1273 make systemd substantially more dynamic then it ever was,
1274 and useful as a general batch manager.
1276 * logind has been updated to make use of scope and slice units
1277 for managing user sessions. As a user logs in he will get
1278 his own private slice unit, to which all sessions are added
1279 as scope units. We also added support for automatically
1280 adding an instance of user@.service for the user into the
1281 slice. Effectively logind will no longer create cgroup
1282 hierarchies on its own now, it will defer entirely to PID 1
1283 for this by means of scope, service and slice units. Since
1284 user sessions this way become entities managed by PID 1
1285 the output of "systemctl" is now a lot more comprehensive.
1287 * A new mini-daemon "systemd-machined" has been added which
1288 may be used by virtualization managers to register local
1289 VMs/containers. nspawn has been updated accordingly, and
1290 libvirt will be updated shortly. machined will collect a bit
1291 of meta information about the VMs/containers, and assign
1292 them their own scope unit (see above). The collected
1293 meta-data is then made available via the "machinectl" tool,
1294 and exposed in "ps" and similar tools. machined/machinectl
1295 is compile-time optional.
1297 * As discussed earlier, the low-level cgroup configuration
1298 options ControlGroup=, ControlGroupModify=,
1299 ControlGroupPersistent=, ControlGroupAttribute= have been
1300 removed. Please use high-level attribute settings instead as
1301 well as slice units.
1303 * A new bus call SetUnitProperties() has been added to alter
1304 various runtime parameters of a unit. This is primarily
1305 useful to alter cgroup parameters dynamically in a nice way,
1306 but will be extended later on to make more properties
1307 modifiable at runtime. systemctl gained a new set-properties
1308 command that wraps this call.
1310 * A new tool "systemd-run" has been added which can be used to
1311 run arbitrary command lines as transient services or scopes,
1312 while configuring a number of settings via the command
1313 line. This tool is currently very basic, however already
1314 very useful. We plan to extend this tool to even allow
1315 queuing of execution jobs with time triggers from the
1316 command line, similar in fashion to "at".
1318 * nspawn will now inform the user explicitly that kernels with
1319 audit enabled break containers, and suggest the user to turn
1322 * Support for detecting the IMA and AppArmor security
1323 frameworks with ConditionSecurity= has been added.
1325 * journalctl gained a new "-k" switch for showing only kernel
1326 messages, mimicking dmesg output; in addition to "--user"
1327 and "--system" switches for showing only user's own logs
1330 * systemd-delta can now show information about drop-in
1331 snippets extending unit files.
1333 * libsystemd-bus has been substantially updated but is still
1334 not available as public API.
1336 * systemd will now look for the "debug" argument on the kernel
1337 command line and enable debug logging, similar to what
1338 "systemd.log_level=debug" already did before.
1340 * "systemctl set-default", "systemctl get-default" has been
1341 added to configure the default.target symlink, which
1342 controls what to boot into by default.
1344 * "systemctl set-log-level" has been added as a convenient
1345 way to raise and lower systemd logging threshold.
1347 * "systemd-analyze plot" will now show the time the various
1348 generators needed for execution, as well as information
1349 about the unit file loading.
1351 * libsystemd-journal gained a new sd_journal_open_files() call
1352 for opening specific journal files. journactl also gained a
1353 new switch to expose this new functionality. Previously we
1354 only supported opening all files from a directory, or all
1355 files from the system, as opening individual files only is
1356 racy due to journal file rotation.
1358 * systemd gained the new DefaultEnvironment= setting in
1359 /etc/systemd/system.conf to set environment variables for
1362 * If a privileged process logs a journal message with the
1363 OBJECT_PID= field set, then journald will automatically
1364 augment this with additional OBJECT_UID=, OBJECT_GID=,
1365 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
1366 system services want to log events about specific client
1367 processes. journactl/systemctl has been updated to make use
1368 of this information if all log messages regarding a specific
1371 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
1372 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
1373 Reisner, David Coppa, David King, David Strauss, Eelco
1374 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
1375 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
1376 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
1377 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
1378 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
1379 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
1380 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
1381 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
1382 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
1383 Łukasz Stelmach, 장동준
1387 * The Python bindings gained some minimal support for the APIs
1388 exposed by libsystemd-logind.
1390 * ConditionSecurity= gained support for detecting SMACK. Since
1391 this condition already supports SELinux and AppArmor we only
1392 miss IMA for this. Patches welcome!
1394 Contributions from: Karol Lewandowski, Lennart Poettering,
1395 Zbigniew Jędrzejewski-Szmek
1399 * systemd-nspawn will now create /etc/resolv.conf if
1400 necessary, before bind-mounting the host's file onto it.
1402 * systemd-nspawn will now store meta information about a
1403 container on the container's cgroup as extended attribute
1404 fields, including the root directory.
1406 * The cgroup hierarchy has been reworked in many ways. All
1407 objects any of the components systemd creates in the cgroup
1408 tree are now suffixed. More specifically, user sessions are
1409 now placed in cgroups suffixed with ".session", users in
1410 cgroups suffixed with ".user", and nspawn containers in
1411 cgroups suffixed with ".nspawn". Furthermore, all cgroup
1412 names are now escaped in a simple scheme to avoid collision
1413 of userspace object names with kernel filenames. This work
1414 is preparation for making these objects relocatable in the
1415 cgroup tree, in order to allow easy resource partitioning of
1416 these objects without causing naming conflicts.
1418 * systemctl list-dependencies gained the new switches
1419 --plain, --reverse, --after and --before.
1421 * systemd-inhibit now shows the process name of processes that
1422 have taken an inhibitor lock.
1424 * nss-myhostname will now also resolve "localhost"
1425 implicitly. This makes /etc/hosts an optional file and
1426 nicely handles that on IPv6 ::1 maps to both "localhost" and
1429 * libsystemd-logind.so gained a new call
1430 sd_get_machine_names() to enumerate running containers and
1431 VMs (currently only supported by very new libvirt and
1432 nspawn). sd_login_monitor can now be used to watch
1433 VMs/containers coming and going.
1435 * .include is not allowed recursively anymore, and only in
1436 unit files. Usually it is better to use drop-in snippets in
1437 .d/*.conf anyway, as introduced with systemd 198.
1439 * systemd-analyze gained a new "critical-chain" command that
1440 determines the slowest chain of units run during system
1441 boot-up. It is very useful for tracking down where
1442 optimizing boot time is the most beneficial.
1444 * systemd will no longer allow manipulating service paths in
1445 the name=systemd:/system cgroup tree using ControlGroup= in
1446 units. (But is still fine with it in all other dirs.)
1448 * There's a new systemd-nspawn@.service service file that may
1449 be used to easily run nspawn containers as system
1450 services. With the container's root directory in
1451 /var/lib/container/foobar it is now sufficient to run
1452 "systemctl start systemd-nspawn@foobar.service" to boot it.
1454 * systemd-cgls gained a new parameter "--machine" to list only
1455 the processes within a certain container.
1457 * ConditionSecurity= now can check for "apparmor". We still
1458 are lacking checks for SMACK and IMA for this condition
1459 check though. Patches welcome!
1461 * A new configuration file /etc/systemd/sleep.conf has been
1462 added that may be used to configure which kernel operation
1463 systemd is supposed to execute when "suspend", "hibernate"
1464 or "hybrid-sleep" is requested. This makes the new kernel
1465 "freeze" state accessible to the user.
1467 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
1468 the passed argument if applicable.
1470 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
1471 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
1472 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
1473 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
1474 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
1475 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
1476 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
1481 * The output of 'systemctl list-jobs' got some polishing. The
1482 '--type=' argument may now be passed more than once. A new
1483 command 'systemctl list-sockets' has been added which shows
1484 a list of kernel sockets systemd is listening on with the
1485 socket units they belong to, plus the units these socket
1488 * The experimental libsystemd-bus library got substantial
1489 updates to work in conjunction with the (also experimental)
1490 kdbus kernel project. It works well enough to exchange
1491 messages with some sophistication. Note that kdbus is not
1492 ready yet, and the library is mostly an elaborate test case
1493 for now, and not installable.
1495 * systemd gained a new unit 'systemd-static-nodes.service'
1496 that generates static device nodes earlier during boot, and
1497 can run in conjunction with udev.
1499 * libsystemd-login gained a new call sd_pid_get_user_unit()
1500 to retrieve the user systemd unit a process is running
1501 in. This is useful for systems where systemd is used as
1504 * systemd-nspawn now places all containers in the new /machine
1505 top-level cgroup directory in the name=systemd
1506 hierarchy. libvirt will soon do the same, so that we get a
1507 uniform separation of /system, /user and /machine for system
1508 services, user processes and containers/virtual
1509 machines. This new cgroup hierarchy is also useful to stick
1510 stable names to specific container instances, which can be
1511 recognized later this way (this name may be controlled
1512 via systemd-nspawn's new -M switch). libsystemd-login also
1513 gained a new call sd_pid_get_machine_name() to retrieve the
1514 name of the container/VM a specific process belongs to.
1516 * bootchart can now store its data in the journal.
1518 * libsystemd-journal gained a new call
1519 sd_journal_add_conjunction() for AND expressions to the
1520 matching logic. This can be used to express more complex
1521 logical expressions.
1523 * journactl can now take multiple --unit= and --user-unit=
1526 * The cryptsetup logic now understands the "luks.key=" kernel
1527 command line switch for specifying a file to read the
1528 decryption key from. Also, if a configured key file is not
1529 found the tool will now automatically fall back to prompting
1532 * Python systemd.journal module was updated to wrap recently
1533 added functions from libsystemd-journal. The interface was
1534 changed to bring the low level interface in s.j._Reader
1535 closer to the C API, and the high level interface in
1536 s.j.Reader was updated to wrap and convert all data about
1539 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
1540 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
1541 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
1542 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
1543 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
1544 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1548 * journalctl --update-catalog now understands a new --root=
1549 option to operate on catalogs found in a different root
1552 * During shutdown after systemd has terminated all running
1553 services a final killing loop kills all remaining left-over
1554 processes. We will now print the name of these processes
1555 when we send SIGKILL to them, since this usually indicates a
1558 * If /etc/crypttab refers to password files stored on
1559 configured mount points automatic dependencies will now be
1560 generated to ensure the specific mount is established first
1561 before the key file is attempted to be read.
1563 * 'systemctl status' will now show information about the
1564 network sockets a socket unit is listening on.
1566 * 'systemctl status' will also shown information about any
1567 drop-in configuration file for units. (Drop-In configuration
1568 files in this context are files such as
1569 /etc/systemd/systemd/foobar.service.d/*.conf)
1571 * systemd-cgtop now optionally shows summed up CPU times of
1572 cgroups. Press '%' while running cgtop to switch between
1573 percentage and absolute mode. This is useful to determine
1574 which cgroups use up the most CPU time over the entire
1575 runtime of the system. systemd-cgtop has also been updated
1576 to be 'pipeable' for processing with further shell tools.
1578 * 'hostnamectl set-hostname' will now allow setting of FQDN
1581 * The formatting and parsing of time span values has been
1582 changed. The parser now understands fractional expressions
1583 such as "5.5h". The formatter will now output fractional
1584 expressions for all time spans under 1min, i.e. "5.123456s"
1585 rather than "5s 123ms 456us". For time spans under 1s
1586 millisecond values are shown, for those under 1ms
1587 microsecond values are shown. This should greatly improve
1588 all time-related output of systemd.
1590 * libsystemd-login and libsystemd-journal gained new
1591 functions for querying the poll() events mask and poll()
1592 timeout value for integration into arbitrary event
1595 * localectl gained the ability to list available X11 keymaps
1596 (models, layouts, variants, options).
1598 * 'systemd-analyze dot' gained the ability to filter for
1599 specific units via shell-style globs, to create smaller,
1600 more useful graphs. I.e. it is now possible to create simple
1601 graphs of all the dependencies between only target units, or
1602 of all units that Avahi has dependencies with.
1604 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
1605 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
1606 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
1607 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
1608 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
1609 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
1610 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
1614 * The boot-time readahead implementation for rotating media
1615 will now read the read-ahead data in multiple passes which
1616 consist of all read requests made in equidistant time
1617 intervals. This means instead of strictly reading read-ahead
1618 data in its physical order on disk we now try to find a
1619 middle ground between physical and access time order.
1621 * /etc/os-release files gained a new BUILD_ID= field for usage
1622 on operating systems that provide continuous builds of OS
1625 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
1626 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
1627 William Douglas, Zbigniew Jędrzejewski-Szmek
1631 * systemd-python gained an API exposing libsystemd-daemon.
1633 * The SMACK setup logic gained support for uploading CIPSO
1636 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
1637 ReadOnlyDirectories= and InaccessibleDirectories= has
1638 changed. The private /tmp and /var/tmp directories are now
1639 shared by all processes of a service (which means
1640 ExecStartPre= may now leave data in /tmp that ExecStart= of
1641 the same service can still access). When a service is
1642 stopped its temporary directories are immediately deleted
1643 (normal clean-up with tmpfiles is still done in addition to
1646 * By default, systemd will now set a couple of sysctl
1647 variables in the kernel: the safe sysrq options are turned
1648 on, IP route verification is turned on, and source routing
1649 disabled. The recently added hardlink and softlink
1650 protection of the kernel is turned on. These settings should
1651 be reasonably safe, and good defaults for all new systems.
1653 * The predictable network naming logic may now be turned off
1654 with a new kernel command line switch: net.ifnames=0.
1656 * A new libsystemd-bus module has been added that implements a
1657 pretty complete D-Bus client library. For details see:
1659 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
1661 * journald will now explicitly flush the journal files to disk
1662 at the latest 5min after each write. The file will then also
1663 be marked offline until the next write. This should increase
1664 reliability in case of a crash. The synchronization delay
1665 can be configured via SyncIntervalSec= in journald.conf.
1667 * There's a new remote-fs-setup.target unit that can be used
1668 to pull in specific services when at least one remote file
1669 system is to be mounted.
1671 * There are new targets timers.target and paths.target as
1672 canonical targets to pull user timer and path units in
1673 from. This complements sockets.target with a similar
1674 purpose for socket units.
1676 * libudev gained a new call udev_device_set_attribute_value()
1677 to set sysfs attributes of a device.
1679 * The udev daemon now sets the default number of worker
1680 processes executed in parallel based on the number of available
1681 CPUs instead of the amount of available RAM. This is supposed
1682 to provide a more reliable default and limit a too aggressive
1683 paralellism for setups with 1000s of devices connected.
1685 Contributions from: Auke Kok, Colin Walters, Cristian
1686 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
1687 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
1688 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
1689 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
1690 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
1691 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
1692 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
1693 Zbigniew Jędrzejewski-Szmek
1697 * Configuration of unit files may now be extended via drop-in
1698 files without having to edit/override the unit files
1699 themselves. More specifically, if the administrator wants to
1700 change one value for a service file foobar.service he can
1701 now do so by dropping in a configuration snippet into
1702 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
1703 will load all these snippets and apply them on top of the
1704 main unit configuration file, possibly extending or
1705 overriding its settings. Using these drop-in snippets is
1706 generally nicer than the two earlier options for changing
1707 unit files locally: copying the files from
1708 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
1709 them there; or creating a new file in /etc/systemd/system/
1710 that incorporates the original one via ".include". Drop-in
1711 snippets into these .d/ directories can be placed in any
1712 directory systemd looks for units in, and the usual
1713 overriding semantics between /usr/lib, /etc and /run apply
1716 * Most unit file settings which take lists of items can now be
1717 reset by assigning the empty string to them. For example,
1718 normally, settings such as Environment=FOO=BAR append a new
1719 environment variable assignment to the environment block,
1720 each time they are used. By assigning Environment= the empty
1721 string the environment block can be reset to empty. This is
1722 particularly useful with the .d/*.conf drop-in snippets
1723 mentioned above, since this adds the ability to reset list
1724 settings from vendor unit files via these drop-ins.
1726 * systemctl gained a new "list-dependencies" command for
1727 listing the dependencies of a unit recursively.
1729 * Inhibitors are now honored and listed by "systemctl
1730 suspend", "systemctl poweroff" (and similar) too, not only
1731 GNOME. These commands will also list active sessions by
1734 * Resource limits (as exposed by the various control group
1735 controllers) can now be controlled dynamically at runtime
1736 for all units. More specifically, you can now use a command
1737 like "systemctl set-cgroup-attr foobar.service cpu.shares
1738 2000" to alter the CPU shares a specific service gets. These
1739 settings are stored persistently on disk, and thus allow the
1740 administrator to easily adjust the resource usage of
1741 services with a few simple commands. This dynamic resource
1742 management logic is also available to other programs via the
1743 bus. Almost any kernel cgroup attribute and controller is
1746 * systemd-vconsole-setup will now copy all font settings to
1747 all allocated VTs, where it previously applied them only to
1750 * libsystemd-login gained the new sd_session_get_tty() API
1753 * This release drops support for a few legacy or
1754 distribution-specific LSB facility names when parsing init
1755 scripts: $x-display-manager, $mail-transfer-agent,
1756 $mail-transport-agent, $mail-transfer-agent, $smtp,
1757 $null. Also, the mail-transfer-agent.target unit backing
1758 this has been removed. Distributions which want to retain
1759 compatibility with this should carry the burden for
1760 supporting this themselves and patch support for these back
1761 in, if they really need to. Also, the facilities $syslog and
1762 $local_fs are now ignored, since systemd does not support
1763 early-boot LSB init scripts anymore, and these facilities
1764 are implied anyway for normal services. syslog.target has
1767 * There are new bus calls on PID1's Manager object for
1768 cancelling jobs, and removing snapshot units. Previously,
1769 both calls were only available on the Job and Snapshot
1772 * systemd-journal-gatewayd gained SSL support.
1774 * The various "environment" files, such as /etc/locale.conf
1775 now support continuation lines with a backslash ("\") as
1776 last character in the line, similarly in style (but different)
1777 to how this is supported in shells.
1779 * For normal user processes the _SYSTEMD_USER_UNIT= field is
1780 now implicitly appended to every log entry logged. systemctl
1781 has been updated to filter by this field when operating on a
1782 user systemd instance.
1784 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
1785 CAP_AUDIT_CONTROL capabilities to the capabilities set for
1786 the container. This makes it easier to boot unmodified
1787 Fedora systems in a container, which however still requires
1788 audit=0 to be passed on the kernel command line. Auditing in
1789 kernel and userspace is unfortunately still too broken in
1790 context of containers, hence we recommend compiling it out
1791 of the kernel or using audit=0. Hopefully this will be fixed
1792 one day for good in the kernel.
1794 * nspawn gained the new --bind= and --bind-ro= parameters to
1795 bind mount specific directories from the host into the
1798 * nspawn will now mount its own devpts file system instance
1799 into the container, in order not to leak pty devices from
1800 the host into the container.
1802 * systemd will now read the firmware boot time performance
1803 information from the EFI variables, if the used boot loader
1804 supports this, and takes it into account for boot performance
1805 analysis via "systemd-analyze". This is currently supported
1806 only in conjunction with Gummiboot, but could be supported
1807 by other boot loaders too. For details see:
1809 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
1811 * A new generator has been added that automatically mounts the
1812 EFI System Partition (ESP) to /boot, if that directory
1813 exists, is empty, and no other file system has been
1814 configured to be mounted there.
1816 * logind will now send out PrepareForSleep(false) out
1817 unconditionally, after coming back from suspend. This may be
1818 used by applications as asynchronous notification for
1819 system resume events.
1821 * "systemctl unlock-sessions" has been added, that allows
1822 unlocking the screens of all user sessions at once, similar
1823 to how "systemctl lock-sessions" already locked all users
1824 sessions. This is backed by a new D-Bus call UnlockSessions().
1826 * "loginctl seat-status" will now show the master device of a
1827 seat. (i.e. the device of a seat that needs to be around for
1828 the seat to be considered available, usually the graphics
1831 * tmpfiles gained a new "X" line type, that allows
1832 configuration of files and directories (with wildcards) that
1833 shall be excluded from automatic cleanup ("aging").
1835 * udev default rules set the device node permissions now only
1836 at "add" events, and do not change them any longer with a
1837 later "change" event.
1839 * The log messages for lid events and power/sleep keypresses
1840 now carry a message ID.
1842 * We now have a substantially larger unit test suite, but this
1843 continues to be work in progress.
1845 * udevadm hwdb gained a new --root= parameter to change the
1846 root directory to operate relative to.
1848 * logind will now issue a background sync() request to the kernel
1849 early at shutdown, so that dirty buffers are flushed to disk early
1850 instead of at the last moment, in order to optimize shutdown
1853 * A new bootctl tool has been added that is an interface for
1854 certain boot loader operations. This is currently a preview
1855 and is likely to be extended into a small mechanism daemon
1856 like timedated, localed, hostnamed, and can be used by
1857 graphical UIs to enumerate available boot options, and
1858 request boot into firmware operations.
1860 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
1861 the rest of the package. It also has been updated to work
1862 correctly in initrds.
1864 * Policykit previously has been runtime optional, and is now
1865 also compile time optional via a configure switch.
1867 * systemd-analyze has been reimplemented in C. Also "systemctl
1868 dot" has moved into systemd-analyze.
1870 * "systemctl status" with no further parameters will now print
1871 the status of all active or failed units.
1873 * Operations such as "systemctl start" can now be executed
1874 with a new mode "--irreversible" which may be used to queue
1875 operations that cannot accidentally be reversed by a later
1876 job queuing. This is by default used to make shutdown
1877 requests more robust.
1879 * The Python API of systemd now gained a new module for
1880 reading journal files.
1882 * A new tool kernel-install has been added that can install
1883 kernel images according to the Boot Loader Specification:
1885 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
1887 * Boot time console output has been improved to provide
1888 animated boot time output for hanging jobs.
1890 * A new tool systemd-activate has been added which can be used
1891 to test socket activation with, directly from the command
1892 line. This should make it much easier to test and debug
1893 socket activation in daemons.
1895 * journalctl gained a new "--reverse" (or -r) option to show
1896 journal output in reverse order (i.e. newest line first).
1898 * journalctl gained a new "--pager-end" (or -e) option to jump
1899 to immediately jump to the end of the journal in the
1900 pager. This is only supported in conjunction with "less".
1902 * journalctl gained a new "--user-unit=" option, that works
1903 similarly to "--unit=" but filters for user units rather than
1906 * A number of unit files to ease adoption of systemd in
1907 initrds has been added. This moves some minimal logic from
1908 the various initrd implementations into systemd proper.
1910 * The journal files are now owned by a new group
1911 "systemd-journal", which exists specifically to allow access
1912 to the journal, and nothing else. Previously, we used the
1913 "adm" group for that, which however possibly covers more
1914 than just journal/log file access. This new group is now
1915 already used by systemd-journal-gatewayd to ensure this
1916 daemon gets access to the journal files and as little else
1917 as possible. Note that "make install" will also set FS ACLs
1918 up for /var/log/journal to give "adm" and "wheel" read
1919 access to it, in addition to "systemd-journal" which owns
1920 the journal files. We recommend that packaging scripts also
1921 add read access to "adm" + "wheel" to /var/log/journal, and
1922 all existing/future journal files. To normal users and
1923 administrators little changes, however packagers need to
1924 ensure to create the "systemd-journal" system group at
1925 package installation time.
1927 * The systemd-journal-gatewayd now runs as unprivileged user
1928 systemd-journal-gateway:systemd-journal-gateway. Packaging
1929 scripts need to create these system user/group at
1932 * timedated now exposes a new boolean property CanNTP that
1933 indicates whether a local NTP service is available or not.
1935 * systemd-detect-virt will now also detect xen PVs
1937 * The pstore file system is now mounted by default, if it is
1940 * In addition to the SELinux and IMA policies we will now also
1941 load SMACK policies at early boot.
1943 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
1944 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
1945 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
1946 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
1947 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
1948 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
1949 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
1950 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
1951 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
1952 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
1953 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
1954 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
1955 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
1956 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
1960 * Timer units now support calendar time events in addition to
1961 monotonic time events. That means you can now trigger a unit
1962 based on a calendar time specification such as "Thu,Fri
1963 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
1964 or fifth day of any month of the year 2013, given that it is
1965 a thursday or friday. This brings timer event support
1966 considerably closer to cron's capabilities. For details on
1967 the supported calendar time specification language see
1970 * udev now supports a number of different naming policies for
1971 network interfaces for predictable names, and a combination
1972 of these policies is now the default. Please see this wiki
1973 document for details:
1975 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
1977 * Auke Kok's bootchart implementation has been added to the
1978 systemd tree. It is an optional component that can graph the
1979 boot in quite some detail. It is one of the best bootchart
1980 implementations around and minimal in its code and
1983 * nss-myhostname has been integrated into the systemd source
1984 tree. nss-myhostname guarantees that the local hostname
1985 always stays resolvable via NSS. It has been a weak
1986 requirement of systemd-hostnamed since a long time, and
1987 since its code is actually trivial we decided to just
1988 include it in systemd's source tree. It can be turned off
1989 with a configure switch.
1991 * The read-ahead logic is now capable of properly detecting
1992 whether a btrfs file system is on SSD or rotating media, in
1993 order to optimize the read-ahead scheme. Previously, it was
1994 only capable of detecting this on traditional file systems
1997 * In udev, additional device properties are now read from the
1998 IAB in addition to the OUI database. Also, Bluetooth company
1999 identities are attached to the devices as well.
2001 * In service files %U may be used as specifier that is
2002 replaced by the configured user name of the service.
2004 * nspawn may now be invoked without a controlling TTY. This
2005 makes it suitable for invocation as its own service. This
2006 may be used to set up a simple containerized server system
2007 using only core OS tools.
2009 * systemd and nspawn can now accept socket file descriptors
2010 when they are started for socket activation. This enables
2011 implementation of socket activated nspawn
2012 containers. i.e. think about autospawning an entire OS image
2013 when the first SSH or HTTP connection is received. We expect
2014 that similar functionality will also be added to libvirt-lxc
2017 * journalctl will now suppress ANSI color codes when
2018 presenting log data.
2020 * systemctl will no longer show control group information for
2021 a unit if a the control group is empty anyway.
2023 * logind can now automatically suspend/hibernate/shutdown the
2026 * /etc/machine-info and hostnamed now also expose the chassis
2027 type of the system. This can be used to determine whether
2028 the local system is a laptop, desktop, handset or
2029 tablet. This information may either be configured by the
2030 user/vendor or is automatically determined from ACPI and DMI
2031 information if possible.
2033 * A number of PolicyKit actions are now bound together with
2034 "imply" rules. This should simplify creating UIs because
2035 many actions will now authenticate similar ones as well.
2037 * Unit files learnt a new condition ConditionACPower= which
2038 may be used to conditionalize a unit depending on whether an
2039 AC power source is connected or not, of whether the system
2040 is running on battery power.
2042 * systemctl gained a new "is-failed" verb that may be used in
2043 shell scripts and suchlike to check whether a specific unit
2044 is in the "failed" state.
2046 * The EnvironmentFile= setting in unit files now supports file
2047 globbing, and can hence be used to easily read a number of
2048 environment files at once.
2050 * systemd will no longer detect and recognize specific
2051 distributions. All distribution-specific #ifdeffery has been
2052 removed, systemd is now fully generic and
2053 distribution-agnostic. Effectively, not too much is lost as
2054 a lot of the code is still accessible via explicit configure
2055 switches. However, support for some distribution specific
2056 legacy configuration file formats has been dropped. We
2057 recommend distributions to simply adopt the configuration
2058 files everybody else uses now and convert the old
2059 configuration from packaging scripts. Most distributions
2060 already did that. If that's not possible or desirable,
2061 distributions are welcome to forward port the specific
2062 pieces of code locally from the git history.
2064 * When logging a message about a unit systemd will now always
2065 log the unit name in the message meta data.
2067 * localectl will now also discover system locale data that is
2068 not stored in locale archives, but directly unpacked.
2070 * logind will no longer unconditionally use framebuffer
2071 devices as seat masters, i.e. as devices that are required
2072 to be existing before a seat is considered preset. Instead,
2073 it will now look for all devices that are tagged as
2074 "seat-master" in udev. By default framebuffer devices will
2075 be marked as such, but depending on local systems other
2076 devices might be marked as well. This may be used to
2077 integrate graphics cards using closed source drivers (such
2078 as NVidia ones) more nicely into logind. Note however, that
2079 we recommend using the open source NVidia drivers instead,
2080 and no udev rules for the closed-source drivers will be
2081 shipped from us upstream.
2083 Contributions from: Adam Williamson, Alessandro Crismani, Auke
2084 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
2085 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
2086 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
2087 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
2088 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
2089 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
2090 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
2091 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
2092 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
2093 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
2094 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
2099 * udev gained support for loading additional device properties
2100 from an indexed database that is keyed by vendor/product IDs
2101 and similar device identifiers. For the beginning this
2102 "hwdb" is populated with data from the well-known PCI and
2103 USB database, but also includes PNP, ACPI and OID data. In
2104 the longer run this indexed database shall grow into
2105 becoming the one central database for non-essential
2106 userspace device metadata. Previously, data from the PCI/USB
2107 database was only attached to select devices, since the
2108 lookup was a relatively expensive operation due to O(n) time
2109 complexity (with n being the number of entries in the
2110 database). Since this is now O(1), we decided to add in this
2111 data for all devices where this is available, by
2112 default. Note that the indexed database needs to be rebuilt
2113 when new data files are installed. To achieve this you need
2114 to update your packaging scripts to invoke "udevadm hwdb
2115 --update" after installation of hwdb data files. For
2116 RPM-based distributions we introduced the new
2117 %udev_hwdb_update macro for this purpose.
2119 * The Journal gained support for the "Message Catalog", an
2120 indexed database to link up additional information with
2121 journal entries. For further details please check:
2123 http://www.freedesktop.org/wiki/Software/systemd/catalog
2125 The indexed message catalog database also needs to be
2126 rebuilt after installation of message catalog files. Use
2127 "journalctl --update-catalog" for this. For RPM-based
2128 distributions we introduced the %journal_catalog_update
2129 macro for this purpose.
2131 * The Python Journal bindings gained support for the standard
2132 Python logging framework.
2134 * The Journal API gained new functions for checking whether
2135 the underlying file system of a journal file is capable of
2136 properly reporting file change notifications, or whether
2137 applications that want to reflect journal changes "live"
2138 need to recheck journal files continuously in appropriate
2141 * It is now possible to set the "age" field for tmpfiles
2142 entries to 0, indicating that files matching this entry
2143 shall always be removed when the directories are cleaned up.
2145 * coredumpctl gained a new "gdb" verb which invokes gdb
2146 right-away on the selected coredump.
2148 * There's now support for "hybrid sleep" on kernels that
2149 support this, in addition to "suspend" and "hibernate". Use
2150 "systemctl hybrid-sleep" to make use of this.
2152 * logind's HandleSuspendKey= setting (and related settings)
2153 now gained support for a new "lock" setting to simply
2154 request the screen lock on all local sessions, instead of
2155 actually executing a suspend or hibernation.
2157 * systemd will now mount the EFI variables file system by
2160 * Socket units now gained support for configuration of the
2161 SMACK security label.
2163 * timedatectl will now output the time of the last and next
2164 daylight saving change.
2166 * We dropped support for various legacy and distro-specific
2167 concepts, such as insserv, early-boot SysV services
2168 (i.e. those for non-standard runlevels such as 'b' or 'S')
2169 or ArchLinux /etc/rc.conf support. We recommend the
2170 distributions who still need support this to either continue
2171 to maintain the necessary patches downstream, or find a
2172 different solution. (Talk to us if you have questions!)
2174 * Various systemd components will now bypass PolicyKit checks
2175 for root and otherwise handle properly if PolicyKit is not
2176 found to be around. This should fix most issues for
2177 PolicyKit-less systems. Quite frankly this should have been
2178 this way since day one. It is absolutely our intention to
2179 make systemd work fine on PolicyKit-less systems, and we
2180 consider it a bug if something does not work as it should if
2181 PolicyKit is not around.
2183 * For embedded systems it is now possible to build udev and
2184 systemd without blkid and/or kmod support.
2186 * "systemctl switch-root" is now capable of switching root
2187 more than once. I.e. in addition to transitions from the
2188 initrd to the host OS it is now possible to transition to
2189 further OS images from the host. This is useful to implement
2190 offline updating tools.
2192 * Various other additions have been made to the RPM macros
2193 shipped with systemd. Use %udev_rules_update() after
2194 installing new udev rules files. %_udevhwdbdir,
2195 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
2196 %_sysctldir are now available which resolve to the right
2197 directories for packages to place various data files in.
2199 * journalctl gained the new --full switch (in addition to
2200 --all, to disable ellipsation for long messages.
2202 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
2203 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
2204 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
2205 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
2206 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
2207 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
2208 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
2209 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
2210 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
2214 * journalctl gained new --since= and --until= switches to
2215 filter by time. It also now supports nice filtering for
2216 units via --unit=/-u.
2218 * Type=oneshot services may use ExecReload= and do the
2221 * The journal daemon now supports time-based rotation and
2222 vacuuming, in addition to the usual disk-space based
2225 * The journal will now index the available field values for
2226 each field name. This enables clients to show pretty drop
2227 downs of available match values when filtering. The bash
2228 completion of journalctl has been updated
2229 accordingly. journalctl gained a new switch -F to list all
2230 values a certain field takes in the journal database.
2232 * More service events are now written as structured messages
2233 to the journal, and made recognizable via message IDs.
2235 * The timedated, localed and hostnamed mini-services which
2236 previously only provided support for changing time, locale
2237 and hostname settings from graphical DEs such as GNOME now
2238 also have a minimal (but very useful) text-based client
2239 utility each. This is probably the nicest way to changing
2240 these settings from the command line now, especially since
2241 it lists available options and is fully integrated with bash
2244 * There's now a new tool "systemd-coredumpctl" to list and
2245 extract coredumps from the journal.
2247 * We now install a README each in /var/log/ and
2248 /etc/rc.d/init.d explaining where the system logs and init
2249 scripts went. This hopefully should help folks who go to
2250 that dirs and look into the otherwise now empty void and
2251 scratch their heads.
2253 * When user-services are invoked (by systemd --user) the
2254 $MANAGERPID env var is set to the PID of systemd.
2256 * SIGRTMIN+24 when sent to a --user instance will now result
2257 in immediate termination of systemd.
2259 * gatewayd received numerous feature additions such as a
2260 "follow" mode, for live syncing and filtering.
2262 * browse.html now allows filtering and showing detailed
2263 information on specific entries. Keyboard navigation and
2264 mouse screen support has been added.
2266 * gatewayd/journalctl now supports HTML5/JSON
2267 Server-Sent-Events as output.
2269 * The SysV init script compatibility logic will now
2270 heuristically determine whether a script supports the
2271 "reload" verb, and only then make this available as
2274 * "systemctl status --follow" has been removed, use "journalctl
2277 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
2278 have been removed since they are hardly useful to be
2281 * And I'd like to take the opportunity to specifically mention
2282 Zbigniew for his great contributions. Zbigniew, you rock!
2284 Contributions from: Andrew Eikum, Christian Hesse, Colin
2285 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
2286 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
2287 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
2288 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
2289 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
2290 Jędrzejewski-Szmek, Сковорода Никита Андреевич
2294 * If /etc/vconsole.conf is non-existent or empty we will no
2295 longer load any console font or key map at boot by
2296 default. Instead the kernel defaults will be left
2297 intact. This is definitely the right thing to do, as no
2298 configuration should mean no configuration, and hard-coding
2299 font names that are different on all archs is probably a bad
2300 idea. Also, the kernel default key map and font should be
2301 good enough for most cases anyway, and mostly identical to
2302 the userspace fonts/key maps we previously overloaded them
2303 with. If distributions want to continue to default to a
2304 non-kernel font or key map they should ship a default
2305 /etc/vconsole.conf with the appropriate contents.
2307 Contributions from: Colin Walters, Daniel J Walsh, Dave
2308 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
2309 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
2313 * journalctl gained a new --cursor= switch to show entries
2314 starting from the specified location in the journal.
2316 * We now enforce a size limit on journal entry fields exported
2317 with "-o json" in journalctl. Fields larger than 4K will be
2318 assigned null. This can be turned off with --all.
2320 * An (optional) journal gateway daemon is now available as
2321 "systemd-journal-gatewayd.service". This service provides
2322 access to the journal via HTTP and JSON. This functionality
2323 will be used to implement live log synchronization in both
2324 pull and push modes, but has various other users too, such
2325 as easy log access for debugging of embedded devices. Right
2326 now it is already useful to retrieve the journal via HTTP:
2328 # systemctl start systemd-journal-gatewayd.service
2329 # wget http://localhost:19531/entries
2331 This will download the journal contents in a
2332 /var/log/messages compatible format. The same as JSON:
2334 # curl -H"Accept: application/json" http://localhost:19531/entries
2336 This service is also accessible via a web browser where a
2337 single static HTML5 app is served that uses the JSON logic
2338 to enable the user to do some basic browsing of the
2339 journal. This will be extended later on. Here's an example
2340 screenshot of this app in its current state:
2342 http://0pointer.de/public/journal-gatewayd
2344 Contributions from: Kay Sievers, Lennart Poettering, Robert
2345 Milasan, Tom Gundersen
2349 * The bash completion logic is now available for journalctl
2352 * We do not mount the "cpuset" controller anymore together with
2353 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
2354 started if no parameters are assigned to it. "cpuset" hence
2355 broke code that assumed it it could create "cpu" groups and
2358 * journalctl -f will now subscribe to terminal size changes,
2359 and line break accordingly.
2361 Contributions from: Dave Reisner, Kay Sievers, Lennart
2362 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
2366 * nspawn will now create a symlink /etc/localtime in the
2367 container environment, copying the host's timezone
2368 setting. Previously this has been done via a bind mount, but
2369 since symlinks cannot be bind mounted this has now been
2370 changed to create/update the appropriate symlink.
2372 * journalctl -n's line number argument is now optional, and
2373 will default to 10 if omitted.
2375 * journald will now log the maximum size the journal files may
2376 take up on disk. This is particularly useful if the default
2377 built-in logic of determining this parameter from the file
2378 system size is used. Use "systemctl status
2379 systemd-journald.service" to see this information.
2381 * The multi-seat X wrapper tool has been stripped down. As X
2382 is now capable of enumerating graphics devices via udev in a
2383 seat-aware way the wrapper is not strictly necessary
2384 anymore. A stripped down temporary stop-gap is still shipped
2385 until the upstream display managers have been updated to
2386 fully support the new X logic. Expect this wrapper to be
2387 removed entirely in one of the next releases.
2389 * HandleSleepKey= in logind.conf has been split up into
2390 HandleSuspendKey= and HandleHibernateKey=. The old setting
2391 is not available anymore. X11 and the kernel are
2392 distuingishing between these keys and we should too. This
2393 also means the inhibition lock for these keys has been split
2396 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
2397 Poettering, Lukas Nykryn, Václav Pavlín
2401 * Whenever a unit changes state we will now log this to the
2402 journal and show along the unit's own log output in
2405 * ConditionPathIsMountPoint= can now properly detect bind
2406 mount points too. (Previously, a bind mount of one file
2407 system to another place in the same file system could not be
2408 detected as mount, since they shared struct stat's st_dev
2411 * We will now mount the cgroup controllers cpu, cpuacct,
2412 cpuset and the controllers net_cls, net_prio together by
2415 * nspawn containers will now have a virtualized boot
2416 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
2417 over with a randomized ID at container initialization). This
2418 has the effect of making "journalctl -b" do the right thing
2421 * The JSON output journal serialization has been updated not
2422 to generate "endless" list objects anymore, but rather one
2423 JSON object per line. This is more in line how most JSON
2424 parsers expect JSON objects. The new output mode
2425 "json-pretty" has been added to provide similar output, but
2426 neatly aligned for readability by humans.
2428 * We dropped all explicit sync() invocations in the shutdown
2429 code. The kernel does this implicitly anyway in the kernel
2430 reboot() syscall. halt(8)'s -n option is now a compatibility
2433 * We now support virtualized reboot() in containers, as
2434 supported by newer kernels. We will fall back to exit() if
2435 CAP_SYS_REBOOT is not available to the container. Also,
2436 nspawn makes use of this now and will actually reboot the
2437 container if the containerized OS asks for that.
2439 * journalctl will only show local log output by default
2440 now. Use --merge (-m) to show remote log output, too.
2442 * libsystemd-journal gained the new sd_journal_get_usage()
2443 call to determine the current disk usage of all journal
2444 files. This is exposed in the new "journalctl --disk-usage"
2447 * journald gained a new configuration setting SplitMode= in
2448 journald.conf which may be used to control how user journals
2449 are split off. See journald.conf(5) for details.
2451 * A new condition type ConditionFileNotEmpty= has been added.
2453 * tmpfiles' "w" lines now support file globbing, to write
2454 multiple files at once.
2456 * We added Python bindings for the journal submission
2457 APIs. More Python APIs for a number of selected APIs will
2458 likely follow. Note that we intend to add native bindings
2459 only for the Python language, as we consider it common
2460 enough to deserve bindings shipped within systemd. There are
2461 various projects outside of systemd that provide bindings
2462 for languages such as PHP or Lua.
2464 * Many conditions will now resolve specifiers such as %i. In
2465 addition, PathChanged= and related directives of .path units
2466 now support specifiers as well.
2468 * There's now a new RPM macro definition for the system preset
2471 * journald will now warn if it ca not forward a message to the
2472 syslog daemon because its socket is full.
2474 * timedated will no longer write or process /etc/timezone,
2475 except on Debian. As we do not support late mounted /usr
2476 anymore /etc/localtime always being a symlink is now safe,
2477 and hence the information in /etc/timezone is not necessary
2480 * logind will now always reserve one VT for a text getty (VT6
2481 by default). Previously if more than 6 X sessions where
2482 started they took up all the VTs with auto-spawned gettys,
2483 so that no text gettys were available anymore.
2485 * udev will now automatically inform the btrfs kernel logic
2486 about btrfs RAID components showing up. This should make
2487 simple hotplug based btrfs RAID assembly work.
2489 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
2490 (but not for its children which will stay at the kernel
2491 default). This should allow setups with a lot more listening
2494 * systemd will now always pass the configured timezone to the
2495 kernel at boot. timedated will do the same when the timezone
2498 * logind's inhibition logic has been updated. By default,
2499 logind will now handle the lid switch, the power and sleep
2500 keys all the time, even in graphical sessions. If DEs want
2501 to handle these events on their own they should take the new
2502 handle-power-key, handle-sleep-key and handle-lid-switch
2503 inhibitors during their runtime. A simple way to achiveve
2504 that is to invoke the DE wrapped in an invocation of:
2506 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
2508 * Access to unit operations is now checked via SELinux taking
2509 the unit file label and client process label into account.
2511 * systemd will now notify the administrator in the journal
2512 when he over-mounts a non-empty directory.
2514 * There are new specifiers that are resolved in unit files,
2515 for the host name (%H), the machine ID (%m) and the boot ID
2518 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
2519 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
2520 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
2521 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
2522 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
2523 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
2524 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
2528 * Support for reading structured kernel messages from
2529 /dev/kmsg has now been added and is enabled by default.
2531 * Support for reading kernel messages from /proc/kmsg has now
2532 been removed. If you want kernel messages in the journal
2533 make sure to run a recent kernel (>= 3.5) that supports
2534 reading structured messages from /dev/kmsg (see
2535 above). /proc/kmsg is now exclusive property of classic
2536 syslog daemons again.
2538 * The libudev API gained the new
2539 udev_device_new_from_device_id() call.
2541 * The logic for file system namespace (ReadOnlyDirectory=,
2542 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
2543 require pivot_root() anymore. This means fewer temporary
2544 directories are created below /tmp for this feature.
2546 * nspawn containers will now see and receive all submounts
2547 made on the host OS below the root file system of the
2550 * Forward Secure Sealing is now supported for Journal files,
2551 which provide cryptographical sealing of journal files so
2552 that attackers cannot alter log history anymore without this
2553 being detectable. Lennart will soon post a blog story about
2554 this explaining it in more detail.
2556 * There are two new service settings RestartPreventExitStatus=
2557 and SuccessExitStatus= which allow configuration of exit
2558 status (exit code or signal) which will be excepted from the
2559 restart logic, resp. consider successful.
2561 * journalctl gained the new --verify switch that can be used
2562 to check the integrity of the structure of journal files and
2563 (if Forward Secure Sealing is enabled) the contents of
2566 * nspawn containers will now be run with /dev/stdin, /dev/fd/
2567 and similar symlinks pre-created. This makes running shells
2568 as container init process a lot more fun.
2570 * The fstab support can now handle PARTUUID= and PARTLABEL=
2573 * A new ConditionHost= condition has been added to match
2574 against the hostname (with globs) and machine ID. This is
2575 useful for clusters where a single OS image is used to
2576 provision a large number of hosts which shall run slightly
2577 different sets of services.
2579 * Services which hit the restart limit will now be placed in a
2582 Contributions from: Bertram Poettering, Dave Reisner, Huang
2583 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
2584 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
2588 * When running in --user mode systemd will now become a
2589 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
2590 tree a lot more organized.
2592 * A new PartOf= unit dependency type has been introduced that
2593 may be used to group services in a natural way.
2595 * "systemctl enable" may now be used to enable instances of
2598 * journalctl now prints error log levels in red, and
2599 warning/notice log levels in bright white. It also supports
2600 filtering by log level now.
2602 * cgtop gained a new -n switch (similar to top), to configure
2603 the maximum number of iterations to run for. It also gained
2604 -b, to run in batch mode (accepting no input).
2606 * The suffix ".service" may now be omitted on most systemctl
2607 command lines involving service unit names.
2609 * There's a new bus call in logind to lock all sessions, as
2610 well as a loginctl verb for it "lock-sessions".
2612 * libsystemd-logind.so gained a new call sd_journal_perror()
2613 that works similar to libc perror() but logs to the journal
2614 and encodes structured information about the error number.
2616 * /etc/crypttab entries now understand the new keyfile-size=
2619 * shutdown(8) now can send a (configurable) wall message when
2620 a shutdown is cancelled.
2622 * The mount propagation mode for the root file system will now
2623 default to "shared", which is useful to make containers work
2624 nicely out-of-the-box so that they receive new mounts from
2625 the host. This can be undone locally by running "mount
2626 --make-rprivate /" if needed.
2628 * The prefdm.service file has been removed. Distributions
2629 should maintain this unit downstream if they intend to keep
2630 it around. However, we recommend writing normal unit files
2631 for display managers instead.
2633 * Since systemd is a crucial part of the OS we will now
2634 default to a number of compiler switches that improve
2635 security (hardening) such as read-only relocations, stack
2636 protection, and suchlike.
2638 * The TimeoutSec= setting for services is now split into
2639 TimeoutStartSec= and TimeoutStopSec= to allow configuration
2640 of individual time outs for the start and the stop phase of
2643 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
2644 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
2645 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
2646 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
2647 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
2648 Gundersen, Zbigniew Jędrzejewski-Szmek
2652 * The journal and id128 C APIs are now fully documented as man
2655 * Extra safety checks have been added when transitioning from
2656 the initial RAM disk to the main system to avoid accidental
2659 * /etc/crypttab entries now understand the new keyfile-offset=
2662 * systemctl -t can now be used to filter by unit load state.
2664 * The journal C API gained the new sd_journal_wait() call to
2665 make writing synchronous journal clients easier.
2667 * journalctl gained the new -D switch to show journals from a
2670 * journalctl now displays a special marker between log
2671 messages of two different boots.
2673 * The journal is now explicitly flushed to /var via a service
2674 systemd-journal-flush.service, rather than implicitly simply
2675 by seeing /var/log/journal to be writable.
2677 * journalctl (and the journal C APIs) can now match for much
2678 more complex expressions, with alternatives and
2681 * When transitioning from the initial RAM disk to the main
2682 system we will now kill all processes in a killing spree to
2683 ensure no processes stay around by accident.
2685 * Three new specifiers may be used in unit files: %u, %h, %s
2686 resolve to the user name, user home directory resp. user
2687 shell. This is useful for running systemd user instances.
2689 * We now automatically rotate journal files if their data
2690 object hash table gets a fill level > 75%. We also size the
2691 hash table based on the configured maximum file size. This
2692 together should lower hash collisions drastically and thus
2693 speed things up a bit.
2695 * journalctl gained the new "--header" switch to introspect
2696 header data of journal files.
2698 * A new setting SystemCallFilters= has been added to services
2699 which may be used to apply blacklists or whitelists to
2700 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
2702 * nspawn gained a new --link-journal= switch (and quicker: -j)
2703 to link the container journal with the host. This makes it
2704 very easy to centralize log viewing on the host for all
2705 guests while still keeping the journal files separated.
2707 * Many bugfixes and optimizations
2709 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
2710 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
2711 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
2716 * Several tools now understand kernel command line arguments,
2717 which are only read when run in an initial RAM disk. They
2718 usually follow closely their normal counterparts, but are
2721 * There's a new tool to analyze the readahead files that are
2722 automatically generated at boot. Use:
2724 /usr/lib/systemd/systemd-readahead analyze /.readahead
2726 * We now provide an early debug shell on tty9 if this enabled. Use:
2728 systemctl enable debug-shell.service
2730 * All plymouth related units have been moved into the Plymouth
2731 package. Please make sure to upgrade your Plymouth version
2734 * systemd-tmpfiles now supports getting passed the basename of
2735 a configuration file only, in which case it will look for it
2736 in all appropriate directories automatically.
2738 * udevadm info now takes a /dev or /sys path as argument, and
2739 does the right thing. Example:
2741 udevadm info /dev/sda
2742 udevadm info /sys/class/block/sda
2744 * systemctl now prints a warning if a unit is stopped but a
2745 unit that might trigger it continues to run. Example: a
2746 service is stopped but the socket that activates it is left
2749 * "systemctl status" will now mention if the log output was
2750 shortened due to rotation since a service has been started.
2752 * The journal API now exposes functions to determine the
2753 "cutoff" times due to rotation.
2755 * journald now understands SIGUSR1 and SIGUSR2 for triggering
2756 immediately flushing of runtime logs to /var if possible,
2757 resp. for triggering immediate rotation of the journal
2760 * It is now considered an error if a service is attempted to
2761 be stopped that is not loaded.
2763 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
2765 * systemd-analyze now supports Python 3
2767 * tmpfiles now supports cleaning up directories via aging
2768 where the first level dirs are always kept around but
2769 directories beneath it automatically aged. This is enabled
2770 by prefixing the age field with '~'.
2772 * Seat objects now expose CanGraphical, CanTTY properties
2773 which is required to deal with very fast bootups where the
2774 display manager might be running before the graphics drivers
2775 completed initialization.
2777 * Seat objects now expose a State property.
2779 * We now include RPM macros for service enabling/disabling
2780 based on the preset logic. We recommend RPM based
2781 distributions to make use of these macros if possible. This
2782 makes it simpler to reuse RPM spec files across
2785 * We now make sure that the collected systemd unit name is
2786 always valid when services log to the journal via
2789 * There's a new man page kernel-command-line(7) detailing all
2790 command line options we understand.
2792 * The fstab generator may now be disabled at boot by passing
2793 fstab=0 on the kernel command line.
2795 * A new kernel command line option modules-load= is now understood
2796 to load a specific kernel module statically, early at boot.
2798 * Unit names specified on the systemctl command line are now
2799 automatically escaped as needed. Also, if file system or
2800 device paths are specified they are automatically turned
2801 into the appropriate mount or device unit names. Example:
2803 systemctl status /home
2804 systemctl status /dev/sda
2806 * The SysVConsole= configuration option has been removed from
2807 system.conf parsing.
2809 * The SysV search path is no longer exported on the D-Bus
2812 * The Names= option is been removed from unit file parsing.
2814 * There's a new man page bootup(7) detailing the boot process.
2816 * Every unit and every generator we ship with systemd now
2817 comes with full documentation. The self-explanatory boot is
2820 * A couple of services gained "systemd-" prefixes in their
2821 name if they wrap systemd code, rather than only external
2822 code. Among them fsck@.service which is now
2823 systemd-fsck@.service.
2825 * The HaveWatchdog property has been removed from the D-Bus
2828 * systemd.confirm_spawn= on the kernel command line should now
2831 * There's a new man page crypttab(5) which details all options
2832 we actually understand.
2834 * systemd-nspawn gained a new --capability= switch to pass
2835 additional capabilities to the container.
2837 * timedated will now read known NTP implementation unit names
2838 from /usr/lib/systemd/ntp-units.d/*.list,
2839 systemd-timedated-ntp.target has been removed.
2841 * journalctl gained a new switch "-b" that lists log data of
2842 the current boot only.
2844 * The notify socket is in the abstract namespace again, in
2845 order to support daemons which chroot() at start-up.
2847 * There is a new Storage= configuration option for journald
2848 which allows configuration of where log data should go. This
2849 also provides a way to disable journal logging entirely, so
2850 that data collected is only forwarded to the console, the
2851 kernel log buffer or another syslog implementation.
2853 * Many bugfixes and optimizations
2855 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
2856 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
2857 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
2858 Shawn Landden, Tom Gundersen
2862 * "systemctl help <unit>" now shows the man page if one is
2865 * Several new man pages have been added.
2867 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
2868 MaxLevelConsole= can now be specified in
2869 journald.conf. These options allow reducing the amount of
2870 data stored on disk or forwarded by the log level.
2872 * TimerSlackNSec= can now be specified in system.conf for
2873 PID1. This allows system-wide power savings.
2875 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
2876 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
2881 * logind is now capable of (optionally) handling power and
2882 sleep keys as well as the lid switch.
2884 * journalctl now understands the syntax "journalctl
2885 /usr/bin/avahi-daemon" to get all log output of a specific
2888 * CapabilityBoundingSet= in system.conf now also influences
2889 the capability bound set of usermode helpers of the kernel.
2891 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
2892 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
2893 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
2894 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
2898 * Note that we skipped 139 releases here in order to set the
2899 new version to something that is greater than both udev's
2900 and systemd's most recent version number.
2902 * udev: all udev sources are merged into the systemd source tree now.
2903 All future udev development will happen in the systemd tree. It
2904 is still fully supported to use the udev daemon and tools without
2905 systemd running, like in initramfs or other init systems. Building
2906 udev though, will require the *build* of the systemd tree, but
2907 udev can be properly *run* without systemd.
2909 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
2910 should be used to create dead device nodes as workarounds for broken
2913 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
2914 no longer supported. udev_monitor_new_from_netlink() needs to be
2915 used to subscribe to events.
2917 * udev: when udevd is started by systemd, processes which are left
2918 behind by forking them off of udev rules, are unconditionally cleaned
2919 up and killed now after the event handling has finished. Services or
2920 daemons must be started as systemd services. Services can be
2921 pulled-in by udev to get started, but they can no longer be directly
2922 forked by udev rules.
2924 * udev: the daemon binary is called systemd-udevd now and installed
2925 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
2926 to adapt to that, create symlink, or rename the binary after building
2929 * libudev no longer provides these symbols:
2930 udev_monitor_from_socket()
2931 udev_queue_get_failed_list_entry()
2932 udev_get_{dev,sys,run}_path()
2933 The versions number was bumped and symbol versioning introduced.
2935 * systemd-loginctl and systemd-journalctl have been renamed
2936 to loginctl and journalctl to match systemctl.
2938 * The config files: /etc/systemd/systemd-logind.conf and
2939 /etc/systemd/systemd-journald.conf have been renamed to
2940 logind.conf and journald.conf. Package updates should rename
2941 the files to the new names on upgrade.
2943 * For almost all files the license is now LGPL2.1+, changed
2944 from the previous GPL2.0+. Exceptions are some minor stuff
2945 of udev (which will be changed to LGPL2.1 eventually, too),
2946 and the MIT licensed sd-daemon.[ch] library that is suitable
2947 to be used as drop-in files.
2949 * systemd and logind now handle system sleep states, in
2950 particular suspending and hibernating.
2952 * logind now implements a sleep/shutdown/idle inhibiting logic
2953 suitable for a variety of uses. Soonishly Lennart will blog
2954 about this in more detail.
2956 * var-run.mount and var-lock.mount are no longer provided
2957 (which prevously bind mounted these directories to their new
2958 places). Distributions which have not converted these
2959 directories to symlinks should consider stealing these files
2960 from git history and add them downstream.
2962 * We introduced the Documentation= field for units and added
2963 this to all our shipped units. This is useful to make it
2964 easier to explore the boot and the purpose of the various
2967 * All smaller setup units (such as
2968 systemd-vconsole-setup.service) now detect properly if they
2969 are run in a container and are skipped when
2970 appropriate. This guarantees an entirely noise-free boot in
2971 Linux container environments such as systemd-nspawn.
2973 * A framework for implementing offline system updates is now
2974 integrated, for details see:
2975 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
2977 * A new service type Type=idle is available now which helps us
2978 avoiding ugly interleaving of getty output and boot status
2981 * There's now a system-wide CapabilityBoundingSet= option to
2982 globally reduce the set of capabilities for the
2983 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
2984 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
2985 even CAP_NET_ADMIN system-wide for secure systems.
2987 * There are now system-wide DefaultLimitXXX= options to
2988 globally change the defaults of the various resource limits
2989 for all units started by PID 1.
2991 * Harald Hoyer's systemd test suite has been integrated into
2992 systemd which allows easy testing of systemd builds in qemu
2993 and nspawn. (This is really awesome! Ask us for details!)
2995 * The fstab parser is now implemented as generator, not inside
2998 * systemctl will now warn you if .mount units generated from
2999 /etc/fstab are out of date due to changes in fstab that
3000 have not been read by systemd yet.
3002 * systemd is now suitable for usage in initrds. Dracut has
3003 already been updated to make use of this. With this in place
3004 initrds get a slight bit faster but primarily are much
3005 easier to introspect and debug since "systemctl status" in
3006 the host system can be used to introspect initrd services,
3007 and the journal from the initrd is kept around too.
3009 * systemd-delta has been added, a tool to explore differences
3010 between user/admin configuration and vendor defaults.
3012 * PrivateTmp= now affects both /tmp and /var/tmp.
3014 * Boot time status messages are now much prettier and feature
3015 proper english language. Booting up systemd has never been
3018 * Read-ahead pack files now include the inode number of all
3019 files to pre-cache. When the inode changes the pre-caching
3020 is not attempted. This should be nicer to deal with updated
3021 packages which might result in changes of read-ahead
3024 * We now temporaritly lower the kernel's read_ahead_kb variable
3025 when collecting read-ahead data to ensure the kernel's
3026 built-in read-ahead does not add noise to our measurements
3027 of necessary blocks to pre-cache.
3029 * There's now RequiresMountsFor= to add automatic dependencies
3030 for all mounts necessary for a specific file system path.
3032 * MountAuto= and SwapAuto= have been removed from
3033 system.conf. Mounting file systems at boot has to take place
3036 * nspawn now learned a new switch --uuid= to set the machine
3037 ID on the command line.
3039 * nspawn now learned the -b switch to automatically search
3042 * vt102 is now the default TERM for serial TTYs, upgraded from
3045 * systemd-logind now works on VT-less systems.
3047 * The build tree has been reorganized. The individual
3048 components now have directories of their own.
3050 * A new condition type ConditionPathIsReadWrite= is now available.
3052 * nspawn learned the new -C switch to create cgroups for the
3053 container in other hierarchies.
3055 * We now have support for hardware watchdogs, configurable in
3058 * The scheduled shutdown logic now has a public API.
3060 * We now mount /tmp as tmpfs by default, but this can be
3061 masked and /etc/fstab can override it.
3063 * Since udisks does not make use of /media anymore we are not
3064 mounting a tmpfs on it anymore.
3066 * journalctl gained a new --local switch to only interleave
3067 locally generated journal files.
3069 * We can now load the IMA policy at boot automatically.
3071 * The GTK tools have been split off into a systemd-ui.
3073 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
3074 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
3075 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
3076 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
3077 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
3078 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
3079 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
3080 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
3081 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
3086 * This is mostly a bugfix release
3088 * Support optional initialization of the machine ID from the
3089 KVM or container configured UUID.
3091 * Support immediate reboots with "systemctl reboot -ff"
3093 * Show /etc/os-release data in systemd-analyze output
3095 * Many bugfixes for the journal, including endianness fixes and
3096 ensuring that disk space enforcement works
3098 * sd-login.h is C++ comptaible again
3100 * Extend the /etc/os-release format on request of the Debian
3103 * We now refuse non-UTF8 strings used in various configuration
3104 and unit files. This is done to ensure we do not pass invalid
3105 data over D-Bus or expose it elsewhere.
3107 * Register Mimo USB Screens as suitable for automatic seat
3110 * Read SELinux client context from journal clients in a race
3113 * Reorder configuration file lookup order. /etc now always
3114 overrides /run in order to allow the administrator to always
3115 and unconditionally override vendor supplied or
3116 automatically generated data.
3118 * The various user visible bits of the journal now have man
3119 pages. We still lack man pages for the journal API calls
3122 * We now ship all man pages in HTML format again in the
3125 Contributions from: Dave Reisner, Dirk Eibach, Frederic
3126 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
3127 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
3132 * This is mostly a bugfix release
3134 * systems lacking /etc/os-release are no longer supported.
3136 * Various functionality updates to libsystemd-login.so
3138 * Track class of PAM logins to distuingish greeters from
3141 Contributions from: Kay Sievers, Lennart Poettering, Michael
3146 * This is an important bugfix release for v41.
3148 * Building man pages is now optional which should be useful
3149 for those building systemd from git but unwilling to install
3152 * Watchdog support for supervising services is now usable. In
3153 a future release support for hardware watchdogs
3154 (i.e. /dev/watchdog) will be added building on this.
3156 * Service start rate limiting is now configurable and can be
3157 turned off per service. When a start rate limit is hit a
3158 reboot can automatically be triggered.
3160 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
3162 Contributions from: Benjamin Franzke, Bill Nottingham,
3163 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
3164 Schmidt, Michał Górny, Piotr Drąg
3168 * The systemd binary is installed /usr/lib/systemd/systemd now;
3169 An existing /sbin/init symlink needs to be adapted with the
3172 * The code that loads kernel modules has been ported to invoke
3173 libkmod directly, instead of modprobe. This means we do not
3174 support systems with module-init-tools anymore.
3176 * Watchdog support is now already useful, but still not
3179 * A new kernel command line option systemd.setenv= is
3180 understood to set system wide environment variables
3181 dynamically at boot.
3183 * We now limit the set of capabilities of systemd-journald.
3185 * We now set SIGPIPE to ignore by default, since it only is
3186 useful in shell pipelines, and has little use in general
3187 code. This can be disabled with IgnoreSIPIPE=no in unit
3190 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
3191 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
3196 * This is mostly a bugfix release
3198 * We now expose the reason why a service failed in the
3199 "Result" D-Bus property.
3201 * Rudimentary service watchdog support (will be completed over
3202 the next few releases.)
3204 * When systemd forks off in order execute some service we will
3205 now immediately changes its argv[0] to reflect which process
3206 it will execute. This is useful to minimize the time window
3207 with a generic argv[0], which makes bootcharts more useful
3209 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
3210 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
3211 Mike Kazantsev, Ray Strode
3215 * This is mostly a test release, but incorporates many
3218 * New systemd-cgtop tool to show control groups by their
3221 * Linking against libacl for ACLs is optional again. If
3222 disabled, support tracking device access for active logins
3223 goes becomes unavailable, and so does access to the user
3224 journals by the respective users.
3226 * If a group "adm" exists, journal files are automatically
3227 owned by them, thus allow members of this group full access
3228 to the system journal as well as all user journals.
3230 * The journal now stores the SELinux context of the logging
3231 client for all entries.
3233 * Add C++ inclusion guards to all public headers
3235 * New output mode "cat" in the journal to print only text
3236 messages, without any meta data like date or time.
3238 * Include tiny X server wrapper as a temporary stop-gap to
3239 teach XOrg udev display enumeration. This is used by display
3240 managers such as gdm, and will go away as soon as XOrg
3241 learned native udev hotplugging for display devices.
3243 * Add new systemd-cat tool for executing arbitrary programs
3244 with STDERR/STDOUT connected to the journal. Can also act as
3245 BSD logger replacement, and does so by default.
3247 * Optionally store all locally generated coredumps in the
3248 journal along with meta data.
3250 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
3251 writing short strings to files (for usage for /sys), and for
3252 creating symlinks, character and block device nodes.
3254 * New unit file option ControlGroupPersistent= to make cgroups
3255 persistent, following the mechanisms outlined in
3256 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
3258 * Support multiple local RTCs in a sane way
3260 * No longer monopolize IO when replaying readahead data on
3261 rotating disks, since we might starve non-file-system IO to
3262 death, since fanotify() will not see accesses done by blkid,
3265 * Do not show kernel threads in systemd-cgls anymore, unless
3266 requested with new -k switch.
3268 Contributions from: Dan Horák, Kay Sievers, Lennart
3269 Poettering, Michal Schmidt
3273 * This is mostly a test release, but incorporates many
3276 * The git repository moved to:
3277 git://anongit.freedesktop.org/systemd/systemd
3278 ssh://git.freedesktop.org/git/systemd/systemd
3280 * First release with the journal
3281 http://0pointer.de/blog/projects/the-journal.html
3283 * The journal replaces both systemd-kmsg-syslogd and
3284 systemd-stdout-bridge.
3286 * New sd_pid_get_unit() API call in libsystemd-logind
3288 * Many systemadm clean-ups
3290 * Introduce remote-fs-pre.target which is ordered before all
3291 remote mounts and may be used to start services before all
3294 * Added Mageia support
3296 * Add bash completion for systemd-loginctl
3298 * Actively monitor PID file creation for daemons which exit in
3299 the parent process before having finished writing the PID
3300 file in the daemon process. Daemons which do this need to be
3301 fixed (i.e. PID file creation must have finished before the
3302 parent exits), but we now react a bit more gracefully to them.
3304 * Add colourful boot output, mimicking the well-known output
3305 of existing distributions.
3307 * New option PassCredentials= for socket units, for
3308 compatibility with a recent kernel ABI breakage.
3310 * /etc/rc.local is now hooked in via a generator binary, and
3311 thus will no longer act as synchronization point during
3314 * systemctl list-unit-files now supports --root=.
3316 * systemd-tmpfiles now understands two new commands: z, Z for
3317 relabelling files according to the SELinux database. This is
3318 useful to apply SELinux labels to specific files in /sys,
3321 * Output of SysV services is now forwarded to both the console
3322 and the journal by default, not only just the console.
3324 * New man pages for all APIs from libsystemd-login.
3326 * The build tree got reorganized and a the build system is a
3327 lot more modular allowing embedded setups to specifically
3328 select the components of systemd they are interested in.
3330 * Support for Linux systems lacking the kernel VT subsystem is
3333 * configure's --with-rootdir= got renamed to
3334 --with-rootprefix= to follow the naming used by udev and
3337 * Unless specified otherwise we will now install to /usr instead
3338 of /usr/local by default.
3340 * Processes with '@' in argv[0][0] are now excluded from the
3341 final shut-down killing spree, following the logic explained
3343 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
3345 * All processes remaining in a service cgroup when we enter
3346 the START or START_PRE states are now killed with
3347 SIGKILL. That means it is no longer possible to spawn
3348 background processes from ExecStart= lines (which was never
3349 supported anyway, and bad style).
3351 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
3352 reloading of units together.
3354 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
3355 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
3356 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
3357 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
3358 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek