1 systemd System and Service Manager
5 * When restoring the screen brightness at boot, stay away from
6 the darkest setting or from the lowest 5% of the available
7 range, depending on which is the larger value of both. This
8 should effectively protect the user from rebooting into a
9 black screen, should the brightness have been set to minimum
12 * sd-login gained a new sd_machine_get_class() call to
13 determine the class ("vm" or "container") of a machine
14 registered with machined.
16 * sd-login gained new calls
17 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
18 to query the identity of the peer of a local AF_UNIX
19 connection. They operate similar to their sd_pid_get_xyz()
22 * PID 1 will now maintain a system-wide system state engine
23 with the states "starting", "running", "degraded",
24 "maintenance", "stopping". These states are bound to system
25 startup, normal runtime, runtime with at least one failed
26 service, rescue/emergency mode and system shutdown. This
27 state is shown in the "systemctl status" output when no unit
28 name is passed. It is useful to determine system state, in
29 particularly when doing so for many systems or containers at
32 * A new command "list-machines" has been added to "systemctl"
33 that lists all local OS containers and shows their system
34 state (see above), if systemd runs inside of them.
36 * systemctl gained a new "-r" switch to recursively enumerate
37 units on all local containers, when used with the
38 "list-unit" command (which is the default one that is
39 executed when no parameters are specified).
41 * The GPT automatic partition discovery logic will now honour
42 two GPT partition flags: one may be set on a partition to
43 cause it to be mounted read-only, and the other may be set
44 on a partition to ignore it during automatic discovery.
46 * Two new GPT type UUIDs have been added for automatic root
47 partition discovery, for 32bit and 64bit ARM. This is not
48 particularly useful for discovering the root directory on
49 these architectures during bare-metal boots (since UEFI is
50 not common there), but still very useful to allow booting of
51 ARM disk images in nspawn with the -i option.
53 * MAC addresses of interfaces created with nspawn's
54 --network-interface= switch will now be generated from the
55 machine name, and thus be stable between multiple invocations
58 * logind will now automatically remove all IPC objects owned
59 by a user if she or he fully logs out. This makes sure that
60 users who are logged out cannot continue to consume IPC
61 resources. This covers SysV memory, semaphores and message
62 queues as well as POSIX shared memory and message
63 queues. Traditionally SysV and POSIX IPC had no life-cycle
64 limits, with this functionality this is corrected. This may
65 be turned off using the RemoveIPC= switch of logind.conf.
67 * The systemd-machine-id-setup and tmpfiles tools gained a
68 --root= switch to operate on a specific root directory,
71 * journald can now forward logged messages to the TTYs of all
72 logged in users ("wall"). This is the default for all
73 emergency messages now.
75 * A new tool systemd-journal-remote has been added to stream
76 journal log messages across the network.
78 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
79 controller trees are mounted into it. Note that the
80 directories mounted beneath it are not read-only. This is a
81 security measure and is particularly useful because glibc
82 actually includes a search logic to pick any tmpfs it can
83 find to implement shm_open() if /dev/shm is not available
84 (which it might very well be in namespaced setups).
86 * machinectl gained a new "poweroff" command to cleanly power
87 down a local OS container.
89 * The PrivateDevices= unit file setting will now also drop the
90 CAP_MKNOD capability from the capability bound set, and
91 imply DevicePolicy=closed.
93 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
94 comprehensively on all long-running systemd services where
97 * systemd-udevd will now run in a disassociated mount
98 namespace. To mount directories from udev rules make sure to
99 pull in mount units via SYSTEMD_WANTS properties.
101 * The kdbus support gained support for uploading policy into
102 the kernel. sd-bus gained support for creating "monitoring"
103 connections that can eavesdrop into all bus communication
104 for debugging purposes.
106 * Timestamps may now be specified in seconds since the UNIX
107 epoch Jan 1st, 1970 by specifying "@" followed by the value
110 * Native tcpwrap support in systemd has been removed. tcpwrap
111 is old code, not really maintained anymore and has serious
112 shortcomings, and better options such as firewalls
113 exist. For setups that require tcpwrap usage, please
114 consider invoking your socket-activated service via tcpd,
115 like on traditional inetd.
117 * A new system.conf configuration option
118 DefaultTimerAccuracySec= has been added that controls the
119 default AccuracySec= setting of .timer units.
121 * Timer units gained a new WakeSystem= switch. If enabled
122 timers configured this way will cause the system to resume
123 from system suspend (if the system supports that, which most
126 * Timer units gained a new Persistent= switch. If enabled
127 timers configured this way will save to disk when they have
128 been last triggered. This information is then used on next
129 reboot to possible execute overdue timer events, that
130 couldn't take place because the system was powered off. This
131 enables simple anacron-like behaviour for timer units.
133 * systemctl's "list-timers" will now also list the time a
134 timer unit was last triggered in addition to the next time
135 it will be triggered.
137 * systemd-networkd will now assign predictable IPv4LL
138 addresses to its local interfaces.
140 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
141 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
142 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
143 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
144 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
145 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
146 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
147 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
150 -- Berlin, 2014-03-25
154 * A new unit file setting RestrictAddressFamilies= has been
155 added to restrict which socket address families unit
156 processes gain access to. This takes address family names
157 like "AF_INET" or "AF_UNIX", and is useful to minimize the
158 attack surface of services via exotic protocol stacks. This
159 is built on seccomp system call filters.
161 * Two new unit file settings RuntimeDirectory= and
162 RuntimeDirectoryMode= have been added that may be used to
163 manage a per-daemon runtime directories below /run. This is
164 an alternative for setting up directory permissions with
165 tmpfiles snippets, and has the advantage that the runtime
166 directory's lifetime is bound to the daemon runtime and that
167 the daemon starts up with an empty directory each time. This
168 is particularly useful when writing services that drop
169 priviliges using the User= or Group= setting.
171 * The DeviceAllow= unit setting now supports globbing for
172 matching against device group names.
174 * The systemd configuration file system.conf gained new
175 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
176 DefaultMemoryAccounting= to globally turn on/off accounting
177 for specific resources (cgroups) for all units. These
178 settings may still be overridden individually in each unit
181 * systemd-gpt-auto-generator is now able to discover /srv and
182 root partitions in addition to /home and swap partitions. It
183 also supports LUKS-encrypted partitions now. With this in
184 place automatic discovery of partitions to mount following
185 the Discoverable Partitions Specification
186 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
187 is now a lot more complete. This allows booting without
188 /etc/fstab and without root= on the kernel command line on
189 appropriately prepared systems.
191 * systemd-nspawn gained a new --image= switch which allows
192 booting up disk images and Linux installations on any block
193 device that follow the Discoverable Partitions Specification
194 (see above). This means that installations made with
195 appropriately updated installers may now be started and
196 deployed using container managers, completely
197 unmodified. (We hope that libvirt-lxc will add support for
198 this feature soon, too.)
200 * systemd-nspawn gained a new --network-macvlan= setting to
201 set up a private macvlan interface for the
202 container. Similar, systemd-networkd gained a new
203 Kind=macvlan setting in .netdev files.
205 * systemd-networkd now supports configuring local addresses
208 * A new tool systemd-network-wait-online has been added to
209 synchronously wait for network connectivity using
212 * The sd-bus.h bus API gained a new sd_bus_track object for
213 tracking the life-cycle of bus peers. Note that sd-bus.h is
214 still not a public API though (unless you specify
215 --enable-kdbus on the configure command line, which however
216 voids your warranty and you get no API stability guarantee).
218 * The $XDG_RUNTIME_DIR runtime directories for each user are
219 now individual tmpfs instances, which has the benefit of
220 introducing separate pools for each user, with individual
221 size limits, and thus making sure that unprivileged clients
222 can no longer negatively impact the system or other users by
223 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
224 RuntimeDirectorySize= has been introduced that allows
225 controlling the default size limit for all users. It
226 defaults to 10% of the available physical memory. This is no
227 replacement for quotas on tmpfs though (which the kernel
228 still does not support), as /dev/shm and /tmp are still
229 shared resources used by both the system and unprivileged
232 * logind will now automatically turn off automatic suspending
233 on laptop lid close when more than one display is
234 connected. This was previously expected to be implemented
235 individually in desktop environments (such as GNOME),
236 however has been added to logind now, in order to fix a
237 boot-time race where a desktop environment might not have
238 been started yet and thus not been able to take an inhibitor
239 lock at the time where logind already suspends the system
242 * logind will now wait at least 30s after each system
243 suspend/resume cycle, and 3min after system boot before
244 suspending the system due to a closed laptop lid. This
245 should give USB docking stations and similar enough time to
246 be probed and configured after system resume and boot in
247 order to then act as suspend blocker.
249 * systemd-run gained a new --property= setting which allows
250 initialization of resource control properties (and others)
251 for the created scope or service unit. Example: "systemd-run
252 --property=BlockIOWeight=10 updatedb" may be used to run
253 updatedb at a low block IO scheduling weight.
255 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
256 now also work in --scope mode.
258 * When systemd is compiled with kdbus support, basic support
259 for enforced policies is now in place. (Note that enabling
260 kdbus still voids your warranty and no API compatibility
263 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
264 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
265 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
266 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
267 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
268 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
269 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
270 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
271 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
272 Zbigniew Jędrzejewski-Szmek
274 -- Berlin, 2014-03-12
278 * systemd will now relabel /dev after loading the SMACK policy
279 according to SMACK rules.
281 * A new unit file option AppArmorProfile= has been added to
282 set the AppArmor profile for the processes of a unit.
284 * A new condition check ConditionArchitecture= has been added
285 to conditionalize units based on the system architecture, as
286 reported by uname()'s "machine" field.
288 * systemd-networkd now supports matching on the system
289 virtualization, architecture, kernel command line, host name
292 * logind is now a lot more aggressive when suspending the
293 machine due to a closed laptop lid. Instead of acting only
294 on the lid close action it will continuously watch the lid
295 status and act on it. This is useful for laptops where the
296 power button is on the outside of the chassis so that it can
297 be reached without opening the lid (such as the Lenovo
298 Yoga). On those machines logind will now immediately
299 re-suspend the machine if the power button has been
300 accidentally pressed while the laptop was suspended and in a
303 * logind will now watch SW_DOCK switches and inhibit reaction
304 to the lid switch if it is pressed. This means that logind
305 will not suspend the machine anymore if the lid is closed
306 and the system is docked, if the laptop supports SW_DOCK
307 notifications via the input layer. Note that ACPI docking
308 stations do not generate this currently. Also note that this
309 logic is usually not fully sufficient and Desktop
310 Environments should take a lid switch inhibitor lock when an
311 external display is connected, as systemd will not watch
314 * nspawn will now make use of the devices cgroup controller by
315 default, and only permit creation of and access to the usual
316 API device nodes like /dev/null or /dev/random, as well as
317 access to (but not creation of) the pty devices.
319 * We will now ship a default .network file for
320 systemd-networkd that automatically configures DHCP for
321 network interfaces created by nspawn's --network-veth or
322 --network-bridge= switches.
324 * systemd will now understand the usual M, K, G, T suffixes
325 according to SI conventions (i.e. to the base 1000) when
326 referring to throughput and hardware metrics. It will stay
327 with IEC conventions (i.e. to the base 1024) for software
328 metrics, according to what is customary according to
329 Wikipedia. We explicitly document which base applies for
330 each configuration option.
332 * The DeviceAllow= setting in unit files now supports a syntax
333 to whitelist an entire group of devices node majors at once,
334 based on the /proc/devices listing. For example, with the
335 string "char-pts" it is now possible to whitelist all
336 current and future pseudo-TTYs at once.
338 * sd-event learned a new "post" event source. Event sources of
339 this type are triggered by the dispatching of any event
340 source of a type that is not "post". This is useful for
341 implementing clean-up and check event sources that are
342 triggered by other work being done in the program.
344 * systemd-networkd is no longer statically enabled, but uses
345 the usual [Install] sections so that it can be
346 enabled/disabled using systemctl. It still is enabled by
349 * When creating a veth interface pair with systemd-nspawn the
350 host side will now be prefixed with "vb-" if
351 --network-bridge= is used, and with "ve-" if --network-veth
352 is used. This way it is easy to distinguish these cases on
353 the host, for example to apply different configuration to
354 them with systemd-networkd.
356 * The compatibility libraries for libsystemd-journal.so,
357 libsystem-id128.so, libsystemd-login.so and
358 libsystemd-daemon.so do not make use of IFUNC
359 anymore. Instead we now build libsystemd.so multiple times
360 under these alternative names. This means that the footprint
361 is drastically increased, but given that these are
362 transitional compatibility libraries this shouldn't matter
363 much. This change has been made necessary to support the ARM
364 platform for these compatibility libraries, as the ARM
365 toolchain isn't really at the same level as the toolchain
366 for other architectures like x86 and does not support
367 IFUNC. Please make sure to use --enable-compat-libs only
368 during a transitional period!
370 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
371 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
372 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
373 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
374 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
375 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
376 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
377 Zbigniew Jędrzejewski-Szmek
379 -- Berlin, 2014-02-24
383 * A new component "systemd-networkd" has been added that can
384 be used to configure local network interfaces statically or
385 via DHCP. It is capable of bringing up bridges, VLANs, and
386 bonding. Currently, no hook-ups for interactive network
387 configuration are provided. Use this for your initrd,
388 container, embedded, or server setup if you need a simple,
389 yet powerful, network configuration solution. This
390 configuration subsystem is quite nifty, as it allows wildcard
391 hotplug matching in interfaces. For example, with a single
392 configuration snippet, you can configure that all Ethernet
393 interfaces showing up are automatically added to a bridge,
394 or similar. It supports link-sensing and more.
396 * A new tool "systemd-socket-proxyd" has been added which can
397 act as a bidirectional proxy for TCP sockets. This is
398 useful for adding socket activation support to services that
399 do not actually support socket activation, including virtual
400 machines and the like.
402 * Add a new tool to save/restore rfkill state on
405 * Save/restore state of keyboard backlights in addition to
406 display backlights on shutdown/boot.
408 * udev learned a new SECLABEL{} construct to label device
409 nodes with a specific security label when they appear. For
410 now, only SECLABEL{selinux} is supported, but the syntax is
411 prepared for additional security frameworks.
413 * udev gained a new scheme to configure link-level attributes
414 from files in /etc/systemd/network/*.link. These files can
415 match against MAC address, device path, driver name and type,
416 and will apply attributes like the naming policy, link speed,
417 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
418 address assignment policy (randomized, ...).
420 * The configuration of network interface naming rules for
421 "permanent interface names" has changed: a new NamePolicy=
422 setting in the [Link] section of .link files determines the
423 priority of possible naming schemes (onboard, slot, mac,
424 path). The default value of this setting is determined by
425 /usr/lib/net/links/99-default.link. Old
426 80-net-name-slot.rules udev configuration file has been
427 removed, so local configuration overriding this file should
428 be adapated to override 99-default.link instead.
430 * When the User= switch is used in a unit file, also
431 initialize $SHELL= based on the user database entry.
433 * systemd no longer depends on libdbus. All communication is
434 now done with sd-bus, systemd's low-level bus library
437 * kdbus support has been added to PID 1 itself. When kdbus is
438 enabled, this causes PID 1 to set up the system bus and
439 enable support for a new ".busname" unit type that
440 encapsulates bus name activation on kdbus. It works a little
441 bit like ".socket" units, except for bus names. A new
442 generator has been added that converts classic dbus1 service
443 activation files automatically into native systemd .busname
446 * sd-bus: add a light-weight vtable implementation that allows
447 defining objects on the bus with a simple static const
448 vtable array of its methods, signals and properties.
450 * systemd will not generate or install static dbus
451 introspection data anymore to /usr/share/dbus-1/interfaces,
452 as the precise format of these files is unclear, and
453 nothing makes use of it.
455 * A proxy daemon is now provided to proxy clients connecting
456 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
457 compatibility with classic D-Bus.
459 * A bus driver implementation has been added that supports the
460 classic D-Bus bus driver calls on kdbus, also for
461 compatibility purposes.
463 * A new API "sd-event.h" has been added that implements a
464 minimal event loop API built around epoll. It provides a
465 couple of features that direct epoll usage is lacking:
466 prioritization of events, scales to large numbers of timer
467 events, per-event timer slack (accuracy), system-wide
468 coalescing of timer events, exit handlers, watchdog
469 supervision support using systemd's sd_notify() API, child
472 * A new API "sd-rntl.h" has been added that provides an API
473 around the route netlink interface of the kernel, similar in
476 * A new API "sd-dhcp-client.h" has been added that provides a
477 small DHCPv4 client-side implementation. This is used by
480 * There is a new kernel command line option
481 "systemd.restore_state=0|1". When set to "0", none of the
482 systemd tools will restore saved runtime state to hardware
483 devices. More specifically, the rfkill and backlight states
486 * The FsckPassNo= compatibility option in mount/service units
487 has been removed. The fstab generator will now add the
488 necessary dependencies automatically, and does not require
489 PID1's support for that anymore.
491 * journalctl gained a new switch, --list-boots, that lists
492 recent boots with their times and boot IDs.
494 * The various tools like systemctl, loginctl, timedatectl,
495 busctl, systemd-run, ... have gained a new switch "-M" to
496 connect to a specific, local OS container (as direct
497 connection, without requiring SSH). This works on any
498 container that is registered with machined, such as those
499 created by libvirt-lxc or nspawn.
501 * systemd-run and systemd-analyze also gained support for "-H"
502 to connect to remote hosts via SSH. This is particularly
503 useful for systemd-run because it enables queuing of jobs
506 * machinectl gained a new command "login" to open a getty
507 login in any local container. This works with any container
508 that is registered with machined (such as those created by
509 libvirt-lxc or nspawn), and which runs systemd inside.
511 * machinectl gained a new "reboot" command that may be used to
512 trigger a reboot on a specific container that is registered
513 with machined. This works on any container that runs an init
516 * systemctl gained a new "list-timers" command to print a nice
517 listing of installed timer units with the times they elapse
520 * Alternative reboot() parameters may now be specified on the
521 "systemctl reboot" command line and are passed to the
522 reboot() system call.
524 * systemctl gained a new --job-mode= switch to configure the
525 mode to queue a job with. This is a more generic version of
526 --fail, --irreversible, and --ignore-dependencies, which are
527 still available but not advertised anymore.
529 * /etc/systemd/system.conf gained new settings to configure
530 various default timeouts of units, as well as the default
531 start limit interval and burst. These may still be overridden
534 * PID1 will now export on the bus profile data of the security
535 policy upload process (such as the SELinux policy upload to
538 * journald: when forwarding logs to the console, include
539 timestamps (following the setting in
540 /sys/module/printk/parameters/time).
542 * OnCalendar= in timer units now understands the special
543 strings "yearly" and "annually". (Both are equivalent)
545 * The accuracy of timer units is now configurable with the new
546 AccuracySec= setting. It defaults to 1min.
548 * A new dependency type JoinsNamespaceOf= has been added that
549 allows running two services within the same /tmp and network
550 namespace, if PrivateNetwork= or PrivateTmp= are used.
552 * A new command "cat" has been added to systemctl. It outputs
553 the original unit file of a unit, and concatenates the
554 contents of additional "drop-in" unit file snippets, so that
555 the full configuration is shown.
557 * systemctl now supports globbing on the various "list-xyz"
558 commands, like "list-units" or "list-sockets", as well as on
559 those commands which take multiple unit names.
561 * journalctl's --unit= switch gained support for globbing.
563 * All systemd daemons now make use of the watchdog logic so
564 that systemd automatically notices when they hang.
566 * If the $container_ttys environment variable is set,
567 getty-generator will automatically spawn a getty for each
568 listed tty. This is useful for container managers to request
569 login gettys to be spawned on as many ttys as needed.
571 * %h, %s, %U specifier support is not available anymore when
572 used in unit files for PID 1. This is because NSS calls are
573 not safe from PID 1. They stay available for --user
574 instances of systemd, and as special case for the root user.
576 * loginctl gained a new "--no-legend" switch to turn off output
579 * The "sd-login.h" API gained three new calls:
580 sd_session_is_remote(), sd_session_get_remote_user(),
581 sd_session_get_remote_host() to query information about
584 * The udev hardware database now also carries vendor/product
585 information of SDIO devices.
587 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
588 determine whether watchdog notifications are requested by
591 * Socket-activated per-connection services now include a
592 short description of the connection parameters in the
595 * tmpfiles gained a new "--boot" option. When this is not used,
596 only lines where the command character is not suffixed with
597 "!" are executed. When this option is specified, those
598 options are executed too. This partitions tmpfiles
599 directives into those that can be safely executed at any
600 time, and those which should be run only at boot (for
601 example, a line that creates /run/nologin).
603 * A new API "sd-resolve.h" has been added which provides a simple
604 asynchronous wrapper around glibc NSS host name resolution
605 calls, such as getaddrinfo(). In contrast to glibc's
606 getaddrinfo_a(), it does not use signals. In contrast to most
607 other asynchronous name resolution libraries, this one does
608 not reimplement DNS, but reuses NSS, so that alternate
609 host name resolution systems continue to work, such as mDNS,
610 LDAP, etc. This API is based on libasyncns, but it has been
611 cleaned up for inclusion in systemd.
613 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
614 "sd-daemon.h" are no longer found in individual libraries
615 libsystemd-journal.so, libsystemd-login.so,
616 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
617 merged them into a single library, libsystemd.so, which
618 provides all symbols. The reason for this is cyclic
619 dependencies, as these libraries tend to use each other's
620 symbols. So far, we've managed to workaround that by linking
621 a copy of a good part of our code into each of these
622 libraries again and again, which, however, makes certain
623 things hard to do, like sharing static variables. Also, it
624 substantially increases footprint. With this change, there
625 is only one library for the basic APIs systemd
626 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
627 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
628 library as well, however are subject to the --enable-kdbus
629 switch (see below). Note that "sd-dhcp-client.h" is not part
630 of this library (this is because it only consumes, never
631 provides, services of/to other APIs). To make the transition
632 easy from the separate libraries to the unified one, we
633 provide the --enable-compat-libs compile-time switch which
634 will generate stub libraries that are compatible with the
635 old ones but redirect all calls to the new one.
637 * All of the kdbus logic and the new APIs "sd-bus.h",
638 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
639 and "sd-utf8.h" are compile-time optional via the
640 "--enable-kdbus" switch, and they are not compiled in by
641 default. To make use of kdbus, you have to explicitly enable
642 the switch. Note however, that neither the kernel nor the
643 userspace API for all of this is considered stable yet. We
644 want to maintain the freedom to still change the APIs for
645 now. By specifying this build-time switch, you acknowledge
646 that you are aware of the instability of the current
649 * Also, note that while kdbus is pretty much complete,
650 it lacks one thing: proper policy support. This means you
651 can build a fully working system with all features; however,
652 it will be highly insecure. Policy support will be added in
653 one of the next releases, at the same time that we will
654 declare the APIs stable.
656 * When the kernel command-line argument "kdbus" is specified,
657 systemd will automatically load the kdbus.ko kernel module. At
658 this stage of development, it is only useful for testing kdbus
659 and should not be used in production. Note: if "--enable-kdbus"
660 is specified, and the kdbus.ko kernel module is available, and
661 "kdbus" is added to the kernel command line, the entire system
662 runs with kdbus instead of dbus-daemon, with the above mentioned
663 problem of missing the system policy enforcement. Also a future
664 version of kdbus.ko or a newer systemd will not be compatible with
665 each other, and will unlikely be able to boot the machine if only
666 one of them is updated.
668 * systemctl gained a new "import-environment" command which
669 uploads the caller's environment (or parts thereof) into the
670 service manager so that it is inherited by services started
671 by the manager. This is useful to upload variables like
672 $DISPLAY into the user service manager.
674 * A new PrivateDevices= switch has been added to service units
675 which allows running a service with a namespaced /dev
676 directory that does not contain any device nodes for
677 physical devices. More specifically, it only includes devices
678 such as /dev/null, /dev/urandom, and /dev/zero which are API
681 * logind has been extended to support behaviour like VT
682 switching on seats that do not support a VT. This makes
683 multi-session available on seats that are not the first seat
684 (seat0), and on systems where kernel support for VTs has
685 been disabled at compile-time.
687 * If a process holds a delay lock for system sleep or shutdown
688 and fails to release it in time, we will now log its
689 identity. This makes it easier to identify processes that
690 cause slow suspends or power-offs.
692 * When parsing /etc/crypttab, support for a new key-slot=
693 option as supported by Debian is added. It allows indicating
694 which LUKS slot to use on disk, speeding up key loading.
696 * The sd_journald_sendv() API call has been checked and
697 officially declared to be async-signal-safe so that it may
698 be invoked from signal handlers for logging purposes.
700 * Boot-time status output is now enabled automatically after a
701 short timeout if boot does not progress, in order to give
702 the user an indication what she or he is waiting for.
704 * The boot-time output has been improved to show how much time
705 remains until jobs expire.
707 * The KillMode= switch in service units gained a new possible
708 value "mixed". If set, and the unit is shut down, then the
709 initial SIGTERM signal is sent only to the main daemon
710 process, while the following SIGKILL signal is sent to
711 all remaining processes of the service.
713 * When a scope unit is registered, a new property "Controller"
714 may be set. If set to a valid bus name, systemd will send a
715 RequestStop() signal to this name when it would like to shut
716 down the scope. This may be used to hook manager logic into
717 the shutdown logic of scope units. Also, scope units may now
718 be put in a special "abandoned" state, in which case the
719 manager process which created them takes no further
720 responsibilities for it.
722 * When reading unit files, systemd will now verify
723 the access mode of these files, and warn about certain
724 suspicious combinations. This has been added to make it
725 easier to track down packaging bugs where unit files are
726 marked executable or world-writable.
728 * systemd-nspawn gained a new "--setenv=" switch to set
729 container-wide environment variables. The similar option in
730 systemd-activate was renamed from "--environment=" to
731 "--setenv=" for consistency.
733 * systemd-nspawn has been updated to create a new kdbus domain
734 for each container that is invoked, thus allowing each
735 container to have its own set of system and user buses,
736 independent of the host.
738 * systemd-nspawn gained a new --drop-capability= switch to run
739 the container with less capabilities than the default. Both
740 --drop-capability= and --capability= now take the special
741 string "all" for dropping or keeping all capabilities.
743 * systemd-nspawn gained new switches for executing containers
744 with specific SELinux labels set.
746 * systemd-nspawn gained a new --quiet switch to not generate
747 any additional output but the container's own console
750 * systemd-nspawn gained a new --share-system switch to run a
751 container without PID namespacing enabled.
753 * systemd-nspawn gained a new --register= switch to control
754 whether the container is registered with systemd-machined or
755 not. This is useful for containers that do not run full
756 OS images, but only specific apps.
758 * systemd-nspawn gained a new --keep-unit which may be used
759 when invoked as the only program from a service unit, and
760 results in registration of the unit service itself in
761 systemd-machined, instead of a newly opened scope unit.
763 * systemd-nspawn gained a new --network-interface= switch for
764 moving arbitrary interfaces to the container. The new
765 --network-veth switch creates a virtual Ethernet connection
766 between host and container. The new --network-bridge=
767 switch then allows assigning the host side of this virtual
768 Ethernet connection to a bridge device.
770 * systemd-nspawn gained a new --personality= switch for
771 setting the kernel personality for the container. This is
772 useful when running a 32bit container on a 64bit host. A
773 similar option Personality= is now also available in service
776 * logind will now also track a "Desktop" identifier for each
777 session which encodes the desktop environment of it. This is
778 useful for desktop environments that want to identify
779 multiple running sessions of itself easily.
781 * A new SELinuxContext= setting for service units has been
782 added that allows setting a specific SELinux execution
783 context for a service.
785 * Most systemd client tools will now honour $SYSTEMD_LESS for
786 settings of the "less" pager. By default, these tools will
787 override $LESS to allow certain operations to work, such as
788 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
789 influence this logic.
791 * systemd's "seccomp" hook-up has been changed to make use of
792 the libseccomp library instead of using its own
793 implementation. This has benefits for portability among
796 * For usage together with SystemCallFilter=, a new
797 SystemCallErrorNumber= setting has been introduced that
798 allows configuration of a system error number to return on
799 filtered system calls, instead of immediately killing the
800 process. Also, SystemCallArchitectures= has been added to
801 limit access to system calls of a particular architecture
802 (in order to turn off support for unused secondary
803 architectures). There is also a global
804 SystemCallArchitectures= setting in system.conf now to turn
805 off support for non-native system calls system-wide.
807 * systemd requires a kernel with a working name_to_handle_at(),
808 please see the kernel config requirements in the README file.
810 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
811 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
812 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
813 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
814 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
815 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
816 Elia Pinto, Florian Weimer, George McCollister, Goffredo
817 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
818 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
819 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
820 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
821 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
822 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
823 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
824 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
825 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
826 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
827 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
828 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
829 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
830 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
831 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
832 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
833 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
835 -- Berlin, 2014-02-20
839 * logind has gained support for facilitating privileged input
840 and drm device access for unprivileged clients. This work is
841 useful to allow Wayland display servers (and similar
842 programs, such as kmscon) to run under the user's ID and
843 access input and drm devices which are normally
844 protected. When this is used (and the kernel is new enough)
845 logind will "mute" IO on the file descriptors passed to
846 Wayland as long as it is in the background and "unmute" it
847 if it returns into the foreground. This allows secure
848 session switching without allowing background sessions to
849 eavesdrop on input and display data. This also introduces
850 session switching support if VT support is turned off in the
851 kernel, and on seats that are not seat0.
853 * A new kernel command line option luks.options= is understood
854 now which allows specifiying LUKS options for usage for LUKS
855 encrypted partitions specified with luks.uuid=.
857 * tmpfiles.d(5) snippets may now use specifier expansion in
858 path names. More specifically %m, %b, %H, %v, are now
859 replaced by the local machine id, boot id, hostname, and
860 kernel version number.
862 * A new tmpfiles.d(5) command "m" has been introduced which
863 may be used to change the owner/group/access mode of a file
864 or directory if it exists, but do nothing if it doesn't.
866 * This release removes high-level support for the
867 MemorySoftLimit= cgroup setting. The underlying kernel
868 cgroup attribute memory.soft_limit= is currently badly
869 designed and likely to be removed from the kernel API in its
870 current form, hence we shouldn't expose it for now.
872 * The memory.use_hierarchy cgroup attribute is now enabled for
873 all cgroups systemd creates in the memory cgroup
874 hierarchy. This option is likely to be come the built-in
875 default in the kernel anyway, and the non-hierarchial mode
876 never made much sense in the intrinsically hierarchial
879 * A new field _SYSTEMD_SLICE= is logged along with all journal
880 messages containing the slice a message was generated
881 from. This is useful to allow easy per-customer filtering of
882 logs among other things.
884 * systemd-journald will no longer adjust the group of journal
885 files it creates to the "systemd-journal" group. Instead we
886 rely on the journal directory to be owned by the
887 "systemd-journal" group, and its setgid bit set, so that the
888 kernel file system layer will automatically enforce that
889 journal files inherit this group assignment. The reason for
890 this change is that we cannot allow NSS look-ups from
891 journald which would be necessary to resolve
892 "systemd-journal" to a numeric GID, because this might
893 create deadlocks if NSS involves synchronous queries to
894 other daemons (such as nscd, or sssd) which in turn are
895 logging clients of journald and might block on it, which
896 would then dead lock. A tmpfiles.d(5) snippet included in
897 systemd will make sure the setgid bit and group are
898 properly set on the journal directory if it exists on every
899 boot. However, we recommend adjusting it manually after
900 upgrades too (or from RPM scriptlets), so that the change is
901 not delayed until next reboot.
903 * Backlight and random seed files in /var/lib/ have moved into
904 the /var/lib/systemd/ directory, in order to centralize all
905 systemd generated files in one directory.
907 * Boot time performance measurements (as displayed by
908 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
909 performance information if that's available to determine how
910 much time BIOS and boot loader initialization required. With
911 a sufficiently new BIOS you hence no longer need to boot
912 with Gummiboot to get access to such information.
914 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
915 Cristian Rodríguez, Dave Reisner, David Herrmann, David
916 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
917 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
918 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
919 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
920 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
922 -- Berlin, 2013-10-02
926 * The Restart= option for services now understands a new
927 on-watchdog setting, which will restart the service
928 automatically if the service stops sending out watchdog keep
929 alive messages (as configured with WatchdogSec=).
931 * The getty generator (which is responsible for bringing up a
932 getty on configured serial consoles) will no longer only
933 start a getty on the primary kernel console but on all
934 others, too. This makes the order in which console= is
935 specified on the kernel command line less important.
937 * libsystemd-logind gained a new sd_session_get_vt() call to
938 retrieve the VT number of a session.
940 * If the option "tries=0" is set for an entry of /etc/crypttab
941 its passphrase is queried indefinitely instead of any
942 maximum number of tries.
944 * If a service with a configure PID file terminates its PID
945 file will now be removed automatically if it still exists
946 afterwards. This should put an end to stale PID files.
948 * systemd-run will now also take relative binary path names
949 for execution and no longer insists on absolute paths.
951 * InaccessibleDirectories= and ReadOnlyDirectories= now take
952 paths that are optionally prefixed with "-" to indicate that
953 it should not be considered a failure if they don't exist.
955 * journalctl -o (and similar commands) now understands a new
956 output mode "short-precise", it is similar to "short" but
957 shows timestamps with usec accuracy.
959 * The option "discard" (as known from Debian) is now
960 synonymous to "allow-discards" in /etc/crypttab. In fact,
961 "discard" is preferred now (since it is easier to remember
964 * Some licensing clean-ups were made, so that more code is now
965 LGPL-2.1 licensed than before.
967 * A minimal tool to save/restore the display backlight
968 brightness across reboots has been added. It will store the
969 backlight setting as late as possible at shutdown, and
970 restore it as early as possible during reboot.
972 * A logic to automatically discover and enable home and swap
973 partitions on GPT disks has been added. With this in place
974 /etc/fstab becomes optional for many setups as systemd can
975 discover certain partitions located on the root disk
976 automatically. Home partitions are recognized under their
977 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
978 partitions are recognized under their GPT type ID
979 0657fd6da4ab43c484e50933c84b4f4f.
981 * systemd will no longer pass any environment from the kernel
982 or initrd to system services. If you want to set an
983 environment for all services, do so via the kernel command
984 line systemd.setenv= assignment.
986 * The systemd-sysctl tool no longer natively reads the file
987 /etc/sysctl.conf. If desired, the file should be symlinked
988 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
989 legacy support by a symlink rather than built-in code, it
990 also makes the otherwise hidden order of application of the
991 different files visible. (Note that this partly reverts to a
992 pre-198 application order of sysctl knobs!)
994 * The "systemctl set-log-level" and "systemctl dump" commands
995 have been moved to systemd-analyze.
997 * systemd-run learned the new --remain-after-exit switch,
998 which causes the scope unit not to be cleaned up
999 automatically after the process terminated.
1001 * tmpfiles learned a new --exclude-prefix= switch to exclude
1002 certain paths from operation.
1004 * journald will now automatically flush all messages to disk
1005 as soon as a message of the log priorities CRIT, ALERT or
1008 Contributions from: Andrew Cook, Brandon Philips, Christian
1009 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
1010 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
1011 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
1012 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
1013 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
1014 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
1015 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
1016 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
1017 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
1018 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
1019 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
1020 William Giokas, Zbigniew Jędrzejewski-Szmek
1022 -- Berlin, 2013-09-13
1026 * The documentation has been updated to cover the various new
1027 concepts introduced with 205.
1029 * Unit files now understand the new %v specifier which
1030 resolves to the kernel version string as returned by "uname
1033 * systemctl now supports filtering the unit list output by
1034 load state, active state and sub state, using the new
1037 * "systemctl status" will now show the results of the
1038 condition checks (like ConditionPathExists= and similar) of
1039 the last start attempts of the unit. They are also logged to
1042 * "journalctl -b" may now be used to look for boot output of a
1043 specific boot. Try "journalctl -b -1" for the previous boot,
1044 but the syntax is substantially more powerful.
1046 * "journalctl --show-cursor" has been added which prints the
1047 cursor string the last shown log line. This may then be used
1048 with the new "journalctl --after-cursor=" switch to continue
1049 browsing logs from that point on.
1051 * "journalctl --force" may now be used to force regeneration
1054 * Creation of "dead" device nodes has been moved from udev
1055 into kmod and tmpfiles. Previously, udev would read the kmod
1056 databases to pre-generate dead device nodes based on meta
1057 information contained in kernel modules, so that these would
1058 be auto-loaded on access rather then at boot. As this
1059 doesn't really have much to do with the exposing actual
1060 kernel devices to userspace this has always been slightly
1061 alien in the udev codebase. Following the new scheme kmod
1062 will now generate a runtime snippet for tmpfiles from the
1063 module meta information and it now is tmpfiles' job to the
1064 create the nodes. This also allows overriding access and
1065 other parameters for the nodes using the usual tmpfiles
1066 facilities. As side effect this allows us to remove the
1067 CAP_SYS_MKNOD capability bit from udevd entirely.
1069 * logind's device ACLs may now be applied to these "dead"
1070 devices nodes too, thus finally allowing managed access to
1071 devices such as /dev/snd/sequencer whithout loading the
1072 backing module right-away.
1074 * A new RPM macro has been added that may be used to apply
1075 tmpfiles configuration during package installation.
1077 * systemd-detect-virt and ConditionVirtualization= now can
1078 detect User-Mode-Linux machines (UML).
1080 * journald will now implicitly log the effective capabilities
1081 set of processes in the message metadata.
1083 * systemd-cryptsetup has gained support for TrueCrypt volumes.
1085 * The initrd interface has been simplified (more specifically,
1086 support for passing performance data via environment
1087 variables and fsck results via files in /run has been
1088 removed). These features were non-essential, and are
1089 nowadays available in a much nicer way by having systemd in
1090 the initrd serialize its state and have the hosts systemd
1091 deserialize it again.
1093 * The udev "keymap" data files and tools to apply keyboard
1094 specific mappings of scan to key codes, and force-release
1095 scan code lists have been entirely replaced by a udev
1096 "keyboard" builtin and a hwdb data file.
1098 * systemd will now honour the kernel's "quiet" command line
1099 argument also during late shutdown, resulting in a
1100 completely silent shutdown when used.
1102 * There's now an option to control the SO_REUSEPORT socket
1103 option in .socket units.
1105 * Instance units will now automatically get a per-template
1106 subslice of system.slice unless something else is explicitly
1107 configured. For example, instances of sshd@.service will now
1108 implicitly be placed in system-sshd.slice rather than
1109 system.slice as before.
1111 * Test coverage support may now be enabled at build time.
1113 Contributions from: Dave Reisner, Frederic Crozat, Harald
1114 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
1115 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
1116 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
1117 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
1118 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
1119 Giokas, Zbigniew Jędrzejewski-Szmek
1121 -- Berlin, 2013-07-23
1125 * Two new unit types have been introduced:
1127 Scope units are very similar to service units, however, are
1128 created out of pre-existing processes -- instead of PID 1
1129 forking off the processes. By using scope units it is
1130 possible for system services and applications to group their
1131 own child processes (worker processes) in a powerful way
1132 which then maybe used to organize them, or kill them
1133 together, or apply resource limits on them.
1135 Slice units may be used to partition system resources in an
1136 hierarchial fashion and then assign other units to them. By
1137 default there are now three slices: system.slice (for all
1138 system services), user.slice (for all user sessions),
1139 machine.slice (for VMs and containers).
1141 Slices and scopes have been introduced primarily in
1142 context of the work to move cgroup handling to a
1143 single-writer scheme, where only PID 1
1144 creates/removes/manages cgroups.
1146 * There's a new concept of "transient" units. In contrast to
1147 normal units these units are created via an API at runtime,
1148 not from configuration from disk. More specifically this
1149 means it is now possible to run arbitrary programs as
1150 independent services, with all execution parameters passed
1151 in via bus APIs rather than read from disk. Transient units
1152 make systemd substantially more dynamic then it ever was,
1153 and useful as a general batch manager.
1155 * logind has been updated to make use of scope and slice units
1156 for managing user sessions. As a user logs in he will get
1157 his own private slice unit, to which all sessions are added
1158 as scope units. We also added support for automatically
1159 adding an instance of user@.service for the user into the
1160 slice. Effectively logind will no longer create cgroup
1161 hierarchies on its own now, it will defer entirely to PID 1
1162 for this by means of scope, service and slice units. Since
1163 user sessions this way become entities managed by PID 1
1164 the output of "systemctl" is now a lot more comprehensive.
1166 * A new mini-daemon "systemd-machined" has been added which
1167 may be used by virtualization managers to register local
1168 VMs/containers. nspawn has been updated accordingly, and
1169 libvirt will be updated shortly. machined will collect a bit
1170 of meta information about the VMs/containers, and assign
1171 them their own scope unit (see above). The collected
1172 meta-data is then made available via the "machinectl" tool,
1173 and exposed in "ps" and similar tools. machined/machinectl
1174 is compile-time optional.
1176 * As discussed earlier, the low-level cgroup configuration
1177 options ControlGroup=, ControlGroupModify=,
1178 ControlGroupPersistent=, ControlGroupAttribute= have been
1179 removed. Please use high-level attribute settings instead as
1180 well as slice units.
1182 * A new bus call SetUnitProperties() has been added to alter
1183 various runtime parameters of a unit. This is primarily
1184 useful to alter cgroup parameters dynamically in a nice way,
1185 but will be extended later on to make more properties
1186 modifiable at runtime. systemctl gained a new set-properties
1187 command that wraps this call.
1189 * A new tool "systemd-run" has been added which can be used to
1190 run arbitrary command lines as transient services or scopes,
1191 while configuring a number of settings via the command
1192 line. This tool is currently very basic, however already
1193 very useful. We plan to extend this tool to even allow
1194 queuing of execution jobs with time triggers from the
1195 command line, similar in fashion to "at".
1197 * nspawn will now inform the user explicitly that kernels with
1198 audit enabled break containers, and suggest the user to turn
1201 * Support for detecting the IMA and AppArmor security
1202 frameworks with ConditionSecurity= has been added.
1204 * journalctl gained a new "-k" switch for showing only kernel
1205 messages, mimicking dmesg output; in addition to "--user"
1206 and "--system" switches for showing only user's own logs
1209 * systemd-delta can now show information about drop-in
1210 snippets extending unit files.
1212 * libsystemd-bus has been substantially updated but is still
1213 not available as public API.
1215 * systemd will now look for the "debug" argument on the kernel
1216 command line and enable debug logging, similar to
1217 "systemd.log_level=debug" already did before.
1219 * "systemctl set-default", "systemctl get-default" has been
1220 added to configure the default.target symlink, which
1221 controls what to boot into by default.
1223 * "systemctl set-log-level" has been added as a convenient
1224 way to raise and lower systemd logging threshold.
1226 * "systemd-analyze plot" will now show the time the various
1227 generators needed for execution, as well as information
1228 about the unit file loading.
1230 * libsystemd-journal gained a new sd_journal_open_files() call
1231 for opening specific journal files. journactl also gained a
1232 new switch to expose this new functionality. Previously we
1233 only supported opening all files from a directory, or all
1234 files from the system, as opening individual files only is
1235 racy due to journal file rotation.
1237 * systemd gained the new DefaultEnvironment= setting in
1238 /etc/systemd/system.conf to set environment variables for
1241 * If a privileged process logs a journal message with the
1242 OBJECT_PID= field set, then journald will automatically
1243 augment this with additional OBJECT_UID=, OBJECT_GID=,
1244 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
1245 system services want to log events about specific client
1246 processes. journactl/systemctl has been updated to make use
1247 of this information if all log messages regarding a specific
1250 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
1251 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
1252 Reisner, David Coppa, David King, David Strauss, Eelco
1253 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
1254 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
1255 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
1256 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
1257 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
1258 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
1259 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
1260 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
1261 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
1262 Łukasz Stelmach, 장동준
1266 * The Python bindings gained some minimal support for the APIs
1267 exposed by libsystemd-logind.
1269 * ConditionSecurity= gained support for detecting SMACK. Since
1270 this condition already supports SELinux and AppArmor we only
1271 miss IMA for this. Patches welcome!
1273 Contributions from: Karol Lewandowski, Lennart Poettering,
1274 Zbigniew Jędrzejewski-Szmek
1278 * systemd-nspawn will now create /etc/resolv.conf if
1279 necessary, before bind-mounting the host's file onto it.
1281 * systemd-nspawn will now store meta information about a
1282 container on the container's cgroup as extended attribute
1283 fields, including the root directory.
1285 * The cgroup hierarchy has been reworked in many ways. All
1286 objects any of the components systemd creates in the cgroup
1287 tree are now suffixed. More specifically, user sessions are
1288 now placed in cgroups suffixed with ".session", users in
1289 cgroups suffixed with ".user", and nspawn containers in
1290 cgroups suffixed with ".nspawn". Furthermore, all cgroup
1291 names are now escaped in a simple scheme to avoid collision
1292 of userspace object names with kernel filenames. This work
1293 is preparation for making these objects relocatable in the
1294 cgroup tree, in order to allow easy resource partitioning of
1295 these objects without causing naming conflicts.
1297 * systemctl list-dependencies gained the new switches
1298 --plain, --reverse, --after and --before.
1300 * systemd-inhibit now shows the process name of processes that
1301 have taken an inhibitor lock.
1303 * nss-myhostname will now also resolve "localhost"
1304 implicitly. This makes /etc/hosts an optional file and
1305 nicely handles that on IPv6 ::1 maps to both "localhost" and
1308 * libsystemd-logind.so gained a new call
1309 sd_get_machine_names() to enumerate running containers and
1310 VMs (currently only supported by very new libvirt and
1311 nspawn). sd_login_monitor can now be used to watch
1312 VMs/containers coming and going.
1314 * .include is not allowed recursively anymore, and only in
1315 unit files. Usually it is better to use drop-in snippets in
1316 .d/*.conf anyway, as introduced with systemd 198.
1318 * systemd-analyze gained a new "critical-chain" command that
1319 determines the slowest chain of units run during system
1320 boot-up. It is very useful for tracking down where
1321 optimizing boot time is the most beneficial.
1323 * systemd will no longer allow manipulating service paths in
1324 the name=systemd:/system cgroup tree using ControlGroup= in
1325 units. (But is still fine with it in all other dirs.)
1327 * There's a new systemd-nspawn@.service service file that may
1328 be used to easily run nspawn containers as system
1329 services. With the container's root directory in
1330 /var/lib/container/foobar it is now sufficient to run
1331 "systemctl start systemd-nspawn@foobar.service" to boot it.
1333 * systemd-cgls gained a new parameter "--machine" to list only
1334 the processes within a certain container.
1336 * ConditionSecurity= now can check for "apparmor". We still
1337 are lacking checks for SMACK and IMA for this condition
1338 check though. Patches welcome!
1340 * A new configuration file /etc/systemd/sleep.conf has been
1341 added that may be used to configure which kernel operation
1342 systemd is supposed to execute when "suspend", "hibernate"
1343 or "hybrid-sleep" is requested. This makes the new kernel
1344 "freeze" state accessible to the user.
1346 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
1347 the passed argument if applicable.
1349 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
1350 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
1351 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
1352 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
1353 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
1354 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
1355 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
1360 * The output of 'systemctl list-jobs' got some polishing. The
1361 '--type=' argument may now be passed more than once. A new
1362 command 'systemctl list-sockets' has been added which shows
1363 a list of kernel sockets systemd is listening on with the
1364 socket units they belong to, plus the units these socket
1367 * The experimental libsystemd-bus library got substantial
1368 updates to work in conjunction with the (also experimental)
1369 kdbus kernel project. It works well enough to exchange
1370 messages with some sophistication. Note that kdbus is not
1371 ready yet, and the library is mostly an elaborate test case
1372 for now, and not installable.
1374 * systemd gained a new unit 'systemd-static-nodes.service'
1375 that generates static device nodes earlier during boot, and
1376 can run in conjunction with udev.
1378 * libsystemd-login gained a new call sd_pid_get_user_unit()
1379 to retrieve the user systemd unit a process is running
1380 in. This is useful for systems where systemd is used as
1383 * systemd-nspawn now places all containers in the new /machine
1384 top-level cgroup directory in the name=systemd
1385 hierarchy. libvirt will soon do the same, so that we get a
1386 uniform separation of /system, /user and /machine for system
1387 services, user processes and containers/virtual
1388 machines. This new cgroup hierarchy is also useful to stick
1389 stable names to specific container instances, which can be
1390 recognized later this way (this name may be controlled
1391 via systemd-nspawn's new -M switch). libsystemd-login also
1392 gained a new call sd_pid_get_machine_name() to retrieve the
1393 name of the container/VM a specific process belongs to.
1395 * bootchart can now store its data in the journal.
1397 * libsystemd-journal gained a new call
1398 sd_journal_add_conjunction() for AND expressions to the
1399 matching logic. This can be used to express more complex
1400 logical expressions.
1402 * journactl can now take multiple --unit= and --user-unit=
1405 * The cryptsetup logic now understands the "luks.key=" kernel
1406 command line switch for specifying a file to read the
1407 decryption key from. Also, if a configured key file is not
1408 found the tool will now automatically fall back to prompting
1411 * Python systemd.journal module was updated to wrap recently
1412 added functions from libsystemd-journal. The interface was
1413 changed to bring the low level interface in s.j._Reader
1414 closer to the C API, and the high level interface in
1415 s.j.Reader was updated to wrap and convert all data about
1418 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
1419 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
1420 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
1421 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
1422 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
1423 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1427 * journalctl --update-catalog now understands a new --root=
1428 option to operate on catalogs found in a different root
1431 * During shutdown after systemd has terminated all running
1432 services a final killing loop kills all remaining left-over
1433 processes. We will now print the name of these processes
1434 when we send SIGKILL to them, since this usually indicates a
1437 * If /etc/crypttab refers to password files stored on
1438 configured mount points automatic dependencies will now be
1439 generated to ensure the specific mount is established first
1440 before the key file is attempted to be read.
1442 * 'systemctl status' will now show information about the
1443 network sockets a socket unit is listening on.
1445 * 'systemctl status' will also shown information about any
1446 drop-in configuration file for units. (Drop-In configuration
1447 files in this context are files such as
1448 /etc/systemd/systemd/foobar.service.d/*.conf)
1450 * systemd-cgtop now optionally shows summed up CPU times of
1451 cgroups. Press '%' while running cgtop to switch between
1452 percentage and absolute mode. This is useful to determine
1453 which cgroups use up the most CPU time over the entire
1454 runtime of the system. systemd-cgtop has also been updated
1455 to be 'pipeable' for processing with further shell tools.
1457 * 'hostnamectl set-hostname' will now allow setting of FQDN
1460 * The formatting and parsing of time span values has been
1461 changed. The parser now understands fractional expressions
1462 such as "5.5h". The formatter will now output fractional
1463 expressions for all time spans under 1min, i.e. "5.123456s"
1464 rather than "5s 123ms 456us". For time spans under 1s
1465 millisecond values are shown, for those under 1ms
1466 microsecond values are shown. This should greatly improve
1467 all time-related output of systemd.
1469 * libsystemd-login and libsystemd-journal gained new
1470 functions for querying the poll() events mask and poll()
1471 timeout value for integration into arbitrary event
1474 * localectl gained the ability to list available X11 keymaps
1475 (models, layouts, variants, options).
1477 * 'systemd-analyze dot' gained the ability to filter for
1478 specific units via shell-style globs, to create smaller,
1479 more useful graphs. I.e. it's now possible to create simple
1480 graphs of all the dependencies between only target units, or
1481 of all units that Avahi has dependencies with.
1483 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
1484 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
1485 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
1486 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
1487 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
1488 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
1489 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
1493 * The boot-time readahead implementation for rotating media
1494 will now read the read-ahead data in multiple passes which
1495 consist of all read requests made in equidistant time
1496 intervals. This means instead of strictly reading read-ahead
1497 data in its physical order on disk we now try to find a
1498 middle ground between physical and access time order.
1500 * /etc/os-release files gained a new BUILD_ID= field for usage
1501 on operating systems that provide continuous builds of OS
1504 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
1505 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
1506 William Douglas, Zbigniew Jędrzejewski-Szmek
1510 * systemd-python gained an API exposing libsystemd-daemon.
1512 * The SMACK setup logic gained support for uploading CIPSO
1515 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
1516 ReadOnlyDirectories= and InaccessibleDirectories= has
1517 changed. The private /tmp and /var/tmp directories are now
1518 shared by all processes of a service (which means
1519 ExecStartPre= may now leave data in /tmp that ExecStart= of
1520 the same service can still access). When a service is
1521 stopped its temporary directories are immediately deleted
1522 (normal clean-up with tmpfiles is still done in addition to
1525 * By default, systemd will now set a couple of sysctl
1526 variables in the kernel: the safe sysrq options are turned
1527 on, IP route verification is turned on, and source routing
1528 disabled. The recently added hardlink and softlink
1529 protection of the kernel is turned on. These settings should
1530 be reasonably safe, and good defaults for all new systems.
1532 * The predictable network naming logic may now be turned off
1533 with a new kernel command line switch: net.ifnames=0.
1535 * A new libsystemd-bus module has been added that implements a
1536 pretty complete D-Bus client library. For details see:
1538 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
1540 * journald will now explicitly flush the journal files to disk
1541 at the latest 5min after each write. The file will then also
1542 be marked offline until the next write. This should increase
1543 reliability in case of a crash. The synchronization delay
1544 can be configured via SyncIntervalSec= in journald.conf.
1546 * There's a new remote-fs-setup.target unit that can be used
1547 to pull in specific services when at least one remote file
1548 system is to be mounted.
1550 * There are new targets timers.target and paths.target as
1551 canonical targets to pull user timer and path units in
1552 from. This complements sockets.target with a similar
1553 purpose for socket units.
1555 * libudev gained a new call udev_device_set_attribute_value()
1556 to set sysfs attributes of a device.
1558 * The udev daemon now sets the default number of worker
1559 processes executed in parallel based on the number of available
1560 CPUs instead of the amount of available RAM. This is supposed
1561 to provide a more reliable default and limit a too aggressive
1562 paralellism for setups with 1000s of devices connected.
1564 Contributions from: Auke Kok, Colin Walters, Cristian
1565 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
1566 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
1567 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
1568 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
1569 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
1570 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
1571 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
1572 Zbigniew Jędrzejewski-Szmek
1576 * Configuration of unit files may now be extended via drop-in
1577 files without having to edit/override the unit files
1578 themselves. More specifically, if the administrator wants to
1579 change one value for a service file foobar.service he can
1580 now do so by dropping in a configuration snippet into
1581 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
1582 will load all these snippets and apply them on top of the
1583 main unit configuration file, possibly extending or
1584 overriding its settings. Using these drop-in snippets is
1585 generally nicer than the two earlier options for changing
1586 unit files locally: copying the files from
1587 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
1588 them there; or creating a new file in /etc/systemd/system/
1589 that incorporates the original one via ".include". Drop-in
1590 snippets into these .d/ directories can be placed in any
1591 directory systemd looks for units in, and the usual
1592 overriding semantics between /usr/lib, /etc and /run apply
1595 * Most unit file settings which take lists of items can now be
1596 reset by assigning the empty string to them. For example,
1597 normally, settings such as Environment=FOO=BAR append a new
1598 environment variable assignment to the environment block,
1599 each time they are used. By assigning Environment= the empty
1600 string the environment block can be reset to empty. This is
1601 particularly useful with the .d/*.conf drop-in snippets
1602 mentioned above, since this adds the ability to reset list
1603 settings from vendor unit files via these drop-ins.
1605 * systemctl gained a new "list-dependencies" command for
1606 listing the dependencies of a unit recursively.
1608 * Inhibitors are now honored and listed by "systemctl
1609 suspend", "systemctl poweroff" (and similar) too, not only
1610 GNOME. These commands will also list active sessions by
1613 * Resource limits (as exposed by the various control group
1614 controllers) can now be controlled dynamically at runtime
1615 for all units. More specifically, you can now use a command
1616 like "systemctl set-cgroup-attr foobar.service cpu.shares
1617 2000" to alter the CPU shares a specific service gets. These
1618 settings are stored persistently on disk, and thus allow the
1619 administrator to easily adjust the resource usage of
1620 services with a few simple commands. This dynamic resource
1621 management logic is also available to other programs via the
1622 bus. Almost any kernel cgroup attribute and controller is
1625 * systemd-vconsole-setup will now copy all font settings to
1626 all allocated VTs, where it previously applied them only to
1629 * libsystemd-login gained the new sd_session_get_tty() API
1632 * This release drops support for a few legacy or
1633 distribution-specific LSB facility names when parsing init
1634 scripts: $x-display-manager, $mail-transfer-agent,
1635 $mail-transport-agent, $mail-transfer-agent, $smtp,
1636 $null. Also, the mail-transfer-agent.target unit backing
1637 this has been removed. Distributions which want to retain
1638 compatibility with this should carry the burden for
1639 supporting this themselves and patch support for these back
1640 in, if they really need to. Also, the facilities $syslog and
1641 $local_fs are now ignored, since systemd does not support
1642 early-boot LSB init scripts anymore, and these facilities
1643 are implied anyway for normal services. syslog.target has
1646 * There are new bus calls on PID1's Manager object for
1647 cancelling jobs, and removing snapshot units. Previously,
1648 both calls were only available on the Job and Snapshot
1651 * systemd-journal-gatewayd gained SSL support.
1653 * The various "environment" files, such as /etc/locale.conf
1654 now support continuation lines with a backslash ("\") as
1655 last character in the line, similar in style (but different)
1656 to how this is supported in shells.
1658 * For normal user processes the _SYSTEMD_USER_UNIT= field is
1659 now implicitly appended to every log entry logged. systemctl
1660 has been updated to filter by this field when operating on a
1661 user systemd instance.
1663 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
1664 CAP_AUDIT_CONTROL capabilities to the capabilities set for
1665 the container. This makes it easier to boot unmodified
1666 Fedora systems in a container, which however still requires
1667 audit=0 to be passed on the kernel command line. Auditing in
1668 kernel and userspace is unfortunately still too broken in
1669 context of containers, hence we recommend compiling it out
1670 of the kernel or using audit=0. Hopefully this will be fixed
1671 one day for good in the kernel.
1673 * nspawn gained the new --bind= and --bind-ro= parameters to
1674 bind mount specific directories from the host into the
1677 * nspawn will now mount its own devpts file system instance
1678 into the container, in order not to leak pty devices from
1679 the host into the container.
1681 * systemd will now read the firmware boot time performance
1682 information from the EFI variables, if the used boot loader
1683 supports this, and takes it into account for boot performance
1684 analysis via "systemd-analyze". This is currently supported
1685 only in conjunction with Gummiboot, but could be supported
1686 by other boot loaders too. For details see:
1688 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
1690 * A new generator has been added that automatically mounts the
1691 EFI System Partition (ESP) to /boot, if that directory
1692 exists, is empty, and no other file system has been
1693 configured to be mounted there.
1695 * logind will now send out PrepareForSleep(false) out
1696 unconditionally, after coming back from suspend. This may be
1697 used by applications as asynchronous notification for
1698 system resume events.
1700 * "systemctl unlock-sessions" has been added, that allows
1701 unlocking the screens of all user sessions at once, similar
1702 how "systemctl lock-sessions" already locked all users
1703 sessions. This is backed by a new D-Bus call UnlockSessions().
1705 * "loginctl seat-status" will now show the master device of a
1706 seat. (i.e. the device of a seat that needs to be around for
1707 the seat to be considered available, usually the graphics
1710 * tmpfiles gained a new "X" line type, that allows
1711 configuration of files and directories (with wildcards) that
1712 shall be excluded from automatic cleanup ("aging").
1714 * udev default rules set the device node permissions now only
1715 at "add" events, and do not change them any longer with a
1716 later "change" event.
1718 * The log messages for lid events and power/sleep keypresses
1719 now carry a message ID.
1721 * We now have a substantially larger unit test suite, but this
1722 continues to be work in progress.
1724 * udevadm hwdb gained a new --root= parameter to change the
1725 root directory to operate relative to.
1727 * logind will now issue a background sync() request to the kernel
1728 early at shutdown, so that dirty buffers are flushed to disk early
1729 instead of at the last moment, in order to optimize shutdown
1732 * A new bootctl tool has been added that is an interface for
1733 certain boot loader operations. This is currently a preview
1734 and is likely to be extended into a small mechanism daemon
1735 like timedated, localed, hostnamed, and can be used by
1736 graphical UIs to enumerate available boot options, and
1737 request boot into firmware operations.
1739 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
1740 the rest of the package. It also has been updated to work
1741 correctly in initrds.
1743 * Policykit previously has been runtime optional, and is now
1744 also compile time optional via a configure switch.
1746 * systemd-analyze has been reimplemented in C. Also "systemctl
1747 dot" has moved into systemd-analyze.
1749 * "systemctl status" with no further parameters will now print
1750 the status of all active or failed units.
1752 * Operations such as "systemctl start" can now be executed
1753 with a new mode "--irreversible" which may be used to queue
1754 operations that cannot accidentally be reversed by a later
1755 job queuing. This is by default used to make shutdown
1756 requests more robust.
1758 * The Python API of systemd now gained a new module for
1759 reading journal files.
1761 * A new tool kernel-install has been added that can install
1762 kernel images according to the Boot Loader Specification:
1764 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
1766 * Boot time console output has been improved to provide
1767 animated boot time output for hanging jobs.
1769 * A new tool systemd-activate has been added which can be used
1770 to test socket activation with, directly from the command
1771 line. This should make it much easier to test and debug
1772 socket activation in daemons.
1774 * journalctl gained a new "--reverse" (or -r) option to show
1775 journal output in reverse order (i.e. newest line first).
1777 * journalctl gained a new "--pager-end" (or -e) option to jump
1778 to immediately jump to the end of the journal in the
1779 pager. This is only supported in conjunction with "less".
1781 * journalctl gained a new "--user-unit=" option, that works
1782 similar to "--unit=" but filters for user units rather than
1785 * A number of unit files to ease adoption of systemd in
1786 initrds has been added. This moves some minimal logic from
1787 the various initrd implementations into systemd proper.
1789 * The journal files are now owned by a new group
1790 "systemd-journal", which exists specifically to allow access
1791 to the journal, and nothing else. Previously, we used the
1792 "adm" group for that, which however possibly covers more
1793 than just journal/log file access. This new group is now
1794 already used by systemd-journal-gatewayd to ensure this
1795 daemon gets access to the journal files and as little else
1796 as possible. Note that "make install" will also set FS ACLs
1797 up for /var/log/journal to give "adm" and "wheel" read
1798 access to it, in addition to "systemd-journal" which owns
1799 the journal files. We recommend that packaging scripts also
1800 add read access to "adm" + "wheel" to /var/log/journal, and
1801 all existing/future journal files. To normal users and
1802 administrators little changes, however packagers need to
1803 ensure to create the "systemd-journal" system group at
1804 package installation time.
1806 * The systemd-journal-gatewayd now runs as unprivileged user
1807 systemd-journal-gateway:systemd-journal-gateway. Packaging
1808 scripts need to create these system user/group at
1811 * timedated now exposes a new boolean property CanNTP that
1812 indicates whether a local NTP service is available or not.
1814 * systemd-detect-virt will now also detect xen PVs
1816 * The pstore file system is now mounted by default, if it is
1819 * In addition to the SELinux and IMA policies we will now also
1820 load SMACK policies at early boot.
1822 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
1823 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
1824 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
1825 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
1826 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
1827 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
1828 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
1829 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
1830 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
1831 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
1832 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
1833 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
1834 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
1835 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
1839 * Timer units now support calendar time events in addition to
1840 monotonic time events. That means you can now trigger a unit
1841 based on a calendar time specification such as "Thu,Fri
1842 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
1843 or fifth day of any month of the year 2013, given that it is
1844 a thursday or friday. This brings timer event support
1845 considerably closer to cron's capabilities. For details on
1846 the supported calendar time specification language see
1849 * udev now supports a number of different naming policies for
1850 network interfaces for predictable names, and a combination
1851 of these policies is now the default. Please see this wiki
1852 document for details:
1854 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
1856 * Auke Kok's bootchart implementation has been added to the
1857 systemd tree. It's an optional component that can graph the
1858 boot in quite some detail. It's one of the best bootchart
1859 implementations around and minimal in its code and
1862 * nss-myhostname has been integrated into the systemd source
1863 tree. nss-myhostname guarantees that the local hostname
1864 always stays resolvable via NSS. It has been a weak
1865 requirement of systemd-hostnamed since a long time, and
1866 since its code is actually trivial we decided to just
1867 include it in systemd's source tree. It can be turned off
1868 with a configure switch.
1870 * The read-ahead logic is now capable of properly detecting
1871 whether a btrfs file system is on SSD or rotating media, in
1872 order to optimize the read-ahead scheme. Previously, it was
1873 only capable of detecting this on traditional file systems
1876 * In udev, additional device properties are now read from the
1877 IAB in addition to the OUI database. Also, Bluetooth company
1878 identities are attached to the devices as well.
1880 * In service files %U may be used as specifier that is
1881 replaced by the configured user name of the service.
1883 * nspawn may now be invoked without a controlling TTY. This
1884 makes it suitable for invocation as its own service. This
1885 may be used to set up a simple containerized server system
1886 using only core OS tools.
1888 * systemd and nspawn can now accept socket file descriptors
1889 when they are started for socket activation. This enables
1890 implementation of socket activated nspawn
1891 containers. i.e. think about autospawning an entire OS image
1892 when the first SSH or HTTP connection is received. We expect
1893 that similar functionality will also be added to libvirt-lxc
1896 * journalctl will now suppress ANSI color codes when
1897 presenting log data.
1899 * systemctl will no longer show control group information for
1900 a unit if a the control group is empty anyway.
1902 * logind can now automatically suspend/hibernate/shutdown the
1905 * /etc/machine-info and hostnamed now also expose the chassis
1906 type of the system. This can be used to determine whether
1907 the local system is a laptop, desktop, handset or
1908 tablet. This information may either be configured by the
1909 user/vendor or is automatically determined from ACPI and DMI
1910 information if possible.
1912 * A number of PolicyKit actions are now bound together with
1913 "imply" rules. This should simplify creating UIs because
1914 many actions will now authenticate similar ones as well.
1916 * Unit files learnt a new condition ConditionACPower= which
1917 may be used to conditionalize a unit depending on whether an
1918 AC power source is connected or not, of whether the system
1919 is running on battery power.
1921 * systemctl gained a new "is-failed" verb that may be used in
1922 shell scripts and suchlike to check whether a specific unit
1923 is in the "failed" state.
1925 * The EnvironmentFile= setting in unit files now supports file
1926 globbing, and can hence be used to easily read a number of
1927 environment files at once.
1929 * systemd will no longer detect and recognize specific
1930 distributions. All distribution-specific #ifdeffery has been
1931 removed, systemd is now fully generic and
1932 distribution-agnostic. Effectively, not too much is lost as
1933 a lot of the code is still accessible via explicit configure
1934 switches. However, support for some distribution specific
1935 legacy configuration file formats has been dropped. We
1936 recommend distributions to simply adopt the configuration
1937 files everybody else uses now and convert the old
1938 configuration from packaging scripts. Most distributions
1939 already did that. If that's not possible or desirable,
1940 distributions are welcome to forward port the specific
1941 pieces of code locally from the git history.
1943 * When logging a message about a unit systemd will now always
1944 log the unit name in the message meta data.
1946 * localectl will now also discover system locale data that is
1947 not stored in locale archives, but directly unpacked.
1949 * logind will no longer unconditionally use framebuffer
1950 devices as seat masters, i.e. as devices that are required
1951 to be existing before a seat is considered preset. Instead,
1952 it will now look for all devices that are tagged as
1953 "seat-master" in udev. By default framebuffer devices will
1954 be marked as such, but depending on local systems other
1955 devices might be marked as well. This may be used to
1956 integrate graphics cards using closed source drivers (such
1957 as NVidia ones) more nicely into logind. Note however, that
1958 we recommend using the open source NVidia drivers instead,
1959 and no udev rules for the closed-source drivers will be
1960 shipped from us upstream.
1962 Contributions from: Adam Williamson, Alessandro Crismani, Auke
1963 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
1964 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
1965 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
1966 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
1967 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
1968 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
1969 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
1970 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
1971 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
1972 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
1973 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
1978 * udev gained support for loading additional device properties
1979 from an indexed database that is keyed by vendor/product IDs
1980 and similar device identifiers. For the beginning this
1981 "hwdb" is populated with data from the well-known PCI and
1982 USB database, but also includes PNP, ACPI and OID data. In
1983 the longer run this indexed database shall grow into
1984 becoming the one central database for non-essential
1985 userspace device metadata. Previously, data from the PCI/USB
1986 database was only attached to select devices, since the
1987 lookup was a relatively expensive operation due to O(n) time
1988 complexity (with n being the number of entries in the
1989 database). Since this is now O(1), we decided to add in this
1990 data for all devices where this is available, by
1991 default. Note that the indexed database needs to be rebuilt
1992 when new data files are installed. To achieve this you need
1993 to update your packaging scripts to invoke "udevadm hwdb
1994 --update" after installation of hwdb data files. For
1995 RPM-based distributions we introduced the new
1996 %udev_hwdb_update macro for this purpose.
1998 * The Journal gained support for the "Message Catalog", an
1999 indexed database to link up additional information with
2000 journal entries. For further details please check:
2002 http://www.freedesktop.org/wiki/Software/systemd/catalog
2004 The indexed message catalog database also needs to be
2005 rebuilt after installation of message catalog files. Use
2006 "journalctl --update-catalog" for this. For RPM-based
2007 distributions we introduced the %journal_catalog_update
2008 macro for this purpose.
2010 * The Python Journal bindings gained support for the standard
2011 Python logging framework.
2013 * The Journal API gained new functions for checking whether
2014 the underlying file system of a journal file is capable of
2015 properly reporting file change notifications, or whether
2016 applications that want to reflect journal changes "live"
2017 need to recheck journal files continuously in appropriate
2020 * It is now possible to set the "age" field for tmpfiles
2021 entries to 0, indicating that files matching this entry
2022 shall always be removed when the directories are cleaned up.
2024 * coredumpctl gained a new "gdb" verb which invokes gdb
2025 right-away on the selected coredump.
2027 * There's now support for "hybrid sleep" on kernels that
2028 support this, in addition to "suspend" and "hibernate". Use
2029 "systemctl hybrid-sleep" to make use of this.
2031 * logind's HandleSuspendKey= setting (and related settings)
2032 now gained support for a new "lock" setting to simply
2033 request the screen lock on all local sessions, instead of
2034 actually executing a suspend or hibernation.
2036 * systemd will now mount the EFI variables file system by
2039 * Socket units now gained support for configuration of the
2040 SMACK security label.
2042 * timedatectl will now output the time of the last and next
2043 daylight saving change.
2045 * We dropped support for various legacy and distro-specific
2046 concepts, such as insserv, early-boot SysV services
2047 (i.e. those for non-standard runlevels such as 'b' or 'S')
2048 or ArchLinux /etc/rc.conf support. We recommend the
2049 distributions who still need support this to either continue
2050 to maintain the necessary patches downstream, or find a
2051 different solution. (Talk to us if you have questions!)
2053 * Various systemd components will now bypass PolicyKit checks
2054 for root and otherwise handle properly if PolicyKit is not
2055 found to be around. This should fix most issues for
2056 PolicyKit-less systems. Quite frankly this should have been
2057 this way since day one. It is absolutely our intention to
2058 make systemd work fine on PolicyKit-less systems, and we
2059 consider it a bug if something doesn't work as it should if
2060 PolicyKit is not around.
2062 * For embedded systems it is now possible to build udev and
2063 systemd without blkid and/or kmod support.
2065 * "systemctl switch-root" is now capable of switching root
2066 more than once. I.e. in addition to transitions from the
2067 initrd to the host OS it is now possible to transition to
2068 further OS images from the host. This is useful to implement
2069 offline updating tools.
2071 * Various other additions have been made to the RPM macros
2072 shipped with systemd. Use %udev_rules_update() after
2073 installing new udev rules files. %_udevhwdbdir,
2074 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
2075 %_sysctldir are now available which resolve to the right
2076 directories for packages to place various data files in.
2078 * journalctl gained the new --full switch (in addition to
2079 --all, to disable ellipsation for long messages.
2081 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
2082 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
2083 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
2084 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
2085 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
2086 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
2087 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
2088 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
2089 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
2093 * journalctl gained new --since= and --until= switches to
2094 filter by time. It also now supports nice filtering for
2095 units via --unit=/-u.
2097 * Type=oneshot services may use ExecReload= and do the
2100 * The journal daemon now supports time-based rotation and
2101 vacuuming, in addition to the usual disk-space based
2104 * The journal will now index the available field values for
2105 each field name. This enables clients to show pretty drop
2106 downs of available match values when filtering. The bash
2107 completion of journalctl has been updated
2108 accordingly. journalctl gained a new switch -F to list all
2109 values a certain field takes in the journal database.
2111 * More service events are now written as structured messages
2112 to the journal, and made recognizable via message IDs.
2114 * The timedated, localed and hostnamed mini-services which
2115 previously only provided support for changing time, locale
2116 and hostname settings from graphical DEs such as GNOME now
2117 also have a minimal (but very useful) text-based client
2118 utility each. This is probably the nicest way to changing
2119 these settings from the command line now, especially since
2120 it lists available options and is fully integrated with bash
2123 * There's now a new tool "systemd-coredumpctl" to list and
2124 extract coredumps from the journal.
2126 * We now install a README each in /var/log/ and
2127 /etc/rc.d/init.d explaining where the system logs and init
2128 scripts went. This hopefully should help folks who go to
2129 that dirs and look into the otherwise now empty void and
2130 scratch their heads.
2132 * When user-services are invoked (by systemd --user) the
2133 $MANAGERPID env var is set to the PID of systemd.
2135 * SIGRTMIN+24 when sent to a --user instance will now result
2136 in immediate termination of systemd.
2138 * gatewayd received numerous feature additions such as a
2139 "follow" mode, for live syncing and filtering.
2141 * browse.html now allows filtering and showing detailed
2142 information on specific entries. Keyboard navigation and
2143 mouse screen support has been added.
2145 * gatewayd/journalctl now supports HTML5/JSON
2146 Server-Sent-Events as output.
2148 * The SysV init script compatibility logic will now
2149 heuristically determine whether a script supports the
2150 "reload" verb, and only then make this available as
2153 * "systemctl status --follow" has been removed, use "journalctl
2156 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
2157 have been removed since they are hardly useful to be
2160 * And I'd like to take the opportunity to specifically mention
2161 Zbigniew for his great contributions. Zbigniew, you rock!
2163 Contributions from: Andrew Eikum, Christian Hesse, Colin
2164 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
2165 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
2166 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
2167 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
2168 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
2169 Jędrzejewski-Szmek, Сковорода Никита Андреевич
2173 * If /etc/vconsole.conf is non-existent or empty we will no
2174 longer load any console font or key map at boot by
2175 default. Instead the kernel defaults will be left
2176 intact. This is definitely the right thing to do, as no
2177 configuration should mean no configuration, and hard-coding
2178 font names that are different on all archs is probably a bad
2179 idea. Also, the kernel default key map and font should be
2180 good enough for most cases anyway, and mostly identical to
2181 the userspace fonts/key maps we previously overloaded them
2182 with. If distributions want to continue to default to a
2183 non-kernel font or key map they should ship a default
2184 /etc/vconsole.conf with the appropriate contents.
2186 Contributions from: Colin Walters, Daniel J Walsh, Dave
2187 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
2188 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
2192 * journalctl gained a new --cursor= switch to show entries
2193 starting from the specified location in the journal.
2195 * We now enforce a size limit on journal entry fields exported
2196 with "-o json" in journalctl. Fields larger than 4K will be
2197 assigned null. This can be turned off with --all.
2199 * An (optional) journal gateway daemon is now available as
2200 "systemd-journal-gatewayd.service". This service provides
2201 access to the journal via HTTP and JSON. This functionality
2202 will be used to implement live log synchronization in both
2203 pull and push modes, but has various other users too, such
2204 as easy log access for debugging of embedded devices. Right
2205 now it is already useful to retrieve the journal via HTTP:
2207 # systemctl start systemd-journal-gatewayd.service
2208 # wget http://localhost:19531/entries
2210 This will download the journal contents in a
2211 /var/log/messages compatible format. The same as JSON:
2213 # curl -H"Accept: application/json" http://localhost:19531/entries
2215 This service is also accessible via a web browser where a
2216 single static HTML5 app is served that uses the JSON logic
2217 to enable the user to do some basic browsing of the
2218 journal. This will be extended later on. Here's an example
2219 screenshot of this app in its current state:
2221 http://0pointer.de/public/journal-gatewayd
2223 Contributions from: Kay Sievers, Lennart Poettering, Robert
2224 Milasan, Tom Gundersen
2228 * The bash completion logic is now available for journalctl
2231 * We don't mount the "cpuset" controller anymore together with
2232 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
2233 started if no parameters are assigned to it. "cpuset" hence
2234 broke code that assumed it it could create "cpu" groups and
2237 * journalctl -f will now subscribe to terminal size changes,
2238 and line break accordingly.
2240 Contributions from: Dave Reisner, Kay Sievers, Lennart
2241 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
2245 * nspawn will now create a symlink /etc/localtime in the
2246 container environment, copying the host's timezone
2247 setting. Previously this has been done via a bind mount, but
2248 since symlinks cannot be bind mounted this has now been
2249 changed to create/update the appropriate symlink.
2251 * journalctl -n's line number argument is now optional, and
2252 will default to 10 if omitted.
2254 * journald will now log the maximum size the journal files may
2255 take up on disk. This is particularly useful if the default
2256 built-in logic of determining this parameter from the file
2257 system size is used. Use "systemctl status
2258 systemd-journald.service" to see this information.
2260 * The multi-seat X wrapper tool has been stripped down. As X
2261 is now capable of enumerating graphics devices via udev in a
2262 seat-aware way the wrapper is not strictly necessary
2263 anymore. A stripped down temporary stop-gap is still shipped
2264 until the upstream display managers have been updated to
2265 fully support the new X logic. Expect this wrapper to be
2266 removed entirely in one of the next releases.
2268 * HandleSleepKey= in logind.conf has been split up into
2269 HandleSuspendKey= and HandleHibernateKey=. The old setting
2270 is not available anymore. X11 and the kernel are
2271 distuingishing between these keys and we should too. This
2272 also means the inhibition lock for these keys has been split
2275 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
2276 Poettering, Lukas Nykryn, Václav Pavlín
2280 * Whenever a unit changes state we'll now log this to the
2281 journal and show along the unit's own log output in
2284 * ConditionPathIsMountPoint= can now properly detect bind
2285 mount points too. (Previously, a bind mount of one file
2286 system to another place in the same file system could not be
2287 detected as mount, since they shared struct stat's st_dev
2290 * We will now mount the cgroup controllers cpu, cpuacct,
2291 cpuset and the controllers net_cls, net_prio together by
2294 * nspawn containers will now have a virtualized boot
2295 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
2296 over with a randomized ID at container initialization). This
2297 has the effect of making "journalctl -b" do the right thing
2300 * The JSON output journal serialization has been updated not
2301 to generate "endless" list objects anymore, but rather one
2302 JSON object per line. This is more in line how most JSON
2303 parsers expect JSON objects. The new output mode
2304 "json-pretty" has been added to provide similar output, but
2305 neatly aligned for readability by humans.
2307 * We dropped all explicit sync() invocations in the shutdown
2308 code. The kernel does this implicitly anyway in the kernel
2309 reboot() syscall. halt(8)'s -n option is now a compatibility
2312 * We now support virtualized reboot() in containers, as
2313 supported by newer kernels. We will fall back to exit() if
2314 CAP_SYS_REBOOT is not available to the container. Also,
2315 nspawn makes use of this now and will actually reboot the
2316 container if the containerized OS asks for that.
2318 * journalctl will only show local log output by default
2319 now. Use --merge (-m) to show remote log output, too.
2321 * libsystemd-journal gained the new sd_journal_get_usage()
2322 call to determine the current disk usage of all journal
2323 files. This is exposed in the new "journalctl --disk-usage"
2326 * journald gained a new configuration setting SplitMode= in
2327 journald.conf which may be used to control how user journals
2328 are split off. See journald.conf(5) for details.
2330 * A new condition type ConditionFileNotEmpty= has been added.
2332 * tmpfiles' "w" lines now support file globbing, to write
2333 multiple files at once.
2335 * We added Python bindings for the journal submission
2336 APIs. More Python APIs for a number of selected APIs will
2337 likely follow. Note that we intend to add native bindings
2338 only for the Python language, as we consider it common
2339 enough to deserve bindings shipped within systemd. There are
2340 various projects outside of systemd that provide bindings
2341 for languages such as PHP or Lua.
2343 * Many conditions will now resolve specifiers such as %i. In
2344 addition, PathChanged= and related directives of .path units
2345 now support specifiers as well.
2347 * There's now a new RPM macro definition for the system preset
2350 * journald will now warn if it can't forward a message to the
2351 syslog daemon because it's socket is full.
2353 * timedated will no longer write or process /etc/timezone,
2354 except on Debian. As we do not support late mounted /usr
2355 anymore /etc/localtime always being a symlink is now safe,
2356 and hence the information in /etc/timezone is not necessary
2359 * logind will now always reserve one VT for a text getty (VT6
2360 by default). Previously if more than 6 X sessions where
2361 started they took up all the VTs with auto-spawned gettys,
2362 so that no text gettys were available anymore.
2364 * udev will now automatically inform the btrfs kernel logic
2365 about btrfs RAID components showing up. This should make
2366 simple hotplug based btrfs RAID assembly work.
2368 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
2369 (but not for its children which will stay at the kernel
2370 default). This should allow setups with a lot more listening
2373 * systemd will now always pass the configured timezone to the
2374 kernel at boot. timedated will do the same when the timezone
2377 * logind's inhibition logic has been updated. By default,
2378 logind will now handle the lid switch, the power and sleep
2379 keys all the time, even in graphical sessions. If DEs want
2380 to handle these events on their own they should take the new
2381 handle-power-key, handle-sleep-key and handle-lid-switch
2382 inhibitors during their runtime. A simple way to achiveve
2383 that is to invoke the DE wrapped in an invocation of:
2385 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
2387 * Access to unit operations is now checked via SELinux taking
2388 the unit file label and client process label into account.
2390 * systemd will now notify the administrator in the journal
2391 when he over-mounts a non-empty directory.
2393 * There are new specifiers that are resolved in unit files,
2394 for the host name (%H), the machine ID (%m) and the boot ID
2397 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
2398 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
2399 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
2400 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
2401 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
2402 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
2403 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
2407 * Support for reading structured kernel messages from
2408 /dev/kmsg has now been added and is enabled by default.
2410 * Support for reading kernel messages from /proc/kmsg has now
2411 been removed. If you want kernel messages in the journal
2412 make sure to run a recent kernel (>= 3.5) that supports
2413 reading structured messages from /dev/kmsg (see
2414 above). /proc/kmsg is now exclusive property of classic
2415 syslog daemons again.
2417 * The libudev API gained the new
2418 udev_device_new_from_device_id() call.
2420 * The logic for file system namespace (ReadOnlyDirectory=,
2421 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
2422 require pivot_root() anymore. This means fewer temporary
2423 directories are created below /tmp for this feature.
2425 * nspawn containers will now see and receive all submounts
2426 made on the host OS below the root file system of the
2429 * Forward Secure Sealing is now supported for Journal files,
2430 which provide cryptographical sealing of journal files so
2431 that attackers cannot alter log history anymore without this
2432 being detectable. Lennart will soon post a blog story about
2433 this explaining it in more detail.
2435 * There are two new service settings RestartPreventExitStatus=
2436 and SuccessExitStatus= which allow configuration of exit
2437 status (exit code or signal) which will be excepted from the
2438 restart logic, resp. consider successful.
2440 * journalctl gained the new --verify switch that can be used
2441 to check the integrity of the structure of journal files and
2442 (if Forward Secure Sealing is enabled) the contents of
2445 * nspawn containers will now be run with /dev/stdin, /dev/fd/
2446 and similar symlinks pre-created. This makes running shells
2447 as container init process a lot more fun.
2449 * The fstab support can now handle PARTUUID= and PARTLABEL=
2452 * A new ConditionHost= condition has been added to match
2453 against the hostname (with globs) and machine ID. This is
2454 useful for clusters where a single OS image is used to
2455 provision a large number of hosts which shall run slightly
2456 different sets of services.
2458 * Services which hit the restart limit will now be placed in a
2461 Contributions from: Bertram Poettering, Dave Reisner, Huang
2462 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
2463 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
2467 * When running in --user mode systemd will now become a
2468 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
2469 tree a lot more organized.
2471 * A new PartOf= unit dependency type has been introduced that
2472 may be used to group services in a natural way.
2474 * "systemctl enable" may now be used to enable instances of
2477 * journalctl now prints error log levels in red, and
2478 warning/notice log levels in bright white. It also supports
2479 filtering by log level now.
2481 * cgtop gained a new -n switch (similar to top), to configure
2482 the maximum number of iterations to run for. It also gained
2483 -b, to run in batch mode (accepting no input).
2485 * The suffix ".service" may now be omitted on most systemctl
2486 command lines involving service unit names.
2488 * There's a new bus call in logind to lock all sessions, as
2489 well as a loginctl verb for it "lock-sessions".
2491 * libsystemd-logind.so gained a new call sd_journal_perror()
2492 that works similar to libc perror() but logs to the journal
2493 and encodes structured information about the error number.
2495 * /etc/crypttab entries now understand the new keyfile-size=
2498 * shutdown(8) now can send a (configurable) wall message when
2499 a shutdown is cancelled.
2501 * The mount propagation mode for the root file system will now
2502 default to "shared", which is useful to make containers work
2503 nicely out-of-the-box so that they receive new mounts from
2504 the host. This can be undone locally by running "mount
2505 --make-rprivate /" if needed.
2507 * The prefdm.service file has been removed. Distributions
2508 should maintain this unit downstream if they intend to keep
2509 it around. However, we recommend writing normal unit files
2510 for display managers instead.
2512 * Since systemd is a crucial part of the OS we will now
2513 default to a number of compiler switches that improve
2514 security (hardening) such as read-only relocations, stack
2515 protection, and suchlike.
2517 * The TimeoutSec= setting for services is now split into
2518 TimeoutStartSec= and TimeoutStopSec= to allow configuration
2519 of individual time outs for the start and the stop phase of
2522 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
2523 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
2524 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
2525 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
2526 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
2527 Gundersen, Zbigniew Jędrzejewski-Szmek
2531 * The journal and id128 C APIs are now fully documented as man
2534 * Extra safety checks have been added when transitioning from
2535 the initial RAM disk to the main system to avoid accidental
2538 * /etc/crypttab entries now understand the new keyfile-offset=
2541 * systemctl -t can now be used to filter by unit load state.
2543 * The journal C API gained the new sd_journal_wait() call to
2544 make writing synchronous journal clients easier.
2546 * journalctl gained the new -D switch to show journals from a
2549 * journalctl now displays a special marker between log
2550 messages of two different boots.
2552 * The journal is now explicitly flushed to /var via a service
2553 systemd-journal-flush.service, rather than implicitly simply
2554 by seeing /var/log/journal to be writable.
2556 * journalctl (and the journal C APIs) can now match for much
2557 more complex expressions, with alternatives and
2560 * When transitioning from the initial RAM disk to the main
2561 system we will now kill all processes in a killing spree to
2562 ensure no processes stay around by accident.
2564 * Three new specifiers may be used in unit files: %u, %h, %s
2565 resolve to the user name, user home directory resp. user
2566 shell. This is useful for running systemd user instances.
2568 * We now automatically rotate journal files if their data
2569 object hash table gets a fill level > 75%. We also size the
2570 hash table based on the configured maximum file size. This
2571 together should lower hash collisions drastically and thus
2572 speed things up a bit.
2574 * journalctl gained the new "--header" switch to introspect
2575 header data of journal files.
2577 * A new setting SystemCallFilters= has been added to services
2578 which may be used to apply blacklists or whitelists to
2579 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
2581 * nspawn gained a new --link-journal= switch (and quicker: -j)
2582 to link the container journal with the host. This makes it
2583 very easy to centralize log viewing on the host for all
2584 guests while still keeping the journal files separated.
2586 * Many bugfixes and optimizations
2588 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
2589 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
2590 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
2595 * Several tools now understand kernel command line arguments,
2596 which are only read when run in an initial RAM disk. They
2597 usually follow closely their normal counterparts, but are
2600 * There's a new tool to analyze the readahead files that are
2601 automatically generated at boot. Use:
2603 /usr/lib/systemd/systemd-readahead analyze /.readahead
2605 * We now provide an early debug shell on tty9 if this enabled. Use:
2607 systemctl enable debug-shell.service
2609 * All plymouth related units have been moved into the Plymouth
2610 package. Please make sure to upgrade your Plymouth version
2613 * systemd-tmpfiles now supports getting passed the basename of
2614 a configuration file only, in which case it will look for it
2615 in all appropriate directories automatically.
2617 * udevadm info now takes a /dev or /sys path as argument, and
2618 does the right thing. Example:
2620 udevadm info /dev/sda
2621 udevadm info /sys/class/block/sda
2623 * systemctl now prints a warning if a unit is stopped but a
2624 unit that might trigger it continues to run. Example: a
2625 service is stopped but the socket that activates it is left
2628 * "systemctl status" will now mention if the log output was
2629 shortened due to rotation since a service has been started.
2631 * The journal API now exposes functions to determine the
2632 "cutoff" times due to rotation.
2634 * journald now understands SIGUSR1 and SIGUSR2 for triggering
2635 immediately flushing of runtime logs to /var if possible,
2636 resp. for triggering immediate rotation of the journal
2639 * It is now considered an error if a service is attempted to
2640 be stopped that is not loaded.
2642 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
2644 * systemd-analyze now supports Python 3
2646 * tmpfiles now supports cleaning up directories via aging
2647 where the first level dirs are always kept around but
2648 directories beneath it automatically aged. This is enabled
2649 by prefixing the age field with '~'.
2651 * Seat objects now expose CanGraphical, CanTTY properties
2652 which is required to deal with very fast bootups where the
2653 display manager might be running before the graphics drivers
2654 completed initialization.
2656 * Seat objects now expose a State property.
2658 * We now include RPM macros for service enabling/disabling
2659 based on the preset logic. We recommend RPM based
2660 distributions to make use of these macros if possible. This
2661 makes it simpler to reuse RPM spec files across
2664 * We now make sure that the collected systemd unit name is
2665 always valid when services log to the journal via
2668 * There's a new man page kernel-command-line(7) detailing all
2669 command line options we understand.
2671 * The fstab generator may now be disabled at boot by passing
2672 fstab=0 on the kernel command line.
2674 * A new kernel command line option modules-load= is now understood
2675 to load a specific kernel module statically, early at boot.
2677 * Unit names specified on the systemctl command line are now
2678 automatically escaped as needed. Also, if file system or
2679 device paths are specified they are automatically turned
2680 into the appropriate mount or device unit names. Example:
2682 systemctl status /home
2683 systemctl status /dev/sda
2685 * The SysVConsole= configuration option has been removed from
2686 system.conf parsing.
2688 * The SysV search path is no longer exported on the D-Bus
2691 * The Names= option is been removed from unit file parsing.
2693 * There's a new man page bootup(7) detailing the boot process.
2695 * Every unit and every generator we ship with systemd now
2696 comes with full documentation. The self-explanatory boot is
2699 * A couple of services gained "systemd-" prefixes in their
2700 name if they wrap systemd code, rather than only external
2701 code. Among them fsck@.service which is now
2702 systemd-fsck@.service.
2704 * The HaveWatchdog property has been removed from the D-Bus
2707 * systemd.confirm_spawn= on the kernel command line should now
2710 * There's a new man page crypttab(5) which details all options
2711 we actually understand.
2713 * systemd-nspawn gained a new --capability= switch to pass
2714 additional capabilities to the container.
2716 * timedated will now read known NTP implementation unit names
2717 from /usr/lib/systemd/ntp-units.d/*.list,
2718 systemd-timedated-ntp.target has been removed.
2720 * journalctl gained a new switch "-b" that lists log data of
2721 the current boot only.
2723 * The notify socket is in the abstract namespace again, in
2724 order to support daemons which chroot() at start-up.
2726 * There is a new Storage= configuration option for journald
2727 which allows configuration of where log data should go. This
2728 also provides a way to disable journal logging entirely, so
2729 that data collected is only forwarded to the console, the
2730 kernel log buffer or another syslog implementation.
2732 * Many bugfixes and optimizations
2734 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
2735 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
2736 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
2737 Shawn Landden, Tom Gundersen
2741 * "systemctl help <unit>" now shows the man page if one is
2744 * Several new man pages have been added.
2746 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
2747 MaxLevelConsole= can now be specified in
2748 journald.conf. These options allow reducing the amount of
2749 data stored on disk or forwarded by the log level.
2751 * TimerSlackNSec= can now be specified in system.conf for
2752 PID1. This allows system-wide power savings.
2754 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
2755 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
2760 * logind is now capable of (optionally) handling power and
2761 sleep keys as well as the lid switch.
2763 * journalctl now understands the syntax "journalctl
2764 /usr/bin/avahi-daemon" to get all log output of a specific
2767 * CapabilityBoundingSet= in system.conf now also influences
2768 the capability bound set of usermode helpers of the kernel.
2770 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
2771 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
2772 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
2773 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
2777 * Note that we skipped 139 releases here in order to set the
2778 new version to something that is greater than both udev's
2779 and systemd's most recent version number.
2781 * udev: all udev sources are merged into the systemd source tree now.
2782 All future udev development will happen in the systemd tree. It
2783 is still fully supported to use the udev daemon and tools without
2784 systemd running, like in initramfs or other init systems. Building
2785 udev though, will require the *build* of the systemd tree, but
2786 udev can be properly *run* without systemd.
2788 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
2789 should be used to create dead device nodes as workarounds for broken
2792 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
2793 no longer supported. udev_monitor_new_from_netlink() needs to be
2794 used to subscribe to events.
2796 * udev: when udevd is started by systemd, processes which are left
2797 behind by forking them off of udev rules, are unconditionally cleaned
2798 up and killed now after the event handling has finished. Services or
2799 daemons must be started as systemd services. Services can be
2800 pulled-in by udev to get started, but they can no longer be directly
2801 forked by udev rules.
2803 * udev: the daemon binary is called systemd-udevd now and installed
2804 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
2805 to adapt to that, create symlink, or rename the binary after building
2808 * libudev no longer provides these symbols:
2809 udev_monitor_from_socket()
2810 udev_queue_get_failed_list_entry()
2811 udev_get_{dev,sys,run}_path()
2812 The versions number was bumped and symbol versioning introduced.
2814 * systemd-loginctl and systemd-journalctl have been renamed
2815 to loginctl and journalctl to match systemctl.
2817 * The config files: /etc/systemd/systemd-logind.conf and
2818 /etc/systemd/systemd-journald.conf have been renamed to
2819 logind.conf and journald.conf. Package updates should rename
2820 the files to the new names on upgrade.
2822 * For almost all files the license is now LGPL2.1+, changed
2823 from the previous GPL2.0+. Exceptions are some minor stuff
2824 of udev (which will be changed to LGPL2.1 eventually, too),
2825 and the MIT licensed sd-daemon.[ch] library that is suitable
2826 to be used as drop-in files.
2828 * systemd and logind now handle system sleep states, in
2829 particular suspending and hibernating.
2831 * logind now implements a sleep/shutdown/idle inhibiting logic
2832 suitable for a variety of uses. Soonishly Lennart will blog
2833 about this in more detail.
2835 * var-run.mount and var-lock.mount are no longer provided
2836 (which prevously bind mounted these directories to their new
2837 places). Distributions which have not converted these
2838 directories to symlinks should consider stealing these files
2839 from git history and add them downstream.
2841 * We introduced the Documentation= field for units and added
2842 this to all our shipped units. This is useful to make it
2843 easier to explore the boot and the purpose of the various
2846 * All smaller setup units (such as
2847 systemd-vconsole-setup.service) now detect properly if they
2848 are run in a container and are skipped when
2849 appropriate. This guarantees an entirely noise-free boot in
2850 Linux container environments such as systemd-nspawn.
2852 * A framework for implementing offline system updates is now
2853 integrated, for details see:
2854 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
2856 * A new service type Type=idle is available now which helps us
2857 avoiding ugly interleaving of getty output and boot status
2860 * There's now a system-wide CapabilityBoundingSet= option to
2861 globally reduce the set of capabilities for the
2862 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
2863 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
2864 even CAP_NET_ADMIN system-wide for secure systems.
2866 * There are now system-wide DefaultLimitXXX= options to
2867 globally change the defaults of the various resource limits
2868 for all units started by PID 1.
2870 * Harald Hoyer's systemd test suite has been integrated into
2871 systemd which allows easy testing of systemd builds in qemu
2872 and nspawn. (This is really awesome! Ask us for details!)
2874 * The fstab parser is now implemented as generator, not inside
2877 * systemctl will now warn you if .mount units generated from
2878 /etc/fstab are out of date due to changes in fstab that
2879 haven't been read by systemd yet.
2881 * systemd is now suitable for usage in initrds. Dracut has
2882 already been updated to make use of this. With this in place
2883 initrds get a slight bit faster but primarily are much
2884 easier to introspect and debug since "systemctl status" in
2885 the host system can be used to introspect initrd services,
2886 and the journal from the initrd is kept around too.
2888 * systemd-delta has been added, a tool to explore differences
2889 between user/admin configuration and vendor defaults.
2891 * PrivateTmp= now affects both /tmp and /var/tmp.
2893 * Boot time status messages are now much prettier and feature
2894 proper english language. Booting up systemd has never been
2897 * Read-ahead pack files now include the inode number of all
2898 files to pre-cache. When the inode changes the pre-caching
2899 is not attempted. This should be nicer to deal with updated
2900 packages which might result in changes of read-ahead
2903 * We now temporaritly lower the kernel's read_ahead_kb variable
2904 when collecting read-ahead data to ensure the kernel's
2905 built-in read-ahead does not add noise to our measurements
2906 of necessary blocks to pre-cache.
2908 * There's now RequiresMountsFor= to add automatic dependencies
2909 for all mounts necessary for a specific file system path.
2911 * MountAuto= and SwapAuto= have been removed from
2912 system.conf. Mounting file systems at boot has to take place
2915 * nspawn now learned a new switch --uuid= to set the machine
2916 ID on the command line.
2918 * nspawn now learned the -b switch to automatically search
2921 * vt102 is now the default TERM for serial TTYs, upgraded from
2924 * systemd-logind now works on VT-less systems.
2926 * The build tree has been reorganized. The individual
2927 components now have directories of their own.
2929 * A new condition type ConditionPathIsReadWrite= is now available.
2931 * nspawn learned the new -C switch to create cgroups for the
2932 container in other hierarchies.
2934 * We now have support for hardware watchdogs, configurable in
2937 * The scheduled shutdown logic now has a public API.
2939 * We now mount /tmp as tmpfs by default, but this can be
2940 masked and /etc/fstab can override it.
2942 * Since udisks doesn't make use of /media anymore we are not
2943 mounting a tmpfs on it anymore.
2945 * journalctl gained a new --local switch to only interleave
2946 locally generated journal files.
2948 * We can now load the IMA policy at boot automatically.
2950 * The GTK tools have been split off into a systemd-ui.
2952 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
2953 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
2954 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
2955 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
2956 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
2957 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
2958 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
2959 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
2960 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
2965 * This is mostly a bugfix release
2967 * Support optional initialization of the machine ID from the
2968 KVM or container configured UUID.
2970 * Support immediate reboots with "systemctl reboot -ff"
2972 * Show /etc/os-release data in systemd-analyze output
2974 * Many bugfixes for the journal, including endianness fixes and
2975 ensuring that disk space enforcement works
2977 * sd-login.h is C++ comptaible again
2979 * Extend the /etc/os-release format on request of the Debian
2982 * We now refuse non-UTF8 strings used in various configuration
2983 and unit files. This is done to ensure we don't pass invalid
2984 data over D-Bus or expose it elsewhere.
2986 * Register Mimo USB Screens as suitable for automatic seat
2989 * Read SELinux client context from journal clients in a race
2992 * Reorder configuration file lookup order. /etc now always
2993 overrides /run in order to allow the administrator to always
2994 and unconditionally override vendor supplied or
2995 automatically generated data.
2997 * The various user visible bits of the journal now have man
2998 pages. We still lack man pages for the journal API calls
3001 * We now ship all man pages in HTML format again in the
3004 Contributions from: Dave Reisner, Dirk Eibach, Frederic
3005 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
3006 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
3011 * This is mostly a bugfix release
3013 * systems lacking /etc/os-release are no longer supported.
3015 * Various functionality updates to libsystemd-login.so
3017 * Track class of PAM logins to distuingish greeters from
3020 Contributions from: Kay Sievers, Lennart Poettering, Michael
3025 * This is an important bugfix release for v41.
3027 * Building man pages is now optional which should be useful
3028 for those building systemd from git but unwilling to install
3031 * Watchdog support for supervising services is now usable. In
3032 a future release support for hardware watchdogs
3033 (i.e. /dev/watchdog) will be added building on this.
3035 * Service start rate limiting is now configurable and can be
3036 turned off per service. When a start rate limit is hit a
3037 reboot can automatically be triggered.
3039 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
3041 Contributions from: Benjamin Franzke, Bill Nottingham,
3042 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
3043 Schmidt, Michał Górny, Piotr Drąg
3047 * The systemd binary is installed /usr/lib/systemd/systemd now;
3048 An existing /sbin/init symlink needs to be adapted with the
3051 * The code that loads kernel modules has been ported to invoke
3052 libkmod directly, instead of modprobe. This means we do not
3053 support systems with module-init-tools anymore.
3055 * Watchdog support is now already useful, but still not
3058 * A new kernel command line option systemd.setenv= is
3059 understood to set system wide environment variables
3060 dynamically at boot.
3062 * We now limit the set of capabilities of systemd-journald.
3064 * We now set SIGPIPE to ignore by default, since it only is
3065 useful in shell pipelines, and has little use in general
3066 code. This can be disabled with IgnoreSIPIPE=no in unit
3069 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
3070 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
3075 * This is mostly a bugfix release
3077 * We now expose the reason why a service failed in the
3078 "Result" D-Bus property.
3080 * Rudimentary service watchdog support (will be completed over
3081 the next few releases.)
3083 * When systemd forks off in order execute some service we will
3084 now immediately changes its argv[0] to reflect which process
3085 it will execute. This is useful to minimize the time window
3086 with a generic argv[0], which makes bootcharts more useful
3088 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
3089 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
3090 Mike Kazantsev, Ray Strode
3094 * This is mostly a test release, but incorporates many
3097 * New systemd-cgtop tool to show control groups by their
3100 * Linking against libacl for ACLs is optional again. If
3101 disabled, support tracking device access for active logins
3102 goes becomes unavailable, and so does access to the user
3103 journals by the respective users.
3105 * If a group "adm" exists, journal files are automatically
3106 owned by them, thus allow members of this group full access
3107 to the system journal as well as all user journals.
3109 * The journal now stores the SELinux context of the logging
3110 client for all entries.
3112 * Add C++ inclusion guards to all public headers
3114 * New output mode "cat" in the journal to print only text
3115 messages, without any meta data like date or time.
3117 * Include tiny X server wrapper as a temporary stop-gap to
3118 teach XOrg udev display enumeration. This is used by display
3119 managers such as gdm, and will go away as soon as XOrg
3120 learned native udev hotplugging for display devices.
3122 * Add new systemd-cat tool for executing arbitrary programs
3123 with STDERR/STDOUT connected to the journal. Can also act as
3124 BSD logger replacement, and does so by default.
3126 * Optionally store all locally generated coredumps in the
3127 journal along with meta data.
3129 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
3130 writing short strings to files (for usage for /sys), and for
3131 creating symlinks, character and block device nodes.
3133 * New unit file option ControlGroupPersistent= to make cgroups
3134 persistent, following the mechanisms outlined in
3135 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
3137 * Support multiple local RTCs in a sane way
3139 * No longer monopolize IO when replaying readahead data on
3140 rotating disks, since we might starve non-file-system IO to
3141 death, since fanotify() will not see accesses done by blkid,
3144 * Don't show kernel threads in systemd-cgls anymore, unless
3145 requested with new -k switch.
3147 Contributions from: Dan Horák, Kay Sievers, Lennart
3148 Poettering, Michal Schmidt
3152 * This is mostly a test release, but incorporates many
3155 * The git repository moved to:
3156 git://anongit.freedesktop.org/systemd/systemd
3157 ssh://git.freedesktop.org/git/systemd/systemd
3159 * First release with the journal
3160 http://0pointer.de/blog/projects/the-journal.html
3162 * The journal replaces both systemd-kmsg-syslogd and
3163 systemd-stdout-bridge.
3165 * New sd_pid_get_unit() API call in libsystemd-logind
3167 * Many systemadm clean-ups
3169 * Introduce remote-fs-pre.target which is ordered before all
3170 remote mounts and may be used to start services before all
3173 * Added Mageia support
3175 * Add bash completion for systemd-loginctl
3177 * Actively monitor PID file creation for daemons which exit in
3178 the parent process before having finished writing the PID
3179 file in the daemon process. Daemons which do this need to be
3180 fixed (i.e. PID file creation must have finished before the
3181 parent exits), but we now react a bit more gracefully to them.
3183 * Add colourful boot output, mimicking the well-known output
3184 of existing distributions.
3186 * New option PassCredentials= for socket units, for
3187 compatibility with a recent kernel ABI breakage.
3189 * /etc/rc.local is now hooked in via a generator binary, and
3190 thus will no longer act as synchronization point during
3193 * systemctl list-unit-files now supports --root=.
3195 * systemd-tmpfiles now understands two new commands: z, Z for
3196 relabelling files according to the SELinux database. This is
3197 useful to apply SELinux labels to specific files in /sys,
3200 * Output of SysV services is now forwarded to both the console
3201 and the journal by default, not only just the console.
3203 * New man pages for all APIs from libsystemd-login.
3205 * The build tree got reorganized and a the build system is a
3206 lot more modular allowing embedded setups to specifically
3207 select the components of systemd they are interested in.
3209 * Support for Linux systems lacking the kernel VT subsystem is
3212 * configure's --with-rootdir= got renamed to
3213 --with-rootprefix= to follow the naming used by udev and
3216 * Unless specified otherwise we'll now install to /usr instead
3217 of /usr/local by default.
3219 * Processes with '@' in argv[0][0] are now excluded from the
3220 final shut-down killing spree, following the logic explained
3222 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
3224 * All processes remaining in a service cgroup when we enter
3225 the START or START_PRE states are now killed with
3226 SIGKILL. That means it is no longer possible to spawn
3227 background processes from ExecStart= lines (which was never
3228 supported anyway, and bad style).
3230 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
3231 reloading of units together.
3233 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
3234 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
3235 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
3236 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
3237 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek