1 systemd System and Service Manager
5 * A new component "systemd-networkd" has been added that can
6 be used to configure local network interfaces statically or
7 via DHCP. It is capable of bringing up bridges, VLANs, and
8 bonding. Currently, no hook-ups for interactive network
9 configuration are provided. Use this for your initrd,
10 container, embedded, or server setup if you need a simple,
11 yet powerful, network configuration solution. This
12 configuration subsystem is quite nifty, as it allows wildcard
13 hotplug matching in interfaces. For example, with a single
14 configuration snippet, you can configure that all Ethernet
15 interfaces showing up are automatically added to a bridge,
16 or similar. It supports link-sensing and more.
18 * A new tool "systemd-socket-proxyd" has been added which can
19 act as a bidirectional proxy for TCP sockets. This is
20 useful for adding socket activation support to services that
21 do not actually support socket activation, including virtual
22 machines and the like.
24 * Add a new tool to save/restore rfkill state on
27 * Save/restore state of keyboard backlights in addition to
28 display backlights on shutdown/boot.
30 * udev learned a new SECLABEL{} construct to label device
31 nodes with a specific security label when they appear. For
32 now, only SECLABEL{selinux} is supported, but the syntax is
33 prepared for additional security frameworks.
35 * udev gained a new scheme to configure link-level attributes
36 from files in /etc/systemd/network/*.link. These files can
37 match against MAC address, device path, driver name and type,
38 and will apply attributes like the naming policy, link speed,
39 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
40 address assignment policy (randomized, ...).
42 * The configuration of network interface naming rules for
43 "permanent interface names" has changed: a new NamePolicy=
44 setting in the [Link] section of .link files determines the
45 priority of possible naming schemes (onboard, slot, mac,
46 path). The default value of this setting is determined by
47 /usr/lib/net/links/99-default.link. Old
48 80-net-name-slot.rules udev configuration file has been
49 removed, so local configuration overriding this file should
50 be adapated to override 99-default.link instead.
52 * When the User= switch is used in a unit file, also
53 initialize $SHELL= based on the user database entry.
55 * systemd no longer depends on libdbus. All communication is
56 now done with sd-bus, systemd's low-level bus library
59 * kdbus support has been added to PID 1 itself. When kdbus is
60 enabled, this causes PID 1 to set up the system bus and
61 enable support for a new ".busname" unit type that
62 encapsulates bus name activation on kdbus. It works a little
63 bit like ".socket" units, except for bus names. A new
64 generator has been added that converts classic dbus1 service
65 activation files automatically into native systemd .busname
68 * sd-bus: add a light-weight vtable implementation that allows
69 defining objects on the bus with a simple static const
70 vtable array of its methods, signals and properties.
72 * systemd will not generate or install static dbus
73 introspection data anymore to /usr/share/dbus-1/interfaces,
74 as the precise format of these files is unclear, and
75 nothing makes use of it.
77 * A proxy daemon is now provided to proxy clients connecting
78 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
79 compatibility with classic D-Bus.
81 * A bus driver implementation has been added that supports the
82 classic D-Bus bus driver calls on kdbus, also for
83 compatibility purposes.
85 * A new API "sd-event.h" has been added that implements a
86 minimal event loop API built around epoll. It provides a
87 couple of features that direct epoll usage is lacking:
88 prioritization of events, scales to large numbers of timer
89 events, per-event timer slack (accuracy), system-wide
90 coalescing of timer events, exit handlers, watchdog
91 supervision support using systemd's sd_notify() API, child
94 * A new API "sd-rntl.h" has been added that provides an API
95 around the route netlink interface of the kernel, similar in
98 * A new API "sd-dhcp-client.h" has been added that provides a
99 small DHCPv4 client-side implementation. This is used by
102 * There is a new kernel command line option
103 "systemd.restore_state=0|1". When set to "0", none of the
104 systemd tools will restore saved runtime state to hardware
105 devices. More specifically, the rfkill and backlight states
108 * The FsckPassNo= compatibility option in mount/service units
109 has been removed. The fstab generator will now add the
110 necessary dependencies automatically, and does not require
111 PID1's support for that anymore.
113 * journalctl gained a new switch, --list-boots, that lists
114 recent boots with their times and boot IDs.
116 * The various tools like systemctl, loginctl, timedatectl,
117 busctl, systemd-run, ... have gained a new switch "-M" to
118 connect to a specific, local OS container (as direct
119 connection, without requiring SSH). This works on any
120 container that is registered with machined, such as those
121 created by libvirt-lxc or nspawn.
123 * systemd-run and systemd-analyze also gained support for "-H"
124 to connect to remote hosts via SSH. This is particularly
125 useful for systemd-run because it enables queuing of jobs
128 * machinectl gained a new command "login" to open a getty
129 login in any local container. This works with any container
130 that is registered with machined (such as those created by
131 libvirt-lxc or nspawn), and which runs systemd inside.
133 * machinectl gained a new "reboot" command that may be used to
134 trigger a reboot on a specific container that is registered
135 with machined. This works on any container that runs an init
138 * systemctl gained a new "list-timers" command to print a nice
139 listing of installed timer units with the times they elapse
142 * Alternative reboot() parameters may now be specified on the
143 "systemctl reboot" command line and are passed to the
144 reboot() system call.
146 * systemctl gained a new --job-mode= switch to configure the
147 mode to queue a job with. This is a more generic version of
148 --fail, --irreversible, and --ignore-dependencies, which are
149 still available but not advertised anymore.
151 * /etc/systemd/system.conf gained new settings to configure
152 various default timeouts of units, as well as the default
153 start limit interval and burst. These may still be overridden
156 * PID1 will now export on the bus profile data of the security
157 policy upload process (such as the SELinux policy upload to
160 * journald: when forwarding logs to the console, include
161 timestamps (following the setting in
162 /sys/module/printk/parameters/time).
164 * OnCalendar= in timer units now understands the special
165 strings "yearly" and "annually". (Both are equivalent)
167 * The accuracy of timer units is now configurable with the new
168 AccuracySec= setting. It defaults to 1min.
170 * A new dependency type JoinsNamespaceOf= has been added that
171 allows running two services within the same /tmp and network
172 namespace, if PrivateNetwork= or PrivateTmp= are used.
174 * A new command "cat" has been added to systemctl. It outputs
175 the original unit file of a unit, and concatenates the
176 contents of additional "drop-in" unit file snippets, so that
177 the full configuration is shown.
179 * systemctl now supports globbing on the various "list-xyz"
180 commands, like "list-units" or "list-sockets", as well as on
181 those commands which take multiple unit names.
183 * journalctl's --unit= switch gained support for globbing.
185 * All systemd daemons now make use of the watchdog logic so
186 that systemd automatically notices when they hang.
188 * If the $container_ttys environment variable is set,
189 getty-generator will automatically spawn a getty for each
190 listed tty. This is useful for container managers to request
191 login gettys to be spawned on as many ttys as needed.
193 * %h, %s, %U specifier support is not available anymore when
194 used in unit files for PID 1. This is because NSS calls are
195 not safe from PID 1. They stay available for --user
196 instances of systemd, and as special case for the root user.
198 * loginctl gained a new "--no-legend" switch to turn off output
201 * The "sd-login.h" API gained three new calls:
202 sd_session_is_remote(), sd_session_get_remote_user(),
203 sd_session_get_remote_host() to query information about
206 * The udev hardware database now also carries vendor/product
207 information of SDIO devices.
209 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
210 determine whether watchdog notifications are requested by
213 * Socket-activated per-connection services now include a
214 short description of the connection parameters in the
217 * tmpfiles gained a new "--boot" option. When this is not used,
218 only lines where the command character is not suffixed with
219 "!" are executed. When this option is specified, those
220 options are executed too. This partitions tmpfiles
221 directives into those that can be safely executed at any
222 time, and those which should be run only at boot (for
223 example, a line that creates /run/nologin).
225 * A new API "sd-resolve.h" has been added which provides a simple
226 asynchronous wrapper around glibc NSS host name resolution
227 calls, such as getaddrinfo(). In contrast to glibc's
228 getaddrinfo_a(), it does not use signals. In contrast to most
229 other asynchronous name resolution libraries, this one does
230 not reimplement DNS, but reuses NSS, so that alternate
231 host name resolution systems continue to work, such as mDNS,
232 LDAP, etc. This API is based on libasyncns, but it has been
233 cleaned up for inclusion in systemd.
235 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
236 "sd-daemon.h" are no longer found in individual libraries
237 libsystemd-journal.so, libsystemd-login.so,
238 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
239 merged them into a single library, libsystemd.so, which
240 provides all symbols. The reason for this is cyclic
241 dependencies, as these libraries tend to use each other's
242 symbols. So far, we've managed to workaround that by linking
243 a copy of a good part of our code into each of these
244 libraries again and again, which, however, makes certain
245 things hard to do, like sharing static variables. Also, it
246 substantially increases footprint. With this change, there
247 is only one library for the basic APIs systemd
248 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
249 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
250 library as well, however are subject to the --enable-kdbus
251 switch (see below). Note that "sd-dhcp-client.h" is not part
252 of this library (this is because it only consumes, never
253 provides, services of/to other APIs). To make the transition
254 easy from the separate libraries to the unified one, we
255 provide the --enable-compat-libs compile-time switch which
256 will generate stub libraries that are compatible with the
257 old ones but redirect all calls to the new one.
259 * All of the kdbus logic and the new APIs "sd-bus.h",
260 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
261 and "sd-utf8.h" are compile-time optional via the
262 "--enable-kdbus" switch, and they are not compiled in by
263 default. To make use of kdbus, you have to explicitly enable
264 the switch. Note however, that neither the kernel nor the
265 userspace API for all of this is considered stable yet. We
266 want to maintain the freedom to still change the APIs for
267 now. By specifying this build-time switch, you acknowledge
268 that you are aware of the instability of the current
271 * Also, note that while kdbus is pretty much complete,
272 it lacks one thing: proper policy support. This means you
273 can build a fully working system with all features; however,
274 it will be highly insecure. Policy support will be added in
275 one of the next releases, at the same time that we will
276 declare the APIs stable.
278 * When the kernel command-line argument "kdbus" is specified,
279 systemd will automatically load the kdbus.ko kernel module. At
280 this stage of development, it is only useful for testing kdbus
281 and should not be used in production. Note: if "--enable-kdbus"
282 is specified, and the kdbus.ko kernel module is available, and
283 "kdbus" is added to the kernel command line, the entire system
284 runs with kdbus instead of dbus-daemon, with the above mentioned
285 problem of missing the system policy enforcement. Also a future
286 version of kdbus.ko or a newer systemd will not be compatible with
287 each other, and will unlikely be able to boot the machine if only
288 one of them is updated.
290 * systemctl gained a new "import-environment" command which
291 uploads the caller's environment (or parts thereof) into the
292 service manager so that it is inherited by services started
293 by the manager. This is useful to upload variables like
294 $DISPLAY into the user service manager.
296 * A new PrivateDevices= switch has been added to service units
297 which allows running a service with a namespaced /dev
298 directory that does not contain any device nodes for
299 physical devices. More specifically, it only includes devices
300 such as /dev/null, /dev/urandom, and /dev/zero which are API
303 * logind has been extended to support behaviour like VT
304 switching on seats that do not support a VT. This makes
305 multi-session available on seats that are not the first seat
306 (seat0), and on systems where kernel support for VTs has
307 been disabled at compile-time.
309 * If a process holds a delay lock for system sleep or shutdown
310 and fails to release it in time, we will now log its
311 identity. This makes it easier to identify processes that
312 cause slow suspends or power-offs.
314 * When parsing /etc/crypttab, support for a new key-slot=
315 option as supported by Debian is added. It allows indicating
316 which LUKS slot to use on disk, speeding up key loading.
318 * The sd_journald_sendv() API call has been checked and
319 officially declared to be async-signal-safe so that it may
320 be invoked from signal handlers for logging purposes.
322 * Boot-time status output is now enabled automatically after a
323 short timeout if boot does not progress, in order to give
324 the user an indication what she or he is waiting for.
326 * The boot-time output has been improved to show how much time
327 remains until jobs expire.
329 * The KillMode= switch in service units gained a new possible
330 value "mixed". If set, and the unit is shut down, then the
331 initial SIGTERM signal is sent only to the main daemon
332 process, while the following SIGKILL signal is sent to
333 all remaining processes of the service.
335 * When a scope unit is registered, a new property "Controller"
336 may be set. If set to a valid bus name, systemd will send a
337 RequestStop() signal to this name when it would like to shut
338 down the scope. This may be used to hook manager logic into
339 the shutdown logic of scope units. Also, scope units may now
340 be put in a special "abandoned" state, in which case the
341 manager process which created them takes no further
342 responsibilities for it.
344 * When reading unit files, systemd will now verify
345 the access mode of these files, and warn about certain
346 suspicious combinations. This has been added to make it
347 easier to track down packaging bugs where unit files are
348 marked executable or world-writable.
350 * systemd-nspawn gained a new "--setenv=" switch to set
351 container-wide environment variables. The similar option in
352 systemd-activate was renamed from "--environment=" to
353 "--setenv=" for consistency.
355 * systemd-nspawn has been updated to create a new kdbus domain
356 for each container that is invoked, thus allowing each
357 container to have its own set of system and user buses,
358 independent of the host.
360 * systemd-nspawn gained a new --drop-capability= switch to run
361 the container with less capabilities than the default. Both
362 --drop-capability= and --capability= now take the special
363 string "all" for dropping or keeping all capabilities.
365 * systemd-nspawn gained new switches for executing containers
366 with specific SELinux labels set.
368 * systemd-nspawn gained a new --quiet switch to not generate
369 any additional output but the container's own console
372 * systemd-nspawn gained a new --share-system switch to run a
373 container without PID namespacing enabled.
375 * systemd-nspawn gained a new --register= switch to control
376 whether the container is registered with systemd-machined or
377 not. This is useful for containers that do not run full
378 OS images, but only specific apps.
380 * systemd-nspawn gained a new --keep-unit which may be used
381 when invoked as the only program from a service unit, and
382 results in registration of the unit service itself in
383 systemd-machined, instead of a newly opened scope unit.
385 * systemd-nspawn gained a new --network-interface= switch for
386 moving arbitrary interfaces to the container. The new
387 --network-veth switch creates a virtual Ethernet connection
388 between host and container. The new --network-bridge=
389 switch then allows assigning the host side of this virtual
390 Ethernet connection to a bridge device.
392 * systemd-nspawn gained a new --personality= switch for
393 setting the kernel personality for the container. This is
394 useful when running a 32bit container on a 64bit host. A
395 similar option Personality= is now also available in service
398 * logind will now also track a "Desktop" identifier for each
399 session which encodes the desktop environment of it. This is
400 useful for desktop environments that want to identify
401 multiple running sessions of itself easily.
403 * A new SELinuxContext= setting for service units has been
404 added that allows setting a specific SELinux execution
405 context for a service.
407 * Most systemd client tools will now honour $SYSTEMD_LESS for
408 settings of the "less" pager. By default, these tools will
409 override $LESS to allow certain operations to work, such as
410 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
411 influence this logic.
413 * systemd's "seccomp" hook-up has been changed to make use of
414 the libseccomp library instead of using its own
415 implementation. This has benefits for portability among
418 * For usage together with SystemCallFilter=, a new
419 SystemCallErrorNumber= setting has been introduced that
420 allows configuration of a system error number to return on
421 filtered system calls, instead of immediately killing the
422 process. Also, SystemCallArchitectures= has been added to
423 limit access to system calls of a particular architecture
424 (in order to turn off support for unused secondary
425 architectures). There is also a global
426 SystemCallArchitectures= setting in system.conf now to turn
427 off support for non-native system calls system-wide.
429 * systemd requires a kernel with a working name_to_handle_at(),
430 please see the kernel config requirements in the README file.
432 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
433 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
434 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
435 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
436 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
437 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
438 Elia Pinto, Florian Weimer, George McCollister, Goffredo
439 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
440 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
441 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
442 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
443 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
444 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
445 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
446 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
447 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
448 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
449 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
450 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
451 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
452 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
453 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
454 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
455 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
457 -- Berlin, 2014-02-20
461 * logind has gained support for facilitating privileged input
462 and drm device access for unprivileged clients. This work is
463 useful to allow Wayland display servers (and similar
464 programs, such as kmscon) to run under the user's ID and
465 access input and drm devices which are normally
466 protected. When this is used (and the kernel is new enough)
467 logind will "mute" IO on the file descriptors passed to
468 Wayland as long as it is in the background and "unmute" it
469 if it returns into the foreground. This allows secure
470 session switching without allowing background sessions to
471 eavesdrop on input and display data. This also introduces
472 session switching support if VT support is turned off in the
473 kernel, and on seats that are not seat0.
475 * A new kernel command line option luks.options= is understood
476 now which allows specifiying LUKS options for usage for LUKS
477 encrypted partitions specified with luks.uuid=.
479 * tmpfiles.d(5) snippets may now use specifier expansion in
480 path names. More specifically %m, %b, %H, %v, are now
481 replaced by the local machine id, boot id, hostname, and
482 kernel version number.
484 * A new tmpfiles.d(5) command "m" has been introduced which
485 may be used to change the owner/group/access mode of a file
486 or directory if it exists, but do nothing if it doesn't.
488 * This release removes high-level support for the
489 MemorySoftLimit= cgroup setting. The underlying kernel
490 cgroup attribute memory.soft_limit= is currently badly
491 designed and likely to be removed from the kernel API in its
492 current form, hence we shouldn't expose it for now.
494 * The memory.use_hierarchy cgroup attribute is now enabled for
495 all cgroups systemd creates in the memory cgroup
496 hierarchy. This option is likely to be come the built-in
497 default in the kernel anyway, and the non-hierarchial mode
498 never made much sense in the intrinsically hierarchial
501 * A new field _SYSTEMD_SLICE= is logged along with all journal
502 messages containing the slice a message was generated
503 from. This is useful to allow easy per-customer filtering of
504 logs among other things.
506 * systemd-journald will no longer adjust the group of journal
507 files it creates to the "systemd-journal" group. Instead we
508 rely on the journal directory to be owned by the
509 "systemd-journal" group, and its setgid bit set, so that the
510 kernel file system layer will automatically enforce that
511 journal files inherit this group assignment. The reason for
512 this change is that we cannot allow NSS look-ups from
513 journald which would be necessary to resolve
514 "systemd-journal" to a numeric GID, because this might
515 create deadlocks if NSS involves synchronous queries to
516 other daemons (such as nscd, or sssd) which in turn are
517 logging clients of journald and might block on it, which
518 would then dead lock. A tmpfiles.d(5) snippet included in
519 systemd will make sure the setgid bit and group are
520 properly set on the journal directory if it exists on every
521 boot. However, we recommend adjusting it manually after
522 upgrades too (or from RPM scriptlets), so that the change is
523 not delayed until next reboot.
525 * Backlight and random seed files in /var/lib/ have moved into
526 the /var/lib/systemd/ directory, in order to centralize all
527 systemd generated files in one directory.
529 * Boot time performance measurements (as displayed by
530 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
531 performance information if that's available to determine how
532 much time BIOS and boot loader initialization required. With
533 a sufficiently new BIOS you hence no longer need to boot
534 with Gummiboot to get access to such information.
536 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
537 Cristian Rodríguez, Dave Reisner, David Herrmann, David
538 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
539 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
540 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
541 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
542 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
544 -- Berlin, 2013-10-02
548 * The Restart= option for services now understands a new
549 on-watchdog setting, which will restart the service
550 automatically if the service stops sending out watchdog keep
551 alive messages (as configured with WatchdogSec=).
553 * The getty generator (which is responsible for bringing up a
554 getty on configured serial consoles) will no longer only
555 start a getty on the primary kernel console but on all
556 others, too. This makes the order in which console= is
557 specified on the kernel command line less important.
559 * libsystemd-logind gained a new sd_session_get_vt() call to
560 retrieve the VT number of a session.
562 * If the option "tries=0" is set for an entry of /etc/crypttab
563 its passphrase is queried indefinitely instead of any
564 maximum number of tries.
566 * If a service with a configure PID file terminates its PID
567 file will now be removed automatically if it still exists
568 afterwards. This should put an end to stale PID files.
570 * systemd-run will now also take relative binary path names
571 for execution and no longer insists on absolute paths.
573 * InaccessibleDirectories= and ReadOnlyDirectories= now take
574 paths that are optionally prefixed with "-" to indicate that
575 it should not be considered a failure if they don't exist.
577 * journalctl -o (and similar commands) now understands a new
578 output mode "short-precise", it is similar to "short" but
579 shows timestamps with usec accuracy.
581 * The option "discard" (as known from Debian) is now
582 synonymous to "allow-discards" in /etc/crypttab. In fact,
583 "discard" is preferred now (since it is easier to remember
586 * Some licensing clean-ups were made, so that more code is now
587 LGPL-2.1 licensed than before.
589 * A minimal tool to save/restore the display backlight
590 brightness across reboots has been added. It will store the
591 backlight setting as late as possible at shutdown, and
592 restore it as early as possible during reboot.
594 * A logic to automatically discover and enable home and swap
595 partitions on GPT disks has been added. With this in place
596 /etc/fstab becomes optional for many setups as systemd can
597 discover certain partitions located on the root disk
598 automatically. Home partitions are recognized under their
599 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
600 partitions are recognized under their GPT type ID
601 0657fd6da4ab43c484e50933c84b4f4f.
603 * systemd will no longer pass any environment from the kernel
604 or initrd to system services. If you want to set an
605 environment for all services, do so via the kernel command
606 line systemd.setenv= assignment.
608 * The systemd-sysctl tool no longer natively reads the file
609 /etc/sysctl.conf. If desired, the file should be symlinked
610 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
611 legacy support by a symlink rather than built-in code, it
612 also makes the otherwise hidden order of application of the
613 different files visible. (Note that this partly reverts to a
614 pre-198 application order of sysctl knobs!)
616 * The "systemctl set-log-level" and "systemctl dump" commands
617 have been moved to systemd-analyze.
619 * systemd-run learned the new --remain-after-exit switch,
620 which causes the scope unit not to be cleaned up
621 automatically after the process terminated.
623 * tmpfiles learned a new --exclude-prefix= switch to exclude
624 certain paths from operation.
626 * journald will now automatically flush all messages to disk
627 as soon as a message of the log priorities CRIT, ALERT or
630 Contributions from: Andrew Cook, Brandon Philips, Christian
631 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
632 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
633 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
634 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
635 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
636 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
637 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
638 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
639 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
640 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
641 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
642 William Giokas, Zbigniew Jędrzejewski-Szmek
644 -- Berlin, 2013-09-13
648 * The documentation has been updated to cover the various new
649 concepts introduced with 205.
651 * Unit files now understand the new %v specifier which
652 resolves to the kernel version string as returned by "uname
655 * systemctl now supports filtering the unit list output by
656 load state, active state and sub state, using the new
659 * "systemctl status" will now show the results of the
660 condition checks (like ConditionPathExists= and similar) of
661 the last start attempts of the unit. They are also logged to
664 * "journalctl -b" may now be used to look for boot output of a
665 specific boot. Try "journalctl -b -1" for the previous boot,
666 but the syntax is substantially more powerful.
668 * "journalctl --show-cursor" has been added which prints the
669 cursor string the last shown log line. This may then be used
670 with the new "journalctl --after-cursor=" switch to continue
671 browsing logs from that point on.
673 * "journalctl --force" may now be used to force regeneration
676 * Creation of "dead" device nodes has been moved from udev
677 into kmod and tmpfiles. Previously, udev would read the kmod
678 databases to pre-generate dead device nodes based on meta
679 information contained in kernel modules, so that these would
680 be auto-loaded on access rather then at boot. As this
681 doesn't really have much to do with the exposing actual
682 kernel devices to userspace this has always been slightly
683 alien in the udev codebase. Following the new scheme kmod
684 will now generate a runtime snippet for tmpfiles from the
685 module meta information and it now is tmpfiles' job to the
686 create the nodes. This also allows overriding access and
687 other parameters for the nodes using the usual tmpfiles
688 facilities. As side effect this allows us to remove the
689 CAP_SYS_MKNOD capability bit from udevd entirely.
691 * logind's device ACLs may now be applied to these "dead"
692 devices nodes too, thus finally allowing managed access to
693 devices such as /dev/snd/sequencer whithout loading the
694 backing module right-away.
696 * A new RPM macro has been added that may be used to apply
697 tmpfiles configuration during package installation.
699 * systemd-detect-virt and ConditionVirtualization= now can
700 detect User-Mode-Linux machines (UML).
702 * journald will now implicitly log the effective capabilities
703 set of processes in the message metadata.
705 * systemd-cryptsetup has gained support for TrueCrypt volumes.
707 * The initrd interface has been simplified (more specifically,
708 support for passing performance data via environment
709 variables and fsck results via files in /run has been
710 removed). These features were non-essential, and are
711 nowadays available in a much nicer way by having systemd in
712 the initrd serialize its state and have the hosts systemd
713 deserialize it again.
715 * The udev "keymap" data files and tools to apply keyboard
716 specific mappings of scan to key codes, and force-release
717 scan code lists have been entirely replaced by a udev
718 "keyboard" builtin and a hwdb data file.
720 * systemd will now honour the kernel's "quiet" command line
721 argument also during late shutdown, resulting in a
722 completely silent shutdown when used.
724 * There's now an option to control the SO_REUSEPORT socket
725 option in .socket units.
727 * Instance units will now automatically get a per-template
728 subslice of system.slice unless something else is explicitly
729 configured. For example, instances of sshd@.service will now
730 implicitly be placed in system-sshd.slice rather than
731 system.slice as before.
733 * Test coverage support may now be enabled at build time.
735 Contributions from: Dave Reisner, Frederic Crozat, Harald
736 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
737 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
738 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
739 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
740 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
741 Giokas, Zbigniew Jędrzejewski-Szmek
743 -- Berlin, 2013-07-23
747 * Two new unit types have been introduced:
749 Scope units are very similar to service units, however, are
750 created out of pre-existing processes -- instead of PID 1
751 forking off the processes. By using scope units it is
752 possible for system services and applications to group their
753 own child processes (worker processes) in a powerful way
754 which then maybe used to organize them, or kill them
755 together, or apply resource limits on them.
757 Slice units may be used to partition system resources in an
758 hierarchial fashion and then assign other units to them. By
759 default there are now three slices: system.slice (for all
760 system services), user.slice (for all user sessions),
761 machine.slice (for VMs and containers).
763 Slices and scopes have been introduced primarily in
764 context of the work to move cgroup handling to a
765 single-writer scheme, where only PID 1
766 creates/removes/manages cgroups.
768 * There's a new concept of "transient" units. In contrast to
769 normal units these units are created via an API at runtime,
770 not from configuration from disk. More specifically this
771 means it is now possible to run arbitrary programs as
772 independent services, with all execution parameters passed
773 in via bus APIs rather than read from disk. Transient units
774 make systemd substantially more dynamic then it ever was,
775 and useful as a general batch manager.
777 * logind has been updated to make use of scope and slice units
778 for managing user sessions. As a user logs in he will get
779 his own private slice unit, to which all sessions are added
780 as scope units. We also added support for automatically
781 adding an instance of user@.service for the user into the
782 slice. Effectively logind will no longer create cgroup
783 hierarchies on its own now, it will defer entirely to PID 1
784 for this by means of scope, service and slice units. Since
785 user sessions this way become entities managed by PID 1
786 the output of "systemctl" is now a lot more comprehensive.
788 * A new mini-daemon "systemd-machined" has been added which
789 may be used by virtualization managers to register local
790 VMs/containers. nspawn has been updated accordingly, and
791 libvirt will be updated shortly. machined will collect a bit
792 of meta information about the VMs/containers, and assign
793 them their own scope unit (see above). The collected
794 meta-data is then made available via the "machinectl" tool,
795 and exposed in "ps" and similar tools. machined/machinectl
796 is compile-time optional.
798 * As discussed earlier, the low-level cgroup configuration
799 options ControlGroup=, ControlGroupModify=,
800 ControlGroupPersistent=, ControlGroupAttribute= have been
801 removed. Please use high-level attribute settings instead as
804 * A new bus call SetUnitProperties() has been added to alter
805 various runtime parameters of a unit. This is primarily
806 useful to alter cgroup parameters dynamically in a nice way,
807 but will be extended later on to make more properties
808 modifiable at runtime. systemctl gained a new set-properties
809 command that wraps this call.
811 * A new tool "systemd-run" has been added which can be used to
812 run arbitrary command lines as transient services or scopes,
813 while configuring a number of settings via the command
814 line. This tool is currently very basic, however already
815 very useful. We plan to extend this tool to even allow
816 queuing of execution jobs with time triggers from the
817 command line, similar in fashion to "at".
819 * nspawn will now inform the user explicitly that kernels with
820 audit enabled break containers, and suggest the user to turn
823 * Support for detecting the IMA and AppArmor security
824 frameworks with ConditionSecurity= has been added.
826 * journalctl gained a new "-k" switch for showing only kernel
827 messages, mimicking dmesg output; in addition to "--user"
828 and "--system" switches for showing only user's own logs
831 * systemd-delta can now show information about drop-in
832 snippets extending unit files.
834 * libsystemd-bus has been substantially updated but is still
835 not available as public API.
837 * systemd will now look for the "debug" argument on the kernel
838 command line and enable debug logging, similar to
839 "systemd.log_level=debug" already did before.
841 * "systemctl set-default", "systemctl get-default" has been
842 added to configure the default.target symlink, which
843 controls what to boot into by default.
845 * "systemctl set-log-level" has been added as a convenient
846 way to raise and lower systemd logging threshold.
848 * "systemd-analyze plot" will now show the time the various
849 generators needed for execution, as well as information
850 about the unit file loading.
852 * libsystemd-journal gained a new sd_journal_open_files() call
853 for opening specific journal files. journactl also gained a
854 new switch to expose this new functionality. Previously we
855 only supported opening all files from a directory, or all
856 files from the system, as opening individual files only is
857 racy due to journal file rotation.
859 * systemd gained the new DefaultEnvironment= setting in
860 /etc/systemd/system.conf to set environment variables for
863 * If a privileged process logs a journal message with the
864 OBJECT_PID= field set, then journald will automatically
865 augment this with additional OBJECT_UID=, OBJECT_GID=,
866 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
867 system services want to log events about specific client
868 processes. journactl/systemctl has been updated to make use
869 of this information if all log messages regarding a specific
872 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
873 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
874 Reisner, David Coppa, David King, David Strauss, Eelco
875 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
876 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
877 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
878 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
879 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
880 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
881 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
882 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
883 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
888 * The Python bindings gained some minimal support for the APIs
889 exposed by libsystemd-logind.
891 * ConditionSecurity= gained support for detecting SMACK. Since
892 this condition already supports SELinux and AppArmor we only
893 miss IMA for this. Patches welcome!
895 Contributions from: Karol Lewandowski, Lennart Poettering,
896 Zbigniew Jędrzejewski-Szmek
900 * systemd-nspawn will now create /etc/resolv.conf if
901 necessary, before bind-mounting the host's file onto it.
903 * systemd-nspawn will now store meta information about a
904 container on the container's cgroup as extended attribute
905 fields, including the root directory.
907 * The cgroup hierarchy has been reworked in many ways. All
908 objects any of the components systemd creates in the cgroup
909 tree are now suffixed. More specifically, user sessions are
910 now placed in cgroups suffixed with ".session", users in
911 cgroups suffixed with ".user", and nspawn containers in
912 cgroups suffixed with ".nspawn". Furthermore, all cgroup
913 names are now escaped in a simple scheme to avoid collision
914 of userspace object names with kernel filenames. This work
915 is preparation for making these objects relocatable in the
916 cgroup tree, in order to allow easy resource partitioning of
917 these objects without causing naming conflicts.
919 * systemctl list-dependencies gained the new switches
920 --plain, --reverse, --after and --before.
922 * systemd-inhibit now shows the process name of processes that
923 have taken an inhibitor lock.
925 * nss-myhostname will now also resolve "localhost"
926 implicitly. This makes /etc/hosts an optional file and
927 nicely handles that on IPv6 ::1 maps to both "localhost" and
930 * libsystemd-logind.so gained a new call
931 sd_get_machine_names() to enumerate running containers and
932 VMs (currently only supported by very new libvirt and
933 nspawn). sd_login_monitor can now be used to watch
934 VMs/containers coming and going.
936 * .include is not allowed recursively anymore, and only in
937 unit files. Usually it is better to use drop-in snippets in
938 .d/*.conf anyway, as introduced with systemd 198.
940 * systemd-analyze gained a new "critical-chain" command that
941 determines the slowest chain of units run during system
942 boot-up. It is very useful for tracking down where
943 optimizing boot time is the most beneficial.
945 * systemd will no longer allow manipulating service paths in
946 the name=systemd:/system cgroup tree using ControlGroup= in
947 units. (But is still fine with it in all other dirs.)
949 * There's a new systemd-nspawn@.service service file that may
950 be used to easily run nspawn containers as system
951 services. With the container's root directory in
952 /var/lib/container/foobar it is now sufficient to run
953 "systemctl start systemd-nspawn@foobar.service" to boot it.
955 * systemd-cgls gained a new parameter "--machine" to list only
956 the processes within a certain container.
958 * ConditionSecurity= now can check for "apparmor". We still
959 are lacking checks for SMACK and IMA for this condition
960 check though. Patches welcome!
962 * A new configuration file /etc/systemd/sleep.conf has been
963 added that may be used to configure which kernel operation
964 systemd is supposed to execute when "suspend", "hibernate"
965 or "hybrid-sleep" is requested. This makes the new kernel
966 "freeze" state accessible to the user.
968 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
969 the passed argument if applicable.
971 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
972 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
973 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
974 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
975 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
976 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
977 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
982 * The output of 'systemctl list-jobs' got some polishing. The
983 '--type=' argument may now be passed more than once. A new
984 command 'systemctl list-sockets' has been added which shows
985 a list of kernel sockets systemd is listening on with the
986 socket units they belong to, plus the units these socket
989 * The experimental libsystemd-bus library got substantial
990 updates to work in conjunction with the (also experimental)
991 kdbus kernel project. It works well enough to exchange
992 messages with some sophistication. Note that kdbus is not
993 ready yet, and the library is mostly an elaborate test case
994 for now, and not installable.
996 * systemd gained a new unit 'systemd-static-nodes.service'
997 that generates static device nodes earlier during boot, and
998 can run in conjunction with udev.
1000 * libsystemd-login gained a new call sd_pid_get_user_unit()
1001 to retrieve the user systemd unit a process is running
1002 in. This is useful for systems where systemd is used as
1005 * systemd-nspawn now places all containers in the new /machine
1006 top-level cgroup directory in the name=systemd
1007 hierarchy. libvirt will soon do the same, so that we get a
1008 uniform separation of /system, /user and /machine for system
1009 services, user processes and containers/virtual
1010 machines. This new cgroup hierarchy is also useful to stick
1011 stable names to specific container instances, which can be
1012 recognized later this way (this name may be controlled
1013 via systemd-nspawn's new -M switch). libsystemd-login also
1014 gained a new call sd_pid_get_machine_name() to retrieve the
1015 name of the container/VM a specific process belongs to.
1017 * bootchart can now store its data in the journal.
1019 * libsystemd-journal gained a new call
1020 sd_journal_add_conjunction() for AND expressions to the
1021 matching logic. This can be used to express more complex
1022 logical expressions.
1024 * journactl can now take multiple --unit= and --user-unit=
1027 * The cryptsetup logic now understands the "luks.key=" kernel
1028 command line switch for specifying a file to read the
1029 decryption key from. Also, if a configured key file is not
1030 found the tool will now automatically fall back to prompting
1033 * Python systemd.journal module was updated to wrap recently
1034 added functions from libsystemd-journal. The interface was
1035 changed to bring the low level interface in s.j._Reader
1036 closer to the C API, and the high level interface in
1037 s.j.Reader was updated to wrap and convert all data about
1040 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
1041 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
1042 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
1043 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
1044 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
1045 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1049 * journalctl --update-catalog now understands a new --root=
1050 option to operate on catalogs found in a different root
1053 * During shutdown after systemd has terminated all running
1054 services a final killing loop kills all remaining left-over
1055 processes. We will now print the name of these processes
1056 when we send SIGKILL to them, since this usually indicates a
1059 * If /etc/crypttab refers to password files stored on
1060 configured mount points automatic dependencies will now be
1061 generated to ensure the specific mount is established first
1062 before the key file is attempted to be read.
1064 * 'systemctl status' will now show information about the
1065 network sockets a socket unit is listening on.
1067 * 'systemctl status' will also shown information about any
1068 drop-in configuration file for units. (Drop-In configuration
1069 files in this context are files such as
1070 /etc/systemd/systemd/foobar.service.d/*.conf)
1072 * systemd-cgtop now optionally shows summed up CPU times of
1073 cgroups. Press '%' while running cgtop to switch between
1074 percentage and absolute mode. This is useful to determine
1075 which cgroups use up the most CPU time over the entire
1076 runtime of the system. systemd-cgtop has also been updated
1077 to be 'pipeable' for processing with further shell tools.
1079 * 'hostnamectl set-hostname' will now allow setting of FQDN
1082 * The formatting and parsing of time span values has been
1083 changed. The parser now understands fractional expressions
1084 such as "5.5h". The formatter will now output fractional
1085 expressions for all time spans under 1min, i.e. "5.123456s"
1086 rather than "5s 123ms 456us". For time spans under 1s
1087 millisecond values are shown, for those under 1ms
1088 microsecond values are shown. This should greatly improve
1089 all time-related output of systemd.
1091 * libsystemd-login and libsystemd-journal gained new
1092 functions for querying the poll() events mask and poll()
1093 timeout value for integration into arbitrary event
1096 * localectl gained the ability to list available X11 keymaps
1097 (models, layouts, variants, options).
1099 * 'systemd-analyze dot' gained the ability to filter for
1100 specific units via shell-style globs, to create smaller,
1101 more useful graphs. I.e. it's now possible to create simple
1102 graphs of all the dependencies between only target units, or
1103 of all units that Avahi has dependencies with.
1105 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
1106 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
1107 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
1108 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
1109 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
1110 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
1111 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
1115 * The boot-time readahead implementation for rotating media
1116 will now read the read-ahead data in multiple passes which
1117 consist of all read requests made in equidistant time
1118 intervals. This means instead of strictly reading read-ahead
1119 data in its physical order on disk we now try to find a
1120 middle ground between physical and access time order.
1122 * /etc/os-release files gained a new BUILD_ID= field for usage
1123 on operating systems that provide continuous builds of OS
1126 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
1127 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
1128 William Douglas, Zbigniew Jędrzejewski-Szmek
1132 * systemd-python gained an API exposing libsystemd-daemon.
1134 * The SMACK setup logic gained support for uploading CIPSO
1137 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
1138 ReadOnlyDirectories= and InaccessibleDirectories= has
1139 changed. The private /tmp and /var/tmp directories are now
1140 shared by all processes of a service (which means
1141 ExecStartPre= may now leave data in /tmp that ExecStart= of
1142 the same service can still access). When a service is
1143 stopped its temporary directories are immediately deleted
1144 (normal clean-up with tmpfiles is still done in addition to
1147 * By default, systemd will now set a couple of sysctl
1148 variables in the kernel: the safe sysrq options are turned
1149 on, IP route verification is turned on, and source routing
1150 disabled. The recently added hardlink and softlink
1151 protection of the kernel is turned on. These settings should
1152 be reasonably safe, and good defaults for all new systems.
1154 * The predictable network naming logic may now be turned off
1155 with a new kernel command line switch: net.ifnames=0.
1157 * A new libsystemd-bus module has been added that implements a
1158 pretty complete D-Bus client library. For details see:
1160 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
1162 * journald will now explicitly flush the journal files to disk
1163 at the latest 5min after each write. The file will then also
1164 be marked offline until the next write. This should increase
1165 reliability in case of a crash. The synchronization delay
1166 can be configured via SyncIntervalSec= in journald.conf.
1168 * There's a new remote-fs-setup.target unit that can be used
1169 to pull in specific services when at least one remote file
1170 system is to be mounted.
1172 * There are new targets timers.target and paths.target as
1173 canonical targets to pull user timer and path units in
1174 from. This complements sockets.target with a similar
1175 purpose for socket units.
1177 * libudev gained a new call udev_device_set_attribute_value()
1178 to set sysfs attributes of a device.
1180 * The udev daemon now sets the default number of worker
1181 processes executed in parallel based on the number of available
1182 CPUs instead of the amount of available RAM. This is supposed
1183 to provide a more reliable default and limit a too aggressive
1184 paralellism for setups with 1000s of devices connected.
1186 Contributions from: Auke Kok, Colin Walters, Cristian
1187 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
1188 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
1189 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
1190 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
1191 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
1192 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
1193 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
1194 Zbigniew Jędrzejewski-Szmek
1198 * Configuration of unit files may now be extended via drop-in
1199 files without having to edit/override the unit files
1200 themselves. More specifically, if the administrator wants to
1201 change one value for a service file foobar.service he can
1202 now do so by dropping in a configuration snippet into
1203 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
1204 will load all these snippets and apply them on top of the
1205 main unit configuration file, possibly extending or
1206 overriding its settings. Using these drop-in snippets is
1207 generally nicer than the two earlier options for changing
1208 unit files locally: copying the files from
1209 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
1210 them there; or creating a new file in /etc/systemd/system/
1211 that incorporates the original one via ".include". Drop-in
1212 snippets into these .d/ directories can be placed in any
1213 directory systemd looks for units in, and the usual
1214 overriding semantics between /usr/lib, /etc and /run apply
1217 * Most unit file settings which take lists of items can now be
1218 reset by assigning the empty string to them. For example,
1219 normally, settings such as Environment=FOO=BAR append a new
1220 environment variable assignment to the environment block,
1221 each time they are used. By assigning Environment= the empty
1222 string the environment block can be reset to empty. This is
1223 particularly useful with the .d/*.conf drop-in snippets
1224 mentioned above, since this adds the ability to reset list
1225 settings from vendor unit files via these drop-ins.
1227 * systemctl gained a new "list-dependencies" command for
1228 listing the dependencies of a unit recursively.
1230 * Inhibitors are now honored and listed by "systemctl
1231 suspend", "systemctl poweroff" (and similar) too, not only
1232 GNOME. These commands will also list active sessions by
1235 * Resource limits (as exposed by the various control group
1236 controllers) can now be controlled dynamically at runtime
1237 for all units. More specifically, you can now use a command
1238 like "systemctl set-cgroup-attr foobar.service cpu.shares
1239 2000" to alter the CPU shares a specific service gets. These
1240 settings are stored persistently on disk, and thus allow the
1241 administrator to easily adjust the resource usage of
1242 services with a few simple commands. This dynamic resource
1243 management logic is also available to other programs via the
1244 bus. Almost any kernel cgroup attribute and controller is
1247 * systemd-vconsole-setup will now copy all font settings to
1248 all allocated VTs, where it previously applied them only to
1251 * libsystemd-login gained the new sd_session_get_tty() API
1254 * This release drops support for a few legacy or
1255 distribution-specific LSB facility names when parsing init
1256 scripts: $x-display-manager, $mail-transfer-agent,
1257 $mail-transport-agent, $mail-transfer-agent, $smtp,
1258 $null. Also, the mail-transfer-agent.target unit backing
1259 this has been removed. Distributions which want to retain
1260 compatibility with this should carry the burden for
1261 supporting this themselves and patch support for these back
1262 in, if they really need to. Also, the facilities $syslog and
1263 $local_fs are now ignored, since systemd does not support
1264 early-boot LSB init scripts anymore, and these facilities
1265 are implied anyway for normal services. syslog.target has
1268 * There are new bus calls on PID1's Manager object for
1269 cancelling jobs, and removing snapshot units. Previously,
1270 both calls were only available on the Job and Snapshot
1273 * systemd-journal-gatewayd gained SSL support.
1275 * The various "environment" files, such as /etc/locale.conf
1276 now support continuation lines with a backslash ("\") as
1277 last character in the line, similar in style (but different)
1278 to how this is supported in shells.
1280 * For normal user processes the _SYSTEMD_USER_UNIT= field is
1281 now implicitly appended to every log entry logged. systemctl
1282 has been updated to filter by this field when operating on a
1283 user systemd instance.
1285 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
1286 CAP_AUDIT_CONTROL capabilities to the capabilities set for
1287 the container. This makes it easier to boot unmodified
1288 Fedora systems in a container, which however still requires
1289 audit=0 to be passed on the kernel command line. Auditing in
1290 kernel and userspace is unfortunately still too broken in
1291 context of containers, hence we recommend compiling it out
1292 of the kernel or using audit=0. Hopefully this will be fixed
1293 one day for good in the kernel.
1295 * nspawn gained the new --bind= and --bind-ro= parameters to
1296 bind mount specific directories from the host into the
1299 * nspawn will now mount its own devpts file system instance
1300 into the container, in order not to leak pty devices from
1301 the host into the container.
1303 * systemd will now read the firmware boot time performance
1304 information from the EFI variables, if the used boot loader
1305 supports this, and takes it into account for boot performance
1306 analysis via "systemd-analyze". This is currently supported
1307 only in conjunction with Gummiboot, but could be supported
1308 by other boot loaders too. For details see:
1310 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
1312 * A new generator has been added that automatically mounts the
1313 EFI System Partition (ESP) to /boot, if that directory
1314 exists, is empty, and no other file system has been
1315 configured to be mounted there.
1317 * logind will now send out PrepareForSleep(false) out
1318 unconditionally, after coming back from suspend. This may be
1319 used by applications as asynchronous notification for
1320 system resume events.
1322 * "systemctl unlock-sessions" has been added, that allows
1323 unlocking the screens of all user sessions at once, similar
1324 how "systemctl lock-sessions" already locked all users
1325 sessions. This is backed by a new D-Bus call UnlockSessions().
1327 * "loginctl seat-status" will now show the master device of a
1328 seat. (i.e. the device of a seat that needs to be around for
1329 the seat to be considered available, usually the graphics
1332 * tmpfiles gained a new "X" line type, that allows
1333 configuration of files and directories (with wildcards) that
1334 shall be excluded from automatic cleanup ("aging").
1336 * udev default rules set the device node permissions now only
1337 at "add" events, and do not change them any longer with a
1338 later "change" event.
1340 * The log messages for lid events and power/sleep keypresses
1341 now carry a message ID.
1343 * We now have a substantially larger unit test suite, but this
1344 continues to be work in progress.
1346 * udevadm hwdb gained a new --root= parameter to change the
1347 root directory to operate relative to.
1349 * logind will now issue a background sync() request to the kernel
1350 early at shutdown, so that dirty buffers are flushed to disk early
1351 instead of at the last moment, in order to optimize shutdown
1354 * A new bootctl tool has been added that is an interface for
1355 certain boot loader operations. This is currently a preview
1356 and is likely to be extended into a small mechanism daemon
1357 like timedated, localed, hostnamed, and can be used by
1358 graphical UIs to enumerate available boot options, and
1359 request boot into firmware operations.
1361 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
1362 the rest of the package. It also has been updated to work
1363 correctly in initrds.
1365 * Policykit previously has been runtime optional, and is now
1366 also compile time optional via a configure switch.
1368 * systemd-analyze has been reimplemented in C. Also "systemctl
1369 dot" has moved into systemd-analyze.
1371 * "systemctl status" with no further parameters will now print
1372 the status of all active or failed units.
1374 * Operations such as "systemctl start" can now be executed
1375 with a new mode "--irreversible" which may be used to queue
1376 operations that cannot accidentally be reversed by a later
1377 job queuing. This is by default used to make shutdown
1378 requests more robust.
1380 * The Python API of systemd now gained a new module for
1381 reading journal files.
1383 * A new tool kernel-install has been added that can install
1384 kernel images according to the Boot Loader Specification:
1386 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
1388 * Boot time console output has been improved to provide
1389 animated boot time output for hanging jobs.
1391 * A new tool systemd-activate has been added which can be used
1392 to test socket activation with, directly from the command
1393 line. This should make it much easier to test and debug
1394 socket activation in daemons.
1396 * journalctl gained a new "--reverse" (or -r) option to show
1397 journal output in reverse order (i.e. newest line first).
1399 * journalctl gained a new "--pager-end" (or -e) option to jump
1400 to immediately jump to the end of the journal in the
1401 pager. This is only supported in conjunction with "less".
1403 * journalctl gained a new "--user-unit=" option, that works
1404 similar to "--unit=" but filters for user units rather than
1407 * A number of unit files to ease adoption of systemd in
1408 initrds has been added. This moves some minimal logic from
1409 the various initrd implementations into systemd proper.
1411 * The journal files are now owned by a new group
1412 "systemd-journal", which exists specifically to allow access
1413 to the journal, and nothing else. Previously, we used the
1414 "adm" group for that, which however possibly covers more
1415 than just journal/log file access. This new group is now
1416 already used by systemd-journal-gatewayd to ensure this
1417 daemon gets access to the journal files and as little else
1418 as possible. Note that "make install" will also set FS ACLs
1419 up for /var/log/journal to give "adm" and "wheel" read
1420 access to it, in addition to "systemd-journal" which owns
1421 the journal files. We recommend that packaging scripts also
1422 add read access to "adm" + "wheel" to /var/log/journal, and
1423 all existing/future journal files. To normal users and
1424 administrators little changes, however packagers need to
1425 ensure to create the "systemd-journal" system group at
1426 package installation time.
1428 * The systemd-journal-gatewayd now runs as unprivileged user
1429 systemd-journal-gateway:systemd-journal-gateway. Packaging
1430 scripts need to create these system user/group at
1433 * timedated now exposes a new boolean property CanNTP that
1434 indicates whether a local NTP service is available or not.
1436 * systemd-detect-virt will now also detect xen PVs
1438 * The pstore file system is now mounted by default, if it is
1441 * In addition to the SELinux and IMA policies we will now also
1442 load SMACK policies at early boot.
1444 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
1445 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
1446 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
1447 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
1448 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
1449 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
1450 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
1451 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
1452 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
1453 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
1454 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
1455 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
1456 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
1457 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
1461 * Timer units now support calendar time events in addition to
1462 monotonic time events. That means you can now trigger a unit
1463 based on a calendar time specification such as "Thu,Fri
1464 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
1465 or fifth day of any month of the year 2013, given that it is
1466 a thursday or friday. This brings timer event support
1467 considerably closer to cron's capabilities. For details on
1468 the supported calendar time specification language see
1471 * udev now supports a number of different naming policies for
1472 network interfaces for predictable names, and a combination
1473 of these policies is now the default. Please see this wiki
1474 document for details:
1476 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
1478 * Auke Kok's bootchart implementation has been added to the
1479 systemd tree. It's an optional component that can graph the
1480 boot in quite some detail. It's one of the best bootchart
1481 implementations around and minimal in its code and
1484 * nss-myhostname has been integrated into the systemd source
1485 tree. nss-myhostname guarantees that the local hostname
1486 always stays resolvable via NSS. It has been a weak
1487 requirement of systemd-hostnamed since a long time, and
1488 since its code is actually trivial we decided to just
1489 include it in systemd's source tree. It can be turned off
1490 with a configure switch.
1492 * The read-ahead logic is now capable of properly detecting
1493 whether a btrfs file system is on SSD or rotating media, in
1494 order to optimize the read-ahead scheme. Previously, it was
1495 only capable of detecting this on traditional file systems
1498 * In udev, additional device properties are now read from the
1499 IAB in addition to the OUI database. Also, Bluetooth company
1500 identities are attached to the devices as well.
1502 * In service files %U may be used as specifier that is
1503 replaced by the configured user name of the service.
1505 * nspawn may now be invoked without a controlling TTY. This
1506 makes it suitable for invocation as its own service. This
1507 may be used to set up a simple containerized server system
1508 using only core OS tools.
1510 * systemd and nspawn can now accept socket file descriptors
1511 when they are started for socket activation. This enables
1512 implementation of socket activated nspawn
1513 containers. i.e. think about autospawning an entire OS image
1514 when the first SSH or HTTP connection is received. We expect
1515 that similar functionality will also be added to libvirt-lxc
1518 * journalctl will now suppress ANSI color codes when
1519 presenting log data.
1521 * systemctl will no longer show control group information for
1522 a unit if a the control group is empty anyway.
1524 * logind can now automatically suspend/hibernate/shutdown the
1527 * /etc/machine-info and hostnamed now also expose the chassis
1528 type of the system. This can be used to determine whether
1529 the local system is a laptop, desktop, handset or
1530 tablet. This information may either be configured by the
1531 user/vendor or is automatically determined from ACPI and DMI
1532 information if possible.
1534 * A number of PolicyKit actions are now bound together with
1535 "imply" rules. This should simplify creating UIs because
1536 many actions will now authenticate similar ones as well.
1538 * Unit files learnt a new condition ConditionACPower= which
1539 may be used to conditionalize a unit depending on whether an
1540 AC power source is connected or not, of whether the system
1541 is running on battery power.
1543 * systemctl gained a new "is-failed" verb that may be used in
1544 shell scripts and suchlike to check whether a specific unit
1545 is in the "failed" state.
1547 * The EnvironmentFile= setting in unit files now supports file
1548 globbing, and can hence be used to easily read a number of
1549 environment files at once.
1551 * systemd will no longer detect and recognize specific
1552 distributions. All distribution-specific #ifdeffery has been
1553 removed, systemd is now fully generic and
1554 distribution-agnostic. Effectively, not too much is lost as
1555 a lot of the code is still accessible via explicit configure
1556 switches. However, support for some distribution specific
1557 legacy configuration file formats has been dropped. We
1558 recommend distributions to simply adopt the configuration
1559 files everybody else uses now and convert the old
1560 configuration from packaging scripts. Most distributions
1561 already did that. If that's not possible or desirable,
1562 distributions are welcome to forward port the specific
1563 pieces of code locally from the git history.
1565 * When logging a message about a unit systemd will now always
1566 log the unit name in the message meta data.
1568 * localectl will now also discover system locale data that is
1569 not stored in locale archives, but directly unpacked.
1571 * logind will no longer unconditionally use framebuffer
1572 devices as seat masters, i.e. as devices that are required
1573 to be existing before a seat is considered preset. Instead,
1574 it will now look for all devices that are tagged as
1575 "seat-master" in udev. By default framebuffer devices will
1576 be marked as such, but depending on local systems other
1577 devices might be marked as well. This may be used to
1578 integrate graphics cards using closed source drivers (such
1579 as NVidia ones) more nicely into logind. Note however, that
1580 we recommend using the open source NVidia drivers instead,
1581 and no udev rules for the closed-source drivers will be
1582 shipped from us upstream.
1584 Contributions from: Adam Williamson, Alessandro Crismani, Auke
1585 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
1586 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
1587 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
1588 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
1589 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
1590 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
1591 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
1592 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
1593 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
1594 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
1595 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
1600 * udev gained support for loading additional device properties
1601 from an indexed database that is keyed by vendor/product IDs
1602 and similar device identifiers. For the beginning this
1603 "hwdb" is populated with data from the well-known PCI and
1604 USB database, but also includes PNP, ACPI and OID data. In
1605 the longer run this indexed database shall grow into
1606 becoming the one central database for non-essential
1607 userspace device metadata. Previously, data from the PCI/USB
1608 database was only attached to select devices, since the
1609 lookup was a relatively expensive operation due to O(n) time
1610 complexity (with n being the number of entries in the
1611 database). Since this is now O(1), we decided to add in this
1612 data for all devices where this is available, by
1613 default. Note that the indexed database needs to be rebuilt
1614 when new data files are installed. To achieve this you need
1615 to update your packaging scripts to invoke "udevadm hwdb
1616 --update" after installation of hwdb data files. For
1617 RPM-based distributions we introduced the new
1618 %udev_hwdb_update macro for this purpose.
1620 * The Journal gained support for the "Message Catalog", an
1621 indexed database to link up additional information with
1622 journal entries. For further details please check:
1624 http://www.freedesktop.org/wiki/Software/systemd/catalog
1626 The indexed message catalog database also needs to be
1627 rebuilt after installation of message catalog files. Use
1628 "journalctl --update-catalog" for this. For RPM-based
1629 distributions we introduced the %journal_catalog_update
1630 macro for this purpose.
1632 * The Python Journal bindings gained support for the standard
1633 Python logging framework.
1635 * The Journal API gained new functions for checking whether
1636 the underlying file system of a journal file is capable of
1637 properly reporting file change notifications, or whether
1638 applications that want to reflect journal changes "live"
1639 need to recheck journal files continuously in appropriate
1642 * It is now possible to set the "age" field for tmpfiles
1643 entries to 0, indicating that files matching this entry
1644 shall always be removed when the directories are cleaned up.
1646 * coredumpctl gained a new "gdb" verb which invokes gdb
1647 right-away on the selected coredump.
1649 * There's now support for "hybrid sleep" on kernels that
1650 support this, in addition to "suspend" and "hibernate". Use
1651 "systemctl hybrid-sleep" to make use of this.
1653 * logind's HandleSuspendKey= setting (and related settings)
1654 now gained support for a new "lock" setting to simply
1655 request the screen lock on all local sessions, instead of
1656 actually executing a suspend or hibernation.
1658 * systemd will now mount the EFI variables file system by
1661 * Socket units now gained support for configuration of the
1662 SMACK security label.
1664 * timedatectl will now output the time of the last and next
1665 daylight saving change.
1667 * We dropped support for various legacy and distro-specific
1668 concepts, such as insserv, early-boot SysV services
1669 (i.e. those for non-standard runlevels such as 'b' or 'S')
1670 or ArchLinux /etc/rc.conf support. We recommend the
1671 distributions who still need support this to either continue
1672 to maintain the necessary patches downstream, or find a
1673 different solution. (Talk to us if you have questions!)
1675 * Various systemd components will now bypass PolicyKit checks
1676 for root and otherwise handle properly if PolicyKit is not
1677 found to be around. This should fix most issues for
1678 PolicyKit-less systems. Quite frankly this should have been
1679 this way since day one. It is absolutely our intention to
1680 make systemd work fine on PolicyKit-less systems, and we
1681 consider it a bug if something doesn't work as it should if
1682 PolicyKit is not around.
1684 * For embedded systems it is now possible to build udev and
1685 systemd without blkid and/or kmod support.
1687 * "systemctl switch-root" is now capable of switching root
1688 more than once. I.e. in addition to transitions from the
1689 initrd to the host OS it is now possible to transition to
1690 further OS images from the host. This is useful to implement
1691 offline updating tools.
1693 * Various other additions have been made to the RPM macros
1694 shipped with systemd. Use %udev_rules_update() after
1695 installing new udev rules files. %_udevhwdbdir,
1696 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
1697 %_sysctldir are now available which resolve to the right
1698 directories for packages to place various data files in.
1700 * journalctl gained the new --full switch (in addition to
1701 --all, to disable ellipsation for long messages.
1703 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
1704 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
1705 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
1706 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
1707 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
1708 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
1709 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
1710 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
1711 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
1715 * journalctl gained new --since= and --until= switches to
1716 filter by time. It also now supports nice filtering for
1717 units via --unit=/-u.
1719 * Type=oneshot services may use ExecReload= and do the
1722 * The journal daemon now supports time-based rotation and
1723 vacuuming, in addition to the usual disk-space based
1726 * The journal will now index the available field values for
1727 each field name. This enables clients to show pretty drop
1728 downs of available match values when filtering. The bash
1729 completion of journalctl has been updated
1730 accordingly. journalctl gained a new switch -F to list all
1731 values a certain field takes in the journal database.
1733 * More service events are now written as structured messages
1734 to the journal, and made recognizable via message IDs.
1736 * The timedated, localed and hostnamed mini-services which
1737 previously only provided support for changing time, locale
1738 and hostname settings from graphical DEs such as GNOME now
1739 also have a minimal (but very useful) text-based client
1740 utility each. This is probably the nicest way to changing
1741 these settings from the command line now, especially since
1742 it lists available options and is fully integrated with bash
1745 * There's now a new tool "systemd-coredumpctl" to list and
1746 extract coredumps from the journal.
1748 * We now install a README each in /var/log/ and
1749 /etc/rc.d/init.d explaining where the system logs and init
1750 scripts went. This hopefully should help folks who go to
1751 that dirs and look into the otherwise now empty void and
1752 scratch their heads.
1754 * When user-services are invoked (by systemd --user) the
1755 $MANAGERPID env var is set to the PID of systemd.
1757 * SIGRTMIN+24 when sent to a --user instance will now result
1758 in immediate termination of systemd.
1760 * gatewayd received numerous feature additions such as a
1761 "follow" mode, for live syncing and filtering.
1763 * browse.html now allows filtering and showing detailed
1764 information on specific entries. Keyboard navigation and
1765 mouse screen support has been added.
1767 * gatewayd/journalctl now supports HTML5/JSON
1768 Server-Sent-Events as output.
1770 * The SysV init script compatibility logic will now
1771 heuristically determine whether a script supports the
1772 "reload" verb, and only then make this available as
1775 * "systemctl status --follow" has been removed, use "journalctl
1778 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
1779 have been removed since they are hardly useful to be
1782 * And I'd like to take the opportunity to specifically mention
1783 Zbigniew for his great contributions. Zbigniew, you rock!
1785 Contributions from: Andrew Eikum, Christian Hesse, Colin
1786 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
1787 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
1788 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
1789 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
1790 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
1791 Jędrzejewski-Szmek, Сковорода Никита Андреевич
1795 * If /etc/vconsole.conf is non-existent or empty we will no
1796 longer load any console font or key map at boot by
1797 default. Instead the kernel defaults will be left
1798 intact. This is definitely the right thing to do, as no
1799 configuration should mean no configuration, and hard-coding
1800 font names that are different on all archs is probably a bad
1801 idea. Also, the kernel default key map and font should be
1802 good enough for most cases anyway, and mostly identical to
1803 the userspace fonts/key maps we previously overloaded them
1804 with. If distributions want to continue to default to a
1805 non-kernel font or key map they should ship a default
1806 /etc/vconsole.conf with the appropriate contents.
1808 Contributions from: Colin Walters, Daniel J Walsh, Dave
1809 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
1810 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1814 * journalctl gained a new --cursor= switch to show entries
1815 starting from the specified location in the journal.
1817 * We now enforce a size limit on journal entry fields exported
1818 with "-o json" in journalctl. Fields larger than 4K will be
1819 assigned null. This can be turned off with --all.
1821 * An (optional) journal gateway daemon is now available as
1822 "systemd-journal-gatewayd.service". This service provides
1823 access to the journal via HTTP and JSON. This functionality
1824 will be used to implement live log synchronization in both
1825 pull and push modes, but has various other users too, such
1826 as easy log access for debugging of embedded devices. Right
1827 now it is already useful to retrieve the journal via HTTP:
1829 # systemctl start systemd-journal-gatewayd.service
1830 # wget http://localhost:19531/entries
1832 This will download the journal contents in a
1833 /var/log/messages compatible format. The same as JSON:
1835 # curl -H"Accept: application/json" http://localhost:19531/entries
1837 This service is also accessible via a web browser where a
1838 single static HTML5 app is served that uses the JSON logic
1839 to enable the user to do some basic browsing of the
1840 journal. This will be extended later on. Here's an example
1841 screenshot of this app in its current state:
1843 http://0pointer.de/public/journal-gatewayd
1845 Contributions from: Kay Sievers, Lennart Poettering, Robert
1846 Milasan, Tom Gundersen
1850 * The bash completion logic is now available for journalctl
1853 * We don't mount the "cpuset" controller anymore together with
1854 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
1855 started if no parameters are assigned to it. "cpuset" hence
1856 broke code that assumed it it could create "cpu" groups and
1859 * journalctl -f will now subscribe to terminal size changes,
1860 and line break accordingly.
1862 Contributions from: Dave Reisner, Kay Sievers, Lennart
1863 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
1867 * nspawn will now create a symlink /etc/localtime in the
1868 container environment, copying the host's timezone
1869 setting. Previously this has been done via a bind mount, but
1870 since symlinks cannot be bind mounted this has now been
1871 changed to create/update the appropriate symlink.
1873 * journalctl -n's line number argument is now optional, and
1874 will default to 10 if omitted.
1876 * journald will now log the maximum size the journal files may
1877 take up on disk. This is particularly useful if the default
1878 built-in logic of determining this parameter from the file
1879 system size is used. Use "systemctl status
1880 systemd-journald.service" to see this information.
1882 * The multi-seat X wrapper tool has been stripped down. As X
1883 is now capable of enumerating graphics devices via udev in a
1884 seat-aware way the wrapper is not strictly necessary
1885 anymore. A stripped down temporary stop-gap is still shipped
1886 until the upstream display managers have been updated to
1887 fully support the new X logic. Expect this wrapper to be
1888 removed entirely in one of the next releases.
1890 * HandleSleepKey= in logind.conf has been split up into
1891 HandleSuspendKey= and HandleHibernateKey=. The old setting
1892 is not available anymore. X11 and the kernel are
1893 distuingishing between these keys and we should too. This
1894 also means the inhibition lock for these keys has been split
1897 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
1898 Poettering, Lukas Nykryn, Václav Pavlín
1902 * Whenever a unit changes state we'll now log this to the
1903 journal and show along the unit's own log output in
1906 * ConditionPathIsMountPoint= can now properly detect bind
1907 mount points too. (Previously, a bind mount of one file
1908 system to another place in the same file system could not be
1909 detected as mount, since they shared struct stat's st_dev
1912 * We will now mount the cgroup controllers cpu, cpuacct,
1913 cpuset and the controllers net_cls, net_prio together by
1916 * nspawn containers will now have a virtualized boot
1917 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
1918 over with a randomized ID at container initialization). This
1919 has the effect of making "journalctl -b" do the right thing
1922 * The JSON output journal serialization has been updated not
1923 to generate "endless" list objects anymore, but rather one
1924 JSON object per line. This is more in line how most JSON
1925 parsers expect JSON objects. The new output mode
1926 "json-pretty" has been added to provide similar output, but
1927 neatly aligned for readability by humans.
1929 * We dropped all explicit sync() invocations in the shutdown
1930 code. The kernel does this implicitly anyway in the kernel
1931 reboot() syscall. halt(8)'s -n option is now a compatibility
1934 * We now support virtualized reboot() in containers, as
1935 supported by newer kernels. We will fall back to exit() if
1936 CAP_SYS_REBOOT is not available to the container. Also,
1937 nspawn makes use of this now and will actually reboot the
1938 container if the containerized OS asks for that.
1940 * journalctl will only show local log output by default
1941 now. Use --merge (-m) to show remote log output, too.
1943 * libsystemd-journal gained the new sd_journal_get_usage()
1944 call to determine the current disk usage of all journal
1945 files. This is exposed in the new "journalctl --disk-usage"
1948 * journald gained a new configuration setting SplitMode= in
1949 journald.conf which may be used to control how user journals
1950 are split off. See journald.conf(5) for details.
1952 * A new condition type ConditionFileNotEmpty= has been added.
1954 * tmpfiles' "w" lines now support file globbing, to write
1955 multiple files at once.
1957 * We added Python bindings for the journal submission
1958 APIs. More Python APIs for a number of selected APIs will
1959 likely follow. Note that we intend to add native bindings
1960 only for the Python language, as we consider it common
1961 enough to deserve bindings shipped within systemd. There are
1962 various projects outside of systemd that provide bindings
1963 for languages such as PHP or Lua.
1965 * Many conditions will now resolve specifiers such as %i. In
1966 addition, PathChanged= and related directives of .path units
1967 now support specifiers as well.
1969 * There's now a new RPM macro definition for the system preset
1972 * journald will now warn if it can't forward a message to the
1973 syslog daemon because it's socket is full.
1975 * timedated will no longer write or process /etc/timezone,
1976 except on Debian. As we do not support late mounted /usr
1977 anymore /etc/localtime always being a symlink is now safe,
1978 and hence the information in /etc/timezone is not necessary
1981 * logind will now always reserve one VT for a text getty (VT6
1982 by default). Previously if more than 6 X sessions where
1983 started they took up all the VTs with auto-spawned gettys,
1984 so that no text gettys were available anymore.
1986 * udev will now automatically inform the btrfs kernel logic
1987 about btrfs RAID components showing up. This should make
1988 simple hotplug based btrfs RAID assembly work.
1990 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
1991 (but not for its children which will stay at the kernel
1992 default). This should allow setups with a lot more listening
1995 * systemd will now always pass the configured timezone to the
1996 kernel at boot. timedated will do the same when the timezone
1999 * logind's inhibition logic has been updated. By default,
2000 logind will now handle the lid switch, the power and sleep
2001 keys all the time, even in graphical sessions. If DEs want
2002 to handle these events on their own they should take the new
2003 handle-power-key, handle-sleep-key and handle-lid-switch
2004 inhibitors during their runtime. A simple way to achiveve
2005 that is to invoke the DE wrapped in an invocation of:
2007 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
2009 * Access to unit operations is now checked via SELinux taking
2010 the unit file label and client process label into account.
2012 * systemd will now notify the administrator in the journal
2013 when he over-mounts a non-empty directory.
2015 * There are new specifiers that are resolved in unit files,
2016 for the host name (%H), the machine ID (%m) and the boot ID
2019 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
2020 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
2021 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
2022 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
2023 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
2024 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
2025 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
2029 * Support for reading structured kernel messages from
2030 /dev/kmsg has now been added and is enabled by default.
2032 * Support for reading kernel messages from /proc/kmsg has now
2033 been removed. If you want kernel messages in the journal
2034 make sure to run a recent kernel (>= 3.5) that supports
2035 reading structured messages from /dev/kmsg (see
2036 above). /proc/kmsg is now exclusive property of classic
2037 syslog daemons again.
2039 * The libudev API gained the new
2040 udev_device_new_from_device_id() call.
2042 * The logic for file system namespace (ReadOnlyDirectory=,
2043 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
2044 require pivot_root() anymore. This means fewer temporary
2045 directories are created below /tmp for this feature.
2047 * nspawn containers will now see and receive all submounts
2048 made on the host OS below the root file system of the
2051 * Forward Secure Sealing is now supported for Journal files,
2052 which provide cryptographical sealing of journal files so
2053 that attackers cannot alter log history anymore without this
2054 being detectable. Lennart will soon post a blog story about
2055 this explaining it in more detail.
2057 * There are two new service settings RestartPreventExitStatus=
2058 and SuccessExitStatus= which allow configuration of exit
2059 status (exit code or signal) which will be excepted from the
2060 restart logic, resp. consider successful.
2062 * journalctl gained the new --verify switch that can be used
2063 to check the integrity of the structure of journal files and
2064 (if Forward Secure Sealing is enabled) the contents of
2067 * nspawn containers will now be run with /dev/stdin, /dev/fd/
2068 and similar symlinks pre-created. This makes running shells
2069 as container init process a lot more fun.
2071 * The fstab support can now handle PARTUUID= and PARTLABEL=
2074 * A new ConditionHost= condition has been added to match
2075 against the hostname (with globs) and machine ID. This is
2076 useful for clusters where a single OS image is used to
2077 provision a large number of hosts which shall run slightly
2078 different sets of services.
2080 * Services which hit the restart limit will now be placed in a
2083 Contributions from: Bertram Poettering, Dave Reisner, Huang
2084 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
2085 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
2089 * When running in --user mode systemd will now become a
2090 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
2091 tree a lot more organized.
2093 * A new PartOf= unit dependency type has been introduced that
2094 may be used to group services in a natural way.
2096 * "systemctl enable" may now be used to enable instances of
2099 * journalctl now prints error log levels in red, and
2100 warning/notice log levels in bright white. It also supports
2101 filtering by log level now.
2103 * cgtop gained a new -n switch (similar to top), to configure
2104 the maximum number of iterations to run for. It also gained
2105 -b, to run in batch mode (accepting no input).
2107 * The suffix ".service" may now be omitted on most systemctl
2108 command lines involving service unit names.
2110 * There's a new bus call in logind to lock all sessions, as
2111 well as a loginctl verb for it "lock-sessions".
2113 * libsystemd-logind.so gained a new call sd_journal_perror()
2114 that works similar to libc perror() but logs to the journal
2115 and encodes structured information about the error number.
2117 * /etc/crypttab entries now understand the new keyfile-size=
2120 * shutdown(8) now can send a (configurable) wall message when
2121 a shutdown is cancelled.
2123 * The mount propagation mode for the root file system will now
2124 default to "shared", which is useful to make containers work
2125 nicely out-of-the-box so that they receive new mounts from
2126 the host. This can be undone locally by running "mount
2127 --make-rprivate /" if needed.
2129 * The prefdm.service file has been removed. Distributions
2130 should maintain this unit downstream if they intend to keep
2131 it around. However, we recommend writing normal unit files
2132 for display managers instead.
2134 * Since systemd is a crucial part of the OS we will now
2135 default to a number of compiler switches that improve
2136 security (hardening) such as read-only relocations, stack
2137 protection, and suchlike.
2139 * The TimeoutSec= setting for services is now split into
2140 TimeoutStartSec= and TimeoutStopSec= to allow configuration
2141 of individual time outs for the start and the stop phase of
2144 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
2145 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
2146 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
2147 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
2148 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
2149 Gundersen, Zbigniew Jędrzejewski-Szmek
2153 * The journal and id128 C APIs are now fully documented as man
2156 * Extra safety checks have been added when transitioning from
2157 the initial RAM disk to the main system to avoid accidental
2160 * /etc/crypttab entries now understand the new keyfile-offset=
2163 * systemctl -t can now be used to filter by unit load state.
2165 * The journal C API gained the new sd_journal_wait() call to
2166 make writing synchronous journal clients easier.
2168 * journalctl gained the new -D switch to show journals from a
2171 * journalctl now displays a special marker between log
2172 messages of two different boots.
2174 * The journal is now explicitly flushed to /var via a service
2175 systemd-journal-flush.service, rather than implicitly simply
2176 by seeing /var/log/journal to be writable.
2178 * journalctl (and the journal C APIs) can now match for much
2179 more complex expressions, with alternatives and
2182 * When transitioning from the initial RAM disk to the main
2183 system we will now kill all processes in a killing spree to
2184 ensure no processes stay around by accident.
2186 * Three new specifiers may be used in unit files: %u, %h, %s
2187 resolve to the user name, user home directory resp. user
2188 shell. This is useful for running systemd user instances.
2190 * We now automatically rotate journal files if their data
2191 object hash table gets a fill level > 75%. We also size the
2192 hash table based on the configured maximum file size. This
2193 together should lower hash collisions drastically and thus
2194 speed things up a bit.
2196 * journalctl gained the new "--header" switch to introspect
2197 header data of journal files.
2199 * A new setting SystemCallFilters= has been added to services
2200 which may be used to apply blacklists or whitelists to
2201 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
2203 * nspawn gained a new --link-journal= switch (and quicker: -j)
2204 to link the container journal with the host. This makes it
2205 very easy to centralize log viewing on the host for all
2206 guests while still keeping the journal files separated.
2208 * Many bugfixes and optimizations
2210 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
2211 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
2212 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
2217 * Several tools now understand kernel command line arguments,
2218 which are only read when run in an initial RAM disk. They
2219 usually follow closely their normal counterparts, but are
2222 * There's a new tool to analyze the readahead files that are
2223 automatically generated at boot. Use:
2225 /usr/lib/systemd/systemd-readahead analyze /.readahead
2227 * We now provide an early debug shell on tty9 if this enabled. Use:
2229 systemctl enable debug-shell.service
2231 * All plymouth related units have been moved into the Plymouth
2232 package. Please make sure to upgrade your Plymouth version
2235 * systemd-tmpfiles now supports getting passed the basename of
2236 a configuration file only, in which case it will look for it
2237 in all appropriate directories automatically.
2239 * udevadm info now takes a /dev or /sys path as argument, and
2240 does the right thing. Example:
2242 udevadm info /dev/sda
2243 udevadm info /sys/class/block/sda
2245 * systemctl now prints a warning if a unit is stopped but a
2246 unit that might trigger it continues to run. Example: a
2247 service is stopped but the socket that activates it is left
2250 * "systemctl status" will now mention if the log output was
2251 shortened due to rotation since a service has been started.
2253 * The journal API now exposes functions to determine the
2254 "cutoff" times due to rotation.
2256 * journald now understands SIGUSR1 and SIGUSR2 for triggering
2257 immediately flushing of runtime logs to /var if possible,
2258 resp. for triggering immediate rotation of the journal
2261 * It is now considered an error if a service is attempted to
2262 be stopped that is not loaded.
2264 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
2266 * systemd-analyze now supports Python 3
2268 * tmpfiles now supports cleaning up directories via aging
2269 where the first level dirs are always kept around but
2270 directories beneath it automatically aged. This is enabled
2271 by prefixing the age field with '~'.
2273 * Seat objects now expose CanGraphical, CanTTY properties
2274 which is required to deal with very fast bootups where the
2275 display manager might be running before the graphics drivers
2276 completed initialization.
2278 * Seat objects now expose a State property.
2280 * We now include RPM macros for service enabling/disabling
2281 based on the preset logic. We recommend RPM based
2282 distributions to make use of these macros if possible. This
2283 makes it simpler to reuse RPM spec files across
2286 * We now make sure that the collected systemd unit name is
2287 always valid when services log to the journal via
2290 * There's a new man page kernel-command-line(7) detailing all
2291 command line options we understand.
2293 * The fstab generator may now be disabled at boot by passing
2294 fstab=0 on the kernel command line.
2296 * A new kernel command line option modules-load= is now understood
2297 to load a specific kernel module statically, early at boot.
2299 * Unit names specified on the systemctl command line are now
2300 automatically escaped as needed. Also, if file system or
2301 device paths are specified they are automatically turned
2302 into the appropriate mount or device unit names. Example:
2304 systemctl status /home
2305 systemctl status /dev/sda
2307 * The SysVConsole= configuration option has been removed from
2308 system.conf parsing.
2310 * The SysV search path is no longer exported on the D-Bus
2313 * The Names= option is been removed from unit file parsing.
2315 * There's a new man page bootup(7) detailing the boot process.
2317 * Every unit and every generator we ship with systemd now
2318 comes with full documentation. The self-explanatory boot is
2321 * A couple of services gained "systemd-" prefixes in their
2322 name if they wrap systemd code, rather than only external
2323 code. Among them fsck@.service which is now
2324 systemd-fsck@.service.
2326 * The HaveWatchdog property has been removed from the D-Bus
2329 * systemd.confirm_spawn= on the kernel command line should now
2332 * There's a new man page crypttab(5) which details all options
2333 we actually understand.
2335 * systemd-nspawn gained a new --capability= switch to pass
2336 additional capabilities to the container.
2338 * timedated will now read known NTP implementation unit names
2339 from /usr/lib/systemd/ntp-units.d/*.list,
2340 systemd-timedated-ntp.target has been removed.
2342 * journalctl gained a new switch "-b" that lists log data of
2343 the current boot only.
2345 * The notify socket is in the abstract namespace again, in
2346 order to support daemons which chroot() at start-up.
2348 * There is a new Storage= configuration option for journald
2349 which allows configuration of where log data should go. This
2350 also provides a way to disable journal logging entirely, so
2351 that data collected is only forwarded to the console, the
2352 kernel log buffer or another syslog implementation.
2354 * Many bugfixes and optimizations
2356 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
2357 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
2358 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
2359 Shawn Landden, Tom Gundersen
2363 * "systemctl help <unit>" now shows the man page if one is
2366 * Several new man pages have been added.
2368 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
2369 MaxLevelConsole= can now be specified in
2370 journald.conf. These options allow reducing the amount of
2371 data stored on disk or forwarded by the log level.
2373 * TimerSlackNSec= can now be specified in system.conf for
2374 PID1. This allows system-wide power savings.
2376 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
2377 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
2382 * logind is now capable of (optionally) handling power and
2383 sleep keys as well as the lid switch.
2385 * journalctl now understands the syntax "journalctl
2386 /usr/bin/avahi-daemon" to get all log output of a specific
2389 * CapabilityBoundingSet= in system.conf now also influences
2390 the capability bound set of usermode helpers of the kernel.
2392 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
2393 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
2394 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
2395 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
2399 * Note that we skipped 139 releases here in order to set the
2400 new version to something that is greater than both udev's
2401 and systemd's most recent version number.
2403 * udev: all udev sources are merged into the systemd source tree now.
2404 All future udev development will happen in the systemd tree. It
2405 is still fully supported to use the udev daemon and tools without
2406 systemd running, like in initramfs or other init systems. Building
2407 udev though, will require the *build* of the systemd tree, but
2408 udev can be properly *run* without systemd.
2410 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
2411 should be used to create dead device nodes as workarounds for broken
2414 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
2415 no longer supported. udev_monitor_new_from_netlink() needs to be
2416 used to subscribe to events.
2418 * udev: when udevd is started by systemd, processes which are left
2419 behind by forking them off of udev rules, are unconditionally cleaned
2420 up and killed now after the event handling has finished. Services or
2421 daemons must be started as systemd services. Services can be
2422 pulled-in by udev to get started, but they can no longer be directly
2423 forked by udev rules.
2425 * udev: the daemon binary is called systemd-udevd now and installed
2426 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
2427 to adapt to that, create symlink, or rename the binary after building
2430 * libudev no longer provides these symbols:
2431 udev_monitor_from_socket()
2432 udev_queue_get_failed_list_entry()
2433 udev_get_{dev,sys,run}_path()
2434 The versions number was bumped and symbol versioning introduced.
2436 * systemd-loginctl and systemd-journalctl have been renamed
2437 to loginctl and journalctl to match systemctl.
2439 * The config files: /etc/systemd/systemd-logind.conf and
2440 /etc/systemd/systemd-journald.conf have been renamed to
2441 logind.conf and journald.conf. Package updates should rename
2442 the files to the new names on upgrade.
2444 * For almost all files the license is now LGPL2.1+, changed
2445 from the previous GPL2.0+. Exceptions are some minor stuff
2446 of udev (which will be changed to LGPL2.1 eventually, too),
2447 and the MIT licensed sd-daemon.[ch] library that is suitable
2448 to be used as drop-in files.
2450 * systemd and logind now handle system sleep states, in
2451 particular suspending and hibernating.
2453 * logind now implements a sleep/shutdown/idle inhibiting logic
2454 suitable for a variety of uses. Soonishly Lennart will blog
2455 about this in more detail.
2457 * var-run.mount and var-lock.mount are no longer provided
2458 (which prevously bind mounted these directories to their new
2459 places). Distributions which have not converted these
2460 directories to symlinks should consider stealing these files
2461 from git history and add them downstream.
2463 * We introduced the Documentation= field for units and added
2464 this to all our shipped units. This is useful to make it
2465 easier to explore the boot and the purpose of the various
2468 * All smaller setup units (such as
2469 systemd-vconsole-setup.service) now detect properly if they
2470 are run in a container and are skipped when
2471 appropriate. This guarantees an entirely noise-free boot in
2472 Linux container environments such as systemd-nspawn.
2474 * A framework for implementing offline system updates is now
2475 integrated, for details see:
2476 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
2478 * A new service type Type=idle is available now which helps us
2479 avoiding ugly interleaving of getty output and boot status
2482 * There's now a system-wide CapabilityBoundingSet= option to
2483 globally reduce the set of capabilities for the
2484 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
2485 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
2486 even CAP_NET_ADMIN system-wide for secure systems.
2488 * There are now system-wide DefaultLimitXXX= options to
2489 globally change the defaults of the various resource limits
2490 for all units started by PID 1.
2492 * Harald Hoyer's systemd test suite has been integrated into
2493 systemd which allows easy testing of systemd builds in qemu
2494 and nspawn. (This is really awesome! Ask us for details!)
2496 * The fstab parser is now implemented as generator, not inside
2499 * systemctl will now warn you if .mount units generated from
2500 /etc/fstab are out of date due to changes in fstab that
2501 haven't been read by systemd yet.
2503 * systemd is now suitable for usage in initrds. Dracut has
2504 already been updated to make use of this. With this in place
2505 initrds get a slight bit faster but primarily are much
2506 easier to introspect and debug since "systemctl status" in
2507 the host system can be used to introspect initrd services,
2508 and the journal from the initrd is kept around too.
2510 * systemd-delta has been added, a tool to explore differences
2511 between user/admin configuration and vendor defaults.
2513 * PrivateTmp= now affects both /tmp and /var/tmp.
2515 * Boot time status messages are now much prettier and feature
2516 proper english language. Booting up systemd has never been
2519 * Read-ahead pack files now include the inode number of all
2520 files to pre-cache. When the inode changes the pre-caching
2521 is not attempted. This should be nicer to deal with updated
2522 packages which might result in changes of read-ahead
2525 * We now temporaritly lower the kernel's read_ahead_kb variable
2526 when collecting read-ahead data to ensure the kernel's
2527 built-in read-ahead does not add noise to our measurements
2528 of necessary blocks to pre-cache.
2530 * There's now RequiresMountsFor= to add automatic dependencies
2531 for all mounts necessary for a specific file system path.
2533 * MountAuto= and SwapAuto= have been removed from
2534 system.conf. Mounting file systems at boot has to take place
2537 * nspawn now learned a new switch --uuid= to set the machine
2538 ID on the command line.
2540 * nspawn now learned the -b switch to automatically search
2543 * vt102 is now the default TERM for serial TTYs, upgraded from
2546 * systemd-logind now works on VT-less systems.
2548 * The build tree has been reorganized. The individual
2549 components now have directories of their own.
2551 * A new condition type ConditionPathIsReadWrite= is now available.
2553 * nspawn learned the new -C switch to create cgroups for the
2554 container in other hierarchies.
2556 * We now have support for hardware watchdogs, configurable in
2559 * The scheduled shutdown logic now has a public API.
2561 * We now mount /tmp as tmpfs by default, but this can be
2562 masked and /etc/fstab can override it.
2564 * Since udisks doesn't make use of /media anymore we are not
2565 mounting a tmpfs on it anymore.
2567 * journalctl gained a new --local switch to only interleave
2568 locally generated journal files.
2570 * We can now load the IMA policy at boot automatically.
2572 * The GTK tools have been split off into a systemd-ui.
2574 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
2575 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
2576 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
2577 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
2578 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
2579 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
2580 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
2581 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
2582 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
2587 * This is mostly a bugfix release
2589 * Support optional initialization of the machine ID from the
2590 KVM or container configured UUID.
2592 * Support immediate reboots with "systemctl reboot -ff"
2594 * Show /etc/os-release data in systemd-analyze output
2596 * Many bugfixes for the journal, including endianness fixes and
2597 ensuring that disk space enforcement works
2599 * sd-login.h is C++ comptaible again
2601 * Extend the /etc/os-release format on request of the Debian
2604 * We now refuse non-UTF8 strings used in various configuration
2605 and unit files. This is done to ensure we don't pass invalid
2606 data over D-Bus or expose it elsewhere.
2608 * Register Mimo USB Screens as suitable for automatic seat
2611 * Read SELinux client context from journal clients in a race
2614 * Reorder configuration file lookup order. /etc now always
2615 overrides /run in order to allow the administrator to always
2616 and unconditionally override vendor supplied or
2617 automatically generated data.
2619 * The various user visible bits of the journal now have man
2620 pages. We still lack man pages for the journal API calls
2623 * We now ship all man pages in HTML format again in the
2626 Contributions from: Dave Reisner, Dirk Eibach, Frederic
2627 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
2628 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
2633 * This is mostly a bugfix release
2635 * systems lacking /etc/os-release are no longer supported.
2637 * Various functionality updates to libsystemd-login.so
2639 * Track class of PAM logins to distuingish greeters from
2642 Contributions from: Kay Sievers, Lennart Poettering, Michael
2647 * This is an important bugfix release for v41.
2649 * Building man pages is now optional which should be useful
2650 for those building systemd from git but unwilling to install
2653 * Watchdog support for supervising services is now usable. In
2654 a future release support for hardware watchdogs
2655 (i.e. /dev/watchdog) will be added building on this.
2657 * Service start rate limiting is now configurable and can be
2658 turned off per service. When a start rate limit is hit a
2659 reboot can automatically be triggered.
2661 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
2663 Contributions from: Benjamin Franzke, Bill Nottingham,
2664 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
2665 Schmidt, Michał Górny, Piotr Drąg
2669 * The systemd binary is installed /usr/lib/systemd/systemd now;
2670 An existing /sbin/init symlink needs to be adapted with the
2673 * The code that loads kernel modules has been ported to invoke
2674 libkmod directly, instead of modprobe. This means we do not
2675 support systems with module-init-tools anymore.
2677 * Watchdog support is now already useful, but still not
2680 * A new kernel command line option systemd.setenv= is
2681 understood to set system wide environment variables
2682 dynamically at boot.
2684 * We now limit the set of capabilities of systemd-journald.
2686 * We now set SIGPIPE to ignore by default, since it only is
2687 useful in shell pipelines, and has little use in general
2688 code. This can be disabled with IgnoreSIPIPE=no in unit
2691 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
2692 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
2697 * This is mostly a bugfix release
2699 * We now expose the reason why a service failed in the
2700 "Result" D-Bus property.
2702 * Rudimentary service watchdog support (will be completed over
2703 the next few releases.)
2705 * When systemd forks off in order execute some service we will
2706 now immediately changes its argv[0] to reflect which process
2707 it will execute. This is useful to minimize the time window
2708 with a generic argv[0], which makes bootcharts more useful
2710 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
2711 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
2712 Mike Kazantsev, Ray Strode
2716 * This is mostly a test release, but incorporates many
2719 * New systemd-cgtop tool to show control groups by their
2722 * Linking against libacl for ACLs is optional again. If
2723 disabled, support tracking device access for active logins
2724 goes becomes unavailable, and so does access to the user
2725 journals by the respective users.
2727 * If a group "adm" exists, journal files are automatically
2728 owned by them, thus allow members of this group full access
2729 to the system journal as well as all user journals.
2731 * The journal now stores the SELinux context of the logging
2732 client for all entries.
2734 * Add C++ inclusion guards to all public headers
2736 * New output mode "cat" in the journal to print only text
2737 messages, without any meta data like date or time.
2739 * Include tiny X server wrapper as a temporary stop-gap to
2740 teach XOrg udev display enumeration. This is used by display
2741 managers such as gdm, and will go away as soon as XOrg
2742 learned native udev hotplugging for display devices.
2744 * Add new systemd-cat tool for executing arbitrary programs
2745 with STDERR/STDOUT connected to the journal. Can also act as
2746 BSD logger replacement, and does so by default.
2748 * Optionally store all locally generated coredumps in the
2749 journal along with meta data.
2751 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
2752 writing short strings to files (for usage for /sys), and for
2753 creating symlinks, character and block device nodes.
2755 * New unit file option ControlGroupPersistent= to make cgroups
2756 persistent, following the mechanisms outlined in
2757 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
2759 * Support multiple local RTCs in a sane way
2761 * No longer monopolize IO when replaying readahead data on
2762 rotating disks, since we might starve non-file-system IO to
2763 death, since fanotify() will not see accesses done by blkid,
2766 * Don't show kernel threads in systemd-cgls anymore, unless
2767 requested with new -k switch.
2769 Contributions from: Dan Horák, Kay Sievers, Lennart
2770 Poettering, Michal Schmidt
2774 * This is mostly a test release, but incorporates many
2777 * The git repository moved to:
2778 git://anongit.freedesktop.org/systemd/systemd
2779 ssh://git.freedesktop.org/git/systemd/systemd
2781 * First release with the journal
2782 http://0pointer.de/blog/projects/the-journal.html
2784 * The journal replaces both systemd-kmsg-syslogd and
2785 systemd-stdout-bridge.
2787 * New sd_pid_get_unit() API call in libsystemd-logind
2789 * Many systemadm clean-ups
2791 * Introduce remote-fs-pre.target which is ordered before all
2792 remote mounts and may be used to start services before all
2795 * Added Mageia support
2797 * Add bash completion for systemd-loginctl
2799 * Actively monitor PID file creation for daemons which exit in
2800 the parent process before having finished writing the PID
2801 file in the daemon process. Daemons which do this need to be
2802 fixed (i.e. PID file creation must have finished before the
2803 parent exits), but we now react a bit more gracefully to them.
2805 * Add colourful boot output, mimicking the well-known output
2806 of existing distributions.
2808 * New option PassCredentials= for socket units, for
2809 compatibility with a recent kernel ABI breakage.
2811 * /etc/rc.local is now hooked in via a generator binary, and
2812 thus will no longer act as synchronization point during
2815 * systemctl list-unit-files now supports --root=.
2817 * systemd-tmpfiles now understands two new commands: z, Z for
2818 relabelling files according to the SELinux database. This is
2819 useful to apply SELinux labels to specific files in /sys,
2822 * Output of SysV services is now forwarded to both the console
2823 and the journal by default, not only just the console.
2825 * New man pages for all APIs from libsystemd-login.
2827 * The build tree got reorganized and a the build system is a
2828 lot more modular allowing embedded setups to specifically
2829 select the components of systemd they are interested in.
2831 * Support for Linux systems lacking the kernel VT subsystem is
2834 * configure's --with-rootdir= got renamed to
2835 --with-rootprefix= to follow the naming used by udev and
2838 * Unless specified otherwise we'll now install to /usr instead
2839 of /usr/local by default.
2841 * Processes with '@' in argv[0][0] are now excluded from the
2842 final shut-down killing spree, following the logic explained
2844 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
2846 * All processes remaining in a service cgroup when we enter
2847 the START or START_PRE states are now killed with
2848 SIGKILL. That means it is no longer possible to spawn
2849 background processes from ExecStart= lines (which was never
2850 supported anyway, and bad style).
2852 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
2853 reloading of units together.
2855 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
2856 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
2857 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
2858 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
2859 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek