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 to set up bridges, VLANs and bonding.
8 This currently provides no hookups for interactive network
9 configuration. Use this for your initrd, container, embedded
10 or server setup, if you need a simple, yet powerful network
11 configuration solution. This configuration subsystem is
12 quite nifty as it allows wildcard hotplug matching in
13 interfaces. For example, with a single configuration snippet
14 you can configure that all ethernet interfaces showing up
15 are automatically added to a bridge, or similar. It
16 optionally 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 kbd backlights in addition to display
28 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, WakeOnLan settings, MAC address, MAC
40 address assignment policy (randomized, ...).
42 * When the User= switch is used in a unit file, also
43 initialize $SHELL= based on the user database entry.
45 * systemd no longer depends on libdbus. All communication is
46 now done with sd-bus, systemd's low-level bus library
49 * kdbus support has been added to PID 1 itself. When kdbus is
50 enabled, this causes PID 1 to set up the system bus and
51 enable support for a new ".busname" unit type that
52 encapsulates bus name activation on kdbus. It works a little
53 bit like ".socket" units, except for bus names. A new
54 generator has been added that converts classic dbus1 service
55 activation files automatically into native systemd .busname
58 * sd-bus: add a light-weight vtable implementation that allows
59 defining objects on the bus with a simple static const
60 vtable array of its methods, signals and properties.
62 * systemd will not generate nor install static dbus
63 introspection data anymore to /usr/share/dbus-1/interfaces,
64 as the precise format of these files are unclear, and
65 nothing makes use of it.
67 * A proxy daemon is now provided to proxy clients connecting
68 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
69 compatibility with classic D-Bus.
71 * A bus driver implementation has been added that supports the
72 classic D-Bus bus driver calls on kdbus, also for
73 compatibility purposes.
75 * A new API "sd-event.h" has been added that implements a
76 minimal event loop API built around epoll. It provides a
77 couple of features that direct epoll usage is lacking:
78 prioritization of events, scales to large numbers of timer
79 events, per-event timer slack (accuracy), system-wide
80 coalescing of timer events, exit handlers, watchdog
81 supervision support using systemd's sd_notify() API, child
84 * A new API "sd-rntl.h" has been added that provides an API
85 around the route netlink interface of the kernel, similar in
88 * A new API "sd-dhcp.h" has been added that provides a small
89 DHCPv4 client-side implementation. This is used by
92 * There is a new kernel command line option
93 "systemd.restore_state". When set, none of the systemd tools
94 will restore saved runtime state to hardware devices. More
95 specifically, the rfkill and backlight states are not
98 * The FsckPassNo= compatibility option in mount/service units
99 has been removed. The fstab generator will now add the
100 necessary dependencies automatically, and does not require
101 PID1's support for that anymore.
103 * journalctl gained a new switch --list-boots, that lists
104 recent boots with their times and boot IDs.
106 * The various tools like systemctl, loginctl, timedatectl,
107 busctl, systemd-run, ... have gained a new switch "-M" to
108 connect to a specific, local OS container (as direct
109 connection, without requiring SSH). This works on any
110 container that is registered with machined, such as those
111 created by libvirt-lxc or nspawn.
113 * systemd-run and systemd-analyze also gained support for "-H"
114 to connect to remote hosts via SSH. This is particularly
115 useful for systemd-run since it enables queuing of jobs onto
118 * machinectl gained a new command "login" to open a getty
119 login in any local container. This works with any container
120 that is registered with machined (such as those created by
121 libvirt-lxc or nspawn), and which run systemd inside.
123 * machinectl gained a new "reboot" command that may be used to
124 trigger a reboot on a specific container that is registered
125 with machined. This works on any container that runs an init
128 * systemctl gained a new "list-timers" command to print a nice
129 listing of installed timer units with the times they elapse
132 * Alternative reboot() parameters may now be specified on the
133 "systemctl reboot" command line and are passed to the
134 reboot() system call.
136 * systemctl gained a new --job-mode= switch to configure the
137 mode to queue a job with. This is a more generic version of
138 --fail, --irreversible, --ignore-dependencies which are
139 still available but not advertised anymore.
141 * systemd-activate gained a new --setenv= parameter to specify
142 additional environment variables to pass to the executed
145 * /etc/systemd/system.conf gained new settings to configure
146 various default timeouts of units, as well as the default
147 start limit interval and burst. These may still be overridden
150 * PID1 will now export on the bus profile data of the security
151 policy upload process (such as the SELinux policy upload to
154 * journald: when forwarding logs to the console, include
157 * OnCalendar= in timer units now understands the special
158 strings "yearly" and "annually". (Both are equivalent)
160 * The accuracy of timer units is now configurable with the new
161 AccuracySec= setting. It defaults to 1min.
163 * A new dependency type JoinsNamespaceOf= has been added that
164 allows running two services within the same /tmp and network
165 namespace, if PrivateNetwork= or PrivateTmp= are used.
167 * A new command "cat" has been added to systemctl. It outputs
168 the original unit file of a unit, and concatenates the
169 contents of addition "drop-in" unit file snippets to it, so
170 that the full configuration is shown.
172 * systemctl now supports globbing on the various "list-xyz"
173 commands, like "list-units" or "list-sockets", as well as on
174 thsoe commands which take multiple unit names.
176 * All systemd daemons now make use of the watchdog logic so
177 that systemd automatically notices when they hang.
179 * If the $container_ttys environment variable is set,
180 getty-generator will automatically spawn a getty for each
181 listed tty. This is useful for container managers to request
182 login gettys to be spawned on as many ttys as needed.
184 * %h, %s, %U specifier support is not available anymore when
185 used in unit files for PID 1. This is because NSS calls are
186 not safe from PID 1. They stay available for --user
187 instances of systemd, and as special case for the root user.
189 * loginctl gained a new "--no-legend" switch to turn off output
192 * The "sd-login.h" API gained three new calls:
193 sd_session_is_remote(), sd_session_get_remote_user(),
194 sd_session_get_remote_host() to query information about
197 * The udev device database now also carries vendor/product
198 information about SDIO devices.
200 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
201 determine whether watchdog notifications are requested by
204 * "systemd-delta" will now also display changes made via .d/
205 drop-ins for unit files.
207 * Socket-activated per-connection services will now include a
208 short description of the connection parameters in the
211 * tmpfiles gained a new "--boot" option. When this is not used,
212 only lines where the command character is not suffixed with
213 "!" are executed. When this option is specified, those
214 options are executed too. This is useful to ensure that
215 specific lines are not executed by accident during runtime,
216 and only at boot. (For example, a line that creates
219 * A new API "sd-resolv.h" has been added which provides a simple
220 asynchronous wrapper around glibc NSS host name resolution
221 calls, such as getaddrinfo(). In contrast to glibc's
222 getaddrinfo_a(), it does not use signals. In contrast to most
223 other asynchronous name resolution libraries, this one does
224 not reimplement DNS, but reuses NSS, so that alternate
225 host name resolution systems continue to work, such as mDNS,
226 LDAP, etc. This API is based on libasyncns, but has been
227 cleaned up for inclusion in systemd.
229 * journalctl's --unit= switch gained support for globbing.
231 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h" are no
232 longer found in individual libraries libsystemd-journal.so,
233 libsystemd-login.so, libsystemd-id128.so. Instead, we have
234 merged them into a single library libsystemd.so which
235 provides all symbols. The reason for this are cyclic
236 dependencies, as these libraries tend to use each other's
237 symbols. So far, we managed to workaround that by linking a
238 copy of a good part of our code into each of these libraries
239 again and again, which however makes certain things hard to
240 do, like sharing static variables. Also, it substantially
241 increases footprint. With this change, there is only one
242 library for the basic APIs systemd provides. Also,
243 "sd-bus.h", "sd-memfd.h", "sd-event.h", "sd-rtnl.h",
244 "sd-resolve.h", "sd-utf8.h" are found in this library as
245 well, however are subject to the --enable-kdbus switch (see
246 below). Note that "sd-dhcp.h" and "sd-daemon.h" are not part
247 of this libraries (the former because it only consumes,
248 never provides services of/to other APIs, and the latter
249 because it is completely standalone). To make the transition
250 from the separate libraries to the unified one easy, we
251 provide the --enable-compat-libs compile-time switch which
252 will generate stub libraries that are compatible with the
253 old ones but redirect all calls to the new one.
255 * All the kdbus logic and the new APIs "sd-bus.h",
256 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
257 "sd-utf8.h" is compile-time optional via the
258 "--enable-kdbus" switch, and is not compiled in by
259 default. To make use of it, you have to explicitly enable
260 the switch. Note however, that neither the kernel nor the
261 userspace API for all of this is considered stable yet. We
262 want to maintain the freedom to still change the APIs for
263 now. By specifying this build-time switch, you acknowledge
264 that you are aware of the instability of the current
267 * Also, note that while kdbus is pretty much complete,
268 it lacks one thing: proper policy support. This means you
269 can build a fully working system with all features, however
270 it will be highly insecure. Policy support will be added in
271 one of the next releases, at the same time that we will
272 declare the APIs stable.
274 * When the kernel command line argument "kdbus" is specified,
275 systemd will automatically load the kdbus.ko kernel module. At
276 this stage of development, it is only useful for testing kdbus,
277 and should not be used in production. Note: if "--enable-kdbus"
278 is specified and the kdbus.ko kernel module is available and
279 "kdbus" is added to the kernel command line, the entire system
280 runs with kdbus instead of dbus-daemon, with the above mentioned
281 problem of missing the system policy enforcement. Also a future
282 version of kdbus.ko or a newer systemd will not be compatible with
283 each other, and will unlikely be able to boot the machine if only
284 one of them is updated.
286 * systemctl gained a new "import-environment" command which
287 uploads the caller's environment (or parts thereof) into the
288 service manager so that it is inherited by services started
289 by the manager. This is useful to upload variables like
290 $DISPLAY into the user service manager.
292 * A new PrivateDevices= switch has been added to service units
293 which allows running a service with a namespaced /dev
294 directory that does not contain any device nodes for
295 physical devices. More specifically, it only includes devices
296 such as /dev/null, /dev/urandom and /dev/zero which are API
299 * logind has been extended to support behaviour like VT
300 switching on seats that do not support a VT. This makes
301 multi-session available on seats that are not the first seat
302 (seat0), and on systems where kernel support for VTs has
303 been disabled at compile time.
305 * If a process holds a delay lock for system sleep or shutdown
306 and fails to release it in time, we will now log about its
307 identity. This makes it easier to identify processes that
308 cause slow suspends or power-offs.
310 * When parsing /etc/crypttab, support a new key-slot= option
311 as supported by Debian, which allows indicating which LUKS
314 * The boot-time has been improved to show information about
315 timeouts that are expiring as they are expiring.
317 * The sd_journald_sendv() API call has been updated to be
318 async-signal-safe so that it may be invoked from signal
319 handlers for logging purposes.
321 * Boot-time status output is now enabled automatically after a
322 short timeout if boot does not progress, in order to give
323 the user an indication what he is waiting for.
325 * The KillMode= switch in service units gained a new possible
326 value "mixed". If set and the unit is shutdown, then the
327 initial SIGTERM signal is sent only to the main daemon
328 process, while the following SIGKILL signal is then sent to
329 all remaining processes of the service.
331 * When a scope unit is registered, a new property "Controller"
332 may be set. If set to a valid bus name, systemd will send a
333 RequestStop() signal to this name when it would like to shut
334 down the scope. This may be used to hook manager logic into
335 the shutdown logic of scope units. Also, scope units may now
336 be put in a special "abandoned" state in which case the
337 manager process which created them takes no further
338 responsibilities for it.
340 * When reading unit files, systemd will now implicitly verify
341 the access mode of these files, and warn about certain
342 suspicious combinations. This has been added to make it
343 easier to track down packaging bugs where unit files are
344 marked executable or world-writable.
346 * systemd-nspawn gained a new "--setenv=" switch to set
347 container-wide environment variables.
349 * systemd-nspawn has been updated to create a new kdbus domain
350 for each container that is invoked, thus allowing each
351 container to have its own set of system and user buses,
352 independently of the host.
354 * systemd-nspawn gained a new --drop-capability= switch to run
355 the container with less capabilities than the default. Both
356 --drop-capability= and --capability= now take the special
357 string "all" for dropping or keeping all capabilities.
359 * systemd-nspawn gained new switches for executing containers
360 with specific SELinux labels set.
362 * systemd-nspawn gained a new --quiet switch to not generate
363 any additional output but the container's own console
366 * systemd-nspawn gained a new --share-system switch to run a
367 container without PID namespacing enabled.
369 * systemd-nspawn gained a new --register= switch to control
370 whether the container is registered with machined or
371 not. This is useful for containers that do not register full
372 OS images, but only specific apps.
374 * systemd-nspawn gained a new --keep-unit which may be used
375 when invoked as only program from a service unit, and
376 results in registration of the unit service itself in
377 machined, instead of a newly opened scope unit.
379 * systemd-nspawn gained a new --network-interface= switch for
380 moving arbitrary interfaces to the container. The new
381 --network-veth switch creates a virtual Ethernet connection
382 between host and container. Thew new --network-bridge=
383 switch then additionally allows assigning the host side of
384 this virtual Ethernet connection to a bridge device.
386 * logind will now also track a "Desktop" identifier for each
387 session which encodes the desktop environment of it. This is
388 useful for desktop environments that want to identify
389 multiple running sessions of itself easily.
391 * A new SELinuxContext= setting for service units has been
392 added that allows setting a specific SELinux execution
393 context for a service.
395 * Most systemd client tools will now honour $SYSTEMD_LESS for
396 settings of the "less" pager. By default, these tools will
397 override $LESS to allow certain operations like
398 jump-to-the-end work. With $SYSTEMD_LESS, it is possible to
399 influence this logic.
401 * systemd's "seccomp" hook-up has been changed to make use of
402 the libseccomp library instead of using its own
403 implementation. This has benefits for portability among
406 * For usage together with SystemCallFilter=, a new
407 SystemCallErrorNumber= setting has been introduce that
408 allows configuration if a system error number to return on
409 filtered syscalls, instead of immediately killing the
410 process. Also, SystemCallArchitectures= has been added to
411 limit access to system calls of a particular architecture
412 (in order to turn off support for unused secondary
413 architectures). There is also a global
414 SystemcallArchitecture= setting in system.conf now to turn
415 off support for non-native system calls system-wide.
417 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
418 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
419 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
420 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
421 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
422 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
423 Elia Pinto, Florian Weimer, George McCollister, Goffredo
424 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
425 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
426 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
427 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
428 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
429 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
430 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
431 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
432 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
433 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
434 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
435 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
436 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
437 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
438 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
439 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
440 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
442 -- Berlin, 2014-02-xx
446 * logind has gained support for facilitating privileged input
447 and drm device access for unprivileged clients. This work is
448 useful to allow Wayland display servers (and similar
449 programs, such as kmscon) to run under the user's ID and
450 access input and drm devices which are normally
451 protected. When this is used (and the kernel is new enough)
452 logind will "mute" IO on the file descriptors passed to
453 Wayland as long as it is in the background and "unmute" it
454 if it returns into the foreground. This allows secure
455 session switching without allowing background sessions to
456 eavesdrop on input and display data. This also introduces
457 session switching support if VT support is turned off in the
458 kernel, and on seats that are not seat0.
460 * A new kernel command line option luks.options= is understood
461 now which allows specifiying LUKS options for usage for LUKS
462 encrypted partitions specified with luks.uuid=.
464 * tmpfiles.d(5) snippets may now use specifier expansion in
465 path names. More specifically %m, %b, %H, %v, are now
466 replaced by the local machine id, boot id, hostname, and
467 kernel version number.
469 * A new tmpfiles.d(5) command "m" has been introduced which
470 may be used to change the owner/group/access mode of a file
471 or directory if it exists, but do nothing if it doesn't.
473 * This release removes high-level support for the
474 MemorySoftLimit= cgroup setting. The underlying kernel
475 cgroup attribute memory.soft_limit= is currently badly
476 designed and likely to be removed from the kernel API in its
477 current form, hence we shouldn't expose it for now.
479 * The memory.use_hierarchy cgroup attribute is now enabled for
480 all cgroups systemd creates in the memory cgroup
481 hierarchy. This option is likely to be come the built-in
482 default in the kernel anyway, and the non-hierarchial mode
483 never made much sense in the intrinsically hierarchial
486 * A new field _SYSTEMD_SLICE= is logged along with all journal
487 messages containing the slice a message was generated
488 from. This is useful to allow easy per-customer filtering of
489 logs among other things.
491 * systemd-journald will no longer adjust the group of journal
492 files it creates to the "systemd-journal" group. Instead we
493 rely on the journal directory to be owned by the
494 "systemd-journal" group, and its setgid bit set, so that the
495 kernel file system layer will automatically enforce that
496 journal files inherit this group assignment. The reason for
497 this change is that we cannot allow NSS look-ups from
498 journald which would be necessary to resolve
499 "systemd-journal" to a numeric GID, because this might
500 create deadlocks if NSS involves synchronous queries to
501 other daemons (such as nscd, or sssd) which in turn are
502 logging clients of journald and might block on it, which
503 would then dead lock. A tmpfiles.d(5) snippet included in
504 systemd will make sure the setgid bit and group are
505 properly set on the journal directory if it exists on every
506 boot. However, we recommend adjusting it manually after
507 upgrades too (or from RPM scriptlets), so that the change is
508 not delayed until next reboot.
510 * Backlight and random seed files in /var/lib/ have moved into
511 the /var/lib/systemd/ directory, in order to centralize all
512 systemd generated files in one directory.
514 * Boot time performance measurements (as displayed by
515 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
516 performance information if that's available to determine how
517 much time BIOS and boot loader initialization required. With
518 a sufficiently new BIOS you hence no longer need to boot
519 with Gummiboot to get access to such information.
521 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
522 Cristian Rodríguez, Dave Reisner, David Herrmann, David
523 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
524 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
525 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
526 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
527 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
529 -- Berlin, 2013-10-02
533 * The Restart= option for services now understands a new
534 on-watchdog setting, which will restart the service
535 automatically if the service stops sending out watchdog keep
536 alive messages (as configured with WatchdogSec=).
538 * The getty generator (which is responsible for bringing up a
539 getty on configured serial consoles) will no longer only
540 start a getty on the primary kernel console but on all
541 others, too. This makes the order in which console= is
542 specified on the kernel command line less important.
544 * libsystemd-logind gained a new sd_session_get_vt() call to
545 retrieve the VT number of a session.
547 * If the option "tries=0" is set for an entry of /etc/crypttab
548 its passphrase is queried indefinitely instead of any
549 maximum number of tries.
551 * If a service with a configure PID file terminates its PID
552 file will now be removed automatically if it still exists
553 afterwards. This should put an end to stale PID files.
555 * systemd-run will now also take relative binary path names
556 for execution and no longer insists on absolute paths.
558 * InaccessibleDirectories= and ReadOnlyDirectories= now take
559 paths that are optionally prefixed with "-" to indicate that
560 it should not be considered a failure if they don't exist.
562 * journalctl -o (and similar commands) now understands a new
563 output mode "short-precise", it is similar to "short" but
564 shows timestamps with usec accuracy.
566 * The option "discard" (as known from Debian) is now
567 synonymous to "allow-discards" in /etc/crypttab. In fact,
568 "discard" is preferred now (since it is easier to remember
571 * Some licensing clean-ups were made, so that more code is now
572 LGPL-2.1 licensed than before.
574 * A minimal tool to save/restore the display backlight
575 brightness across reboots has been added. It will store the
576 backlight setting as late as possible at shutdown, and
577 restore it as early as possible during reboot.
579 * A logic to automatically discover and enable home and swap
580 partitions on GPT disks has been added. With this in place
581 /etc/fstab becomes optional for many setups as systemd can
582 discover certain partitions located on the root disk
583 automatically. Home partitions are recognized under their
584 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
585 partitions are recognized under their GPT type ID
586 0657fd6da4ab43c484e50933c84b4f4f.
588 * systemd will no longer pass any environment from the kernel
589 or initrd to system services. If you want to set an
590 environment for all services, do so via the kernel command
591 line systemd.setenv= assignment.
593 * The systemd-sysctl tool no longer natively reads the file
594 /etc/sysctl.conf. If desired, the file should be symlinked
595 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
596 legacy support by a symlink rather than built-in code, it
597 also makes the otherwise hidden order of application of the
598 different files visible. (Note that this partly reverts to a
599 pre-198 application order of sysctl knobs!)
601 * The "systemctl set-log-level" and "systemctl dump" commands
602 have been moved to systemd-analyze.
604 * systemd-run learned the new --remain-after-exit switch,
605 which causes the scope unit not to be cleaned up
606 automatically after the process terminated.
608 * tmpfiles learned a new --exclude-prefix= switch to exclude
609 certain paths from operation.
611 * journald will now automatically flush all messages to disk
612 as soon as a message of the log priorities CRIT, ALERT or
615 Contributions from: Andrew Cook, Brandon Philips, Christian
616 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
617 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
618 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
619 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
620 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
621 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
622 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
623 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
624 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
625 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
626 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
627 William Giokas, Zbigniew Jędrzejewski-Szmek
629 -- Berlin, 2013-09-13
633 * The documentation has been updated to cover the various new
634 concepts introduced with 205.
636 * Unit files now understand the new %v specifier which
637 resolves to the kernel version string as returned by "uname
640 * systemctl now supports filtering the unit list output by
641 load state, active state and sub state, using the new
644 * "systemctl status" will now show the results of the
645 condition checks (like ConditionPathExists= and similar) of
646 the last start attempts of the unit. They are also logged to
649 * "journalctl -b" may now be used to look for boot output of a
650 specific boot. Try "journalctl -b -1" for the previous boot,
651 but the syntax is substantially more powerful.
653 * "journalctl --show-cursor" has been added which prints the
654 cursor string the last shown log line. This may then be used
655 with the new "journalctl --after-cursor=" switch to continue
656 browsing logs from that point on.
658 * "journalctl --force" may now be used to force regeneration
661 * Creation of "dead" device nodes has been moved from udev
662 into kmod and tmpfiles. Previously, udev would read the kmod
663 databases to pre-generate dead device nodes based on meta
664 information contained in kernel modules, so that these would
665 be auto-loaded on access rather then at boot. As this
666 doesn't really have much to do with the exposing actual
667 kernel devices to userspace this has always been slightly
668 alien in the udev codebase. Following the new scheme kmod
669 will now generate a runtime snippet for tmpfiles from the
670 module meta information and it now is tmpfiles' job to the
671 create the nodes. This also allows overriding access and
672 other parameters for the nodes using the usual tmpfiles
673 facilities. As side effect this allows us to remove the
674 CAP_SYS_MKNOD capability bit from udevd entirely.
676 * logind's device ACLs may now be applied to these "dead"
677 devices nodes too, thus finally allowing managed access to
678 devices such as /dev/snd/sequencer whithout loading the
679 backing module right-away.
681 * A new RPM macro has been added that may be used to apply
682 tmpfiles configuration during package installation.
684 * systemd-detect-virt and ConditionVirtualization= now can
685 detect User-Mode-Linux machines (UML).
687 * journald will now implicitly log the effective capabilities
688 set of processes in the message metadata.
690 * systemd-cryptsetup has gained support for TrueCrypt volumes.
692 * The initrd interface has been simplified (more specifically,
693 support for passing performance data via environment
694 variables and fsck results via files in /run has been
695 removed). These features were non-essential, and are
696 nowadays available in a much nicer way by having systemd in
697 the initrd serialize its state and have the hosts systemd
698 deserialize it again.
700 * The udev "keymap" data files and tools to apply keyboard
701 specific mappings of scan to key codes, and force-release
702 scan code lists have been entirely replaced by a udev
703 "keyboard" builtin and a hwdb data file.
705 * systemd will now honour the kernel's "quiet" command line
706 argument also during late shutdown, resulting in a
707 completely silent shutdown when used.
709 * There's now an option to control the SO_REUSEPORT socket
710 option in .socket units.
712 * Instance units will now automatically get a per-template
713 subslice of system.slice unless something else is explicitly
714 configured. For example, instances of sshd@.service will now
715 implicitly be placed in system-sshd.slice rather than
716 system.slice as before.
718 * Test coverage support may now be enabled at build time.
720 Contributions from: Dave Reisner, Frederic Crozat, Harald
721 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
722 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
723 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
724 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
725 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
726 Giokas, Zbigniew Jędrzejewski-Szmek
728 -- Berlin, 2013-07-23
732 * Two new unit types have been introduced:
734 Scope units are very similar to service units, however, are
735 created out of pre-existing processes -- instead of PID 1
736 forking off the processes. By using scope units it is
737 possible for system services and applications to group their
738 own child processes (worker processes) in a powerful way
739 which then maybe used to organize them, or kill them
740 together, or apply resource limits on them.
742 Slice units may be used to partition system resources in an
743 hierarchial fashion and then assign other units to them. By
744 default there are now three slices: system.slice (for all
745 system services), user.slice (for all user sessions),
746 machine.slice (for VMs and containers).
748 Slices and scopes have been introduced primarily in
749 context of the work to move cgroup handling to a
750 single-writer scheme, where only PID 1
751 creates/removes/manages cgroups.
753 * There's a new concept of "transient" units. In contrast to
754 normal units these units are created via an API at runtime,
755 not from configuration from disk. More specifically this
756 means it is now possible to run arbitrary programs as
757 independent services, with all execution parameters passed
758 in via bus APIs rather than read from disk. Transient units
759 make systemd substantially more dynamic then it ever was,
760 and useful as a general batch manager.
762 * logind has been updated to make use of scope and slice units
763 for managing user sessions. As a user logs in he will get
764 his own private slice unit, to which all sessions are added
765 as scope units. We also added support for automatically
766 adding an instance of user@.service for the user into the
767 slice. Effectively logind will no longer create cgroup
768 hierarchies on its own now, it will defer entirely to PID 1
769 for this by means of scope, service and slice units. Since
770 user sessions this way become entities managed by PID 1
771 the output of "systemctl" is now a lot more comprehensive.
773 * A new mini-daemon "systemd-machined" has been added which
774 may be used by virtualization managers to register local
775 VMs/containers. nspawn has been updated accordingly, and
776 libvirt will be updated shortly. machined will collect a bit
777 of meta information about the VMs/containers, and assign
778 them their own scope unit (see above). The collected
779 meta-data is then made available via the "machinectl" tool,
780 and exposed in "ps" and similar tools. machined/machinectl
781 is compile-time optional.
783 * As discussed earlier, the low-level cgroup configuration
784 options ControlGroup=, ControlGroupModify=,
785 ControlGroupPersistent=, ControlGroupAttribute= have been
786 removed. Please use high-level attribute settings instead as
789 * A new bus call SetUnitProperties() has been added to alter
790 various runtime parameters of a unit. This is primarily
791 useful to alter cgroup parameters dynamically in a nice way,
792 but will be extended later on to make more properties
793 modifiable at runtime. systemctl gained a new set-properties
794 command that wraps this call.
796 * A new tool "systemd-run" has been added which can be used to
797 run arbitrary command lines as transient services or scopes,
798 while configuring a number of settings via the command
799 line. This tool is currently very basic, however already
800 very useful. We plan to extend this tool to even allow
801 queuing of execution jobs with time triggers from the
802 command line, similar in fashion to "at".
804 * nspawn will now inform the user explicitly that kernels with
805 audit enabled break containers, and suggest the user to turn
808 * Support for detecting the IMA and AppArmor security
809 frameworks with ConditionSecurity= has been added.
811 * journalctl gained a new "-k" switch for showing only kernel
812 messages, mimicking dmesg output; in addition to "--user"
813 and "--system" switches for showing only user's own logs
816 * systemd-delta can now show information about drop-in
817 snippets extending unit files.
819 * libsystemd-bus has been substantially updated but is still
820 not available as public API.
822 * systemd will now look for the "debug" argument on the kernel
823 command line and enable debug logging, similar to
824 "systemd.log_level=debug" already did before.
826 * "systemctl set-default", "systemctl get-default" has been
827 added to configure the default.target symlink, which
828 controls what to boot into by default.
830 * "systemctl set-log-level" has been added as a convenient
831 way to raise and lower systemd logging threshold.
833 * "systemd-analyze plot" will now show the time the various
834 generators needed for execution, as well as information
835 about the unit file loading.
837 * libsystemd-journal gained a new sd_journal_open_files() call
838 for opening specific journal files. journactl also gained a
839 new switch to expose this new functionality. Previously we
840 only supported opening all files from a directory, or all
841 files from the system, as opening individual files only is
842 racy due to journal file rotation.
844 * systemd gained the new DefaultEnvironment= setting in
845 /etc/systemd/system.conf to set environment variables for
848 * If a privileged process logs a journal message with the
849 OBJECT_PID= field set, then journald will automatically
850 augment this with additional OBJECT_UID=, OBJECT_GID=,
851 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
852 system services want to log events about specific client
853 processes. journactl/systemctl has been updated to make use
854 of this information if all log messages regarding a specific
857 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
858 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
859 Reisner, David Coppa, David King, David Strauss, Eelco
860 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
861 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
862 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
863 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
864 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
865 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
866 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
867 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
868 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
873 * The Python bindings gained some minimal support for the APIs
874 exposed by libsystemd-logind.
876 * ConditionSecurity= gained support for detecting SMACK. Since
877 this condition already supports SELinux and AppArmor we only
878 miss IMA for this. Patches welcome!
880 Contributions from: Karol Lewandowski, Lennart Poettering,
881 Zbigniew Jędrzejewski-Szmek
885 * systemd-nspawn will now create /etc/resolv.conf if
886 necessary, before bind-mounting the host's file onto it.
888 * systemd-nspawn will now store meta information about a
889 container on the container's cgroup as extended attribute
890 fields, including the root directory.
892 * The cgroup hierarchy has been reworked in many ways. All
893 objects any of the components systemd creates in the cgroup
894 tree are now suffixed. More specifically, user sessions are
895 now placed in cgroups suffixed with ".session", users in
896 cgroups suffixed with ".user", and nspawn containers in
897 cgroups suffixed with ".nspawn". Furthermore, all cgroup
898 names are now escaped in a simple scheme to avoid collision
899 of userspace object names with kernel filenames. This work
900 is preparation for making these objects relocatable in the
901 cgroup tree, in order to allow easy resource partitioning of
902 these objects without causing naming conflicts.
904 * systemctl list-dependencies gained the new switches
905 --plain, --reverse, --after and --before.
907 * systemd-inhibit now shows the process name of processes that
908 have taken an inhibitor lock.
910 * nss-myhostname will now also resolve "localhost"
911 implicitly. This makes /etc/hosts an optional file and
912 nicely handles that on IPv6 ::1 maps to both "localhost" and
915 * libsystemd-logind.so gained a new call
916 sd_get_machine_names() to enumerate running containers and
917 VMs (currently only supported by very new libvirt and
918 nspawn). sd_login_monitor can now be used to watch
919 VMs/containers coming and going.
921 * .include is not allowed recursively anymore, and only in
922 unit files. Usually it is better to use drop-in snippets in
923 .d/*.conf anyway, as introduced with systemd 198.
925 * systemd-analyze gained a new "critical-chain" command that
926 determines the slowest chain of units run during system
927 boot-up. It is very useful for tracking down where
928 optimizing boot time is the most beneficial.
930 * systemd will no longer allow manipulating service paths in
931 the name=systemd:/system cgroup tree using ControlGroup= in
932 units. (But is still fine with it in all other dirs.)
934 * There's a new systemd-nspawn@.service service file that may
935 be used to easily run nspawn containers as system
936 services. With the container's root directory in
937 /var/lib/container/foobar it is now sufficient to run
938 "systemctl start systemd-nspawn@foobar.service" to boot it.
940 * systemd-cgls gained a new parameter "--machine" to list only
941 the processes within a certain container.
943 * ConditionSecurity= now can check for "apparmor". We still
944 are lacking checks for SMACK and IMA for this condition
945 check though. Patches welcome!
947 * A new configuration file /etc/systemd/sleep.conf has been
948 added that may be used to configure which kernel operation
949 systemd is supposed to execute when "suspend", "hibernate"
950 or "hybrid-sleep" is requested. This makes the new kernel
951 "freeze" state accessible to the user.
953 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
954 the passed argument if applicable.
956 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
957 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
958 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
959 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
960 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
961 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
962 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
967 * The output of 'systemctl list-jobs' got some polishing. The
968 '--type=' argument may now be passed more than once. A new
969 command 'systemctl list-sockets' has been added which shows
970 a list of kernel sockets systemd is listening on with the
971 socket units they belong to, plus the units these socket
974 * The experimental libsystemd-bus library got substantial
975 updates to work in conjunction with the (also experimental)
976 kdbus kernel project. It works well enough to exchange
977 messages with some sophistication. Note that kdbus is not
978 ready yet, and the library is mostly an elaborate test case
979 for now, and not installable.
981 * systemd gained a new unit 'systemd-static-nodes.service'
982 that generates static device nodes earlier during boot, and
983 can run in conjunction with udev.
985 * libsystemd-login gained a new call sd_pid_get_user_unit()
986 to retrieve the user systemd unit a process is running
987 in. This is useful for systems where systemd is used as
990 * systemd-nspawn now places all containers in the new /machine
991 top-level cgroup directory in the name=systemd
992 hierarchy. libvirt will soon do the same, so that we get a
993 uniform separation of /system, /user and /machine for system
994 services, user processes and containers/virtual
995 machines. This new cgroup hierarchy is also useful to stick
996 stable names to specific container instances, which can be
997 recognized later this way (this name may be controlled
998 via systemd-nspawn's new -M switch). libsystemd-login also
999 gained a new call sd_pid_get_machine_name() to retrieve the
1000 name of the container/VM a specific process belongs to.
1002 * bootchart can now store its data in the journal.
1004 * libsystemd-journal gained a new call
1005 sd_journal_add_conjunction() for AND expressions to the
1006 matching logic. This can be used to express more complex
1007 logical expressions.
1009 * journactl can now take multiple --unit= and --user-unit=
1012 * The cryptsetup logic now understands the "luks.key=" kernel
1013 command line switch for specifying a file to read the
1014 decryption key from. Also, if a configured key file is not
1015 found the tool will now automatically fall back to prompting
1018 * Python systemd.journal module was updated to wrap recently
1019 added functions from libsystemd-journal. The interface was
1020 changed to bring the low level interface in s.j._Reader
1021 closer to the C API, and the high level interface in
1022 s.j.Reader was updated to wrap and convert all data about
1025 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
1026 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
1027 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
1028 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
1029 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
1030 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1034 * journalctl --update-catalog now understands a new --root=
1035 option to operate on catalogs found in a different root
1038 * During shutdown after systemd has terminated all running
1039 services a final killing loop kills all remaining left-over
1040 processes. We will now print the name of these processes
1041 when we send SIGKILL to them, since this usually indicates a
1044 * If /etc/crypttab refers to password files stored on
1045 configured mount points automatic dependencies will now be
1046 generated to ensure the specific mount is established first
1047 before the key file is attempted to be read.
1049 * 'systemctl status' will now show information about the
1050 network sockets a socket unit is listening on.
1052 * 'systemctl status' will also shown information about any
1053 drop-in configuration file for units. (Drop-In configuration
1054 files in this context are files such as
1055 /etc/systemd/systemd/foobar.service.d/*.conf)
1057 * systemd-cgtop now optionally shows summed up CPU times of
1058 cgroups. Press '%' while running cgtop to switch between
1059 percentage and absolute mode. This is useful to determine
1060 which cgroups use up the most CPU time over the entire
1061 runtime of the system. systemd-cgtop has also been updated
1062 to be 'pipeable' for processing with further shell tools.
1064 * 'hostnamectl set-hostname' will now allow setting of FQDN
1067 * The formatting and parsing of time span values has been
1068 changed. The parser now understands fractional expressions
1069 such as "5.5h". The formatter will now output fractional
1070 expressions for all time spans under 1min, i.e. "5.123456s"
1071 rather than "5s 123ms 456us". For time spans under 1s
1072 millisecond values are shown, for those under 1ms
1073 microsecond values are shown. This should greatly improve
1074 all time-related output of systemd.
1076 * libsystemd-login and libsystemd-journal gained new
1077 functions for querying the poll() events mask and poll()
1078 timeout value for integration into arbitrary event
1081 * localectl gained the ability to list available X11 keymaps
1082 (models, layouts, variants, options).
1084 * 'systemd-analyze dot' gained the ability to filter for
1085 specific units via shell-style globs, to create smaller,
1086 more useful graphs. I.e. it's now possible to create simple
1087 graphs of all the dependencies between only target units, or
1088 of all units that Avahi has dependencies with.
1090 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
1091 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
1092 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
1093 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
1094 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
1095 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
1096 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
1100 * The boot-time readahead implementation for rotating media
1101 will now read the read-ahead data in multiple passes which
1102 consist of all read requests made in equidistant time
1103 intervals. This means instead of strictly reading read-ahead
1104 data in its physical order on disk we now try to find a
1105 middle ground between physical and access time order.
1107 * /etc/os-release files gained a new BUILD_ID= field for usage
1108 on operating systems that provide continuous builds of OS
1111 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
1112 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
1113 William Douglas, Zbigniew Jędrzejewski-Szmek
1117 * systemd-python gained an API exposing libsystemd-daemon.
1119 * The SMACK setup logic gained support for uploading CIPSO
1122 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
1123 ReadOnlyDirectories= and InaccessibleDirectories= has
1124 changed. The private /tmp and /var/tmp directories are now
1125 shared by all processes of a service (which means
1126 ExecStartPre= may now leave data in /tmp that ExecStart= of
1127 the same service can still access). When a service is
1128 stopped its temporary directories are immediately deleted
1129 (normal clean-up with tmpfiles is still done in addition to
1132 * By default, systemd will now set a couple of sysctl
1133 variables in the kernel: the safe sysrq options are turned
1134 on, IP route verification is turned on, and source routing
1135 disabled. The recently added hardlink and softlink
1136 protection of the kernel is turned on. These settings should
1137 be reasonably safe, and good defaults for all new systems.
1139 * The predictable network naming logic may now be turned off
1140 with a new kernel command line switch: net.ifnames=0.
1142 * A new libsystemd-bus module has been added that implements a
1143 pretty complete D-Bus client library. For details see:
1145 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
1147 * journald will now explicitly flush the journal files to disk
1148 at the latest 5min after each write. The file will then also
1149 be marked offline until the next write. This should increase
1150 reliability in case of a crash. The synchronization delay
1151 can be configured via SyncIntervalSec= in journald.conf.
1153 * There's a new remote-fs-setup.target unit that can be used
1154 to pull in specific services when at least one remote file
1155 system is to be mounted.
1157 * There are new targets timers.target and paths.target as
1158 canonical targets to pull user timer and path units in
1159 from. This complements sockets.target with a similar
1160 purpose for socket units.
1162 * libudev gained a new call udev_device_set_attribute_value()
1163 to set sysfs attributes of a device.
1165 * The udev daemon now sets the default number of worker
1166 processes executed in parallel based on the number of available
1167 CPUs instead of the amount of available RAM. This is supposed
1168 to provide a more reliable default and limit a too aggressive
1169 paralellism for setups with 1000s of devices connected.
1171 Contributions from: Auke Kok, Colin Walters, Cristian
1172 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
1173 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
1174 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
1175 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
1176 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
1177 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
1178 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
1179 Zbigniew Jędrzejewski-Szmek
1183 * Configuration of unit files may now be extended via drop-in
1184 files without having to edit/override the unit files
1185 themselves. More specifically, if the administrator wants to
1186 change one value for a service file foobar.service he can
1187 now do so by dropping in a configuration snippet into
1188 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
1189 will load all these snippets and apply them on top of the
1190 main unit configuration file, possibly extending or
1191 overriding its settings. Using these drop-in snippets is
1192 generally nicer than the two earlier options for changing
1193 unit files locally: copying the files from
1194 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
1195 them there; or creating a new file in /etc/systemd/system/
1196 that incorporates the original one via ".include". Drop-in
1197 snippets into these .d/ directories can be placed in any
1198 directory systemd looks for units in, and the usual
1199 overriding semantics between /usr/lib, /etc and /run apply
1202 * Most unit file settings which take lists of items can now be
1203 reset by assigning the empty string to them. For example,
1204 normally, settings such as Environment=FOO=BAR append a new
1205 environment variable assignment to the environment block,
1206 each time they are used. By assigning Environment= the empty
1207 string the environment block can be reset to empty. This is
1208 particularly useful with the .d/*.conf drop-in snippets
1209 mentioned above, since this adds the ability to reset list
1210 settings from vendor unit files via these drop-ins.
1212 * systemctl gained a new "list-dependencies" command for
1213 listing the dependencies of a unit recursively.
1215 * Inhibitors are now honored and listed by "systemctl
1216 suspend", "systemctl poweroff" (and similar) too, not only
1217 GNOME. These commands will also list active sessions by
1220 * Resource limits (as exposed by the various control group
1221 controllers) can now be controlled dynamically at runtime
1222 for all units. More specifically, you can now use a command
1223 like "systemctl set-cgroup-attr foobar.service cpu.shares
1224 2000" to alter the CPU shares a specific service gets. These
1225 settings are stored persistently on disk, and thus allow the
1226 administrator to easily adjust the resource usage of
1227 services with a few simple commands. This dynamic resource
1228 management logic is also available to other programs via the
1229 bus. Almost any kernel cgroup attribute and controller is
1232 * systemd-vconsole-setup will now copy all font settings to
1233 all allocated VTs, where it previously applied them only to
1236 * libsystemd-login gained the new sd_session_get_tty() API
1239 * This release drops support for a few legacy or
1240 distribution-specific LSB facility names when parsing init
1241 scripts: $x-display-manager, $mail-transfer-agent,
1242 $mail-transport-agent, $mail-transfer-agent, $smtp,
1243 $null. Also, the mail-transfer-agent.target unit backing
1244 this has been removed. Distributions which want to retain
1245 compatibility with this should carry the burden for
1246 supporting this themselves and patch support for these back
1247 in, if they really need to. Also, the facilities $syslog and
1248 $local_fs are now ignored, since systemd does not support
1249 early-boot LSB init scripts anymore, and these facilities
1250 are implied anyway for normal services. syslog.target has
1253 * There are new bus calls on PID1's Manager object for
1254 cancelling jobs, and removing snapshot units. Previously,
1255 both calls were only available on the Job and Snapshot
1258 * systemd-journal-gatewayd gained SSL support.
1260 * The various "environment" files, such as /etc/locale.conf
1261 now support continuation lines with a backslash ("\") as
1262 last character in the line, similar in style (but different)
1263 to how this is supported in shells.
1265 * For normal user processes the _SYSTEMD_USER_UNIT= field is
1266 now implicitly appended to every log entry logged. systemctl
1267 has been updated to filter by this field when operating on a
1268 user systemd instance.
1270 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
1271 CAP_AUDIT_CONTROL capabilities to the capabilities set for
1272 the container. This makes it easier to boot unmodified
1273 Fedora systems in a container, which however still requires
1274 audit=0 to be passed on the kernel command line. Auditing in
1275 kernel and userspace is unfortunately still too broken in
1276 context of containers, hence we recommend compiling it out
1277 of the kernel or using audit=0. Hopefully this will be fixed
1278 one day for good in the kernel.
1280 * nspawn gained the new --bind= and --bind-ro= parameters to
1281 bind mount specific directories from the host into the
1284 * nspawn will now mount its own devpts file system instance
1285 into the container, in order not to leak pty devices from
1286 the host into the container.
1288 * systemd will now read the firmware boot time performance
1289 information from the EFI variables, if the used boot loader
1290 supports this, and takes it into account for boot performance
1291 analysis via "systemd-analyze". This is currently supported
1292 only in conjunction with Gummiboot, but could be supported
1293 by other boot loaders too. For details see:
1295 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
1297 * A new generator has been added that automatically mounts the
1298 EFI System Partition (ESP) to /boot, if that directory
1299 exists, is empty, and no other file system has been
1300 configured to be mounted there.
1302 * logind will now send out PrepareForSleep(false) out
1303 unconditionally, after coming back from suspend. This may be
1304 used by applications as asynchronous notification for
1305 system resume events.
1307 * "systemctl unlock-sessions" has been added, that allows
1308 unlocking the screens of all user sessions at once, similar
1309 how "systemctl lock-sessions" already locked all users
1310 sessions. This is backed by a new D-Bus call UnlockSessions().
1312 * "loginctl seat-status" will now show the master device of a
1313 seat. (i.e. the device of a seat that needs to be around for
1314 the seat to be considered available, usually the graphics
1317 * tmpfiles gained a new "X" line type, that allows
1318 configuration of files and directories (with wildcards) that
1319 shall be excluded from automatic cleanup ("aging").
1321 * udev default rules set the device node permissions now only
1322 at "add" events, and do not change them any longer with a
1323 later "change" event.
1325 * The log messages for lid events and power/sleep keypresses
1326 now carry a message ID.
1328 * We now have a substantially larger unit test suite, but this
1329 continues to be work in progress.
1331 * udevadm hwdb gained a new --root= parameter to change the
1332 root directory to operate relative to.
1334 * logind will now issue a background sync() request to the kernel
1335 early at shutdown, so that dirty buffers are flushed to disk early
1336 instead of at the last moment, in order to optimize shutdown
1339 * A new bootctl tool has been added that is an interface for
1340 certain boot loader operations. This is currently a preview
1341 and is likely to be extended into a small mechanism daemon
1342 like timedated, localed, hostnamed, and can be used by
1343 graphical UIs to enumerate available boot options, and
1344 request boot into firmware operations.
1346 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
1347 the rest of the package. It also has been updated to work
1348 correctly in initrds.
1350 * Policykit previously has been runtime optional, and is now
1351 also compile time optional via a configure switch.
1353 * systemd-analyze has been reimplemented in C. Also "systemctl
1354 dot" has moved into systemd-analyze.
1356 * "systemctl status" with no further parameters will now print
1357 the status of all active or failed units.
1359 * Operations such as "systemctl start" can now be executed
1360 with a new mode "--irreversible" which may be used to queue
1361 operations that cannot accidentally be reversed by a later
1362 job queuing. This is by default used to make shutdown
1363 requests more robust.
1365 * The Python API of systemd now gained a new module for
1366 reading journal files.
1368 * A new tool kernel-install has been added that can install
1369 kernel images according to the Boot Loader Specification:
1371 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
1373 * Boot time console output has been improved to provide
1374 animated boot time output for hanging jobs.
1376 * A new tool systemd-activate has been added which can be used
1377 to test socket activation with, directly from the command
1378 line. This should make it much easier to test and debug
1379 socket activation in daemons.
1381 * journalctl gained a new "--reverse" (or -r) option to show
1382 journal output in reverse order (i.e. newest line first).
1384 * journalctl gained a new "--pager-end" (or -e) option to jump
1385 to immediately jump to the end of the journal in the
1386 pager. This is only supported in conjunction with "less".
1388 * journalctl gained a new "--user-unit=" option, that works
1389 similar to "--unit=" but filters for user units rather than
1392 * A number of unit files to ease adoption of systemd in
1393 initrds has been added. This moves some minimal logic from
1394 the various initrd implementations into systemd proper.
1396 * The journal files are now owned by a new group
1397 "systemd-journal", which exists specifically to allow access
1398 to the journal, and nothing else. Previously, we used the
1399 "adm" group for that, which however possibly covers more
1400 than just journal/log file access. This new group is now
1401 already used by systemd-journal-gatewayd to ensure this
1402 daemon gets access to the journal files and as little else
1403 as possible. Note that "make install" will also set FS ACLs
1404 up for /var/log/journal to give "adm" and "wheel" read
1405 access to it, in addition to "systemd-journal" which owns
1406 the journal files. We recommend that packaging scripts also
1407 add read access to "adm" + "wheel" to /var/log/journal, and
1408 all existing/future journal files. To normal users and
1409 administrators little changes, however packagers need to
1410 ensure to create the "systemd-journal" system group at
1411 package installation time.
1413 * The systemd-journal-gatewayd now runs as unprivileged user
1414 systemd-journal-gateway:systemd-journal-gateway. Packaging
1415 scripts need to create these system user/group at
1418 * timedated now exposes a new boolean property CanNTP that
1419 indicates whether a local NTP service is available or not.
1421 * systemd-detect-virt will now also detect xen PVs
1423 * The pstore file system is now mounted by default, if it is
1426 * In addition to the SELinux and IMA policies we will now also
1427 load SMACK policies at early boot.
1429 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
1430 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
1431 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
1432 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
1433 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
1434 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
1435 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
1436 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
1437 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
1438 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
1439 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
1440 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
1441 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
1442 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
1446 * Timer units now support calendar time events in addition to
1447 monotonic time events. That means you can now trigger a unit
1448 based on a calendar time specification such as "Thu,Fri
1449 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
1450 or fifth day of any month of the year 2013, given that it is
1451 a thursday or friday. This brings timer event support
1452 considerably closer to cron's capabilities. For details on
1453 the supported calendar time specification language see
1456 * udev now supports a number of different naming policies for
1457 network interfaces for predictable names, and a combination
1458 of these policies is now the default. Please see this wiki
1459 document for details:
1461 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
1463 * Auke Kok's bootchart implementation has been added to the
1464 systemd tree. It's an optional component that can graph the
1465 boot in quite some detail. It's one of the best bootchart
1466 implementations around and minimal in its code and
1469 * nss-myhostname has been integrated into the systemd source
1470 tree. nss-myhostname guarantees that the local hostname
1471 always stays resolvable via NSS. It has been a weak
1472 requirement of systemd-hostnamed since a long time, and
1473 since its code is actually trivial we decided to just
1474 include it in systemd's source tree. It can be turned off
1475 with a configure switch.
1477 * The read-ahead logic is now capable of properly detecting
1478 whether a btrfs file system is on SSD or rotating media, in
1479 order to optimize the read-ahead scheme. Previously, it was
1480 only capable of detecting this on traditional file systems
1483 * In udev, additional device properties are now read from the
1484 IAB in addition to the OUI database. Also, Bluetooth company
1485 identities are attached to the devices as well.
1487 * In service files %U may be used as specifier that is
1488 replaced by the configured user name of the service.
1490 * nspawn may now be invoked without a controlling TTY. This
1491 makes it suitable for invocation as its own service. This
1492 may be used to set up a simple containerized server system
1493 using only core OS tools.
1495 * systemd and nspawn can now accept socket file descriptors
1496 when they are started for socket activation. This enables
1497 implementation of socket activated nspawn
1498 containers. i.e. think about autospawning an entire OS image
1499 when the first SSH or HTTP connection is received. We expect
1500 that similar functionality will also be added to libvirt-lxc
1503 * journalctl will now suppress ANSI color codes when
1504 presenting log data.
1506 * systemctl will no longer show control group information for
1507 a unit if a the control group is empty anyway.
1509 * logind can now automatically suspend/hibernate/shutdown the
1512 * /etc/machine-info and hostnamed now also expose the chassis
1513 type of the system. This can be used to determine whether
1514 the local system is a laptop, desktop, handset or
1515 tablet. This information may either be configured by the
1516 user/vendor or is automatically determined from ACPI and DMI
1517 information if possible.
1519 * A number of PolicyKit actions are now bound together with
1520 "imply" rules. This should simplify creating UIs because
1521 many actions will now authenticate similar ones as well.
1523 * Unit files learnt a new condition ConditionACPower= which
1524 may be used to conditionalize a unit depending on whether an
1525 AC power source is connected or not, of whether the system
1526 is running on battery power.
1528 * systemctl gained a new "is-failed" verb that may be used in
1529 shell scripts and suchlike to check whether a specific unit
1530 is in the "failed" state.
1532 * The EnvironmentFile= setting in unit files now supports file
1533 globbing, and can hence be used to easily read a number of
1534 environment files at once.
1536 * systemd will no longer detect and recognize specific
1537 distributions. All distribution-specific #ifdeffery has been
1538 removed, systemd is now fully generic and
1539 distribution-agnostic. Effectively, not too much is lost as
1540 a lot of the code is still accessible via explicit configure
1541 switches. However, support for some distribution specific
1542 legacy configuration file formats has been dropped. We
1543 recommend distributions to simply adopt the configuration
1544 files everybody else uses now and convert the old
1545 configuration from packaging scripts. Most distributions
1546 already did that. If that's not possible or desirable,
1547 distributions are welcome to forward port the specific
1548 pieces of code locally from the git history.
1550 * When logging a message about a unit systemd will now always
1551 log the unit name in the message meta data.
1553 * localectl will now also discover system locale data that is
1554 not stored in locale archives, but directly unpacked.
1556 * logind will no longer unconditionally use framebuffer
1557 devices as seat masters, i.e. as devices that are required
1558 to be existing before a seat is considered preset. Instead,
1559 it will now look for all devices that are tagged as
1560 "seat-master" in udev. By default framebuffer devices will
1561 be marked as such, but depending on local systems other
1562 devices might be marked as well. This may be used to
1563 integrate graphics cards using closed source drivers (such
1564 as NVidia ones) more nicely into logind. Note however, that
1565 we recommend using the open source NVidia drivers instead,
1566 and no udev rules for the closed-source drivers will be
1567 shipped from us upstream.
1569 Contributions from: Adam Williamson, Alessandro Crismani, Auke
1570 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
1571 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
1572 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
1573 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
1574 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
1575 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
1576 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
1577 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
1578 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
1579 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
1580 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
1585 * udev gained support for loading additional device properties
1586 from an indexed database that is keyed by vendor/product IDs
1587 and similar device identifiers. For the beginning this
1588 "hwdb" is populated with data from the well-known PCI and
1589 USB database, but also includes PNP, ACPI and OID data. In
1590 the longer run this indexed database shall grow into
1591 becoming the one central database for non-essential
1592 userspace device metadata. Previously, data from the PCI/USB
1593 database was only attached to select devices, since the
1594 lookup was a relatively expensive operation due to O(n) time
1595 complexity (with n being the number of entries in the
1596 database). Since this is now O(1), we decided to add in this
1597 data for all devices where this is available, by
1598 default. Note that the indexed database needs to be rebuilt
1599 when new data files are installed. To achieve this you need
1600 to update your packaging scripts to invoke "udevadm hwdb
1601 --update" after installation of hwdb data files. For
1602 RPM-based distributions we introduced the new
1603 %udev_hwdb_update macro for this purpose.
1605 * The Journal gained support for the "Message Catalog", an
1606 indexed database to link up additional information with
1607 journal entries. For further details please check:
1609 http://www.freedesktop.org/wiki/Software/systemd/catalog
1611 The indexed message catalog database also needs to be
1612 rebuilt after installation of message catalog files. Use
1613 "journalctl --update-catalog" for this. For RPM-based
1614 distributions we introduced the %journal_catalog_update
1615 macro for this purpose.
1617 * The Python Journal bindings gained support for the standard
1618 Python logging framework.
1620 * The Journal API gained new functions for checking whether
1621 the underlying file system of a journal file is capable of
1622 properly reporting file change notifications, or whether
1623 applications that want to reflect journal changes "live"
1624 need to recheck journal files continuously in appropriate
1627 * It is now possible to set the "age" field for tmpfiles
1628 entries to 0, indicating that files matching this entry
1629 shall always be removed when the directories are cleaned up.
1631 * coredumpctl gained a new "gdb" verb which invokes gdb
1632 right-away on the selected coredump.
1634 * There's now support for "hybrid sleep" on kernels that
1635 support this, in addition to "suspend" and "hibernate". Use
1636 "systemctl hybrid-sleep" to make use of this.
1638 * logind's HandleSuspendKey= setting (and related settings)
1639 now gained support for a new "lock" setting to simply
1640 request the screen lock on all local sessions, instead of
1641 actually executing a suspend or hibernation.
1643 * systemd will now mount the EFI variables file system by
1646 * Socket units now gained support for configuration of the
1647 SMACK security label.
1649 * timedatectl will now output the time of the last and next
1650 daylight saving change.
1652 * We dropped support for various legacy and distro-specific
1653 concepts, such as insserv, early-boot SysV services
1654 (i.e. those for non-standard runlevels such as 'b' or 'S')
1655 or ArchLinux /etc/rc.conf support. We recommend the
1656 distributions who still need support this to either continue
1657 to maintain the necessary patches downstream, or find a
1658 different solution. (Talk to us if you have questions!)
1660 * Various systemd components will now bypass PolicyKit checks
1661 for root and otherwise handle properly if PolicyKit is not
1662 found to be around. This should fix most issues for
1663 PolicyKit-less systems. Quite frankly this should have been
1664 this way since day one. It is absolutely our intention to
1665 make systemd work fine on PolicyKit-less systems, and we
1666 consider it a bug if something doesn't work as it should if
1667 PolicyKit is not around.
1669 * For embedded systems it is now possible to build udev and
1670 systemd without blkid and/or kmod support.
1672 * "systemctl switch-root" is now capable of switching root
1673 more than once. I.e. in addition to transitions from the
1674 initrd to the host OS it is now possible to transition to
1675 further OS images from the host. This is useful to implement
1676 offline updating tools.
1678 * Various other additions have been made to the RPM macros
1679 shipped with systemd. Use %udev_rules_update() after
1680 installing new udev rules files. %_udevhwdbdir,
1681 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
1682 %_sysctldir are now available which resolve to the right
1683 directories for packages to place various data files in.
1685 * journalctl gained the new --full switch (in addition to
1686 --all, to disable ellipsation for long messages.
1688 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
1689 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
1690 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
1691 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
1692 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
1693 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
1694 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
1695 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
1696 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
1700 * journalctl gained new --since= and --until= switches to
1701 filter by time. It also now supports nice filtering for
1702 units via --unit=/-u.
1704 * Type=oneshot services may use ExecReload= and do the
1707 * The journal daemon now supports time-based rotation and
1708 vacuuming, in addition to the usual disk-space based
1711 * The journal will now index the available field values for
1712 each field name. This enables clients to show pretty drop
1713 downs of available match values when filtering. The bash
1714 completion of journalctl has been updated
1715 accordingly. journalctl gained a new switch -F to list all
1716 values a certain field takes in the journal database.
1718 * More service events are now written as structured messages
1719 to the journal, and made recognizable via message IDs.
1721 * The timedated, localed and hostnamed mini-services which
1722 previously only provided support for changing time, locale
1723 and hostname settings from graphical DEs such as GNOME now
1724 also have a minimal (but very useful) text-based client
1725 utility each. This is probably the nicest way to changing
1726 these settings from the command line now, especially since
1727 it lists available options and is fully integrated with bash
1730 * There's now a new tool "systemd-coredumpctl" to list and
1731 extract coredumps from the journal.
1733 * We now install a README each in /var/log/ and
1734 /etc/rc.d/init.d explaining where the system logs and init
1735 scripts went. This hopefully should help folks who go to
1736 that dirs and look into the otherwise now empty void and
1737 scratch their heads.
1739 * When user-services are invoked (by systemd --user) the
1740 $MANAGERPID env var is set to the PID of systemd.
1742 * SIGRTMIN+24 when sent to a --user instance will now result
1743 in immediate termination of systemd.
1745 * gatewayd received numerous feature additions such as a
1746 "follow" mode, for live syncing and filtering.
1748 * browse.html now allows filtering and showing detailed
1749 information on specific entries. Keyboard navigation and
1750 mouse screen support has been added.
1752 * gatewayd/journalctl now supports HTML5/JSON
1753 Server-Sent-Events as output.
1755 * The SysV init script compatibility logic will now
1756 heuristically determine whether a script supports the
1757 "reload" verb, and only then make this available as
1760 * "systemctl status --follow" has been removed, use "journalctl
1763 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
1764 have been removed since they are hardly useful to be
1767 * And I'd like to take the opportunity to specifically mention
1768 Zbigniew for his great contributions. Zbigniew, you rock!
1770 Contributions from: Andrew Eikum, Christian Hesse, Colin
1771 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
1772 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
1773 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
1774 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
1775 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
1776 Jędrzejewski-Szmek, Сковорода Никита Андреевич
1780 * If /etc/vconsole.conf is non-existent or empty we will no
1781 longer load any console font or key map at boot by
1782 default. Instead the kernel defaults will be left
1783 intact. This is definitely the right thing to do, as no
1784 configuration should mean no configuration, and hard-coding
1785 font names that are different on all archs is probably a bad
1786 idea. Also, the kernel default key map and font should be
1787 good enough for most cases anyway, and mostly identical to
1788 the userspace fonts/key maps we previously overloaded them
1789 with. If distributions want to continue to default to a
1790 non-kernel font or key map they should ship a default
1791 /etc/vconsole.conf with the appropriate contents.
1793 Contributions from: Colin Walters, Daniel J Walsh, Dave
1794 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
1795 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1799 * journalctl gained a new --cursor= switch to show entries
1800 starting from the specified location in the journal.
1802 * We now enforce a size limit on journal entry fields exported
1803 with "-o json" in journalctl. Fields larger than 4K will be
1804 assigned null. This can be turned off with --all.
1806 * An (optional) journal gateway daemon is now available as
1807 "systemd-journal-gatewayd.service". This service provides
1808 access to the journal via HTTP and JSON. This functionality
1809 will be used to implement live log synchronization in both
1810 pull and push modes, but has various other users too, such
1811 as easy log access for debugging of embedded devices. Right
1812 now it is already useful to retrieve the journal via HTTP:
1814 # systemctl start systemd-journal-gatewayd.service
1815 # wget http://localhost:19531/entries
1817 This will download the journal contents in a
1818 /var/log/messages compatible format. The same as JSON:
1820 # curl -H"Accept: application/json" http://localhost:19531/entries
1822 This service is also accessible via a web browser where a
1823 single static HTML5 app is served that uses the JSON logic
1824 to enable the user to do some basic browsing of the
1825 journal. This will be extended later on. Here's an example
1826 screenshot of this app in its current state:
1828 http://0pointer.de/public/journal-gatewayd
1830 Contributions from: Kay Sievers, Lennart Poettering, Robert
1831 Milasan, Tom Gundersen
1835 * The bash completion logic is now available for journalctl
1838 * We don't mount the "cpuset" controller anymore together with
1839 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
1840 started if no parameters are assigned to it. "cpuset" hence
1841 broke code that assumed it it could create "cpu" groups and
1844 * journalctl -f will now subscribe to terminal size changes,
1845 and line break accordingly.
1847 Contributions from: Dave Reisner, Kay Sievers, Lennart
1848 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
1852 * nspawn will now create a symlink /etc/localtime in the
1853 container environment, copying the host's timezone
1854 setting. Previously this has been done via a bind mount, but
1855 since symlinks cannot be bind mounted this has now been
1856 changed to create/update the appropriate symlink.
1858 * journalctl -n's line number argument is now optional, and
1859 will default to 10 if omitted.
1861 * journald will now log the maximum size the journal files may
1862 take up on disk. This is particularly useful if the default
1863 built-in logic of determining this parameter from the file
1864 system size is used. Use "systemctl status
1865 systemd-journald.service" to see this information.
1867 * The multi-seat X wrapper tool has been stripped down. As X
1868 is now capable of enumerating graphics devices via udev in a
1869 seat-aware way the wrapper is not strictly necessary
1870 anymore. A stripped down temporary stop-gap is still shipped
1871 until the upstream display managers have been updated to
1872 fully support the new X logic. Expect this wrapper to be
1873 removed entirely in one of the next releases.
1875 * HandleSleepKey= in logind.conf has been split up into
1876 HandleSuspendKey= and HandleHibernateKey=. The old setting
1877 is not available anymore. X11 and the kernel are
1878 distuingishing between these keys and we should too. This
1879 also means the inhibition lock for these keys has been split
1882 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
1883 Poettering, Lukas Nykryn, Václav Pavlín
1887 * Whenever a unit changes state we'll now log this to the
1888 journal and show along the unit's own log output in
1891 * ConditionPathIsMountPoint= can now properly detect bind
1892 mount points too. (Previously, a bind mount of one file
1893 system to another place in the same file system could not be
1894 detected as mount, since they shared struct stat's st_dev
1897 * We will now mount the cgroup controllers cpu, cpuacct,
1898 cpuset and the controllers net_cls, net_prio together by
1901 * nspawn containers will now have a virtualized boot
1902 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
1903 over with a randomized ID at container initialization). This
1904 has the effect of making "journalctl -b" do the right thing
1907 * The JSON output journal serialization has been updated not
1908 to generate "endless" list objects anymore, but rather one
1909 JSON object per line. This is more in line how most JSON
1910 parsers expect JSON objects. The new output mode
1911 "json-pretty" has been added to provide similar output, but
1912 neatly aligned for readability by humans.
1914 * We dropped all explicit sync() invocations in the shutdown
1915 code. The kernel does this implicitly anyway in the kernel
1916 reboot() syscall. halt(8)'s -n option is now a compatibility
1919 * We now support virtualized reboot() in containers, as
1920 supported by newer kernels. We will fall back to exit() if
1921 CAP_SYS_REBOOT is not available to the container. Also,
1922 nspawn makes use of this now and will actually reboot the
1923 container if the containerized OS asks for that.
1925 * journalctl will only show local log output by default
1926 now. Use --merge (-m) to show remote log output, too.
1928 * libsystemd-journal gained the new sd_journal_get_usage()
1929 call to determine the current disk usage of all journal
1930 files. This is exposed in the new "journalctl --disk-usage"
1933 * journald gained a new configuration setting SplitMode= in
1934 journald.conf which may be used to control how user journals
1935 are split off. See journald.conf(5) for details.
1937 * A new condition type ConditionFileNotEmpty= has been added.
1939 * tmpfiles' "w" lines now support file globbing, to write
1940 multiple files at once.
1942 * We added Python bindings for the journal submission
1943 APIs. More Python APIs for a number of selected APIs will
1944 likely follow. Note that we intend to add native bindings
1945 only for the Python language, as we consider it common
1946 enough to deserve bindings shipped within systemd. There are
1947 various projects outside of systemd that provide bindings
1948 for languages such as PHP or Lua.
1950 * Many conditions will now resolve specifiers such as %i. In
1951 addition, PathChanged= and related directives of .path units
1952 now support specifiers as well.
1954 * There's now a new RPM macro definition for the system preset
1957 * journald will now warn if it can't forward a message to the
1958 syslog daemon because it's socket is full.
1960 * timedated will no longer write or process /etc/timezone,
1961 except on Debian. As we do not support late mounted /usr
1962 anymore /etc/localtime always being a symlink is now safe,
1963 and hence the information in /etc/timezone is not necessary
1966 * logind will now always reserve one VT for a text getty (VT6
1967 by default). Previously if more than 6 X sessions where
1968 started they took up all the VTs with auto-spawned gettys,
1969 so that no text gettys were available anymore.
1971 * udev will now automatically inform the btrfs kernel logic
1972 about btrfs RAID components showing up. This should make
1973 simple hotplug based btrfs RAID assembly work.
1975 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
1976 (but not for its children which will stay at the kernel
1977 default). This should allow setups with a lot more listening
1980 * systemd will now always pass the configured timezone to the
1981 kernel at boot. timedated will do the same when the timezone
1984 * logind's inhibition logic has been updated. By default,
1985 logind will now handle the lid switch, the power and sleep
1986 keys all the time, even in graphical sessions. If DEs want
1987 to handle these events on their own they should take the new
1988 handle-power-key, handle-sleep-key and handle-lid-switch
1989 inhibitors during their runtime. A simple way to achiveve
1990 that is to invoke the DE wrapped in an invocation of:
1992 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
1994 * Access to unit operations is now checked via SELinux taking
1995 the unit file label and client process label into account.
1997 * systemd will now notify the administrator in the journal
1998 when he over-mounts a non-empty directory.
2000 * There are new specifiers that are resolved in unit files,
2001 for the host name (%H), the machine ID (%m) and the boot ID
2004 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
2005 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
2006 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
2007 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
2008 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
2009 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
2010 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
2014 * Support for reading structured kernel messages from
2015 /dev/kmsg has now been added and is enabled by default.
2017 * Support for reading kernel messages from /proc/kmsg has now
2018 been removed. If you want kernel messages in the journal
2019 make sure to run a recent kernel (>= 3.5) that supports
2020 reading structured messages from /dev/kmsg (see
2021 above). /proc/kmsg is now exclusive property of classic
2022 syslog daemons again.
2024 * The libudev API gained the new
2025 udev_device_new_from_device_id() call.
2027 * The logic for file system namespace (ReadOnlyDirectory=,
2028 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
2029 require pivot_root() anymore. This means fewer temporary
2030 directories are created below /tmp for this feature.
2032 * nspawn containers will now see and receive all submounts
2033 made on the host OS below the root file system of the
2036 * Forward Secure Sealing is now supported for Journal files,
2037 which provide cryptographical sealing of journal files so
2038 that attackers cannot alter log history anymore without this
2039 being detectable. Lennart will soon post a blog story about
2040 this explaining it in more detail.
2042 * There are two new service settings RestartPreventExitStatus=
2043 and SuccessExitStatus= which allow configuration of exit
2044 status (exit code or signal) which will be excepted from the
2045 restart logic, resp. consider successful.
2047 * journalctl gained the new --verify switch that can be used
2048 to check the integrity of the structure of journal files and
2049 (if Forward Secure Sealing is enabled) the contents of
2052 * nspawn containers will now be run with /dev/stdin, /dev/fd/
2053 and similar symlinks pre-created. This makes running shells
2054 as container init process a lot more fun.
2056 * The fstab support can now handle PARTUUID= and PARTLABEL=
2059 * A new ConditionHost= condition has been added to match
2060 against the hostname (with globs) and machine ID. This is
2061 useful for clusters where a single OS image is used to
2062 provision a large number of hosts which shall run slightly
2063 different sets of services.
2065 * Services which hit the restart limit will now be placed in a
2068 Contributions from: Bertram Poettering, Dave Reisner, Huang
2069 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
2070 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
2074 * When running in --user mode systemd will now become a
2075 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
2076 tree a lot more organized.
2078 * A new PartOf= unit dependency type has been introduced that
2079 may be used to group services in a natural way.
2081 * "systemctl enable" may now be used to enable instances of
2084 * journalctl now prints error log levels in red, and
2085 warning/notice log levels in bright white. It also supports
2086 filtering by log level now.
2088 * cgtop gained a new -n switch (similar to top), to configure
2089 the maximum number of iterations to run for. It also gained
2090 -b, to run in batch mode (accepting no input).
2092 * The suffix ".service" may now be omitted on most systemctl
2093 command lines involving service unit names.
2095 * There's a new bus call in logind to lock all sessions, as
2096 well as a loginctl verb for it "lock-sessions".
2098 * libsystemd-logind.so gained a new call sd_journal_perror()
2099 that works similar to libc perror() but logs to the journal
2100 and encodes structured information about the error number.
2102 * /etc/crypttab entries now understand the new keyfile-size=
2105 * shutdown(8) now can send a (configurable) wall message when
2106 a shutdown is cancelled.
2108 * The mount propagation mode for the root file system will now
2109 default to "shared", which is useful to make containers work
2110 nicely out-of-the-box so that they receive new mounts from
2111 the host. This can be undone locally by running "mount
2112 --make-rprivate /" if needed.
2114 * The prefdm.service file has been removed. Distributions
2115 should maintain this unit downstream if they intend to keep
2116 it around. However, we recommend writing normal unit files
2117 for display managers instead.
2119 * Since systemd is a crucial part of the OS we will now
2120 default to a number of compiler switches that improve
2121 security (hardening) such as read-only relocations, stack
2122 protection, and suchlike.
2124 * The TimeoutSec= setting for services is now split into
2125 TimeoutStartSec= and TimeoutStopSec= to allow configuration
2126 of individual time outs for the start and the stop phase of
2129 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
2130 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
2131 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
2132 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
2133 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
2134 Gundersen, Zbigniew Jędrzejewski-Szmek
2138 * The journal and id128 C APIs are now fully documented as man
2141 * Extra safety checks have been added when transitioning from
2142 the initial RAM disk to the main system to avoid accidental
2145 * /etc/crypttab entries now understand the new keyfile-offset=
2148 * systemctl -t can now be used to filter by unit load state.
2150 * The journal C API gained the new sd_journal_wait() call to
2151 make writing synchronous journal clients easier.
2153 * journalctl gained the new -D switch to show journals from a
2156 * journalctl now displays a special marker between log
2157 messages of two different boots.
2159 * The journal is now explicitly flushed to /var via a service
2160 systemd-journal-flush.service, rather than implicitly simply
2161 by seeing /var/log/journal to be writable.
2163 * journalctl (and the journal C APIs) can now match for much
2164 more complex expressions, with alternatives and
2167 * When transitioning from the initial RAM disk to the main
2168 system we will now kill all processes in a killing spree to
2169 ensure no processes stay around by accident.
2171 * Three new specifiers may be used in unit files: %u, %h, %s
2172 resolve to the user name, user home directory resp. user
2173 shell. This is useful for running systemd user instances.
2175 * We now automatically rotate journal files if their data
2176 object hash table gets a fill level > 75%. We also size the
2177 hash table based on the configured maximum file size. This
2178 together should lower hash collisions drastically and thus
2179 speed things up a bit.
2181 * journalctl gained the new "--header" switch to introspect
2182 header data of journal files.
2184 * A new setting SystemCallFilters= has been added to services
2185 which may be used to apply blacklists or whitelists to
2186 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
2188 * nspawn gained a new --link-journal= switch (and quicker: -j)
2189 to link the container journal with the host. This makes it
2190 very easy to centralize log viewing on the host for all
2191 guests while still keeping the journal files separated.
2193 * Many bugfixes and optimizations
2195 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
2196 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
2197 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
2202 * Several tools now understand kernel command line arguments,
2203 which are only read when run in an initial RAM disk. They
2204 usually follow closely their normal counterparts, but are
2207 * There's a new tool to analyze the readahead files that are
2208 automatically generated at boot. Use:
2210 /usr/lib/systemd/systemd-readahead analyze /.readahead
2212 * We now provide an early debug shell on tty9 if this enabled. Use:
2214 systemctl enable debug-shell.service
2216 * All plymouth related units have been moved into the Plymouth
2217 package. Please make sure to upgrade your Plymouth version
2220 * systemd-tmpfiles now supports getting passed the basename of
2221 a configuration file only, in which case it will look for it
2222 in all appropriate directories automatically.
2224 * udevadm info now takes a /dev or /sys path as argument, and
2225 does the right thing. Example:
2227 udevadm info /dev/sda
2228 udevadm info /sys/class/block/sda
2230 * systemctl now prints a warning if a unit is stopped but a
2231 unit that might trigger it continues to run. Example: a
2232 service is stopped but the socket that activates it is left
2235 * "systemctl status" will now mention if the log output was
2236 shortened due to rotation since a service has been started.
2238 * The journal API now exposes functions to determine the
2239 "cutoff" times due to rotation.
2241 * journald now understands SIGUSR1 and SIGUSR2 for triggering
2242 immediately flushing of runtime logs to /var if possible,
2243 resp. for triggering immediate rotation of the journal
2246 * It is now considered an error if a service is attempted to
2247 be stopped that is not loaded.
2249 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
2251 * systemd-analyze now supports Python 3
2253 * tmpfiles now supports cleaning up directories via aging
2254 where the first level dirs are always kept around but
2255 directories beneath it automatically aged. This is enabled
2256 by prefixing the age field with '~'.
2258 * Seat objects now expose CanGraphical, CanTTY properties
2259 which is required to deal with very fast bootups where the
2260 display manager might be running before the graphics drivers
2261 completed initialization.
2263 * Seat objects now expose a State property.
2265 * We now include RPM macros for service enabling/disabling
2266 based on the preset logic. We recommend RPM based
2267 distributions to make use of these macros if possible. This
2268 makes it simpler to reuse RPM spec files across
2271 * We now make sure that the collected systemd unit name is
2272 always valid when services log to the journal via
2275 * There's a new man page kernel-command-line(7) detailing all
2276 command line options we understand.
2278 * The fstab generator may now be disabled at boot by passing
2279 fstab=0 on the kernel command line.
2281 * A new kernel command line option modules-load= is now understood
2282 to load a specific kernel module statically, early at boot.
2284 * Unit names specified on the systemctl command line are now
2285 automatically escaped as needed. Also, if file system or
2286 device paths are specified they are automatically turned
2287 into the appropriate mount or device unit names. Example:
2289 systemctl status /home
2290 systemctl status /dev/sda
2292 * The SysVConsole= configuration option has been removed from
2293 system.conf parsing.
2295 * The SysV search path is no longer exported on the D-Bus
2298 * The Names= option is been removed from unit file parsing.
2300 * There's a new man page bootup(7) detailing the boot process.
2302 * Every unit and every generator we ship with systemd now
2303 comes with full documentation. The self-explanatory boot is
2306 * A couple of services gained "systemd-" prefixes in their
2307 name if they wrap systemd code, rather than only external
2308 code. Among them fsck@.service which is now
2309 systemd-fsck@.service.
2311 * The HaveWatchdog property has been removed from the D-Bus
2314 * systemd.confirm_spawn= on the kernel command line should now
2317 * There's a new man page crypttab(5) which details all options
2318 we actually understand.
2320 * systemd-nspawn gained a new --capability= switch to pass
2321 additional capabilities to the container.
2323 * timedated will now read known NTP implementation unit names
2324 from /usr/lib/systemd/ntp-units.d/*.list,
2325 systemd-timedated-ntp.target has been removed.
2327 * journalctl gained a new switch "-b" that lists log data of
2328 the current boot only.
2330 * The notify socket is in the abstract namespace again, in
2331 order to support daemons which chroot() at start-up.
2333 * There is a new Storage= configuration option for journald
2334 which allows configuration of where log data should go. This
2335 also provides a way to disable journal logging entirely, so
2336 that data collected is only forwarded to the console, the
2337 kernel log buffer or another syslog implementation.
2339 * Many bugfixes and optimizations
2341 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
2342 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
2343 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
2344 Shawn Landden, Tom Gundersen
2348 * "systemctl help <unit>" now shows the man page if one is
2351 * Several new man pages have been added.
2353 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
2354 MaxLevelConsole= can now be specified in
2355 journald.conf. These options allow reducing the amount of
2356 data stored on disk or forwarded by the log level.
2358 * TimerSlackNSec= can now be specified in system.conf for
2359 PID1. This allows system-wide power savings.
2361 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
2362 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
2367 * logind is now capable of (optionally) handling power and
2368 sleep keys as well as the lid switch.
2370 * journalctl now understands the syntax "journalctl
2371 /usr/bin/avahi-daemon" to get all log output of a specific
2374 * CapabilityBoundingSet= in system.conf now also influences
2375 the capability bound set of usermode helpers of the kernel.
2377 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
2378 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
2379 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
2380 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
2384 * Note that we skipped 139 releases here in order to set the
2385 new version to something that is greater than both udev's
2386 and systemd's most recent version number.
2388 * udev: all udev sources are merged into the systemd source tree now.
2389 All future udev development will happen in the systemd tree. It
2390 is still fully supported to use the udev daemon and tools without
2391 systemd running, like in initramfs or other init systems. Building
2392 udev though, will require the *build* of the systemd tree, but
2393 udev can be properly *run* without systemd.
2395 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
2396 should be used to create dead device nodes as workarounds for broken
2399 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
2400 no longer supported. udev_monitor_new_from_netlink() needs to be
2401 used to subscribe to events.
2403 * udev: when udevd is started by systemd, processes which are left
2404 behind by forking them off of udev rules, are unconditionally cleaned
2405 up and killed now after the event handling has finished. Services or
2406 daemons must be started as systemd services. Services can be
2407 pulled-in by udev to get started, but they can no longer be directly
2408 forked by udev rules.
2410 * udev: the daemon binary is called systemd-udevd now and installed
2411 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
2412 to adapt to that, create symlink, or rename the binary after building
2415 * libudev no longer provides these symbols:
2416 udev_monitor_from_socket()
2417 udev_queue_get_failed_list_entry()
2418 udev_get_{dev,sys,run}_path()
2419 The versions number was bumped and symbol versioning introduced.
2421 * systemd-loginctl and systemd-journalctl have been renamed
2422 to loginctl and journalctl to match systemctl.
2424 * The config files: /etc/systemd/systemd-logind.conf and
2425 /etc/systemd/systemd-journald.conf have been renamed to
2426 logind.conf and journald.conf. Package updates should rename
2427 the files to the new names on upgrade.
2429 * For almost all files the license is now LGPL2.1+, changed
2430 from the previous GPL2.0+. Exceptions are some minor stuff
2431 of udev (which will be changed to LGPL2.1 eventually, too),
2432 and the MIT licensed sd-daemon.[ch] library that is suitable
2433 to be used as drop-in files.
2435 * systemd and logind now handle system sleep states, in
2436 particular suspending and hibernating.
2438 * logind now implements a sleep/shutdown/idle inhibiting logic
2439 suitable for a variety of uses. Soonishly Lennart will blog
2440 about this in more detail.
2442 * var-run.mount and var-lock.mount are no longer provided
2443 (which prevously bind mounted these directories to their new
2444 places). Distributions which have not converted these
2445 directories to symlinks should consider stealing these files
2446 from git history and add them downstream.
2448 * We introduced the Documentation= field for units and added
2449 this to all our shipped units. This is useful to make it
2450 easier to explore the boot and the purpose of the various
2453 * All smaller setup units (such as
2454 systemd-vconsole-setup.service) now detect properly if they
2455 are run in a container and are skipped when
2456 appropriate. This guarantees an entirely noise-free boot in
2457 Linux container environments such as systemd-nspawn.
2459 * A framework for implementing offline system updates is now
2460 integrated, for details see:
2461 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
2463 * A new service type Type=idle is available now which helps us
2464 avoiding ugly interleaving of getty output and boot status
2467 * There's now a system-wide CapabilityBoundingSet= option to
2468 globally reduce the set of capabilities for the
2469 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
2470 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
2471 even CAP_NET_ADMIN system-wide for secure systems.
2473 * There are now system-wide DefaultLimitXXX= options to
2474 globally change the defaults of the various resource limits
2475 for all units started by PID 1.
2477 * Harald Hoyer's systemd test suite has been integrated into
2478 systemd which allows easy testing of systemd builds in qemu
2479 and nspawn. (This is really awesome! Ask us for details!)
2481 * The fstab parser is now implemented as generator, not inside
2484 * systemctl will now warn you if .mount units generated from
2485 /etc/fstab are out of date due to changes in fstab that
2486 haven't been read by systemd yet.
2488 * systemd is now suitable for usage in initrds. Dracut has
2489 already been updated to make use of this. With this in place
2490 initrds get a slight bit faster but primarily are much
2491 easier to introspect and debug since "systemctl status" in
2492 the host system can be used to introspect initrd services,
2493 and the journal from the initrd is kept around too.
2495 * systemd-delta has been added, a tool to explore differences
2496 between user/admin configuration and vendor defaults.
2498 * PrivateTmp= now affects both /tmp and /var/tmp.
2500 * Boot time status messages are now much prettier and feature
2501 proper english language. Booting up systemd has never been
2504 * Read-ahead pack files now include the inode number of all
2505 files to pre-cache. When the inode changes the pre-caching
2506 is not attempted. This should be nicer to deal with updated
2507 packages which might result in changes of read-ahead
2510 * We now temporaritly lower the kernel's read_ahead_kb variable
2511 when collecting read-ahead data to ensure the kernel's
2512 built-in read-ahead does not add noise to our measurements
2513 of necessary blocks to pre-cache.
2515 * There's now RequiresMountsFor= to add automatic dependencies
2516 for all mounts necessary for a specific file system path.
2518 * MountAuto= and SwapAuto= have been removed from
2519 system.conf. Mounting file systems at boot has to take place
2522 * nspawn now learned a new switch --uuid= to set the machine
2523 ID on the command line.
2525 * nspawn now learned the -b switch to automatically search
2528 * vt102 is now the default TERM for serial TTYs, upgraded from
2531 * systemd-logind now works on VT-less systems.
2533 * The build tree has been reorganized. The individual
2534 components now have directories of their own.
2536 * A new condition type ConditionPathIsReadWrite= is now available.
2538 * nspawn learned the new -C switch to create cgroups for the
2539 container in other hierarchies.
2541 * We now have support for hardware watchdogs, configurable in
2544 * The scheduled shutdown logic now has a public API.
2546 * We now mount /tmp as tmpfs by default, but this can be
2547 masked and /etc/fstab can override it.
2549 * Since udisks doesn't make use of /media anymore we are not
2550 mounting a tmpfs on it anymore.
2552 * journalctl gained a new --local switch to only interleave
2553 locally generated journal files.
2555 * We can now load the IMA policy at boot automatically.
2557 * The GTK tools have been split off into a systemd-ui.
2559 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
2560 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
2561 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
2562 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
2563 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
2564 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
2565 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
2566 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
2567 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
2572 * This is mostly a bugfix release
2574 * Support optional initialization of the machine ID from the
2575 KVM or container configured UUID.
2577 * Support immediate reboots with "systemctl reboot -ff"
2579 * Show /etc/os-release data in systemd-analyze output
2581 * Many bugfixes for the journal, including endianness fixes and
2582 ensuring that disk space enforcement works
2584 * sd-login.h is C++ comptaible again
2586 * Extend the /etc/os-release format on request of the Debian
2589 * We now refuse non-UTF8 strings used in various configuration
2590 and unit files. This is done to ensure we don't pass invalid
2591 data over D-Bus or expose it elsewhere.
2593 * Register Mimo USB Screens as suitable for automatic seat
2596 * Read SELinux client context from journal clients in a race
2599 * Reorder configuration file lookup order. /etc now always
2600 overrides /run in order to allow the administrator to always
2601 and unconditionally override vendor supplied or
2602 automatically generated data.
2604 * The various user visible bits of the journal now have man
2605 pages. We still lack man pages for the journal API calls
2608 * We now ship all man pages in HTML format again in the
2611 Contributions from: Dave Reisner, Dirk Eibach, Frederic
2612 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
2613 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
2618 * This is mostly a bugfix release
2620 * systems lacking /etc/os-release are no longer supported.
2622 * Various functionality updates to libsystemd-login.so
2624 * Track class of PAM logins to distuingish greeters from
2627 Contributions from: Kay Sievers, Lennart Poettering, Michael
2632 * This is an important bugfix release for v41.
2634 * Building man pages is now optional which should be useful
2635 for those building systemd from git but unwilling to install
2638 * Watchdog support for supervising services is now usable. In
2639 a future release support for hardware watchdogs
2640 (i.e. /dev/watchdog) will be added building on this.
2642 * Service start rate limiting is now configurable and can be
2643 turned off per service. When a start rate limit is hit a
2644 reboot can automatically be triggered.
2646 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
2648 Contributions from: Benjamin Franzke, Bill Nottingham,
2649 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
2650 Schmidt, Michał Górny, Piotr Drąg
2654 * The systemd binary is installed /usr/lib/systemd/systemd now;
2655 An existing /sbin/init symlink needs to be adapted with the
2658 * The code that loads kernel modules has been ported to invoke
2659 libkmod directly, instead of modprobe. This means we do not
2660 support systems with module-init-tools anymore.
2662 * Watchdog support is now already useful, but still not
2665 * A new kernel command line option systemd.setenv= is
2666 understood to set system wide environment variables
2667 dynamically at boot.
2669 * We now limit the set of capabilities of systemd-journald.
2671 * We now set SIGPIPE to ignore by default, since it only is
2672 useful in shell pipelines, and has little use in general
2673 code. This can be disabled with IgnoreSIPIPE=no in unit
2676 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
2677 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
2682 * This is mostly a bugfix release
2684 * We now expose the reason why a service failed in the
2685 "Result" D-Bus property.
2687 * Rudimentary service watchdog support (will be completed over
2688 the next few releases.)
2690 * When systemd forks off in order execute some service we will
2691 now immediately changes its argv[0] to reflect which process
2692 it will execute. This is useful to minimize the time window
2693 with a generic argv[0], which makes bootcharts more useful
2695 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
2696 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
2697 Mike Kazantsev, Ray Strode
2701 * This is mostly a test release, but incorporates many
2704 * New systemd-cgtop tool to show control groups by their
2707 * Linking against libacl for ACLs is optional again. If
2708 disabled, support tracking device access for active logins
2709 goes becomes unavailable, and so does access to the user
2710 journals by the respective users.
2712 * If a group "adm" exists, journal files are automatically
2713 owned by them, thus allow members of this group full access
2714 to the system journal as well as all user journals.
2716 * The journal now stores the SELinux context of the logging
2717 client for all entries.
2719 * Add C++ inclusion guards to all public headers
2721 * New output mode "cat" in the journal to print only text
2722 messages, without any meta data like date or time.
2724 * Include tiny X server wrapper as a temporary stop-gap to
2725 teach XOrg udev display enumeration. This is used by display
2726 managers such as gdm, and will go away as soon as XOrg
2727 learned native udev hotplugging for display devices.
2729 * Add new systemd-cat tool for executing arbitrary programs
2730 with STDERR/STDOUT connected to the journal. Can also act as
2731 BSD logger replacement, and does so by default.
2733 * Optionally store all locally generated coredumps in the
2734 journal along with meta data.
2736 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
2737 writing short strings to files (for usage for /sys), and for
2738 creating symlinks, character and block device nodes.
2740 * New unit file option ControlGroupPersistent= to make cgroups
2741 persistent, following the mechanisms outlined in
2742 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
2744 * Support multiple local RTCs in a sane way
2746 * No longer monopolize IO when replaying readahead data on
2747 rotating disks, since we might starve non-file-system IO to
2748 death, since fanotify() will not see accesses done by blkid,
2751 * Don't show kernel threads in systemd-cgls anymore, unless
2752 requested with new -k switch.
2754 Contributions from: Dan Horák, Kay Sievers, Lennart
2755 Poettering, Michal Schmidt
2759 * This is mostly a test release, but incorporates many
2762 * The git repository moved to:
2763 git://anongit.freedesktop.org/systemd/systemd
2764 ssh://git.freedesktop.org/git/systemd/systemd
2766 * First release with the journal
2767 http://0pointer.de/blog/projects/the-journal.html
2769 * The journal replaces both systemd-kmsg-syslogd and
2770 systemd-stdout-bridge.
2772 * New sd_pid_get_unit() API call in libsystemd-logind
2774 * Many systemadm clean-ups
2776 * Introduce remote-fs-pre.target which is ordered before all
2777 remote mounts and may be used to start services before all
2780 * Added Mageia support
2782 * Add bash completion for systemd-loginctl
2784 * Actively monitor PID file creation for daemons which exit in
2785 the parent process before having finished writing the PID
2786 file in the daemon process. Daemons which do this need to be
2787 fixed (i.e. PID file creation must have finished before the
2788 parent exits), but we now react a bit more gracefully to them.
2790 * Add colourful boot output, mimicking the well-known output
2791 of existing distributions.
2793 * New option PassCredentials= for socket units, for
2794 compatibility with a recent kernel ABI breakage.
2796 * /etc/rc.local is now hooked in via a generator binary, and
2797 thus will no longer act as synchronization point during
2800 * systemctl list-unit-files now supports --root=.
2802 * systemd-tmpfiles now understands two new commands: z, Z for
2803 relabelling files according to the SELinux database. This is
2804 useful to apply SELinux labels to specific files in /sys,
2807 * Output of SysV services is now forwarded to both the console
2808 and the journal by default, not only just the console.
2810 * New man pages for all APIs from libsystemd-login.
2812 * The build tree got reorganized and a the build system is a
2813 lot more modular allowing embedded setups to specifically
2814 select the components of systemd they are interested in.
2816 * Support for Linux systems lacking the kernel VT subsystem is
2819 * configure's --with-rootdir= got renamed to
2820 --with-rootprefix= to follow the naming used by udev and
2823 * Unless specified otherwise we'll now install to /usr instead
2824 of /usr/local by default.
2826 * Processes with '@' in argv[0][0] are now excluded from the
2827 final shut-down killing spree, following the logic explained
2829 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
2831 * All processes remaining in a service cgroup when we enter
2832 the START or START_PRE states are now killed with
2833 SIGKILL. That means it is no longer possible to spawn
2834 background processes from ExecStart= lines (which was never
2835 supported anyway, and bad style).
2837 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
2838 reloading of units together.
2840 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
2841 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
2842 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
2843 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
2844 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek