1 systemd System and Service Manager
5 * A new unit file setting RestrictAddressFamilies= has been
6 added to restrict which socket address families unit
7 processes gain access to. This takes address family names
8 like "AF_INET" or "AF_UNIX", and is useful to minimize the
9 attack surface of services via exotic protocol stacks. This
10 is built on seccomp system call filters.
12 * Two new unit file settings RuntimeDirectory= and
13 RuntimeDirectoryMode= have been added that may be used to
14 manage a per-daemon runtime directories below /run. This is
15 an alternative for setting up directory permissions with
16 tmpfiles snippets, and has the advantage that the runtime
17 directory's lifetime is bound to the daemon runtime and that
18 the daemon starts up with an empty directory each time. This
19 is particularly useful when writing services that drop
20 priviliges using the User= or Group= setting.
22 * The DeviceAllow= unit setting now supports globbing for
23 matching against device group names.
25 * The systemd configuration file system.conf gained new
26 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
27 DefaultMemoryAccounting= to globally turn on/off accounting
28 for specific resources (cgroups) for all units. These
29 settings may still be overridden individually in each unit
32 * systemd-gpt-auto-generator is now able to discover /srv and
33 root partitions in addition to /home and swap partitions. It
34 also supports LUKS-encrypted partitions now. With this in
35 place automatic discovery of partitions to mount following
36 the Discoverable Partitions Specification
37 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
38 is now a lot more complete. This allows booting without
39 /etc/fstab and without root= on the kernel command line on
40 appropriately prepared systems.
42 * systemd-nspawn gained a new --image= switch which allows
43 booting up disk images and Linux installations on any block
44 device that follow the Discoverable Partitions Specification
45 (see above). This means that installations made with
46 appropriately updated installers may now be started and
47 deployed using container managers, completely
48 unmodified. (We hope that libvirt-lxc will add support for
49 this feature soon, too.)
51 * systemd-nspawn gained a new --network-macvlan= setting to
52 set up a private macvlan interface for the
53 container. Similar, systemd-networkd gained a new
54 Kind=macvlan setting in .netdev files.
56 * systemd-networkd now supports configuring local addresses
59 * A new tool systemd-network-wait-online has been added to
60 synchronously wait for network connectivity using
63 * The sd-bus.h bus API gained a new sd_bus_track object for
64 tracking the life-cycle of bus peers. Note that sd-bus.h is
65 still not a public API though (unless you specify
66 --enable-kdbus on the configure command line, which however
67 voids your warranty and you get no API stability guarantee).
69 * The $XDG_RUNTIME_DIR runtime directories for each user are
70 now individual tmpfs instances, which has the benefit of
71 introducing separate pools for each user, with individual
72 size limits, and thus making sure that unprivileged clients
73 can no longer negatively impact the system or other users by
74 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
75 RuntimeDirectorySize= has been introduced that allows
76 controlling the default size limit for all users. It
77 defaults to 10% of the available physical memory. This is no
78 replacement for quotas on tmpfs though (which the kernel
79 still does not support), as /dev/shm and /tmp are still
80 shared resources used by both the system and unprivileged
83 * logind will now automatically turn off automatic suspending
84 on laptop lid close when more than one display is
85 connected. This was previously expected to be implemented
86 individually in desktop environments (such as GNOME),
87 however has been added to logind now, in order to fix a
88 boot-time race where a desktop environment might not have
89 been started yet and thus not been able to take an inhibitor
90 lock at the time where logind already suspends the system
93 * logind will now wait at least 30s after each system
94 suspend/resume cycle, and 3min after system boot before
95 suspending the system due to a closed laptop lid. This
96 should give USB docking stations and similar enough time to
97 be probed and configured after system resume and boot in
98 order to then act as suspend blocker.
100 * systemd-run gained a new --property= setting which allows
101 initialization of resource control properties (and others)
102 for the created scope or service unit. Example: "systemd-run
103 --property=BlockIOWeight=10 updatedb" may be used to run
104 updatedb at a low block IO scheduling weight.
106 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
107 now also work in --scope mode.
109 * When systemd is compiled with kdbus support, basic support
110 for enforced policies is now in place. (Note that enabling
111 kdbus still voids your warranty and no API compatibility
114 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
115 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
116 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
117 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
118 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
119 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
120 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
121 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
122 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
123 Zbigniew Jędrzejewski-Szmek
125 -- Berlin, 2014-03-12
129 * systemd will now relabel /dev after loading the SMACK policy
130 according to SMACK rules.
132 * A new unit file option AppArmorProfile= has been added to
133 set the AppArmor profile for the processes of a unit.
135 * A new condition check ConditionArchitecture= has been added
136 to conditionalize units based on the system architecture, as
137 reported by uname()'s "machine" field.
139 * systemd-networkd now supports matching on the system
140 virtualization, architecture, kernel command line, host name
143 * logind is now a lot more aggressive when suspending the
144 machine due to a closed laptop lid. Instead of acting only
145 on the lid close action it will continuously watch the lid
146 status and act on it. This is useful for laptops where the
147 power button is on the outside of the chassis so that it can
148 be reached without opening the lid (such as the Lenovo
149 Yoga). On those machines logind will now immediately
150 re-suspend the machine if the power button has been
151 accidentally pressed while the laptop was suspended and in a
154 * logind will now watch SW_DOCK switches and inhibit reaction
155 to the lid switch if it is pressed. This means that logind
156 will not suspend the machine anymore if the lid is closed
157 and the system is docked, if the laptop supports SW_DOCK
158 notifications via the input layer. Note that ACPI docking
159 stations do not generate this currently. Also note that this
160 logic is usually not fully sufficient and Desktop
161 Environments should take a lid switch inhibitor lock when an
162 external display is connected, as systemd will not watch
165 * nspawn will now make use of the devices cgroup controller by
166 default, and only permit creation of and access to the usual
167 API device nodes like /dev/null or /dev/random, as well as
168 access to (but not creation of) the pty devices.
170 * We will now ship a default .network file for
171 systemd-networkd that automatically configures DHCP for
172 network interfaces created by nspawn's --network-veth or
173 --network-bridge= switches.
175 * systemd will now understand the usual M, K, G, T suffixes
176 according to SI conventions (i.e. to the base 1000) when
177 referring to throughput and hardware metrics. It will stay
178 with IEC conventions (i.e. to the base 1024) for software
179 metrics, according to what is customary according to
180 Wikipedia. We explicitly document which base applies for
181 each configuration option.
183 * The DeviceAllow= setting in unit files now supports a syntax
184 to whitelist an entire group of devices node majors at once,
185 based on the /proc/devices listing. For example, with the
186 string "char-pts" it is now possible to whitelist all
187 current and future pseudo-TTYs at once.
189 * sd-event learned a new "post" event source. Event sources of
190 this type are triggered by the dispatching of any event
191 source of a type that is not "post". This is useful for
192 implementing clean-up and check event sources that are
193 triggered by other work being done in the program.
195 * systemd-networkd is no longer statically enabled, but uses
196 the usual [Install] sections so that it can be
197 enabled/disabled using systemctl. It still is enabled by
200 * When creating a veth interface pair with systemd-nspawn the
201 host side will now be prefixed with "vb-" if
202 --network-bridge= is used, and with "ve-" if --network-veth
203 is used. This way it is easy to distinguish these cases on
204 the host, for example to apply different configuration to
205 them with systemd-networkd.
207 * The compatibility libraries for libsystemd-journal.so,
208 libsystem-id128.so, libsystemd-login.so and
209 libsystemd-daemon.so do not make use of IFUNC
210 anymore. Instead we now build libsystemd.so multiple times
211 under these alternative names. This means that the footprint
212 is drastically increased, but given that these are
213 transitional compatibility libraries this shouldn't matter
214 much. This change has been made necessary to support the ARM
215 platform for these compatibility libraries, as the ARM
216 toolchain isn't really at the same level as the toolchain
217 for other architectures like x86 and does not support
218 IFUNC. Please make sure to use --enable-compat-libs only
219 during a transitional period!
221 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
222 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
223 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
224 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
225 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
226 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
227 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
228 Zbigniew Jędrzejewski-Szmek
230 -- Berlin, 2014-02-24
234 * A new component "systemd-networkd" has been added that can
235 be used to configure local network interfaces statically or
236 via DHCP. It is capable of bringing up bridges, VLANs, and
237 bonding. Currently, no hook-ups for interactive network
238 configuration are provided. Use this for your initrd,
239 container, embedded, or server setup if you need a simple,
240 yet powerful, network configuration solution. This
241 configuration subsystem is quite nifty, as it allows wildcard
242 hotplug matching in interfaces. For example, with a single
243 configuration snippet, you can configure that all Ethernet
244 interfaces showing up are automatically added to a bridge,
245 or similar. It supports link-sensing and more.
247 * A new tool "systemd-socket-proxyd" has been added which can
248 act as a bidirectional proxy for TCP sockets. This is
249 useful for adding socket activation support to services that
250 do not actually support socket activation, including virtual
251 machines and the like.
253 * Add a new tool to save/restore rfkill state on
256 * Save/restore state of keyboard backlights in addition to
257 display backlights on shutdown/boot.
259 * udev learned a new SECLABEL{} construct to label device
260 nodes with a specific security label when they appear. For
261 now, only SECLABEL{selinux} is supported, but the syntax is
262 prepared for additional security frameworks.
264 * udev gained a new scheme to configure link-level attributes
265 from files in /etc/systemd/network/*.link. These files can
266 match against MAC address, device path, driver name and type,
267 and will apply attributes like the naming policy, link speed,
268 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
269 address assignment policy (randomized, ...).
271 * The configuration of network interface naming rules for
272 "permanent interface names" has changed: a new NamePolicy=
273 setting in the [Link] section of .link files determines the
274 priority of possible naming schemes (onboard, slot, mac,
275 path). The default value of this setting is determined by
276 /usr/lib/net/links/99-default.link. Old
277 80-net-name-slot.rules udev configuration file has been
278 removed, so local configuration overriding this file should
279 be adapated to override 99-default.link instead.
281 * When the User= switch is used in a unit file, also
282 initialize $SHELL= based on the user database entry.
284 * systemd no longer depends on libdbus. All communication is
285 now done with sd-bus, systemd's low-level bus library
288 * kdbus support has been added to PID 1 itself. When kdbus is
289 enabled, this causes PID 1 to set up the system bus and
290 enable support for a new ".busname" unit type that
291 encapsulates bus name activation on kdbus. It works a little
292 bit like ".socket" units, except for bus names. A new
293 generator has been added that converts classic dbus1 service
294 activation files automatically into native systemd .busname
297 * sd-bus: add a light-weight vtable implementation that allows
298 defining objects on the bus with a simple static const
299 vtable array of its methods, signals and properties.
301 * systemd will not generate or install static dbus
302 introspection data anymore to /usr/share/dbus-1/interfaces,
303 as the precise format of these files is unclear, and
304 nothing makes use of it.
306 * A proxy daemon is now provided to proxy clients connecting
307 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
308 compatibility with classic D-Bus.
310 * A bus driver implementation has been added that supports the
311 classic D-Bus bus driver calls on kdbus, also for
312 compatibility purposes.
314 * A new API "sd-event.h" has been added that implements a
315 minimal event loop API built around epoll. It provides a
316 couple of features that direct epoll usage is lacking:
317 prioritization of events, scales to large numbers of timer
318 events, per-event timer slack (accuracy), system-wide
319 coalescing of timer events, exit handlers, watchdog
320 supervision support using systemd's sd_notify() API, child
323 * A new API "sd-rntl.h" has been added that provides an API
324 around the route netlink interface of the kernel, similar in
327 * A new API "sd-dhcp-client.h" has been added that provides a
328 small DHCPv4 client-side implementation. This is used by
331 * There is a new kernel command line option
332 "systemd.restore_state=0|1". When set to "0", none of the
333 systemd tools will restore saved runtime state to hardware
334 devices. More specifically, the rfkill and backlight states
337 * The FsckPassNo= compatibility option in mount/service units
338 has been removed. The fstab generator will now add the
339 necessary dependencies automatically, and does not require
340 PID1's support for that anymore.
342 * journalctl gained a new switch, --list-boots, that lists
343 recent boots with their times and boot IDs.
345 * The various tools like systemctl, loginctl, timedatectl,
346 busctl, systemd-run, ... have gained a new switch "-M" to
347 connect to a specific, local OS container (as direct
348 connection, without requiring SSH). This works on any
349 container that is registered with machined, such as those
350 created by libvirt-lxc or nspawn.
352 * systemd-run and systemd-analyze also gained support for "-H"
353 to connect to remote hosts via SSH. This is particularly
354 useful for systemd-run because it enables queuing of jobs
357 * machinectl gained a new command "login" to open a getty
358 login in any local container. This works with any container
359 that is registered with machined (such as those created by
360 libvirt-lxc or nspawn), and which runs systemd inside.
362 * machinectl gained a new "reboot" command that may be used to
363 trigger a reboot on a specific container that is registered
364 with machined. This works on any container that runs an init
367 * systemctl gained a new "list-timers" command to print a nice
368 listing of installed timer units with the times they elapse
371 * Alternative reboot() parameters may now be specified on the
372 "systemctl reboot" command line and are passed to the
373 reboot() system call.
375 * systemctl gained a new --job-mode= switch to configure the
376 mode to queue a job with. This is a more generic version of
377 --fail, --irreversible, and --ignore-dependencies, which are
378 still available but not advertised anymore.
380 * /etc/systemd/system.conf gained new settings to configure
381 various default timeouts of units, as well as the default
382 start limit interval and burst. These may still be overridden
385 * PID1 will now export on the bus profile data of the security
386 policy upload process (such as the SELinux policy upload to
389 * journald: when forwarding logs to the console, include
390 timestamps (following the setting in
391 /sys/module/printk/parameters/time).
393 * OnCalendar= in timer units now understands the special
394 strings "yearly" and "annually". (Both are equivalent)
396 * The accuracy of timer units is now configurable with the new
397 AccuracySec= setting. It defaults to 1min.
399 * A new dependency type JoinsNamespaceOf= has been added that
400 allows running two services within the same /tmp and network
401 namespace, if PrivateNetwork= or PrivateTmp= are used.
403 * A new command "cat" has been added to systemctl. It outputs
404 the original unit file of a unit, and concatenates the
405 contents of additional "drop-in" unit file snippets, so that
406 the full configuration is shown.
408 * systemctl now supports globbing on the various "list-xyz"
409 commands, like "list-units" or "list-sockets", as well as on
410 those commands which take multiple unit names.
412 * journalctl's --unit= switch gained support for globbing.
414 * All systemd daemons now make use of the watchdog logic so
415 that systemd automatically notices when they hang.
417 * If the $container_ttys environment variable is set,
418 getty-generator will automatically spawn a getty for each
419 listed tty. This is useful for container managers to request
420 login gettys to be spawned on as many ttys as needed.
422 * %h, %s, %U specifier support is not available anymore when
423 used in unit files for PID 1. This is because NSS calls are
424 not safe from PID 1. They stay available for --user
425 instances of systemd, and as special case for the root user.
427 * loginctl gained a new "--no-legend" switch to turn off output
430 * The "sd-login.h" API gained three new calls:
431 sd_session_is_remote(), sd_session_get_remote_user(),
432 sd_session_get_remote_host() to query information about
435 * The udev hardware database now also carries vendor/product
436 information of SDIO devices.
438 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
439 determine whether watchdog notifications are requested by
442 * Socket-activated per-connection services now include a
443 short description of the connection parameters in the
446 * tmpfiles gained a new "--boot" option. When this is not used,
447 only lines where the command character is not suffixed with
448 "!" are executed. When this option is specified, those
449 options are executed too. This partitions tmpfiles
450 directives into those that can be safely executed at any
451 time, and those which should be run only at boot (for
452 example, a line that creates /run/nologin).
454 * A new API "sd-resolve.h" has been added which provides a simple
455 asynchronous wrapper around glibc NSS host name resolution
456 calls, such as getaddrinfo(). In contrast to glibc's
457 getaddrinfo_a(), it does not use signals. In contrast to most
458 other asynchronous name resolution libraries, this one does
459 not reimplement DNS, but reuses NSS, so that alternate
460 host name resolution systems continue to work, such as mDNS,
461 LDAP, etc. This API is based on libasyncns, but it has been
462 cleaned up for inclusion in systemd.
464 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
465 "sd-daemon.h" are no longer found in individual libraries
466 libsystemd-journal.so, libsystemd-login.so,
467 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
468 merged them into a single library, libsystemd.so, which
469 provides all symbols. The reason for this is cyclic
470 dependencies, as these libraries tend to use each other's
471 symbols. So far, we've managed to workaround that by linking
472 a copy of a good part of our code into each of these
473 libraries again and again, which, however, makes certain
474 things hard to do, like sharing static variables. Also, it
475 substantially increases footprint. With this change, there
476 is only one library for the basic APIs systemd
477 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
478 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
479 library as well, however are subject to the --enable-kdbus
480 switch (see below). Note that "sd-dhcp-client.h" is not part
481 of this library (this is because it only consumes, never
482 provides, services of/to other APIs). To make the transition
483 easy from the separate libraries to the unified one, we
484 provide the --enable-compat-libs compile-time switch which
485 will generate stub libraries that are compatible with the
486 old ones but redirect all calls to the new one.
488 * All of the kdbus logic and the new APIs "sd-bus.h",
489 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
490 and "sd-utf8.h" are compile-time optional via the
491 "--enable-kdbus" switch, and they are not compiled in by
492 default. To make use of kdbus, you have to explicitly enable
493 the switch. Note however, that neither the kernel nor the
494 userspace API for all of this is considered stable yet. We
495 want to maintain the freedom to still change the APIs for
496 now. By specifying this build-time switch, you acknowledge
497 that you are aware of the instability of the current
500 * Also, note that while kdbus is pretty much complete,
501 it lacks one thing: proper policy support. This means you
502 can build a fully working system with all features; however,
503 it will be highly insecure. Policy support will be added in
504 one of the next releases, at the same time that we will
505 declare the APIs stable.
507 * When the kernel command-line argument "kdbus" is specified,
508 systemd will automatically load the kdbus.ko kernel module. At
509 this stage of development, it is only useful for testing kdbus
510 and should not be used in production. Note: if "--enable-kdbus"
511 is specified, and the kdbus.ko kernel module is available, and
512 "kdbus" is added to the kernel command line, the entire system
513 runs with kdbus instead of dbus-daemon, with the above mentioned
514 problem of missing the system policy enforcement. Also a future
515 version of kdbus.ko or a newer systemd will not be compatible with
516 each other, and will unlikely be able to boot the machine if only
517 one of them is updated.
519 * systemctl gained a new "import-environment" command which
520 uploads the caller's environment (or parts thereof) into the
521 service manager so that it is inherited by services started
522 by the manager. This is useful to upload variables like
523 $DISPLAY into the user service manager.
525 * A new PrivateDevices= switch has been added to service units
526 which allows running a service with a namespaced /dev
527 directory that does not contain any device nodes for
528 physical devices. More specifically, it only includes devices
529 such as /dev/null, /dev/urandom, and /dev/zero which are API
532 * logind has been extended to support behaviour like VT
533 switching on seats that do not support a VT. This makes
534 multi-session available on seats that are not the first seat
535 (seat0), and on systems where kernel support for VTs has
536 been disabled at compile-time.
538 * If a process holds a delay lock for system sleep or shutdown
539 and fails to release it in time, we will now log its
540 identity. This makes it easier to identify processes that
541 cause slow suspends or power-offs.
543 * When parsing /etc/crypttab, support for a new key-slot=
544 option as supported by Debian is added. It allows indicating
545 which LUKS slot to use on disk, speeding up key loading.
547 * The sd_journald_sendv() API call has been checked and
548 officially declared to be async-signal-safe so that it may
549 be invoked from signal handlers for logging purposes.
551 * Boot-time status output is now enabled automatically after a
552 short timeout if boot does not progress, in order to give
553 the user an indication what she or he is waiting for.
555 * The boot-time output has been improved to show how much time
556 remains until jobs expire.
558 * The KillMode= switch in service units gained a new possible
559 value "mixed". If set, and the unit is shut down, then the
560 initial SIGTERM signal is sent only to the main daemon
561 process, while the following SIGKILL signal is sent to
562 all remaining processes of the service.
564 * When a scope unit is registered, a new property "Controller"
565 may be set. If set to a valid bus name, systemd will send a
566 RequestStop() signal to this name when it would like to shut
567 down the scope. This may be used to hook manager logic into
568 the shutdown logic of scope units. Also, scope units may now
569 be put in a special "abandoned" state, in which case the
570 manager process which created them takes no further
571 responsibilities for it.
573 * When reading unit files, systemd will now verify
574 the access mode of these files, and warn about certain
575 suspicious combinations. This has been added to make it
576 easier to track down packaging bugs where unit files are
577 marked executable or world-writable.
579 * systemd-nspawn gained a new "--setenv=" switch to set
580 container-wide environment variables. The similar option in
581 systemd-activate was renamed from "--environment=" to
582 "--setenv=" for consistency.
584 * systemd-nspawn has been updated to create a new kdbus domain
585 for each container that is invoked, thus allowing each
586 container to have its own set of system and user buses,
587 independent of the host.
589 * systemd-nspawn gained a new --drop-capability= switch to run
590 the container with less capabilities than the default. Both
591 --drop-capability= and --capability= now take the special
592 string "all" for dropping or keeping all capabilities.
594 * systemd-nspawn gained new switches for executing containers
595 with specific SELinux labels set.
597 * systemd-nspawn gained a new --quiet switch to not generate
598 any additional output but the container's own console
601 * systemd-nspawn gained a new --share-system switch to run a
602 container without PID namespacing enabled.
604 * systemd-nspawn gained a new --register= switch to control
605 whether the container is registered with systemd-machined or
606 not. This is useful for containers that do not run full
607 OS images, but only specific apps.
609 * systemd-nspawn gained a new --keep-unit which may be used
610 when invoked as the only program from a service unit, and
611 results in registration of the unit service itself in
612 systemd-machined, instead of a newly opened scope unit.
614 * systemd-nspawn gained a new --network-interface= switch for
615 moving arbitrary interfaces to the container. The new
616 --network-veth switch creates a virtual Ethernet connection
617 between host and container. The new --network-bridge=
618 switch then allows assigning the host side of this virtual
619 Ethernet connection to a bridge device.
621 * systemd-nspawn gained a new --personality= switch for
622 setting the kernel personality for the container. This is
623 useful when running a 32bit container on a 64bit host. A
624 similar option Personality= is now also available in service
627 * logind will now also track a "Desktop" identifier for each
628 session which encodes the desktop environment of it. This is
629 useful for desktop environments that want to identify
630 multiple running sessions of itself easily.
632 * A new SELinuxContext= setting for service units has been
633 added that allows setting a specific SELinux execution
634 context for a service.
636 * Most systemd client tools will now honour $SYSTEMD_LESS for
637 settings of the "less" pager. By default, these tools will
638 override $LESS to allow certain operations to work, such as
639 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
640 influence this logic.
642 * systemd's "seccomp" hook-up has been changed to make use of
643 the libseccomp library instead of using its own
644 implementation. This has benefits for portability among
647 * For usage together with SystemCallFilter=, a new
648 SystemCallErrorNumber= setting has been introduced that
649 allows configuration of a system error number to return on
650 filtered system calls, instead of immediately killing the
651 process. Also, SystemCallArchitectures= has been added to
652 limit access to system calls of a particular architecture
653 (in order to turn off support for unused secondary
654 architectures). There is also a global
655 SystemCallArchitectures= setting in system.conf now to turn
656 off support for non-native system calls system-wide.
658 * systemd requires a kernel with a working name_to_handle_at(),
659 please see the kernel config requirements in the README file.
661 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
662 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
663 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
664 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
665 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
666 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
667 Elia Pinto, Florian Weimer, George McCollister, Goffredo
668 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
669 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
670 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
671 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
672 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
673 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
674 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
675 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
676 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
677 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
678 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
679 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
680 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
681 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
682 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
683 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
684 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
686 -- Berlin, 2014-02-20
690 * logind has gained support for facilitating privileged input
691 and drm device access for unprivileged clients. This work is
692 useful to allow Wayland display servers (and similar
693 programs, such as kmscon) to run under the user's ID and
694 access input and drm devices which are normally
695 protected. When this is used (and the kernel is new enough)
696 logind will "mute" IO on the file descriptors passed to
697 Wayland as long as it is in the background and "unmute" it
698 if it returns into the foreground. This allows secure
699 session switching without allowing background sessions to
700 eavesdrop on input and display data. This also introduces
701 session switching support if VT support is turned off in the
702 kernel, and on seats that are not seat0.
704 * A new kernel command line option luks.options= is understood
705 now which allows specifiying LUKS options for usage for LUKS
706 encrypted partitions specified with luks.uuid=.
708 * tmpfiles.d(5) snippets may now use specifier expansion in
709 path names. More specifically %m, %b, %H, %v, are now
710 replaced by the local machine id, boot id, hostname, and
711 kernel version number.
713 * A new tmpfiles.d(5) command "m" has been introduced which
714 may be used to change the owner/group/access mode of a file
715 or directory if it exists, but do nothing if it doesn't.
717 * This release removes high-level support for the
718 MemorySoftLimit= cgroup setting. The underlying kernel
719 cgroup attribute memory.soft_limit= is currently badly
720 designed and likely to be removed from the kernel API in its
721 current form, hence we shouldn't expose it for now.
723 * The memory.use_hierarchy cgroup attribute is now enabled for
724 all cgroups systemd creates in the memory cgroup
725 hierarchy. This option is likely to be come the built-in
726 default in the kernel anyway, and the non-hierarchial mode
727 never made much sense in the intrinsically hierarchial
730 * A new field _SYSTEMD_SLICE= is logged along with all journal
731 messages containing the slice a message was generated
732 from. This is useful to allow easy per-customer filtering of
733 logs among other things.
735 * systemd-journald will no longer adjust the group of journal
736 files it creates to the "systemd-journal" group. Instead we
737 rely on the journal directory to be owned by the
738 "systemd-journal" group, and its setgid bit set, so that the
739 kernel file system layer will automatically enforce that
740 journal files inherit this group assignment. The reason for
741 this change is that we cannot allow NSS look-ups from
742 journald which would be necessary to resolve
743 "systemd-journal" to a numeric GID, because this might
744 create deadlocks if NSS involves synchronous queries to
745 other daemons (such as nscd, or sssd) which in turn are
746 logging clients of journald and might block on it, which
747 would then dead lock. A tmpfiles.d(5) snippet included in
748 systemd will make sure the setgid bit and group are
749 properly set on the journal directory if it exists on every
750 boot. However, we recommend adjusting it manually after
751 upgrades too (or from RPM scriptlets), so that the change is
752 not delayed until next reboot.
754 * Backlight and random seed files in /var/lib/ have moved into
755 the /var/lib/systemd/ directory, in order to centralize all
756 systemd generated files in one directory.
758 * Boot time performance measurements (as displayed by
759 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
760 performance information if that's available to determine how
761 much time BIOS and boot loader initialization required. With
762 a sufficiently new BIOS you hence no longer need to boot
763 with Gummiboot to get access to such information.
765 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
766 Cristian Rodríguez, Dave Reisner, David Herrmann, David
767 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
768 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
769 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
770 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
771 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
773 -- Berlin, 2013-10-02
777 * The Restart= option for services now understands a new
778 on-watchdog setting, which will restart the service
779 automatically if the service stops sending out watchdog keep
780 alive messages (as configured with WatchdogSec=).
782 * The getty generator (which is responsible for bringing up a
783 getty on configured serial consoles) will no longer only
784 start a getty on the primary kernel console but on all
785 others, too. This makes the order in which console= is
786 specified on the kernel command line less important.
788 * libsystemd-logind gained a new sd_session_get_vt() call to
789 retrieve the VT number of a session.
791 * If the option "tries=0" is set for an entry of /etc/crypttab
792 its passphrase is queried indefinitely instead of any
793 maximum number of tries.
795 * If a service with a configure PID file terminates its PID
796 file will now be removed automatically if it still exists
797 afterwards. This should put an end to stale PID files.
799 * systemd-run will now also take relative binary path names
800 for execution and no longer insists on absolute paths.
802 * InaccessibleDirectories= and ReadOnlyDirectories= now take
803 paths that are optionally prefixed with "-" to indicate that
804 it should not be considered a failure if they don't exist.
806 * journalctl -o (and similar commands) now understands a new
807 output mode "short-precise", it is similar to "short" but
808 shows timestamps with usec accuracy.
810 * The option "discard" (as known from Debian) is now
811 synonymous to "allow-discards" in /etc/crypttab. In fact,
812 "discard" is preferred now (since it is easier to remember
815 * Some licensing clean-ups were made, so that more code is now
816 LGPL-2.1 licensed than before.
818 * A minimal tool to save/restore the display backlight
819 brightness across reboots has been added. It will store the
820 backlight setting as late as possible at shutdown, and
821 restore it as early as possible during reboot.
823 * A logic to automatically discover and enable home and swap
824 partitions on GPT disks has been added. With this in place
825 /etc/fstab becomes optional for many setups as systemd can
826 discover certain partitions located on the root disk
827 automatically. Home partitions are recognized under their
828 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
829 partitions are recognized under their GPT type ID
830 0657fd6da4ab43c484e50933c84b4f4f.
832 * systemd will no longer pass any environment from the kernel
833 or initrd to system services. If you want to set an
834 environment for all services, do so via the kernel command
835 line systemd.setenv= assignment.
837 * The systemd-sysctl tool no longer natively reads the file
838 /etc/sysctl.conf. If desired, the file should be symlinked
839 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
840 legacy support by a symlink rather than built-in code, it
841 also makes the otherwise hidden order of application of the
842 different files visible. (Note that this partly reverts to a
843 pre-198 application order of sysctl knobs!)
845 * The "systemctl set-log-level" and "systemctl dump" commands
846 have been moved to systemd-analyze.
848 * systemd-run learned the new --remain-after-exit switch,
849 which causes the scope unit not to be cleaned up
850 automatically after the process terminated.
852 * tmpfiles learned a new --exclude-prefix= switch to exclude
853 certain paths from operation.
855 * journald will now automatically flush all messages to disk
856 as soon as a message of the log priorities CRIT, ALERT or
859 Contributions from: Andrew Cook, Brandon Philips, Christian
860 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
861 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
862 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
863 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
864 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
865 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
866 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
867 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
868 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
869 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
870 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
871 William Giokas, Zbigniew Jędrzejewski-Szmek
873 -- Berlin, 2013-09-13
877 * The documentation has been updated to cover the various new
878 concepts introduced with 205.
880 * Unit files now understand the new %v specifier which
881 resolves to the kernel version string as returned by "uname
884 * systemctl now supports filtering the unit list output by
885 load state, active state and sub state, using the new
888 * "systemctl status" will now show the results of the
889 condition checks (like ConditionPathExists= and similar) of
890 the last start attempts of the unit. They are also logged to
893 * "journalctl -b" may now be used to look for boot output of a
894 specific boot. Try "journalctl -b -1" for the previous boot,
895 but the syntax is substantially more powerful.
897 * "journalctl --show-cursor" has been added which prints the
898 cursor string the last shown log line. This may then be used
899 with the new "journalctl --after-cursor=" switch to continue
900 browsing logs from that point on.
902 * "journalctl --force" may now be used to force regeneration
905 * Creation of "dead" device nodes has been moved from udev
906 into kmod and tmpfiles. Previously, udev would read the kmod
907 databases to pre-generate dead device nodes based on meta
908 information contained in kernel modules, so that these would
909 be auto-loaded on access rather then at boot. As this
910 doesn't really have much to do with the exposing actual
911 kernel devices to userspace this has always been slightly
912 alien in the udev codebase. Following the new scheme kmod
913 will now generate a runtime snippet for tmpfiles from the
914 module meta information and it now is tmpfiles' job to the
915 create the nodes. This also allows overriding access and
916 other parameters for the nodes using the usual tmpfiles
917 facilities. As side effect this allows us to remove the
918 CAP_SYS_MKNOD capability bit from udevd entirely.
920 * logind's device ACLs may now be applied to these "dead"
921 devices nodes too, thus finally allowing managed access to
922 devices such as /dev/snd/sequencer whithout loading the
923 backing module right-away.
925 * A new RPM macro has been added that may be used to apply
926 tmpfiles configuration during package installation.
928 * systemd-detect-virt and ConditionVirtualization= now can
929 detect User-Mode-Linux machines (UML).
931 * journald will now implicitly log the effective capabilities
932 set of processes in the message metadata.
934 * systemd-cryptsetup has gained support for TrueCrypt volumes.
936 * The initrd interface has been simplified (more specifically,
937 support for passing performance data via environment
938 variables and fsck results via files in /run has been
939 removed). These features were non-essential, and are
940 nowadays available in a much nicer way by having systemd in
941 the initrd serialize its state and have the hosts systemd
942 deserialize it again.
944 * The udev "keymap" data files and tools to apply keyboard
945 specific mappings of scan to key codes, and force-release
946 scan code lists have been entirely replaced by a udev
947 "keyboard" builtin and a hwdb data file.
949 * systemd will now honour the kernel's "quiet" command line
950 argument also during late shutdown, resulting in a
951 completely silent shutdown when used.
953 * There's now an option to control the SO_REUSEPORT socket
954 option in .socket units.
956 * Instance units will now automatically get a per-template
957 subslice of system.slice unless something else is explicitly
958 configured. For example, instances of sshd@.service will now
959 implicitly be placed in system-sshd.slice rather than
960 system.slice as before.
962 * Test coverage support may now be enabled at build time.
964 Contributions from: Dave Reisner, Frederic Crozat, Harald
965 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
966 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
967 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
968 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
969 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
970 Giokas, Zbigniew Jędrzejewski-Szmek
972 -- Berlin, 2013-07-23
976 * Two new unit types have been introduced:
978 Scope units are very similar to service units, however, are
979 created out of pre-existing processes -- instead of PID 1
980 forking off the processes. By using scope units it is
981 possible for system services and applications to group their
982 own child processes (worker processes) in a powerful way
983 which then maybe used to organize them, or kill them
984 together, or apply resource limits on them.
986 Slice units may be used to partition system resources in an
987 hierarchial fashion and then assign other units to them. By
988 default there are now three slices: system.slice (for all
989 system services), user.slice (for all user sessions),
990 machine.slice (for VMs and containers).
992 Slices and scopes have been introduced primarily in
993 context of the work to move cgroup handling to a
994 single-writer scheme, where only PID 1
995 creates/removes/manages cgroups.
997 * There's a new concept of "transient" units. In contrast to
998 normal units these units are created via an API at runtime,
999 not from configuration from disk. More specifically this
1000 means it is now possible to run arbitrary programs as
1001 independent services, with all execution parameters passed
1002 in via bus APIs rather than read from disk. Transient units
1003 make systemd substantially more dynamic then it ever was,
1004 and useful as a general batch manager.
1006 * logind has been updated to make use of scope and slice units
1007 for managing user sessions. As a user logs in he will get
1008 his own private slice unit, to which all sessions are added
1009 as scope units. We also added support for automatically
1010 adding an instance of user@.service for the user into the
1011 slice. Effectively logind will no longer create cgroup
1012 hierarchies on its own now, it will defer entirely to PID 1
1013 for this by means of scope, service and slice units. Since
1014 user sessions this way become entities managed by PID 1
1015 the output of "systemctl" is now a lot more comprehensive.
1017 * A new mini-daemon "systemd-machined" has been added which
1018 may be used by virtualization managers to register local
1019 VMs/containers. nspawn has been updated accordingly, and
1020 libvirt will be updated shortly. machined will collect a bit
1021 of meta information about the VMs/containers, and assign
1022 them their own scope unit (see above). The collected
1023 meta-data is then made available via the "machinectl" tool,
1024 and exposed in "ps" and similar tools. machined/machinectl
1025 is compile-time optional.
1027 * As discussed earlier, the low-level cgroup configuration
1028 options ControlGroup=, ControlGroupModify=,
1029 ControlGroupPersistent=, ControlGroupAttribute= have been
1030 removed. Please use high-level attribute settings instead as
1031 well as slice units.
1033 * A new bus call SetUnitProperties() has been added to alter
1034 various runtime parameters of a unit. This is primarily
1035 useful to alter cgroup parameters dynamically in a nice way,
1036 but will be extended later on to make more properties
1037 modifiable at runtime. systemctl gained a new set-properties
1038 command that wraps this call.
1040 * A new tool "systemd-run" has been added which can be used to
1041 run arbitrary command lines as transient services or scopes,
1042 while configuring a number of settings via the command
1043 line. This tool is currently very basic, however already
1044 very useful. We plan to extend this tool to even allow
1045 queuing of execution jobs with time triggers from the
1046 command line, similar in fashion to "at".
1048 * nspawn will now inform the user explicitly that kernels with
1049 audit enabled break containers, and suggest the user to turn
1052 * Support for detecting the IMA and AppArmor security
1053 frameworks with ConditionSecurity= has been added.
1055 * journalctl gained a new "-k" switch for showing only kernel
1056 messages, mimicking dmesg output; in addition to "--user"
1057 and "--system" switches for showing only user's own logs
1060 * systemd-delta can now show information about drop-in
1061 snippets extending unit files.
1063 * libsystemd-bus has been substantially updated but is still
1064 not available as public API.
1066 * systemd will now look for the "debug" argument on the kernel
1067 command line and enable debug logging, similar to
1068 "systemd.log_level=debug" already did before.
1070 * "systemctl set-default", "systemctl get-default" has been
1071 added to configure the default.target symlink, which
1072 controls what to boot into by default.
1074 * "systemctl set-log-level" has been added as a convenient
1075 way to raise and lower systemd logging threshold.
1077 * "systemd-analyze plot" will now show the time the various
1078 generators needed for execution, as well as information
1079 about the unit file loading.
1081 * libsystemd-journal gained a new sd_journal_open_files() call
1082 for opening specific journal files. journactl also gained a
1083 new switch to expose this new functionality. Previously we
1084 only supported opening all files from a directory, or all
1085 files from the system, as opening individual files only is
1086 racy due to journal file rotation.
1088 * systemd gained the new DefaultEnvironment= setting in
1089 /etc/systemd/system.conf to set environment variables for
1092 * If a privileged process logs a journal message with the
1093 OBJECT_PID= field set, then journald will automatically
1094 augment this with additional OBJECT_UID=, OBJECT_GID=,
1095 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
1096 system services want to log events about specific client
1097 processes. journactl/systemctl has been updated to make use
1098 of this information if all log messages regarding a specific
1101 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
1102 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
1103 Reisner, David Coppa, David King, David Strauss, Eelco
1104 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
1105 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
1106 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
1107 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
1108 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
1109 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
1110 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
1111 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
1112 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
1113 Łukasz Stelmach, 장동준
1117 * The Python bindings gained some minimal support for the APIs
1118 exposed by libsystemd-logind.
1120 * ConditionSecurity= gained support for detecting SMACK. Since
1121 this condition already supports SELinux and AppArmor we only
1122 miss IMA for this. Patches welcome!
1124 Contributions from: Karol Lewandowski, Lennart Poettering,
1125 Zbigniew Jędrzejewski-Szmek
1129 * systemd-nspawn will now create /etc/resolv.conf if
1130 necessary, before bind-mounting the host's file onto it.
1132 * systemd-nspawn will now store meta information about a
1133 container on the container's cgroup as extended attribute
1134 fields, including the root directory.
1136 * The cgroup hierarchy has been reworked in many ways. All
1137 objects any of the components systemd creates in the cgroup
1138 tree are now suffixed. More specifically, user sessions are
1139 now placed in cgroups suffixed with ".session", users in
1140 cgroups suffixed with ".user", and nspawn containers in
1141 cgroups suffixed with ".nspawn". Furthermore, all cgroup
1142 names are now escaped in a simple scheme to avoid collision
1143 of userspace object names with kernel filenames. This work
1144 is preparation for making these objects relocatable in the
1145 cgroup tree, in order to allow easy resource partitioning of
1146 these objects without causing naming conflicts.
1148 * systemctl list-dependencies gained the new switches
1149 --plain, --reverse, --after and --before.
1151 * systemd-inhibit now shows the process name of processes that
1152 have taken an inhibitor lock.
1154 * nss-myhostname will now also resolve "localhost"
1155 implicitly. This makes /etc/hosts an optional file and
1156 nicely handles that on IPv6 ::1 maps to both "localhost" and
1159 * libsystemd-logind.so gained a new call
1160 sd_get_machine_names() to enumerate running containers and
1161 VMs (currently only supported by very new libvirt and
1162 nspawn). sd_login_monitor can now be used to watch
1163 VMs/containers coming and going.
1165 * .include is not allowed recursively anymore, and only in
1166 unit files. Usually it is better to use drop-in snippets in
1167 .d/*.conf anyway, as introduced with systemd 198.
1169 * systemd-analyze gained a new "critical-chain" command that
1170 determines the slowest chain of units run during system
1171 boot-up. It is very useful for tracking down where
1172 optimizing boot time is the most beneficial.
1174 * systemd will no longer allow manipulating service paths in
1175 the name=systemd:/system cgroup tree using ControlGroup= in
1176 units. (But is still fine with it in all other dirs.)
1178 * There's a new systemd-nspawn@.service service file that may
1179 be used to easily run nspawn containers as system
1180 services. With the container's root directory in
1181 /var/lib/container/foobar it is now sufficient to run
1182 "systemctl start systemd-nspawn@foobar.service" to boot it.
1184 * systemd-cgls gained a new parameter "--machine" to list only
1185 the processes within a certain container.
1187 * ConditionSecurity= now can check for "apparmor". We still
1188 are lacking checks for SMACK and IMA for this condition
1189 check though. Patches welcome!
1191 * A new configuration file /etc/systemd/sleep.conf has been
1192 added that may be used to configure which kernel operation
1193 systemd is supposed to execute when "suspend", "hibernate"
1194 or "hybrid-sleep" is requested. This makes the new kernel
1195 "freeze" state accessible to the user.
1197 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
1198 the passed argument if applicable.
1200 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
1201 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
1202 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
1203 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
1204 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
1205 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
1206 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
1211 * The output of 'systemctl list-jobs' got some polishing. The
1212 '--type=' argument may now be passed more than once. A new
1213 command 'systemctl list-sockets' has been added which shows
1214 a list of kernel sockets systemd is listening on with the
1215 socket units they belong to, plus the units these socket
1218 * The experimental libsystemd-bus library got substantial
1219 updates to work in conjunction with the (also experimental)
1220 kdbus kernel project. It works well enough to exchange
1221 messages with some sophistication. Note that kdbus is not
1222 ready yet, and the library is mostly an elaborate test case
1223 for now, and not installable.
1225 * systemd gained a new unit 'systemd-static-nodes.service'
1226 that generates static device nodes earlier during boot, and
1227 can run in conjunction with udev.
1229 * libsystemd-login gained a new call sd_pid_get_user_unit()
1230 to retrieve the user systemd unit a process is running
1231 in. This is useful for systems where systemd is used as
1234 * systemd-nspawn now places all containers in the new /machine
1235 top-level cgroup directory in the name=systemd
1236 hierarchy. libvirt will soon do the same, so that we get a
1237 uniform separation of /system, /user and /machine for system
1238 services, user processes and containers/virtual
1239 machines. This new cgroup hierarchy is also useful to stick
1240 stable names to specific container instances, which can be
1241 recognized later this way (this name may be controlled
1242 via systemd-nspawn's new -M switch). libsystemd-login also
1243 gained a new call sd_pid_get_machine_name() to retrieve the
1244 name of the container/VM a specific process belongs to.
1246 * bootchart can now store its data in the journal.
1248 * libsystemd-journal gained a new call
1249 sd_journal_add_conjunction() for AND expressions to the
1250 matching logic. This can be used to express more complex
1251 logical expressions.
1253 * journactl can now take multiple --unit= and --user-unit=
1256 * The cryptsetup logic now understands the "luks.key=" kernel
1257 command line switch for specifying a file to read the
1258 decryption key from. Also, if a configured key file is not
1259 found the tool will now automatically fall back to prompting
1262 * Python systemd.journal module was updated to wrap recently
1263 added functions from libsystemd-journal. The interface was
1264 changed to bring the low level interface in s.j._Reader
1265 closer to the C API, and the high level interface in
1266 s.j.Reader was updated to wrap and convert all data about
1269 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
1270 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
1271 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
1272 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
1273 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
1274 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1278 * journalctl --update-catalog now understands a new --root=
1279 option to operate on catalogs found in a different root
1282 * During shutdown after systemd has terminated all running
1283 services a final killing loop kills all remaining left-over
1284 processes. We will now print the name of these processes
1285 when we send SIGKILL to them, since this usually indicates a
1288 * If /etc/crypttab refers to password files stored on
1289 configured mount points automatic dependencies will now be
1290 generated to ensure the specific mount is established first
1291 before the key file is attempted to be read.
1293 * 'systemctl status' will now show information about the
1294 network sockets a socket unit is listening on.
1296 * 'systemctl status' will also shown information about any
1297 drop-in configuration file for units. (Drop-In configuration
1298 files in this context are files such as
1299 /etc/systemd/systemd/foobar.service.d/*.conf)
1301 * systemd-cgtop now optionally shows summed up CPU times of
1302 cgroups. Press '%' while running cgtop to switch between
1303 percentage and absolute mode. This is useful to determine
1304 which cgroups use up the most CPU time over the entire
1305 runtime of the system. systemd-cgtop has also been updated
1306 to be 'pipeable' for processing with further shell tools.
1308 * 'hostnamectl set-hostname' will now allow setting of FQDN
1311 * The formatting and parsing of time span values has been
1312 changed. The parser now understands fractional expressions
1313 such as "5.5h". The formatter will now output fractional
1314 expressions for all time spans under 1min, i.e. "5.123456s"
1315 rather than "5s 123ms 456us". For time spans under 1s
1316 millisecond values are shown, for those under 1ms
1317 microsecond values are shown. This should greatly improve
1318 all time-related output of systemd.
1320 * libsystemd-login and libsystemd-journal gained new
1321 functions for querying the poll() events mask and poll()
1322 timeout value for integration into arbitrary event
1325 * localectl gained the ability to list available X11 keymaps
1326 (models, layouts, variants, options).
1328 * 'systemd-analyze dot' gained the ability to filter for
1329 specific units via shell-style globs, to create smaller,
1330 more useful graphs. I.e. it's now possible to create simple
1331 graphs of all the dependencies between only target units, or
1332 of all units that Avahi has dependencies with.
1334 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
1335 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
1336 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
1337 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
1338 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
1339 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
1340 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
1344 * The boot-time readahead implementation for rotating media
1345 will now read the read-ahead data in multiple passes which
1346 consist of all read requests made in equidistant time
1347 intervals. This means instead of strictly reading read-ahead
1348 data in its physical order on disk we now try to find a
1349 middle ground between physical and access time order.
1351 * /etc/os-release files gained a new BUILD_ID= field for usage
1352 on operating systems that provide continuous builds of OS
1355 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
1356 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
1357 William Douglas, Zbigniew Jędrzejewski-Szmek
1361 * systemd-python gained an API exposing libsystemd-daemon.
1363 * The SMACK setup logic gained support for uploading CIPSO
1366 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
1367 ReadOnlyDirectories= and InaccessibleDirectories= has
1368 changed. The private /tmp and /var/tmp directories are now
1369 shared by all processes of a service (which means
1370 ExecStartPre= may now leave data in /tmp that ExecStart= of
1371 the same service can still access). When a service is
1372 stopped its temporary directories are immediately deleted
1373 (normal clean-up with tmpfiles is still done in addition to
1376 * By default, systemd will now set a couple of sysctl
1377 variables in the kernel: the safe sysrq options are turned
1378 on, IP route verification is turned on, and source routing
1379 disabled. The recently added hardlink and softlink
1380 protection of the kernel is turned on. These settings should
1381 be reasonably safe, and good defaults for all new systems.
1383 * The predictable network naming logic may now be turned off
1384 with a new kernel command line switch: net.ifnames=0.
1386 * A new libsystemd-bus module has been added that implements a
1387 pretty complete D-Bus client library. For details see:
1389 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
1391 * journald will now explicitly flush the journal files to disk
1392 at the latest 5min after each write. The file will then also
1393 be marked offline until the next write. This should increase
1394 reliability in case of a crash. The synchronization delay
1395 can be configured via SyncIntervalSec= in journald.conf.
1397 * There's a new remote-fs-setup.target unit that can be used
1398 to pull in specific services when at least one remote file
1399 system is to be mounted.
1401 * There are new targets timers.target and paths.target as
1402 canonical targets to pull user timer and path units in
1403 from. This complements sockets.target with a similar
1404 purpose for socket units.
1406 * libudev gained a new call udev_device_set_attribute_value()
1407 to set sysfs attributes of a device.
1409 * The udev daemon now sets the default number of worker
1410 processes executed in parallel based on the number of available
1411 CPUs instead of the amount of available RAM. This is supposed
1412 to provide a more reliable default and limit a too aggressive
1413 paralellism for setups with 1000s of devices connected.
1415 Contributions from: Auke Kok, Colin Walters, Cristian
1416 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
1417 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
1418 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
1419 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
1420 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
1421 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
1422 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
1423 Zbigniew Jędrzejewski-Szmek
1427 * Configuration of unit files may now be extended via drop-in
1428 files without having to edit/override the unit files
1429 themselves. More specifically, if the administrator wants to
1430 change one value for a service file foobar.service he can
1431 now do so by dropping in a configuration snippet into
1432 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
1433 will load all these snippets and apply them on top of the
1434 main unit configuration file, possibly extending or
1435 overriding its settings. Using these drop-in snippets is
1436 generally nicer than the two earlier options for changing
1437 unit files locally: copying the files from
1438 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
1439 them there; or creating a new file in /etc/systemd/system/
1440 that incorporates the original one via ".include". Drop-in
1441 snippets into these .d/ directories can be placed in any
1442 directory systemd looks for units in, and the usual
1443 overriding semantics between /usr/lib, /etc and /run apply
1446 * Most unit file settings which take lists of items can now be
1447 reset by assigning the empty string to them. For example,
1448 normally, settings such as Environment=FOO=BAR append a new
1449 environment variable assignment to the environment block,
1450 each time they are used. By assigning Environment= the empty
1451 string the environment block can be reset to empty. This is
1452 particularly useful with the .d/*.conf drop-in snippets
1453 mentioned above, since this adds the ability to reset list
1454 settings from vendor unit files via these drop-ins.
1456 * systemctl gained a new "list-dependencies" command for
1457 listing the dependencies of a unit recursively.
1459 * Inhibitors are now honored and listed by "systemctl
1460 suspend", "systemctl poweroff" (and similar) too, not only
1461 GNOME. These commands will also list active sessions by
1464 * Resource limits (as exposed by the various control group
1465 controllers) can now be controlled dynamically at runtime
1466 for all units. More specifically, you can now use a command
1467 like "systemctl set-cgroup-attr foobar.service cpu.shares
1468 2000" to alter the CPU shares a specific service gets. These
1469 settings are stored persistently on disk, and thus allow the
1470 administrator to easily adjust the resource usage of
1471 services with a few simple commands. This dynamic resource
1472 management logic is also available to other programs via the
1473 bus. Almost any kernel cgroup attribute and controller is
1476 * systemd-vconsole-setup will now copy all font settings to
1477 all allocated VTs, where it previously applied them only to
1480 * libsystemd-login gained the new sd_session_get_tty() API
1483 * This release drops support for a few legacy or
1484 distribution-specific LSB facility names when parsing init
1485 scripts: $x-display-manager, $mail-transfer-agent,
1486 $mail-transport-agent, $mail-transfer-agent, $smtp,
1487 $null. Also, the mail-transfer-agent.target unit backing
1488 this has been removed. Distributions which want to retain
1489 compatibility with this should carry the burden for
1490 supporting this themselves and patch support for these back
1491 in, if they really need to. Also, the facilities $syslog and
1492 $local_fs are now ignored, since systemd does not support
1493 early-boot LSB init scripts anymore, and these facilities
1494 are implied anyway for normal services. syslog.target has
1497 * There are new bus calls on PID1's Manager object for
1498 cancelling jobs, and removing snapshot units. Previously,
1499 both calls were only available on the Job and Snapshot
1502 * systemd-journal-gatewayd gained SSL support.
1504 * The various "environment" files, such as /etc/locale.conf
1505 now support continuation lines with a backslash ("\") as
1506 last character in the line, similar in style (but different)
1507 to how this is supported in shells.
1509 * For normal user processes the _SYSTEMD_USER_UNIT= field is
1510 now implicitly appended to every log entry logged. systemctl
1511 has been updated to filter by this field when operating on a
1512 user systemd instance.
1514 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
1515 CAP_AUDIT_CONTROL capabilities to the capabilities set for
1516 the container. This makes it easier to boot unmodified
1517 Fedora systems in a container, which however still requires
1518 audit=0 to be passed on the kernel command line. Auditing in
1519 kernel and userspace is unfortunately still too broken in
1520 context of containers, hence we recommend compiling it out
1521 of the kernel or using audit=0. Hopefully this will be fixed
1522 one day for good in the kernel.
1524 * nspawn gained the new --bind= and --bind-ro= parameters to
1525 bind mount specific directories from the host into the
1528 * nspawn will now mount its own devpts file system instance
1529 into the container, in order not to leak pty devices from
1530 the host into the container.
1532 * systemd will now read the firmware boot time performance
1533 information from the EFI variables, if the used boot loader
1534 supports this, and takes it into account for boot performance
1535 analysis via "systemd-analyze". This is currently supported
1536 only in conjunction with Gummiboot, but could be supported
1537 by other boot loaders too. For details see:
1539 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
1541 * A new generator has been added that automatically mounts the
1542 EFI System Partition (ESP) to /boot, if that directory
1543 exists, is empty, and no other file system has been
1544 configured to be mounted there.
1546 * logind will now send out PrepareForSleep(false) out
1547 unconditionally, after coming back from suspend. This may be
1548 used by applications as asynchronous notification for
1549 system resume events.
1551 * "systemctl unlock-sessions" has been added, that allows
1552 unlocking the screens of all user sessions at once, similar
1553 how "systemctl lock-sessions" already locked all users
1554 sessions. This is backed by a new D-Bus call UnlockSessions().
1556 * "loginctl seat-status" will now show the master device of a
1557 seat. (i.e. the device of a seat that needs to be around for
1558 the seat to be considered available, usually the graphics
1561 * tmpfiles gained a new "X" line type, that allows
1562 configuration of files and directories (with wildcards) that
1563 shall be excluded from automatic cleanup ("aging").
1565 * udev default rules set the device node permissions now only
1566 at "add" events, and do not change them any longer with a
1567 later "change" event.
1569 * The log messages for lid events and power/sleep keypresses
1570 now carry a message ID.
1572 * We now have a substantially larger unit test suite, but this
1573 continues to be work in progress.
1575 * udevadm hwdb gained a new --root= parameter to change the
1576 root directory to operate relative to.
1578 * logind will now issue a background sync() request to the kernel
1579 early at shutdown, so that dirty buffers are flushed to disk early
1580 instead of at the last moment, in order to optimize shutdown
1583 * A new bootctl tool has been added that is an interface for
1584 certain boot loader operations. This is currently a preview
1585 and is likely to be extended into a small mechanism daemon
1586 like timedated, localed, hostnamed, and can be used by
1587 graphical UIs to enumerate available boot options, and
1588 request boot into firmware operations.
1590 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
1591 the rest of the package. It also has been updated to work
1592 correctly in initrds.
1594 * Policykit previously has been runtime optional, and is now
1595 also compile time optional via a configure switch.
1597 * systemd-analyze has been reimplemented in C. Also "systemctl
1598 dot" has moved into systemd-analyze.
1600 * "systemctl status" with no further parameters will now print
1601 the status of all active or failed units.
1603 * Operations such as "systemctl start" can now be executed
1604 with a new mode "--irreversible" which may be used to queue
1605 operations that cannot accidentally be reversed by a later
1606 job queuing. This is by default used to make shutdown
1607 requests more robust.
1609 * The Python API of systemd now gained a new module for
1610 reading journal files.
1612 * A new tool kernel-install has been added that can install
1613 kernel images according to the Boot Loader Specification:
1615 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
1617 * Boot time console output has been improved to provide
1618 animated boot time output for hanging jobs.
1620 * A new tool systemd-activate has been added which can be used
1621 to test socket activation with, directly from the command
1622 line. This should make it much easier to test and debug
1623 socket activation in daemons.
1625 * journalctl gained a new "--reverse" (or -r) option to show
1626 journal output in reverse order (i.e. newest line first).
1628 * journalctl gained a new "--pager-end" (or -e) option to jump
1629 to immediately jump to the end of the journal in the
1630 pager. This is only supported in conjunction with "less".
1632 * journalctl gained a new "--user-unit=" option, that works
1633 similar to "--unit=" but filters for user units rather than
1636 * A number of unit files to ease adoption of systemd in
1637 initrds has been added. This moves some minimal logic from
1638 the various initrd implementations into systemd proper.
1640 * The journal files are now owned by a new group
1641 "systemd-journal", which exists specifically to allow access
1642 to the journal, and nothing else. Previously, we used the
1643 "adm" group for that, which however possibly covers more
1644 than just journal/log file access. This new group is now
1645 already used by systemd-journal-gatewayd to ensure this
1646 daemon gets access to the journal files and as little else
1647 as possible. Note that "make install" will also set FS ACLs
1648 up for /var/log/journal to give "adm" and "wheel" read
1649 access to it, in addition to "systemd-journal" which owns
1650 the journal files. We recommend that packaging scripts also
1651 add read access to "adm" + "wheel" to /var/log/journal, and
1652 all existing/future journal files. To normal users and
1653 administrators little changes, however packagers need to
1654 ensure to create the "systemd-journal" system group at
1655 package installation time.
1657 * The systemd-journal-gatewayd now runs as unprivileged user
1658 systemd-journal-gateway:systemd-journal-gateway. Packaging
1659 scripts need to create these system user/group at
1662 * timedated now exposes a new boolean property CanNTP that
1663 indicates whether a local NTP service is available or not.
1665 * systemd-detect-virt will now also detect xen PVs
1667 * The pstore file system is now mounted by default, if it is
1670 * In addition to the SELinux and IMA policies we will now also
1671 load SMACK policies at early boot.
1673 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
1674 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
1675 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
1676 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
1677 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
1678 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
1679 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
1680 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
1681 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
1682 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
1683 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
1684 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
1685 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
1686 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
1690 * Timer units now support calendar time events in addition to
1691 monotonic time events. That means you can now trigger a unit
1692 based on a calendar time specification such as "Thu,Fri
1693 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
1694 or fifth day of any month of the year 2013, given that it is
1695 a thursday or friday. This brings timer event support
1696 considerably closer to cron's capabilities. For details on
1697 the supported calendar time specification language see
1700 * udev now supports a number of different naming policies for
1701 network interfaces for predictable names, and a combination
1702 of these policies is now the default. Please see this wiki
1703 document for details:
1705 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
1707 * Auke Kok's bootchart implementation has been added to the
1708 systemd tree. It's an optional component that can graph the
1709 boot in quite some detail. It's one of the best bootchart
1710 implementations around and minimal in its code and
1713 * nss-myhostname has been integrated into the systemd source
1714 tree. nss-myhostname guarantees that the local hostname
1715 always stays resolvable via NSS. It has been a weak
1716 requirement of systemd-hostnamed since a long time, and
1717 since its code is actually trivial we decided to just
1718 include it in systemd's source tree. It can be turned off
1719 with a configure switch.
1721 * The read-ahead logic is now capable of properly detecting
1722 whether a btrfs file system is on SSD or rotating media, in
1723 order to optimize the read-ahead scheme. Previously, it was
1724 only capable of detecting this on traditional file systems
1727 * In udev, additional device properties are now read from the
1728 IAB in addition to the OUI database. Also, Bluetooth company
1729 identities are attached to the devices as well.
1731 * In service files %U may be used as specifier that is
1732 replaced by the configured user name of the service.
1734 * nspawn may now be invoked without a controlling TTY. This
1735 makes it suitable for invocation as its own service. This
1736 may be used to set up a simple containerized server system
1737 using only core OS tools.
1739 * systemd and nspawn can now accept socket file descriptors
1740 when they are started for socket activation. This enables
1741 implementation of socket activated nspawn
1742 containers. i.e. think about autospawning an entire OS image
1743 when the first SSH or HTTP connection is received. We expect
1744 that similar functionality will also be added to libvirt-lxc
1747 * journalctl will now suppress ANSI color codes when
1748 presenting log data.
1750 * systemctl will no longer show control group information for
1751 a unit if a the control group is empty anyway.
1753 * logind can now automatically suspend/hibernate/shutdown the
1756 * /etc/machine-info and hostnamed now also expose the chassis
1757 type of the system. This can be used to determine whether
1758 the local system is a laptop, desktop, handset or
1759 tablet. This information may either be configured by the
1760 user/vendor or is automatically determined from ACPI and DMI
1761 information if possible.
1763 * A number of PolicyKit actions are now bound together with
1764 "imply" rules. This should simplify creating UIs because
1765 many actions will now authenticate similar ones as well.
1767 * Unit files learnt a new condition ConditionACPower= which
1768 may be used to conditionalize a unit depending on whether an
1769 AC power source is connected or not, of whether the system
1770 is running on battery power.
1772 * systemctl gained a new "is-failed" verb that may be used in
1773 shell scripts and suchlike to check whether a specific unit
1774 is in the "failed" state.
1776 * The EnvironmentFile= setting in unit files now supports file
1777 globbing, and can hence be used to easily read a number of
1778 environment files at once.
1780 * systemd will no longer detect and recognize specific
1781 distributions. All distribution-specific #ifdeffery has been
1782 removed, systemd is now fully generic and
1783 distribution-agnostic. Effectively, not too much is lost as
1784 a lot of the code is still accessible via explicit configure
1785 switches. However, support for some distribution specific
1786 legacy configuration file formats has been dropped. We
1787 recommend distributions to simply adopt the configuration
1788 files everybody else uses now and convert the old
1789 configuration from packaging scripts. Most distributions
1790 already did that. If that's not possible or desirable,
1791 distributions are welcome to forward port the specific
1792 pieces of code locally from the git history.
1794 * When logging a message about a unit systemd will now always
1795 log the unit name in the message meta data.
1797 * localectl will now also discover system locale data that is
1798 not stored in locale archives, but directly unpacked.
1800 * logind will no longer unconditionally use framebuffer
1801 devices as seat masters, i.e. as devices that are required
1802 to be existing before a seat is considered preset. Instead,
1803 it will now look for all devices that are tagged as
1804 "seat-master" in udev. By default framebuffer devices will
1805 be marked as such, but depending on local systems other
1806 devices might be marked as well. This may be used to
1807 integrate graphics cards using closed source drivers (such
1808 as NVidia ones) more nicely into logind. Note however, that
1809 we recommend using the open source NVidia drivers instead,
1810 and no udev rules for the closed-source drivers will be
1811 shipped from us upstream.
1813 Contributions from: Adam Williamson, Alessandro Crismani, Auke
1814 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
1815 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
1816 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
1817 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
1818 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
1819 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
1820 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
1821 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
1822 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
1823 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
1824 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
1829 * udev gained support for loading additional device properties
1830 from an indexed database that is keyed by vendor/product IDs
1831 and similar device identifiers. For the beginning this
1832 "hwdb" is populated with data from the well-known PCI and
1833 USB database, but also includes PNP, ACPI and OID data. In
1834 the longer run this indexed database shall grow into
1835 becoming the one central database for non-essential
1836 userspace device metadata. Previously, data from the PCI/USB
1837 database was only attached to select devices, since the
1838 lookup was a relatively expensive operation due to O(n) time
1839 complexity (with n being the number of entries in the
1840 database). Since this is now O(1), we decided to add in this
1841 data for all devices where this is available, by
1842 default. Note that the indexed database needs to be rebuilt
1843 when new data files are installed. To achieve this you need
1844 to update your packaging scripts to invoke "udevadm hwdb
1845 --update" after installation of hwdb data files. For
1846 RPM-based distributions we introduced the new
1847 %udev_hwdb_update macro for this purpose.
1849 * The Journal gained support for the "Message Catalog", an
1850 indexed database to link up additional information with
1851 journal entries. For further details please check:
1853 http://www.freedesktop.org/wiki/Software/systemd/catalog
1855 The indexed message catalog database also needs to be
1856 rebuilt after installation of message catalog files. Use
1857 "journalctl --update-catalog" for this. For RPM-based
1858 distributions we introduced the %journal_catalog_update
1859 macro for this purpose.
1861 * The Python Journal bindings gained support for the standard
1862 Python logging framework.
1864 * The Journal API gained new functions for checking whether
1865 the underlying file system of a journal file is capable of
1866 properly reporting file change notifications, or whether
1867 applications that want to reflect journal changes "live"
1868 need to recheck journal files continuously in appropriate
1871 * It is now possible to set the "age" field for tmpfiles
1872 entries to 0, indicating that files matching this entry
1873 shall always be removed when the directories are cleaned up.
1875 * coredumpctl gained a new "gdb" verb which invokes gdb
1876 right-away on the selected coredump.
1878 * There's now support for "hybrid sleep" on kernels that
1879 support this, in addition to "suspend" and "hibernate". Use
1880 "systemctl hybrid-sleep" to make use of this.
1882 * logind's HandleSuspendKey= setting (and related settings)
1883 now gained support for a new "lock" setting to simply
1884 request the screen lock on all local sessions, instead of
1885 actually executing a suspend or hibernation.
1887 * systemd will now mount the EFI variables file system by
1890 * Socket units now gained support for configuration of the
1891 SMACK security label.
1893 * timedatectl will now output the time of the last and next
1894 daylight saving change.
1896 * We dropped support for various legacy and distro-specific
1897 concepts, such as insserv, early-boot SysV services
1898 (i.e. those for non-standard runlevels such as 'b' or 'S')
1899 or ArchLinux /etc/rc.conf support. We recommend the
1900 distributions who still need support this to either continue
1901 to maintain the necessary patches downstream, or find a
1902 different solution. (Talk to us if you have questions!)
1904 * Various systemd components will now bypass PolicyKit checks
1905 for root and otherwise handle properly if PolicyKit is not
1906 found to be around. This should fix most issues for
1907 PolicyKit-less systems. Quite frankly this should have been
1908 this way since day one. It is absolutely our intention to
1909 make systemd work fine on PolicyKit-less systems, and we
1910 consider it a bug if something doesn't work as it should if
1911 PolicyKit is not around.
1913 * For embedded systems it is now possible to build udev and
1914 systemd without blkid and/or kmod support.
1916 * "systemctl switch-root" is now capable of switching root
1917 more than once. I.e. in addition to transitions from the
1918 initrd to the host OS it is now possible to transition to
1919 further OS images from the host. This is useful to implement
1920 offline updating tools.
1922 * Various other additions have been made to the RPM macros
1923 shipped with systemd. Use %udev_rules_update() after
1924 installing new udev rules files. %_udevhwdbdir,
1925 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
1926 %_sysctldir are now available which resolve to the right
1927 directories for packages to place various data files in.
1929 * journalctl gained the new --full switch (in addition to
1930 --all, to disable ellipsation for long messages.
1932 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
1933 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
1934 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
1935 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
1936 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
1937 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
1938 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
1939 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
1940 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
1944 * journalctl gained new --since= and --until= switches to
1945 filter by time. It also now supports nice filtering for
1946 units via --unit=/-u.
1948 * Type=oneshot services may use ExecReload= and do the
1951 * The journal daemon now supports time-based rotation and
1952 vacuuming, in addition to the usual disk-space based
1955 * The journal will now index the available field values for
1956 each field name. This enables clients to show pretty drop
1957 downs of available match values when filtering. The bash
1958 completion of journalctl has been updated
1959 accordingly. journalctl gained a new switch -F to list all
1960 values a certain field takes in the journal database.
1962 * More service events are now written as structured messages
1963 to the journal, and made recognizable via message IDs.
1965 * The timedated, localed and hostnamed mini-services which
1966 previously only provided support for changing time, locale
1967 and hostname settings from graphical DEs such as GNOME now
1968 also have a minimal (but very useful) text-based client
1969 utility each. This is probably the nicest way to changing
1970 these settings from the command line now, especially since
1971 it lists available options and is fully integrated with bash
1974 * There's now a new tool "systemd-coredumpctl" to list and
1975 extract coredumps from the journal.
1977 * We now install a README each in /var/log/ and
1978 /etc/rc.d/init.d explaining where the system logs and init
1979 scripts went. This hopefully should help folks who go to
1980 that dirs and look into the otherwise now empty void and
1981 scratch their heads.
1983 * When user-services are invoked (by systemd --user) the
1984 $MANAGERPID env var is set to the PID of systemd.
1986 * SIGRTMIN+24 when sent to a --user instance will now result
1987 in immediate termination of systemd.
1989 * gatewayd received numerous feature additions such as a
1990 "follow" mode, for live syncing and filtering.
1992 * browse.html now allows filtering and showing detailed
1993 information on specific entries. Keyboard navigation and
1994 mouse screen support has been added.
1996 * gatewayd/journalctl now supports HTML5/JSON
1997 Server-Sent-Events as output.
1999 * The SysV init script compatibility logic will now
2000 heuristically determine whether a script supports the
2001 "reload" verb, and only then make this available as
2004 * "systemctl status --follow" has been removed, use "journalctl
2007 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
2008 have been removed since they are hardly useful to be
2011 * And I'd like to take the opportunity to specifically mention
2012 Zbigniew for his great contributions. Zbigniew, you rock!
2014 Contributions from: Andrew Eikum, Christian Hesse, Colin
2015 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
2016 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
2017 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
2018 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
2019 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
2020 Jędrzejewski-Szmek, Сковорода Никита Андреевич
2024 * If /etc/vconsole.conf is non-existent or empty we will no
2025 longer load any console font or key map at boot by
2026 default. Instead the kernel defaults will be left
2027 intact. This is definitely the right thing to do, as no
2028 configuration should mean no configuration, and hard-coding
2029 font names that are different on all archs is probably a bad
2030 idea. Also, the kernel default key map and font should be
2031 good enough for most cases anyway, and mostly identical to
2032 the userspace fonts/key maps we previously overloaded them
2033 with. If distributions want to continue to default to a
2034 non-kernel font or key map they should ship a default
2035 /etc/vconsole.conf with the appropriate contents.
2037 Contributions from: Colin Walters, Daniel J Walsh, Dave
2038 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
2039 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
2043 * journalctl gained a new --cursor= switch to show entries
2044 starting from the specified location in the journal.
2046 * We now enforce a size limit on journal entry fields exported
2047 with "-o json" in journalctl. Fields larger than 4K will be
2048 assigned null. This can be turned off with --all.
2050 * An (optional) journal gateway daemon is now available as
2051 "systemd-journal-gatewayd.service". This service provides
2052 access to the journal via HTTP and JSON. This functionality
2053 will be used to implement live log synchronization in both
2054 pull and push modes, but has various other users too, such
2055 as easy log access for debugging of embedded devices. Right
2056 now it is already useful to retrieve the journal via HTTP:
2058 # systemctl start systemd-journal-gatewayd.service
2059 # wget http://localhost:19531/entries
2061 This will download the journal contents in a
2062 /var/log/messages compatible format. The same as JSON:
2064 # curl -H"Accept: application/json" http://localhost:19531/entries
2066 This service is also accessible via a web browser where a
2067 single static HTML5 app is served that uses the JSON logic
2068 to enable the user to do some basic browsing of the
2069 journal. This will be extended later on. Here's an example
2070 screenshot of this app in its current state:
2072 http://0pointer.de/public/journal-gatewayd
2074 Contributions from: Kay Sievers, Lennart Poettering, Robert
2075 Milasan, Tom Gundersen
2079 * The bash completion logic is now available for journalctl
2082 * We don't mount the "cpuset" controller anymore together with
2083 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
2084 started if no parameters are assigned to it. "cpuset" hence
2085 broke code that assumed it it could create "cpu" groups and
2088 * journalctl -f will now subscribe to terminal size changes,
2089 and line break accordingly.
2091 Contributions from: Dave Reisner, Kay Sievers, Lennart
2092 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
2096 * nspawn will now create a symlink /etc/localtime in the
2097 container environment, copying the host's timezone
2098 setting. Previously this has been done via a bind mount, but
2099 since symlinks cannot be bind mounted this has now been
2100 changed to create/update the appropriate symlink.
2102 * journalctl -n's line number argument is now optional, and
2103 will default to 10 if omitted.
2105 * journald will now log the maximum size the journal files may
2106 take up on disk. This is particularly useful if the default
2107 built-in logic of determining this parameter from the file
2108 system size is used. Use "systemctl status
2109 systemd-journald.service" to see this information.
2111 * The multi-seat X wrapper tool has been stripped down. As X
2112 is now capable of enumerating graphics devices via udev in a
2113 seat-aware way the wrapper is not strictly necessary
2114 anymore. A stripped down temporary stop-gap is still shipped
2115 until the upstream display managers have been updated to
2116 fully support the new X logic. Expect this wrapper to be
2117 removed entirely in one of the next releases.
2119 * HandleSleepKey= in logind.conf has been split up into
2120 HandleSuspendKey= and HandleHibernateKey=. The old setting
2121 is not available anymore. X11 and the kernel are
2122 distuingishing between these keys and we should too. This
2123 also means the inhibition lock for these keys has been split
2126 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
2127 Poettering, Lukas Nykryn, Václav Pavlín
2131 * Whenever a unit changes state we'll now log this to the
2132 journal and show along the unit's own log output in
2135 * ConditionPathIsMountPoint= can now properly detect bind
2136 mount points too. (Previously, a bind mount of one file
2137 system to another place in the same file system could not be
2138 detected as mount, since they shared struct stat's st_dev
2141 * We will now mount the cgroup controllers cpu, cpuacct,
2142 cpuset and the controllers net_cls, net_prio together by
2145 * nspawn containers will now have a virtualized boot
2146 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
2147 over with a randomized ID at container initialization). This
2148 has the effect of making "journalctl -b" do the right thing
2151 * The JSON output journal serialization has been updated not
2152 to generate "endless" list objects anymore, but rather one
2153 JSON object per line. This is more in line how most JSON
2154 parsers expect JSON objects. The new output mode
2155 "json-pretty" has been added to provide similar output, but
2156 neatly aligned for readability by humans.
2158 * We dropped all explicit sync() invocations in the shutdown
2159 code. The kernel does this implicitly anyway in the kernel
2160 reboot() syscall. halt(8)'s -n option is now a compatibility
2163 * We now support virtualized reboot() in containers, as
2164 supported by newer kernels. We will fall back to exit() if
2165 CAP_SYS_REBOOT is not available to the container. Also,
2166 nspawn makes use of this now and will actually reboot the
2167 container if the containerized OS asks for that.
2169 * journalctl will only show local log output by default
2170 now. Use --merge (-m) to show remote log output, too.
2172 * libsystemd-journal gained the new sd_journal_get_usage()
2173 call to determine the current disk usage of all journal
2174 files. This is exposed in the new "journalctl --disk-usage"
2177 * journald gained a new configuration setting SplitMode= in
2178 journald.conf which may be used to control how user journals
2179 are split off. See journald.conf(5) for details.
2181 * A new condition type ConditionFileNotEmpty= has been added.
2183 * tmpfiles' "w" lines now support file globbing, to write
2184 multiple files at once.
2186 * We added Python bindings for the journal submission
2187 APIs. More Python APIs for a number of selected APIs will
2188 likely follow. Note that we intend to add native bindings
2189 only for the Python language, as we consider it common
2190 enough to deserve bindings shipped within systemd. There are
2191 various projects outside of systemd that provide bindings
2192 for languages such as PHP or Lua.
2194 * Many conditions will now resolve specifiers such as %i. In
2195 addition, PathChanged= and related directives of .path units
2196 now support specifiers as well.
2198 * There's now a new RPM macro definition for the system preset
2201 * journald will now warn if it can't forward a message to the
2202 syslog daemon because it's socket is full.
2204 * timedated will no longer write or process /etc/timezone,
2205 except on Debian. As we do not support late mounted /usr
2206 anymore /etc/localtime always being a symlink is now safe,
2207 and hence the information in /etc/timezone is not necessary
2210 * logind will now always reserve one VT for a text getty (VT6
2211 by default). Previously if more than 6 X sessions where
2212 started they took up all the VTs with auto-spawned gettys,
2213 so that no text gettys were available anymore.
2215 * udev will now automatically inform the btrfs kernel logic
2216 about btrfs RAID components showing up. This should make
2217 simple hotplug based btrfs RAID assembly work.
2219 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
2220 (but not for its children which will stay at the kernel
2221 default). This should allow setups with a lot more listening
2224 * systemd will now always pass the configured timezone to the
2225 kernel at boot. timedated will do the same when the timezone
2228 * logind's inhibition logic has been updated. By default,
2229 logind will now handle the lid switch, the power and sleep
2230 keys all the time, even in graphical sessions. If DEs want
2231 to handle these events on their own they should take the new
2232 handle-power-key, handle-sleep-key and handle-lid-switch
2233 inhibitors during their runtime. A simple way to achiveve
2234 that is to invoke the DE wrapped in an invocation of:
2236 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
2238 * Access to unit operations is now checked via SELinux taking
2239 the unit file label and client process label into account.
2241 * systemd will now notify the administrator in the journal
2242 when he over-mounts a non-empty directory.
2244 * There are new specifiers that are resolved in unit files,
2245 for the host name (%H), the machine ID (%m) and the boot ID
2248 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
2249 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
2250 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
2251 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
2252 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
2253 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
2254 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
2258 * Support for reading structured kernel messages from
2259 /dev/kmsg has now been added and is enabled by default.
2261 * Support for reading kernel messages from /proc/kmsg has now
2262 been removed. If you want kernel messages in the journal
2263 make sure to run a recent kernel (>= 3.5) that supports
2264 reading structured messages from /dev/kmsg (see
2265 above). /proc/kmsg is now exclusive property of classic
2266 syslog daemons again.
2268 * The libudev API gained the new
2269 udev_device_new_from_device_id() call.
2271 * The logic for file system namespace (ReadOnlyDirectory=,
2272 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
2273 require pivot_root() anymore. This means fewer temporary
2274 directories are created below /tmp for this feature.
2276 * nspawn containers will now see and receive all submounts
2277 made on the host OS below the root file system of the
2280 * Forward Secure Sealing is now supported for Journal files,
2281 which provide cryptographical sealing of journal files so
2282 that attackers cannot alter log history anymore without this
2283 being detectable. Lennart will soon post a blog story about
2284 this explaining it in more detail.
2286 * There are two new service settings RestartPreventExitStatus=
2287 and SuccessExitStatus= which allow configuration of exit
2288 status (exit code or signal) which will be excepted from the
2289 restart logic, resp. consider successful.
2291 * journalctl gained the new --verify switch that can be used
2292 to check the integrity of the structure of journal files and
2293 (if Forward Secure Sealing is enabled) the contents of
2296 * nspawn containers will now be run with /dev/stdin, /dev/fd/
2297 and similar symlinks pre-created. This makes running shells
2298 as container init process a lot more fun.
2300 * The fstab support can now handle PARTUUID= and PARTLABEL=
2303 * A new ConditionHost= condition has been added to match
2304 against the hostname (with globs) and machine ID. This is
2305 useful for clusters where a single OS image is used to
2306 provision a large number of hosts which shall run slightly
2307 different sets of services.
2309 * Services which hit the restart limit will now be placed in a
2312 Contributions from: Bertram Poettering, Dave Reisner, Huang
2313 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
2314 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
2318 * When running in --user mode systemd will now become a
2319 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
2320 tree a lot more organized.
2322 * A new PartOf= unit dependency type has been introduced that
2323 may be used to group services in a natural way.
2325 * "systemctl enable" may now be used to enable instances of
2328 * journalctl now prints error log levels in red, and
2329 warning/notice log levels in bright white. It also supports
2330 filtering by log level now.
2332 * cgtop gained a new -n switch (similar to top), to configure
2333 the maximum number of iterations to run for. It also gained
2334 -b, to run in batch mode (accepting no input).
2336 * The suffix ".service" may now be omitted on most systemctl
2337 command lines involving service unit names.
2339 * There's a new bus call in logind to lock all sessions, as
2340 well as a loginctl verb for it "lock-sessions".
2342 * libsystemd-logind.so gained a new call sd_journal_perror()
2343 that works similar to libc perror() but logs to the journal
2344 and encodes structured information about the error number.
2346 * /etc/crypttab entries now understand the new keyfile-size=
2349 * shutdown(8) now can send a (configurable) wall message when
2350 a shutdown is cancelled.
2352 * The mount propagation mode for the root file system will now
2353 default to "shared", which is useful to make containers work
2354 nicely out-of-the-box so that they receive new mounts from
2355 the host. This can be undone locally by running "mount
2356 --make-rprivate /" if needed.
2358 * The prefdm.service file has been removed. Distributions
2359 should maintain this unit downstream if they intend to keep
2360 it around. However, we recommend writing normal unit files
2361 for display managers instead.
2363 * Since systemd is a crucial part of the OS we will now
2364 default to a number of compiler switches that improve
2365 security (hardening) such as read-only relocations, stack
2366 protection, and suchlike.
2368 * The TimeoutSec= setting for services is now split into
2369 TimeoutStartSec= and TimeoutStopSec= to allow configuration
2370 of individual time outs for the start and the stop phase of
2373 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
2374 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
2375 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
2376 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
2377 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
2378 Gundersen, Zbigniew Jędrzejewski-Szmek
2382 * The journal and id128 C APIs are now fully documented as man
2385 * Extra safety checks have been added when transitioning from
2386 the initial RAM disk to the main system to avoid accidental
2389 * /etc/crypttab entries now understand the new keyfile-offset=
2392 * systemctl -t can now be used to filter by unit load state.
2394 * The journal C API gained the new sd_journal_wait() call to
2395 make writing synchronous journal clients easier.
2397 * journalctl gained the new -D switch to show journals from a
2400 * journalctl now displays a special marker between log
2401 messages of two different boots.
2403 * The journal is now explicitly flushed to /var via a service
2404 systemd-journal-flush.service, rather than implicitly simply
2405 by seeing /var/log/journal to be writable.
2407 * journalctl (and the journal C APIs) can now match for much
2408 more complex expressions, with alternatives and
2411 * When transitioning from the initial RAM disk to the main
2412 system we will now kill all processes in a killing spree to
2413 ensure no processes stay around by accident.
2415 * Three new specifiers may be used in unit files: %u, %h, %s
2416 resolve to the user name, user home directory resp. user
2417 shell. This is useful for running systemd user instances.
2419 * We now automatically rotate journal files if their data
2420 object hash table gets a fill level > 75%. We also size the
2421 hash table based on the configured maximum file size. This
2422 together should lower hash collisions drastically and thus
2423 speed things up a bit.
2425 * journalctl gained the new "--header" switch to introspect
2426 header data of journal files.
2428 * A new setting SystemCallFilters= has been added to services
2429 which may be used to apply blacklists or whitelists to
2430 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
2432 * nspawn gained a new --link-journal= switch (and quicker: -j)
2433 to link the container journal with the host. This makes it
2434 very easy to centralize log viewing on the host for all
2435 guests while still keeping the journal files separated.
2437 * Many bugfixes and optimizations
2439 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
2440 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
2441 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
2446 * Several tools now understand kernel command line arguments,
2447 which are only read when run in an initial RAM disk. They
2448 usually follow closely their normal counterparts, but are
2451 * There's a new tool to analyze the readahead files that are
2452 automatically generated at boot. Use:
2454 /usr/lib/systemd/systemd-readahead analyze /.readahead
2456 * We now provide an early debug shell on tty9 if this enabled. Use:
2458 systemctl enable debug-shell.service
2460 * All plymouth related units have been moved into the Plymouth
2461 package. Please make sure to upgrade your Plymouth version
2464 * systemd-tmpfiles now supports getting passed the basename of
2465 a configuration file only, in which case it will look for it
2466 in all appropriate directories automatically.
2468 * udevadm info now takes a /dev or /sys path as argument, and
2469 does the right thing. Example:
2471 udevadm info /dev/sda
2472 udevadm info /sys/class/block/sda
2474 * systemctl now prints a warning if a unit is stopped but a
2475 unit that might trigger it continues to run. Example: a
2476 service is stopped but the socket that activates it is left
2479 * "systemctl status" will now mention if the log output was
2480 shortened due to rotation since a service has been started.
2482 * The journal API now exposes functions to determine the
2483 "cutoff" times due to rotation.
2485 * journald now understands SIGUSR1 and SIGUSR2 for triggering
2486 immediately flushing of runtime logs to /var if possible,
2487 resp. for triggering immediate rotation of the journal
2490 * It is now considered an error if a service is attempted to
2491 be stopped that is not loaded.
2493 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
2495 * systemd-analyze now supports Python 3
2497 * tmpfiles now supports cleaning up directories via aging
2498 where the first level dirs are always kept around but
2499 directories beneath it automatically aged. This is enabled
2500 by prefixing the age field with '~'.
2502 * Seat objects now expose CanGraphical, CanTTY properties
2503 which is required to deal with very fast bootups where the
2504 display manager might be running before the graphics drivers
2505 completed initialization.
2507 * Seat objects now expose a State property.
2509 * We now include RPM macros for service enabling/disabling
2510 based on the preset logic. We recommend RPM based
2511 distributions to make use of these macros if possible. This
2512 makes it simpler to reuse RPM spec files across
2515 * We now make sure that the collected systemd unit name is
2516 always valid when services log to the journal via
2519 * There's a new man page kernel-command-line(7) detailing all
2520 command line options we understand.
2522 * The fstab generator may now be disabled at boot by passing
2523 fstab=0 on the kernel command line.
2525 * A new kernel command line option modules-load= is now understood
2526 to load a specific kernel module statically, early at boot.
2528 * Unit names specified on the systemctl command line are now
2529 automatically escaped as needed. Also, if file system or
2530 device paths are specified they are automatically turned
2531 into the appropriate mount or device unit names. Example:
2533 systemctl status /home
2534 systemctl status /dev/sda
2536 * The SysVConsole= configuration option has been removed from
2537 system.conf parsing.
2539 * The SysV search path is no longer exported on the D-Bus
2542 * The Names= option is been removed from unit file parsing.
2544 * There's a new man page bootup(7) detailing the boot process.
2546 * Every unit and every generator we ship with systemd now
2547 comes with full documentation. The self-explanatory boot is
2550 * A couple of services gained "systemd-" prefixes in their
2551 name if they wrap systemd code, rather than only external
2552 code. Among them fsck@.service which is now
2553 systemd-fsck@.service.
2555 * The HaveWatchdog property has been removed from the D-Bus
2558 * systemd.confirm_spawn= on the kernel command line should now
2561 * There's a new man page crypttab(5) which details all options
2562 we actually understand.
2564 * systemd-nspawn gained a new --capability= switch to pass
2565 additional capabilities to the container.
2567 * timedated will now read known NTP implementation unit names
2568 from /usr/lib/systemd/ntp-units.d/*.list,
2569 systemd-timedated-ntp.target has been removed.
2571 * journalctl gained a new switch "-b" that lists log data of
2572 the current boot only.
2574 * The notify socket is in the abstract namespace again, in
2575 order to support daemons which chroot() at start-up.
2577 * There is a new Storage= configuration option for journald
2578 which allows configuration of where log data should go. This
2579 also provides a way to disable journal logging entirely, so
2580 that data collected is only forwarded to the console, the
2581 kernel log buffer or another syslog implementation.
2583 * Many bugfixes and optimizations
2585 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
2586 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
2587 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
2588 Shawn Landden, Tom Gundersen
2592 * "systemctl help <unit>" now shows the man page if one is
2595 * Several new man pages have been added.
2597 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
2598 MaxLevelConsole= can now be specified in
2599 journald.conf. These options allow reducing the amount of
2600 data stored on disk or forwarded by the log level.
2602 * TimerSlackNSec= can now be specified in system.conf for
2603 PID1. This allows system-wide power savings.
2605 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
2606 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
2611 * logind is now capable of (optionally) handling power and
2612 sleep keys as well as the lid switch.
2614 * journalctl now understands the syntax "journalctl
2615 /usr/bin/avahi-daemon" to get all log output of a specific
2618 * CapabilityBoundingSet= in system.conf now also influences
2619 the capability bound set of usermode helpers of the kernel.
2621 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
2622 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
2623 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
2624 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
2628 * Note that we skipped 139 releases here in order to set the
2629 new version to something that is greater than both udev's
2630 and systemd's most recent version number.
2632 * udev: all udev sources are merged into the systemd source tree now.
2633 All future udev development will happen in the systemd tree. It
2634 is still fully supported to use the udev daemon and tools without
2635 systemd running, like in initramfs or other init systems. Building
2636 udev though, will require the *build* of the systemd tree, but
2637 udev can be properly *run* without systemd.
2639 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
2640 should be used to create dead device nodes as workarounds for broken
2643 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
2644 no longer supported. udev_monitor_new_from_netlink() needs to be
2645 used to subscribe to events.
2647 * udev: when udevd is started by systemd, processes which are left
2648 behind by forking them off of udev rules, are unconditionally cleaned
2649 up and killed now after the event handling has finished. Services or
2650 daemons must be started as systemd services. Services can be
2651 pulled-in by udev to get started, but they can no longer be directly
2652 forked by udev rules.
2654 * udev: the daemon binary is called systemd-udevd now and installed
2655 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
2656 to adapt to that, create symlink, or rename the binary after building
2659 * libudev no longer provides these symbols:
2660 udev_monitor_from_socket()
2661 udev_queue_get_failed_list_entry()
2662 udev_get_{dev,sys,run}_path()
2663 The versions number was bumped and symbol versioning introduced.
2665 * systemd-loginctl and systemd-journalctl have been renamed
2666 to loginctl and journalctl to match systemctl.
2668 * The config files: /etc/systemd/systemd-logind.conf and
2669 /etc/systemd/systemd-journald.conf have been renamed to
2670 logind.conf and journald.conf. Package updates should rename
2671 the files to the new names on upgrade.
2673 * For almost all files the license is now LGPL2.1+, changed
2674 from the previous GPL2.0+. Exceptions are some minor stuff
2675 of udev (which will be changed to LGPL2.1 eventually, too),
2676 and the MIT licensed sd-daemon.[ch] library that is suitable
2677 to be used as drop-in files.
2679 * systemd and logind now handle system sleep states, in
2680 particular suspending and hibernating.
2682 * logind now implements a sleep/shutdown/idle inhibiting logic
2683 suitable for a variety of uses. Soonishly Lennart will blog
2684 about this in more detail.
2686 * var-run.mount and var-lock.mount are no longer provided
2687 (which prevously bind mounted these directories to their new
2688 places). Distributions which have not converted these
2689 directories to symlinks should consider stealing these files
2690 from git history and add them downstream.
2692 * We introduced the Documentation= field for units and added
2693 this to all our shipped units. This is useful to make it
2694 easier to explore the boot and the purpose of the various
2697 * All smaller setup units (such as
2698 systemd-vconsole-setup.service) now detect properly if they
2699 are run in a container and are skipped when
2700 appropriate. This guarantees an entirely noise-free boot in
2701 Linux container environments such as systemd-nspawn.
2703 * A framework for implementing offline system updates is now
2704 integrated, for details see:
2705 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
2707 * A new service type Type=idle is available now which helps us
2708 avoiding ugly interleaving of getty output and boot status
2711 * There's now a system-wide CapabilityBoundingSet= option to
2712 globally reduce the set of capabilities for the
2713 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
2714 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
2715 even CAP_NET_ADMIN system-wide for secure systems.
2717 * There are now system-wide DefaultLimitXXX= options to
2718 globally change the defaults of the various resource limits
2719 for all units started by PID 1.
2721 * Harald Hoyer's systemd test suite has been integrated into
2722 systemd which allows easy testing of systemd builds in qemu
2723 and nspawn. (This is really awesome! Ask us for details!)
2725 * The fstab parser is now implemented as generator, not inside
2728 * systemctl will now warn you if .mount units generated from
2729 /etc/fstab are out of date due to changes in fstab that
2730 haven't been read by systemd yet.
2732 * systemd is now suitable for usage in initrds. Dracut has
2733 already been updated to make use of this. With this in place
2734 initrds get a slight bit faster but primarily are much
2735 easier to introspect and debug since "systemctl status" in
2736 the host system can be used to introspect initrd services,
2737 and the journal from the initrd is kept around too.
2739 * systemd-delta has been added, a tool to explore differences
2740 between user/admin configuration and vendor defaults.
2742 * PrivateTmp= now affects both /tmp and /var/tmp.
2744 * Boot time status messages are now much prettier and feature
2745 proper english language. Booting up systemd has never been
2748 * Read-ahead pack files now include the inode number of all
2749 files to pre-cache. When the inode changes the pre-caching
2750 is not attempted. This should be nicer to deal with updated
2751 packages which might result in changes of read-ahead
2754 * We now temporaritly lower the kernel's read_ahead_kb variable
2755 when collecting read-ahead data to ensure the kernel's
2756 built-in read-ahead does not add noise to our measurements
2757 of necessary blocks to pre-cache.
2759 * There's now RequiresMountsFor= to add automatic dependencies
2760 for all mounts necessary for a specific file system path.
2762 * MountAuto= and SwapAuto= have been removed from
2763 system.conf. Mounting file systems at boot has to take place
2766 * nspawn now learned a new switch --uuid= to set the machine
2767 ID on the command line.
2769 * nspawn now learned the -b switch to automatically search
2772 * vt102 is now the default TERM for serial TTYs, upgraded from
2775 * systemd-logind now works on VT-less systems.
2777 * The build tree has been reorganized. The individual
2778 components now have directories of their own.
2780 * A new condition type ConditionPathIsReadWrite= is now available.
2782 * nspawn learned the new -C switch to create cgroups for the
2783 container in other hierarchies.
2785 * We now have support for hardware watchdogs, configurable in
2788 * The scheduled shutdown logic now has a public API.
2790 * We now mount /tmp as tmpfs by default, but this can be
2791 masked and /etc/fstab can override it.
2793 * Since udisks doesn't make use of /media anymore we are not
2794 mounting a tmpfs on it anymore.
2796 * journalctl gained a new --local switch to only interleave
2797 locally generated journal files.
2799 * We can now load the IMA policy at boot automatically.
2801 * The GTK tools have been split off into a systemd-ui.
2803 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
2804 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
2805 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
2806 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
2807 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
2808 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
2809 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
2810 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
2811 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
2816 * This is mostly a bugfix release
2818 * Support optional initialization of the machine ID from the
2819 KVM or container configured UUID.
2821 * Support immediate reboots with "systemctl reboot -ff"
2823 * Show /etc/os-release data in systemd-analyze output
2825 * Many bugfixes for the journal, including endianness fixes and
2826 ensuring that disk space enforcement works
2828 * sd-login.h is C++ comptaible again
2830 * Extend the /etc/os-release format on request of the Debian
2833 * We now refuse non-UTF8 strings used in various configuration
2834 and unit files. This is done to ensure we don't pass invalid
2835 data over D-Bus or expose it elsewhere.
2837 * Register Mimo USB Screens as suitable for automatic seat
2840 * Read SELinux client context from journal clients in a race
2843 * Reorder configuration file lookup order. /etc now always
2844 overrides /run in order to allow the administrator to always
2845 and unconditionally override vendor supplied or
2846 automatically generated data.
2848 * The various user visible bits of the journal now have man
2849 pages. We still lack man pages for the journal API calls
2852 * We now ship all man pages in HTML format again in the
2855 Contributions from: Dave Reisner, Dirk Eibach, Frederic
2856 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
2857 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
2862 * This is mostly a bugfix release
2864 * systems lacking /etc/os-release are no longer supported.
2866 * Various functionality updates to libsystemd-login.so
2868 * Track class of PAM logins to distuingish greeters from
2871 Contributions from: Kay Sievers, Lennart Poettering, Michael
2876 * This is an important bugfix release for v41.
2878 * Building man pages is now optional which should be useful
2879 for those building systemd from git but unwilling to install
2882 * Watchdog support for supervising services is now usable. In
2883 a future release support for hardware watchdogs
2884 (i.e. /dev/watchdog) will be added building on this.
2886 * Service start rate limiting is now configurable and can be
2887 turned off per service. When a start rate limit is hit a
2888 reboot can automatically be triggered.
2890 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
2892 Contributions from: Benjamin Franzke, Bill Nottingham,
2893 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
2894 Schmidt, Michał Górny, Piotr Drąg
2898 * The systemd binary is installed /usr/lib/systemd/systemd now;
2899 An existing /sbin/init symlink needs to be adapted with the
2902 * The code that loads kernel modules has been ported to invoke
2903 libkmod directly, instead of modprobe. This means we do not
2904 support systems with module-init-tools anymore.
2906 * Watchdog support is now already useful, but still not
2909 * A new kernel command line option systemd.setenv= is
2910 understood to set system wide environment variables
2911 dynamically at boot.
2913 * We now limit the set of capabilities of systemd-journald.
2915 * We now set SIGPIPE to ignore by default, since it only is
2916 useful in shell pipelines, and has little use in general
2917 code. This can be disabled with IgnoreSIPIPE=no in unit
2920 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
2921 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
2926 * This is mostly a bugfix release
2928 * We now expose the reason why a service failed in the
2929 "Result" D-Bus property.
2931 * Rudimentary service watchdog support (will be completed over
2932 the next few releases.)
2934 * When systemd forks off in order execute some service we will
2935 now immediately changes its argv[0] to reflect which process
2936 it will execute. This is useful to minimize the time window
2937 with a generic argv[0], which makes bootcharts more useful
2939 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
2940 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
2941 Mike Kazantsev, Ray Strode
2945 * This is mostly a test release, but incorporates many
2948 * New systemd-cgtop tool to show control groups by their
2951 * Linking against libacl for ACLs is optional again. If
2952 disabled, support tracking device access for active logins
2953 goes becomes unavailable, and so does access to the user
2954 journals by the respective users.
2956 * If a group "adm" exists, journal files are automatically
2957 owned by them, thus allow members of this group full access
2958 to the system journal as well as all user journals.
2960 * The journal now stores the SELinux context of the logging
2961 client for all entries.
2963 * Add C++ inclusion guards to all public headers
2965 * New output mode "cat" in the journal to print only text
2966 messages, without any meta data like date or time.
2968 * Include tiny X server wrapper as a temporary stop-gap to
2969 teach XOrg udev display enumeration. This is used by display
2970 managers such as gdm, and will go away as soon as XOrg
2971 learned native udev hotplugging for display devices.
2973 * Add new systemd-cat tool for executing arbitrary programs
2974 with STDERR/STDOUT connected to the journal. Can also act as
2975 BSD logger replacement, and does so by default.
2977 * Optionally store all locally generated coredumps in the
2978 journal along with meta data.
2980 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
2981 writing short strings to files (for usage for /sys), and for
2982 creating symlinks, character and block device nodes.
2984 * New unit file option ControlGroupPersistent= to make cgroups
2985 persistent, following the mechanisms outlined in
2986 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
2988 * Support multiple local RTCs in a sane way
2990 * No longer monopolize IO when replaying readahead data on
2991 rotating disks, since we might starve non-file-system IO to
2992 death, since fanotify() will not see accesses done by blkid,
2995 * Don't show kernel threads in systemd-cgls anymore, unless
2996 requested with new -k switch.
2998 Contributions from: Dan Horák, Kay Sievers, Lennart
2999 Poettering, Michal Schmidt
3003 * This is mostly a test release, but incorporates many
3006 * The git repository moved to:
3007 git://anongit.freedesktop.org/systemd/systemd
3008 ssh://git.freedesktop.org/git/systemd/systemd
3010 * First release with the journal
3011 http://0pointer.de/blog/projects/the-journal.html
3013 * The journal replaces both systemd-kmsg-syslogd and
3014 systemd-stdout-bridge.
3016 * New sd_pid_get_unit() API call in libsystemd-logind
3018 * Many systemadm clean-ups
3020 * Introduce remote-fs-pre.target which is ordered before all
3021 remote mounts and may be used to start services before all
3024 * Added Mageia support
3026 * Add bash completion for systemd-loginctl
3028 * Actively monitor PID file creation for daemons which exit in
3029 the parent process before having finished writing the PID
3030 file in the daemon process. Daemons which do this need to be
3031 fixed (i.e. PID file creation must have finished before the
3032 parent exits), but we now react a bit more gracefully to them.
3034 * Add colourful boot output, mimicking the well-known output
3035 of existing distributions.
3037 * New option PassCredentials= for socket units, for
3038 compatibility with a recent kernel ABI breakage.
3040 * /etc/rc.local is now hooked in via a generator binary, and
3041 thus will no longer act as synchronization point during
3044 * systemctl list-unit-files now supports --root=.
3046 * systemd-tmpfiles now understands two new commands: z, Z for
3047 relabelling files according to the SELinux database. This is
3048 useful to apply SELinux labels to specific files in /sys,
3051 * Output of SysV services is now forwarded to both the console
3052 and the journal by default, not only just the console.
3054 * New man pages for all APIs from libsystemd-login.
3056 * The build tree got reorganized and a the build system is a
3057 lot more modular allowing embedded setups to specifically
3058 select the components of systemd they are interested in.
3060 * Support for Linux systems lacking the kernel VT subsystem is
3063 * configure's --with-rootdir= got renamed to
3064 --with-rootprefix= to follow the naming used by udev and
3067 * Unless specified otherwise we'll now install to /usr instead
3068 of /usr/local by default.
3070 * Processes with '@' in argv[0][0] are now excluded from the
3071 final shut-down killing spree, following the logic explained
3073 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
3075 * All processes remaining in a service cgroup when we enter
3076 the START or START_PRE states are now killed with
3077 SIGKILL. That means it is no longer possible to spawn
3078 background processes from ExecStart= lines (which was never
3079 supported anyway, and bad style).
3081 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
3082 reloading of units together.
3084 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
3085 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
3086 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
3087 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
3088 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek