1 systemd System and Service Manager
5 * systemd will now relabel /dev after loading the SMACK policy
6 according to SMACK rules.
8 * A new unit file option AppArmoreProfile= has been added to
9 set the AppArmor profile for the processes of a unit.
11 * A new condition check ConditionArchitecture= has been added
12 to conditionalize units based on the system architecture, as
13 reported by uname()'s "machine" field.
15 * systemd-networkd now supports matching on the system
16 virtualization, architecture, kernel command line, host name
19 * logind is now a lot more aggressive when suspending the
20 machine due to a closed laptop lid. Instead of acting only
21 on the lid close action it will continuously watch the lid
22 status and act on it. This is useful for laptops where the
23 power button is on the outside of the chassis so that it can
24 be reached without opening the lid (such as the Lenovo
25 Yoga). On those machines logind will now immediately
26 re-suspend the machine if the power button has been
27 accidentally pressed while the laptop was suspended and in a
30 * logind will now watch SW_DOCK switches and inhibit reaction
31 to the lid switch if it is pressed. This means that logind
32 will not suspend the machine anymore if the lid is closed
33 and the systemd is docked, if the laptop supports SW_DOCK
34 notifications via the input layer. Note that ACPI docking
35 stations do not generate this currently. Also note that this
36 logic is usually not fully sufficient and Desktop
37 Environments should take a lid switch inhibitor lock when an
38 external display is connected, as systemd will not watch
41 * nspawn will now make use of the devices cgroup controller by
42 default, and only permit creation of and access to the usual
43 API device nodes like /dev/null or /dev/random, as well as
44 access to (but not creation of) the pty devices.
46 * We will now ship a default .network file for
47 systemd-networkd that automatically configures DHCP for
48 network interfaces created by nspawn's --network-veth or
49 --network-bridge= switches.
51 * systemd will now understand the usual M, K, G, T suffixes
52 according to SI conventions (i.e. to the base 1000) when
53 referring to throughput and hardware metrics. It will stay
54 with IEC conventions (i.e. to the base 1024) for software
55 metrics, according to what is customary according to
56 Wikipedia. We explicitly document which base applies for
57 each configuration option.
59 * The DeviceAllow= setting in unit files now supports a syntax
60 to whitelist an entire group of devices node majors at once,
61 based on the /proc/devices listing. For example, with the
62 string "char-pts" it is now possible to whitelist all
63 current and future pseudo-TTYs at once.
65 * sd-event learned a new "post" event source. Event sources of
66 this type are triggered by the dispatching of any event
67 source of a type that is not "post". This is useful for
68 implementing clean-up and check event sources that are
69 triggered by other work being done in the program.
71 * systemd-networkd is no longer statically enabled, but uses
72 the usual [Install] sections so that it can be
73 enabled/disabled using systemctl. It still is enabled by
76 * When creating a veth interface pair with systemd-nspawn the
77 host side will now be prefixed with "vb-" if
78 --network-bridge= is used, and with "ve-" if --network-veth
79 is used. This way it is easy to distinguish these cases on
80 the host, for example to apply different configuration to
81 them with systemd-networkd.
83 * The compatibility libraries for libsystemd-journal.so,
84 libsystem-id128.so, libsystemd-login.so and
85 libsystemd-daemon.so do not make use of IFUNC
86 anymore. Instead we now build libsystemd.so multiple times
87 under these alternative names. This means that the footprint
88 is drastically increased, but given that these are
89 transitional compatibility libraries this shouldn't matter
90 much. This change has been made necessary to support the ARM
91 platform for these compatibility libraries, as the ARM
92 toolchain isn't really at the same level as the toolchain
93 for other architectures like x86 and does not support
94 IFUNC. Please make sure to use --enable-compat-libs only
95 during a transitional period!
97 Contributions from: Andreas Fuchs, Armin K, Colin Walters,
98 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
99 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
100 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
101 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
102 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
103 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
104 Zbigniew Jędrzejewski-Szmek
106 -- Berlin, 2014-02-24
110 * A new component "systemd-networkd" has been added that can
111 be used to configure local network interfaces statically or
112 via DHCP. It is capable of bringing up bridges, VLANs, and
113 bonding. Currently, no hook-ups for interactive network
114 configuration are provided. Use this for your initrd,
115 container, embedded, or server setup if you need a simple,
116 yet powerful, network configuration solution. This
117 configuration subsystem is quite nifty, as it allows wildcard
118 hotplug matching in interfaces. For example, with a single
119 configuration snippet, you can configure that all Ethernet
120 interfaces showing up are automatically added to a bridge,
121 or similar. It supports link-sensing and more.
123 * A new tool "systemd-socket-proxyd" has been added which can
124 act as a bidirectional proxy for TCP sockets. This is
125 useful for adding socket activation support to services that
126 do not actually support socket activation, including virtual
127 machines and the like.
129 * Add a new tool to save/restore rfkill state on
132 * Save/restore state of keyboard backlights in addition to
133 display backlights on shutdown/boot.
135 * udev learned a new SECLABEL{} construct to label device
136 nodes with a specific security label when they appear. For
137 now, only SECLABEL{selinux} is supported, but the syntax is
138 prepared for additional security frameworks.
140 * udev gained a new scheme to configure link-level attributes
141 from files in /etc/systemd/network/*.link. These files can
142 match against MAC address, device path, driver name and type,
143 and will apply attributes like the naming policy, link speed,
144 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
145 address assignment policy (randomized, ...).
147 * The configuration of network interface naming rules for
148 "permanent interface names" has changed: a new NamePolicy=
149 setting in the [Link] section of .link files determines the
150 priority of possible naming schemes (onboard, slot, mac,
151 path). The default value of this setting is determined by
152 /usr/lib/net/links/99-default.link. Old
153 80-net-name-slot.rules udev configuration file has been
154 removed, so local configuration overriding this file should
155 be adapated to override 99-default.link instead.
157 * When the User= switch is used in a unit file, also
158 initialize $SHELL= based on the user database entry.
160 * systemd no longer depends on libdbus. All communication is
161 now done with sd-bus, systemd's low-level bus library
164 * kdbus support has been added to PID 1 itself. When kdbus is
165 enabled, this causes PID 1 to set up the system bus and
166 enable support for a new ".busname" unit type that
167 encapsulates bus name activation on kdbus. It works a little
168 bit like ".socket" units, except for bus names. A new
169 generator has been added that converts classic dbus1 service
170 activation files automatically into native systemd .busname
173 * sd-bus: add a light-weight vtable implementation that allows
174 defining objects on the bus with a simple static const
175 vtable array of its methods, signals and properties.
177 * systemd will not generate or install static dbus
178 introspection data anymore to /usr/share/dbus-1/interfaces,
179 as the precise format of these files is unclear, and
180 nothing makes use of it.
182 * A proxy daemon is now provided to proxy clients connecting
183 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
184 compatibility with classic D-Bus.
186 * A bus driver implementation has been added that supports the
187 classic D-Bus bus driver calls on kdbus, also for
188 compatibility purposes.
190 * A new API "sd-event.h" has been added that implements a
191 minimal event loop API built around epoll. It provides a
192 couple of features that direct epoll usage is lacking:
193 prioritization of events, scales to large numbers of timer
194 events, per-event timer slack (accuracy), system-wide
195 coalescing of timer events, exit handlers, watchdog
196 supervision support using systemd's sd_notify() API, child
199 * A new API "sd-rntl.h" has been added that provides an API
200 around the route netlink interface of the kernel, similar in
203 * A new API "sd-dhcp-client.h" has been added that provides a
204 small DHCPv4 client-side implementation. This is used by
207 * There is a new kernel command line option
208 "systemd.restore_state=0|1". When set to "0", none of the
209 systemd tools will restore saved runtime state to hardware
210 devices. More specifically, the rfkill and backlight states
213 * The FsckPassNo= compatibility option in mount/service units
214 has been removed. The fstab generator will now add the
215 necessary dependencies automatically, and does not require
216 PID1's support for that anymore.
218 * journalctl gained a new switch, --list-boots, that lists
219 recent boots with their times and boot IDs.
221 * The various tools like systemctl, loginctl, timedatectl,
222 busctl, systemd-run, ... have gained a new switch "-M" to
223 connect to a specific, local OS container (as direct
224 connection, without requiring SSH). This works on any
225 container that is registered with machined, such as those
226 created by libvirt-lxc or nspawn.
228 * systemd-run and systemd-analyze also gained support for "-H"
229 to connect to remote hosts via SSH. This is particularly
230 useful for systemd-run because it enables queuing of jobs
233 * machinectl gained a new command "login" to open a getty
234 login in any local container. This works with any container
235 that is registered with machined (such as those created by
236 libvirt-lxc or nspawn), and which runs systemd inside.
238 * machinectl gained a new "reboot" command that may be used to
239 trigger a reboot on a specific container that is registered
240 with machined. This works on any container that runs an init
243 * systemctl gained a new "list-timers" command to print a nice
244 listing of installed timer units with the times they elapse
247 * Alternative reboot() parameters may now be specified on the
248 "systemctl reboot" command line and are passed to the
249 reboot() system call.
251 * systemctl gained a new --job-mode= switch to configure the
252 mode to queue a job with. This is a more generic version of
253 --fail, --irreversible, and --ignore-dependencies, which are
254 still available but not advertised anymore.
256 * /etc/systemd/system.conf gained new settings to configure
257 various default timeouts of units, as well as the default
258 start limit interval and burst. These may still be overridden
261 * PID1 will now export on the bus profile data of the security
262 policy upload process (such as the SELinux policy upload to
265 * journald: when forwarding logs to the console, include
266 timestamps (following the setting in
267 /sys/module/printk/parameters/time).
269 * OnCalendar= in timer units now understands the special
270 strings "yearly" and "annually". (Both are equivalent)
272 * The accuracy of timer units is now configurable with the new
273 AccuracySec= setting. It defaults to 1min.
275 * A new dependency type JoinsNamespaceOf= has been added that
276 allows running two services within the same /tmp and network
277 namespace, if PrivateNetwork= or PrivateTmp= are used.
279 * A new command "cat" has been added to systemctl. It outputs
280 the original unit file of a unit, and concatenates the
281 contents of additional "drop-in" unit file snippets, so that
282 the full configuration is shown.
284 * systemctl now supports globbing on the various "list-xyz"
285 commands, like "list-units" or "list-sockets", as well as on
286 those commands which take multiple unit names.
288 * journalctl's --unit= switch gained support for globbing.
290 * All systemd daemons now make use of the watchdog logic so
291 that systemd automatically notices when they hang.
293 * If the $container_ttys environment variable is set,
294 getty-generator will automatically spawn a getty for each
295 listed tty. This is useful for container managers to request
296 login gettys to be spawned on as many ttys as needed.
298 * %h, %s, %U specifier support is not available anymore when
299 used in unit files for PID 1. This is because NSS calls are
300 not safe from PID 1. They stay available for --user
301 instances of systemd, and as special case for the root user.
303 * loginctl gained a new "--no-legend" switch to turn off output
306 * The "sd-login.h" API gained three new calls:
307 sd_session_is_remote(), sd_session_get_remote_user(),
308 sd_session_get_remote_host() to query information about
311 * The udev hardware database now also carries vendor/product
312 information of SDIO devices.
314 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
315 determine whether watchdog notifications are requested by
318 * Socket-activated per-connection services now include a
319 short description of the connection parameters in the
322 * tmpfiles gained a new "--boot" option. When this is not used,
323 only lines where the command character is not suffixed with
324 "!" are executed. When this option is specified, those
325 options are executed too. This partitions tmpfiles
326 directives into those that can be safely executed at any
327 time, and those which should be run only at boot (for
328 example, a line that creates /run/nologin).
330 * A new API "sd-resolve.h" has been added which provides a simple
331 asynchronous wrapper around glibc NSS host name resolution
332 calls, such as getaddrinfo(). In contrast to glibc's
333 getaddrinfo_a(), it does not use signals. In contrast to most
334 other asynchronous name resolution libraries, this one does
335 not reimplement DNS, but reuses NSS, so that alternate
336 host name resolution systems continue to work, such as mDNS,
337 LDAP, etc. This API is based on libasyncns, but it has been
338 cleaned up for inclusion in systemd.
340 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
341 "sd-daemon.h" are no longer found in individual libraries
342 libsystemd-journal.so, libsystemd-login.so,
343 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
344 merged them into a single library, libsystemd.so, which
345 provides all symbols. The reason for this is cyclic
346 dependencies, as these libraries tend to use each other's
347 symbols. So far, we've managed to workaround that by linking
348 a copy of a good part of our code into each of these
349 libraries again and again, which, however, makes certain
350 things hard to do, like sharing static variables. Also, it
351 substantially increases footprint. With this change, there
352 is only one library for the basic APIs systemd
353 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
354 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
355 library as well, however are subject to the --enable-kdbus
356 switch (see below). Note that "sd-dhcp-client.h" is not part
357 of this library (this is because it only consumes, never
358 provides, services of/to other APIs). To make the transition
359 easy from the separate libraries to the unified one, we
360 provide the --enable-compat-libs compile-time switch which
361 will generate stub libraries that are compatible with the
362 old ones but redirect all calls to the new one.
364 * All of the kdbus logic and the new APIs "sd-bus.h",
365 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
366 and "sd-utf8.h" are compile-time optional via the
367 "--enable-kdbus" switch, and they are not compiled in by
368 default. To make use of kdbus, you have to explicitly enable
369 the switch. Note however, that neither the kernel nor the
370 userspace API for all of this is considered stable yet. We
371 want to maintain the freedom to still change the APIs for
372 now. By specifying this build-time switch, you acknowledge
373 that you are aware of the instability of the current
376 * Also, note that while kdbus is pretty much complete,
377 it lacks one thing: proper policy support. This means you
378 can build a fully working system with all features; however,
379 it will be highly insecure. Policy support will be added in
380 one of the next releases, at the same time that we will
381 declare the APIs stable.
383 * When the kernel command-line argument "kdbus" is specified,
384 systemd will automatically load the kdbus.ko kernel module. At
385 this stage of development, it is only useful for testing kdbus
386 and should not be used in production. Note: if "--enable-kdbus"
387 is specified, and the kdbus.ko kernel module is available, and
388 "kdbus" is added to the kernel command line, the entire system
389 runs with kdbus instead of dbus-daemon, with the above mentioned
390 problem of missing the system policy enforcement. Also a future
391 version of kdbus.ko or a newer systemd will not be compatible with
392 each other, and will unlikely be able to boot the machine if only
393 one of them is updated.
395 * systemctl gained a new "import-environment" command which
396 uploads the caller's environment (or parts thereof) into the
397 service manager so that it is inherited by services started
398 by the manager. This is useful to upload variables like
399 $DISPLAY into the user service manager.
401 * A new PrivateDevices= switch has been added to service units
402 which allows running a service with a namespaced /dev
403 directory that does not contain any device nodes for
404 physical devices. More specifically, it only includes devices
405 such as /dev/null, /dev/urandom, and /dev/zero which are API
408 * logind has been extended to support behaviour like VT
409 switching on seats that do not support a VT. This makes
410 multi-session available on seats that are not the first seat
411 (seat0), and on systems where kernel support for VTs has
412 been disabled at compile-time.
414 * If a process holds a delay lock for system sleep or shutdown
415 and fails to release it in time, we will now log its
416 identity. This makes it easier to identify processes that
417 cause slow suspends or power-offs.
419 * When parsing /etc/crypttab, support for a new key-slot=
420 option as supported by Debian is added. It allows indicating
421 which LUKS slot to use on disk, speeding up key loading.
423 * The sd_journald_sendv() API call has been checked and
424 officially declared to be async-signal-safe so that it may
425 be invoked from signal handlers for logging purposes.
427 * Boot-time status output is now enabled automatically after a
428 short timeout if boot does not progress, in order to give
429 the user an indication what she or he is waiting for.
431 * The boot-time output has been improved to show how much time
432 remains until jobs expire.
434 * The KillMode= switch in service units gained a new possible
435 value "mixed". If set, and the unit is shut down, then the
436 initial SIGTERM signal is sent only to the main daemon
437 process, while the following SIGKILL signal is sent to
438 all remaining processes of the service.
440 * When a scope unit is registered, a new property "Controller"
441 may be set. If set to a valid bus name, systemd will send a
442 RequestStop() signal to this name when it would like to shut
443 down the scope. This may be used to hook manager logic into
444 the shutdown logic of scope units. Also, scope units may now
445 be put in a special "abandoned" state, in which case the
446 manager process which created them takes no further
447 responsibilities for it.
449 * When reading unit files, systemd will now verify
450 the access mode of these files, and warn about certain
451 suspicious combinations. This has been added to make it
452 easier to track down packaging bugs where unit files are
453 marked executable or world-writable.
455 * systemd-nspawn gained a new "--setenv=" switch to set
456 container-wide environment variables. The similar option in
457 systemd-activate was renamed from "--environment=" to
458 "--setenv=" for consistency.
460 * systemd-nspawn has been updated to create a new kdbus domain
461 for each container that is invoked, thus allowing each
462 container to have its own set of system and user buses,
463 independent of the host.
465 * systemd-nspawn gained a new --drop-capability= switch to run
466 the container with less capabilities than the default. Both
467 --drop-capability= and --capability= now take the special
468 string "all" for dropping or keeping all capabilities.
470 * systemd-nspawn gained new switches for executing containers
471 with specific SELinux labels set.
473 * systemd-nspawn gained a new --quiet switch to not generate
474 any additional output but the container's own console
477 * systemd-nspawn gained a new --share-system switch to run a
478 container without PID namespacing enabled.
480 * systemd-nspawn gained a new --register= switch to control
481 whether the container is registered with systemd-machined or
482 not. This is useful for containers that do not run full
483 OS images, but only specific apps.
485 * systemd-nspawn gained a new --keep-unit which may be used
486 when invoked as the only program from a service unit, and
487 results in registration of the unit service itself in
488 systemd-machined, instead of a newly opened scope unit.
490 * systemd-nspawn gained a new --network-interface= switch for
491 moving arbitrary interfaces to the container. The new
492 --network-veth switch creates a virtual Ethernet connection
493 between host and container. The new --network-bridge=
494 switch then allows assigning the host side of this virtual
495 Ethernet connection to a bridge device.
497 * systemd-nspawn gained a new --personality= switch for
498 setting the kernel personality for the container. This is
499 useful when running a 32bit container on a 64bit host. A
500 similar option Personality= is now also available in service
503 * logind will now also track a "Desktop" identifier for each
504 session which encodes the desktop environment of it. This is
505 useful for desktop environments that want to identify
506 multiple running sessions of itself easily.
508 * A new SELinuxContext= setting for service units has been
509 added that allows setting a specific SELinux execution
510 context for a service.
512 * Most systemd client tools will now honour $SYSTEMD_LESS for
513 settings of the "less" pager. By default, these tools will
514 override $LESS to allow certain operations to work, such as
515 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
516 influence this logic.
518 * systemd's "seccomp" hook-up has been changed to make use of
519 the libseccomp library instead of using its own
520 implementation. This has benefits for portability among
523 * For usage together with SystemCallFilter=, a new
524 SystemCallErrorNumber= setting has been introduced that
525 allows configuration of a system error number to return on
526 filtered system calls, instead of immediately killing the
527 process. Also, SystemCallArchitectures= has been added to
528 limit access to system calls of a particular architecture
529 (in order to turn off support for unused secondary
530 architectures). There is also a global
531 SystemCallArchitectures= setting in system.conf now to turn
532 off support for non-native system calls system-wide.
534 * systemd requires a kernel with a working name_to_handle_at(),
535 please see the kernel config requirements in the README file.
537 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
538 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
539 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
540 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
541 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
542 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
543 Elia Pinto, Florian Weimer, George McCollister, Goffredo
544 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
545 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
546 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
547 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
548 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
549 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
550 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
551 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
552 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
553 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
554 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
555 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
556 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
557 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
558 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
559 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
560 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
562 -- Berlin, 2014-02-20
566 * logind has gained support for facilitating privileged input
567 and drm device access for unprivileged clients. This work is
568 useful to allow Wayland display servers (and similar
569 programs, such as kmscon) to run under the user's ID and
570 access input and drm devices which are normally
571 protected. When this is used (and the kernel is new enough)
572 logind will "mute" IO on the file descriptors passed to
573 Wayland as long as it is in the background and "unmute" it
574 if it returns into the foreground. This allows secure
575 session switching without allowing background sessions to
576 eavesdrop on input and display data. This also introduces
577 session switching support if VT support is turned off in the
578 kernel, and on seats that are not seat0.
580 * A new kernel command line option luks.options= is understood
581 now which allows specifiying LUKS options for usage for LUKS
582 encrypted partitions specified with luks.uuid=.
584 * tmpfiles.d(5) snippets may now use specifier expansion in
585 path names. More specifically %m, %b, %H, %v, are now
586 replaced by the local machine id, boot id, hostname, and
587 kernel version number.
589 * A new tmpfiles.d(5) command "m" has been introduced which
590 may be used to change the owner/group/access mode of a file
591 or directory if it exists, but do nothing if it doesn't.
593 * This release removes high-level support for the
594 MemorySoftLimit= cgroup setting. The underlying kernel
595 cgroup attribute memory.soft_limit= is currently badly
596 designed and likely to be removed from the kernel API in its
597 current form, hence we shouldn't expose it for now.
599 * The memory.use_hierarchy cgroup attribute is now enabled for
600 all cgroups systemd creates in the memory cgroup
601 hierarchy. This option is likely to be come the built-in
602 default in the kernel anyway, and the non-hierarchial mode
603 never made much sense in the intrinsically hierarchial
606 * A new field _SYSTEMD_SLICE= is logged along with all journal
607 messages containing the slice a message was generated
608 from. This is useful to allow easy per-customer filtering of
609 logs among other things.
611 * systemd-journald will no longer adjust the group of journal
612 files it creates to the "systemd-journal" group. Instead we
613 rely on the journal directory to be owned by the
614 "systemd-journal" group, and its setgid bit set, so that the
615 kernel file system layer will automatically enforce that
616 journal files inherit this group assignment. The reason for
617 this change is that we cannot allow NSS look-ups from
618 journald which would be necessary to resolve
619 "systemd-journal" to a numeric GID, because this might
620 create deadlocks if NSS involves synchronous queries to
621 other daemons (such as nscd, or sssd) which in turn are
622 logging clients of journald and might block on it, which
623 would then dead lock. A tmpfiles.d(5) snippet included in
624 systemd will make sure the setgid bit and group are
625 properly set on the journal directory if it exists on every
626 boot. However, we recommend adjusting it manually after
627 upgrades too (or from RPM scriptlets), so that the change is
628 not delayed until next reboot.
630 * Backlight and random seed files in /var/lib/ have moved into
631 the /var/lib/systemd/ directory, in order to centralize all
632 systemd generated files in one directory.
634 * Boot time performance measurements (as displayed by
635 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
636 performance information if that's available to determine how
637 much time BIOS and boot loader initialization required. With
638 a sufficiently new BIOS you hence no longer need to boot
639 with Gummiboot to get access to such information.
641 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
642 Cristian Rodríguez, Dave Reisner, David Herrmann, David
643 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
644 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
645 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
646 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
647 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
649 -- Berlin, 2013-10-02
653 * The Restart= option for services now understands a new
654 on-watchdog setting, which will restart the service
655 automatically if the service stops sending out watchdog keep
656 alive messages (as configured with WatchdogSec=).
658 * The getty generator (which is responsible for bringing up a
659 getty on configured serial consoles) will no longer only
660 start a getty on the primary kernel console but on all
661 others, too. This makes the order in which console= is
662 specified on the kernel command line less important.
664 * libsystemd-logind gained a new sd_session_get_vt() call to
665 retrieve the VT number of a session.
667 * If the option "tries=0" is set for an entry of /etc/crypttab
668 its passphrase is queried indefinitely instead of any
669 maximum number of tries.
671 * If a service with a configure PID file terminates its PID
672 file will now be removed automatically if it still exists
673 afterwards. This should put an end to stale PID files.
675 * systemd-run will now also take relative binary path names
676 for execution and no longer insists on absolute paths.
678 * InaccessibleDirectories= and ReadOnlyDirectories= now take
679 paths that are optionally prefixed with "-" to indicate that
680 it should not be considered a failure if they don't exist.
682 * journalctl -o (and similar commands) now understands a new
683 output mode "short-precise", it is similar to "short" but
684 shows timestamps with usec accuracy.
686 * The option "discard" (as known from Debian) is now
687 synonymous to "allow-discards" in /etc/crypttab. In fact,
688 "discard" is preferred now (since it is easier to remember
691 * Some licensing clean-ups were made, so that more code is now
692 LGPL-2.1 licensed than before.
694 * A minimal tool to save/restore the display backlight
695 brightness across reboots has been added. It will store the
696 backlight setting as late as possible at shutdown, and
697 restore it as early as possible during reboot.
699 * A logic to automatically discover and enable home and swap
700 partitions on GPT disks has been added. With this in place
701 /etc/fstab becomes optional for many setups as systemd can
702 discover certain partitions located on the root disk
703 automatically. Home partitions are recognized under their
704 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
705 partitions are recognized under their GPT type ID
706 0657fd6da4ab43c484e50933c84b4f4f.
708 * systemd will no longer pass any environment from the kernel
709 or initrd to system services. If you want to set an
710 environment for all services, do so via the kernel command
711 line systemd.setenv= assignment.
713 * The systemd-sysctl tool no longer natively reads the file
714 /etc/sysctl.conf. If desired, the file should be symlinked
715 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
716 legacy support by a symlink rather than built-in code, it
717 also makes the otherwise hidden order of application of the
718 different files visible. (Note that this partly reverts to a
719 pre-198 application order of sysctl knobs!)
721 * The "systemctl set-log-level" and "systemctl dump" commands
722 have been moved to systemd-analyze.
724 * systemd-run learned the new --remain-after-exit switch,
725 which causes the scope unit not to be cleaned up
726 automatically after the process terminated.
728 * tmpfiles learned a new --exclude-prefix= switch to exclude
729 certain paths from operation.
731 * journald will now automatically flush all messages to disk
732 as soon as a message of the log priorities CRIT, ALERT or
735 Contributions from: Andrew Cook, Brandon Philips, Christian
736 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
737 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
738 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
739 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
740 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
741 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
742 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
743 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
744 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
745 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
746 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
747 William Giokas, Zbigniew Jędrzejewski-Szmek
749 -- Berlin, 2013-09-13
753 * The documentation has been updated to cover the various new
754 concepts introduced with 205.
756 * Unit files now understand the new %v specifier which
757 resolves to the kernel version string as returned by "uname
760 * systemctl now supports filtering the unit list output by
761 load state, active state and sub state, using the new
764 * "systemctl status" will now show the results of the
765 condition checks (like ConditionPathExists= and similar) of
766 the last start attempts of the unit. They are also logged to
769 * "journalctl -b" may now be used to look for boot output of a
770 specific boot. Try "journalctl -b -1" for the previous boot,
771 but the syntax is substantially more powerful.
773 * "journalctl --show-cursor" has been added which prints the
774 cursor string the last shown log line. This may then be used
775 with the new "journalctl --after-cursor=" switch to continue
776 browsing logs from that point on.
778 * "journalctl --force" may now be used to force regeneration
781 * Creation of "dead" device nodes has been moved from udev
782 into kmod and tmpfiles. Previously, udev would read the kmod
783 databases to pre-generate dead device nodes based on meta
784 information contained in kernel modules, so that these would
785 be auto-loaded on access rather then at boot. As this
786 doesn't really have much to do with the exposing actual
787 kernel devices to userspace this has always been slightly
788 alien in the udev codebase. Following the new scheme kmod
789 will now generate a runtime snippet for tmpfiles from the
790 module meta information and it now is tmpfiles' job to the
791 create the nodes. This also allows overriding access and
792 other parameters for the nodes using the usual tmpfiles
793 facilities. As side effect this allows us to remove the
794 CAP_SYS_MKNOD capability bit from udevd entirely.
796 * logind's device ACLs may now be applied to these "dead"
797 devices nodes too, thus finally allowing managed access to
798 devices such as /dev/snd/sequencer whithout loading the
799 backing module right-away.
801 * A new RPM macro has been added that may be used to apply
802 tmpfiles configuration during package installation.
804 * systemd-detect-virt and ConditionVirtualization= now can
805 detect User-Mode-Linux machines (UML).
807 * journald will now implicitly log the effective capabilities
808 set of processes in the message metadata.
810 * systemd-cryptsetup has gained support for TrueCrypt volumes.
812 * The initrd interface has been simplified (more specifically,
813 support for passing performance data via environment
814 variables and fsck results via files in /run has been
815 removed). These features were non-essential, and are
816 nowadays available in a much nicer way by having systemd in
817 the initrd serialize its state and have the hosts systemd
818 deserialize it again.
820 * The udev "keymap" data files and tools to apply keyboard
821 specific mappings of scan to key codes, and force-release
822 scan code lists have been entirely replaced by a udev
823 "keyboard" builtin and a hwdb data file.
825 * systemd will now honour the kernel's "quiet" command line
826 argument also during late shutdown, resulting in a
827 completely silent shutdown when used.
829 * There's now an option to control the SO_REUSEPORT socket
830 option in .socket units.
832 * Instance units will now automatically get a per-template
833 subslice of system.slice unless something else is explicitly
834 configured. For example, instances of sshd@.service will now
835 implicitly be placed in system-sshd.slice rather than
836 system.slice as before.
838 * Test coverage support may now be enabled at build time.
840 Contributions from: Dave Reisner, Frederic Crozat, Harald
841 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
842 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
843 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
844 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
845 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
846 Giokas, Zbigniew Jędrzejewski-Szmek
848 -- Berlin, 2013-07-23
852 * Two new unit types have been introduced:
854 Scope units are very similar to service units, however, are
855 created out of pre-existing processes -- instead of PID 1
856 forking off the processes. By using scope units it is
857 possible for system services and applications to group their
858 own child processes (worker processes) in a powerful way
859 which then maybe used to organize them, or kill them
860 together, or apply resource limits on them.
862 Slice units may be used to partition system resources in an
863 hierarchial fashion and then assign other units to them. By
864 default there are now three slices: system.slice (for all
865 system services), user.slice (for all user sessions),
866 machine.slice (for VMs and containers).
868 Slices and scopes have been introduced primarily in
869 context of the work to move cgroup handling to a
870 single-writer scheme, where only PID 1
871 creates/removes/manages cgroups.
873 * There's a new concept of "transient" units. In contrast to
874 normal units these units are created via an API at runtime,
875 not from configuration from disk. More specifically this
876 means it is now possible to run arbitrary programs as
877 independent services, with all execution parameters passed
878 in via bus APIs rather than read from disk. Transient units
879 make systemd substantially more dynamic then it ever was,
880 and useful as a general batch manager.
882 * logind has been updated to make use of scope and slice units
883 for managing user sessions. As a user logs in he will get
884 his own private slice unit, to which all sessions are added
885 as scope units. We also added support for automatically
886 adding an instance of user@.service for the user into the
887 slice. Effectively logind will no longer create cgroup
888 hierarchies on its own now, it will defer entirely to PID 1
889 for this by means of scope, service and slice units. Since
890 user sessions this way become entities managed by PID 1
891 the output of "systemctl" is now a lot more comprehensive.
893 * A new mini-daemon "systemd-machined" has been added which
894 may be used by virtualization managers to register local
895 VMs/containers. nspawn has been updated accordingly, and
896 libvirt will be updated shortly. machined will collect a bit
897 of meta information about the VMs/containers, and assign
898 them their own scope unit (see above). The collected
899 meta-data is then made available via the "machinectl" tool,
900 and exposed in "ps" and similar tools. machined/machinectl
901 is compile-time optional.
903 * As discussed earlier, the low-level cgroup configuration
904 options ControlGroup=, ControlGroupModify=,
905 ControlGroupPersistent=, ControlGroupAttribute= have been
906 removed. Please use high-level attribute settings instead as
909 * A new bus call SetUnitProperties() has been added to alter
910 various runtime parameters of a unit. This is primarily
911 useful to alter cgroup parameters dynamically in a nice way,
912 but will be extended later on to make more properties
913 modifiable at runtime. systemctl gained a new set-properties
914 command that wraps this call.
916 * A new tool "systemd-run" has been added which can be used to
917 run arbitrary command lines as transient services or scopes,
918 while configuring a number of settings via the command
919 line. This tool is currently very basic, however already
920 very useful. We plan to extend this tool to even allow
921 queuing of execution jobs with time triggers from the
922 command line, similar in fashion to "at".
924 * nspawn will now inform the user explicitly that kernels with
925 audit enabled break containers, and suggest the user to turn
928 * Support for detecting the IMA and AppArmor security
929 frameworks with ConditionSecurity= has been added.
931 * journalctl gained a new "-k" switch for showing only kernel
932 messages, mimicking dmesg output; in addition to "--user"
933 and "--system" switches for showing only user's own logs
936 * systemd-delta can now show information about drop-in
937 snippets extending unit files.
939 * libsystemd-bus has been substantially updated but is still
940 not available as public API.
942 * systemd will now look for the "debug" argument on the kernel
943 command line and enable debug logging, similar to
944 "systemd.log_level=debug" already did before.
946 * "systemctl set-default", "systemctl get-default" has been
947 added to configure the default.target symlink, which
948 controls what to boot into by default.
950 * "systemctl set-log-level" has been added as a convenient
951 way to raise and lower systemd logging threshold.
953 * "systemd-analyze plot" will now show the time the various
954 generators needed for execution, as well as information
955 about the unit file loading.
957 * libsystemd-journal gained a new sd_journal_open_files() call
958 for opening specific journal files. journactl also gained a
959 new switch to expose this new functionality. Previously we
960 only supported opening all files from a directory, or all
961 files from the system, as opening individual files only is
962 racy due to journal file rotation.
964 * systemd gained the new DefaultEnvironment= setting in
965 /etc/systemd/system.conf to set environment variables for
968 * If a privileged process logs a journal message with the
969 OBJECT_PID= field set, then journald will automatically
970 augment this with additional OBJECT_UID=, OBJECT_GID=,
971 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
972 system services want to log events about specific client
973 processes. journactl/systemctl has been updated to make use
974 of this information if all log messages regarding a specific
977 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
978 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
979 Reisner, David Coppa, David King, David Strauss, Eelco
980 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
981 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
982 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
983 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
984 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
985 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
986 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
987 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
988 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
993 * The Python bindings gained some minimal support for the APIs
994 exposed by libsystemd-logind.
996 * ConditionSecurity= gained support for detecting SMACK. Since
997 this condition already supports SELinux and AppArmor we only
998 miss IMA for this. Patches welcome!
1000 Contributions from: Karol Lewandowski, Lennart Poettering,
1001 Zbigniew Jędrzejewski-Szmek
1005 * systemd-nspawn will now create /etc/resolv.conf if
1006 necessary, before bind-mounting the host's file onto it.
1008 * systemd-nspawn will now store meta information about a
1009 container on the container's cgroup as extended attribute
1010 fields, including the root directory.
1012 * The cgroup hierarchy has been reworked in many ways. All
1013 objects any of the components systemd creates in the cgroup
1014 tree are now suffixed. More specifically, user sessions are
1015 now placed in cgroups suffixed with ".session", users in
1016 cgroups suffixed with ".user", and nspawn containers in
1017 cgroups suffixed with ".nspawn". Furthermore, all cgroup
1018 names are now escaped in a simple scheme to avoid collision
1019 of userspace object names with kernel filenames. This work
1020 is preparation for making these objects relocatable in the
1021 cgroup tree, in order to allow easy resource partitioning of
1022 these objects without causing naming conflicts.
1024 * systemctl list-dependencies gained the new switches
1025 --plain, --reverse, --after and --before.
1027 * systemd-inhibit now shows the process name of processes that
1028 have taken an inhibitor lock.
1030 * nss-myhostname will now also resolve "localhost"
1031 implicitly. This makes /etc/hosts an optional file and
1032 nicely handles that on IPv6 ::1 maps to both "localhost" and
1035 * libsystemd-logind.so gained a new call
1036 sd_get_machine_names() to enumerate running containers and
1037 VMs (currently only supported by very new libvirt and
1038 nspawn). sd_login_monitor can now be used to watch
1039 VMs/containers coming and going.
1041 * .include is not allowed recursively anymore, and only in
1042 unit files. Usually it is better to use drop-in snippets in
1043 .d/*.conf anyway, as introduced with systemd 198.
1045 * systemd-analyze gained a new "critical-chain" command that
1046 determines the slowest chain of units run during system
1047 boot-up. It is very useful for tracking down where
1048 optimizing boot time is the most beneficial.
1050 * systemd will no longer allow manipulating service paths in
1051 the name=systemd:/system cgroup tree using ControlGroup= in
1052 units. (But is still fine with it in all other dirs.)
1054 * There's a new systemd-nspawn@.service service file that may
1055 be used to easily run nspawn containers as system
1056 services. With the container's root directory in
1057 /var/lib/container/foobar it is now sufficient to run
1058 "systemctl start systemd-nspawn@foobar.service" to boot it.
1060 * systemd-cgls gained a new parameter "--machine" to list only
1061 the processes within a certain container.
1063 * ConditionSecurity= now can check for "apparmor". We still
1064 are lacking checks for SMACK and IMA for this condition
1065 check though. Patches welcome!
1067 * A new configuration file /etc/systemd/sleep.conf has been
1068 added that may be used to configure which kernel operation
1069 systemd is supposed to execute when "suspend", "hibernate"
1070 or "hybrid-sleep" is requested. This makes the new kernel
1071 "freeze" state accessible to the user.
1073 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
1074 the passed argument if applicable.
1076 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
1077 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
1078 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
1079 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
1080 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
1081 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
1082 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
1087 * The output of 'systemctl list-jobs' got some polishing. The
1088 '--type=' argument may now be passed more than once. A new
1089 command 'systemctl list-sockets' has been added which shows
1090 a list of kernel sockets systemd is listening on with the
1091 socket units they belong to, plus the units these socket
1094 * The experimental libsystemd-bus library got substantial
1095 updates to work in conjunction with the (also experimental)
1096 kdbus kernel project. It works well enough to exchange
1097 messages with some sophistication. Note that kdbus is not
1098 ready yet, and the library is mostly an elaborate test case
1099 for now, and not installable.
1101 * systemd gained a new unit 'systemd-static-nodes.service'
1102 that generates static device nodes earlier during boot, and
1103 can run in conjunction with udev.
1105 * libsystemd-login gained a new call sd_pid_get_user_unit()
1106 to retrieve the user systemd unit a process is running
1107 in. This is useful for systems where systemd is used as
1110 * systemd-nspawn now places all containers in the new /machine
1111 top-level cgroup directory in the name=systemd
1112 hierarchy. libvirt will soon do the same, so that we get a
1113 uniform separation of /system, /user and /machine for system
1114 services, user processes and containers/virtual
1115 machines. This new cgroup hierarchy is also useful to stick
1116 stable names to specific container instances, which can be
1117 recognized later this way (this name may be controlled
1118 via systemd-nspawn's new -M switch). libsystemd-login also
1119 gained a new call sd_pid_get_machine_name() to retrieve the
1120 name of the container/VM a specific process belongs to.
1122 * bootchart can now store its data in the journal.
1124 * libsystemd-journal gained a new call
1125 sd_journal_add_conjunction() for AND expressions to the
1126 matching logic. This can be used to express more complex
1127 logical expressions.
1129 * journactl can now take multiple --unit= and --user-unit=
1132 * The cryptsetup logic now understands the "luks.key=" kernel
1133 command line switch for specifying a file to read the
1134 decryption key from. Also, if a configured key file is not
1135 found the tool will now automatically fall back to prompting
1138 * Python systemd.journal module was updated to wrap recently
1139 added functions from libsystemd-journal. The interface was
1140 changed to bring the low level interface in s.j._Reader
1141 closer to the C API, and the high level interface in
1142 s.j.Reader was updated to wrap and convert all data about
1145 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
1146 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
1147 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
1148 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
1149 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
1150 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1154 * journalctl --update-catalog now understands a new --root=
1155 option to operate on catalogs found in a different root
1158 * During shutdown after systemd has terminated all running
1159 services a final killing loop kills all remaining left-over
1160 processes. We will now print the name of these processes
1161 when we send SIGKILL to them, since this usually indicates a
1164 * If /etc/crypttab refers to password files stored on
1165 configured mount points automatic dependencies will now be
1166 generated to ensure the specific mount is established first
1167 before the key file is attempted to be read.
1169 * 'systemctl status' will now show information about the
1170 network sockets a socket unit is listening on.
1172 * 'systemctl status' will also shown information about any
1173 drop-in configuration file for units. (Drop-In configuration
1174 files in this context are files such as
1175 /etc/systemd/systemd/foobar.service.d/*.conf)
1177 * systemd-cgtop now optionally shows summed up CPU times of
1178 cgroups. Press '%' while running cgtop to switch between
1179 percentage and absolute mode. This is useful to determine
1180 which cgroups use up the most CPU time over the entire
1181 runtime of the system. systemd-cgtop has also been updated
1182 to be 'pipeable' for processing with further shell tools.
1184 * 'hostnamectl set-hostname' will now allow setting of FQDN
1187 * The formatting and parsing of time span values has been
1188 changed. The parser now understands fractional expressions
1189 such as "5.5h". The formatter will now output fractional
1190 expressions for all time spans under 1min, i.e. "5.123456s"
1191 rather than "5s 123ms 456us". For time spans under 1s
1192 millisecond values are shown, for those under 1ms
1193 microsecond values are shown. This should greatly improve
1194 all time-related output of systemd.
1196 * libsystemd-login and libsystemd-journal gained new
1197 functions for querying the poll() events mask and poll()
1198 timeout value for integration into arbitrary event
1201 * localectl gained the ability to list available X11 keymaps
1202 (models, layouts, variants, options).
1204 * 'systemd-analyze dot' gained the ability to filter for
1205 specific units via shell-style globs, to create smaller,
1206 more useful graphs. I.e. it's now possible to create simple
1207 graphs of all the dependencies between only target units, or
1208 of all units that Avahi has dependencies with.
1210 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
1211 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
1212 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
1213 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
1214 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
1215 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
1216 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
1220 * The boot-time readahead implementation for rotating media
1221 will now read the read-ahead data in multiple passes which
1222 consist of all read requests made in equidistant time
1223 intervals. This means instead of strictly reading read-ahead
1224 data in its physical order on disk we now try to find a
1225 middle ground between physical and access time order.
1227 * /etc/os-release files gained a new BUILD_ID= field for usage
1228 on operating systems that provide continuous builds of OS
1231 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
1232 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
1233 William Douglas, Zbigniew Jędrzejewski-Szmek
1237 * systemd-python gained an API exposing libsystemd-daemon.
1239 * The SMACK setup logic gained support for uploading CIPSO
1242 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
1243 ReadOnlyDirectories= and InaccessibleDirectories= has
1244 changed. The private /tmp and /var/tmp directories are now
1245 shared by all processes of a service (which means
1246 ExecStartPre= may now leave data in /tmp that ExecStart= of
1247 the same service can still access). When a service is
1248 stopped its temporary directories are immediately deleted
1249 (normal clean-up with tmpfiles is still done in addition to
1252 * By default, systemd will now set a couple of sysctl
1253 variables in the kernel: the safe sysrq options are turned
1254 on, IP route verification is turned on, and source routing
1255 disabled. The recently added hardlink and softlink
1256 protection of the kernel is turned on. These settings should
1257 be reasonably safe, and good defaults for all new systems.
1259 * The predictable network naming logic may now be turned off
1260 with a new kernel command line switch: net.ifnames=0.
1262 * A new libsystemd-bus module has been added that implements a
1263 pretty complete D-Bus client library. For details see:
1265 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
1267 * journald will now explicitly flush the journal files to disk
1268 at the latest 5min after each write. The file will then also
1269 be marked offline until the next write. This should increase
1270 reliability in case of a crash. The synchronization delay
1271 can be configured via SyncIntervalSec= in journald.conf.
1273 * There's a new remote-fs-setup.target unit that can be used
1274 to pull in specific services when at least one remote file
1275 system is to be mounted.
1277 * There are new targets timers.target and paths.target as
1278 canonical targets to pull user timer and path units in
1279 from. This complements sockets.target with a similar
1280 purpose for socket units.
1282 * libudev gained a new call udev_device_set_attribute_value()
1283 to set sysfs attributes of a device.
1285 * The udev daemon now sets the default number of worker
1286 processes executed in parallel based on the number of available
1287 CPUs instead of the amount of available RAM. This is supposed
1288 to provide a more reliable default and limit a too aggressive
1289 paralellism for setups with 1000s of devices connected.
1291 Contributions from: Auke Kok, Colin Walters, Cristian
1292 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
1293 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
1294 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
1295 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
1296 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
1297 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
1298 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
1299 Zbigniew Jędrzejewski-Szmek
1303 * Configuration of unit files may now be extended via drop-in
1304 files without having to edit/override the unit files
1305 themselves. More specifically, if the administrator wants to
1306 change one value for a service file foobar.service he can
1307 now do so by dropping in a configuration snippet into
1308 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
1309 will load all these snippets and apply them on top of the
1310 main unit configuration file, possibly extending or
1311 overriding its settings. Using these drop-in snippets is
1312 generally nicer than the two earlier options for changing
1313 unit files locally: copying the files from
1314 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
1315 them there; or creating a new file in /etc/systemd/system/
1316 that incorporates the original one via ".include". Drop-in
1317 snippets into these .d/ directories can be placed in any
1318 directory systemd looks for units in, and the usual
1319 overriding semantics between /usr/lib, /etc and /run apply
1322 * Most unit file settings which take lists of items can now be
1323 reset by assigning the empty string to them. For example,
1324 normally, settings such as Environment=FOO=BAR append a new
1325 environment variable assignment to the environment block,
1326 each time they are used. By assigning Environment= the empty
1327 string the environment block can be reset to empty. This is
1328 particularly useful with the .d/*.conf drop-in snippets
1329 mentioned above, since this adds the ability to reset list
1330 settings from vendor unit files via these drop-ins.
1332 * systemctl gained a new "list-dependencies" command for
1333 listing the dependencies of a unit recursively.
1335 * Inhibitors are now honored and listed by "systemctl
1336 suspend", "systemctl poweroff" (and similar) too, not only
1337 GNOME. These commands will also list active sessions by
1340 * Resource limits (as exposed by the various control group
1341 controllers) can now be controlled dynamically at runtime
1342 for all units. More specifically, you can now use a command
1343 like "systemctl set-cgroup-attr foobar.service cpu.shares
1344 2000" to alter the CPU shares a specific service gets. These
1345 settings are stored persistently on disk, and thus allow the
1346 administrator to easily adjust the resource usage of
1347 services with a few simple commands. This dynamic resource
1348 management logic is also available to other programs via the
1349 bus. Almost any kernel cgroup attribute and controller is
1352 * systemd-vconsole-setup will now copy all font settings to
1353 all allocated VTs, where it previously applied them only to
1356 * libsystemd-login gained the new sd_session_get_tty() API
1359 * This release drops support for a few legacy or
1360 distribution-specific LSB facility names when parsing init
1361 scripts: $x-display-manager, $mail-transfer-agent,
1362 $mail-transport-agent, $mail-transfer-agent, $smtp,
1363 $null. Also, the mail-transfer-agent.target unit backing
1364 this has been removed. Distributions which want to retain
1365 compatibility with this should carry the burden for
1366 supporting this themselves and patch support for these back
1367 in, if they really need to. Also, the facilities $syslog and
1368 $local_fs are now ignored, since systemd does not support
1369 early-boot LSB init scripts anymore, and these facilities
1370 are implied anyway for normal services. syslog.target has
1373 * There are new bus calls on PID1's Manager object for
1374 cancelling jobs, and removing snapshot units. Previously,
1375 both calls were only available on the Job and Snapshot
1378 * systemd-journal-gatewayd gained SSL support.
1380 * The various "environment" files, such as /etc/locale.conf
1381 now support continuation lines with a backslash ("\") as
1382 last character in the line, similar in style (but different)
1383 to how this is supported in shells.
1385 * For normal user processes the _SYSTEMD_USER_UNIT= field is
1386 now implicitly appended to every log entry logged. systemctl
1387 has been updated to filter by this field when operating on a
1388 user systemd instance.
1390 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
1391 CAP_AUDIT_CONTROL capabilities to the capabilities set for
1392 the container. This makes it easier to boot unmodified
1393 Fedora systems in a container, which however still requires
1394 audit=0 to be passed on the kernel command line. Auditing in
1395 kernel and userspace is unfortunately still too broken in
1396 context of containers, hence we recommend compiling it out
1397 of the kernel or using audit=0. Hopefully this will be fixed
1398 one day for good in the kernel.
1400 * nspawn gained the new --bind= and --bind-ro= parameters to
1401 bind mount specific directories from the host into the
1404 * nspawn will now mount its own devpts file system instance
1405 into the container, in order not to leak pty devices from
1406 the host into the container.
1408 * systemd will now read the firmware boot time performance
1409 information from the EFI variables, if the used boot loader
1410 supports this, and takes it into account for boot performance
1411 analysis via "systemd-analyze". This is currently supported
1412 only in conjunction with Gummiboot, but could be supported
1413 by other boot loaders too. For details see:
1415 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
1417 * A new generator has been added that automatically mounts the
1418 EFI System Partition (ESP) to /boot, if that directory
1419 exists, is empty, and no other file system has been
1420 configured to be mounted there.
1422 * logind will now send out PrepareForSleep(false) out
1423 unconditionally, after coming back from suspend. This may be
1424 used by applications as asynchronous notification for
1425 system resume events.
1427 * "systemctl unlock-sessions" has been added, that allows
1428 unlocking the screens of all user sessions at once, similar
1429 how "systemctl lock-sessions" already locked all users
1430 sessions. This is backed by a new D-Bus call UnlockSessions().
1432 * "loginctl seat-status" will now show the master device of a
1433 seat. (i.e. the device of a seat that needs to be around for
1434 the seat to be considered available, usually the graphics
1437 * tmpfiles gained a new "X" line type, that allows
1438 configuration of files and directories (with wildcards) that
1439 shall be excluded from automatic cleanup ("aging").
1441 * udev default rules set the device node permissions now only
1442 at "add" events, and do not change them any longer with a
1443 later "change" event.
1445 * The log messages for lid events and power/sleep keypresses
1446 now carry a message ID.
1448 * We now have a substantially larger unit test suite, but this
1449 continues to be work in progress.
1451 * udevadm hwdb gained a new --root= parameter to change the
1452 root directory to operate relative to.
1454 * logind will now issue a background sync() request to the kernel
1455 early at shutdown, so that dirty buffers are flushed to disk early
1456 instead of at the last moment, in order to optimize shutdown
1459 * A new bootctl tool has been added that is an interface for
1460 certain boot loader operations. This is currently a preview
1461 and is likely to be extended into a small mechanism daemon
1462 like timedated, localed, hostnamed, and can be used by
1463 graphical UIs to enumerate available boot options, and
1464 request boot into firmware operations.
1466 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
1467 the rest of the package. It also has been updated to work
1468 correctly in initrds.
1470 * Policykit previously has been runtime optional, and is now
1471 also compile time optional via a configure switch.
1473 * systemd-analyze has been reimplemented in C. Also "systemctl
1474 dot" has moved into systemd-analyze.
1476 * "systemctl status" with no further parameters will now print
1477 the status of all active or failed units.
1479 * Operations such as "systemctl start" can now be executed
1480 with a new mode "--irreversible" which may be used to queue
1481 operations that cannot accidentally be reversed by a later
1482 job queuing. This is by default used to make shutdown
1483 requests more robust.
1485 * The Python API of systemd now gained a new module for
1486 reading journal files.
1488 * A new tool kernel-install has been added that can install
1489 kernel images according to the Boot Loader Specification:
1491 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
1493 * Boot time console output has been improved to provide
1494 animated boot time output for hanging jobs.
1496 * A new tool systemd-activate has been added which can be used
1497 to test socket activation with, directly from the command
1498 line. This should make it much easier to test and debug
1499 socket activation in daemons.
1501 * journalctl gained a new "--reverse" (or -r) option to show
1502 journal output in reverse order (i.e. newest line first).
1504 * journalctl gained a new "--pager-end" (or -e) option to jump
1505 to immediately jump to the end of the journal in the
1506 pager. This is only supported in conjunction with "less".
1508 * journalctl gained a new "--user-unit=" option, that works
1509 similar to "--unit=" but filters for user units rather than
1512 * A number of unit files to ease adoption of systemd in
1513 initrds has been added. This moves some minimal logic from
1514 the various initrd implementations into systemd proper.
1516 * The journal files are now owned by a new group
1517 "systemd-journal", which exists specifically to allow access
1518 to the journal, and nothing else. Previously, we used the
1519 "adm" group for that, which however possibly covers more
1520 than just journal/log file access. This new group is now
1521 already used by systemd-journal-gatewayd to ensure this
1522 daemon gets access to the journal files and as little else
1523 as possible. Note that "make install" will also set FS ACLs
1524 up for /var/log/journal to give "adm" and "wheel" read
1525 access to it, in addition to "systemd-journal" which owns
1526 the journal files. We recommend that packaging scripts also
1527 add read access to "adm" + "wheel" to /var/log/journal, and
1528 all existing/future journal files. To normal users and
1529 administrators little changes, however packagers need to
1530 ensure to create the "systemd-journal" system group at
1531 package installation time.
1533 * The systemd-journal-gatewayd now runs as unprivileged user
1534 systemd-journal-gateway:systemd-journal-gateway. Packaging
1535 scripts need to create these system user/group at
1538 * timedated now exposes a new boolean property CanNTP that
1539 indicates whether a local NTP service is available or not.
1541 * systemd-detect-virt will now also detect xen PVs
1543 * The pstore file system is now mounted by default, if it is
1546 * In addition to the SELinux and IMA policies we will now also
1547 load SMACK policies at early boot.
1549 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
1550 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
1551 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
1552 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
1553 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
1554 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
1555 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
1556 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
1557 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
1558 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
1559 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
1560 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
1561 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
1562 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
1566 * Timer units now support calendar time events in addition to
1567 monotonic time events. That means you can now trigger a unit
1568 based on a calendar time specification such as "Thu,Fri
1569 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
1570 or fifth day of any month of the year 2013, given that it is
1571 a thursday or friday. This brings timer event support
1572 considerably closer to cron's capabilities. For details on
1573 the supported calendar time specification language see
1576 * udev now supports a number of different naming policies for
1577 network interfaces for predictable names, and a combination
1578 of these policies is now the default. Please see this wiki
1579 document for details:
1581 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
1583 * Auke Kok's bootchart implementation has been added to the
1584 systemd tree. It's an optional component that can graph the
1585 boot in quite some detail. It's one of the best bootchart
1586 implementations around and minimal in its code and
1589 * nss-myhostname has been integrated into the systemd source
1590 tree. nss-myhostname guarantees that the local hostname
1591 always stays resolvable via NSS. It has been a weak
1592 requirement of systemd-hostnamed since a long time, and
1593 since its code is actually trivial we decided to just
1594 include it in systemd's source tree. It can be turned off
1595 with a configure switch.
1597 * The read-ahead logic is now capable of properly detecting
1598 whether a btrfs file system is on SSD or rotating media, in
1599 order to optimize the read-ahead scheme. Previously, it was
1600 only capable of detecting this on traditional file systems
1603 * In udev, additional device properties are now read from the
1604 IAB in addition to the OUI database. Also, Bluetooth company
1605 identities are attached to the devices as well.
1607 * In service files %U may be used as specifier that is
1608 replaced by the configured user name of the service.
1610 * nspawn may now be invoked without a controlling TTY. This
1611 makes it suitable for invocation as its own service. This
1612 may be used to set up a simple containerized server system
1613 using only core OS tools.
1615 * systemd and nspawn can now accept socket file descriptors
1616 when they are started for socket activation. This enables
1617 implementation of socket activated nspawn
1618 containers. i.e. think about autospawning an entire OS image
1619 when the first SSH or HTTP connection is received. We expect
1620 that similar functionality will also be added to libvirt-lxc
1623 * journalctl will now suppress ANSI color codes when
1624 presenting log data.
1626 * systemctl will no longer show control group information for
1627 a unit if a the control group is empty anyway.
1629 * logind can now automatically suspend/hibernate/shutdown the
1632 * /etc/machine-info and hostnamed now also expose the chassis
1633 type of the system. This can be used to determine whether
1634 the local system is a laptop, desktop, handset or
1635 tablet. This information may either be configured by the
1636 user/vendor or is automatically determined from ACPI and DMI
1637 information if possible.
1639 * A number of PolicyKit actions are now bound together with
1640 "imply" rules. This should simplify creating UIs because
1641 many actions will now authenticate similar ones as well.
1643 * Unit files learnt a new condition ConditionACPower= which
1644 may be used to conditionalize a unit depending on whether an
1645 AC power source is connected or not, of whether the system
1646 is running on battery power.
1648 * systemctl gained a new "is-failed" verb that may be used in
1649 shell scripts and suchlike to check whether a specific unit
1650 is in the "failed" state.
1652 * The EnvironmentFile= setting in unit files now supports file
1653 globbing, and can hence be used to easily read a number of
1654 environment files at once.
1656 * systemd will no longer detect and recognize specific
1657 distributions. All distribution-specific #ifdeffery has been
1658 removed, systemd is now fully generic and
1659 distribution-agnostic. Effectively, not too much is lost as
1660 a lot of the code is still accessible via explicit configure
1661 switches. However, support for some distribution specific
1662 legacy configuration file formats has been dropped. We
1663 recommend distributions to simply adopt the configuration
1664 files everybody else uses now and convert the old
1665 configuration from packaging scripts. Most distributions
1666 already did that. If that's not possible or desirable,
1667 distributions are welcome to forward port the specific
1668 pieces of code locally from the git history.
1670 * When logging a message about a unit systemd will now always
1671 log the unit name in the message meta data.
1673 * localectl will now also discover system locale data that is
1674 not stored in locale archives, but directly unpacked.
1676 * logind will no longer unconditionally use framebuffer
1677 devices as seat masters, i.e. as devices that are required
1678 to be existing before a seat is considered preset. Instead,
1679 it will now look for all devices that are tagged as
1680 "seat-master" in udev. By default framebuffer devices will
1681 be marked as such, but depending on local systems other
1682 devices might be marked as well. This may be used to
1683 integrate graphics cards using closed source drivers (such
1684 as NVidia ones) more nicely into logind. Note however, that
1685 we recommend using the open source NVidia drivers instead,
1686 and no udev rules for the closed-source drivers will be
1687 shipped from us upstream.
1689 Contributions from: Adam Williamson, Alessandro Crismani, Auke
1690 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
1691 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
1692 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
1693 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
1694 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
1695 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
1696 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
1697 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
1698 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
1699 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
1700 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
1705 * udev gained support for loading additional device properties
1706 from an indexed database that is keyed by vendor/product IDs
1707 and similar device identifiers. For the beginning this
1708 "hwdb" is populated with data from the well-known PCI and
1709 USB database, but also includes PNP, ACPI and OID data. In
1710 the longer run this indexed database shall grow into
1711 becoming the one central database for non-essential
1712 userspace device metadata. Previously, data from the PCI/USB
1713 database was only attached to select devices, since the
1714 lookup was a relatively expensive operation due to O(n) time
1715 complexity (with n being the number of entries in the
1716 database). Since this is now O(1), we decided to add in this
1717 data for all devices where this is available, by
1718 default. Note that the indexed database needs to be rebuilt
1719 when new data files are installed. To achieve this you need
1720 to update your packaging scripts to invoke "udevadm hwdb
1721 --update" after installation of hwdb data files. For
1722 RPM-based distributions we introduced the new
1723 %udev_hwdb_update macro for this purpose.
1725 * The Journal gained support for the "Message Catalog", an
1726 indexed database to link up additional information with
1727 journal entries. For further details please check:
1729 http://www.freedesktop.org/wiki/Software/systemd/catalog
1731 The indexed message catalog database also needs to be
1732 rebuilt after installation of message catalog files. Use
1733 "journalctl --update-catalog" for this. For RPM-based
1734 distributions we introduced the %journal_catalog_update
1735 macro for this purpose.
1737 * The Python Journal bindings gained support for the standard
1738 Python logging framework.
1740 * The Journal API gained new functions for checking whether
1741 the underlying file system of a journal file is capable of
1742 properly reporting file change notifications, or whether
1743 applications that want to reflect journal changes "live"
1744 need to recheck journal files continuously in appropriate
1747 * It is now possible to set the "age" field for tmpfiles
1748 entries to 0, indicating that files matching this entry
1749 shall always be removed when the directories are cleaned up.
1751 * coredumpctl gained a new "gdb" verb which invokes gdb
1752 right-away on the selected coredump.
1754 * There's now support for "hybrid sleep" on kernels that
1755 support this, in addition to "suspend" and "hibernate". Use
1756 "systemctl hybrid-sleep" to make use of this.
1758 * logind's HandleSuspendKey= setting (and related settings)
1759 now gained support for a new "lock" setting to simply
1760 request the screen lock on all local sessions, instead of
1761 actually executing a suspend or hibernation.
1763 * systemd will now mount the EFI variables file system by
1766 * Socket units now gained support for configuration of the
1767 SMACK security label.
1769 * timedatectl will now output the time of the last and next
1770 daylight saving change.
1772 * We dropped support for various legacy and distro-specific
1773 concepts, such as insserv, early-boot SysV services
1774 (i.e. those for non-standard runlevels such as 'b' or 'S')
1775 or ArchLinux /etc/rc.conf support. We recommend the
1776 distributions who still need support this to either continue
1777 to maintain the necessary patches downstream, or find a
1778 different solution. (Talk to us if you have questions!)
1780 * Various systemd components will now bypass PolicyKit checks
1781 for root and otherwise handle properly if PolicyKit is not
1782 found to be around. This should fix most issues for
1783 PolicyKit-less systems. Quite frankly this should have been
1784 this way since day one. It is absolutely our intention to
1785 make systemd work fine on PolicyKit-less systems, and we
1786 consider it a bug if something doesn't work as it should if
1787 PolicyKit is not around.
1789 * For embedded systems it is now possible to build udev and
1790 systemd without blkid and/or kmod support.
1792 * "systemctl switch-root" is now capable of switching root
1793 more than once. I.e. in addition to transitions from the
1794 initrd to the host OS it is now possible to transition to
1795 further OS images from the host. This is useful to implement
1796 offline updating tools.
1798 * Various other additions have been made to the RPM macros
1799 shipped with systemd. Use %udev_rules_update() after
1800 installing new udev rules files. %_udevhwdbdir,
1801 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
1802 %_sysctldir are now available which resolve to the right
1803 directories for packages to place various data files in.
1805 * journalctl gained the new --full switch (in addition to
1806 --all, to disable ellipsation for long messages.
1808 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
1809 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
1810 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
1811 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
1812 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
1813 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
1814 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
1815 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
1816 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
1820 * journalctl gained new --since= and --until= switches to
1821 filter by time. It also now supports nice filtering for
1822 units via --unit=/-u.
1824 * Type=oneshot services may use ExecReload= and do the
1827 * The journal daemon now supports time-based rotation and
1828 vacuuming, in addition to the usual disk-space based
1831 * The journal will now index the available field values for
1832 each field name. This enables clients to show pretty drop
1833 downs of available match values when filtering. The bash
1834 completion of journalctl has been updated
1835 accordingly. journalctl gained a new switch -F to list all
1836 values a certain field takes in the journal database.
1838 * More service events are now written as structured messages
1839 to the journal, and made recognizable via message IDs.
1841 * The timedated, localed and hostnamed mini-services which
1842 previously only provided support for changing time, locale
1843 and hostname settings from graphical DEs such as GNOME now
1844 also have a minimal (but very useful) text-based client
1845 utility each. This is probably the nicest way to changing
1846 these settings from the command line now, especially since
1847 it lists available options and is fully integrated with bash
1850 * There's now a new tool "systemd-coredumpctl" to list and
1851 extract coredumps from the journal.
1853 * We now install a README each in /var/log/ and
1854 /etc/rc.d/init.d explaining where the system logs and init
1855 scripts went. This hopefully should help folks who go to
1856 that dirs and look into the otherwise now empty void and
1857 scratch their heads.
1859 * When user-services are invoked (by systemd --user) the
1860 $MANAGERPID env var is set to the PID of systemd.
1862 * SIGRTMIN+24 when sent to a --user instance will now result
1863 in immediate termination of systemd.
1865 * gatewayd received numerous feature additions such as a
1866 "follow" mode, for live syncing and filtering.
1868 * browse.html now allows filtering and showing detailed
1869 information on specific entries. Keyboard navigation and
1870 mouse screen support has been added.
1872 * gatewayd/journalctl now supports HTML5/JSON
1873 Server-Sent-Events as output.
1875 * The SysV init script compatibility logic will now
1876 heuristically determine whether a script supports the
1877 "reload" verb, and only then make this available as
1880 * "systemctl status --follow" has been removed, use "journalctl
1883 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
1884 have been removed since they are hardly useful to be
1887 * And I'd like to take the opportunity to specifically mention
1888 Zbigniew for his great contributions. Zbigniew, you rock!
1890 Contributions from: Andrew Eikum, Christian Hesse, Colin
1891 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
1892 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
1893 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
1894 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
1895 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
1896 Jędrzejewski-Szmek, Сковорода Никита Андреевич
1900 * If /etc/vconsole.conf is non-existent or empty we will no
1901 longer load any console font or key map at boot by
1902 default. Instead the kernel defaults will be left
1903 intact. This is definitely the right thing to do, as no
1904 configuration should mean no configuration, and hard-coding
1905 font names that are different on all archs is probably a bad
1906 idea. Also, the kernel default key map and font should be
1907 good enough for most cases anyway, and mostly identical to
1908 the userspace fonts/key maps we previously overloaded them
1909 with. If distributions want to continue to default to a
1910 non-kernel font or key map they should ship a default
1911 /etc/vconsole.conf with the appropriate contents.
1913 Contributions from: Colin Walters, Daniel J Walsh, Dave
1914 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
1915 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1919 * journalctl gained a new --cursor= switch to show entries
1920 starting from the specified location in the journal.
1922 * We now enforce a size limit on journal entry fields exported
1923 with "-o json" in journalctl. Fields larger than 4K will be
1924 assigned null. This can be turned off with --all.
1926 * An (optional) journal gateway daemon is now available as
1927 "systemd-journal-gatewayd.service". This service provides
1928 access to the journal via HTTP and JSON. This functionality
1929 will be used to implement live log synchronization in both
1930 pull and push modes, but has various other users too, such
1931 as easy log access for debugging of embedded devices. Right
1932 now it is already useful to retrieve the journal via HTTP:
1934 # systemctl start systemd-journal-gatewayd.service
1935 # wget http://localhost:19531/entries
1937 This will download the journal contents in a
1938 /var/log/messages compatible format. The same as JSON:
1940 # curl -H"Accept: application/json" http://localhost:19531/entries
1942 This service is also accessible via a web browser where a
1943 single static HTML5 app is served that uses the JSON logic
1944 to enable the user to do some basic browsing of the
1945 journal. This will be extended later on. Here's an example
1946 screenshot of this app in its current state:
1948 http://0pointer.de/public/journal-gatewayd
1950 Contributions from: Kay Sievers, Lennart Poettering, Robert
1951 Milasan, Tom Gundersen
1955 * The bash completion logic is now available for journalctl
1958 * We don't mount the "cpuset" controller anymore together with
1959 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
1960 started if no parameters are assigned to it. "cpuset" hence
1961 broke code that assumed it it could create "cpu" groups and
1964 * journalctl -f will now subscribe to terminal size changes,
1965 and line break accordingly.
1967 Contributions from: Dave Reisner, Kay Sievers, Lennart
1968 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
1972 * nspawn will now create a symlink /etc/localtime in the
1973 container environment, copying the host's timezone
1974 setting. Previously this has been done via a bind mount, but
1975 since symlinks cannot be bind mounted this has now been
1976 changed to create/update the appropriate symlink.
1978 * journalctl -n's line number argument is now optional, and
1979 will default to 10 if omitted.
1981 * journald will now log the maximum size the journal files may
1982 take up on disk. This is particularly useful if the default
1983 built-in logic of determining this parameter from the file
1984 system size is used. Use "systemctl status
1985 systemd-journald.service" to see this information.
1987 * The multi-seat X wrapper tool has been stripped down. As X
1988 is now capable of enumerating graphics devices via udev in a
1989 seat-aware way the wrapper is not strictly necessary
1990 anymore. A stripped down temporary stop-gap is still shipped
1991 until the upstream display managers have been updated to
1992 fully support the new X logic. Expect this wrapper to be
1993 removed entirely in one of the next releases.
1995 * HandleSleepKey= in logind.conf has been split up into
1996 HandleSuspendKey= and HandleHibernateKey=. The old setting
1997 is not available anymore. X11 and the kernel are
1998 distuingishing between these keys and we should too. This
1999 also means the inhibition lock for these keys has been split
2002 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
2003 Poettering, Lukas Nykryn, Václav Pavlín
2007 * Whenever a unit changes state we'll now log this to the
2008 journal and show along the unit's own log output in
2011 * ConditionPathIsMountPoint= can now properly detect bind
2012 mount points too. (Previously, a bind mount of one file
2013 system to another place in the same file system could not be
2014 detected as mount, since they shared struct stat's st_dev
2017 * We will now mount the cgroup controllers cpu, cpuacct,
2018 cpuset and the controllers net_cls, net_prio together by
2021 * nspawn containers will now have a virtualized boot
2022 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
2023 over with a randomized ID at container initialization). This
2024 has the effect of making "journalctl -b" do the right thing
2027 * The JSON output journal serialization has been updated not
2028 to generate "endless" list objects anymore, but rather one
2029 JSON object per line. This is more in line how most JSON
2030 parsers expect JSON objects. The new output mode
2031 "json-pretty" has been added to provide similar output, but
2032 neatly aligned for readability by humans.
2034 * We dropped all explicit sync() invocations in the shutdown
2035 code. The kernel does this implicitly anyway in the kernel
2036 reboot() syscall. halt(8)'s -n option is now a compatibility
2039 * We now support virtualized reboot() in containers, as
2040 supported by newer kernels. We will fall back to exit() if
2041 CAP_SYS_REBOOT is not available to the container. Also,
2042 nspawn makes use of this now and will actually reboot the
2043 container if the containerized OS asks for that.
2045 * journalctl will only show local log output by default
2046 now. Use --merge (-m) to show remote log output, too.
2048 * libsystemd-journal gained the new sd_journal_get_usage()
2049 call to determine the current disk usage of all journal
2050 files. This is exposed in the new "journalctl --disk-usage"
2053 * journald gained a new configuration setting SplitMode= in
2054 journald.conf which may be used to control how user journals
2055 are split off. See journald.conf(5) for details.
2057 * A new condition type ConditionFileNotEmpty= has been added.
2059 * tmpfiles' "w" lines now support file globbing, to write
2060 multiple files at once.
2062 * We added Python bindings for the journal submission
2063 APIs. More Python APIs for a number of selected APIs will
2064 likely follow. Note that we intend to add native bindings
2065 only for the Python language, as we consider it common
2066 enough to deserve bindings shipped within systemd. There are
2067 various projects outside of systemd that provide bindings
2068 for languages such as PHP or Lua.
2070 * Many conditions will now resolve specifiers such as %i. In
2071 addition, PathChanged= and related directives of .path units
2072 now support specifiers as well.
2074 * There's now a new RPM macro definition for the system preset
2077 * journald will now warn if it can't forward a message to the
2078 syslog daemon because it's socket is full.
2080 * timedated will no longer write or process /etc/timezone,
2081 except on Debian. As we do not support late mounted /usr
2082 anymore /etc/localtime always being a symlink is now safe,
2083 and hence the information in /etc/timezone is not necessary
2086 * logind will now always reserve one VT for a text getty (VT6
2087 by default). Previously if more than 6 X sessions where
2088 started they took up all the VTs with auto-spawned gettys,
2089 so that no text gettys were available anymore.
2091 * udev will now automatically inform the btrfs kernel logic
2092 about btrfs RAID components showing up. This should make
2093 simple hotplug based btrfs RAID assembly work.
2095 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
2096 (but not for its children which will stay at the kernel
2097 default). This should allow setups with a lot more listening
2100 * systemd will now always pass the configured timezone to the
2101 kernel at boot. timedated will do the same when the timezone
2104 * logind's inhibition logic has been updated. By default,
2105 logind will now handle the lid switch, the power and sleep
2106 keys all the time, even in graphical sessions. If DEs want
2107 to handle these events on their own they should take the new
2108 handle-power-key, handle-sleep-key and handle-lid-switch
2109 inhibitors during their runtime. A simple way to achiveve
2110 that is to invoke the DE wrapped in an invocation of:
2112 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
2114 * Access to unit operations is now checked via SELinux taking
2115 the unit file label and client process label into account.
2117 * systemd will now notify the administrator in the journal
2118 when he over-mounts a non-empty directory.
2120 * There are new specifiers that are resolved in unit files,
2121 for the host name (%H), the machine ID (%m) and the boot ID
2124 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
2125 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
2126 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
2127 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
2128 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
2129 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
2130 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
2134 * Support for reading structured kernel messages from
2135 /dev/kmsg has now been added and is enabled by default.
2137 * Support for reading kernel messages from /proc/kmsg has now
2138 been removed. If you want kernel messages in the journal
2139 make sure to run a recent kernel (>= 3.5) that supports
2140 reading structured messages from /dev/kmsg (see
2141 above). /proc/kmsg is now exclusive property of classic
2142 syslog daemons again.
2144 * The libudev API gained the new
2145 udev_device_new_from_device_id() call.
2147 * The logic for file system namespace (ReadOnlyDirectory=,
2148 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
2149 require pivot_root() anymore. This means fewer temporary
2150 directories are created below /tmp for this feature.
2152 * nspawn containers will now see and receive all submounts
2153 made on the host OS below the root file system of the
2156 * Forward Secure Sealing is now supported for Journal files,
2157 which provide cryptographical sealing of journal files so
2158 that attackers cannot alter log history anymore without this
2159 being detectable. Lennart will soon post a blog story about
2160 this explaining it in more detail.
2162 * There are two new service settings RestartPreventExitStatus=
2163 and SuccessExitStatus= which allow configuration of exit
2164 status (exit code or signal) which will be excepted from the
2165 restart logic, resp. consider successful.
2167 * journalctl gained the new --verify switch that can be used
2168 to check the integrity of the structure of journal files and
2169 (if Forward Secure Sealing is enabled) the contents of
2172 * nspawn containers will now be run with /dev/stdin, /dev/fd/
2173 and similar symlinks pre-created. This makes running shells
2174 as container init process a lot more fun.
2176 * The fstab support can now handle PARTUUID= and PARTLABEL=
2179 * A new ConditionHost= condition has been added to match
2180 against the hostname (with globs) and machine ID. This is
2181 useful for clusters where a single OS image is used to
2182 provision a large number of hosts which shall run slightly
2183 different sets of services.
2185 * Services which hit the restart limit will now be placed in a
2188 Contributions from: Bertram Poettering, Dave Reisner, Huang
2189 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
2190 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
2194 * When running in --user mode systemd will now become a
2195 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
2196 tree a lot more organized.
2198 * A new PartOf= unit dependency type has been introduced that
2199 may be used to group services in a natural way.
2201 * "systemctl enable" may now be used to enable instances of
2204 * journalctl now prints error log levels in red, and
2205 warning/notice log levels in bright white. It also supports
2206 filtering by log level now.
2208 * cgtop gained a new -n switch (similar to top), to configure
2209 the maximum number of iterations to run for. It also gained
2210 -b, to run in batch mode (accepting no input).
2212 * The suffix ".service" may now be omitted on most systemctl
2213 command lines involving service unit names.
2215 * There's a new bus call in logind to lock all sessions, as
2216 well as a loginctl verb for it "lock-sessions".
2218 * libsystemd-logind.so gained a new call sd_journal_perror()
2219 that works similar to libc perror() but logs to the journal
2220 and encodes structured information about the error number.
2222 * /etc/crypttab entries now understand the new keyfile-size=
2225 * shutdown(8) now can send a (configurable) wall message when
2226 a shutdown is cancelled.
2228 * The mount propagation mode for the root file system will now
2229 default to "shared", which is useful to make containers work
2230 nicely out-of-the-box so that they receive new mounts from
2231 the host. This can be undone locally by running "mount
2232 --make-rprivate /" if needed.
2234 * The prefdm.service file has been removed. Distributions
2235 should maintain this unit downstream if they intend to keep
2236 it around. However, we recommend writing normal unit files
2237 for display managers instead.
2239 * Since systemd is a crucial part of the OS we will now
2240 default to a number of compiler switches that improve
2241 security (hardening) such as read-only relocations, stack
2242 protection, and suchlike.
2244 * The TimeoutSec= setting for services is now split into
2245 TimeoutStartSec= and TimeoutStopSec= to allow configuration
2246 of individual time outs for the start and the stop phase of
2249 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
2250 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
2251 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
2252 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
2253 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
2254 Gundersen, Zbigniew Jędrzejewski-Szmek
2258 * The journal and id128 C APIs are now fully documented as man
2261 * Extra safety checks have been added when transitioning from
2262 the initial RAM disk to the main system to avoid accidental
2265 * /etc/crypttab entries now understand the new keyfile-offset=
2268 * systemctl -t can now be used to filter by unit load state.
2270 * The journal C API gained the new sd_journal_wait() call to
2271 make writing synchronous journal clients easier.
2273 * journalctl gained the new -D switch to show journals from a
2276 * journalctl now displays a special marker between log
2277 messages of two different boots.
2279 * The journal is now explicitly flushed to /var via a service
2280 systemd-journal-flush.service, rather than implicitly simply
2281 by seeing /var/log/journal to be writable.
2283 * journalctl (and the journal C APIs) can now match for much
2284 more complex expressions, with alternatives and
2287 * When transitioning from the initial RAM disk to the main
2288 system we will now kill all processes in a killing spree to
2289 ensure no processes stay around by accident.
2291 * Three new specifiers may be used in unit files: %u, %h, %s
2292 resolve to the user name, user home directory resp. user
2293 shell. This is useful for running systemd user instances.
2295 * We now automatically rotate journal files if their data
2296 object hash table gets a fill level > 75%. We also size the
2297 hash table based on the configured maximum file size. This
2298 together should lower hash collisions drastically and thus
2299 speed things up a bit.
2301 * journalctl gained the new "--header" switch to introspect
2302 header data of journal files.
2304 * A new setting SystemCallFilters= has been added to services
2305 which may be used to apply blacklists or whitelists to
2306 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
2308 * nspawn gained a new --link-journal= switch (and quicker: -j)
2309 to link the container journal with the host. This makes it
2310 very easy to centralize log viewing on the host for all
2311 guests while still keeping the journal files separated.
2313 * Many bugfixes and optimizations
2315 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
2316 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
2317 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
2322 * Several tools now understand kernel command line arguments,
2323 which are only read when run in an initial RAM disk. They
2324 usually follow closely their normal counterparts, but are
2327 * There's a new tool to analyze the readahead files that are
2328 automatically generated at boot. Use:
2330 /usr/lib/systemd/systemd-readahead analyze /.readahead
2332 * We now provide an early debug shell on tty9 if this enabled. Use:
2334 systemctl enable debug-shell.service
2336 * All plymouth related units have been moved into the Plymouth
2337 package. Please make sure to upgrade your Plymouth version
2340 * systemd-tmpfiles now supports getting passed the basename of
2341 a configuration file only, in which case it will look for it
2342 in all appropriate directories automatically.
2344 * udevadm info now takes a /dev or /sys path as argument, and
2345 does the right thing. Example:
2347 udevadm info /dev/sda
2348 udevadm info /sys/class/block/sda
2350 * systemctl now prints a warning if a unit is stopped but a
2351 unit that might trigger it continues to run. Example: a
2352 service is stopped but the socket that activates it is left
2355 * "systemctl status" will now mention if the log output was
2356 shortened due to rotation since a service has been started.
2358 * The journal API now exposes functions to determine the
2359 "cutoff" times due to rotation.
2361 * journald now understands SIGUSR1 and SIGUSR2 for triggering
2362 immediately flushing of runtime logs to /var if possible,
2363 resp. for triggering immediate rotation of the journal
2366 * It is now considered an error if a service is attempted to
2367 be stopped that is not loaded.
2369 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
2371 * systemd-analyze now supports Python 3
2373 * tmpfiles now supports cleaning up directories via aging
2374 where the first level dirs are always kept around but
2375 directories beneath it automatically aged. This is enabled
2376 by prefixing the age field with '~'.
2378 * Seat objects now expose CanGraphical, CanTTY properties
2379 which is required to deal with very fast bootups where the
2380 display manager might be running before the graphics drivers
2381 completed initialization.
2383 * Seat objects now expose a State property.
2385 * We now include RPM macros for service enabling/disabling
2386 based on the preset logic. We recommend RPM based
2387 distributions to make use of these macros if possible. This
2388 makes it simpler to reuse RPM spec files across
2391 * We now make sure that the collected systemd unit name is
2392 always valid when services log to the journal via
2395 * There's a new man page kernel-command-line(7) detailing all
2396 command line options we understand.
2398 * The fstab generator may now be disabled at boot by passing
2399 fstab=0 on the kernel command line.
2401 * A new kernel command line option modules-load= is now understood
2402 to load a specific kernel module statically, early at boot.
2404 * Unit names specified on the systemctl command line are now
2405 automatically escaped as needed. Also, if file system or
2406 device paths are specified they are automatically turned
2407 into the appropriate mount or device unit names. Example:
2409 systemctl status /home
2410 systemctl status /dev/sda
2412 * The SysVConsole= configuration option has been removed from
2413 system.conf parsing.
2415 * The SysV search path is no longer exported on the D-Bus
2418 * The Names= option is been removed from unit file parsing.
2420 * There's a new man page bootup(7) detailing the boot process.
2422 * Every unit and every generator we ship with systemd now
2423 comes with full documentation. The self-explanatory boot is
2426 * A couple of services gained "systemd-" prefixes in their
2427 name if they wrap systemd code, rather than only external
2428 code. Among them fsck@.service which is now
2429 systemd-fsck@.service.
2431 * The HaveWatchdog property has been removed from the D-Bus
2434 * systemd.confirm_spawn= on the kernel command line should now
2437 * There's a new man page crypttab(5) which details all options
2438 we actually understand.
2440 * systemd-nspawn gained a new --capability= switch to pass
2441 additional capabilities to the container.
2443 * timedated will now read known NTP implementation unit names
2444 from /usr/lib/systemd/ntp-units.d/*.list,
2445 systemd-timedated-ntp.target has been removed.
2447 * journalctl gained a new switch "-b" that lists log data of
2448 the current boot only.
2450 * The notify socket is in the abstract namespace again, in
2451 order to support daemons which chroot() at start-up.
2453 * There is a new Storage= configuration option for journald
2454 which allows configuration of where log data should go. This
2455 also provides a way to disable journal logging entirely, so
2456 that data collected is only forwarded to the console, the
2457 kernel log buffer or another syslog implementation.
2459 * Many bugfixes and optimizations
2461 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
2462 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
2463 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
2464 Shawn Landden, Tom Gundersen
2468 * "systemctl help <unit>" now shows the man page if one is
2471 * Several new man pages have been added.
2473 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
2474 MaxLevelConsole= can now be specified in
2475 journald.conf. These options allow reducing the amount of
2476 data stored on disk or forwarded by the log level.
2478 * TimerSlackNSec= can now be specified in system.conf for
2479 PID1. This allows system-wide power savings.
2481 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
2482 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
2487 * logind is now capable of (optionally) handling power and
2488 sleep keys as well as the lid switch.
2490 * journalctl now understands the syntax "journalctl
2491 /usr/bin/avahi-daemon" to get all log output of a specific
2494 * CapabilityBoundingSet= in system.conf now also influences
2495 the capability bound set of usermode helpers of the kernel.
2497 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
2498 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
2499 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
2500 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
2504 * Note that we skipped 139 releases here in order to set the
2505 new version to something that is greater than both udev's
2506 and systemd's most recent version number.
2508 * udev: all udev sources are merged into the systemd source tree now.
2509 All future udev development will happen in the systemd tree. It
2510 is still fully supported to use the udev daemon and tools without
2511 systemd running, like in initramfs or other init systems. Building
2512 udev though, will require the *build* of the systemd tree, but
2513 udev can be properly *run* without systemd.
2515 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
2516 should be used to create dead device nodes as workarounds for broken
2519 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
2520 no longer supported. udev_monitor_new_from_netlink() needs to be
2521 used to subscribe to events.
2523 * udev: when udevd is started by systemd, processes which are left
2524 behind by forking them off of udev rules, are unconditionally cleaned
2525 up and killed now after the event handling has finished. Services or
2526 daemons must be started as systemd services. Services can be
2527 pulled-in by udev to get started, but they can no longer be directly
2528 forked by udev rules.
2530 * udev: the daemon binary is called systemd-udevd now and installed
2531 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
2532 to adapt to that, create symlink, or rename the binary after building
2535 * libudev no longer provides these symbols:
2536 udev_monitor_from_socket()
2537 udev_queue_get_failed_list_entry()
2538 udev_get_{dev,sys,run}_path()
2539 The versions number was bumped and symbol versioning introduced.
2541 * systemd-loginctl and systemd-journalctl have been renamed
2542 to loginctl and journalctl to match systemctl.
2544 * The config files: /etc/systemd/systemd-logind.conf and
2545 /etc/systemd/systemd-journald.conf have been renamed to
2546 logind.conf and journald.conf. Package updates should rename
2547 the files to the new names on upgrade.
2549 * For almost all files the license is now LGPL2.1+, changed
2550 from the previous GPL2.0+. Exceptions are some minor stuff
2551 of udev (which will be changed to LGPL2.1 eventually, too),
2552 and the MIT licensed sd-daemon.[ch] library that is suitable
2553 to be used as drop-in files.
2555 * systemd and logind now handle system sleep states, in
2556 particular suspending and hibernating.
2558 * logind now implements a sleep/shutdown/idle inhibiting logic
2559 suitable for a variety of uses. Soonishly Lennart will blog
2560 about this in more detail.
2562 * var-run.mount and var-lock.mount are no longer provided
2563 (which prevously bind mounted these directories to their new
2564 places). Distributions which have not converted these
2565 directories to symlinks should consider stealing these files
2566 from git history and add them downstream.
2568 * We introduced the Documentation= field for units and added
2569 this to all our shipped units. This is useful to make it
2570 easier to explore the boot and the purpose of the various
2573 * All smaller setup units (such as
2574 systemd-vconsole-setup.service) now detect properly if they
2575 are run in a container and are skipped when
2576 appropriate. This guarantees an entirely noise-free boot in
2577 Linux container environments such as systemd-nspawn.
2579 * A framework for implementing offline system updates is now
2580 integrated, for details see:
2581 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
2583 * A new service type Type=idle is available now which helps us
2584 avoiding ugly interleaving of getty output and boot status
2587 * There's now a system-wide CapabilityBoundingSet= option to
2588 globally reduce the set of capabilities for the
2589 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
2590 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
2591 even CAP_NET_ADMIN system-wide for secure systems.
2593 * There are now system-wide DefaultLimitXXX= options to
2594 globally change the defaults of the various resource limits
2595 for all units started by PID 1.
2597 * Harald Hoyer's systemd test suite has been integrated into
2598 systemd which allows easy testing of systemd builds in qemu
2599 and nspawn. (This is really awesome! Ask us for details!)
2601 * The fstab parser is now implemented as generator, not inside
2604 * systemctl will now warn you if .mount units generated from
2605 /etc/fstab are out of date due to changes in fstab that
2606 haven't been read by systemd yet.
2608 * systemd is now suitable for usage in initrds. Dracut has
2609 already been updated to make use of this. With this in place
2610 initrds get a slight bit faster but primarily are much
2611 easier to introspect and debug since "systemctl status" in
2612 the host system can be used to introspect initrd services,
2613 and the journal from the initrd is kept around too.
2615 * systemd-delta has been added, a tool to explore differences
2616 between user/admin configuration and vendor defaults.
2618 * PrivateTmp= now affects both /tmp and /var/tmp.
2620 * Boot time status messages are now much prettier and feature
2621 proper english language. Booting up systemd has never been
2624 * Read-ahead pack files now include the inode number of all
2625 files to pre-cache. When the inode changes the pre-caching
2626 is not attempted. This should be nicer to deal with updated
2627 packages which might result in changes of read-ahead
2630 * We now temporaritly lower the kernel's read_ahead_kb variable
2631 when collecting read-ahead data to ensure the kernel's
2632 built-in read-ahead does not add noise to our measurements
2633 of necessary blocks to pre-cache.
2635 * There's now RequiresMountsFor= to add automatic dependencies
2636 for all mounts necessary for a specific file system path.
2638 * MountAuto= and SwapAuto= have been removed from
2639 system.conf. Mounting file systems at boot has to take place
2642 * nspawn now learned a new switch --uuid= to set the machine
2643 ID on the command line.
2645 * nspawn now learned the -b switch to automatically search
2648 * vt102 is now the default TERM for serial TTYs, upgraded from
2651 * systemd-logind now works on VT-less systems.
2653 * The build tree has been reorganized. The individual
2654 components now have directories of their own.
2656 * A new condition type ConditionPathIsReadWrite= is now available.
2658 * nspawn learned the new -C switch to create cgroups for the
2659 container in other hierarchies.
2661 * We now have support for hardware watchdogs, configurable in
2664 * The scheduled shutdown logic now has a public API.
2666 * We now mount /tmp as tmpfs by default, but this can be
2667 masked and /etc/fstab can override it.
2669 * Since udisks doesn't make use of /media anymore we are not
2670 mounting a tmpfs on it anymore.
2672 * journalctl gained a new --local switch to only interleave
2673 locally generated journal files.
2675 * We can now load the IMA policy at boot automatically.
2677 * The GTK tools have been split off into a systemd-ui.
2679 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
2680 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
2681 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
2682 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
2683 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
2684 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
2685 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
2686 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
2687 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
2692 * This is mostly a bugfix release
2694 * Support optional initialization of the machine ID from the
2695 KVM or container configured UUID.
2697 * Support immediate reboots with "systemctl reboot -ff"
2699 * Show /etc/os-release data in systemd-analyze output
2701 * Many bugfixes for the journal, including endianness fixes and
2702 ensuring that disk space enforcement works
2704 * sd-login.h is C++ comptaible again
2706 * Extend the /etc/os-release format on request of the Debian
2709 * We now refuse non-UTF8 strings used in various configuration
2710 and unit files. This is done to ensure we don't pass invalid
2711 data over D-Bus or expose it elsewhere.
2713 * Register Mimo USB Screens as suitable for automatic seat
2716 * Read SELinux client context from journal clients in a race
2719 * Reorder configuration file lookup order. /etc now always
2720 overrides /run in order to allow the administrator to always
2721 and unconditionally override vendor supplied or
2722 automatically generated data.
2724 * The various user visible bits of the journal now have man
2725 pages. We still lack man pages for the journal API calls
2728 * We now ship all man pages in HTML format again in the
2731 Contributions from: Dave Reisner, Dirk Eibach, Frederic
2732 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
2733 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
2738 * This is mostly a bugfix release
2740 * systems lacking /etc/os-release are no longer supported.
2742 * Various functionality updates to libsystemd-login.so
2744 * Track class of PAM logins to distuingish greeters from
2747 Contributions from: Kay Sievers, Lennart Poettering, Michael
2752 * This is an important bugfix release for v41.
2754 * Building man pages is now optional which should be useful
2755 for those building systemd from git but unwilling to install
2758 * Watchdog support for supervising services is now usable. In
2759 a future release support for hardware watchdogs
2760 (i.e. /dev/watchdog) will be added building on this.
2762 * Service start rate limiting is now configurable and can be
2763 turned off per service. When a start rate limit is hit a
2764 reboot can automatically be triggered.
2766 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
2768 Contributions from: Benjamin Franzke, Bill Nottingham,
2769 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
2770 Schmidt, Michał Górny, Piotr Drąg
2774 * The systemd binary is installed /usr/lib/systemd/systemd now;
2775 An existing /sbin/init symlink needs to be adapted with the
2778 * The code that loads kernel modules has been ported to invoke
2779 libkmod directly, instead of modprobe. This means we do not
2780 support systems with module-init-tools anymore.
2782 * Watchdog support is now already useful, but still not
2785 * A new kernel command line option systemd.setenv= is
2786 understood to set system wide environment variables
2787 dynamically at boot.
2789 * We now limit the set of capabilities of systemd-journald.
2791 * We now set SIGPIPE to ignore by default, since it only is
2792 useful in shell pipelines, and has little use in general
2793 code. This can be disabled with IgnoreSIPIPE=no in unit
2796 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
2797 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
2802 * This is mostly a bugfix release
2804 * We now expose the reason why a service failed in the
2805 "Result" D-Bus property.
2807 * Rudimentary service watchdog support (will be completed over
2808 the next few releases.)
2810 * When systemd forks off in order execute some service we will
2811 now immediately changes its argv[0] to reflect which process
2812 it will execute. This is useful to minimize the time window
2813 with a generic argv[0], which makes bootcharts more useful
2815 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
2816 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
2817 Mike Kazantsev, Ray Strode
2821 * This is mostly a test release, but incorporates many
2824 * New systemd-cgtop tool to show control groups by their
2827 * Linking against libacl for ACLs is optional again. If
2828 disabled, support tracking device access for active logins
2829 goes becomes unavailable, and so does access to the user
2830 journals by the respective users.
2832 * If a group "adm" exists, journal files are automatically
2833 owned by them, thus allow members of this group full access
2834 to the system journal as well as all user journals.
2836 * The journal now stores the SELinux context of the logging
2837 client for all entries.
2839 * Add C++ inclusion guards to all public headers
2841 * New output mode "cat" in the journal to print only text
2842 messages, without any meta data like date or time.
2844 * Include tiny X server wrapper as a temporary stop-gap to
2845 teach XOrg udev display enumeration. This is used by display
2846 managers such as gdm, and will go away as soon as XOrg
2847 learned native udev hotplugging for display devices.
2849 * Add new systemd-cat tool for executing arbitrary programs
2850 with STDERR/STDOUT connected to the journal. Can also act as
2851 BSD logger replacement, and does so by default.
2853 * Optionally store all locally generated coredumps in the
2854 journal along with meta data.
2856 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
2857 writing short strings to files (for usage for /sys), and for
2858 creating symlinks, character and block device nodes.
2860 * New unit file option ControlGroupPersistent= to make cgroups
2861 persistent, following the mechanisms outlined in
2862 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
2864 * Support multiple local RTCs in a sane way
2866 * No longer monopolize IO when replaying readahead data on
2867 rotating disks, since we might starve non-file-system IO to
2868 death, since fanotify() will not see accesses done by blkid,
2871 * Don't show kernel threads in systemd-cgls anymore, unless
2872 requested with new -k switch.
2874 Contributions from: Dan Horák, Kay Sievers, Lennart
2875 Poettering, Michal Schmidt
2879 * This is mostly a test release, but incorporates many
2882 * The git repository moved to:
2883 git://anongit.freedesktop.org/systemd/systemd
2884 ssh://git.freedesktop.org/git/systemd/systemd
2886 * First release with the journal
2887 http://0pointer.de/blog/projects/the-journal.html
2889 * The journal replaces both systemd-kmsg-syslogd and
2890 systemd-stdout-bridge.
2892 * New sd_pid_get_unit() API call in libsystemd-logind
2894 * Many systemadm clean-ups
2896 * Introduce remote-fs-pre.target which is ordered before all
2897 remote mounts and may be used to start services before all
2900 * Added Mageia support
2902 * Add bash completion for systemd-loginctl
2904 * Actively monitor PID file creation for daemons which exit in
2905 the parent process before having finished writing the PID
2906 file in the daemon process. Daemons which do this need to be
2907 fixed (i.e. PID file creation must have finished before the
2908 parent exits), but we now react a bit more gracefully to them.
2910 * Add colourful boot output, mimicking the well-known output
2911 of existing distributions.
2913 * New option PassCredentials= for socket units, for
2914 compatibility with a recent kernel ABI breakage.
2916 * /etc/rc.local is now hooked in via a generator binary, and
2917 thus will no longer act as synchronization point during
2920 * systemctl list-unit-files now supports --root=.
2922 * systemd-tmpfiles now understands two new commands: z, Z for
2923 relabelling files according to the SELinux database. This is
2924 useful to apply SELinux labels to specific files in /sys,
2927 * Output of SysV services is now forwarded to both the console
2928 and the journal by default, not only just the console.
2930 * New man pages for all APIs from libsystemd-login.
2932 * The build tree got reorganized and a the build system is a
2933 lot more modular allowing embedded setups to specifically
2934 select the components of systemd they are interested in.
2936 * Support for Linux systems lacking the kernel VT subsystem is
2939 * configure's --with-rootdir= got renamed to
2940 --with-rootprefix= to follow the naming used by udev and
2943 * Unless specified otherwise we'll now install to /usr instead
2944 of /usr/local by default.
2946 * Processes with '@' in argv[0][0] are now excluded from the
2947 final shut-down killing spree, following the logic explained
2949 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
2951 * All processes remaining in a service cgroup when we enter
2952 the START or START_PRE states are now killed with
2953 SIGKILL. That means it is no longer possible to spawn
2954 background processes from ExecStart= lines (which was never
2955 supported anyway, and bad style).
2957 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
2958 reloading of units together.
2960 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
2961 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
2962 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
2963 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
2964 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
~ianmdlvl / elogind.git / blob