1 systemd System and Service Manager
5 * As an experimental feature, udev now tries to lock the
6 disk device node (flock(LOCK_SH|LOCK_NB)) while it
7 executes events for the disk or any of its partitions.
8 Applications like partitioning programs can lock the
9 disk device node (flock(LOCK_EX)) and claim temporary
10 device ownership that way; udev will entirely skip all event
11 handling for this disk and its partitions. If the disk
12 was opened for writing, the close will trigger a partition
13 table rescan in udev's "watch" facility, and if needed
14 synthesize "change" events for the disk and all its partitions.
15 This is now unconditionally enabled, if it turns out to
16 cause major problems, we might turn it on only for specific
17 devices, or might need to disable it entirely. Device-mapper
18 devices are excluded from this logic.
20 * We temporarily dropped the "-l" switch for fsck invocations,
21 since they collide with the flock() logic above. util-linux
22 upstream has been changed already to avoid this conflict,
23 and we will readd "-l" as soon as util-linux with this
24 change has been released.
26 * The dependency on libattr has been removed. Since a long
27 time the extended attribute calls have moved to glibc, and
28 libattr is thus unnecessary.
30 * Virtualization detection works without priviliges now. This
31 means the systemd-detect-virt binary no longer requires
32 CAP_SYS_PTRACE file capabilities, and our daemons can run
33 with fewer privileges.
35 * systemd-networkd now runs under its own "systemd-network"
36 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
37 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
38 loses the ability to write to files owned by root this way.
40 * Similar, systemd-resolved now runs under its own
41 "systemd-resolve" user with no capabilities remaining.
43 * Similar, systemd-bus-proxyd now runs under its own
44 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
46 * systemd-networkd gained support for setting up "veth"
47 virtual ethernet devices for container connectivity, as well
48 as GRE and VTI tunnels.
50 * systemd-networkd will no longer automatically attempt to
51 manually load kernel modules necessary for certain tunnel
52 transports. Instead it is assumed the kernel loads them
53 automatically when required. This only works correctly on
54 very new kernels. On older kernels, please consider adding
55 the kernel modules to /etc/load-modules.d/ as a work-around.
57 * Two new service settings ProtectedHome= and ProtectedSystem=
58 have been added. When enabled they will make the user data
59 (such as /home) inaccessible or read-only and the system
60 (such as /usr) read-only, for specific services. This allows
61 very light-weight per-service sandboxing to avoid
62 modifications of user data or system files from
63 services. These two new switches have been enabled for all
64 of systemd's long-running services, where appropriate.
66 * Socket units gained new SocketUser= and SocketGroup=
67 settings to set the owner user and group of AF_UNIX sockets
68 and FIFOs in the file system.
70 * Socket units gained a new RemoveOnStop= setting. If enabled
71 all FIFOS and sockets in the file system will be removed
72 when the specific socket unit is stopped.
74 * Socket units gained a new Symlinks= setting. It takes a list
75 of symlinks to create to file system sockets or FIFOs
76 created by the specific unix sockets. This is useful to
77 manage symlinks to socket nodes with the same life-cycle as
80 * The /dev/log socket and /dev/initctl FIFO have been moved to
81 /run, and have been replaced by symlinks. This allows
82 connecting to these facilities even if PrivateDevices=yes is
83 used for a service (which makes /dev/log itself unavailable,
84 but /run is left). This also has the benefit of ensuring
85 that /dev only contains device nodes, directories and
86 symlinks, and nothing else.
88 * sd-daemon gained two new calls sd_pid_notify() and
89 sd_pid_notifyf(). They are similar to sd_notify() and
90 sd_notifyf(), but allow overriding of the source PID of
91 notification messages if permissions permit this. This is
92 useful to send notify messages on behalf of a different
93 process (for example, the parent process). The
94 systemd-notify tool has been updated to make use of this
95 when sending messages (so that notification messages now
96 originate from the shell script invoking systemd-notify and
97 not the systemd-notify process itself. This should minimize
98 a race where systemd fails to associate notification
99 messages to services when the originating process already
102 * A new "on-abnormal" setting for Restart= has been added. If
103 set it will result in automatic restarts on all "abnormal"
104 reasons for a process to exit, which includes unclean
105 signals, core dumps, timeouts and watchdog timeouts, but
106 does not include clean and unclean exit codes or clean
107 signals. Restart=on-abnormal is an alternative for
108 Restart=on-failure for services that shall be able to
109 terminate and avoid restarts on certain errors, by
110 indicating so with an unclean exit code. Restart=on-failure
111 or Restart=on-abnormal is now the recommended setting for
112 all long-running services.
114 * If the InaccessibleDirectories= service setting points to a
115 mount point (or if there are any submounts contained within
116 it), it is now attempted to completely unmount it, to make
117 the file systems truly unavailable for the respective
120 * The ReadOnlyDirectories= service setting and
121 systemd-nspawn's --read-only parameter are now recursively
122 applied to all submounts, too.
124 * Mount units may now be created transiently via the bus APIs.
126 * The support for SysV and LSB init scripts has been removed
127 from the systemd daemon itself. Instead, it is now
128 implemented as a generator that creates native systemd units
129 from these scripts when needed. This enables us to remove a
130 substantial amount of legacy code from PID 1, following the
131 fact that many distributions only ship a very small number
132 of LSB/SysV init scripts nowadays.
134 * Priviliged Xen (dom0) domains are not considered
135 virtualization anymore by the virtualization detection
136 logic. After all, they generally have unrestricted access to
137 the hardware and usually are used to manage the unprivileged
140 * systemd-tmpfiles gained a new "C" line type, for copying
141 files or entire directories.
143 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
144 lines. So far they have been non-globbing versions of the
145 latter, and have thus been redundant. In future it is
146 recommended to only use "z"; and "m" has hence been removed
147 from the documentation, even though it stays supported.
149 * A tmpfiles snippet to recreate the most basic structure in
150 /var has been added. This is enough to create the /var/run →
151 /run symlink and create a couple of structural
152 directories. This allows systems to boot up with an empty or
153 volatile /var. Of course, while with this change the core OS
154 now is capable with dealing with a volatile /var not all
155 user services are ready for it. However, we hope that sooner
156 or later many service daemons will be changed upstream so
157 that they are able to automatically create their necessary
158 directories in /var at boot, should they be missing. This is
159 the first step to allow state-less systems that only require
160 the vendor image for /usr to boot.
162 * systemd-nspawn has gained a new --tmpfs= switch to mount an
163 empty tmpfs instance to a specific directory. This is
164 particularly useful for making use of the automatic
165 reconstruction of /var (see above), by passing --tmpfs=/var.
167 * Access modes specified in tmpfiles snippets may now be
168 prefixed with "~", which indicates that they shall be masked
169 by whether the existing file or directly is currently
170 writable, readable or executable at all. Also, if specified
171 the sgid/suid/sticky bits will be masked for all
174 * A new passive target unit "network-pre.target" has been
175 added which is useful for services that shall run before any
176 network is configured, for example firewall scripts.
180 * A new "systemd-timesyncd" daemon has been added for
181 synchronizing the system clock across the network. It
182 implements an SNTP client. In contrast to NTP
183 implementations such as chrony or the NTP reference server
184 this only implements a client side, and does not bother with
185 the full NTP complexity, focusing only on querying time from
186 one remote server and synchronizing the local clock to
187 it. Unless you intend to serve NTP to networked clients or
188 want to connect to local hardware clocks this simple NTP
189 client should be more than appropriate for most
190 installations. The daemon runs with minimal privileges, and
191 has been hooked up with networkd to only operate when
192 network connectivity is available. The daemon saves the
193 current clock to disk every time a new NTP sync has been
194 acquired, and uses this to possibly correct the system clock
195 early at bootup, in order to accommodate for systems that
196 lack an RTC such as the Raspberry Pi and embedded devices,
197 and make sure that time monotonically progresses on these
198 systems, even if it is not always correct. To make use of
199 this daemon a new system user and group "systemd-timesync"
200 needs to be created on installation of systemd.
202 * The queue "seqnum" interface of libudev has been disabled, as
203 it was generally incompatible with device namespacing as
204 sequence numbers of devices go "missing" if the devices are
205 part of a different namespace.
207 * "systemctl list-timers" and "systemctl list-sockets" gained
208 a --recursive switch for showing units of these types also
209 for all local containers, similar in style to the already
210 supported --recursive switch for "systemctl list-units".
212 * A new RebootArgument= setting has been added for service
213 units, which may be used to specify a kernel reboot argument
214 to use when triggering reboots with StartLimitAction=.
216 * A new FailureAction= setting has been added for service
217 units which may be used to specify an operation to trigger
218 when a service fails. This works similarly to
219 StartLimitAction=, but unlike it controls what is done
220 immediately rather than only after several attempts to
221 restart the service in question.
223 * hostnamed got updated to also expose the kernel name,
224 release, and version on the bus. This is useful for
225 executing commands like hostnamectl with the -H switch.
226 systemd-analyze makes use of this to properly display
227 details when running non-locally.
229 * The bootchart tool can now show cgroup information in the
232 * The CFS CPU quota cgroup attribute is now exposed for
233 services. The new CPUQuota= switch has been added for this
234 which takes a percentage value. Setting this will have the
235 result that a service may never get more CPU time than the
236 specified percentage, even if the machine is otherwise idle.
238 * systemd-networkd learned IPIP and SIT tunnel support.
240 * LSB init scripts exposing a dependency on $network will now
241 get a dependency on network-online.target rather than simply
242 network.target. This should bring LSB handling closer to
243 what it was on SysV systems.
245 * A new fsck.repair= kernel option has been added to control
246 how fsck shall deal with unclean file systems at boot.
248 * The (.ini) configuration file parser will now silently
249 ignore sections whose name begins with "X-". This may be
250 used to maintain application-specific extension sections in unit
253 * machined gained a new API to query the IP addresses of
254 registered containers. "machinectl status" has been updated
255 to show these addresses in its output.
257 * A new call sd_uid_get_display() has been added to the
258 sd-login APIs for querying the "primary" session of a
259 user. The "primary" session of the user is elected from the
260 user's sessions and generally a graphical session is
261 preferred over a text one.
263 * A minimal systemd-resolved daemon has been added. It
264 currently simply acts as a companion to systemd-networkd and
265 manages resolv.conf based on per-interface DNS
266 configuration, possibly supplied via DHCP. In the long run
267 we hope to extend this into a local DNSSEC enabled DNS and
270 * The systemd-networkd-wait-online tool is now enabled by
271 default. It will delay network-online.target until a network
272 connection has been configured. The tool primarily integrates
273 with networkd, but will also make a best effort to make sense
274 of network configuration performed in some other way.
276 * Two new service options StartupCPUShares= and
277 StartupBlockIOWeight= have been added that work similarly to
278 CPUShares= and BlockIOWeight= however only apply during
279 system startup. This is useful to prioritize certain services
280 differently during bootup than during normal runtime.
282 * hostnamed has been changed to prefer the statically
283 configured hostname in /etc/hostname (unless set to
284 'localhost' or empty) over any dynamic one supplied by
285 dhcp. With this change the rules for picking the hostname
286 match more closely the rules of other configuration settings
287 where the local administrator's configuration in /etc always
288 overrides any other settings.
290 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
291 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
292 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
293 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
294 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
295 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
296 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
297 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
298 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
299 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
300 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
301 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
302 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
303 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
304 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
305 Lindskog, WaLyong Cho, Will Woods, Zbigniew
308 -- Beijing, 2014-05-28
312 * When restoring the screen brightness at boot, stay away from
313 the darkest setting or from the lowest 5% of the available
314 range, depending on which is the larger value of both. This
315 should effectively protect the user from rebooting into a
316 black screen, should the brightness have been set to minimum
319 * sd-login gained a new sd_machine_get_class() call to
320 determine the class ("vm" or "container") of a machine
321 registered with machined.
323 * sd-login gained new calls
324 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
325 to query the identity of the peer of a local AF_UNIX
326 connection. They operate similarly to their sd_pid_get_xyz()
329 * PID 1 will now maintain a system-wide system state engine
330 with the states "starting", "running", "degraded",
331 "maintenance", "stopping". These states are bound to system
332 startup, normal runtime, runtime with at least one failed
333 service, rescue/emergency mode and system shutdown. This
334 state is shown in the "systemctl status" output when no unit
335 name is passed. It is useful to determine system state, in
336 particularly when doing so for many systems or containers at
339 * A new command "list-machines" has been added to "systemctl"
340 that lists all local OS containers and shows their system
341 state (see above), if systemd runs inside of them.
343 * systemctl gained a new "-r" switch to recursively enumerate
344 units on all local containers, when used with the
345 "list-unit" command (which is the default one that is
346 executed when no parameters are specified).
348 * The GPT automatic partition discovery logic will now honour
349 two GPT partition flags: one may be set on a partition to
350 cause it to be mounted read-only, and the other may be set
351 on a partition to ignore it during automatic discovery.
353 * Two new GPT type UUIDs have been added for automatic root
354 partition discovery, for 32-bit and 64-bit ARM. This is not
355 particularly useful for discovering the root directory on
356 these architectures during bare-metal boots (since UEFI is
357 not common there), but still very useful to allow booting of
358 ARM disk images in nspawn with the -i option.
360 * MAC addresses of interfaces created with nspawn's
361 --network-interface= switch will now be generated from the
362 machine name, and thus be stable between multiple invocations
365 * logind will now automatically remove all IPC objects owned
366 by a user if she or he fully logs out. This makes sure that
367 users who are logged out cannot continue to consume IPC
368 resources. This covers SysV memory, semaphores and message
369 queues as well as POSIX shared memory and message
370 queues. Traditionally, SysV and POSIX IPC had no life-cycle
371 limits. With this functionality, that is corrected. This may
372 be turned off by using the RemoveIPC= switch of logind.conf.
374 * The systemd-machine-id-setup and tmpfiles tools gained a
375 --root= switch to operate on a specific root directory,
378 * journald can now forward logged messages to the TTYs of all
379 logged in users ("wall"). This is the default for all
380 emergency messages now.
382 * A new tool systemd-journal-remote has been added to stream
383 journal log messages across the network.
385 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
386 controller trees are mounted into it. Note that the
387 directories mounted beneath it are not read-only. This is a
388 security measure and is particularly useful because glibc
389 actually includes a search logic to pick any tmpfs it can
390 find to implement shm_open() if /dev/shm is not available
391 (which it might very well be in namespaced setups).
393 * machinectl gained a new "poweroff" command to cleanly power
394 down a local OS container.
396 * The PrivateDevices= unit file setting will now also drop the
397 CAP_MKNOD capability from the capability bound set, and
398 imply DevicePolicy=closed.
400 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
401 comprehensively on all long-running systemd services where
404 * systemd-udevd will now run in a disassociated mount
405 namespace. To mount directories from udev rules, make sure to
406 pull in mount units via SYSTEMD_WANTS properties.
408 * The kdbus support gained support for uploading policy into
409 the kernel. sd-bus gained support for creating "monitoring"
410 connections that can eavesdrop into all bus communication
411 for debugging purposes.
413 * Timestamps may now be specified in seconds since the UNIX
414 epoch Jan 1st, 1970 by specifying "@" followed by the value
417 * Native tcpwrap support in systemd has been removed. tcpwrap
418 is old code, not really maintained anymore and has serious
419 shortcomings, and better options such as firewalls
420 exist. For setups that require tcpwrap usage, please
421 consider invoking your socket-activated service via tcpd,
422 like on traditional inetd.
424 * A new system.conf configuration option
425 DefaultTimerAccuracySec= has been added that controls the
426 default AccuracySec= setting of .timer units.
428 * Timer units gained a new WakeSystem= switch. If enabled,
429 timers configured this way will cause the system to resume
430 from system suspend (if the system supports that, which most
433 * Timer units gained a new Persistent= switch. If enabled,
434 timers configured this way will save to disk when they have
435 been last triggered. This information is then used on next
436 reboot to possible execute overdue timer events, that
437 could not take place because the system was powered off.
438 This enables simple anacron-like behaviour for timer units.
440 * systemctl's "list-timers" will now also list the time a
441 timer unit was last triggered in addition to the next time
442 it will be triggered.
444 * systemd-networkd will now assign predictable IPv4LL
445 addresses to its local interfaces.
447 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
448 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
449 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
450 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
451 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
452 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
453 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
454 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
457 -- Berlin, 2014-03-25
461 * A new unit file setting RestrictAddressFamilies= has been
462 added to restrict which socket address families unit
463 processes gain access to. This takes address family names
464 like "AF_INET" or "AF_UNIX", and is useful to minimize the
465 attack surface of services via exotic protocol stacks. This
466 is built on seccomp system call filters.
468 * Two new unit file settings RuntimeDirectory= and
469 RuntimeDirectoryMode= have been added that may be used to
470 manage a per-daemon runtime directories below /run. This is
471 an alternative for setting up directory permissions with
472 tmpfiles snippets, and has the advantage that the runtime
473 directory's lifetime is bound to the daemon runtime and that
474 the daemon starts up with an empty directory each time. This
475 is particularly useful when writing services that drop
476 privileges using the User= or Group= setting.
478 * The DeviceAllow= unit setting now supports globbing for
479 matching against device group names.
481 * The systemd configuration file system.conf gained new
482 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
483 DefaultMemoryAccounting= to globally turn on/off accounting
484 for specific resources (cgroups) for all units. These
485 settings may still be overridden individually in each unit
488 * systemd-gpt-auto-generator is now able to discover /srv and
489 root partitions in addition to /home and swap partitions. It
490 also supports LUKS-encrypted partitions now. With this in
491 place, automatic discovery of partitions to mount following
492 the Discoverable Partitions Specification
493 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
494 is now a lot more complete. This allows booting without
495 /etc/fstab and without root= on the kernel command line on
496 systems prepared appropriately.
498 * systemd-nspawn gained a new --image= switch which allows
499 booting up disk images and Linux installations on any block
500 device that follow the Discoverable Partitions Specification
501 (see above). This means that installations made with
502 appropriately updated installers may now be started and
503 deployed using container managers, completely
504 unmodified. (We hope that libvirt-lxc will add support for
505 this feature soon, too.)
507 * systemd-nspawn gained a new --network-macvlan= setting to
508 set up a private macvlan interface for the
509 container. Similarly, systemd-networkd gained a new
510 Kind=macvlan setting in .netdev files.
512 * systemd-networkd now supports configuring local addresses
515 * A new tool systemd-network-wait-online has been added to
516 synchronously wait for network connectivity using
519 * The sd-bus.h bus API gained a new sd_bus_track object for
520 tracking the life-cycle of bus peers. Note that sd-bus.h is
521 still not a public API though (unless you specify
522 --enable-kdbus on the configure command line, which however
523 voids your warranty and you get no API stability guarantee).
525 * The $XDG_RUNTIME_DIR runtime directories for each user are
526 now individual tmpfs instances, which has the benefit of
527 introducing separate pools for each user, with individual
528 size limits, and thus making sure that unprivileged clients
529 can no longer negatively impact the system or other users by
530 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
531 RuntimeDirectorySize= has been introduced that allows
532 controlling the default size limit for all users. It
533 defaults to 10% of the available physical memory. This is no
534 replacement for quotas on tmpfs though (which the kernel
535 still does not support), as /dev/shm and /tmp are still
536 shared resources used by both the system and unprivileged
539 * logind will now automatically turn off automatic suspending
540 on laptop lid close when more than one display is
541 connected. This was previously expected to be implemented
542 individually in desktop environments (such as GNOME),
543 however has been added to logind now, in order to fix a
544 boot-time race where a desktop environment might not have
545 been started yet and thus not been able to take an inhibitor
546 lock at the time where logind already suspends the system
549 * logind will now wait at least 30s after each system
550 suspend/resume cycle, and 3min after system boot before
551 suspending the system due to a closed laptop lid. This
552 should give USB docking stations and similar enough time to
553 be probed and configured after system resume and boot in
554 order to then act as suspend blocker.
556 * systemd-run gained a new --property= setting which allows
557 initialization of resource control properties (and others)
558 for the created scope or service unit. Example: "systemd-run
559 --property=BlockIOWeight=10 updatedb" may be used to run
560 updatedb at a low block IO scheduling weight.
562 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
563 now also work in --scope mode.
565 * When systemd is compiled with kdbus support, basic support
566 for enforced policies is now in place. (Note that enabling
567 kdbus still voids your warranty and no API compatibility
570 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
571 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
572 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
573 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
574 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
575 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
576 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
577 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
578 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
579 Zbigniew Jędrzejewski-Szmek
581 -- Berlin, 2014-03-12
585 * systemd will now relabel /dev after loading the SMACK policy
586 according to SMACK rules.
588 * A new unit file option AppArmorProfile= has been added to
589 set the AppArmor profile for the processes of a unit.
591 * A new condition check ConditionArchitecture= has been added
592 to conditionalize units based on the system architecture, as
593 reported by uname()'s "machine" field.
595 * systemd-networkd now supports matching on the system
596 virtualization, architecture, kernel command line, host name
599 * logind is now a lot more aggressive when suspending the
600 machine due to a closed laptop lid. Instead of acting only
601 on the lid close action, it will continuously watch the lid
602 status and act on it. This is useful for laptops where the
603 power button is on the outside of the chassis so that it can
604 be reached without opening the lid (such as the Lenovo
605 Yoga). On those machines, logind will now immediately
606 re-suspend the machine if the power button has been
607 accidentally pressed while the laptop was suspended and in a
610 * logind will now watch SW_DOCK switches and inhibit reaction
611 to the lid switch if it is pressed. This means that logind
612 will not suspend the machine anymore if the lid is closed
613 and the system is docked, if the laptop supports SW_DOCK
614 notifications via the input layer. Note that ACPI docking
615 stations do not generate this currently. Also note that this
616 logic is usually not fully sufficient and Desktop
617 Environments should take a lid switch inhibitor lock when an
618 external display is connected, as systemd will not watch
621 * nspawn will now make use of the devices cgroup controller by
622 default, and only permit creation of and access to the usual
623 API device nodes like /dev/null or /dev/random, as well as
624 access to (but not creation of) the pty devices.
626 * We will now ship a default .network file for
627 systemd-networkd that automatically configures DHCP for
628 network interfaces created by nspawn's --network-veth or
629 --network-bridge= switches.
631 * systemd will now understand the usual M, K, G, T suffixes
632 according to SI conventions (i.e. to the base 1000) when
633 referring to throughput and hardware metrics. It will stay
634 with IEC conventions (i.e. to the base 1024) for software
635 metrics, according to what is customary according to
636 Wikipedia. We explicitly document which base applies for
637 each configuration option.
639 * The DeviceAllow= setting in unit files now supports a syntax
640 to whitelist an entire group of devices node majors at once,
641 based on the /proc/devices listing. For example, with the
642 string "char-pts", it is now possible to whitelist all
643 current and future pseudo-TTYs at once.
645 * sd-event learned a new "post" event source. Event sources of
646 this type are triggered by the dispatching of any event
647 source of a type that is not "post". This is useful for
648 implementing clean-up and check event sources that are
649 triggered by other work being done in the program.
651 * systemd-networkd is no longer statically enabled, but uses
652 the usual [Install] sections so that it can be
653 enabled/disabled using systemctl. It still is enabled by
656 * When creating a veth interface pair with systemd-nspawn, the
657 host side will now be prefixed with "vb-" if
658 --network-bridge= is used, and with "ve-" if --network-veth
659 is used. This way, it is easy to distinguish these cases on
660 the host, for example to apply different configuration to
661 them with systemd-networkd.
663 * The compatibility libraries for libsystemd-journal.so,
664 libsystem-id128.so, libsystemd-login.so and
665 libsystemd-daemon.so do not make use of IFUNC
666 anymore. Instead, we now build libsystemd.so multiple times
667 under these alternative names. This means that the footprint
668 is drastically increased, but given that these are
669 transitional compatibility libraries, this should not matter
670 much. This change has been made necessary to support the ARM
671 platform for these compatibility libraries, as the ARM
672 toolchain is not really at the same level as the toolchain
673 for other architectures like x86 and does not support
674 IFUNC. Please make sure to use --enable-compat-libs only
675 during a transitional period!
677 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
678 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
679 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
680 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
681 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
682 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
683 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
684 Zbigniew Jędrzejewski-Szmek
686 -- Berlin, 2014-02-24
690 * A new component "systemd-networkd" has been added that can
691 be used to configure local network interfaces statically or
692 via DHCP. It is capable of bringing up bridges, VLANs, and
693 bonding. Currently, no hook-ups for interactive network
694 configuration are provided. Use this for your initrd,
695 container, embedded, or server setup if you need a simple,
696 yet powerful, network configuration solution. This
697 configuration subsystem is quite nifty, as it allows wildcard
698 hotplug matching in interfaces. For example, with a single
699 configuration snippet, you can configure that all Ethernet
700 interfaces showing up are automatically added to a bridge,
701 or similar. It supports link-sensing and more.
703 * A new tool "systemd-socket-proxyd" has been added which can
704 act as a bidirectional proxy for TCP sockets. This is
705 useful for adding socket activation support to services that
706 do not actually support socket activation, including virtual
707 machines and the like.
709 * Add a new tool to save/restore rfkill state on
712 * Save/restore state of keyboard backlights in addition to
713 display backlights on shutdown/boot.
715 * udev learned a new SECLABEL{} construct to label device
716 nodes with a specific security label when they appear. For
717 now, only SECLABEL{selinux} is supported, but the syntax is
718 prepared for additional security frameworks.
720 * udev gained a new scheme to configure link-level attributes
721 from files in /etc/systemd/network/*.link. These files can
722 match against MAC address, device path, driver name and type,
723 and will apply attributes like the naming policy, link speed,
724 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
725 address assignment policy (randomized, ...).
727 * The configuration of network interface naming rules for
728 "permanent interface names" has changed: a new NamePolicy=
729 setting in the [Link] section of .link files determines the
730 priority of possible naming schemes (onboard, slot, mac,
731 path). The default value of this setting is determined by
732 /usr/lib/net/links/99-default.link. Old
733 80-net-name-slot.rules udev configuration file has been
734 removed, so local configuration overriding this file should
735 be adapated to override 99-default.link instead.
737 * When the User= switch is used in a unit file, also
738 initialize $SHELL= based on the user database entry.
740 * systemd no longer depends on libdbus. All communication is
741 now done with sd-bus, systemd's low-level bus library
744 * kdbus support has been added to PID 1 itself. When kdbus is
745 enabled, this causes PID 1 to set up the system bus and
746 enable support for a new ".busname" unit type that
747 encapsulates bus name activation on kdbus. It works a little
748 bit like ".socket" units, except for bus names. A new
749 generator has been added that converts classic dbus1 service
750 activation files automatically into native systemd .busname
753 * sd-bus: add a light-weight vtable implementation that allows
754 defining objects on the bus with a simple static const
755 vtable array of its methods, signals and properties.
757 * systemd will not generate or install static dbus
758 introspection data anymore to /usr/share/dbus-1/interfaces,
759 as the precise format of these files is unclear, and
760 nothing makes use of it.
762 * A proxy daemon is now provided to proxy clients connecting
763 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
764 compatibility with classic D-Bus.
766 * A bus driver implementation has been added that supports the
767 classic D-Bus bus driver calls on kdbus, also for
768 compatibility purposes.
770 * A new API "sd-event.h" has been added that implements a
771 minimal event loop API built around epoll. It provides a
772 couple of features that direct epoll usage is lacking:
773 prioritization of events, scales to large numbers of timer
774 events, per-event timer slack (accuracy), system-wide
775 coalescing of timer events, exit handlers, watchdog
776 supervision support using systemd's sd_notify() API, child
779 * A new API "sd-rntl.h" has been added that provides an API
780 around the route netlink interface of the kernel, similar in
783 * A new API "sd-dhcp-client.h" has been added that provides a
784 small DHCPv4 client-side implementation. This is used by
787 * There is a new kernel command line option
788 "systemd.restore_state=0|1". When set to "0", none of the
789 systemd tools will restore saved runtime state to hardware
790 devices. More specifically, the rfkill and backlight states
793 * The FsckPassNo= compatibility option in mount/service units
794 has been removed. The fstab generator will now add the
795 necessary dependencies automatically, and does not require
796 PID1's support for that anymore.
798 * journalctl gained a new switch, --list-boots, that lists
799 recent boots with their times and boot IDs.
801 * The various tools like systemctl, loginctl, timedatectl,
802 busctl, systemd-run, ... have gained a new switch "-M" to
803 connect to a specific, local OS container (as direct
804 connection, without requiring SSH). This works on any
805 container that is registered with machined, such as those
806 created by libvirt-lxc or nspawn.
808 * systemd-run and systemd-analyze also gained support for "-H"
809 to connect to remote hosts via SSH. This is particularly
810 useful for systemd-run because it enables queuing of jobs
813 * machinectl gained a new command "login" to open a getty
814 login in any local container. This works with any container
815 that is registered with machined (such as those created by
816 libvirt-lxc or nspawn), and which runs systemd inside.
818 * machinectl gained a new "reboot" command that may be used to
819 trigger a reboot on a specific container that is registered
820 with machined. This works on any container that runs an init
823 * systemctl gained a new "list-timers" command to print a nice
824 listing of installed timer units with the times they elapse
827 * Alternative reboot() parameters may now be specified on the
828 "systemctl reboot" command line and are passed to the
829 reboot() system call.
831 * systemctl gained a new --job-mode= switch to configure the
832 mode to queue a job with. This is a more generic version of
833 --fail, --irreversible, and --ignore-dependencies, which are
834 still available but not advertised anymore.
836 * /etc/systemd/system.conf gained new settings to configure
837 various default timeouts of units, as well as the default
838 start limit interval and burst. These may still be overridden
841 * PID1 will now export on the bus profile data of the security
842 policy upload process (such as the SELinux policy upload to
845 * journald: when forwarding logs to the console, include
846 timestamps (following the setting in
847 /sys/module/printk/parameters/time).
849 * OnCalendar= in timer units now understands the special
850 strings "yearly" and "annually". (Both are equivalent)
852 * The accuracy of timer units is now configurable with the new
853 AccuracySec= setting. It defaults to 1min.
855 * A new dependency type JoinsNamespaceOf= has been added that
856 allows running two services within the same /tmp and network
857 namespace, if PrivateNetwork= or PrivateTmp= are used.
859 * A new command "cat" has been added to systemctl. It outputs
860 the original unit file of a unit, and concatenates the
861 contents of additional "drop-in" unit file snippets, so that
862 the full configuration is shown.
864 * systemctl now supports globbing on the various "list-xyz"
865 commands, like "list-units" or "list-sockets", as well as on
866 those commands which take multiple unit names.
868 * journalctl's --unit= switch gained support for globbing.
870 * All systemd daemons now make use of the watchdog logic so
871 that systemd automatically notices when they hang.
873 * If the $container_ttys environment variable is set,
874 getty-generator will automatically spawn a getty for each
875 listed tty. This is useful for container managers to request
876 login gettys to be spawned on as many ttys as needed.
878 * %h, %s, %U specifier support is not available anymore when
879 used in unit files for PID 1. This is because NSS calls are
880 not safe from PID 1. They stay available for --user
881 instances of systemd, and as special case for the root user.
883 * loginctl gained a new "--no-legend" switch to turn off output
886 * The "sd-login.h" API gained three new calls:
887 sd_session_is_remote(), sd_session_get_remote_user(),
888 sd_session_get_remote_host() to query information about
891 * The udev hardware database now also carries vendor/product
892 information of SDIO devices.
894 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
895 determine whether watchdog notifications are requested by
898 * Socket-activated per-connection services now include a
899 short description of the connection parameters in the
902 * tmpfiles gained a new "--boot" option. When this is not used,
903 only lines where the command character is not suffixed with
904 "!" are executed. When this option is specified, those
905 options are executed too. This partitions tmpfiles
906 directives into those that can be safely executed at any
907 time, and those which should be run only at boot (for
908 example, a line that creates /run/nologin).
910 * A new API "sd-resolve.h" has been added which provides a simple
911 asynchronous wrapper around glibc NSS host name resolution
912 calls, such as getaddrinfo(). In contrast to glibc's
913 getaddrinfo_a(), it does not use signals. In contrast to most
914 other asynchronous name resolution libraries, this one does
915 not reimplement DNS, but reuses NSS, so that alternate
916 host name resolution systems continue to work, such as mDNS,
917 LDAP, etc. This API is based on libasyncns, but it has been
918 cleaned up for inclusion in systemd.
920 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
921 "sd-daemon.h" are no longer found in individual libraries
922 libsystemd-journal.so, libsystemd-login.so,
923 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
924 merged them into a single library, libsystemd.so, which
925 provides all symbols. The reason for this is cyclic
926 dependencies, as these libraries tend to use each other's
927 symbols. So far, we have managed to workaround that by linking
928 a copy of a good part of our code into each of these
929 libraries again and again, which, however, makes certain
930 things hard to do, like sharing static variables. Also, it
931 substantially increases footprint. With this change, there
932 is only one library for the basic APIs systemd
933 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
934 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
935 library as well, however are subject to the --enable-kdbus
936 switch (see below). Note that "sd-dhcp-client.h" is not part
937 of this library (this is because it only consumes, never
938 provides, services of/to other APIs). To make the transition
939 easy from the separate libraries to the unified one, we
940 provide the --enable-compat-libs compile-time switch which
941 will generate stub libraries that are compatible with the
942 old ones but redirect all calls to the new one.
944 * All of the kdbus logic and the new APIs "sd-bus.h",
945 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
946 and "sd-utf8.h" are compile-time optional via the
947 "--enable-kdbus" switch, and they are not compiled in by
948 default. To make use of kdbus, you have to explicitly enable
949 the switch. Note however, that neither the kernel nor the
950 userspace API for all of this is considered stable yet. We
951 want to maintain the freedom to still change the APIs for
952 now. By specifying this build-time switch, you acknowledge
953 that you are aware of the instability of the current
956 * Also, note that while kdbus is pretty much complete,
957 it lacks one thing: proper policy support. This means you
958 can build a fully working system with all features; however,
959 it will be highly insecure. Policy support will be added in
960 one of the next releases, at the same time that we will
961 declare the APIs stable.
963 * When the kernel command-line argument "kdbus" is specified,
964 systemd will automatically load the kdbus.ko kernel module. At
965 this stage of development, it is only useful for testing kdbus
966 and should not be used in production. Note: if "--enable-kdbus"
967 is specified, and the kdbus.ko kernel module is available, and
968 "kdbus" is added to the kernel command line, the entire system
969 runs with kdbus instead of dbus-daemon, with the above mentioned
970 problem of missing the system policy enforcement. Also a future
971 version of kdbus.ko or a newer systemd will not be compatible with
972 each other, and will unlikely be able to boot the machine if only
973 one of them is updated.
975 * systemctl gained a new "import-environment" command which
976 uploads the caller's environment (or parts thereof) into the
977 service manager so that it is inherited by services started
978 by the manager. This is useful to upload variables like
979 $DISPLAY into the user service manager.
981 * A new PrivateDevices= switch has been added to service units
982 which allows running a service with a namespaced /dev
983 directory that does not contain any device nodes for
984 physical devices. More specifically, it only includes devices
985 such as /dev/null, /dev/urandom, and /dev/zero which are API
988 * logind has been extended to support behaviour like VT
989 switching on seats that do not support a VT. This makes
990 multi-session available on seats that are not the first seat
991 (seat0), and on systems where kernel support for VTs has
992 been disabled at compile-time.
994 * If a process holds a delay lock for system sleep or shutdown
995 and fails to release it in time, we will now log its
996 identity. This makes it easier to identify processes that
997 cause slow suspends or power-offs.
999 * When parsing /etc/crypttab, support for a new key-slot=
1000 option as supported by Debian is added. It allows indicating
1001 which LUKS slot to use on disk, speeding up key loading.
1003 * The sd_journald_sendv() API call has been checked and
1004 officially declared to be async-signal-safe so that it may
1005 be invoked from signal handlers for logging purposes.
1007 * Boot-time status output is now enabled automatically after a
1008 short timeout if boot does not progress, in order to give
1009 the user an indication what she or he is waiting for.
1011 * The boot-time output has been improved to show how much time
1012 remains until jobs expire.
1014 * The KillMode= switch in service units gained a new possible
1015 value "mixed". If set, and the unit is shut down, then the
1016 initial SIGTERM signal is sent only to the main daemon
1017 process, while the following SIGKILL signal is sent to
1018 all remaining processes of the service.
1020 * When a scope unit is registered, a new property "Controller"
1021 may be set. If set to a valid bus name, systemd will send a
1022 RequestStop() signal to this name when it would like to shut
1023 down the scope. This may be used to hook manager logic into
1024 the shutdown logic of scope units. Also, scope units may now
1025 be put in a special "abandoned" state, in which case the
1026 manager process which created them takes no further
1027 responsibilities for it.
1029 * When reading unit files, systemd will now verify
1030 the access mode of these files, and warn about certain
1031 suspicious combinations. This has been added to make it
1032 easier to track down packaging bugs where unit files are
1033 marked executable or world-writable.
1035 * systemd-nspawn gained a new "--setenv=" switch to set
1036 container-wide environment variables. The similar option in
1037 systemd-activate was renamed from "--environment=" to
1038 "--setenv=" for consistency.
1040 * systemd-nspawn has been updated to create a new kdbus domain
1041 for each container that is invoked, thus allowing each
1042 container to have its own set of system and user buses,
1043 independent of the host.
1045 * systemd-nspawn gained a new --drop-capability= switch to run
1046 the container with less capabilities than the default. Both
1047 --drop-capability= and --capability= now take the special
1048 string "all" for dropping or keeping all capabilities.
1050 * systemd-nspawn gained new switches for executing containers
1051 with specific SELinux labels set.
1053 * systemd-nspawn gained a new --quiet switch to not generate
1054 any additional output but the container's own console
1057 * systemd-nspawn gained a new --share-system switch to run a
1058 container without PID namespacing enabled.
1060 * systemd-nspawn gained a new --register= switch to control
1061 whether the container is registered with systemd-machined or
1062 not. This is useful for containers that do not run full
1063 OS images, but only specific apps.
1065 * systemd-nspawn gained a new --keep-unit which may be used
1066 when invoked as the only program from a service unit, and
1067 results in registration of the unit service itself in
1068 systemd-machined, instead of a newly opened scope unit.
1070 * systemd-nspawn gained a new --network-interface= switch for
1071 moving arbitrary interfaces to the container. The new
1072 --network-veth switch creates a virtual Ethernet connection
1073 between host and container. The new --network-bridge=
1074 switch then allows assigning the host side of this virtual
1075 Ethernet connection to a bridge device.
1077 * systemd-nspawn gained a new --personality= switch for
1078 setting the kernel personality for the container. This is
1079 useful when running a 32-bit container on a 64-bit host. A
1080 similar option Personality= is now also available for service
1083 * logind will now also track a "Desktop" identifier for each
1084 session which encodes the desktop environment of it. This is
1085 useful for desktop environments that want to identify
1086 multiple running sessions of itself easily.
1088 * A new SELinuxContext= setting for service units has been
1089 added that allows setting a specific SELinux execution
1090 context for a service.
1092 * Most systemd client tools will now honour $SYSTEMD_LESS for
1093 settings of the "less" pager. By default, these tools will
1094 override $LESS to allow certain operations to work, such as
1095 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
1096 influence this logic.
1098 * systemd's "seccomp" hook-up has been changed to make use of
1099 the libseccomp library instead of using its own
1100 implementation. This has benefits for portability among
1103 * For usage together with SystemCallFilter=, a new
1104 SystemCallErrorNumber= setting has been introduced that
1105 allows configuration of a system error number to be returned
1106 on filtered system calls, instead of immediately killing the
1107 process. Also, SystemCallArchitectures= has been added to
1108 limit access to system calls of a particular architecture
1109 (in order to turn off support for unused secondary
1110 architectures). There is also a global
1111 SystemCallArchitectures= setting in system.conf now to turn
1112 off support for non-native system calls system-wide.
1114 * systemd requires a kernel with a working name_to_handle_at(),
1115 please see the kernel config requirements in the README file.
1117 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
1118 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
1119 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
1120 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
1121 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
1122 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
1123 Elia Pinto, Florian Weimer, George McCollister, Goffredo
1124 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
1125 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
1126 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
1127 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
1128 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
1129 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
1130 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
1131 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
1132 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
1133 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
1134 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
1135 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
1136 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
1137 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
1138 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
1139 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
1140 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
1142 -- Berlin, 2014-02-20
1146 * logind has gained support for facilitating privileged input
1147 and drm device access for unprivileged clients. This work is
1148 useful to allow Wayland display servers (and similar
1149 programs, such as kmscon) to run under the user's ID and
1150 access input and drm devices which are normally
1151 protected. When this is used (and the kernel is new enough)
1152 logind will "mute" IO on the file descriptors passed to
1153 Wayland as long as it is in the background and "unmute" it
1154 if it returns into the foreground. This allows secure
1155 session switching without allowing background sessions to
1156 eavesdrop on input and display data. This also introduces
1157 session switching support if VT support is turned off in the
1158 kernel, and on seats that are not seat0.
1160 * A new kernel command line option luks.options= is understood
1161 now which allows specifiying LUKS options for usage for LUKS
1162 encrypted partitions specified with luks.uuid=.
1164 * tmpfiles.d(5) snippets may now use specifier expansion in
1165 path names. More specifically %m, %b, %H, %v, are now
1166 replaced by the local machine id, boot id, hostname, and
1167 kernel version number.
1169 * A new tmpfiles.d(5) command "m" has been introduced which
1170 may be used to change the owner/group/access mode of a file
1171 or directory if it exists, but do nothing if it does not.
1173 * This release removes high-level support for the
1174 MemorySoftLimit= cgroup setting. The underlying kernel
1175 cgroup attribute memory.soft_limit= is currently badly
1176 designed and likely to be removed from the kernel API in its
1177 current form, hence we should not expose it for now.
1179 * The memory.use_hierarchy cgroup attribute is now enabled for
1180 all cgroups systemd creates in the memory cgroup
1181 hierarchy. This option is likely to be come the built-in
1182 default in the kernel anyway, and the non-hierarchial mode
1183 never made much sense in the intrinsically hierarchial
1186 * A new field _SYSTEMD_SLICE= is logged along with all journal
1187 messages containing the slice a message was generated
1188 from. This is useful to allow easy per-customer filtering of
1189 logs among other things.
1191 * systemd-journald will no longer adjust the group of journal
1192 files it creates to the "systemd-journal" group. Instead we
1193 rely on the journal directory to be owned by the
1194 "systemd-journal" group, and its setgid bit set, so that the
1195 kernel file system layer will automatically enforce that
1196 journal files inherit this group assignment. The reason for
1197 this change is that we cannot allow NSS look-ups from
1198 journald which would be necessary to resolve
1199 "systemd-journal" to a numeric GID, because this might
1200 create deadlocks if NSS involves synchronous queries to
1201 other daemons (such as nscd, or sssd) which in turn are
1202 logging clients of journald and might block on it, which
1203 would then dead lock. A tmpfiles.d(5) snippet included in
1204 systemd will make sure the setgid bit and group are
1205 properly set on the journal directory if it exists on every
1206 boot. However, we recommend adjusting it manually after
1207 upgrades too (or from RPM scriptlets), so that the change is
1208 not delayed until next reboot.
1210 * Backlight and random seed files in /var/lib/ have moved into
1211 the /var/lib/systemd/ directory, in order to centralize all
1212 systemd generated files in one directory.
1214 * Boot time performance measurements (as displayed by
1215 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
1216 performance information if that's available to determine how
1217 much time BIOS and boot loader initialization required. With
1218 a sufficiently new BIOS you hence no longer need to boot
1219 with Gummiboot to get access to such information.
1221 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
1222 Cristian Rodríguez, Dave Reisner, David Herrmann, David
1223 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
1224 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
1225 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
1226 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
1227 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1229 -- Berlin, 2013-10-02
1233 * The Restart= option for services now understands a new
1234 on-watchdog setting, which will restart the service
1235 automatically if the service stops sending out watchdog keep
1236 alive messages (as configured with WatchdogSec=).
1238 * The getty generator (which is responsible for bringing up a
1239 getty on configured serial consoles) will no longer only
1240 start a getty on the primary kernel console but on all
1241 others, too. This makes the order in which console= is
1242 specified on the kernel command line less important.
1244 * libsystemd-logind gained a new sd_session_get_vt() call to
1245 retrieve the VT number of a session.
1247 * If the option "tries=0" is set for an entry of /etc/crypttab
1248 its passphrase is queried indefinitely instead of any
1249 maximum number of tries.
1251 * If a service with a configure PID file terminates its PID
1252 file will now be removed automatically if it still exists
1253 afterwards. This should put an end to stale PID files.
1255 * systemd-run will now also take relative binary path names
1256 for execution and no longer insists on absolute paths.
1258 * InaccessibleDirectories= and ReadOnlyDirectories= now take
1259 paths that are optionally prefixed with "-" to indicate that
1260 it should not be considered a failure if they do not exist.
1262 * journalctl -o (and similar commands) now understands a new
1263 output mode "short-precise", it is similar to "short" but
1264 shows timestamps with usec accuracy.
1266 * The option "discard" (as known from Debian) is now
1267 synonymous to "allow-discards" in /etc/crypttab. In fact,
1268 "discard" is preferred now (since it is easier to remember
1271 * Some licensing clean-ups were made, so that more code is now
1272 LGPL-2.1 licensed than before.
1274 * A minimal tool to save/restore the display backlight
1275 brightness across reboots has been added. It will store the
1276 backlight setting as late as possible at shutdown, and
1277 restore it as early as possible during reboot.
1279 * A logic to automatically discover and enable home and swap
1280 partitions on GPT disks has been added. With this in place
1281 /etc/fstab becomes optional for many setups as systemd can
1282 discover certain partitions located on the root disk
1283 automatically. Home partitions are recognized under their
1284 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
1285 partitions are recognized under their GPT type ID
1286 0657fd6da4ab43c484e50933c84b4f4f.
1288 * systemd will no longer pass any environment from the kernel
1289 or initrd to system services. If you want to set an
1290 environment for all services, do so via the kernel command
1291 line systemd.setenv= assignment.
1293 * The systemd-sysctl tool no longer natively reads the file
1294 /etc/sysctl.conf. If desired, the file should be symlinked
1295 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
1296 legacy support by a symlink rather than built-in code, it
1297 also makes the otherwise hidden order of application of the
1298 different files visible. (Note that this partly reverts to a
1299 pre-198 application order of sysctl knobs!)
1301 * The "systemctl set-log-level" and "systemctl dump" commands
1302 have been moved to systemd-analyze.
1304 * systemd-run learned the new --remain-after-exit switch,
1305 which causes the scope unit not to be cleaned up
1306 automatically after the process terminated.
1308 * tmpfiles learned a new --exclude-prefix= switch to exclude
1309 certain paths from operation.
1311 * journald will now automatically flush all messages to disk
1312 as soon as a message of the log priorities CRIT, ALERT or
1315 Contributions from: Andrew Cook, Brandon Philips, Christian
1316 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
1317 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
1318 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
1319 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
1320 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
1321 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
1322 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
1323 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
1324 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
1325 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
1326 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
1327 William Giokas, Zbigniew Jędrzejewski-Szmek
1329 -- Berlin, 2013-09-13
1333 * The documentation has been updated to cover the various new
1334 concepts introduced with 205.
1336 * Unit files now understand the new %v specifier which
1337 resolves to the kernel version string as returned by "uname
1340 * systemctl now supports filtering the unit list output by
1341 load state, active state and sub state, using the new
1344 * "systemctl status" will now show the results of the
1345 condition checks (like ConditionPathExists= and similar) of
1346 the last start attempts of the unit. They are also logged to
1349 * "journalctl -b" may now be used to look for boot output of a
1350 specific boot. Try "journalctl -b -1" for the previous boot,
1351 but the syntax is substantially more powerful.
1353 * "journalctl --show-cursor" has been added which prints the
1354 cursor string the last shown log line. This may then be used
1355 with the new "journalctl --after-cursor=" switch to continue
1356 browsing logs from that point on.
1358 * "journalctl --force" may now be used to force regeneration
1361 * Creation of "dead" device nodes has been moved from udev
1362 into kmod and tmpfiles. Previously, udev would read the kmod
1363 databases to pre-generate dead device nodes based on meta
1364 information contained in kernel modules, so that these would
1365 be auto-loaded on access rather then at boot. As this
1366 does not really have much to do with the exposing actual
1367 kernel devices to userspace this has always been slightly
1368 alien in the udev codebase. Following the new scheme kmod
1369 will now generate a runtime snippet for tmpfiles from the
1370 module meta information and it now is tmpfiles' job to the
1371 create the nodes. This also allows overriding access and
1372 other parameters for the nodes using the usual tmpfiles
1373 facilities. As side effect this allows us to remove the
1374 CAP_SYS_MKNOD capability bit from udevd entirely.
1376 * logind's device ACLs may now be applied to these "dead"
1377 devices nodes too, thus finally allowing managed access to
1378 devices such as /dev/snd/sequencer whithout loading the
1379 backing module right-away.
1381 * A new RPM macro has been added that may be used to apply
1382 tmpfiles configuration during package installation.
1384 * systemd-detect-virt and ConditionVirtualization= now can
1385 detect User-Mode-Linux machines (UML).
1387 * journald will now implicitly log the effective capabilities
1388 set of processes in the message metadata.
1390 * systemd-cryptsetup has gained support for TrueCrypt volumes.
1392 * The initrd interface has been simplified (more specifically,
1393 support for passing performance data via environment
1394 variables and fsck results via files in /run has been
1395 removed). These features were non-essential, and are
1396 nowadays available in a much nicer way by having systemd in
1397 the initrd serialize its state and have the hosts systemd
1398 deserialize it again.
1400 * The udev "keymap" data files and tools to apply keyboard
1401 specific mappings of scan to key codes, and force-release
1402 scan code lists have been entirely replaced by a udev
1403 "keyboard" builtin and a hwdb data file.
1405 * systemd will now honour the kernel's "quiet" command line
1406 argument also during late shutdown, resulting in a
1407 completely silent shutdown when used.
1409 * There's now an option to control the SO_REUSEPORT socket
1410 option in .socket units.
1412 * Instance units will now automatically get a per-template
1413 subslice of system.slice unless something else is explicitly
1414 configured. For example, instances of sshd@.service will now
1415 implicitly be placed in system-sshd.slice rather than
1416 system.slice as before.
1418 * Test coverage support may now be enabled at build time.
1420 Contributions from: Dave Reisner, Frederic Crozat, Harald
1421 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
1422 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
1423 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
1424 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
1425 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
1426 Giokas, Zbigniew Jędrzejewski-Szmek
1428 -- Berlin, 2013-07-23
1432 * Two new unit types have been introduced:
1434 Scope units are very similar to service units, however, are
1435 created out of pre-existing processes -- instead of PID 1
1436 forking off the processes. By using scope units it is
1437 possible for system services and applications to group their
1438 own child processes (worker processes) in a powerful way
1439 which then maybe used to organize them, or kill them
1440 together, or apply resource limits on them.
1442 Slice units may be used to partition system resources in an
1443 hierarchial fashion and then assign other units to them. By
1444 default there are now three slices: system.slice (for all
1445 system services), user.slice (for all user sessions),
1446 machine.slice (for VMs and containers).
1448 Slices and scopes have been introduced primarily in
1449 context of the work to move cgroup handling to a
1450 single-writer scheme, where only PID 1
1451 creates/removes/manages cgroups.
1453 * There's a new concept of "transient" units. In contrast to
1454 normal units these units are created via an API at runtime,
1455 not from configuration from disk. More specifically this
1456 means it is now possible to run arbitrary programs as
1457 independent services, with all execution parameters passed
1458 in via bus APIs rather than read from disk. Transient units
1459 make systemd substantially more dynamic then it ever was,
1460 and useful as a general batch manager.
1462 * logind has been updated to make use of scope and slice units
1463 for managing user sessions. As a user logs in he will get
1464 his own private slice unit, to which all sessions are added
1465 as scope units. We also added support for automatically
1466 adding an instance of user@.service for the user into the
1467 slice. Effectively logind will no longer create cgroup
1468 hierarchies on its own now, it will defer entirely to PID 1
1469 for this by means of scope, service and slice units. Since
1470 user sessions this way become entities managed by PID 1
1471 the output of "systemctl" is now a lot more comprehensive.
1473 * A new mini-daemon "systemd-machined" has been added which
1474 may be used by virtualization managers to register local
1475 VMs/containers. nspawn has been updated accordingly, and
1476 libvirt will be updated shortly. machined will collect a bit
1477 of meta information about the VMs/containers, and assign
1478 them their own scope unit (see above). The collected
1479 meta-data is then made available via the "machinectl" tool,
1480 and exposed in "ps" and similar tools. machined/machinectl
1481 is compile-time optional.
1483 * As discussed earlier, the low-level cgroup configuration
1484 options ControlGroup=, ControlGroupModify=,
1485 ControlGroupPersistent=, ControlGroupAttribute= have been
1486 removed. Please use high-level attribute settings instead as
1487 well as slice units.
1489 * A new bus call SetUnitProperties() has been added to alter
1490 various runtime parameters of a unit. This is primarily
1491 useful to alter cgroup parameters dynamically in a nice way,
1492 but will be extended later on to make more properties
1493 modifiable at runtime. systemctl gained a new set-properties
1494 command that wraps this call.
1496 * A new tool "systemd-run" has been added which can be used to
1497 run arbitrary command lines as transient services or scopes,
1498 while configuring a number of settings via the command
1499 line. This tool is currently very basic, however already
1500 very useful. We plan to extend this tool to even allow
1501 queuing of execution jobs with time triggers from the
1502 command line, similar in fashion to "at".
1504 * nspawn will now inform the user explicitly that kernels with
1505 audit enabled break containers, and suggest the user to turn
1508 * Support for detecting the IMA and AppArmor security
1509 frameworks with ConditionSecurity= has been added.
1511 * journalctl gained a new "-k" switch for showing only kernel
1512 messages, mimicking dmesg output; in addition to "--user"
1513 and "--system" switches for showing only user's own logs
1516 * systemd-delta can now show information about drop-in
1517 snippets extending unit files.
1519 * libsystemd-bus has been substantially updated but is still
1520 not available as public API.
1522 * systemd will now look for the "debug" argument on the kernel
1523 command line and enable debug logging, similar to what
1524 "systemd.log_level=debug" already did before.
1526 * "systemctl set-default", "systemctl get-default" has been
1527 added to configure the default.target symlink, which
1528 controls what to boot into by default.
1530 * "systemctl set-log-level" has been added as a convenient
1531 way to raise and lower systemd logging threshold.
1533 * "systemd-analyze plot" will now show the time the various
1534 generators needed for execution, as well as information
1535 about the unit file loading.
1537 * libsystemd-journal gained a new sd_journal_open_files() call
1538 for opening specific journal files. journactl also gained a
1539 new switch to expose this new functionality. Previously we
1540 only supported opening all files from a directory, or all
1541 files from the system, as opening individual files only is
1542 racy due to journal file rotation.
1544 * systemd gained the new DefaultEnvironment= setting in
1545 /etc/systemd/system.conf to set environment variables for
1548 * If a privileged process logs a journal message with the
1549 OBJECT_PID= field set, then journald will automatically
1550 augment this with additional OBJECT_UID=, OBJECT_GID=,
1551 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
1552 system services want to log events about specific client
1553 processes. journactl/systemctl has been updated to make use
1554 of this information if all log messages regarding a specific
1557 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
1558 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
1559 Reisner, David Coppa, David King, David Strauss, Eelco
1560 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
1561 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
1562 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
1563 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
1564 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
1565 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
1566 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
1567 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
1568 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
1569 Łukasz Stelmach, 장동준
1573 * The Python bindings gained some minimal support for the APIs
1574 exposed by libsystemd-logind.
1576 * ConditionSecurity= gained support for detecting SMACK. Since
1577 this condition already supports SELinux and AppArmor we only
1578 miss IMA for this. Patches welcome!
1580 Contributions from: Karol Lewandowski, Lennart Poettering,
1581 Zbigniew Jędrzejewski-Szmek
1585 * systemd-nspawn will now create /etc/resolv.conf if
1586 necessary, before bind-mounting the host's file onto it.
1588 * systemd-nspawn will now store meta information about a
1589 container on the container's cgroup as extended attribute
1590 fields, including the root directory.
1592 * The cgroup hierarchy has been reworked in many ways. All
1593 objects any of the components systemd creates in the cgroup
1594 tree are now suffixed. More specifically, user sessions are
1595 now placed in cgroups suffixed with ".session", users in
1596 cgroups suffixed with ".user", and nspawn containers in
1597 cgroups suffixed with ".nspawn". Furthermore, all cgroup
1598 names are now escaped in a simple scheme to avoid collision
1599 of userspace object names with kernel filenames. This work
1600 is preparation for making these objects relocatable in the
1601 cgroup tree, in order to allow easy resource partitioning of
1602 these objects without causing naming conflicts.
1604 * systemctl list-dependencies gained the new switches
1605 --plain, --reverse, --after and --before.
1607 * systemd-inhibit now shows the process name of processes that
1608 have taken an inhibitor lock.
1610 * nss-myhostname will now also resolve "localhost"
1611 implicitly. This makes /etc/hosts an optional file and
1612 nicely handles that on IPv6 ::1 maps to both "localhost" and
1615 * libsystemd-logind.so gained a new call
1616 sd_get_machine_names() to enumerate running containers and
1617 VMs (currently only supported by very new libvirt and
1618 nspawn). sd_login_monitor can now be used to watch
1619 VMs/containers coming and going.
1621 * .include is not allowed recursively anymore, and only in
1622 unit files. Usually it is better to use drop-in snippets in
1623 .d/*.conf anyway, as introduced with systemd 198.
1625 * systemd-analyze gained a new "critical-chain" command that
1626 determines the slowest chain of units run during system
1627 boot-up. It is very useful for tracking down where
1628 optimizing boot time is the most beneficial.
1630 * systemd will no longer allow manipulating service paths in
1631 the name=systemd:/system cgroup tree using ControlGroup= in
1632 units. (But is still fine with it in all other dirs.)
1634 * There's a new systemd-nspawn@.service service file that may
1635 be used to easily run nspawn containers as system
1636 services. With the container's root directory in
1637 /var/lib/container/foobar it is now sufficient to run
1638 "systemctl start systemd-nspawn@foobar.service" to boot it.
1640 * systemd-cgls gained a new parameter "--machine" to list only
1641 the processes within a certain container.
1643 * ConditionSecurity= now can check for "apparmor". We still
1644 are lacking checks for SMACK and IMA for this condition
1645 check though. Patches welcome!
1647 * A new configuration file /etc/systemd/sleep.conf has been
1648 added that may be used to configure which kernel operation
1649 systemd is supposed to execute when "suspend", "hibernate"
1650 or "hybrid-sleep" is requested. This makes the new kernel
1651 "freeze" state accessible to the user.
1653 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
1654 the passed argument if applicable.
1656 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
1657 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
1658 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
1659 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
1660 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
1661 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
1662 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
1667 * The output of 'systemctl list-jobs' got some polishing. The
1668 '--type=' argument may now be passed more than once. A new
1669 command 'systemctl list-sockets' has been added which shows
1670 a list of kernel sockets systemd is listening on with the
1671 socket units they belong to, plus the units these socket
1674 * The experimental libsystemd-bus library got substantial
1675 updates to work in conjunction with the (also experimental)
1676 kdbus kernel project. It works well enough to exchange
1677 messages with some sophistication. Note that kdbus is not
1678 ready yet, and the library is mostly an elaborate test case
1679 for now, and not installable.
1681 * systemd gained a new unit 'systemd-static-nodes.service'
1682 that generates static device nodes earlier during boot, and
1683 can run in conjunction with udev.
1685 * libsystemd-login gained a new call sd_pid_get_user_unit()
1686 to retrieve the user systemd unit a process is running
1687 in. This is useful for systems where systemd is used as
1690 * systemd-nspawn now places all containers in the new /machine
1691 top-level cgroup directory in the name=systemd
1692 hierarchy. libvirt will soon do the same, so that we get a
1693 uniform separation of /system, /user and /machine for system
1694 services, user processes and containers/virtual
1695 machines. This new cgroup hierarchy is also useful to stick
1696 stable names to specific container instances, which can be
1697 recognized later this way (this name may be controlled
1698 via systemd-nspawn's new -M switch). libsystemd-login also
1699 gained a new call sd_pid_get_machine_name() to retrieve the
1700 name of the container/VM a specific process belongs to.
1702 * bootchart can now store its data in the journal.
1704 * libsystemd-journal gained a new call
1705 sd_journal_add_conjunction() for AND expressions to the
1706 matching logic. This can be used to express more complex
1707 logical expressions.
1709 * journactl can now take multiple --unit= and --user-unit=
1712 * The cryptsetup logic now understands the "luks.key=" kernel
1713 command line switch for specifying a file to read the
1714 decryption key from. Also, if a configured key file is not
1715 found the tool will now automatically fall back to prompting
1718 * Python systemd.journal module was updated to wrap recently
1719 added functions from libsystemd-journal. The interface was
1720 changed to bring the low level interface in s.j._Reader
1721 closer to the C API, and the high level interface in
1722 s.j.Reader was updated to wrap and convert all data about
1725 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
1726 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
1727 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
1728 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
1729 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
1730 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1734 * journalctl --update-catalog now understands a new --root=
1735 option to operate on catalogs found in a different root
1738 * During shutdown after systemd has terminated all running
1739 services a final killing loop kills all remaining left-over
1740 processes. We will now print the name of these processes
1741 when we send SIGKILL to them, since this usually indicates a
1744 * If /etc/crypttab refers to password files stored on
1745 configured mount points automatic dependencies will now be
1746 generated to ensure the specific mount is established first
1747 before the key file is attempted to be read.
1749 * 'systemctl status' will now show information about the
1750 network sockets a socket unit is listening on.
1752 * 'systemctl status' will also shown information about any
1753 drop-in configuration file for units. (Drop-In configuration
1754 files in this context are files such as
1755 /etc/systemd/systemd/foobar.service.d/*.conf)
1757 * systemd-cgtop now optionally shows summed up CPU times of
1758 cgroups. Press '%' while running cgtop to switch between
1759 percentage and absolute mode. This is useful to determine
1760 which cgroups use up the most CPU time over the entire
1761 runtime of the system. systemd-cgtop has also been updated
1762 to be 'pipeable' for processing with further shell tools.
1764 * 'hostnamectl set-hostname' will now allow setting of FQDN
1767 * The formatting and parsing of time span values has been
1768 changed. The parser now understands fractional expressions
1769 such as "5.5h". The formatter will now output fractional
1770 expressions for all time spans under 1min, i.e. "5.123456s"
1771 rather than "5s 123ms 456us". For time spans under 1s
1772 millisecond values are shown, for those under 1ms
1773 microsecond values are shown. This should greatly improve
1774 all time-related output of systemd.
1776 * libsystemd-login and libsystemd-journal gained new
1777 functions for querying the poll() events mask and poll()
1778 timeout value for integration into arbitrary event
1781 * localectl gained the ability to list available X11 keymaps
1782 (models, layouts, variants, options).
1784 * 'systemd-analyze dot' gained the ability to filter for
1785 specific units via shell-style globs, to create smaller,
1786 more useful graphs. I.e. it is now possible to create simple
1787 graphs of all the dependencies between only target units, or
1788 of all units that Avahi has dependencies with.
1790 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
1791 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
1792 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
1793 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
1794 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
1795 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
1796 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
1800 * The boot-time readahead implementation for rotating media
1801 will now read the read-ahead data in multiple passes which
1802 consist of all read requests made in equidistant time
1803 intervals. This means instead of strictly reading read-ahead
1804 data in its physical order on disk we now try to find a
1805 middle ground between physical and access time order.
1807 * /etc/os-release files gained a new BUILD_ID= field for usage
1808 on operating systems that provide continuous builds of OS
1811 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
1812 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
1813 William Douglas, Zbigniew Jędrzejewski-Szmek
1817 * systemd-python gained an API exposing libsystemd-daemon.
1819 * The SMACK setup logic gained support for uploading CIPSO
1822 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
1823 ReadOnlyDirectories= and InaccessibleDirectories= has
1824 changed. The private /tmp and /var/tmp directories are now
1825 shared by all processes of a service (which means
1826 ExecStartPre= may now leave data in /tmp that ExecStart= of
1827 the same service can still access). When a service is
1828 stopped its temporary directories are immediately deleted
1829 (normal clean-up with tmpfiles is still done in addition to
1832 * By default, systemd will now set a couple of sysctl
1833 variables in the kernel: the safe sysrq options are turned
1834 on, IP route verification is turned on, and source routing
1835 disabled. The recently added hardlink and softlink
1836 protection of the kernel is turned on. These settings should
1837 be reasonably safe, and good defaults for all new systems.
1839 * The predictable network naming logic may now be turned off
1840 with a new kernel command line switch: net.ifnames=0.
1842 * A new libsystemd-bus module has been added that implements a
1843 pretty complete D-Bus client library. For details see:
1845 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
1847 * journald will now explicitly flush the journal files to disk
1848 at the latest 5min after each write. The file will then also
1849 be marked offline until the next write. This should increase
1850 reliability in case of a crash. The synchronization delay
1851 can be configured via SyncIntervalSec= in journald.conf.
1853 * There's a new remote-fs-setup.target unit that can be used
1854 to pull in specific services when at least one remote file
1855 system is to be mounted.
1857 * There are new targets timers.target and paths.target as
1858 canonical targets to pull user timer and path units in
1859 from. This complements sockets.target with a similar
1860 purpose for socket units.
1862 * libudev gained a new call udev_device_set_attribute_value()
1863 to set sysfs attributes of a device.
1865 * The udev daemon now sets the default number of worker
1866 processes executed in parallel based on the number of available
1867 CPUs instead of the amount of available RAM. This is supposed
1868 to provide a more reliable default and limit a too aggressive
1869 paralellism for setups with 1000s of devices connected.
1871 Contributions from: Auke Kok, Colin Walters, Cristian
1872 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
1873 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
1874 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
1875 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
1876 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
1877 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
1878 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
1879 Zbigniew Jędrzejewski-Szmek
1883 * Configuration of unit files may now be extended via drop-in
1884 files without having to edit/override the unit files
1885 themselves. More specifically, if the administrator wants to
1886 change one value for a service file foobar.service he can
1887 now do so by dropping in a configuration snippet into
1888 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
1889 will load all these snippets and apply them on top of the
1890 main unit configuration file, possibly extending or
1891 overriding its settings. Using these drop-in snippets is
1892 generally nicer than the two earlier options for changing
1893 unit files locally: copying the files from
1894 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
1895 them there; or creating a new file in /etc/systemd/system/
1896 that incorporates the original one via ".include". Drop-in
1897 snippets into these .d/ directories can be placed in any
1898 directory systemd looks for units in, and the usual
1899 overriding semantics between /usr/lib, /etc and /run apply
1902 * Most unit file settings which take lists of items can now be
1903 reset by assigning the empty string to them. For example,
1904 normally, settings such as Environment=FOO=BAR append a new
1905 environment variable assignment to the environment block,
1906 each time they are used. By assigning Environment= the empty
1907 string the environment block can be reset to empty. This is
1908 particularly useful with the .d/*.conf drop-in snippets
1909 mentioned above, since this adds the ability to reset list
1910 settings from vendor unit files via these drop-ins.
1912 * systemctl gained a new "list-dependencies" command for
1913 listing the dependencies of a unit recursively.
1915 * Inhibitors are now honored and listed by "systemctl
1916 suspend", "systemctl poweroff" (and similar) too, not only
1917 GNOME. These commands will also list active sessions by
1920 * Resource limits (as exposed by the various control group
1921 controllers) can now be controlled dynamically at runtime
1922 for all units. More specifically, you can now use a command
1923 like "systemctl set-cgroup-attr foobar.service cpu.shares
1924 2000" to alter the CPU shares a specific service gets. These
1925 settings are stored persistently on disk, and thus allow the
1926 administrator to easily adjust the resource usage of
1927 services with a few simple commands. This dynamic resource
1928 management logic is also available to other programs via the
1929 bus. Almost any kernel cgroup attribute and controller is
1932 * systemd-vconsole-setup will now copy all font settings to
1933 all allocated VTs, where it previously applied them only to
1936 * libsystemd-login gained the new sd_session_get_tty() API
1939 * This release drops support for a few legacy or
1940 distribution-specific LSB facility names when parsing init
1941 scripts: $x-display-manager, $mail-transfer-agent,
1942 $mail-transport-agent, $mail-transfer-agent, $smtp,
1943 $null. Also, the mail-transfer-agent.target unit backing
1944 this has been removed. Distributions which want to retain
1945 compatibility with this should carry the burden for
1946 supporting this themselves and patch support for these back
1947 in, if they really need to. Also, the facilities $syslog and
1948 $local_fs are now ignored, since systemd does not support
1949 early-boot LSB init scripts anymore, and these facilities
1950 are implied anyway for normal services. syslog.target has
1953 * There are new bus calls on PID1's Manager object for
1954 cancelling jobs, and removing snapshot units. Previously,
1955 both calls were only available on the Job and Snapshot
1958 * systemd-journal-gatewayd gained SSL support.
1960 * The various "environment" files, such as /etc/locale.conf
1961 now support continuation lines with a backslash ("\") as
1962 last character in the line, similarly in style (but different)
1963 to how this is supported in shells.
1965 * For normal user processes the _SYSTEMD_USER_UNIT= field is
1966 now implicitly appended to every log entry logged. systemctl
1967 has been updated to filter by this field when operating on a
1968 user systemd instance.
1970 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
1971 CAP_AUDIT_CONTROL capabilities to the capabilities set for
1972 the container. This makes it easier to boot unmodified
1973 Fedora systems in a container, which however still requires
1974 audit=0 to be passed on the kernel command line. Auditing in
1975 kernel and userspace is unfortunately still too broken in
1976 context of containers, hence we recommend compiling it out
1977 of the kernel or using audit=0. Hopefully this will be fixed
1978 one day for good in the kernel.
1980 * nspawn gained the new --bind= and --bind-ro= parameters to
1981 bind mount specific directories from the host into the
1984 * nspawn will now mount its own devpts file system instance
1985 into the container, in order not to leak pty devices from
1986 the host into the container.
1988 * systemd will now read the firmware boot time performance
1989 information from the EFI variables, if the used boot loader
1990 supports this, and takes it into account for boot performance
1991 analysis via "systemd-analyze". This is currently supported
1992 only in conjunction with Gummiboot, but could be supported
1993 by other boot loaders too. For details see:
1995 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
1997 * A new generator has been added that automatically mounts the
1998 EFI System Partition (ESP) to /boot, if that directory
1999 exists, is empty, and no other file system has been
2000 configured to be mounted there.
2002 * logind will now send out PrepareForSleep(false) out
2003 unconditionally, after coming back from suspend. This may be
2004 used by applications as asynchronous notification for
2005 system resume events.
2007 * "systemctl unlock-sessions" has been added, that allows
2008 unlocking the screens of all user sessions at once, similar
2009 to how "systemctl lock-sessions" already locked all users
2010 sessions. This is backed by a new D-Bus call UnlockSessions().
2012 * "loginctl seat-status" will now show the master device of a
2013 seat. (i.e. the device of a seat that needs to be around for
2014 the seat to be considered available, usually the graphics
2017 * tmpfiles gained a new "X" line type, that allows
2018 configuration of files and directories (with wildcards) that
2019 shall be excluded from automatic cleanup ("aging").
2021 * udev default rules set the device node permissions now only
2022 at "add" events, and do not change them any longer with a
2023 later "change" event.
2025 * The log messages for lid events and power/sleep keypresses
2026 now carry a message ID.
2028 * We now have a substantially larger unit test suite, but this
2029 continues to be work in progress.
2031 * udevadm hwdb gained a new --root= parameter to change the
2032 root directory to operate relative to.
2034 * logind will now issue a background sync() request to the kernel
2035 early at shutdown, so that dirty buffers are flushed to disk early
2036 instead of at the last moment, in order to optimize shutdown
2039 * A new bootctl tool has been added that is an interface for
2040 certain boot loader operations. This is currently a preview
2041 and is likely to be extended into a small mechanism daemon
2042 like timedated, localed, hostnamed, and can be used by
2043 graphical UIs to enumerate available boot options, and
2044 request boot into firmware operations.
2046 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
2047 the rest of the package. It also has been updated to work
2048 correctly in initrds.
2050 * Policykit previously has been runtime optional, and is now
2051 also compile time optional via a configure switch.
2053 * systemd-analyze has been reimplemented in C. Also "systemctl
2054 dot" has moved into systemd-analyze.
2056 * "systemctl status" with no further parameters will now print
2057 the status of all active or failed units.
2059 * Operations such as "systemctl start" can now be executed
2060 with a new mode "--irreversible" which may be used to queue
2061 operations that cannot accidentally be reversed by a later
2062 job queuing. This is by default used to make shutdown
2063 requests more robust.
2065 * The Python API of systemd now gained a new module for
2066 reading journal files.
2068 * A new tool kernel-install has been added that can install
2069 kernel images according to the Boot Loader Specification:
2071 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
2073 * Boot time console output has been improved to provide
2074 animated boot time output for hanging jobs.
2076 * A new tool systemd-activate has been added which can be used
2077 to test socket activation with, directly from the command
2078 line. This should make it much easier to test and debug
2079 socket activation in daemons.
2081 * journalctl gained a new "--reverse" (or -r) option to show
2082 journal output in reverse order (i.e. newest line first).
2084 * journalctl gained a new "--pager-end" (or -e) option to jump
2085 to immediately jump to the end of the journal in the
2086 pager. This is only supported in conjunction with "less".
2088 * journalctl gained a new "--user-unit=" option, that works
2089 similarly to "--unit=" but filters for user units rather than
2092 * A number of unit files to ease adoption of systemd in
2093 initrds has been added. This moves some minimal logic from
2094 the various initrd implementations into systemd proper.
2096 * The journal files are now owned by a new group
2097 "systemd-journal", which exists specifically to allow access
2098 to the journal, and nothing else. Previously, we used the
2099 "adm" group for that, which however possibly covers more
2100 than just journal/log file access. This new group is now
2101 already used by systemd-journal-gatewayd to ensure this
2102 daemon gets access to the journal files and as little else
2103 as possible. Note that "make install" will also set FS ACLs
2104 up for /var/log/journal to give "adm" and "wheel" read
2105 access to it, in addition to "systemd-journal" which owns
2106 the journal files. We recommend that packaging scripts also
2107 add read access to "adm" + "wheel" to /var/log/journal, and
2108 all existing/future journal files. To normal users and
2109 administrators little changes, however packagers need to
2110 ensure to create the "systemd-journal" system group at
2111 package installation time.
2113 * The systemd-journal-gatewayd now runs as unprivileged user
2114 systemd-journal-gateway:systemd-journal-gateway. Packaging
2115 scripts need to create these system user/group at
2118 * timedated now exposes a new boolean property CanNTP that
2119 indicates whether a local NTP service is available or not.
2121 * systemd-detect-virt will now also detect xen PVs
2123 * The pstore file system is now mounted by default, if it is
2126 * In addition to the SELinux and IMA policies we will now also
2127 load SMACK policies at early boot.
2129 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
2130 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
2131 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
2132 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
2133 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
2134 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
2135 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
2136 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
2137 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
2138 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
2139 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
2140 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
2141 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
2142 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
2146 * Timer units now support calendar time events in addition to
2147 monotonic time events. That means you can now trigger a unit
2148 based on a calendar time specification such as "Thu,Fri
2149 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
2150 or fifth day of any month of the year 2013, given that it is
2151 a thursday or friday. This brings timer event support
2152 considerably closer to cron's capabilities. For details on
2153 the supported calendar time specification language see
2156 * udev now supports a number of different naming policies for
2157 network interfaces for predictable names, and a combination
2158 of these policies is now the default. Please see this wiki
2159 document for details:
2161 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
2163 * Auke Kok's bootchart implementation has been added to the
2164 systemd tree. It is an optional component that can graph the
2165 boot in quite some detail. It is one of the best bootchart
2166 implementations around and minimal in its code and
2169 * nss-myhostname has been integrated into the systemd source
2170 tree. nss-myhostname guarantees that the local hostname
2171 always stays resolvable via NSS. It has been a weak
2172 requirement of systemd-hostnamed since a long time, and
2173 since its code is actually trivial we decided to just
2174 include it in systemd's source tree. It can be turned off
2175 with a configure switch.
2177 * The read-ahead logic is now capable of properly detecting
2178 whether a btrfs file system is on SSD or rotating media, in
2179 order to optimize the read-ahead scheme. Previously, it was
2180 only capable of detecting this on traditional file systems
2183 * In udev, additional device properties are now read from the
2184 IAB in addition to the OUI database. Also, Bluetooth company
2185 identities are attached to the devices as well.
2187 * In service files %U may be used as specifier that is
2188 replaced by the configured user name of the service.
2190 * nspawn may now be invoked without a controlling TTY. This
2191 makes it suitable for invocation as its own service. This
2192 may be used to set up a simple containerized server system
2193 using only core OS tools.
2195 * systemd and nspawn can now accept socket file descriptors
2196 when they are started for socket activation. This enables
2197 implementation of socket activated nspawn
2198 containers. i.e. think about autospawning an entire OS image
2199 when the first SSH or HTTP connection is received. We expect
2200 that similar functionality will also be added to libvirt-lxc
2203 * journalctl will now suppress ANSI color codes when
2204 presenting log data.
2206 * systemctl will no longer show control group information for
2207 a unit if a the control group is empty anyway.
2209 * logind can now automatically suspend/hibernate/shutdown the
2212 * /etc/machine-info and hostnamed now also expose the chassis
2213 type of the system. This can be used to determine whether
2214 the local system is a laptop, desktop, handset or
2215 tablet. This information may either be configured by the
2216 user/vendor or is automatically determined from ACPI and DMI
2217 information if possible.
2219 * A number of PolicyKit actions are now bound together with
2220 "imply" rules. This should simplify creating UIs because
2221 many actions will now authenticate similar ones as well.
2223 * Unit files learnt a new condition ConditionACPower= which
2224 may be used to conditionalize a unit depending on whether an
2225 AC power source is connected or not, of whether the system
2226 is running on battery power.
2228 * systemctl gained a new "is-failed" verb that may be used in
2229 shell scripts and suchlike to check whether a specific unit
2230 is in the "failed" state.
2232 * The EnvironmentFile= setting in unit files now supports file
2233 globbing, and can hence be used to easily read a number of
2234 environment files at once.
2236 * systemd will no longer detect and recognize specific
2237 distributions. All distribution-specific #ifdeffery has been
2238 removed, systemd is now fully generic and
2239 distribution-agnostic. Effectively, not too much is lost as
2240 a lot of the code is still accessible via explicit configure
2241 switches. However, support for some distribution specific
2242 legacy configuration file formats has been dropped. We
2243 recommend distributions to simply adopt the configuration
2244 files everybody else uses now and convert the old
2245 configuration from packaging scripts. Most distributions
2246 already did that. If that's not possible or desirable,
2247 distributions are welcome to forward port the specific
2248 pieces of code locally from the git history.
2250 * When logging a message about a unit systemd will now always
2251 log the unit name in the message meta data.
2253 * localectl will now also discover system locale data that is
2254 not stored in locale archives, but directly unpacked.
2256 * logind will no longer unconditionally use framebuffer
2257 devices as seat masters, i.e. as devices that are required
2258 to be existing before a seat is considered preset. Instead,
2259 it will now look for all devices that are tagged as
2260 "seat-master" in udev. By default framebuffer devices will
2261 be marked as such, but depending on local systems other
2262 devices might be marked as well. This may be used to
2263 integrate graphics cards using closed source drivers (such
2264 as NVidia ones) more nicely into logind. Note however, that
2265 we recommend using the open source NVidia drivers instead,
2266 and no udev rules for the closed-source drivers will be
2267 shipped from us upstream.
2269 Contributions from: Adam Williamson, Alessandro Crismani, Auke
2270 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
2271 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
2272 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
2273 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
2274 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
2275 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
2276 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
2277 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
2278 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
2279 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
2280 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
2285 * udev gained support for loading additional device properties
2286 from an indexed database that is keyed by vendor/product IDs
2287 and similar device identifiers. For the beginning this
2288 "hwdb" is populated with data from the well-known PCI and
2289 USB database, but also includes PNP, ACPI and OID data. In
2290 the longer run this indexed database shall grow into
2291 becoming the one central database for non-essential
2292 userspace device metadata. Previously, data from the PCI/USB
2293 database was only attached to select devices, since the
2294 lookup was a relatively expensive operation due to O(n) time
2295 complexity (with n being the number of entries in the
2296 database). Since this is now O(1), we decided to add in this
2297 data for all devices where this is available, by
2298 default. Note that the indexed database needs to be rebuilt
2299 when new data files are installed. To achieve this you need
2300 to update your packaging scripts to invoke "udevadm hwdb
2301 --update" after installation of hwdb data files. For
2302 RPM-based distributions we introduced the new
2303 %udev_hwdb_update macro for this purpose.
2305 * The Journal gained support for the "Message Catalog", an
2306 indexed database to link up additional information with
2307 journal entries. For further details please check:
2309 http://www.freedesktop.org/wiki/Software/systemd/catalog
2311 The indexed message catalog database also needs to be
2312 rebuilt after installation of message catalog files. Use
2313 "journalctl --update-catalog" for this. For RPM-based
2314 distributions we introduced the %journal_catalog_update
2315 macro for this purpose.
2317 * The Python Journal bindings gained support for the standard
2318 Python logging framework.
2320 * The Journal API gained new functions for checking whether
2321 the underlying file system of a journal file is capable of
2322 properly reporting file change notifications, or whether
2323 applications that want to reflect journal changes "live"
2324 need to recheck journal files continuously in appropriate
2327 * It is now possible to set the "age" field for tmpfiles
2328 entries to 0, indicating that files matching this entry
2329 shall always be removed when the directories are cleaned up.
2331 * coredumpctl gained a new "gdb" verb which invokes gdb
2332 right-away on the selected coredump.
2334 * There's now support for "hybrid sleep" on kernels that
2335 support this, in addition to "suspend" and "hibernate". Use
2336 "systemctl hybrid-sleep" to make use of this.
2338 * logind's HandleSuspendKey= setting (and related settings)
2339 now gained support for a new "lock" setting to simply
2340 request the screen lock on all local sessions, instead of
2341 actually executing a suspend or hibernation.
2343 * systemd will now mount the EFI variables file system by
2346 * Socket units now gained support for configuration of the
2347 SMACK security label.
2349 * timedatectl will now output the time of the last and next
2350 daylight saving change.
2352 * We dropped support for various legacy and distro-specific
2353 concepts, such as insserv, early-boot SysV services
2354 (i.e. those for non-standard runlevels such as 'b' or 'S')
2355 or ArchLinux /etc/rc.conf support. We recommend the
2356 distributions who still need support this to either continue
2357 to maintain the necessary patches downstream, or find a
2358 different solution. (Talk to us if you have questions!)
2360 * Various systemd components will now bypass PolicyKit checks
2361 for root and otherwise handle properly if PolicyKit is not
2362 found to be around. This should fix most issues for
2363 PolicyKit-less systems. Quite frankly this should have been
2364 this way since day one. It is absolutely our intention to
2365 make systemd work fine on PolicyKit-less systems, and we
2366 consider it a bug if something does not work as it should if
2367 PolicyKit is not around.
2369 * For embedded systems it is now possible to build udev and
2370 systemd without blkid and/or kmod support.
2372 * "systemctl switch-root" is now capable of switching root
2373 more than once. I.e. in addition to transitions from the
2374 initrd to the host OS it is now possible to transition to
2375 further OS images from the host. This is useful to implement
2376 offline updating tools.
2378 * Various other additions have been made to the RPM macros
2379 shipped with systemd. Use %udev_rules_update() after
2380 installing new udev rules files. %_udevhwdbdir,
2381 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
2382 %_sysctldir are now available which resolve to the right
2383 directories for packages to place various data files in.
2385 * journalctl gained the new --full switch (in addition to
2386 --all, to disable ellipsation for long messages.
2388 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
2389 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
2390 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
2391 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
2392 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
2393 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
2394 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
2395 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
2396 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
2400 * journalctl gained new --since= and --until= switches to
2401 filter by time. It also now supports nice filtering for
2402 units via --unit=/-u.
2404 * Type=oneshot services may use ExecReload= and do the
2407 * The journal daemon now supports time-based rotation and
2408 vacuuming, in addition to the usual disk-space based
2411 * The journal will now index the available field values for
2412 each field name. This enables clients to show pretty drop
2413 downs of available match values when filtering. The bash
2414 completion of journalctl has been updated
2415 accordingly. journalctl gained a new switch -F to list all
2416 values a certain field takes in the journal database.
2418 * More service events are now written as structured messages
2419 to the journal, and made recognizable via message IDs.
2421 * The timedated, localed and hostnamed mini-services which
2422 previously only provided support for changing time, locale
2423 and hostname settings from graphical DEs such as GNOME now
2424 also have a minimal (but very useful) text-based client
2425 utility each. This is probably the nicest way to changing
2426 these settings from the command line now, especially since
2427 it lists available options and is fully integrated with bash
2430 * There's now a new tool "systemd-coredumpctl" to list and
2431 extract coredumps from the journal.
2433 * We now install a README each in /var/log/ and
2434 /etc/rc.d/init.d explaining where the system logs and init
2435 scripts went. This hopefully should help folks who go to
2436 that dirs and look into the otherwise now empty void and
2437 scratch their heads.
2439 * When user-services are invoked (by systemd --user) the
2440 $MANAGERPID env var is set to the PID of systemd.
2442 * SIGRTMIN+24 when sent to a --user instance will now result
2443 in immediate termination of systemd.
2445 * gatewayd received numerous feature additions such as a
2446 "follow" mode, for live syncing and filtering.
2448 * browse.html now allows filtering and showing detailed
2449 information on specific entries. Keyboard navigation and
2450 mouse screen support has been added.
2452 * gatewayd/journalctl now supports HTML5/JSON
2453 Server-Sent-Events as output.
2455 * The SysV init script compatibility logic will now
2456 heuristically determine whether a script supports the
2457 "reload" verb, and only then make this available as
2460 * "systemctl status --follow" has been removed, use "journalctl
2463 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
2464 have been removed since they are hardly useful to be
2467 * And I'd like to take the opportunity to specifically mention
2468 Zbigniew for his great contributions. Zbigniew, you rock!
2470 Contributions from: Andrew Eikum, Christian Hesse, Colin
2471 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
2472 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
2473 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
2474 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
2475 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
2476 Jędrzejewski-Szmek, Сковорода Никита Андреевич
2480 * If /etc/vconsole.conf is non-existent or empty we will no
2481 longer load any console font or key map at boot by
2482 default. Instead the kernel defaults will be left
2483 intact. This is definitely the right thing to do, as no
2484 configuration should mean no configuration, and hard-coding
2485 font names that are different on all archs is probably a bad
2486 idea. Also, the kernel default key map and font should be
2487 good enough for most cases anyway, and mostly identical to
2488 the userspace fonts/key maps we previously overloaded them
2489 with. If distributions want to continue to default to a
2490 non-kernel font or key map they should ship a default
2491 /etc/vconsole.conf with the appropriate contents.
2493 Contributions from: Colin Walters, Daniel J Walsh, Dave
2494 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
2495 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
2499 * journalctl gained a new --cursor= switch to show entries
2500 starting from the specified location in the journal.
2502 * We now enforce a size limit on journal entry fields exported
2503 with "-o json" in journalctl. Fields larger than 4K will be
2504 assigned null. This can be turned off with --all.
2506 * An (optional) journal gateway daemon is now available as
2507 "systemd-journal-gatewayd.service". This service provides
2508 access to the journal via HTTP and JSON. This functionality
2509 will be used to implement live log synchronization in both
2510 pull and push modes, but has various other users too, such
2511 as easy log access for debugging of embedded devices. Right
2512 now it is already useful to retrieve the journal via HTTP:
2514 # systemctl start systemd-journal-gatewayd.service
2515 # wget http://localhost:19531/entries
2517 This will download the journal contents in a
2518 /var/log/messages compatible format. The same as JSON:
2520 # curl -H"Accept: application/json" http://localhost:19531/entries
2522 This service is also accessible via a web browser where a
2523 single static HTML5 app is served that uses the JSON logic
2524 to enable the user to do some basic browsing of the
2525 journal. This will be extended later on. Here's an example
2526 screenshot of this app in its current state:
2528 http://0pointer.de/public/journal-gatewayd
2530 Contributions from: Kay Sievers, Lennart Poettering, Robert
2531 Milasan, Tom Gundersen
2535 * The bash completion logic is now available for journalctl
2538 * We do not mount the "cpuset" controller anymore together with
2539 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
2540 started if no parameters are assigned to it. "cpuset" hence
2541 broke code that assumed it it could create "cpu" groups and
2544 * journalctl -f will now subscribe to terminal size changes,
2545 and line break accordingly.
2547 Contributions from: Dave Reisner, Kay Sievers, Lennart
2548 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
2552 * nspawn will now create a symlink /etc/localtime in the
2553 container environment, copying the host's timezone
2554 setting. Previously this has been done via a bind mount, but
2555 since symlinks cannot be bind mounted this has now been
2556 changed to create/update the appropriate symlink.
2558 * journalctl -n's line number argument is now optional, and
2559 will default to 10 if omitted.
2561 * journald will now log the maximum size the journal files may
2562 take up on disk. This is particularly useful if the default
2563 built-in logic of determining this parameter from the file
2564 system size is used. Use "systemctl status
2565 systemd-journald.service" to see this information.
2567 * The multi-seat X wrapper tool has been stripped down. As X
2568 is now capable of enumerating graphics devices via udev in a
2569 seat-aware way the wrapper is not strictly necessary
2570 anymore. A stripped down temporary stop-gap is still shipped
2571 until the upstream display managers have been updated to
2572 fully support the new X logic. Expect this wrapper to be
2573 removed entirely in one of the next releases.
2575 * HandleSleepKey= in logind.conf has been split up into
2576 HandleSuspendKey= and HandleHibernateKey=. The old setting
2577 is not available anymore. X11 and the kernel are
2578 distuingishing between these keys and we should too. This
2579 also means the inhibition lock for these keys has been split
2582 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
2583 Poettering, Lukas Nykryn, Václav Pavlín
2587 * Whenever a unit changes state we will now log this to the
2588 journal and show along the unit's own log output in
2591 * ConditionPathIsMountPoint= can now properly detect bind
2592 mount points too. (Previously, a bind mount of one file
2593 system to another place in the same file system could not be
2594 detected as mount, since they shared struct stat's st_dev
2597 * We will now mount the cgroup controllers cpu, cpuacct,
2598 cpuset and the controllers net_cls, net_prio together by
2601 * nspawn containers will now have a virtualized boot
2602 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
2603 over with a randomized ID at container initialization). This
2604 has the effect of making "journalctl -b" do the right thing
2607 * The JSON output journal serialization has been updated not
2608 to generate "endless" list objects anymore, but rather one
2609 JSON object per line. This is more in line how most JSON
2610 parsers expect JSON objects. The new output mode
2611 "json-pretty" has been added to provide similar output, but
2612 neatly aligned for readability by humans.
2614 * We dropped all explicit sync() invocations in the shutdown
2615 code. The kernel does this implicitly anyway in the kernel
2616 reboot() syscall. halt(8)'s -n option is now a compatibility
2619 * We now support virtualized reboot() in containers, as
2620 supported by newer kernels. We will fall back to exit() if
2621 CAP_SYS_REBOOT is not available to the container. Also,
2622 nspawn makes use of this now and will actually reboot the
2623 container if the containerized OS asks for that.
2625 * journalctl will only show local log output by default
2626 now. Use --merge (-m) to show remote log output, too.
2628 * libsystemd-journal gained the new sd_journal_get_usage()
2629 call to determine the current disk usage of all journal
2630 files. This is exposed in the new "journalctl --disk-usage"
2633 * journald gained a new configuration setting SplitMode= in
2634 journald.conf which may be used to control how user journals
2635 are split off. See journald.conf(5) for details.
2637 * A new condition type ConditionFileNotEmpty= has been added.
2639 * tmpfiles' "w" lines now support file globbing, to write
2640 multiple files at once.
2642 * We added Python bindings for the journal submission
2643 APIs. More Python APIs for a number of selected APIs will
2644 likely follow. Note that we intend to add native bindings
2645 only for the Python language, as we consider it common
2646 enough to deserve bindings shipped within systemd. There are
2647 various projects outside of systemd that provide bindings
2648 for languages such as PHP or Lua.
2650 * Many conditions will now resolve specifiers such as %i. In
2651 addition, PathChanged= and related directives of .path units
2652 now support specifiers as well.
2654 * There's now a new RPM macro definition for the system preset
2657 * journald will now warn if it ca not forward a message to the
2658 syslog daemon because its socket is full.
2660 * timedated will no longer write or process /etc/timezone,
2661 except on Debian. As we do not support late mounted /usr
2662 anymore /etc/localtime always being a symlink is now safe,
2663 and hence the information in /etc/timezone is not necessary
2666 * logind will now always reserve one VT for a text getty (VT6
2667 by default). Previously if more than 6 X sessions where
2668 started they took up all the VTs with auto-spawned gettys,
2669 so that no text gettys were available anymore.
2671 * udev will now automatically inform the btrfs kernel logic
2672 about btrfs RAID components showing up. This should make
2673 simple hotplug based btrfs RAID assembly work.
2675 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
2676 (but not for its children which will stay at the kernel
2677 default). This should allow setups with a lot more listening
2680 * systemd will now always pass the configured timezone to the
2681 kernel at boot. timedated will do the same when the timezone
2684 * logind's inhibition logic has been updated. By default,
2685 logind will now handle the lid switch, the power and sleep
2686 keys all the time, even in graphical sessions. If DEs want
2687 to handle these events on their own they should take the new
2688 handle-power-key, handle-sleep-key and handle-lid-switch
2689 inhibitors during their runtime. A simple way to achiveve
2690 that is to invoke the DE wrapped in an invocation of:
2692 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
2694 * Access to unit operations is now checked via SELinux taking
2695 the unit file label and client process label into account.
2697 * systemd will now notify the administrator in the journal
2698 when he over-mounts a non-empty directory.
2700 * There are new specifiers that are resolved in unit files,
2701 for the host name (%H), the machine ID (%m) and the boot ID
2704 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
2705 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
2706 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
2707 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
2708 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
2709 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
2710 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
2714 * Support for reading structured kernel messages from
2715 /dev/kmsg has now been added and is enabled by default.
2717 * Support for reading kernel messages from /proc/kmsg has now
2718 been removed. If you want kernel messages in the journal
2719 make sure to run a recent kernel (>= 3.5) that supports
2720 reading structured messages from /dev/kmsg (see
2721 above). /proc/kmsg is now exclusive property of classic
2722 syslog daemons again.
2724 * The libudev API gained the new
2725 udev_device_new_from_device_id() call.
2727 * The logic for file system namespace (ReadOnlyDirectory=,
2728 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
2729 require pivot_root() anymore. This means fewer temporary
2730 directories are created below /tmp for this feature.
2732 * nspawn containers will now see and receive all submounts
2733 made on the host OS below the root file system of the
2736 * Forward Secure Sealing is now supported for Journal files,
2737 which provide cryptographical sealing of journal files so
2738 that attackers cannot alter log history anymore without this
2739 being detectable. Lennart will soon post a blog story about
2740 this explaining it in more detail.
2742 * There are two new service settings RestartPreventExitStatus=
2743 and SuccessExitStatus= which allow configuration of exit
2744 status (exit code or signal) which will be excepted from the
2745 restart logic, resp. consider successful.
2747 * journalctl gained the new --verify switch that can be used
2748 to check the integrity of the structure of journal files and
2749 (if Forward Secure Sealing is enabled) the contents of
2752 * nspawn containers will now be run with /dev/stdin, /dev/fd/
2753 and similar symlinks pre-created. This makes running shells
2754 as container init process a lot more fun.
2756 * The fstab support can now handle PARTUUID= and PARTLABEL=
2759 * A new ConditionHost= condition has been added to match
2760 against the hostname (with globs) and machine ID. This is
2761 useful for clusters where a single OS image is used to
2762 provision a large number of hosts which shall run slightly
2763 different sets of services.
2765 * Services which hit the restart limit will now be placed in a
2768 Contributions from: Bertram Poettering, Dave Reisner, Huang
2769 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
2770 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
2774 * When running in --user mode systemd will now become a
2775 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
2776 tree a lot more organized.
2778 * A new PartOf= unit dependency type has been introduced that
2779 may be used to group services in a natural way.
2781 * "systemctl enable" may now be used to enable instances of
2784 * journalctl now prints error log levels in red, and
2785 warning/notice log levels in bright white. It also supports
2786 filtering by log level now.
2788 * cgtop gained a new -n switch (similar to top), to configure
2789 the maximum number of iterations to run for. It also gained
2790 -b, to run in batch mode (accepting no input).
2792 * The suffix ".service" may now be omitted on most systemctl
2793 command lines involving service unit names.
2795 * There's a new bus call in logind to lock all sessions, as
2796 well as a loginctl verb for it "lock-sessions".
2798 * libsystemd-logind.so gained a new call sd_journal_perror()
2799 that works similar to libc perror() but logs to the journal
2800 and encodes structured information about the error number.
2802 * /etc/crypttab entries now understand the new keyfile-size=
2805 * shutdown(8) now can send a (configurable) wall message when
2806 a shutdown is cancelled.
2808 * The mount propagation mode for the root file system will now
2809 default to "shared", which is useful to make containers work
2810 nicely out-of-the-box so that they receive new mounts from
2811 the host. This can be undone locally by running "mount
2812 --make-rprivate /" if needed.
2814 * The prefdm.service file has been removed. Distributions
2815 should maintain this unit downstream if they intend to keep
2816 it around. However, we recommend writing normal unit files
2817 for display managers instead.
2819 * Since systemd is a crucial part of the OS we will now
2820 default to a number of compiler switches that improve
2821 security (hardening) such as read-only relocations, stack
2822 protection, and suchlike.
2824 * The TimeoutSec= setting for services is now split into
2825 TimeoutStartSec= and TimeoutStopSec= to allow configuration
2826 of individual time outs for the start and the stop phase of
2829 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
2830 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
2831 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
2832 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
2833 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
2834 Gundersen, Zbigniew Jędrzejewski-Szmek
2838 * The journal and id128 C APIs are now fully documented as man
2841 * Extra safety checks have been added when transitioning from
2842 the initial RAM disk to the main system to avoid accidental
2845 * /etc/crypttab entries now understand the new keyfile-offset=
2848 * systemctl -t can now be used to filter by unit load state.
2850 * The journal C API gained the new sd_journal_wait() call to
2851 make writing synchronous journal clients easier.
2853 * journalctl gained the new -D switch to show journals from a
2856 * journalctl now displays a special marker between log
2857 messages of two different boots.
2859 * The journal is now explicitly flushed to /var via a service
2860 systemd-journal-flush.service, rather than implicitly simply
2861 by seeing /var/log/journal to be writable.
2863 * journalctl (and the journal C APIs) can now match for much
2864 more complex expressions, with alternatives and
2867 * When transitioning from the initial RAM disk to the main
2868 system we will now kill all processes in a killing spree to
2869 ensure no processes stay around by accident.
2871 * Three new specifiers may be used in unit files: %u, %h, %s
2872 resolve to the user name, user home directory resp. user
2873 shell. This is useful for running systemd user instances.
2875 * We now automatically rotate journal files if their data
2876 object hash table gets a fill level > 75%. We also size the
2877 hash table based on the configured maximum file size. This
2878 together should lower hash collisions drastically and thus
2879 speed things up a bit.
2881 * journalctl gained the new "--header" switch to introspect
2882 header data of journal files.
2884 * A new setting SystemCallFilters= has been added to services
2885 which may be used to apply blacklists or whitelists to
2886 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
2888 * nspawn gained a new --link-journal= switch (and quicker: -j)
2889 to link the container journal with the host. This makes it
2890 very easy to centralize log viewing on the host for all
2891 guests while still keeping the journal files separated.
2893 * Many bugfixes and optimizations
2895 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
2896 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
2897 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
2902 * Several tools now understand kernel command line arguments,
2903 which are only read when run in an initial RAM disk. They
2904 usually follow closely their normal counterparts, but are
2907 * There's a new tool to analyze the readahead files that are
2908 automatically generated at boot. Use:
2910 /usr/lib/systemd/systemd-readahead analyze /.readahead
2912 * We now provide an early debug shell on tty9 if this enabled. Use:
2914 systemctl enable debug-shell.service
2916 * All plymouth related units have been moved into the Plymouth
2917 package. Please make sure to upgrade your Plymouth version
2920 * systemd-tmpfiles now supports getting passed the basename of
2921 a configuration file only, in which case it will look for it
2922 in all appropriate directories automatically.
2924 * udevadm info now takes a /dev or /sys path as argument, and
2925 does the right thing. Example:
2927 udevadm info /dev/sda
2928 udevadm info /sys/class/block/sda
2930 * systemctl now prints a warning if a unit is stopped but a
2931 unit that might trigger it continues to run. Example: a
2932 service is stopped but the socket that activates it is left
2935 * "systemctl status" will now mention if the log output was
2936 shortened due to rotation since a service has been started.
2938 * The journal API now exposes functions to determine the
2939 "cutoff" times due to rotation.
2941 * journald now understands SIGUSR1 and SIGUSR2 for triggering
2942 immediately flushing of runtime logs to /var if possible,
2943 resp. for triggering immediate rotation of the journal
2946 * It is now considered an error if a service is attempted to
2947 be stopped that is not loaded.
2949 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
2951 * systemd-analyze now supports Python 3
2953 * tmpfiles now supports cleaning up directories via aging
2954 where the first level dirs are always kept around but
2955 directories beneath it automatically aged. This is enabled
2956 by prefixing the age field with '~'.
2958 * Seat objects now expose CanGraphical, CanTTY properties
2959 which is required to deal with very fast bootups where the
2960 display manager might be running before the graphics drivers
2961 completed initialization.
2963 * Seat objects now expose a State property.
2965 * We now include RPM macros for service enabling/disabling
2966 based on the preset logic. We recommend RPM based
2967 distributions to make use of these macros if possible. This
2968 makes it simpler to reuse RPM spec files across
2971 * We now make sure that the collected systemd unit name is
2972 always valid when services log to the journal via
2975 * There's a new man page kernel-command-line(7) detailing all
2976 command line options we understand.
2978 * The fstab generator may now be disabled at boot by passing
2979 fstab=0 on the kernel command line.
2981 * A new kernel command line option modules-load= is now understood
2982 to load a specific kernel module statically, early at boot.
2984 * Unit names specified on the systemctl command line are now
2985 automatically escaped as needed. Also, if file system or
2986 device paths are specified they are automatically turned
2987 into the appropriate mount or device unit names. Example:
2989 systemctl status /home
2990 systemctl status /dev/sda
2992 * The SysVConsole= configuration option has been removed from
2993 system.conf parsing.
2995 * The SysV search path is no longer exported on the D-Bus
2998 * The Names= option is been removed from unit file parsing.
3000 * There's a new man page bootup(7) detailing the boot process.
3002 * Every unit and every generator we ship with systemd now
3003 comes with full documentation. The self-explanatory boot is
3006 * A couple of services gained "systemd-" prefixes in their
3007 name if they wrap systemd code, rather than only external
3008 code. Among them fsck@.service which is now
3009 systemd-fsck@.service.
3011 * The HaveWatchdog property has been removed from the D-Bus
3014 * systemd.confirm_spawn= on the kernel command line should now
3017 * There's a new man page crypttab(5) which details all options
3018 we actually understand.
3020 * systemd-nspawn gained a new --capability= switch to pass
3021 additional capabilities to the container.
3023 * timedated will now read known NTP implementation unit names
3024 from /usr/lib/systemd/ntp-units.d/*.list,
3025 systemd-timedated-ntp.target has been removed.
3027 * journalctl gained a new switch "-b" that lists log data of
3028 the current boot only.
3030 * The notify socket is in the abstract namespace again, in
3031 order to support daemons which chroot() at start-up.
3033 * There is a new Storage= configuration option for journald
3034 which allows configuration of where log data should go. This
3035 also provides a way to disable journal logging entirely, so
3036 that data collected is only forwarded to the console, the
3037 kernel log buffer or another syslog implementation.
3039 * Many bugfixes and optimizations
3041 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
3042 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
3043 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
3044 Shawn Landden, Tom Gundersen
3048 * "systemctl help <unit>" now shows the man page if one is
3051 * Several new man pages have been added.
3053 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
3054 MaxLevelConsole= can now be specified in
3055 journald.conf. These options allow reducing the amount of
3056 data stored on disk or forwarded by the log level.
3058 * TimerSlackNSec= can now be specified in system.conf for
3059 PID1. This allows system-wide power savings.
3061 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
3062 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
3067 * logind is now capable of (optionally) handling power and
3068 sleep keys as well as the lid switch.
3070 * journalctl now understands the syntax "journalctl
3071 /usr/bin/avahi-daemon" to get all log output of a specific
3074 * CapabilityBoundingSet= in system.conf now also influences
3075 the capability bound set of usermode helpers of the kernel.
3077 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
3078 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
3079 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
3080 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
3084 * Note that we skipped 139 releases here in order to set the
3085 new version to something that is greater than both udev's
3086 and systemd's most recent version number.
3088 * udev: all udev sources are merged into the systemd source tree now.
3089 All future udev development will happen in the systemd tree. It
3090 is still fully supported to use the udev daemon and tools without
3091 systemd running, like in initramfs or other init systems. Building
3092 udev though, will require the *build* of the systemd tree, but
3093 udev can be properly *run* without systemd.
3095 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
3096 should be used to create dead device nodes as workarounds for broken
3099 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
3100 no longer supported. udev_monitor_new_from_netlink() needs to be
3101 used to subscribe to events.
3103 * udev: when udevd is started by systemd, processes which are left
3104 behind by forking them off of udev rules, are unconditionally cleaned
3105 up and killed now after the event handling has finished. Services or
3106 daemons must be started as systemd services. Services can be
3107 pulled-in by udev to get started, but they can no longer be directly
3108 forked by udev rules.
3110 * udev: the daemon binary is called systemd-udevd now and installed
3111 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
3112 to adapt to that, create symlink, or rename the binary after building
3115 * libudev no longer provides these symbols:
3116 udev_monitor_from_socket()
3117 udev_queue_get_failed_list_entry()
3118 udev_get_{dev,sys,run}_path()
3119 The versions number was bumped and symbol versioning introduced.
3121 * systemd-loginctl and systemd-journalctl have been renamed
3122 to loginctl and journalctl to match systemctl.
3124 * The config files: /etc/systemd/systemd-logind.conf and
3125 /etc/systemd/systemd-journald.conf have been renamed to
3126 logind.conf and journald.conf. Package updates should rename
3127 the files to the new names on upgrade.
3129 * For almost all files the license is now LGPL2.1+, changed
3130 from the previous GPL2.0+. Exceptions are some minor stuff
3131 of udev (which will be changed to LGPL2.1 eventually, too),
3132 and the MIT licensed sd-daemon.[ch] library that is suitable
3133 to be used as drop-in files.
3135 * systemd and logind now handle system sleep states, in
3136 particular suspending and hibernating.
3138 * logind now implements a sleep/shutdown/idle inhibiting logic
3139 suitable for a variety of uses. Soonishly Lennart will blog
3140 about this in more detail.
3142 * var-run.mount and var-lock.mount are no longer provided
3143 (which prevously bind mounted these directories to their new
3144 places). Distributions which have not converted these
3145 directories to symlinks should consider stealing these files
3146 from git history and add them downstream.
3148 * We introduced the Documentation= field for units and added
3149 this to all our shipped units. This is useful to make it
3150 easier to explore the boot and the purpose of the various
3153 * All smaller setup units (such as
3154 systemd-vconsole-setup.service) now detect properly if they
3155 are run in a container and are skipped when
3156 appropriate. This guarantees an entirely noise-free boot in
3157 Linux container environments such as systemd-nspawn.
3159 * A framework for implementing offline system updates is now
3160 integrated, for details see:
3161 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
3163 * A new service type Type=idle is available now which helps us
3164 avoiding ugly interleaving of getty output and boot status
3167 * There's now a system-wide CapabilityBoundingSet= option to
3168 globally reduce the set of capabilities for the
3169 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
3170 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
3171 even CAP_NET_ADMIN system-wide for secure systems.
3173 * There are now system-wide DefaultLimitXXX= options to
3174 globally change the defaults of the various resource limits
3175 for all units started by PID 1.
3177 * Harald Hoyer's systemd test suite has been integrated into
3178 systemd which allows easy testing of systemd builds in qemu
3179 and nspawn. (This is really awesome! Ask us for details!)
3181 * The fstab parser is now implemented as generator, not inside
3184 * systemctl will now warn you if .mount units generated from
3185 /etc/fstab are out of date due to changes in fstab that
3186 have not been read by systemd yet.
3188 * systemd is now suitable for usage in initrds. Dracut has
3189 already been updated to make use of this. With this in place
3190 initrds get a slight bit faster but primarily are much
3191 easier to introspect and debug since "systemctl status" in
3192 the host system can be used to introspect initrd services,
3193 and the journal from the initrd is kept around too.
3195 * systemd-delta has been added, a tool to explore differences
3196 between user/admin configuration and vendor defaults.
3198 * PrivateTmp= now affects both /tmp and /var/tmp.
3200 * Boot time status messages are now much prettier and feature
3201 proper english language. Booting up systemd has never been
3204 * Read-ahead pack files now include the inode number of all
3205 files to pre-cache. When the inode changes the pre-caching
3206 is not attempted. This should be nicer to deal with updated
3207 packages which might result in changes of read-ahead
3210 * We now temporaritly lower the kernel's read_ahead_kb variable
3211 when collecting read-ahead data to ensure the kernel's
3212 built-in read-ahead does not add noise to our measurements
3213 of necessary blocks to pre-cache.
3215 * There's now RequiresMountsFor= to add automatic dependencies
3216 for all mounts necessary for a specific file system path.
3218 * MountAuto= and SwapAuto= have been removed from
3219 system.conf. Mounting file systems at boot has to take place
3222 * nspawn now learned a new switch --uuid= to set the machine
3223 ID on the command line.
3225 * nspawn now learned the -b switch to automatically search
3228 * vt102 is now the default TERM for serial TTYs, upgraded from
3231 * systemd-logind now works on VT-less systems.
3233 * The build tree has been reorganized. The individual
3234 components now have directories of their own.
3236 * A new condition type ConditionPathIsReadWrite= is now available.
3238 * nspawn learned the new -C switch to create cgroups for the
3239 container in other hierarchies.
3241 * We now have support for hardware watchdogs, configurable in
3244 * The scheduled shutdown logic now has a public API.
3246 * We now mount /tmp as tmpfs by default, but this can be
3247 masked and /etc/fstab can override it.
3249 * Since udisks does not make use of /media anymore we are not
3250 mounting a tmpfs on it anymore.
3252 * journalctl gained a new --local switch to only interleave
3253 locally generated journal files.
3255 * We can now load the IMA policy at boot automatically.
3257 * The GTK tools have been split off into a systemd-ui.
3259 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
3260 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
3261 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
3262 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
3263 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
3264 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
3265 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
3266 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
3267 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
3272 * This is mostly a bugfix release
3274 * Support optional initialization of the machine ID from the
3275 KVM or container configured UUID.
3277 * Support immediate reboots with "systemctl reboot -ff"
3279 * Show /etc/os-release data in systemd-analyze output
3281 * Many bugfixes for the journal, including endianness fixes and
3282 ensuring that disk space enforcement works
3284 * sd-login.h is C++ comptaible again
3286 * Extend the /etc/os-release format on request of the Debian
3289 * We now refuse non-UTF8 strings used in various configuration
3290 and unit files. This is done to ensure we do not pass invalid
3291 data over D-Bus or expose it elsewhere.
3293 * Register Mimo USB Screens as suitable for automatic seat
3296 * Read SELinux client context from journal clients in a race
3299 * Reorder configuration file lookup order. /etc now always
3300 overrides /run in order to allow the administrator to always
3301 and unconditionally override vendor supplied or
3302 automatically generated data.
3304 * The various user visible bits of the journal now have man
3305 pages. We still lack man pages for the journal API calls
3308 * We now ship all man pages in HTML format again in the
3311 Contributions from: Dave Reisner, Dirk Eibach, Frederic
3312 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
3313 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
3318 * This is mostly a bugfix release
3320 * systems lacking /etc/os-release are no longer supported.
3322 * Various functionality updates to libsystemd-login.so
3324 * Track class of PAM logins to distuingish greeters from
3327 Contributions from: Kay Sievers, Lennart Poettering, Michael
3332 * This is an important bugfix release for v41.
3334 * Building man pages is now optional which should be useful
3335 for those building systemd from git but unwilling to install
3338 * Watchdog support for supervising services is now usable. In
3339 a future release support for hardware watchdogs
3340 (i.e. /dev/watchdog) will be added building on this.
3342 * Service start rate limiting is now configurable and can be
3343 turned off per service. When a start rate limit is hit a
3344 reboot can automatically be triggered.
3346 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
3348 Contributions from: Benjamin Franzke, Bill Nottingham,
3349 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
3350 Schmidt, Michał Górny, Piotr Drąg
3354 * The systemd binary is installed /usr/lib/systemd/systemd now;
3355 An existing /sbin/init symlink needs to be adapted with the
3358 * The code that loads kernel modules has been ported to invoke
3359 libkmod directly, instead of modprobe. This means we do not
3360 support systems with module-init-tools anymore.
3362 * Watchdog support is now already useful, but still not
3365 * A new kernel command line option systemd.setenv= is
3366 understood to set system wide environment variables
3367 dynamically at boot.
3369 * We now limit the set of capabilities of systemd-journald.
3371 * We now set SIGPIPE to ignore by default, since it only is
3372 useful in shell pipelines, and has little use in general
3373 code. This can be disabled with IgnoreSIPIPE=no in unit
3376 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
3377 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
3382 * This is mostly a bugfix release
3384 * We now expose the reason why a service failed in the
3385 "Result" D-Bus property.
3387 * Rudimentary service watchdog support (will be completed over
3388 the next few releases.)
3390 * When systemd forks off in order execute some service we will
3391 now immediately changes its argv[0] to reflect which process
3392 it will execute. This is useful to minimize the time window
3393 with a generic argv[0], which makes bootcharts more useful
3395 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
3396 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
3397 Mike Kazantsev, Ray Strode
3401 * This is mostly a test release, but incorporates many
3404 * New systemd-cgtop tool to show control groups by their
3407 * Linking against libacl for ACLs is optional again. If
3408 disabled, support tracking device access for active logins
3409 goes becomes unavailable, and so does access to the user
3410 journals by the respective users.
3412 * If a group "adm" exists, journal files are automatically
3413 owned by them, thus allow members of this group full access
3414 to the system journal as well as all user journals.
3416 * The journal now stores the SELinux context of the logging
3417 client for all entries.
3419 * Add C++ inclusion guards to all public headers
3421 * New output mode "cat" in the journal to print only text
3422 messages, without any meta data like date or time.
3424 * Include tiny X server wrapper as a temporary stop-gap to
3425 teach XOrg udev display enumeration. This is used by display
3426 managers such as gdm, and will go away as soon as XOrg
3427 learned native udev hotplugging for display devices.
3429 * Add new systemd-cat tool for executing arbitrary programs
3430 with STDERR/STDOUT connected to the journal. Can also act as
3431 BSD logger replacement, and does so by default.
3433 * Optionally store all locally generated coredumps in the
3434 journal along with meta data.
3436 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
3437 writing short strings to files (for usage for /sys), and for
3438 creating symlinks, character and block device nodes.
3440 * New unit file option ControlGroupPersistent= to make cgroups
3441 persistent, following the mechanisms outlined in
3442 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
3444 * Support multiple local RTCs in a sane way
3446 * No longer monopolize IO when replaying readahead data on
3447 rotating disks, since we might starve non-file-system IO to
3448 death, since fanotify() will not see accesses done by blkid,
3451 * Do not show kernel threads in systemd-cgls anymore, unless
3452 requested with new -k switch.
3454 Contributions from: Dan Horák, Kay Sievers, Lennart
3455 Poettering, Michal Schmidt
3459 * This is mostly a test release, but incorporates many
3462 * The git repository moved to:
3463 git://anongit.freedesktop.org/systemd/systemd
3464 ssh://git.freedesktop.org/git/systemd/systemd
3466 * First release with the journal
3467 http://0pointer.de/blog/projects/the-journal.html
3469 * The journal replaces both systemd-kmsg-syslogd and
3470 systemd-stdout-bridge.
3472 * New sd_pid_get_unit() API call in libsystemd-logind
3474 * Many systemadm clean-ups
3476 * Introduce remote-fs-pre.target which is ordered before all
3477 remote mounts and may be used to start services before all
3480 * Added Mageia support
3482 * Add bash completion for systemd-loginctl
3484 * Actively monitor PID file creation for daemons which exit in
3485 the parent process before having finished writing the PID
3486 file in the daemon process. Daemons which do this need to be
3487 fixed (i.e. PID file creation must have finished before the
3488 parent exits), but we now react a bit more gracefully to them.
3490 * Add colourful boot output, mimicking the well-known output
3491 of existing distributions.
3493 * New option PassCredentials= for socket units, for
3494 compatibility with a recent kernel ABI breakage.
3496 * /etc/rc.local is now hooked in via a generator binary, and
3497 thus will no longer act as synchronization point during
3500 * systemctl list-unit-files now supports --root=.
3502 * systemd-tmpfiles now understands two new commands: z, Z for
3503 relabelling files according to the SELinux database. This is
3504 useful to apply SELinux labels to specific files in /sys,
3507 * Output of SysV services is now forwarded to both the console
3508 and the journal by default, not only just the console.
3510 * New man pages for all APIs from libsystemd-login.
3512 * The build tree got reorganized and a the build system is a
3513 lot more modular allowing embedded setups to specifically
3514 select the components of systemd they are interested in.
3516 * Support for Linux systems lacking the kernel VT subsystem is
3519 * configure's --with-rootdir= got renamed to
3520 --with-rootprefix= to follow the naming used by udev and
3523 * Unless specified otherwise we will now install to /usr instead
3524 of /usr/local by default.
3526 * Processes with '@' in argv[0][0] are now excluded from the
3527 final shut-down killing spree, following the logic explained
3529 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
3531 * All processes remaining in a service cgroup when we enter
3532 the START or START_PRE states are now killed with
3533 SIGKILL. That means it is no longer possible to spawn
3534 background processes from ExecStart= lines (which was never
3535 supported anyway, and bad style).
3537 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
3538 reloading of units together.
3540 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
3541 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
3542 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
3543 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
3544 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek