1 systemd System and Service Manager
5 * systemd now depends on util-linux v2.27. More specifically,
6 the newly added mount monitor feature in libmount now
7 replaces systemd's former own implementation.
9 * libmount mandates /etc/mtab not to be regular file, and
10 systemd now enforces this condition at early boot.
11 /etc/mtab has been deprecated and warned about for a very
12 long time, so systems running systemd should already have
13 stopped having this file around as anything else than a
14 symlink to /proc/self/mounts.
16 * Support for the "pids" cgroup controller has been added.
17 It allows accounting the number of tasks in a cgroup and
18 enforcing limits on it. This adds two new setting
19 TasksAccounting= and TasksMax= to each unit, as well as a
20 gloabl option DefaultTasksAccounting=.
22 * Support for the "net_cls" cgroup controller has been added.
23 It allows assigning a net class ID to each task in the cgroup,
24 which can then be used in firewall rules and traffic shaping
25 configurations. Note that the kernel netfilter code does not
26 currently work reliably for ingress packets on unestablished
29 This adds a new config directive called NetClass= to CGroup
30 enabled units. Allowed values are positive numbers for fix
31 assignments and "auto" for picking a free value
34 * 'systemctl is-system-running' now returns 'offline' if the
35 system is not booted with systemd. This command can now be
36 used as a substitute for 'systemd-notify --booted'.
38 * Watchdog timeouts have been increased to 3 minutes for all
39 in-tree service files. Apparently, disk IO issues are more
40 frequent than we hoped, and user reported >1 minute waiting
43 * 'machine-id-commit' functionality has been merged into
44 'machine-id-setup --commit'. The separate binary has been
47 * The WorkingDirectory= directive in unit files may now be
48 set to the special value '~'. In this case, the working
49 directory is set to the home directory of the user configured
52 * A new systemd.crash_reboot=1 kernel command line option has
53 been added that triggers a reboot after crashing. This can
54 also be set through CrashReboot=no in systemd.conf.
56 * The CrashChVT= configuration file setting is renamed to
57 CrashChangeVT=, following our usual logic of not abbreviating
58 unnecessarily. The old directive is still suppored for compat
59 reasons. Also, this directive now takes an interger value
60 between 1 and 63, or a boolean value. The formerly supported
61 '-1' value for disabling stays around for compat reasons.
63 * The PrivateTmp, PrivateDevices, PrivateNetwork,
64 NoNewPrivileges, TTYPath, WorkingDirectory and RootDirectory
65 properties can now be set for transient units.
67 * Galician, Turkish and Korean translations were added.
75 * The DHCP implementation of systemd-networkd gained a set of
78 - The DHCP server now supports emitting DNS and NTP
79 information. It may be enabled and configured via
80 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
81 and NTP information is enabled, but no servers are
82 configured, the corresponding uplink information (if there
83 is any) is propagated.
85 - Server and client now support transmission and reception
86 of timezone information. It can be configured via the
87 newly introduced network options UseTimezone=,
88 EmitTimezone=, and Timezone=. Transmission of timezone
89 information is enabled between host and containers by
90 default now: the container will change its local timezone
91 to what the host has set.
93 - Lease timeouts can now be configured via
94 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
96 - The DHCP server improved on the stability of
97 leases. Clients are more likely to get the same lease
98 information back, even if the server loses state.
100 - The DHCP server supports two new configuration options to
101 control the lease address pool metrics, PoolOffset= and
104 * The encapsulation limit of tunnels in systemd-networkd may
105 now be configured via 'EncapsulationLimit='. It allows
106 modifying the maximum additional levels of encapsulation
107 that are permitted to be prepended to a packet.
109 * systemd now supports the concept of user buses replacing
110 session buses, if used with dbus-1.10 (and enabled via dbus
111 --enable-user-session). It previously only supported this on
112 kdbus-enabled systems, and this release expands this to
113 'dbus-daemon' systems.
115 * systemd-networkd now supports predictable interface names
118 * systemd now optionally supports the new Linux kernel
119 "unified" control group hierarchy. If enabled via the kernel
120 command-line option 'systemd.unified_cgroup_hierarchy=1',
121 systemd will try to mount the unified cgroup hierarchy
122 directly on /sys/fs/cgroup. If not enabled, or not
123 available, systemd will fall back to the legacy cgroup
124 hierarchy setup, as before. Host system and containers can
125 mix and match legacy and unified hierarchies as they
126 wish. nspawn understands the $UNIFIED_CROUP_HIERARCHY
127 environment variable to individually select the hierarchy to
128 use for executed containers. By default, nspawn will use the
129 unified hierarchy for the containers if the host uses the
130 unified hierarchy, and the legacy hierarchy otherwise.
131 Please note that at this point the unified hierarchy is an
132 experimental kernel feature and is likely to change in one
133 of the next kernel releases. Therefore, it should not be
134 enabled by default in downstream distributions yet. The
135 minimum required kernel version for the unified hierarchy to
136 work is 4.2. Note that when the unified hierarchy is used
137 for the first time delegated access to controllers is
138 safe. Because of this systemd-nspawn containers will get
139 access to controllers now, as will systemd user
140 sessions. This means containers and user sessions may now
141 manage their own resources, partitioning up what the system
144 * A new special scope unit "init.scope" has been introduced
145 that encapsulates PID 1 of the system. It may be used to
146 determine resource usage and enforce resource limits on PID
147 1 itself. PID 1 hence moved out of the root of the control
150 * The cgtop tool gained support for filtering out kernel
151 threads when counting tasks in a control group. Also, the
152 count of processes is now recursively summed up by
153 default. Two options -k and --recursive= have been added to
154 revert to old behaviour. The tool has also been updated to
155 work correctly in containers now.
157 * systemd-nspawn's --bind= and --bind-ro= options have been
158 extended to allow creation of non-recursive bind mounts.
160 * libsystemd gained two new calls sd_pid_get_cgroup() and
161 sd_peer_get_cgroup() which return the control group path of
162 a process or peer of a connected AF_UNIX socket. This
163 function call is particularly useful when implementing
164 delegated subtrees support in the control group hierarchy.
166 * The "sd-event" event loop API of libsystemd now supports
167 correct dequeuing of real-time signals, without losing
170 * When systemd requests a PolicyKit decision when managing
171 units it will now add additional fields to the request,
172 including unit name and desired operation. This enables more
173 powerful PolicyKit policies, that make decisions depending
176 * nspawn learnt support for .nspawn settings files, that may
177 accompany the image files or directories of containers, and
178 may contain additional settings for the container. This is
179 an alternative to configuring container parameters via the
182 Contributions from: Cristian Rodríguez, Daniel Mack, David
183 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
184 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
185 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
186 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
187 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
188 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
189 Andersen, Tom Gundersen, Torstein Husebø
191 -- Berlin, 2015-09-08
195 * machinectl gained a new verb 'shell' which opens a fresh
196 shell on the target container or the host. It is similar to
197 the existing 'login' command of machinectl, but spawns the
198 shell directly without prompting for username or
199 password. The pseudo machine '.host' now refers to the local
200 host and is used by default. Hence, 'machinectl shell' can
201 be used as replacement for 'su -' which spawns a session as
202 a fresh systemd unit in a way that is fully isolated from
203 the originating session.
205 * systemd-networkd learned to cope with private-zone DHCP
206 options and allows other programs to query the values.
208 * SELinux access control when enabling/disabling units is no
209 longer enforced with this release. The previous
210 implementation was incorrect, and a new corrected
211 implementation is not yet available. As unit file operations
212 are still protected via PolicyKit and D-Bus policy this is
213 not a security problem. Yet, distributions which care about
214 optimal SELinux support should probably not stabilize on
217 * sd-bus gained support for matches of type "arg0has=", that
218 test for membership of strings in string arrays sent in bus
221 * systemd-resolved now dumps the contents of its DNS and LLMNR
222 caches to the logs on reception of the SIGUSR1 signal. This
223 is useful to debug DNS behaviour.
225 * The coredumpctl tool gained a new --directory= option to
226 operate on journal files in a specific directory.
228 * "systemctl reboot" and related commands gained a new
229 "--message=" option which may be used to set a free-text
230 wall message when shutting down or rebooting the
231 system. This message is also logged, which is useful for
232 figuring out the reason for a reboot or shutdown a
235 * The "systemd-resolve-host" tool's -i switch now takes
236 network interface numbers as alternative to interface names.
238 * A new unit file setting for services has been introduced:
239 UtmpMode= allows configuration of how precisely systemd
240 handles utmp and wtmp entries for the service if this is
241 enabled. This allows writing services that appear similar to
242 user sessions in the output of the "w", "who", "last" and
245 * systemd-resolved will now locally synthesize DNS resource
246 records for the "localhost" and "gateway" domains as well as
247 the local hostname. This should ensure that clients querying
248 RRs via resolved will get similar results as those going via
249 NSS, if nss-myhostname is enabled.
251 Contributions from: Alastair Hughes, Alex Crawford, Daniel
252 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
253 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
254 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
255 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
256 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
257 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
258 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
259 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
260 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
261 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
262 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
264 -- Berlin, 2015-08-27
268 * The systemd-efi-boot-generator functionality was merged into
269 systemd-gpt-auto-generator.
271 * systemd-networkd now supports Group Policy for vxlan
272 devices. It can be enabled via the new boolean configuration
273 option called 'GroupPolicyExtension='.
275 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
276 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
277 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
279 -- Berlin, 2015-07-31
283 * The python-systemd code has been removed from the systemd repository.
284 A new repository has been created which accommodates the code from
285 now on, and we kindly ask distributions to create a separate package
286 for this: https://github.com/systemd/python-systemd
288 * The systemd daemon will now reload its main configuration
289 (/etc/systemd/system.conf) on daemon-reload.
291 * sd-dhcp now exposes vendor specific extensions via
292 sd_dhcp_lease_get_vendor_specific().
294 * systemd-networkd gained a number of new configuration options.
296 - A new boolean configuration option for TAP devices called
297 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
298 device, thus allowing to send and receive GSO packets.
300 - A new tunnel configuration option called 'CopyDSCP='.
301 If enabled, the DSCP field of ip6 tunnels is copied into the
304 - A set of boolean bridge configuration options were added.
305 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
306 and 'UnicastFlood=' are now parsed by networkd and applied to the
307 respective bridge link device via the respective IFLA_BRPORT_*
310 - A new string configuration option to override the hostname sent
311 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
312 is true, networkd will use the configured hostname instead of the
313 system hostname when sending DHCP requests.
315 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
316 networkd will configure the IPv6 flow-label of the tunnel device
317 according to RFC2460.
319 - The 'macvtap' virtual network devices are now supported, similar to
320 the already supported 'macvlan' devices.
322 * systemd-resolved now implements RFC5452 to improve resilience against
323 cache poisoning. Additionally, source port randomization is enabled
324 by default to further protect against DNS spoofing attacks.
326 * nss-mymachines now supports translating UIDs and GIDs of running
327 containers with user-namespaces enabled. If a container 'foo'
328 translates a host uid 'UID' to the container uid 'TUID', then
329 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
330 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
331 mapped as 'vg-foo-TGID'.
333 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
334 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
335 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
336 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
337 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
338 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
339 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
340 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
341 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
342 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
344 -- Berlin, 2015-07-29
348 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
349 There are no known issues with current sysfs, and udev does not need
350 or should be used to work around such bugs.
352 * udev does no longer enable USB HID power management. Several reports
353 indicate, that some devices cannot handle that setting.
355 * The udev accelerometer helper was removed. The functionality
356 is now fully included in iio-sensor-proxy. But this means,
357 older iio-sensor-proxy versions will no longer provide
358 accelerometer/orientation data with this systemd version.
359 Please upgrade iio-sensor-proxy to version 1.0.
361 * networkd gained a new configuration option IPv6PrivacyExtensions=
362 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
363 for Stateless Address") on selected networks.
365 * For the sake of fewer build-time dependencies and less code in the
366 main repository, the python bindings are about to be removed in the
367 next release. A new repository has been created which accommodates
368 the code from now on, and we kindly ask distributions to create a
369 separate package for this. The removal will take place in v223.
371 https://github.com/systemd/python-systemd
373 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
374 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
375 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
376 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
377 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
378 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
379 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
380 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
381 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
382 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
384 -- Berlin, 2015-07-07
388 * The sd-bus.h and sd-event.h APIs have now been declared
389 stable and have been added to the official interface of
390 libsystemd.so. sd-bus implements an alternative D-Bus client
391 library, that is relatively easy to use, very efficient and
392 supports both classic D-Bus as well as kdbus as transport
393 backend. sd-event is a generic event loop abstraction that
394 is built around Linux epoll, but adds features such as event
395 prioritization or efficient timer handling. Both APIs are good
396 choices for C programs looking for a bus and/or event loop
397 implementation that is minimal and does not have to be
398 portable to other kernels.
400 * kdbus support is no longer compile-time optional. It is now
401 always built-in. However, it can still be disabled at
402 runtime using the kdbus=0 kernel command line setting, and
403 that setting may be changed to default to off, by specifying
404 --disable-kdbus at build-time. Note though that the kernel
405 command line setting has no effect if the kdbus.ko kernel
406 module is not installed, in which case kdbus is (obviously)
407 also disabled. We encourage all downstream distributions to
408 begin testing kdbus by adding it to the kernel images in the
409 development distributions, and leaving kdbus support in
412 * The minimal required util-linux version has been bumped to
415 * Support for chkconfig (--enable-chkconfig) was removed in
416 favor of calling an abstraction tool
417 /lib/systemd/systemd-sysv-install. This needs to be
418 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
419 in README for details.
421 * If there's a systemd unit and a SysV init script for the
422 same service name, and the user executes "systemctl enable"
423 for it (or a related call), then this will now enable both
424 (or execute the related operation on both), not just the
427 * The libudev API documentation has been converted from gtkdoc
430 * gudev has been removed from the systemd tree, it is now an
433 * The systemd-cgtop tool learnt a new --raw switch to generate
434 "raw" (machine parsable) output.
436 * networkd's IPForwarding= .network file setting learnt the
437 new setting "kernel", which ensures that networkd does not
438 change the IP forwarding sysctl from the default kernel
441 * The systemd-logind bus API now exposes a new boolean
442 property "Docked" that reports whether logind considers the
443 system "docked", i.e. connected to a docking station or not.
445 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
446 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
447 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
448 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
449 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
450 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
451 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
452 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
453 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
454 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
455 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
456 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
457 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
458 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
459 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
460 Fink, Zbigniew Jędrzejewski-Szmek
462 -- Berlin, 2015-06-19
466 * The gudev library has been extracted into a separate repository
467 available at: https://git.gnome.org/browse/libgudev/
468 It is now managed as part of the Gnome project. Distributions
469 are recommended to pass --disable-gudev to systemd and use
470 gudev from the Gnome project instead. gudev is still included
471 in systemd, for now. It will be removed soon, though. Please
472 also see the announcement-thread on systemd-devel:
473 http://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
475 * systemd now exposes a CPUUsageNSec= property for each
476 service unit on the bus, that contains the overall consumed
477 CPU time of a service (the sum of what each process of the
478 service consumed). This value is only available if
479 CPUAccounting= is turned on for a service, and is then shown
480 in the "systemctl status" output.
482 * Support for configuring alternative mappings of the old SysV
483 runlevels to systemd targets has been removed. They are now
484 hardcoded in a way that runlevels 2, 3, 4 all map to
485 multi-user.target and 5 to graphical.target (which
486 previously was already the default behaviour).
488 * The auto-mounter logic gained support for mount point
489 expiry, using a new TimeoutIdleSec= setting in .automount
490 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
492 * The EFI System Partition (ESP) as mounted to /boot by
493 systemd-efi-boot-generator will now be unmounted
494 automatically after 2 minutes of not being used. This should
495 minimize the risk of ESP corruptions.
497 * New /etc/fstab options x-systemd.requires= and
498 x-systemd.requires-mounts-for= are now supported to express
499 additional dependencies for mounts. This is useful for
500 journalling file systems that support external journal
501 devices or overlay file systems that require underlying file
502 systems to be mounted.
504 * systemd does not support direct live-upgrades (via systemctl
505 daemon-reexec) from versions older than v44 anymore. As no
506 distribution we are aware of shipped such old versions in a
507 stable release this should not be problematic.
509 * When systemd forks off a new per-connection service instance
510 it will now set the $REMOTE_ADDR environment variable to the
511 remote IP address, and $REMOTE_PORT environment variable to
512 the remote IP port. This behaviour is similar to the
513 corresponding environment variables defined by CGI.
515 * systemd-networkd gained support for uplink failure
516 detection. The BindCarrier= option allows binding interface
517 configuration dynamically to the link sense of other
518 interfaces. This is useful to achieve behaviour like in
521 * systemd-networkd gained support for configuring the DHCP
522 client identifier to use when requesting leases.
524 * systemd-networkd now has a per-network UseNTP= option to
525 configure whether NTP server information acquired via DHCP
526 is passed on to services like systemd-timesyncd.
528 * systemd-networkd gained support for vti6 tunnels.
530 * Note that systemd-networkd manages the sysctl variable
531 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
532 it is configured for since v219. The variable controls IP
533 forwarding, and is a per-interface alternative to the global
534 /proc/sys/net/ipv[46]/ip_forward. This setting is
535 configurable in the IPForward= option, which defaults to
536 "no". This means if networkd is used for an interface it is
537 no longer sufficient to set the global sysctl option to turn
538 on IP forwarding! Instead, the .network file option
539 IPForward= needs to be turned on! Note that the
540 implementation of this behaviour was broken in v219 and has
543 * Many bonding and vxlan options are now configurable in
546 * systemd-nspawn gained a new --property= setting to set unit
547 properties for the container scope. This is useful for
548 setting resource parameters (e.g "CPUShares=500") on
549 containers started from the command line.
551 * systemd-nspawn gained a new --private-users= switch to make
552 use of user namespacing available on recent Linux kernels.
554 * systemd-nspawn may now be called as part of a shell pipeline
555 in which case the pipes used for stdin and stdout are passed
556 directly to the process invoked in the container, without
557 indirection via a pseudo tty.
559 * systemd-nspawn gained a new switch to control the UNIX
560 signal to use when killing the init process of the container
563 * systemd-nspawn gained a new --overlay= switch for mounting
564 overlay file systems into the container using the new kernel
567 * When a container image is imported via systemd-importd and
568 the host file system is not btrfs, a loopback block device
569 file is created in /var/lib/machines.raw with a btrfs file
570 system inside. It is then mounted to /var/lib/machines to
571 enable btrfs features for container management. The loopback
572 file and btrfs file system is grown as needed when container
573 images are imported via systemd-importd.
575 * systemd-machined/systemd-importd gained support for btrfs
576 quota, to enforce container disk space limits on disk. This
577 is exposed in "machinectl set-limit".
579 * systemd-importd now can import containers from local .tar,
580 .raw and .qcow2 images, and export them to .tar and .raw. It
581 can also import dkr v2 images now from the network (on top
584 * systemd-importd gained support for verifying downloaded
585 images with gpg2 (previously only gpg1 was supported).
587 * systemd-machined, systemd-logind, systemd: most bus calls
588 are now accessible to unprivileged processes via
589 PolicyKit. Also, systemd-logind will now allow users to kill
590 their own sessions without further privileges or
593 * systemd-shutdownd has been removed. This service was
594 previously responsible for implementing scheduled shutdowns
595 as exposed in /usr/bin/shutdown's time parameter. This
596 functionality has now been moved into systemd-logind and is
597 accessible via a bus interface.
599 * "systemctl reboot" gained a new switch --firmware-setup that
600 can be used to reboot into the EFI firmware setup, if that
601 is available. systemd-logind now exposes an API on the bus
602 to trigger such reboots, in case graphical desktop UIs want
603 to cover this functionality.
605 * "systemctl enable", "systemctl disable" and "systemctl mask"
606 now support a new "--now" switch. If specified the units
607 that are enabled will also be started, and the ones
608 disabled/masked also stopped.
610 * The Gummiboot EFI boot loader tool has been merged into
611 systemd, and renamed to "systemd-boot". The bootctl tool has been
612 updated to support systemd-boot.
614 * An EFI kernel stub has been added that may be used to create
615 kernel EFI binaries that contain not only the actual kernel,
616 but also an initrd, boot splash, command line and OS release
617 information. This combined binary can then be signed as a
618 single image, so that the firmware can verify it all in one
619 step. systemd-boot has special support for EFI binaries created
620 like this and can extract OS release information from them
621 and show them in the boot menu. This functionality is useful
622 to implement cryptographically verified boot schemes.
624 * Optional support has been added to systemd-fsck to pass
625 fsck's progress report to an AF_UNIX socket in the file
628 * udev will no longer create device symlinks for all block
629 devices by default. A blacklist for excluding special block
630 devices from this logic has been turned into a whitelist
631 that requires picking block devices explicitly that require
634 * A new (currently still internal) API sd-device.h has been
635 added to libsystemd. This modernized API is supposed to
636 replace libudev eventually. In fact, already much of libudev
637 is now just a wrapper around sd-device.h.
639 * A new hwdb database for storing metadata about pointing
640 stick devices has been added.
642 * systemd-tmpfiles gained support for setting file attributes
643 similar to the "chattr" tool with new 'h' and 'H' lines.
645 * systemd-journald will no longer unconditionally set the
646 btrfs NOCOW flag on new journal files. This is instead done
647 with tmpfiles snippet using the new 'h' line type. This
648 allows easy disabling of this logic, by masking the
649 journal-nocow.conf tmpfiles file.
651 * systemd-journald will now translate audit message types to
652 human readable identifiers when writing them to the
653 journal. This should improve readability of audit messages.
655 * The LUKS logic gained support for the offset= and skip=
656 options in /etc/crypttab, as previously implemented by
659 * /usr/lib/os-release gained a new optional field VARIANT= for
660 distributions that support multiple variants (such as a
661 desktop edition, a server edition, ...)
663 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
664 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
665 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
666 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
667 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
668 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
669 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
670 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
671 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
672 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
673 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
674 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
675 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
676 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
677 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
678 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
679 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
680 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
681 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
682 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
683 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
684 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
685 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
686 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
687 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
688 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
689 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
691 -- Berlin, 2015-05-22
695 * Introduce a new API "sd-hwdb.h" for querying the hardware
696 metadata database. With this minimal interface one can query
697 and enumerate the udev hwdb, decoupled from the old libudev
698 library. libudev's interface for this is now only a wrapper
699 around sd-hwdb. A new tool systemd-hwdb has been added to
700 interface with and update the database.
702 * When any of systemd's tools copies files (for example due to
703 tmpfiles' C lines) a btrfs reflink will attempted first,
704 before bytewise copying is done.
706 * systemd-nspawn gained a new --ephemeral switch. When
707 specified a btrfs snapshot is taken of the container's root
708 directory, and immediately removed when the container
709 terminates again. Thus, a container can be started whose
710 changes never alter the container's root directory, and are
711 lost on container termination. This switch can also be used
712 for starting a container off the root file system of the
713 host without affecting the host OS. This switch is only
714 available on btrfs file systems.
716 * systemd-nspawn gained a new --template= switch. It takes the
717 path to a container tree to use as template for the tree
718 specified via --directory=, should that directory be
719 missing. This allows instantiating containers dynamically,
720 on first run. This switch is only available on btrfs file
723 * When a .mount unit refers to a mount point on which multiple
724 mounts are stacked, and the .mount unit is stopped all of
725 the stacked mount points will now be unmounted until no
728 * systemd now has an explicit notion of supported and
729 unsupported unit types. Jobs enqueued for unsupported unit
730 types will now fail with an "unsupported" error code. More
731 specifically .swap, .automount and .device units are not
732 supported in containers, .busname units are not supported on
733 non-kdbus systems. .swap and .automount are also not
734 supported if their respective kernel compile time options
737 * machinectl gained support for two new "copy-from" and
738 "copy-to" commands for copying files from a running
739 container to the host or vice versa.
741 * machinectl gained support for a new "bind" command to bind
742 mount host directories into local containers. This is
743 currently only supported for nspawn containers.
745 * networkd gained support for configuring bridge forwarding
746 database entries (fdb) from .network files.
748 * A new tiny daemon "systemd-importd" has been added that can
749 download container images in tar, raw, qcow2 or dkr formats,
750 and make them available locally in /var/lib/machines, so
751 that they can run as nspawn containers. The daemon can GPG
752 verify the downloads (not supported for dkr, since it has no
753 provisions for verifying downloads). It will transparently
754 decompress bz2, xz, gzip compressed downloads if necessary,
755 and restore sparse files on disk. The daemon uses privilege
756 separation to ensure the actual download logic runs with
757 fewer privileges than the daemon itself. machinectl has
758 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
759 make the functionality of importd available to the
760 user. With this in place the Fedora and Ubuntu "Cloud"
761 images can be downloaded and booted as containers unmodified
762 (the Fedora images lack the appropriate GPG signature files
763 currently, so they cannot be verified, but this will change
764 soon, hopefully). Note that downloading images is currently
765 only fully supported on btrfs.
767 * machinectl is now able to list container images found in
768 /var/lib/machines, along with some metadata about sizes of
769 disk and similar. If the directory is located on btrfs and
770 quota is enabled, this includes quota display. A new command
771 "image-status" has been added that shows additional
772 information about images.
774 * machinectl is now able to clone container images
775 efficiently, if the underlying file system (btrfs) supports
776 it, with the new "machinectl list-images" command. It also
777 gained commands for renaming and removing images, as well as
778 marking them read-only or read-write (supported also on
779 legacy file systems).
781 * networkd gained support for collecting LLDP network
782 announcements, from hardware that supports this. This is
783 shown in networkctl output.
785 * systemd-run gained support for a new -t (--pty) switch for
786 invoking a binary on a pty whose input and output is
787 connected to the invoking terminal. This allows executing
788 processes as system services while interactively
789 communicating with them via the terminal. Most interestingly
790 this is supported across container boundaries. Invoking
791 "systemd-run -t /bin/bash" is an alternative to running a
792 full login session, the difference being that the former
793 will not register a session, nor go through the PAM session
796 * tmpfiles gained support for a new "v" line type for creating
797 btrfs subvolumes. If the underlying file system is a legacy
798 file system, this automatically degrades to creating a
799 normal directory. Among others /var/lib/machines is now
800 created like this at boot, should it be missing.
802 * The directory /var/lib/containers/ has been deprecated and
803 been replaced by /var/lib/machines. The term "machines" has
804 been used in the systemd context as generic term for both
805 VMs and containers, and hence appears more appropriate for
806 this, as the directory can also contain raw images bootable
809 * systemd-nspawn when invoked with -M but without --directory=
810 or --image= is now capable of searching for the container
811 root directory, subvolume or disk image automatically, in
812 /var/lib/machines. systemd-nspawn@.service has been updated
813 to make use of this, thus allowing it to be used for raw
816 * A new machines.target unit has been introduced that is
817 supposed to group all containers/VMs invoked as services on
818 the system. systemd-nspawn@.service has been updated to
821 * machinectl gained a new "start" command, for invoking a
822 container as a service. "machinectl start foo" is mostly
823 equivalent to "systemctl start systemd-nspawn@foo.service",
824 but handles escaping in a nicer way.
826 * systemd-nspawn will now mount most of the cgroupfs tree
827 read-only into each container, with the exception of the
828 container's own subtree in the name=systemd hierarchy.
830 * journald now sets the special FS_NOCOW file flag for its
831 journal files. This should improve performance on btrfs, by
832 avoiding heavy fragmentation when journald's write-pattern
833 is used on COW file systems. It degrades btrfs' data
834 integrity guarantees for the files to the same levels as for
835 ext3/ext4 however. This should be OK though as journald does
836 its own data integrity checks and all its objects are
837 checksummed on disk. Also, journald should handle btrfs disk
838 full events a lot more gracefully now, by processing SIGBUS
839 errors, and not relying on fallocate() anymore.
841 * When journald detects that journal files it is writing to
842 have been deleted it will immediately start new journal
845 * systemd now provides a way to store file descriptors
846 per-service in PID 1.This is useful for daemons to ensure
847 that fds they require are not lost during a daemon
848 restart. The fds are passed to the daemon on the next
849 invocation in the same way socket activation fds are
850 passed. This is now used by journald to ensure that the
851 various sockets connected to all the system's stdout/stderr
852 are not lost when journald is restarted. File descriptors
853 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
854 an extension to sd_notify(). Note that a limit is enforced
855 on the number of fds a service can store in PID 1, and it
856 defaults to 0, so that no fds may be stored, unless this is
857 explicitly turned on.
859 * The default TERM variable to use for units connected to a
860 terminal, when no other value is explicitly is set is now
861 vt220 rather than vt102. This should be fairly safe still,
862 but allows PgUp/PgDn work.
864 * The /etc/crypttab option header= as known from Debian is now
867 * "loginctl user-status" and "loginctl session-status" will
868 now show the last 10 lines of log messages of the
869 user/session following the status output. Similar,
870 "machinectl status" will show the last 10 log lines
871 associated with a virtual machine or container
872 service. (Note that this is usually not the log messages
873 done in the VM/container itself, but simply what the
874 container manager logs. For nspawn this includes all console
877 * "loginctl session-status" without further argument will now
878 show the status of the session of the caller. Similar,
879 "lock-session", "unlock-session", "activate",
880 "enable-linger", "disable-linger" may now be called without
881 session/user parameter in which case they apply to the
882 caller's session/user.
884 * An X11 session scriptlet is now shipped that uploads
885 $DISPLAY and $XAUTHORITY into the environment of the systemd
886 --user daemon if a session begins. This should improve
887 compatibility with X11 enabled applications run as systemd
890 * Generators are now subject to masking via /etc and /run, the
891 same way as unit files.
893 * networkd .network files gained support for configuring
894 per-link IPv4/IPv6 packet forwarding as well as IPv4
895 masquerading. This is by default turned on for veth links to
896 containers, as registered by systemd-nspawn. This means that
897 nspawn containers run with --network-veth will now get
898 automatic routed access to the host's networks without any
899 further configuration or setup, as long as networkd runs on
902 * systemd-nspawn gained the --port= (-p) switch to expose TCP
903 or UDP posts of a container on the host. With this in place
904 it is possible to run containers with private veth links
905 (--network-veth), and have their functionality exposed on
906 the host as if their services were running directly on the
909 * systemd-nspawn's --network-veth switch now gained a short
910 version "-n", since with the changes above it is now truly
911 useful out-of-the-box. The systemd-nspawn@.service has been
912 updated to make use of it too by default.
914 * systemd-nspawn will now maintain a per-image R/W lock, to
915 ensure that the same image is not started more than once
916 writable. (It's OK to run an image multiple times
917 simultaneously in read-only mode.)
919 * systemd-nspawn's --image= option is now capable of
920 dissecting and booting MBR and GPT disk images that contain
921 only a single active Linux partition. Previously it
922 supported only GPT disk images with proper GPT type
923 IDs. This allows running cloud images from major
924 distributions directly with systemd-nspawn, without
927 * In addition to collecting mouse dpi data in the udev
928 hardware database, there's now support for collecting angle
929 information for mouse scroll wheels. The database is
930 supposed to guarantee similar scrolling behavior on mice
931 that it knows about. There's also support for collecting
932 information about Touchpad types.
934 * udev's input_id built-in will now also collect touch screen
935 dimension data and attach it to probed devices.
937 * /etc/os-release gained support for a Distribution Privacy
940 * networkd gained support for creating "ipvlan", "gretap",
941 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
943 * systemd-tmpfiles gained support for "a" lines for setting
946 * systemd-nspawn will now mount /tmp in the container to
947 tmpfs, automatically.
949 * systemd now exposes the memory.usage_in_bytes cgroup
950 attribute and shows it for each service in the "systemctl
951 status" output, if available.
953 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
954 immediate reboot is triggered. This useful if shutdown is
955 hung and is unable to complete, to expedite the
956 operation. Note that this kind of reboot will still unmount
957 all file systems, and hence should not result in fsck being
960 * A .device unit for an optical block device will now be
961 considered active only when a medium is in the drive. Also,
962 mount units are now bound to their backing devices thus
963 triggering automatic unmounting when devices become
964 unavailable. With this in place systemd will now
965 automatically unmount left-over mounts when a CD-ROM is
966 ejected or an USB stick is yanked from the system.
968 * networkd-wait-online now has support for waiting for
969 specific interfaces only (with globbing), and for giving up
970 after a configurable timeout.
972 * networkd now exits when idle. It will be automatically
973 restarted as soon as interfaces show up, are removed or
974 change state. networkd will stay around as long as there is
975 at least one DHCP state machine or similar around, that keep
978 * networkd may now configure IPv6 link-local addressing in
979 addition to IPv4 link-local addressing.
981 * The IPv6 "token" for use in SLAAC may now be configured for
982 each .network interface in networkd.
984 * Routes configured with networkd may now be assigned a scope
987 * networkd's [Match] sections now support globbing and lists
988 of multiple space-separated matches per item.
990 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
991 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
992 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
993 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
994 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
995 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
996 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
997 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
998 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
999 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
1000 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
1001 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
1002 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
1003 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
1004 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
1005 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
1006 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
1007 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
1008 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
1009 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
1010 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
1011 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
1012 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
1013 Hoffmann, Zbigniew Jędrzejewski-Szmek
1015 -- Berlin, 2015-02-16
1019 * When querying unit file enablement status (for example via
1020 "systemctl is-enabled"), a new state "indirect" is now known
1021 which indicates that a unit might not be enabled itself, but
1022 another unit listed in its Also= setting might be.
1024 * Similar to the various existing ConditionXYZ= settings for
1025 units there are now matching AssertXYZ= settings. While
1026 failing conditions cause a unit to be skipped, but its job
1027 to succeed, failing assertions declared like this will cause
1028 a unit start operation and its job to fail.
1030 * hostnamed now knows a new chassis type "embedded".
1032 * systemctl gained a new "edit" command. When used on a unit
1033 file this allows extending unit files with .d/ drop-in
1034 configuration snippets or editing the full file (after
1035 copying it from /usr/lib to /etc). This will invoke the
1036 user's editor (as configured with $EDITOR), and reload the
1037 modified configuration after editing.
1039 * "systemctl status" now shows the suggested enablement state
1040 for a unit, as declared in the (usually vendor-supplied)
1041 system preset files.
1043 * nss-myhostname will now resolve the single-label host name
1044 "gateway" to the locally configured default IP routing
1045 gateways, ordered by their metrics. This assigns a stable
1046 name to the used gateways, regardless which ones are
1047 currently configured. Note that the name will only be
1048 resolved after all other name sources (if nss-myhostname is
1049 configured properly) and should hence not negatively impact
1050 systems that use the single-label host name "gateway" in
1053 * systemd-inhibit now allows filtering by mode when listing
1056 * Scope and service units gained a new "Delegate" boolean
1057 property, which when set allows processes running inside the
1058 unit to further partition resources. This is primarily
1059 useful for systemd user instances as well as container
1062 * journald will now pick up audit messages directly from
1063 the kernel, and log them like any other log message. The
1064 audit fields are split up and fully indexed. This means that
1065 journalctl in many ways is now a (nicer!) alternative to
1066 ausearch, the traditional audit client. Note that this
1067 implements only a minimal audit client, if you want the
1068 special audit modes like reboot-on-log-overflow, please use
1069 the traditional auditd instead, which can be used in
1070 parallel to journald.
1072 * The ConditionSecurity= unit file option now understands the
1073 special string "audit" to check whether auditing is
1076 * journalctl gained two new commands --vacuum-size= and
1077 --vacuum-time= to delete old journal files until the
1078 remaining ones take up no more the specified size on disk,
1079 or are not older than the specified time.
1081 * A new, native PPPoE library has been added to sd-network,
1082 systemd's library of light-weight networking protocols. This
1083 library will be used in a future version of networkd to
1084 enable PPPoE communication without an external pppd daemon.
1086 * The busctl tool now understands a new "capture" verb that
1087 works similar to "monitor", but writes a packet capture
1088 trace to STDOUT that can be redirected to a file which is
1089 compatible with libcap's capture file format. This can then
1090 be loaded in Wireshark and similar tools to inspect bus
1093 * The busctl tool now understands a new "tree" verb that shows
1094 the object trees of a specific service on the bus, or of all
1097 * The busctl tool now understands a new "introspect" verb that
1098 shows all interfaces and members of objects on the bus,
1099 including their signature and values. This is particularly
1100 useful to get more information about bus objects shown by
1101 the new "busctl tree" command.
1103 * The busctl tool now understands new verbs "call",
1104 "set-property" and "get-property" for invoking bus method
1105 calls, setting and getting bus object properties in a
1108 * busctl gained a new --augment-creds= argument that controls
1109 whether the tool shall augment credential information it
1110 gets from the bus with data from /proc, in a possibly
1113 * nspawn's --link-journal= switch gained two new values
1114 "try-guest" and "try-host" that work like "guest" and
1115 "host", but do not fail if the host has no persistent
1116 journalling enabled. -j is now equivalent to
1117 --link-journal=try-guest.
1119 * macvlan network devices created by nspawn will now have
1120 stable MAC addresses.
1122 * A new SmackProcessLabel= unit setting has been added, which
1123 controls the SMACK security label processes forked off by
1124 the respective unit shall use.
1126 * If compiled with --enable-xkbcommon, systemd-localed will
1127 verify x11 keymap settings by compiling the given keymap. It
1128 will spew out warnings if the compilation fails. This
1129 requires libxkbcommon to be installed.
1131 * When a coredump is collected a larger number of metadata
1132 fields is now collected and included in the journal records
1133 created for it. More specifically control group membership,
1134 environment variables, memory maps, working directory,
1135 chroot directory, /proc/$PID/status, and a list of open file
1136 descriptors is now stored in the log entry.
1138 * The udev hwdb now contains DPI information for mice. For
1141 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
1143 * All systemd programs that read standalone configuration
1144 files in /etc now also support a corresponding series of
1145 .conf.d configuration directories in /etc/, /run/,
1146 /usr/local/lib/, /usr/lib/, and (if configured with
1147 --enable-split-usr) /lib/. In particular, the following
1148 configuration files now have corresponding configuration
1149 directories: system.conf user.conf, logind.conf,
1150 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
1151 resolved.conf, timesyncd.conf, journal-remote.conf, and
1152 journal-upload.conf. Note that distributions should use the
1153 configuration directories in /usr/lib/; the directories in
1154 /etc/ are reserved for the system administrator.
1156 * systemd-rfkill will no longer take the rfkill device name
1157 into account when storing rfkill state on disk, as the name
1158 might be dynamically assigned and not stable. Instead, the
1159 ID_PATH udev variable combined with the rfkill type (wlan,
1160 bluetooth, ...) is used.
1162 * A new service systemd-machine-id-commit.service has been
1163 added. When used on systems where /etc is read-only during
1164 boot, and /etc/machine-id is not initialized (but an empty
1165 file), this service will copy the temporary machine ID
1166 created as replacement into /etc after the system is fully
1167 booted up. This is useful for systems that are freshly
1168 installed with a non-initialized machine ID, but should get
1169 a fixed machine ID for subsequent boots.
1171 * networkd's .netdev files now provide a large set of
1172 configuration parameters for VXLAN devices. Similar, the
1173 bridge port cost parameter is now configurable in .network
1174 files. There's also new support for configuring IP source
1175 routing. networkd .link files gained support for a new
1176 OriginalName= match that is useful to match against the
1177 original interface name the kernel assigned. .network files
1178 may include MTU= and MACAddress= fields for altering the MTU
1179 and MAC address while being connected to a specific network
1182 * The LUKS logic gained supported for configuring
1183 UUID-specific key files. There's also new support for naming
1184 LUKS device from the kernel command line, using the new
1185 luks.name= argument.
1187 * Timer units may now be transiently created via the bus API
1188 (this was previously already available for scope and service
1189 units). In addition it is now possible to create multiple
1190 transient units at the same time with a single bus call. The
1191 "systemd-run" tool has been updated to make use of this for
1192 running commands on a specified time, in at(1)-style.
1194 * tmpfiles gained support for "t" lines, for assigning
1195 extended attributes to files. Among other uses this may be
1196 used to assign SMACK labels to files.
1198 Contributions from: Alin Rauta, Alison Chaiken, Andrej
1199 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
1200 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
1201 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
1202 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
1203 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
1204 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
1205 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
1206 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
1207 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
1208 Michael Biebl, Michael Chapman, Michael Marineau, Michal
1209 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
1210 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
1211 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
1212 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
1213 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
1214 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
1215 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
1217 -- Berlin, 2014-12-10
1221 * journalctl gained the new options -t/--identifier= to match
1222 on the syslog identifier (aka "tag"), as well as --utc to
1223 show log timestamps in the UTC timezone. journalctl now also
1224 accepts -n/--lines=all to disable line capping in a pager.
1226 * journalctl gained a new switch, --flush, that synchronously
1227 flushes logs from /run/log/journal to /var/log/journal if
1228 persistent storage is enabled. systemd-journal-flush.service
1229 now waits until the operation is complete.
1231 * Services can notify the manager before they start a reload
1232 (by sending RELOADING=1) or shutdown (by sending
1233 STOPPING=1). This allows the manager to track and show the
1234 internal state of daemons and closes a race condition when
1235 the process is still running but has closed its D-Bus
1238 * Services with Type=oneshot do not have to have any ExecStart
1241 * User units are now loaded also from
1242 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
1243 /run/systemd/user directory that was already previously
1244 supported, but is under the control of the user.
1246 * Job timeouts (i.e. time-outs on the time a job that is
1247 queued stays in the run queue) can now optionally result in
1248 immediate reboot or power-off actions (JobTimeoutAction= and
1249 JobTimeoutRebootArgument=). This is useful on ".target"
1250 units, to limit the maximum time a target remains
1251 undispatched in the run queue, and to trigger an emergency
1252 operation in such a case. This is now used by default to
1253 turn off the system if boot-up (as defined by everything in
1254 basic.target) hangs and does not complete for at least
1255 15min. Also, if power-off or reboot hang for at least 30min
1256 an immediate power-off/reboot operation is triggered. This
1257 functionality is particularly useful to increase reliability
1258 on embedded devices, but also on laptops which might
1259 accidentally get powered on when carried in a backpack and
1260 whose boot stays stuck in a hard disk encryption passphrase
1263 * systemd-logind can be configured to also handle lid switch
1264 events even when the machine is docked or multiple displays
1265 are attached (HandleLidSwitchDocked= option).
1267 * A helper binary and a service have been added which can be
1268 used to resume from hibernation in the initramfs. A
1269 generator will parse the resume= option on the kernel
1270 command line to trigger resume.
1272 * A user console daemon systemd-consoled has been
1273 added. Currently, it is a preview, and will so far open a
1274 single terminal on each session of the user marked as
1275 Desktop=systemd-console.
1277 * Route metrics can be specified for DHCP routes added by
1280 * The SELinux context of socket-activated services can be set
1281 from the information provided by the networking stack
1282 (SELinuxContextFromNet= option).
1284 * Userspace firmware loading support has been removed and
1285 the minimum supported kernel version is thus bumped to 3.7.
1287 * Timeout for udev workers has been increased from 1 to 3
1288 minutes, but a warning will be printed after 1 minute to
1289 help diagnose kernel modules that take a long time to load.
1291 * Udev rules can now remove tags on devices with TAG-="foobar".
1293 * systemd's readahead implementation has been removed. In many
1294 circumstances it didn't give expected benefits even for
1295 rotational disk drives and was becoming less relevant in the
1296 age of SSDs. As none of the developers has been using
1297 rotating media anymore, and nobody stepped up to actively
1298 maintain this component of systemd it has now been removed.
1300 * Swap units can use Options= to specify discard options.
1301 Discard options specified for swaps in /etc/fstab are now
1304 * Docker containers are now detected as a separate type of
1307 * The Password Agent protocol gained support for queries where
1308 the user input is shown, useful e.g. for user names.
1309 systemd-ask-password gained a new --echo option to turn that
1312 * The default sysctl.d/ snippets will now set:
1314 net.core.default_qdisc = fq_codel
1316 This selects Fair Queuing Controlled Delay as the default
1317 queuing discipline for network interfaces. fq_codel helps
1318 fight the network bufferbloat problem. It is believed to be
1319 a good default with no tuning required for most workloads.
1320 Downstream distributions may override this choice. On 10Gbit
1321 servers that do not do forwarding, "fq" may perform better.
1322 Systems without a good clocksource should use "pfifo_fast".
1324 * If kdbus is enabled during build a new option BusPolicy= is
1325 available for service units, that allows locking all service
1326 processes into a stricter bus policy, in order to limit
1327 access to various bus services, or even hide most of them
1328 from the service's view entirely.
1330 * networkctl will now show the .network and .link file
1331 networkd has applied to a specific interface.
1333 * sd-login gained a new API call sd_session_get_desktop() to
1334 query which desktop environment has been selected for a
1337 * UNIX utmp support is now compile-time optional to support
1338 legacy-free systems.
1340 * systemctl gained two new commands "add-wants" and
1341 "add-requires" for pulling in units from specific targets
1344 * If the word "rescue" is specified on the kernel command line
1345 the system will now boot into rescue mode (aka
1346 rescue.target), which was previously available only by
1347 specifying "1" or "systemd.unit=rescue.target" on the kernel
1348 command line. This new kernel command line option nicely
1349 mirrors the already existing "emergency" kernel command line
1352 * New kernel command line options mount.usr=, mount.usrflags=,
1353 mount.usrfstype= have been added that match root=, rootflags=,
1354 rootfstype= but allow mounting a specific file system to
1357 * The $NOTIFY_SOCKET is now also passed to control processes of
1358 services, not only the main process.
1360 * This version reenables support for fsck's -l switch. This
1361 means at least version v2.25 of util-linux is required for
1362 operation, otherwise dead-locks on device nodes may
1363 occur. Again: you need to update util-linux to at least
1364 v2.25 when updating systemd to v217.
1366 * The "multi-seat-x" tool has been removed from systemd, as
1367 its functionality has been integrated into X servers 1.16,
1368 and the tool is hence redundant. It is recommended to update
1369 display managers invoking this tool to simply invoke X
1370 directly from now on, again.
1372 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
1373 message flag has been added for all of systemd's PolicyKit
1374 authenticated method calls has been added. In particular
1375 this now allows optional interactive authorization via
1376 PolicyKit for many of PID1's privileged operations such as
1377 unit file enabling and disabling.
1379 * "udevadm hwdb --update" learnt a new switch "--usr" for
1380 placing the rebuilt hardware database in /usr instead of
1381 /etc. When used only hardware database entries stored in
1382 /usr will be used, and any user database entries in /etc are
1383 ignored. This functionality is useful for vendors to ship a
1384 pre-built database on systems where local configuration is
1385 unnecessary or unlikely.
1387 * Calendar time specifications in .timer units now also
1388 understand the strings "semi-annually", "quarterly" and
1389 "minutely" as shortcuts (in addition to the preexisting
1390 "anually", "hourly", ...).
1392 * systemd-tmpfiles will now correctly create files in /dev
1393 at boot which are marked for creation only at boot. It is
1394 recommended to always create static device nodes with 'c!'
1395 and 'b!', so that they are created only at boot and not
1396 overwritten at runtime.
1398 * When the watchdog logic is used for a service (WatchdogSec=)
1399 and the watchdog timeout is hit the service will now be
1400 terminated with SIGABRT (instead of just SIGTERM), in order
1401 to make sure a proper coredump and backtrace is
1402 generated. This ensures that hanging services will result in
1403 similar coredump/backtrace behaviour as services that hit a
1406 Contributions from: Andreas Henriksson, Andrei Borzenkov,
1407 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
1408 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
1409 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
1410 Herrmann, David Sommerseth, David Strauss, Emil Renner
1411 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
1412 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
1413 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
1414 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
1415 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
1416 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
1417 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
1418 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
1419 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
1420 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
1421 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
1422 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
1423 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
1424 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
1425 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
1426 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
1429 -- Berlin, 2014-10-28
1433 * timedated no longer reads NTP implementation unit names from
1434 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
1435 implementations should add a
1437 Conflicts=systemd-timesyncd.service
1439 to their unit files to take over and replace systemd's NTP
1440 default functionality.
1442 * systemd-sysusers gained a new line type "r" for configuring
1443 which UID/GID ranges to allocate system users/groups
1444 from. Lines of type "u" may now add an additional column
1445 that specifies the home directory for the system user to be
1446 created. Also, systemd-sysusers may now optionally read user
1447 information from STDIN instead of a file. This is useful for
1448 invoking it from RPM preinst scriptlets that need to create
1449 users before the first RPM file is installed since these
1450 files might need to be owned by them. A new
1451 %sysusers_create_inline RPM macro has been introduced to do
1452 just that. systemd-sysusers now updates the shadow files as
1453 well as the user/group databases, which should enhance
1454 compatibility with certain tools like grpck.
1456 * A number of bus APIs of PID 1 now optionally consult
1457 PolicyKit to permit access for otherwise unprivileged
1458 clients under certain conditions. Note that this currently
1459 doesn't support interactive authentication yet, but this is
1460 expected to be added eventually, too.
1462 * /etc/machine-info now has new fields for configuring the
1463 deployment environment of the machine, as well as the
1464 location of the machine. hostnamectl has been updated with
1465 new command to update these fields.
1467 * systemd-timesyncd has been updated to automatically acquire
1468 NTP server information from systemd-networkd, which might
1469 have been discovered via DHCP.
1471 * systemd-resolved now includes a caching DNS stub resolver
1472 and a complete LLMNR name resolution implementation. A new
1473 NSS module "nss-resolve" has been added which can be used
1474 instead of glibc's own "nss-dns" to resolve hostnames via
1475 systemd-resolved. Hostnames, addresses and arbitrary RRs may
1476 be resolved via systemd-resolved D-Bus APIs. In contrast to
1477 the glibc internal resolver systemd-resolved is aware of
1478 multi-homed system, and keeps DNS server and caches separate
1479 and per-interface. Queries are sent simultaneously on all
1480 interfaces that have DNS servers configured, in order to
1481 properly handle VPNs and local LANs which might resolve
1482 separate sets of domain names. systemd-resolved may acquire
1483 DNS server information from systemd-networkd automatically,
1484 which in turn might have discovered them via DHCP. A tool
1485 "systemd-resolve-host" has been added that may be used to
1486 query the DNS logic in resolved. systemd-resolved implements
1487 IDNA and automatically uses IDNA or UTF-8 encoding depending
1488 on whether classic DNS or LLMNR is used as transport. In the
1489 next releases we intend to add a DNSSEC and mDNS/DNS-SD
1490 implementation to systemd-resolved.
1492 * A new NSS module nss-mymachines has been added, that
1493 automatically resolves the names of all local registered
1494 containers to their respective IP addresses.
1496 * A new client tool "networkctl" for systemd-networkd has been
1497 added. It currently is entirely passive and will query
1498 networking configuration from udev, rtnetlink and networkd,
1499 and present it to the user in a very friendly
1500 way. Eventually, we hope to extend it to become a full
1501 control utility for networkd.
1503 * .socket units gained a new DeferAcceptSec= setting that
1504 controls the kernels' TCP_DEFER_ACCEPT sockopt for
1505 TCP. Similar, support for controlling TCP keep-alive
1506 settings has been added (KeepAliveTimeSec=,
1507 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
1508 turning off Nagle's algorithm on TCP has been added
1511 * logind learned a new session type "web", for use in projects
1512 like Cockpit which register web clients as PAM sessions.
1514 * timer units with at least one OnCalendar= setting will now
1515 be started only after timer-sync.target has been
1516 reached. This way they will not elapse before the system
1517 clock has been corrected by a local NTP client or
1518 similar. This is particular useful on RTC-less embedded
1519 machines, that come up with an invalid system clock.
1521 * systemd-nspawn's --network-veth= switch should now result in
1522 stable MAC addresses for both the outer and the inner side
1525 * systemd-nspawn gained a new --volatile= switch for running
1526 container instances with /etc or /var unpopulated.
1528 * The kdbus client code has been updated to use the new Linux
1529 3.17 memfd subsystem instead of the old kdbus-specific one.
1531 * systemd-networkd's DHCP client and server now support
1532 FORCERENEW. There are also new configuration options to
1533 configure the vendor client identifier and broadcast mode
1536 * systemd will no longer inform the kernel about the current
1537 timezone, as this is necessarily incorrect and racy as the
1538 kernel has no understanding of DST and similar
1539 concepts. This hence means FAT timestamps will be always
1540 considered UTC, similar to what Android is already
1541 doing. Also, when the RTC is configured to the local time
1542 (rather than UTC) systemd will never synchronize back to it,
1543 as this might confuse Windows at a later boot.
1545 * systemd-analyze gained a new command "verify" for offline
1546 validation of unit files.
1548 * systemd-networkd gained support for a couple of additional
1549 settings for bonding networking setups. Also, the metric for
1550 statically configured routes may now be configured. For
1551 network interfaces where this is appropriate the peer IP
1552 address may now be configured.
1554 * systemd-networkd's DHCP client will no longer request
1555 broadcasting by default, as this tripped up some networks.
1556 For hardware where broadcast is required the feature should
1557 be switched back on using RequestBroadcast=yes.
1559 * systemd-networkd will now set up IPv4LL addresses (when
1560 enabled) even if DHCP is configured successfully.
1562 * udev will now default to respect network device names given
1563 by the kernel when the kernel indicates that these are
1564 predictable. This behavior can be tweaked by changing
1565 NamePolicy= in the relevant .link file.
1567 * A new library systemd-terminal has been added that
1568 implements full TTY stream parsing and rendering. This
1569 library is supposed to be used later on for implementing a
1570 full userspace VT subsystem, replacing the current kernel
1573 * A new tool systemd-journal-upload has been added to push
1574 journal data to a remote system running
1575 systemd-journal-remote.
1577 * journald will no longer forward all local data to another
1578 running syslog daemon. This change has been made because
1579 rsyslog (which appears to be the most commonly used syslog
1580 implementation these days) no longer makes use of this, and
1581 instead pulls the data out of the journal on its own. Since
1582 forwarding the messages to a non-existent syslog server is
1583 more expensive than we assumed we have now turned this
1584 off. If you run a syslog server that is not a recent rsyslog
1585 version, you have to turn this option on again
1586 (ForwardToSyslog= in journald.conf).
1588 * journald now optionally supports the LZ4 compressor for
1589 larger journal fields. This compressor should perform much
1590 better than XZ which was the previous default.
1592 * machinectl now shows the IP addresses of local containers,
1593 if it knows them, plus the interface name of the container.
1595 * A new tool "systemd-escape" has been added that makes it
1596 easy to escape strings to build unit names and similar.
1598 * sd_notify() messages may now include a new ERRNO= field
1599 which is parsed and collected by systemd and shown among the
1600 "systemctl status" output for a service.
1602 * A new component "systemd-firstboot" has been added that
1603 queries the most basic systemd information (timezone,
1604 hostname, root password) interactively on first
1605 boot. Alternatively it may also be used to provision these
1606 things offline on OS images installed into directories.
1608 * The default sysctl.d/ snippets will now set
1610 net.ipv4.conf.default.promote_secondaries=1
1612 This has the benefit of no flushing secondary IP addresses
1613 when primary addresses are removed.
1615 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
1616 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
1617 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
1618 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
1619 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
1620 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
1621 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
1622 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
1623 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
1624 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
1625 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
1626 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
1627 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
1628 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
1629 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
1631 -- Berlin, 2014-08-19
1635 * A new tool systemd-sysusers has been added. This tool
1636 creates system users and groups in /etc/passwd and
1637 /etc/group, based on static declarative system user/group
1638 definitions in /usr/lib/sysusers.d/. This is useful to
1639 enable factory resets and volatile systems that boot up with
1640 an empty /etc directory, and thus need system users and
1641 groups created during early boot. systemd now also ships
1642 with two default sysusers.d/ files for the most basic
1643 users and groups systemd and the core operating system
1646 * A new tmpfiles snippet has been added that rebuilds the
1647 essential files in /etc on boot, should they be missing.
1649 * A directive for ensuring automatic clean-up of
1650 /var/cache/man/ has been removed from the default
1651 configuration. This line should now be shipped by the man
1652 implementation. The necessary change has been made to the
1653 man-db implementation. Note that you need to update your man
1654 implementation to one that ships this line, otherwise no
1655 automatic clean-up of /var/cache/man will take place.
1657 * A new condition ConditionNeedsUpdate= has been added that
1658 may conditionalize services to only run when /etc or /var
1659 are "older" than the vendor operating system resources in
1660 /usr. This is useful for reconstructing or updating /etc
1661 after an offline update of /usr or a factory reset, on the
1662 next reboot. Services that want to run once after such an
1663 update or reset should use this condition and order
1664 themselves before the new systemd-update-done.service, which
1665 will mark the two directories as fully updated. A number of
1666 service files have been added making use of this, to rebuild
1667 the udev hardware database, the journald message catalog and
1668 dynamic loader cache (ldconfig). The systemd-sysusers tool
1669 described above also makes use of this now. With this in
1670 place it is now possible to start up a minimal operating
1671 system with /etc empty cleanly. For more information on the
1672 concepts involved see this recent blog story:
1674 http://0pointer.de/blog/projects/stateless.html
1676 * A new system group "input" has been introduced, and all
1677 input device nodes get this group assigned. This is useful
1678 for system-level software to get access to input devices. It
1679 complements what is already done for "audio" and "video".
1681 * systemd-networkd learnt minimal DHCPv4 server support in
1682 addition to the existing DHCPv4 client support. It also
1683 learnt DHCPv6 client and IPv6 Router Solicitation client
1684 support. The DHCPv4 client gained support for static routes
1685 passed in from the server. Note that the [DHCPv4] section
1686 known in older systemd-networkd versions has been renamed to
1687 [DHCP] and is now also used by the DHCPv6 client. Existing
1688 .network files using settings of this section should be
1689 updated, though compatibility is maintained. Optionally, the
1690 client hostname may now be sent to the DHCP server.
1692 * networkd gained support for vxlan virtual networks as well
1693 as tun/tap and dummy devices.
1695 * networkd gained support for automatic allocation of address
1696 ranges for interfaces from a system-wide pool of
1697 addresses. This is useful for dynamically managing a large
1698 number of interfaces with a single network configuration
1699 file. In particular this is useful to easily assign
1700 appropriate IP addresses to the veth links of a large number
1701 of nspawn instances.
1703 * RPM macros for processing sysusers, sysctl and binfmt
1704 drop-in snippets at package installation time have been
1707 * The /etc/os-release file should now be placed in
1708 /usr/lib/os-release. The old location is automatically
1709 created as symlink. /usr/lib is the more appropriate
1710 location of this file, since it shall actually describe the
1711 vendor operating system shipped in /usr, and not the
1712 configuration stored in /etc.
1714 * .mount units gained a new boolean SloppyOptions= setting
1715 that maps to mount(8)'s -s option which enables permissive
1716 parsing of unknown mount options.
1718 * tmpfiles learnt a new "L+" directive which creates a symlink
1719 but (unlike "L") deletes a pre-existing file first, should
1720 it already exist and not already be the correct
1721 symlink. Similar, "b+", "c+" and "p+" directives have been
1722 added as well, which create block and character devices, as
1723 well as fifos in the filesystem, possibly removing any
1724 pre-existing files of different types.
1726 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
1727 'argument' field (which so far specified the source to
1728 symlink/copy the files from) is now optional. If omitted the
1729 same file os copied from /usr/share/factory/ suffixed by the
1730 full destination path. This is useful for populating /etc
1731 with essential files, by copying them from vendor defaults
1732 shipped in /usr/share/factory/etc.
1734 * A new command "systemctl preset-all" has been added that
1735 applies the service preset settings to all installed unit
1736 files. A new switch --preset-mode= has been added that
1737 controls whether only enable or only disable operations
1740 * A new command "systemctl is-system-running" has been added
1741 that allows checking the overall state of the system, for
1742 example whether it is fully up and running.
1744 * When the system boots up with an empty /etc, the equivalent
1745 to "systemctl preset-all" is executed during early boot, to
1746 make sure all default services are enabled after a factory
1749 * systemd now contains a minimal preset file that enables the
1750 most basic services systemd ships by default.
1752 * Unit files' [Install] section gained a new DefaultInstance=
1753 field for defining the default instance to create if a
1754 template unit is enabled with no instance specified.
1756 * A new passive target cryptsetup-pre.target has been added
1757 that may be used by services that need to make they run and
1758 finish before the first LUKS cryptographic device is set up.
1760 * The /dev/loop-control and /dev/btrfs-control device nodes
1761 are now owned by the "disk" group by default, opening up
1762 access to this group.
1764 * systemd-coredump will now automatically generate a
1765 stack trace of all core dumps taking place on the system,
1766 based on elfutils' libdw library. This stack trace is logged
1769 * systemd-coredump may now optionally store coredumps directly
1770 on disk (in /var/lib/systemd/coredump, possibly compressed),
1771 instead of storing them unconditionally in the journal. This
1772 mode is the new default. A new configuration file
1773 /etc/systemd/coredump.conf has been added to configure this
1774 and other parameters of systemd-coredump.
1776 * coredumpctl gained a new "info" verb to show details about a
1777 specific coredump. A new switch "-1" has also been added
1778 that makes sure to only show information about the most
1779 recent entry instead of all entries. Also, as the tool is
1780 generally useful now the "systemd-" prefix of the binary
1781 name has been removed. Distributions that want to maintain
1782 compatibility with the old name should add a symlink from
1783 the old name to the new name.
1785 * journald's SplitMode= now defaults to "uid". This makes sure
1786 that unprivileged users can access their own coredumps with
1787 coredumpctl without restrictions.
1789 * New kernel command line options "systemd.wants=" (for
1790 pulling an additional unit during boot), "systemd.mask="
1791 (for masking a specific unit for the boot), and
1792 "systemd.debug-shell" (for enabling the debug shell on tty9)
1793 have been added. This is implemented in the new generator
1794 "systemd-debug-generator".
1796 * systemd-nspawn will now by default filter a couple of
1797 syscalls for containers, among them those required for
1798 kernel module loading, direct x86 IO port access, swap
1799 management, and kexec. Most importantly though
1800 open_by_handle_at() is now prohibited for containers,
1801 closing a hole similar to a recently discussed vulnerability
1802 in docker regarding access to files on file hierarchies the
1803 container should normally not have access to. Note that for
1804 nspawn we generally make no security claims anyway (and
1805 this is explicitly documented in the man page), so this is
1806 just a fix for one of the most obvious problems.
1808 * A new man page file-hierarchy(7) has been added that
1809 contains a minimized, modernized version of the file system
1810 layout systemd expects, similar in style to the FHS
1811 specification or hier(5). A new tool systemd-path(1) has
1812 been added to query many of these paths for the local
1815 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
1816 longer done. Since the directory now has a per-user size
1817 limit, and is cleaned on logout this appears unnecessary,
1818 in particular since this now brings the lifecycle of this
1819 directory closer in line with how IPC objects are handled.
1821 * systemd.pc now exports a number of additional directories,
1822 including $libdir (which is useful to identify the library
1823 path for the primary architecture of the system), and a
1824 couple of drop-in directories.
1826 * udev's predictable network interface names now use the dev_port
1827 sysfs attribute, introduced in linux 3.15 instead of dev_id to
1828 distinguish between ports of the same PCI function. dev_id should
1829 only be used for ports using the same HW address, hence the need
1832 * machined has been updated to export the OS version of a
1833 container (read from /etc/os-release and
1834 /usr/lib/os-release) on the bus. This is now shown in
1835 "machinectl status" for a machine.
1837 * A new service setting RestartForceExitStatus= has been
1838 added. If configured to a set of exit signals or process
1839 return values, the service will be restarted when the main
1840 daemon process exits with any of them, regardless of the
1843 * systemctl's -H switch for connecting to remote systemd
1844 machines has been extended so that it may be used to
1845 directly connect to a specific container on the
1846 host. "systemctl -H root@foobar:waldi" will now connect as
1847 user "root" to host "foobar", and then proceed directly to
1848 the container named "waldi". Note that currently you have to
1849 authenticate as user "root" for this to work, as entering
1850 containers is a privileged operation.
1852 Contributions from: Andreas Henriksson, Benjamin Steinwender,
1853 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
1854 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
1855 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
1856 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
1857 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
1858 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
1859 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
1860 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
1861 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
1862 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
1863 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
1865 -- Berlin, 2014-07-03
1869 * As an experimental feature, udev now tries to lock the
1870 disk device node (flock(LOCK_SH|LOCK_NB)) while it
1871 executes events for the disk or any of its partitions.
1872 Applications like partitioning programs can lock the
1873 disk device node (flock(LOCK_EX)) and claim temporary
1874 device ownership that way; udev will entirely skip all event
1875 handling for this disk and its partitions. If the disk
1876 was opened for writing, the close will trigger a partition
1877 table rescan in udev's "watch" facility, and if needed
1878 synthesize "change" events for the disk and all its partitions.
1879 This is now unconditionally enabled, and if it turns out to
1880 cause major problems, we might turn it on only for specific
1881 devices, or might need to disable it entirely. Device Mapper
1882 devices are excluded from this logic.
1884 * We temporarily dropped the "-l" switch for fsck invocations,
1885 since they collide with the flock() logic above. util-linux
1886 upstream has been changed already to avoid this conflict,
1887 and we will readd "-l" as soon as util-linux with this
1888 change has been released.
1890 * The dependency on libattr has been removed. Since a long
1891 time, the extended attribute calls have moved to glibc, and
1892 libattr is thus unnecessary.
1894 * Virtualization detection works without priviliges now. This
1895 means the systemd-detect-virt binary no longer requires
1896 CAP_SYS_PTRACE file capabilities, and our daemons can run
1897 with fewer privileges.
1899 * systemd-networkd now runs under its own "systemd-network"
1900 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
1901 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
1902 loses the ability to write to files owned by root this way.
1904 * Similar, systemd-resolved now runs under its own
1905 "systemd-resolve" user with no capabilities remaining.
1907 * Similar, systemd-bus-proxyd now runs under its own
1908 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
1910 * systemd-networkd gained support for setting up "veth"
1911 virtual ethernet devices for container connectivity, as well
1912 as GRE and VTI tunnels.
1914 * systemd-networkd will no longer automatically attempt to
1915 manually load kernel modules necessary for certain tunnel
1916 transports. Instead, it is assumed the kernel loads them
1917 automatically when required. This only works correctly on
1918 very new kernels. On older kernels, please consider adding
1919 the kernel modules to /etc/modules-load.d/ as a work-around.
1921 * The resolv.conf file systemd-resolved generates has been
1922 moved to /run/systemd/resolve/. If you have a symlink from
1923 /etc/resolv.conf, it might be necessary to correct it.
1925 * Two new service settings, ProtectHome= and ProtectSystem=,
1926 have been added. When enabled, they will make the user data
1927 (such as /home) inaccessible or read-only and the system
1928 (such as /usr) read-only, for specific services. This allows
1929 very light-weight per-service sandboxing to avoid
1930 modifications of user data or system files from
1931 services. These two new switches have been enabled for all
1932 of systemd's long-running services, where appropriate.
1934 * Socket units gained new SocketUser= and SocketGroup=
1935 settings to set the owner user and group of AF_UNIX sockets
1936 and FIFOs in the file system.
1938 * Socket units gained a new RemoveOnStop= setting. If enabled,
1939 all FIFOS and sockets in the file system will be removed
1940 when the specific socket unit is stopped.
1942 * Socket units gained a new Symlinks= setting. It takes a list
1943 of symlinks to create to file system sockets or FIFOs
1944 created by the specific Unix sockets. This is useful to
1945 manage symlinks to socket nodes with the same life-cycle as
1948 * The /dev/log socket and /dev/initctl FIFO have been moved to
1949 /run, and have been replaced by symlinks. This allows
1950 connecting to these facilities even if PrivateDevices=yes is
1951 used for a service (which makes /dev/log itself unavailable,
1952 but /run is left). This also has the benefit of ensuring
1953 that /dev only contains device nodes, directories and
1954 symlinks, and nothing else.
1956 * sd-daemon gained two new calls sd_pid_notify() and
1957 sd_pid_notifyf(). They are similar to sd_notify() and
1958 sd_notifyf(), but allow overriding of the source PID of
1959 notification messages if permissions permit this. This is
1960 useful to send notify messages on behalf of a different
1961 process (for example, the parent process). The
1962 systemd-notify tool has been updated to make use of this
1963 when sending messages (so that notification messages now
1964 originate from the shell script invoking systemd-notify and
1965 not the systemd-notify process itself. This should minimize
1966 a race where systemd fails to associate notification
1967 messages to services when the originating process already
1970 * A new "on-abnormal" setting for Restart= has been added. If
1971 set, it will result in automatic restarts on all "abnormal"
1972 reasons for a process to exit, which includes unclean
1973 signals, core dumps, timeouts and watchdog timeouts, but
1974 does not include clean and unclean exit codes or clean
1975 signals. Restart=on-abnormal is an alternative for
1976 Restart=on-failure for services that shall be able to
1977 terminate and avoid restarts on certain errors, by
1978 indicating so with an unclean exit code. Restart=on-failure
1979 or Restart=on-abnormal is now the recommended setting for
1980 all long-running services.
1982 * If the InaccessibleDirectories= service setting points to a
1983 mount point (or if there are any submounts contained within
1984 it), it is now attempted to completely unmount it, to make
1985 the file systems truly unavailable for the respective
1988 * The ReadOnlyDirectories= service setting and
1989 systemd-nspawn's --read-only parameter are now recursively
1990 applied to all submounts, too.
1992 * Mount units may now be created transiently via the bus APIs.
1994 * The support for SysV and LSB init scripts has been removed
1995 from the systemd daemon itself. Instead, it is now
1996 implemented as a generator that creates native systemd units
1997 from these scripts when needed. This enables us to remove a
1998 substantial amount of legacy code from PID 1, following the
1999 fact that many distributions only ship a very small number
2000 of LSB/SysV init scripts nowadays.
2002 * Privileged Xen (dom0) domains are not considered
2003 virtualization anymore by the virtualization detection
2004 logic. After all, they generally have unrestricted access to
2005 the hardware and usually are used to manage the unprivileged
2008 * systemd-tmpfiles gained a new "C" line type, for copying
2009 files or entire directories.
2011 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
2012 lines. So far, they have been non-globbing versions of the
2013 latter, and have thus been redundant. In future, it is
2014 recommended to only use "z". "m" has hence been removed
2015 from the documentation, even though it stays supported.
2017 * A tmpfiles snippet to recreate the most basic structure in
2018 /var has been added. This is enough to create the /var/run →
2019 /run symlink and create a couple of structural
2020 directories. This allows systems to boot up with an empty or
2021 volatile /var. Of course, while with this change, the core OS
2022 now is capable with dealing with a volatile /var, not all
2023 user services are ready for it. However, we hope that sooner
2024 or later, many service daemons will be changed upstream so
2025 that they are able to automatically create their necessary
2026 directories in /var at boot, should they be missing. This is
2027 the first step to allow state-less systems that only require
2028 the vendor image for /usr to boot.
2030 * systemd-nspawn has gained a new --tmpfs= switch to mount an
2031 empty tmpfs instance to a specific directory. This is
2032 particularly useful for making use of the automatic
2033 reconstruction of /var (see above), by passing --tmpfs=/var.
2035 * Access modes specified in tmpfiles snippets may now be
2036 prefixed with "~", which indicates that they shall be masked
2037 by whether the existing file or directory is currently
2038 writable, readable or executable at all. Also, if specified,
2039 the sgid/suid/sticky bits will be masked for all
2042 * A new passive target unit "network-pre.target" has been
2043 added which is useful for services that shall run before any
2044 network is configured, for example firewall scripts.
2046 * The "floppy" group that previously owned the /dev/fd*
2047 devices is no longer used. The "disk" group is now used
2048 instead. Distributions should probably deprecate usage of
2051 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
2052 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
2053 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
2054 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
2055 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
2056 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
2059 -- Berlin, 2014-06-11
2063 * A new "systemd-timesyncd" daemon has been added for
2064 synchronizing the system clock across the network. It
2065 implements an SNTP client. In contrast to NTP
2066 implementations such as chrony or the NTP reference server,
2067 this only implements a client side, and does not bother with
2068 the full NTP complexity, focusing only on querying time from
2069 one remote server and synchronizing the local clock to
2070 it. Unless you intend to serve NTP to networked clients or
2071 want to connect to local hardware clocks, this simple NTP
2072 client should be more than appropriate for most
2073 installations. The daemon runs with minimal privileges, and
2074 has been hooked up with networkd to only operate when
2075 network connectivity is available. The daemon saves the
2076 current clock to disk every time a new NTP sync has been
2077 acquired, and uses this to possibly correct the system clock
2078 early at bootup, in order to accommodate for systems that
2079 lack an RTC such as the Raspberry Pi and embedded devices,
2080 and to make sure that time monotonically progresses on these
2081 systems, even if it is not always correct. To make use of
2082 this daemon, a new system user and group "systemd-timesync"
2083 needs to be created on installation of systemd.
2085 * The queue "seqnum" interface of libudev has been disabled, as
2086 it was generally incompatible with device namespacing as
2087 sequence numbers of devices go "missing" if the devices are
2088 part of a different namespace.
2090 * "systemctl list-timers" and "systemctl list-sockets" gained
2091 a --recursive switch for showing units of these types also
2092 for all local containers, similar in style to the already
2093 supported --recursive switch for "systemctl list-units".
2095 * A new RebootArgument= setting has been added for service
2096 units, which may be used to specify a kernel reboot argument
2097 to use when triggering reboots with StartLimitAction=.
2099 * A new FailureAction= setting has been added for service
2100 units which may be used to specify an operation to trigger
2101 when a service fails. This works similarly to
2102 StartLimitAction=, but unlike it, controls what is done
2103 immediately rather than only after several attempts to
2104 restart the service in question.
2106 * hostnamed got updated to also expose the kernel name,
2107 release, and version on the bus. This is useful for
2108 executing commands like hostnamectl with the -H switch.
2109 systemd-analyze makes use of this to properly display
2110 details when running non-locally.
2112 * The bootchart tool can now show cgroup information in the
2113 graphs it generates.
2115 * The CFS CPU quota cgroup attribute is now exposed for
2116 services. The new CPUQuota= switch has been added for this
2117 which takes a percentage value. Setting this will have the
2118 result that a service may never get more CPU time than the
2119 specified percentage, even if the machine is otherwise idle.
2121 * systemd-networkd learned IPIP and SIT tunnel support.
2123 * LSB init scripts exposing a dependency on $network will now
2124 get a dependency on network-online.target rather than simply
2125 network.target. This should bring LSB handling closer to
2126 what it was on SysV systems.
2128 * A new fsck.repair= kernel option has been added to control
2129 how fsck shall deal with unclean file systems at boot.
2131 * The (.ini) configuration file parser will now silently
2132 ignore sections whose name begins with "X-". This may be
2133 used to maintain application-specific extension sections in unit
2136 * machined gained a new API to query the IP addresses of
2137 registered containers. "machinectl status" has been updated
2138 to show these addresses in its output.
2140 * A new call sd_uid_get_display() has been added to the
2141 sd-login APIs for querying the "primary" session of a
2142 user. The "primary" session of the user is elected from the
2143 user's sessions and generally a graphical session is
2144 preferred over a text one.
2146 * A minimal systemd-resolved daemon has been added. It
2147 currently simply acts as a companion to systemd-networkd and
2148 manages resolv.conf based on per-interface DNS
2149 configuration, possibly supplied via DHCP. In the long run
2150 we hope to extend this into a local DNSSEC enabled DNS and
2153 * The systemd-networkd-wait-online tool is now enabled by
2154 default. It will delay network-online.target until a network
2155 connection has been configured. The tool primarily integrates
2156 with networkd, but will also make a best effort to make sense
2157 of network configuration performed in some other way.
2159 * Two new service options StartupCPUShares= and
2160 StartupBlockIOWeight= have been added that work similarly to
2161 CPUShares= and BlockIOWeight= however only apply during
2162 system startup. This is useful to prioritize certain services
2163 differently during bootup than during normal runtime.
2165 * hostnamed has been changed to prefer the statically
2166 configured hostname in /etc/hostname (unless set to
2167 'localhost' or empty) over any dynamic one supplied by
2168 dhcp. With this change, the rules for picking the hostname
2169 match more closely the rules of other configuration settings
2170 where the local administrator's configuration in /etc always
2171 overrides any other settings.
2173 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
2174 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
2175 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
2176 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
2177 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
2178 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
2179 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
2180 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
2181 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
2182 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
2183 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
2184 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
2185 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
2186 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
2187 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
2188 Lindskog, WaLyong Cho, Will Woods, Zbigniew
2191 -- Beijing, 2014-05-28
2195 * When restoring the screen brightness at boot, stay away from
2196 the darkest setting or from the lowest 5% of the available
2197 range, depending on which is the larger value of both. This
2198 should effectively protect the user from rebooting into a
2199 black screen, should the brightness have been set to minimum
2202 * sd-login gained a new sd_machine_get_class() call to
2203 determine the class ("vm" or "container") of a machine
2204 registered with machined.
2206 * sd-login gained new calls
2207 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
2208 to query the identity of the peer of a local AF_UNIX
2209 connection. They operate similarly to their sd_pid_get_xyz()
2212 * PID 1 will now maintain a system-wide system state engine
2213 with the states "starting", "running", "degraded",
2214 "maintenance", "stopping". These states are bound to system
2215 startup, normal runtime, runtime with at least one failed
2216 service, rescue/emergency mode and system shutdown. This
2217 state is shown in the "systemctl status" output when no unit
2218 name is passed. It is useful to determine system state, in
2219 particularly when doing so for many systems or containers at
2222 * A new command "list-machines" has been added to "systemctl"
2223 that lists all local OS containers and shows their system
2224 state (see above), if systemd runs inside of them.
2226 * systemctl gained a new "-r" switch to recursively enumerate
2227 units on all local containers, when used with the
2228 "list-unit" command (which is the default one that is
2229 executed when no parameters are specified).
2231 * The GPT automatic partition discovery logic will now honour
2232 two GPT partition flags: one may be set on a partition to
2233 cause it to be mounted read-only, and the other may be set
2234 on a partition to ignore it during automatic discovery.
2236 * Two new GPT type UUIDs have been added for automatic root
2237 partition discovery, for 32-bit and 64-bit ARM. This is not
2238 particularly useful for discovering the root directory on
2239 these architectures during bare-metal boots (since UEFI is
2240 not common there), but still very useful to allow booting of
2241 ARM disk images in nspawn with the -i option.
2243 * MAC addresses of interfaces created with nspawn's
2244 --network-interface= switch will now be generated from the
2245 machine name, and thus be stable between multiple invocations
2248 * logind will now automatically remove all IPC objects owned
2249 by a user if she or he fully logs out. This makes sure that
2250 users who are logged out cannot continue to consume IPC
2251 resources. This covers SysV memory, semaphores and message
2252 queues as well as POSIX shared memory and message
2253 queues. Traditionally, SysV and POSIX IPC had no life-cycle
2254 limits. With this functionality, that is corrected. This may
2255 be turned off by using the RemoveIPC= switch of logind.conf.
2257 * The systemd-machine-id-setup and tmpfiles tools gained a
2258 --root= switch to operate on a specific root directory,
2261 * journald can now forward logged messages to the TTYs of all
2262 logged in users ("wall"). This is the default for all
2263 emergency messages now.
2265 * A new tool systemd-journal-remote has been added to stream
2266 journal log messages across the network.
2268 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
2269 controller trees are mounted into it. Note that the
2270 directories mounted beneath it are not read-only. This is a
2271 security measure and is particularly useful because glibc
2272 actually includes a search logic to pick any tmpfs it can
2273 find to implement shm_open() if /dev/shm is not available
2274 (which it might very well be in namespaced setups).
2276 * machinectl gained a new "poweroff" command to cleanly power
2277 down a local OS container.
2279 * The PrivateDevices= unit file setting will now also drop the
2280 CAP_MKNOD capability from the capability bound set, and
2281 imply DevicePolicy=closed.
2283 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
2284 comprehensively on all long-running systemd services where
2285 this is appropriate.
2287 * systemd-udevd will now run in a disassociated mount
2288 namespace. To mount directories from udev rules, make sure to
2289 pull in mount units via SYSTEMD_WANTS properties.
2291 * The kdbus support gained support for uploading policy into
2292 the kernel. sd-bus gained support for creating "monitoring"
2293 connections that can eavesdrop into all bus communication
2294 for debugging purposes.
2296 * Timestamps may now be specified in seconds since the UNIX
2297 epoch Jan 1st, 1970 by specifying "@" followed by the value
2300 * Native tcpwrap support in systemd has been removed. tcpwrap
2301 is old code, not really maintained anymore and has serious
2302 shortcomings, and better options such as firewalls
2303 exist. For setups that require tcpwrap usage, please
2304 consider invoking your socket-activated service via tcpd,
2305 like on traditional inetd.
2307 * A new system.conf configuration option
2308 DefaultTimerAccuracySec= has been added that controls the
2309 default AccuracySec= setting of .timer units.
2311 * Timer units gained a new WakeSystem= switch. If enabled,
2312 timers configured this way will cause the system to resume
2313 from system suspend (if the system supports that, which most
2316 * Timer units gained a new Persistent= switch. If enabled,
2317 timers configured this way will save to disk when they have
2318 been last triggered. This information is then used on next
2319 reboot to possible execute overdue timer events, that
2320 could not take place because the system was powered off.
2321 This enables simple anacron-like behaviour for timer units.
2323 * systemctl's "list-timers" will now also list the time a
2324 timer unit was last triggered in addition to the next time
2325 it will be triggered.
2327 * systemd-networkd will now assign predictable IPv4LL
2328 addresses to its local interfaces.
2330 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
2331 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
2332 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
2333 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
2334 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
2335 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
2336 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
2337 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
2340 -- Berlin, 2014-03-25
2344 * A new unit file setting RestrictAddressFamilies= has been
2345 added to restrict which socket address families unit
2346 processes gain access to. This takes address family names
2347 like "AF_INET" or "AF_UNIX", and is useful to minimize the
2348 attack surface of services via exotic protocol stacks. This
2349 is built on seccomp system call filters.
2351 * Two new unit file settings RuntimeDirectory= and
2352 RuntimeDirectoryMode= have been added that may be used to
2353 manage a per-daemon runtime directories below /run. This is
2354 an alternative for setting up directory permissions with
2355 tmpfiles snippets, and has the advantage that the runtime
2356 directory's lifetime is bound to the daemon runtime and that
2357 the daemon starts up with an empty directory each time. This
2358 is particularly useful when writing services that drop
2359 privileges using the User= or Group= setting.
2361 * The DeviceAllow= unit setting now supports globbing for
2362 matching against device group names.
2364 * The systemd configuration file system.conf gained new
2365 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
2366 DefaultMemoryAccounting= to globally turn on/off accounting
2367 for specific resources (cgroups) for all units. These
2368 settings may still be overridden individually in each unit
2371 * systemd-gpt-auto-generator is now able to discover /srv and
2372 root partitions in addition to /home and swap partitions. It
2373 also supports LUKS-encrypted partitions now. With this in
2374 place, automatic discovery of partitions to mount following
2375 the Discoverable Partitions Specification
2376 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
2377 is now a lot more complete. This allows booting without
2378 /etc/fstab and without root= on the kernel command line on
2379 systems prepared appropriately.
2381 * systemd-nspawn gained a new --image= switch which allows
2382 booting up disk images and Linux installations on any block
2383 device that follow the Discoverable Partitions Specification
2384 (see above). This means that installations made with
2385 appropriately updated installers may now be started and
2386 deployed using container managers, completely
2387 unmodified. (We hope that libvirt-lxc will add support for
2388 this feature soon, too.)
2390 * systemd-nspawn gained a new --network-macvlan= setting to
2391 set up a private macvlan interface for the
2392 container. Similarly, systemd-networkd gained a new
2393 Kind=macvlan setting in .netdev files.
2395 * systemd-networkd now supports configuring local addresses
2398 * A new tool systemd-network-wait-online has been added to
2399 synchronously wait for network connectivity using
2402 * The sd-bus.h bus API gained a new sd_bus_track object for
2403 tracking the life-cycle of bus peers. Note that sd-bus.h is
2404 still not a public API though (unless you specify
2405 --enable-kdbus on the configure command line, which however
2406 voids your warranty and you get no API stability guarantee).
2408 * The $XDG_RUNTIME_DIR runtime directories for each user are
2409 now individual tmpfs instances, which has the benefit of
2410 introducing separate pools for each user, with individual
2411 size limits, and thus making sure that unprivileged clients
2412 can no longer negatively impact the system or other users by
2413 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
2414 RuntimeDirectorySize= has been introduced that allows
2415 controlling the default size limit for all users. It
2416 defaults to 10% of the available physical memory. This is no
2417 replacement for quotas on tmpfs though (which the kernel
2418 still does not support), as /dev/shm and /tmp are still
2419 shared resources used by both the system and unprivileged
2422 * logind will now automatically turn off automatic suspending
2423 on laptop lid close when more than one display is
2424 connected. This was previously expected to be implemented
2425 individually in desktop environments (such as GNOME),
2426 however has been added to logind now, in order to fix a
2427 boot-time race where a desktop environment might not have
2428 been started yet and thus not been able to take an inhibitor
2429 lock at the time where logind already suspends the system
2430 due to a closed lid.
2432 * logind will now wait at least 30s after each system
2433 suspend/resume cycle, and 3min after system boot before
2434 suspending the system due to a closed laptop lid. This
2435 should give USB docking stations and similar enough time to
2436 be probed and configured after system resume and boot in
2437 order to then act as suspend blocker.
2439 * systemd-run gained a new --property= setting which allows
2440 initialization of resource control properties (and others)
2441 for the created scope or service unit. Example: "systemd-run
2442 --property=BlockIOWeight=10 updatedb" may be used to run
2443 updatedb at a low block IO scheduling weight.
2445 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
2446 now also work in --scope mode.
2448 * When systemd is compiled with kdbus support, basic support
2449 for enforced policies is now in place. (Note that enabling
2450 kdbus still voids your warranty and no API compatibility
2453 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
2454 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
2455 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
2456 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
2457 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
2458 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
2459 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
2460 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
2461 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
2462 Zbigniew Jędrzejewski-Szmek
2464 -- Berlin, 2014-03-12
2468 * systemd will now relabel /dev after loading the SMACK policy
2469 according to SMACK rules.
2471 * A new unit file option AppArmorProfile= has been added to
2472 set the AppArmor profile for the processes of a unit.
2474 * A new condition check ConditionArchitecture= has been added
2475 to conditionalize units based on the system architecture, as
2476 reported by uname()'s "machine" field.
2478 * systemd-networkd now supports matching on the system
2479 virtualization, architecture, kernel command line, host name
2482 * logind is now a lot more aggressive when suspending the
2483 machine due to a closed laptop lid. Instead of acting only
2484 on the lid close action, it will continuously watch the lid
2485 status and act on it. This is useful for laptops where the
2486 power button is on the outside of the chassis so that it can
2487 be reached without opening the lid (such as the Lenovo
2488 Yoga). On those machines, logind will now immediately
2489 re-suspend the machine if the power button has been
2490 accidentally pressed while the laptop was suspended and in a
2491 backpack or similar.
2493 * logind will now watch SW_DOCK switches and inhibit reaction
2494 to the lid switch if it is pressed. This means that logind
2495 will not suspend the machine anymore if the lid is closed
2496 and the system is docked, if the laptop supports SW_DOCK
2497 notifications via the input layer. Note that ACPI docking
2498 stations do not generate this currently. Also note that this
2499 logic is usually not fully sufficient and Desktop
2500 Environments should take a lid switch inhibitor lock when an
2501 external display is connected, as systemd will not watch
2504 * nspawn will now make use of the devices cgroup controller by
2505 default, and only permit creation of and access to the usual
2506 API device nodes like /dev/null or /dev/random, as well as
2507 access to (but not creation of) the pty devices.
2509 * We will now ship a default .network file for
2510 systemd-networkd that automatically configures DHCP for
2511 network interfaces created by nspawn's --network-veth or
2512 --network-bridge= switches.
2514 * systemd will now understand the usual M, K, G, T suffixes
2515 according to SI conventions (i.e. to the base 1000) when
2516 referring to throughput and hardware metrics. It will stay
2517 with IEC conventions (i.e. to the base 1024) for software
2518 metrics, according to what is customary according to
2519 Wikipedia. We explicitly document which base applies for
2520 each configuration option.
2522 * The DeviceAllow= setting in unit files now supports a syntax
2523 to whitelist an entire group of devices node majors at once,
2524 based on the /proc/devices listing. For example, with the
2525 string "char-pts", it is now possible to whitelist all
2526 current and future pseudo-TTYs at once.
2528 * sd-event learned a new "post" event source. Event sources of
2529 this type are triggered by the dispatching of any event
2530 source of a type that is not "post". This is useful for
2531 implementing clean-up and check event sources that are
2532 triggered by other work being done in the program.
2534 * systemd-networkd is no longer statically enabled, but uses
2535 the usual [Install] sections so that it can be
2536 enabled/disabled using systemctl. It still is enabled by
2539 * When creating a veth interface pair with systemd-nspawn, the
2540 host side will now be prefixed with "vb-" if
2541 --network-bridge= is used, and with "ve-" if --network-veth
2542 is used. This way, it is easy to distinguish these cases on
2543 the host, for example to apply different configuration to
2544 them with systemd-networkd.
2546 * The compatibility libraries for libsystemd-journal.so,
2547 libsystem-id128.so, libsystemd-login.so and
2548 libsystemd-daemon.so do not make use of IFUNC
2549 anymore. Instead, we now build libsystemd.so multiple times
2550 under these alternative names. This means that the footprint
2551 is drastically increased, but given that these are
2552 transitional compatibility libraries, this should not matter
2553 much. This change has been made necessary to support the ARM
2554 platform for these compatibility libraries, as the ARM
2555 toolchain is not really at the same level as the toolchain
2556 for other architectures like x86 and does not support
2557 IFUNC. Please make sure to use --enable-compat-libs only
2558 during a transitional period!
2560 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
2561 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
2562 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
2563 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
2564 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
2565 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
2566 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
2567 Zbigniew Jędrzejewski-Szmek
2569 -- Berlin, 2014-02-24
2573 * A new component "systemd-networkd" has been added that can
2574 be used to configure local network interfaces statically or
2575 via DHCP. It is capable of bringing up bridges, VLANs, and
2576 bonding. Currently, no hook-ups for interactive network
2577 configuration are provided. Use this for your initrd,
2578 container, embedded, or server setup if you need a simple,
2579 yet powerful, network configuration solution. This
2580 configuration subsystem is quite nifty, as it allows wildcard
2581 hotplug matching in interfaces. For example, with a single
2582 configuration snippet, you can configure that all Ethernet
2583 interfaces showing up are automatically added to a bridge,
2584 or similar. It supports link-sensing and more.
2586 * A new tool "systemd-socket-proxyd" has been added which can
2587 act as a bidirectional proxy for TCP sockets. This is
2588 useful for adding socket activation support to services that
2589 do not actually support socket activation, including virtual
2590 machines and the like.
2592 * Add a new tool to save/restore rfkill state on
2595 * Save/restore state of keyboard backlights in addition to
2596 display backlights on shutdown/boot.
2598 * udev learned a new SECLABEL{} construct to label device
2599 nodes with a specific security label when they appear. For
2600 now, only SECLABEL{selinux} is supported, but the syntax is
2601 prepared for additional security frameworks.
2603 * udev gained a new scheme to configure link-level attributes
2604 from files in /etc/systemd/network/*.link. These files can
2605 match against MAC address, device path, driver name and type,
2606 and will apply attributes like the naming policy, link speed,
2607 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
2608 address assignment policy (randomized, ...).
2610 * The configuration of network interface naming rules for
2611 "permanent interface names" has changed: a new NamePolicy=
2612 setting in the [Link] section of .link files determines the
2613 priority of possible naming schemes (onboard, slot, mac,
2614 path). The default value of this setting is determined by
2615 /usr/lib/net/links/99-default.link. Old
2616 80-net-name-slot.rules udev configuration file has been
2617 removed, so local configuration overriding this file should
2618 be adapated to override 99-default.link instead.
2620 * When the User= switch is used in a unit file, also
2621 initialize $SHELL= based on the user database entry.
2623 * systemd no longer depends on libdbus. All communication is
2624 now done with sd-bus, systemd's low-level bus library
2627 * kdbus support has been added to PID 1 itself. When kdbus is
2628 enabled, this causes PID 1 to set up the system bus and
2629 enable support for a new ".busname" unit type that
2630 encapsulates bus name activation on kdbus. It works a little
2631 bit like ".socket" units, except for bus names. A new
2632 generator has been added that converts classic dbus1 service
2633 activation files automatically into native systemd .busname
2636 * sd-bus: add a light-weight vtable implementation that allows
2637 defining objects on the bus with a simple static const
2638 vtable array of its methods, signals and properties.
2640 * systemd will not generate or install static dbus
2641 introspection data anymore to /usr/share/dbus-1/interfaces,
2642 as the precise format of these files is unclear, and
2643 nothing makes use of it.
2645 * A proxy daemon is now provided to proxy clients connecting
2646 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
2647 compatibility with classic D-Bus.
2649 * A bus driver implementation has been added that supports the
2650 classic D-Bus bus driver calls on kdbus, also for
2651 compatibility purposes.
2653 * A new API "sd-event.h" has been added that implements a
2654 minimal event loop API built around epoll. It provides a
2655 couple of features that direct epoll usage is lacking:
2656 prioritization of events, scales to large numbers of timer
2657 events, per-event timer slack (accuracy), system-wide
2658 coalescing of timer events, exit handlers, watchdog
2659 supervision support using systemd's sd_notify() API, child
2662 * A new API "sd-rntl.h" has been added that provides an API
2663 around the route netlink interface of the kernel, similar in
2664 style to "sd-bus.h".
2666 * A new API "sd-dhcp-client.h" has been added that provides a
2667 small DHCPv4 client-side implementation. This is used by
2670 * There is a new kernel command line option
2671 "systemd.restore_state=0|1". When set to "0", none of the
2672 systemd tools will restore saved runtime state to hardware
2673 devices. More specifically, the rfkill and backlight states
2676 * The FsckPassNo= compatibility option in mount/service units
2677 has been removed. The fstab generator will now add the
2678 necessary dependencies automatically, and does not require
2679 PID1's support for that anymore.
2681 * journalctl gained a new switch, --list-boots, that lists
2682 recent boots with their times and boot IDs.
2684 * The various tools like systemctl, loginctl, timedatectl,
2685 busctl, systemd-run, ... have gained a new switch "-M" to
2686 connect to a specific, local OS container (as direct
2687 connection, without requiring SSH). This works on any
2688 container that is registered with machined, such as those
2689 created by libvirt-lxc or nspawn.
2691 * systemd-run and systemd-analyze also gained support for "-H"
2692 to connect to remote hosts via SSH. This is particularly
2693 useful for systemd-run because it enables queuing of jobs
2694 onto remote systems.
2696 * machinectl gained a new command "login" to open a getty
2697 login in any local container. This works with any container
2698 that is registered with machined (such as those created by
2699 libvirt-lxc or nspawn), and which runs systemd inside.
2701 * machinectl gained a new "reboot" command that may be used to
2702 trigger a reboot on a specific container that is registered
2703 with machined. This works on any container that runs an init
2704 system of some kind.
2706 * systemctl gained a new "list-timers" command to print a nice
2707 listing of installed timer units with the times they elapse
2710 * Alternative reboot() parameters may now be specified on the
2711 "systemctl reboot" command line and are passed to the
2712 reboot() system call.
2714 * systemctl gained a new --job-mode= switch to configure the
2715 mode to queue a job with. This is a more generic version of
2716 --fail, --irreversible, and --ignore-dependencies, which are
2717 still available but not advertised anymore.
2719 * /etc/systemd/system.conf gained new settings to configure
2720 various default timeouts of units, as well as the default
2721 start limit interval and burst. These may still be overridden
2724 * PID1 will now export on the bus profile data of the security
2725 policy upload process (such as the SELinux policy upload to
2728 * journald: when forwarding logs to the console, include
2729 timestamps (following the setting in
2730 /sys/module/printk/parameters/time).
2732 * OnCalendar= in timer units now understands the special
2733 strings "yearly" and "annually". (Both are equivalent)
2735 * The accuracy of timer units is now configurable with the new
2736 AccuracySec= setting. It defaults to 1min.
2738 * A new dependency type JoinsNamespaceOf= has been added that
2739 allows running two services within the same /tmp and network
2740 namespace, if PrivateNetwork= or PrivateTmp= are used.
2742 * A new command "cat" has been added to systemctl. It outputs
2743 the original unit file of a unit, and concatenates the
2744 contents of additional "drop-in" unit file snippets, so that
2745 the full configuration is shown.
2747 * systemctl now supports globbing on the various "list-xyz"
2748 commands, like "list-units" or "list-sockets", as well as on
2749 those commands which take multiple unit names.
2751 * journalctl's --unit= switch gained support for globbing.
2753 * All systemd daemons now make use of the watchdog logic so
2754 that systemd automatically notices when they hang.
2756 * If the $container_ttys environment variable is set,
2757 getty-generator will automatically spawn a getty for each
2758 listed tty. This is useful for container managers to request
2759 login gettys to be spawned on as many ttys as needed.
2761 * %h, %s, %U specifier support is not available anymore when
2762 used in unit files for PID 1. This is because NSS calls are
2763 not safe from PID 1. They stay available for --user
2764 instances of systemd, and as special case for the root user.
2766 * loginctl gained a new "--no-legend" switch to turn off output
2769 * The "sd-login.h" API gained three new calls:
2770 sd_session_is_remote(), sd_session_get_remote_user(),
2771 sd_session_get_remote_host() to query information about
2774 * The udev hardware database now also carries vendor/product
2775 information of SDIO devices.
2777 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
2778 determine whether watchdog notifications are requested by
2781 * Socket-activated per-connection services now include a
2782 short description of the connection parameters in the
2785 * tmpfiles gained a new "--boot" option. When this is not used,
2786 only lines where the command character is not suffixed with
2787 "!" are executed. When this option is specified, those
2788 options are executed too. This partitions tmpfiles
2789 directives into those that can be safely executed at any
2790 time, and those which should be run only at boot (for
2791 example, a line that creates /run/nologin).
2793 * A new API "sd-resolve.h" has been added which provides a simple
2794 asynchronous wrapper around glibc NSS host name resolution
2795 calls, such as getaddrinfo(). In contrast to glibc's
2796 getaddrinfo_a(), it does not use signals. In contrast to most
2797 other asynchronous name resolution libraries, this one does
2798 not reimplement DNS, but reuses NSS, so that alternate
2799 host name resolution systems continue to work, such as mDNS,
2800 LDAP, etc. This API is based on libasyncns, but it has been
2801 cleaned up for inclusion in systemd.
2803 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
2804 "sd-daemon.h" are no longer found in individual libraries
2805 libsystemd-journal.so, libsystemd-login.so,
2806 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
2807 merged them into a single library, libsystemd.so, which
2808 provides all symbols. The reason for this is cyclic
2809 dependencies, as these libraries tend to use each other's
2810 symbols. So far, we have managed to workaround that by linking
2811 a copy of a good part of our code into each of these
2812 libraries again and again, which, however, makes certain
2813 things hard to do, like sharing static variables. Also, it
2814 substantially increases footprint. With this change, there
2815 is only one library for the basic APIs systemd
2816 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
2817 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
2818 library as well, however are subject to the --enable-kdbus
2819 switch (see below). Note that "sd-dhcp-client.h" is not part
2820 of this library (this is because it only consumes, never
2821 provides, services of/to other APIs). To make the transition
2822 easy from the separate libraries to the unified one, we
2823 provide the --enable-compat-libs compile-time switch which
2824 will generate stub libraries that are compatible with the
2825 old ones but redirect all calls to the new one.
2827 * All of the kdbus logic and the new APIs "sd-bus.h",
2828 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
2829 and "sd-utf8.h" are compile-time optional via the
2830 "--enable-kdbus" switch, and they are not compiled in by
2831 default. To make use of kdbus, you have to explicitly enable
2832 the switch. Note however, that neither the kernel nor the
2833 userspace API for all of this is considered stable yet. We
2834 want to maintain the freedom to still change the APIs for
2835 now. By specifying this build-time switch, you acknowledge
2836 that you are aware of the instability of the current
2839 * Also, note that while kdbus is pretty much complete,
2840 it lacks one thing: proper policy support. This means you
2841 can build a fully working system with all features; however,
2842 it will be highly insecure. Policy support will be added in
2843 one of the next releases, at the same time that we will
2844 declare the APIs stable.
2846 * When the kernel command line argument "kdbus" is specified,
2847 systemd will automatically load the kdbus.ko kernel module. At
2848 this stage of development, it is only useful for testing kdbus
2849 and should not be used in production. Note: if "--enable-kdbus"
2850 is specified, and the kdbus.ko kernel module is available, and
2851 "kdbus" is added to the kernel command line, the entire system
2852 runs with kdbus instead of dbus-daemon, with the above mentioned
2853 problem of missing the system policy enforcement. Also a future
2854 version of kdbus.ko or a newer systemd will not be compatible with
2855 each other, and will unlikely be able to boot the machine if only
2856 one of them is updated.
2858 * systemctl gained a new "import-environment" command which
2859 uploads the caller's environment (or parts thereof) into the
2860 service manager so that it is inherited by services started
2861 by the manager. This is useful to upload variables like
2862 $DISPLAY into the user service manager.
2864 * A new PrivateDevices= switch has been added to service units
2865 which allows running a service with a namespaced /dev
2866 directory that does not contain any device nodes for
2867 physical devices. More specifically, it only includes devices
2868 such as /dev/null, /dev/urandom, and /dev/zero which are API
2871 * logind has been extended to support behaviour like VT
2872 switching on seats that do not support a VT. This makes
2873 multi-session available on seats that are not the first seat
2874 (seat0), and on systems where kernel support for VTs has
2875 been disabled at compile-time.
2877 * If a process holds a delay lock for system sleep or shutdown
2878 and fails to release it in time, we will now log its
2879 identity. This makes it easier to identify processes that
2880 cause slow suspends or power-offs.
2882 * When parsing /etc/crypttab, support for a new key-slot=
2883 option as supported by Debian is added. It allows indicating
2884 which LUKS slot to use on disk, speeding up key loading.
2886 * The sd_journald_sendv() API call has been checked and
2887 officially declared to be async-signal-safe so that it may
2888 be invoked from signal handlers for logging purposes.
2890 * Boot-time status output is now enabled automatically after a
2891 short timeout if boot does not progress, in order to give
2892 the user an indication what she or he is waiting for.
2894 * The boot-time output has been improved to show how much time
2895 remains until jobs expire.
2897 * The KillMode= switch in service units gained a new possible
2898 value "mixed". If set, and the unit is shut down, then the
2899 initial SIGTERM signal is sent only to the main daemon
2900 process, while the following SIGKILL signal is sent to
2901 all remaining processes of the service.
2903 * When a scope unit is registered, a new property "Controller"
2904 may be set. If set to a valid bus name, systemd will send a
2905 RequestStop() signal to this name when it would like to shut
2906 down the scope. This may be used to hook manager logic into
2907 the shutdown logic of scope units. Also, scope units may now
2908 be put in a special "abandoned" state, in which case the
2909 manager process which created them takes no further
2910 responsibilities for it.
2912 * When reading unit files, systemd will now verify
2913 the access mode of these files, and warn about certain
2914 suspicious combinations. This has been added to make it
2915 easier to track down packaging bugs where unit files are
2916 marked executable or world-writable.
2918 * systemd-nspawn gained a new "--setenv=" switch to set
2919 container-wide environment variables. The similar option in
2920 systemd-activate was renamed from "--environment=" to
2921 "--setenv=" for consistency.
2923 * systemd-nspawn has been updated to create a new kdbus domain
2924 for each container that is invoked, thus allowing each
2925 container to have its own set of system and user buses,
2926 independent of the host.
2928 * systemd-nspawn gained a new --drop-capability= switch to run
2929 the container with less capabilities than the default. Both
2930 --drop-capability= and --capability= now take the special
2931 string "all" for dropping or keeping all capabilities.
2933 * systemd-nspawn gained new switches for executing containers
2934 with specific SELinux labels set.
2936 * systemd-nspawn gained a new --quiet switch to not generate
2937 any additional output but the container's own console
2940 * systemd-nspawn gained a new --share-system switch to run a
2941 container without PID namespacing enabled.
2943 * systemd-nspawn gained a new --register= switch to control
2944 whether the container is registered with systemd-machined or
2945 not. This is useful for containers that do not run full
2946 OS images, but only specific apps.
2948 * systemd-nspawn gained a new --keep-unit which may be used
2949 when invoked as the only program from a service unit, and
2950 results in registration of the unit service itself in
2951 systemd-machined, instead of a newly opened scope unit.
2953 * systemd-nspawn gained a new --network-interface= switch for
2954 moving arbitrary interfaces to the container. The new
2955 --network-veth switch creates a virtual Ethernet connection
2956 between host and container. The new --network-bridge=
2957 switch then allows assigning the host side of this virtual
2958 Ethernet connection to a bridge device.
2960 * systemd-nspawn gained a new --personality= switch for
2961 setting the kernel personality for the container. This is
2962 useful when running a 32-bit container on a 64-bit host. A
2963 similar option Personality= is now also available for service
2966 * logind will now also track a "Desktop" identifier for each
2967 session which encodes the desktop environment of it. This is
2968 useful for desktop environments that want to identify
2969 multiple running sessions of itself easily.
2971 * A new SELinuxContext= setting for service units has been
2972 added that allows setting a specific SELinux execution
2973 context for a service.
2975 * Most systemd client tools will now honour $SYSTEMD_LESS for
2976 settings of the "less" pager. By default, these tools will
2977 override $LESS to allow certain operations to work, such as
2978 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
2979 influence this logic.
2981 * systemd's "seccomp" hook-up has been changed to make use of
2982 the libseccomp library instead of using its own
2983 implementation. This has benefits for portability among
2986 * For usage together with SystemCallFilter=, a new
2987 SystemCallErrorNumber= setting has been introduced that
2988 allows configuration of a system error number to be returned
2989 on filtered system calls, instead of immediately killing the
2990 process. Also, SystemCallArchitectures= has been added to
2991 limit access to system calls of a particular architecture
2992 (in order to turn off support for unused secondary
2993 architectures). There is also a global
2994 SystemCallArchitectures= setting in system.conf now to turn
2995 off support for non-native system calls system-wide.
2997 * systemd requires a kernel with a working name_to_handle_at(),
2998 please see the kernel config requirements in the README file.
3000 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
3001 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
3002 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
3003 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
3004 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
3005 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
3006 Elia Pinto, Florian Weimer, George McCollister, Goffredo
3007 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
3008 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
3009 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
3010 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
3011 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
3012 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
3013 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
3014 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
3015 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
3016 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
3017 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
3018 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
3019 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
3020 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
3021 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
3022 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
3023 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
3025 -- Berlin, 2014-02-20
3029 * logind has gained support for facilitating privileged input
3030 and drm device access for unprivileged clients. This work is
3031 useful to allow Wayland display servers (and similar
3032 programs, such as kmscon) to run under the user's ID and
3033 access input and drm devices which are normally
3034 protected. When this is used (and the kernel is new enough)
3035 logind will "mute" IO on the file descriptors passed to
3036 Wayland as long as it is in the background and "unmute" it
3037 if it returns into the foreground. This allows secure
3038 session switching without allowing background sessions to
3039 eavesdrop on input and display data. This also introduces
3040 session switching support if VT support is turned off in the
3041 kernel, and on seats that are not seat0.
3043 * A new kernel command line option luks.options= is understood
3044 now which allows specifying LUKS options for usage for LUKS
3045 encrypted partitions specified with luks.uuid=.
3047 * tmpfiles.d(5) snippets may now use specifier expansion in
3048 path names. More specifically %m, %b, %H, %v, are now
3049 replaced by the local machine id, boot id, hostname, and
3050 kernel version number.
3052 * A new tmpfiles.d(5) command "m" has been introduced which
3053 may be used to change the owner/group/access mode of a file
3054 or directory if it exists, but do nothing if it does not.
3056 * This release removes high-level support for the
3057 MemorySoftLimit= cgroup setting. The underlying kernel
3058 cgroup attribute memory.soft_limit= is currently badly
3059 designed and likely to be removed from the kernel API in its
3060 current form, hence we should not expose it for now.
3062 * The memory.use_hierarchy cgroup attribute is now enabled for
3063 all cgroups systemd creates in the memory cgroup
3064 hierarchy. This option is likely to be come the built-in
3065 default in the kernel anyway, and the non-hierarchical mode
3066 never made much sense in the intrinsically hierarchical
3069 * A new field _SYSTEMD_SLICE= is logged along with all journal
3070 messages containing the slice a message was generated
3071 from. This is useful to allow easy per-customer filtering of
3072 logs among other things.
3074 * systemd-journald will no longer adjust the group of journal
3075 files it creates to the "systemd-journal" group. Instead we
3076 rely on the journal directory to be owned by the
3077 "systemd-journal" group, and its setgid bit set, so that the
3078 kernel file system layer will automatically enforce that
3079 journal files inherit this group assignment. The reason for
3080 this change is that we cannot allow NSS look-ups from
3081 journald which would be necessary to resolve
3082 "systemd-journal" to a numeric GID, because this might
3083 create deadlocks if NSS involves synchronous queries to
3084 other daemons (such as nscd, or sssd) which in turn are
3085 logging clients of journald and might block on it, which
3086 would then dead lock. A tmpfiles.d(5) snippet included in
3087 systemd will make sure the setgid bit and group are
3088 properly set on the journal directory if it exists on every
3089 boot. However, we recommend adjusting it manually after
3090 upgrades too (or from RPM scriptlets), so that the change is
3091 not delayed until next reboot.
3093 * Backlight and random seed files in /var/lib/ have moved into
3094 the /var/lib/systemd/ directory, in order to centralize all
3095 systemd generated files in one directory.
3097 * Boot time performance measurements (as displayed by
3098 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
3099 performance information if that's available to determine how
3100 much time BIOS and boot loader initialization required. With
3101 a sufficiently new BIOS you hence no longer need to boot
3102 with Gummiboot to get access to such information.
3104 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
3105 Cristian Rodríguez, Dave Reisner, David Herrmann, David
3106 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
3107 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
3108 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
3109 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
3110 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3112 -- Berlin, 2013-10-02
3116 * The Restart= option for services now understands a new
3117 on-watchdog setting, which will restart the service
3118 automatically if the service stops sending out watchdog keep
3119 alive messages (as configured with WatchdogSec=).
3121 * The getty generator (which is responsible for bringing up a
3122 getty on configured serial consoles) will no longer only
3123 start a getty on the primary kernel console but on all
3124 others, too. This makes the order in which console= is
3125 specified on the kernel command line less important.
3127 * libsystemd-logind gained a new sd_session_get_vt() call to
3128 retrieve the VT number of a session.
3130 * If the option "tries=0" is set for an entry of /etc/crypttab
3131 its passphrase is queried indefinitely instead of any
3132 maximum number of tries.
3134 * If a service with a configure PID file terminates its PID
3135 file will now be removed automatically if it still exists
3136 afterwards. This should put an end to stale PID files.
3138 * systemd-run will now also take relative binary path names
3139 for execution and no longer insists on absolute paths.
3141 * InaccessibleDirectories= and ReadOnlyDirectories= now take
3142 paths that are optionally prefixed with "-" to indicate that
3143 it should not be considered a failure if they do not exist.
3145 * journalctl -o (and similar commands) now understands a new
3146 output mode "short-precise", it is similar to "short" but
3147 shows timestamps with usec accuracy.
3149 * The option "discard" (as known from Debian) is now
3150 synonymous to "allow-discards" in /etc/crypttab. In fact,
3151 "discard" is preferred now (since it is easier to remember
3154 * Some licensing clean-ups were made, so that more code is now
3155 LGPL-2.1 licensed than before.
3157 * A minimal tool to save/restore the display backlight
3158 brightness across reboots has been added. It will store the
3159 backlight setting as late as possible at shutdown, and
3160 restore it as early as possible during reboot.
3162 * A logic to automatically discover and enable home and swap
3163 partitions on GPT disks has been added. With this in place
3164 /etc/fstab becomes optional for many setups as systemd can
3165 discover certain partitions located on the root disk
3166 automatically. Home partitions are recognized under their
3167 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
3168 partitions are recognized under their GPT type ID
3169 0657fd6da4ab43c484e50933c84b4f4f.
3171 * systemd will no longer pass any environment from the kernel
3172 or initrd to system services. If you want to set an
3173 environment for all services, do so via the kernel command
3174 line systemd.setenv= assignment.
3176 * The systemd-sysctl tool no longer natively reads the file
3177 /etc/sysctl.conf. If desired, the file should be symlinked
3178 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
3179 legacy support by a symlink rather than built-in code, it
3180 also makes the otherwise hidden order of application of the
3181 different files visible. (Note that this partly reverts to a
3182 pre-198 application order of sysctl knobs!)
3184 * The "systemctl set-log-level" and "systemctl dump" commands
3185 have been moved to systemd-analyze.
3187 * systemd-run learned the new --remain-after-exit switch,
3188 which causes the scope unit not to be cleaned up
3189 automatically after the process terminated.
3191 * tmpfiles learned a new --exclude-prefix= switch to exclude
3192 certain paths from operation.
3194 * journald will now automatically flush all messages to disk
3195 as soon as a message at the log level CRIT, ALERT or EMERG
3198 Contributions from: Andrew Cook, Brandon Philips, Christian
3199 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
3200 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
3201 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
3202 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
3203 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
3204 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
3205 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
3206 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
3207 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
3208 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
3209 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
3210 William Giokas, Zbigniew Jędrzejewski-Szmek
3212 -- Berlin, 2013-09-13
3216 * The documentation has been updated to cover the various new
3217 concepts introduced with 205.
3219 * Unit files now understand the new %v specifier which
3220 resolves to the kernel version string as returned by "uname
3223 * systemctl now supports filtering the unit list output by
3224 load state, active state and sub state, using the new
3227 * "systemctl status" will now show the results of the
3228 condition checks (like ConditionPathExists= and similar) of
3229 the last start attempts of the unit. They are also logged to
3232 * "journalctl -b" may now be used to look for boot output of a
3233 specific boot. Try "journalctl -b -1" for the previous boot,
3234 but the syntax is substantially more powerful.
3236 * "journalctl --show-cursor" has been added which prints the
3237 cursor string the last shown log line. This may then be used
3238 with the new "journalctl --after-cursor=" switch to continue
3239 browsing logs from that point on.
3241 * "journalctl --force" may now be used to force regeneration
3244 * Creation of "dead" device nodes has been moved from udev
3245 into kmod and tmpfiles. Previously, udev would read the kmod
3246 databases to pre-generate dead device nodes based on meta
3247 information contained in kernel modules, so that these would
3248 be auto-loaded on access rather then at boot. As this
3249 does not really have much to do with the exposing actual
3250 kernel devices to userspace this has always been slightly
3251 alien in the udev codebase. Following the new scheme kmod
3252 will now generate a runtime snippet for tmpfiles from the
3253 module meta information and it now is tmpfiles' job to the
3254 create the nodes. This also allows overriding access and
3255 other parameters for the nodes using the usual tmpfiles
3256 facilities. As side effect this allows us to remove the
3257 CAP_SYS_MKNOD capability bit from udevd entirely.
3259 * logind's device ACLs may now be applied to these "dead"
3260 devices nodes too, thus finally allowing managed access to
3261 devices such as /dev/snd/sequencer whithout loading the
3262 backing module right-away.
3264 * A new RPM macro has been added that may be used to apply
3265 tmpfiles configuration during package installation.
3267 * systemd-detect-virt and ConditionVirtualization= now can
3268 detect User-Mode-Linux machines (UML).
3270 * journald will now implicitly log the effective capabilities
3271 set of processes in the message metadata.
3273 * systemd-cryptsetup has gained support for TrueCrypt volumes.
3275 * The initrd interface has been simplified (more specifically,
3276 support for passing performance data via environment
3277 variables and fsck results via files in /run has been
3278 removed). These features were non-essential, and are
3279 nowadays available in a much nicer way by having systemd in
3280 the initrd serialize its state and have the hosts systemd
3281 deserialize it again.
3283 * The udev "keymap" data files and tools to apply keyboard
3284 specific mappings of scan to key codes, and force-release
3285 scan code lists have been entirely replaced by a udev
3286 "keyboard" builtin and a hwdb data file.
3288 * systemd will now honour the kernel's "quiet" command line
3289 argument also during late shutdown, resulting in a
3290 completely silent shutdown when used.
3292 * There's now an option to control the SO_REUSEPORT socket
3293 option in .socket units.
3295 * Instance units will now automatically get a per-template
3296 subslice of system.slice unless something else is explicitly
3297 configured. For example, instances of sshd@.service will now
3298 implicitly be placed in system-sshd.slice rather than
3299 system.slice as before.
3301 * Test coverage support may now be enabled at build time.
3303 Contributions from: Dave Reisner, Frederic Crozat, Harald
3304 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
3305 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
3306 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
3307 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
3308 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
3309 Giokas, Zbigniew Jędrzejewski-Szmek
3311 -- Berlin, 2013-07-23
3315 * Two new unit types have been introduced:
3317 Scope units are very similar to service units, however, are
3318 created out of pre-existing processes -- instead of PID 1
3319 forking off the processes. By using scope units it is
3320 possible for system services and applications to group their
3321 own child processes (worker processes) in a powerful way
3322 which then maybe used to organize them, or kill them
3323 together, or apply resource limits on them.
3325 Slice units may be used to partition system resources in an
3326 hierarchical fashion and then assign other units to them. By
3327 default there are now three slices: system.slice (for all
3328 system services), user.slice (for all user sessions),
3329 machine.slice (for VMs and containers).
3331 Slices and scopes have been introduced primarily in
3332 context of the work to move cgroup handling to a
3333 single-writer scheme, where only PID 1
3334 creates/removes/manages cgroups.
3336 * There's a new concept of "transient" units. In contrast to
3337 normal units these units are created via an API at runtime,
3338 not from configuration from disk. More specifically this
3339 means it is now possible to run arbitrary programs as
3340 independent services, with all execution parameters passed
3341 in via bus APIs rather than read from disk. Transient units
3342 make systemd substantially more dynamic then it ever was,
3343 and useful as a general batch manager.
3345 * logind has been updated to make use of scope and slice units
3346 for managing user sessions. As a user logs in he will get
3347 his own private slice unit, to which all sessions are added
3348 as scope units. We also added support for automatically
3349 adding an instance of user@.service for the user into the
3350 slice. Effectively logind will no longer create cgroup
3351 hierarchies on its own now, it will defer entirely to PID 1
3352 for this by means of scope, service and slice units. Since
3353 user sessions this way become entities managed by PID 1
3354 the output of "systemctl" is now a lot more comprehensive.
3356 * A new mini-daemon "systemd-machined" has been added which
3357 may be used by virtualization managers to register local
3358 VMs/containers. nspawn has been updated accordingly, and
3359 libvirt will be updated shortly. machined will collect a bit
3360 of meta information about the VMs/containers, and assign
3361 them their own scope unit (see above). The collected
3362 meta-data is then made available via the "machinectl" tool,
3363 and exposed in "ps" and similar tools. machined/machinectl
3364 is compile-time optional.
3366 * As discussed earlier, the low-level cgroup configuration
3367 options ControlGroup=, ControlGroupModify=,
3368 ControlGroupPersistent=, ControlGroupAttribute= have been
3369 removed. Please use high-level attribute settings instead as
3370 well as slice units.
3372 * A new bus call SetUnitProperties() has been added to alter
3373 various runtime parameters of a unit. This is primarily
3374 useful to alter cgroup parameters dynamically in a nice way,
3375 but will be extended later on to make more properties
3376 modifiable at runtime. systemctl gained a new set-properties
3377 command that wraps this call.
3379 * A new tool "systemd-run" has been added which can be used to
3380 run arbitrary command lines as transient services or scopes,
3381 while configuring a number of settings via the command
3382 line. This tool is currently very basic, however already
3383 very useful. We plan to extend this tool to even allow
3384 queuing of execution jobs with time triggers from the
3385 command line, similar in fashion to "at".
3387 * nspawn will now inform the user explicitly that kernels with
3388 audit enabled break containers, and suggest the user to turn
3391 * Support for detecting the IMA and AppArmor security
3392 frameworks with ConditionSecurity= has been added.
3394 * journalctl gained a new "-k" switch for showing only kernel
3395 messages, mimicking dmesg output; in addition to "--user"
3396 and "--system" switches for showing only user's own logs
3399 * systemd-delta can now show information about drop-in
3400 snippets extending unit files.
3402 * libsystemd-bus has been substantially updated but is still
3403 not available as public API.
3405 * systemd will now look for the "debug" argument on the kernel
3406 command line and enable debug logging, similar to what
3407 "systemd.log_level=debug" already did before.
3409 * "systemctl set-default", "systemctl get-default" has been
3410 added to configure the default.target symlink, which
3411 controls what to boot into by default.
3413 * "systemctl set-log-level" has been added as a convenient
3414 way to raise and lower systemd logging threshold.
3416 * "systemd-analyze plot" will now show the time the various
3417 generators needed for execution, as well as information
3418 about the unit file loading.
3420 * libsystemd-journal gained a new sd_journal_open_files() call
3421 for opening specific journal files. journactl also gained a
3422 new switch to expose this new functionality. Previously we
3423 only supported opening all files from a directory, or all
3424 files from the system, as opening individual files only is
3425 racy due to journal file rotation.
3427 * systemd gained the new DefaultEnvironment= setting in
3428 /etc/systemd/system.conf to set environment variables for
3431 * If a privileged process logs a journal message with the
3432 OBJECT_PID= field set, then journald will automatically
3433 augment this with additional OBJECT_UID=, OBJECT_GID=,
3434 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
3435 system services want to log events about specific client
3436 processes. journactl/systemctl has been updated to make use
3437 of this information if all log messages regarding a specific
3440 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
3441 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
3442 Reisner, David Coppa, David King, David Strauss, Eelco
3443 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
3444 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
3445 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
3446 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
3447 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
3448 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
3449 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
3450 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
3451 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
3452 Łukasz Stelmach, 장동준
3456 * The Python bindings gained some minimal support for the APIs
3457 exposed by libsystemd-logind.
3459 * ConditionSecurity= gained support for detecting SMACK. Since
3460 this condition already supports SELinux and AppArmor we only
3461 miss IMA for this. Patches welcome!
3463 Contributions from: Karol Lewandowski, Lennart Poettering,
3464 Zbigniew Jędrzejewski-Szmek
3468 * systemd-nspawn will now create /etc/resolv.conf if
3469 necessary, before bind-mounting the host's file onto it.
3471 * systemd-nspawn will now store meta information about a
3472 container on the container's cgroup as extended attribute
3473 fields, including the root directory.
3475 * The cgroup hierarchy has been reworked in many ways. All
3476 objects any of the components systemd creates in the cgroup
3477 tree are now suffixed. More specifically, user sessions are
3478 now placed in cgroups suffixed with ".session", users in
3479 cgroups suffixed with ".user", and nspawn containers in
3480 cgroups suffixed with ".nspawn". Furthermore, all cgroup
3481 names are now escaped in a simple scheme to avoid collision
3482 of userspace object names with kernel filenames. This work
3483 is preparation for making these objects relocatable in the
3484 cgroup tree, in order to allow easy resource partitioning of
3485 these objects without causing naming conflicts.
3487 * systemctl list-dependencies gained the new switches
3488 --plain, --reverse, --after and --before.
3490 * systemd-inhibit now shows the process name of processes that
3491 have taken an inhibitor lock.
3493 * nss-myhostname will now also resolve "localhost"
3494 implicitly. This makes /etc/hosts an optional file and
3495 nicely handles that on IPv6 ::1 maps to both "localhost" and
3498 * libsystemd-logind.so gained a new call
3499 sd_get_machine_names() to enumerate running containers and
3500 VMs (currently only supported by very new libvirt and
3501 nspawn). sd_login_monitor can now be used to watch
3502 VMs/containers coming and going.
3504 * .include is not allowed recursively anymore, and only in
3505 unit files. Usually it is better to use drop-in snippets in
3506 .d/*.conf anyway, as introduced with systemd 198.
3508 * systemd-analyze gained a new "critical-chain" command that
3509 determines the slowest chain of units run during system
3510 boot-up. It is very useful for tracking down where
3511 optimizing boot time is the most beneficial.
3513 * systemd will no longer allow manipulating service paths in
3514 the name=systemd:/system cgroup tree using ControlGroup= in
3515 units. (But is still fine with it in all other dirs.)
3517 * There's a new systemd-nspawn@.service service file that may
3518 be used to easily run nspawn containers as system
3519 services. With the container's root directory in
3520 /var/lib/container/foobar it is now sufficient to run
3521 "systemctl start systemd-nspawn@foobar.service" to boot it.
3523 * systemd-cgls gained a new parameter "--machine" to list only
3524 the processes within a certain container.
3526 * ConditionSecurity= now can check for "apparmor". We still
3527 are lacking checks for SMACK and IMA for this condition
3528 check though. Patches welcome!
3530 * A new configuration file /etc/systemd/sleep.conf has been
3531 added that may be used to configure which kernel operation
3532 systemd is supposed to execute when "suspend", "hibernate"
3533 or "hybrid-sleep" is requested. This makes the new kernel
3534 "freeze" state accessible to the user.
3536 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
3537 the passed argument if applicable.
3539 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
3540 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
3541 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
3542 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
3543 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
3544 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
3545 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
3550 * The output of 'systemctl list-jobs' got some polishing. The
3551 '--type=' argument may now be passed more than once. A new
3552 command 'systemctl list-sockets' has been added which shows
3553 a list of kernel sockets systemd is listening on with the
3554 socket units they belong to, plus the units these socket
3557 * The experimental libsystemd-bus library got substantial
3558 updates to work in conjunction with the (also experimental)
3559 kdbus kernel project. It works well enough to exchange
3560 messages with some sophistication. Note that kdbus is not
3561 ready yet, and the library is mostly an elaborate test case
3562 for now, and not installable.
3564 * systemd gained a new unit 'systemd-static-nodes.service'
3565 that generates static device nodes earlier during boot, and
3566 can run in conjunction with udev.
3568 * libsystemd-login gained a new call sd_pid_get_user_unit()
3569 to retrieve the user systemd unit a process is running
3570 in. This is useful for systems where systemd is used as
3573 * systemd-nspawn now places all containers in the new /machine
3574 top-level cgroup directory in the name=systemd
3575 hierarchy. libvirt will soon do the same, so that we get a
3576 uniform separation of /system, /user and /machine for system
3577 services, user processes and containers/virtual
3578 machines. This new cgroup hierarchy is also useful to stick
3579 stable names to specific container instances, which can be
3580 recognized later this way (this name may be controlled
3581 via systemd-nspawn's new -M switch). libsystemd-login also
3582 gained a new call sd_pid_get_machine_name() to retrieve the
3583 name of the container/VM a specific process belongs to.
3585 * bootchart can now store its data in the journal.
3587 * libsystemd-journal gained a new call
3588 sd_journal_add_conjunction() for AND expressions to the
3589 matching logic. This can be used to express more complex
3590 logical expressions.
3592 * journactl can now take multiple --unit= and --user-unit=
3595 * The cryptsetup logic now understands the "luks.key=" kernel
3596 command line switch for specifying a file to read the
3597 decryption key from. Also, if a configured key file is not
3598 found the tool will now automatically fall back to prompting
3601 * Python systemd.journal module was updated to wrap recently
3602 added functions from libsystemd-journal. The interface was
3603 changed to bring the low level interface in s.j._Reader
3604 closer to the C API, and the high level interface in
3605 s.j.Reader was updated to wrap and convert all data about
3608 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
3609 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
3610 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
3611 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
3612 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
3613 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3617 * journalctl --update-catalog now understands a new --root=
3618 option to operate on catalogs found in a different root
3621 * During shutdown after systemd has terminated all running
3622 services a final killing loop kills all remaining left-over
3623 processes. We will now print the name of these processes
3624 when we send SIGKILL to them, since this usually indicates a
3627 * If /etc/crypttab refers to password files stored on
3628 configured mount points automatic dependencies will now be
3629 generated to ensure the specific mount is established first
3630 before the key file is attempted to be read.
3632 * 'systemctl status' will now show information about the
3633 network sockets a socket unit is listening on.
3635 * 'systemctl status' will also shown information about any
3636 drop-in configuration file for units. (Drop-In configuration
3637 files in this context are files such as
3638 /etc/systemd/systemd/foobar.service.d/*.conf)
3640 * systemd-cgtop now optionally shows summed up CPU times of
3641 cgroups. Press '%' while running cgtop to switch between
3642 percentage and absolute mode. This is useful to determine
3643 which cgroups use up the most CPU time over the entire
3644 runtime of the system. systemd-cgtop has also been updated
3645 to be 'pipeable' for processing with further shell tools.
3647 * 'hostnamectl set-hostname' will now allow setting of FQDN
3650 * The formatting and parsing of time span values has been
3651 changed. The parser now understands fractional expressions
3652 such as "5.5h". The formatter will now output fractional
3653 expressions for all time spans under 1min, i.e. "5.123456s"
3654 rather than "5s 123ms 456us". For time spans under 1s
3655 millisecond values are shown, for those under 1ms
3656 microsecond values are shown. This should greatly improve
3657 all time-related output of systemd.
3659 * libsystemd-login and libsystemd-journal gained new
3660 functions for querying the poll() events mask and poll()
3661 timeout value for integration into arbitrary event
3664 * localectl gained the ability to list available X11 keymaps
3665 (models, layouts, variants, options).
3667 * 'systemd-analyze dot' gained the ability to filter for
3668 specific units via shell-style globs, to create smaller,
3669 more useful graphs. I.e. it is now possible to create simple
3670 graphs of all the dependencies between only target units, or
3671 of all units that Avahi has dependencies with.
3673 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
3674 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
3675 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
3676 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
3677 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
3678 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
3679 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
3683 * The boot-time readahead implementation for rotating media
3684 will now read the read-ahead data in multiple passes which
3685 consist of all read requests made in equidistant time
3686 intervals. This means instead of strictly reading read-ahead
3687 data in its physical order on disk we now try to find a
3688 middle ground between physical and access time order.
3690 * /etc/os-release files gained a new BUILD_ID= field for usage
3691 on operating systems that provide continuous builds of OS
3694 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
3695 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
3696 William Douglas, Zbigniew Jędrzejewski-Szmek
3700 * systemd-python gained an API exposing libsystemd-daemon.
3702 * The SMACK setup logic gained support for uploading CIPSO
3705 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
3706 ReadOnlyDirectories= and InaccessibleDirectories= has
3707 changed. The private /tmp and /var/tmp directories are now
3708 shared by all processes of a service (which means
3709 ExecStartPre= may now leave data in /tmp that ExecStart= of
3710 the same service can still access). When a service is
3711 stopped its temporary directories are immediately deleted
3712 (normal clean-up with tmpfiles is still done in addition to
3715 * By default, systemd will now set a couple of sysctl
3716 variables in the kernel: the safe sysrq options are turned
3717 on, IP route verification is turned on, and source routing
3718 disabled. The recently added hardlink and softlink
3719 protection of the kernel is turned on. These settings should
3720 be reasonably safe, and good defaults for all new systems.
3722 * The predictable network naming logic may now be turned off
3723 with a new kernel command line switch: net.ifnames=0.
3725 * A new libsystemd-bus module has been added that implements a
3726 pretty complete D-Bus client library. For details see:
3728 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
3730 * journald will now explicitly flush the journal files to disk
3731 at the latest 5min after each write. The file will then also
3732 be marked offline until the next write. This should increase
3733 reliability in case of a crash. The synchronization delay
3734 can be configured via SyncIntervalSec= in journald.conf.
3736 * There's a new remote-fs-setup.target unit that can be used
3737 to pull in specific services when at least one remote file
3738 system is to be mounted.
3740 * There are new targets timers.target and paths.target as
3741 canonical targets to pull user timer and path units in
3742 from. This complements sockets.target with a similar
3743 purpose for socket units.
3745 * libudev gained a new call udev_device_set_attribute_value()
3746 to set sysfs attributes of a device.
3748 * The udev daemon now sets the default number of worker
3749 processes executed in parallel based on the number of available
3750 CPUs instead of the amount of available RAM. This is supposed
3751 to provide a more reliable default and limit a too aggressive
3752 paralellism for setups with 1000s of devices connected.
3754 Contributions from: Auke Kok, Colin Walters, Cristian
3755 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
3756 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
3757 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
3758 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
3759 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
3760 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
3761 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
3762 Zbigniew Jędrzejewski-Szmek
3766 * Configuration of unit files may now be extended via drop-in
3767 files without having to edit/override the unit files
3768 themselves. More specifically, if the administrator wants to
3769 change one value for a service file foobar.service he can
3770 now do so by dropping in a configuration snippet into
3771 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
3772 will load all these snippets and apply them on top of the
3773 main unit configuration file, possibly extending or
3774 overriding its settings. Using these drop-in snippets is
3775 generally nicer than the two earlier options for changing
3776 unit files locally: copying the files from
3777 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
3778 them there; or creating a new file in /etc/systemd/system/
3779 that incorporates the original one via ".include". Drop-in
3780 snippets into these .d/ directories can be placed in any
3781 directory systemd looks for units in, and the usual
3782 overriding semantics between /usr/lib, /etc and /run apply
3785 * Most unit file settings which take lists of items can now be
3786 reset by assigning the empty string to them. For example,
3787 normally, settings such as Environment=FOO=BAR append a new
3788 environment variable assignment to the environment block,
3789 each time they are used. By assigning Environment= the empty
3790 string the environment block can be reset to empty. This is
3791 particularly useful with the .d/*.conf drop-in snippets
3792 mentioned above, since this adds the ability to reset list
3793 settings from vendor unit files via these drop-ins.
3795 * systemctl gained a new "list-dependencies" command for
3796 listing the dependencies of a unit recursively.
3798 * Inhibitors are now honored and listed by "systemctl
3799 suspend", "systemctl poweroff" (and similar) too, not only
3800 GNOME. These commands will also list active sessions by
3803 * Resource limits (as exposed by the various control group
3804 controllers) can now be controlled dynamically at runtime
3805 for all units. More specifically, you can now use a command
3806 like "systemctl set-cgroup-attr foobar.service cpu.shares
3807 2000" to alter the CPU shares a specific service gets. These
3808 settings are stored persistently on disk, and thus allow the
3809 administrator to easily adjust the resource usage of
3810 services with a few simple commands. This dynamic resource
3811 management logic is also available to other programs via the
3812 bus. Almost any kernel cgroup attribute and controller is
3815 * systemd-vconsole-setup will now copy all font settings to
3816 all allocated VTs, where it previously applied them only to
3819 * libsystemd-login gained the new sd_session_get_tty() API
3822 * This release drops support for a few legacy or
3823 distribution-specific LSB facility names when parsing init
3824 scripts: $x-display-manager, $mail-transfer-agent,
3825 $mail-transport-agent, $mail-transfer-agent, $smtp,
3826 $null. Also, the mail-transfer-agent.target unit backing
3827 this has been removed. Distributions which want to retain
3828 compatibility with this should carry the burden for
3829 supporting this themselves and patch support for these back
3830 in, if they really need to. Also, the facilities $syslog and
3831 $local_fs are now ignored, since systemd does not support
3832 early-boot LSB init scripts anymore, and these facilities
3833 are implied anyway for normal services. syslog.target has
3836 * There are new bus calls on PID1's Manager object for
3837 cancelling jobs, and removing snapshot units. Previously,
3838 both calls were only available on the Job and Snapshot
3841 * systemd-journal-gatewayd gained SSL support.
3843 * The various "environment" files, such as /etc/locale.conf
3844 now support continuation lines with a backslash ("\") as
3845 last character in the line, similarly in style (but different)
3846 to how this is supported in shells.
3848 * For normal user processes the _SYSTEMD_USER_UNIT= field is
3849 now implicitly appended to every log entry logged. systemctl
3850 has been updated to filter by this field when operating on a
3851 user systemd instance.
3853 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
3854 CAP_AUDIT_CONTROL capabilities to the capabilities set for
3855 the container. This makes it easier to boot unmodified
3856 Fedora systems in a container, which however still requires
3857 audit=0 to be passed on the kernel command line. Auditing in
3858 kernel and userspace is unfortunately still too broken in
3859 context of containers, hence we recommend compiling it out
3860 of the kernel or using audit=0. Hopefully this will be fixed
3861 one day for good in the kernel.
3863 * nspawn gained the new --bind= and --bind-ro= parameters to
3864 bind mount specific directories from the host into the
3867 * nspawn will now mount its own devpts file system instance
3868 into the container, in order not to leak pty devices from
3869 the host into the container.
3871 * systemd will now read the firmware boot time performance
3872 information from the EFI variables, if the used boot loader
3873 supports this, and takes it into account for boot performance
3874 analysis via "systemd-analyze". This is currently supported
3875 only in conjunction with Gummiboot, but could be supported
3876 by other boot loaders too. For details see:
3878 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
3880 * A new generator has been added that automatically mounts the
3881 EFI System Partition (ESP) to /boot, if that directory
3882 exists, is empty, and no other file system has been
3883 configured to be mounted there.
3885 * logind will now send out PrepareForSleep(false) out
3886 unconditionally, after coming back from suspend. This may be
3887 used by applications as asynchronous notification for
3888 system resume events.
3890 * "systemctl unlock-sessions" has been added, that allows
3891 unlocking the screens of all user sessions at once, similar
3892 to how "systemctl lock-sessions" already locked all users
3893 sessions. This is backed by a new D-Bus call UnlockSessions().
3895 * "loginctl seat-status" will now show the master device of a
3896 seat. (i.e. the device of a seat that needs to be around for
3897 the seat to be considered available, usually the graphics
3900 * tmpfiles gained a new "X" line type, that allows
3901 configuration of files and directories (with wildcards) that
3902 shall be excluded from automatic cleanup ("aging").
3904 * udev default rules set the device node permissions now only
3905 at "add" events, and do not change them any longer with a
3906 later "change" event.
3908 * The log messages for lid events and power/sleep keypresses
3909 now carry a message ID.
3911 * We now have a substantially larger unit test suite, but this
3912 continues to be work in progress.
3914 * udevadm hwdb gained a new --root= parameter to change the
3915 root directory to operate relative to.
3917 * logind will now issue a background sync() request to the kernel
3918 early at shutdown, so that dirty buffers are flushed to disk early
3919 instead of at the last moment, in order to optimize shutdown
3922 * A new bootctl tool has been added that is an interface for
3923 certain boot loader operations. This is currently a preview
3924 and is likely to be extended into a small mechanism daemon
3925 like timedated, localed, hostnamed, and can be used by
3926 graphical UIs to enumerate available boot options, and
3927 request boot into firmware operations.
3929 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
3930 the rest of the package. It also has been updated to work
3931 correctly in initrds.
3933 * Policykit previously has been runtime optional, and is now
3934 also compile time optional via a configure switch.
3936 * systemd-analyze has been reimplemented in C. Also "systemctl
3937 dot" has moved into systemd-analyze.
3939 * "systemctl status" with no further parameters will now print
3940 the status of all active or failed units.
3942 * Operations such as "systemctl start" can now be executed
3943 with a new mode "--irreversible" which may be used to queue
3944 operations that cannot accidentally be reversed by a later
3945 job queuing. This is by default used to make shutdown
3946 requests more robust.
3948 * The Python API of systemd now gained a new module for
3949 reading journal files.
3951 * A new tool kernel-install has been added that can install
3952 kernel images according to the Boot Loader Specification:
3954 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
3956 * Boot time console output has been improved to provide
3957 animated boot time output for hanging jobs.
3959 * A new tool systemd-activate has been added which can be used
3960 to test socket activation with, directly from the command
3961 line. This should make it much easier to test and debug
3962 socket activation in daemons.
3964 * journalctl gained a new "--reverse" (or -r) option to show
3965 journal output in reverse order (i.e. newest line first).
3967 * journalctl gained a new "--pager-end" (or -e) option to jump
3968 to immediately jump to the end of the journal in the
3969 pager. This is only supported in conjunction with "less".
3971 * journalctl gained a new "--user-unit=" option, that works
3972 similarly to "--unit=" but filters for user units rather than
3975 * A number of unit files to ease adoption of systemd in
3976 initrds has been added. This moves some minimal logic from
3977 the various initrd implementations into systemd proper.
3979 * The journal files are now owned by a new group
3980 "systemd-journal", which exists specifically to allow access
3981 to the journal, and nothing else. Previously, we used the
3982 "adm" group for that, which however possibly covers more
3983 than just journal/log file access. This new group is now
3984 already used by systemd-journal-gatewayd to ensure this
3985 daemon gets access to the journal files and as little else
3986 as possible. Note that "make install" will also set FS ACLs
3987 up for /var/log/journal to give "adm" and "wheel" read
3988 access to it, in addition to "systemd-journal" which owns
3989 the journal files. We recommend that packaging scripts also
3990 add read access to "adm" + "wheel" to /var/log/journal, and
3991 all existing/future journal files. To normal users and
3992 administrators little changes, however packagers need to
3993 ensure to create the "systemd-journal" system group at
3994 package installation time.
3996 * The systemd-journal-gatewayd now runs as unprivileged user
3997 systemd-journal-gateway:systemd-journal-gateway. Packaging
3998 scripts need to create these system user/group at
4001 * timedated now exposes a new boolean property CanNTP that
4002 indicates whether a local NTP service is available or not.
4004 * systemd-detect-virt will now also detect xen PVs
4006 * The pstore file system is now mounted by default, if it is
4009 * In addition to the SELinux and IMA policies we will now also
4010 load SMACK policies at early boot.
4012 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
4013 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
4014 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
4015 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
4016 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
4017 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
4018 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
4019 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
4020 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
4021 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
4022 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
4023 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
4024 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
4025 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
4029 * Timer units now support calendar time events in addition to
4030 monotonic time events. That means you can now trigger a unit
4031 based on a calendar time specification such as "Thu,Fri
4032 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
4033 or fifth day of any month of the year 2013, given that it is
4034 a thursday or friday. This brings timer event support
4035 considerably closer to cron's capabilities. For details on
4036 the supported calendar time specification language see
4039 * udev now supports a number of different naming policies for
4040 network interfaces for predictable names, and a combination
4041 of these policies is now the default. Please see this wiki
4042 document for details:
4044 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
4046 * Auke Kok's bootchart implementation has been added to the
4047 systemd tree. It is an optional component that can graph the
4048 boot in quite some detail. It is one of the best bootchart
4049 implementations around and minimal in its code and
4052 * nss-myhostname has been integrated into the systemd source
4053 tree. nss-myhostname guarantees that the local hostname
4054 always stays resolvable via NSS. It has been a weak
4055 requirement of systemd-hostnamed since a long time, and
4056 since its code is actually trivial we decided to just
4057 include it in systemd's source tree. It can be turned off
4058 with a configure switch.
4060 * The read-ahead logic is now capable of properly detecting
4061 whether a btrfs file system is on SSD or rotating media, in
4062 order to optimize the read-ahead scheme. Previously, it was
4063 only capable of detecting this on traditional file systems
4066 * In udev, additional device properties are now read from the
4067 IAB in addition to the OUI database. Also, Bluetooth company
4068 identities are attached to the devices as well.
4070 * In service files %U may be used as specifier that is
4071 replaced by the configured user name of the service.
4073 * nspawn may now be invoked without a controlling TTY. This
4074 makes it suitable for invocation as its own service. This
4075 may be used to set up a simple containerized server system
4076 using only core OS tools.
4078 * systemd and nspawn can now accept socket file descriptors
4079 when they are started for socket activation. This enables
4080 implementation of socket activated nspawn
4081 containers. i.e. think about autospawning an entire OS image
4082 when the first SSH or HTTP connection is received. We expect
4083 that similar functionality will also be added to libvirt-lxc
4086 * journalctl will now suppress ANSI color codes when
4087 presenting log data.
4089 * systemctl will no longer show control group information for
4090 a unit if a the control group is empty anyway.
4092 * logind can now automatically suspend/hibernate/shutdown the
4095 * /etc/machine-info and hostnamed now also expose the chassis
4096 type of the system. This can be used to determine whether
4097 the local system is a laptop, desktop, handset or
4098 tablet. This information may either be configured by the
4099 user/vendor or is automatically determined from ACPI and DMI
4100 information if possible.
4102 * A number of PolicyKit actions are now bound together with
4103 "imply" rules. This should simplify creating UIs because
4104 many actions will now authenticate similar ones as well.
4106 * Unit files learnt a new condition ConditionACPower= which
4107 may be used to conditionalize a unit depending on whether an
4108 AC power source is connected or not, of whether the system
4109 is running on battery power.
4111 * systemctl gained a new "is-failed" verb that may be used in
4112 shell scripts and suchlike to check whether a specific unit
4113 is in the "failed" state.
4115 * The EnvironmentFile= setting in unit files now supports file
4116 globbing, and can hence be used to easily read a number of
4117 environment files at once.
4119 * systemd will no longer detect and recognize specific
4120 distributions. All distribution-specific #ifdeffery has been
4121 removed, systemd is now fully generic and
4122 distribution-agnostic. Effectively, not too much is lost as
4123 a lot of the code is still accessible via explicit configure
4124 switches. However, support for some distribution specific
4125 legacy configuration file formats has been dropped. We
4126 recommend distributions to simply adopt the configuration
4127 files everybody else uses now and convert the old
4128 configuration from packaging scripts. Most distributions
4129 already did that. If that's not possible or desirable,
4130 distributions are welcome to forward port the specific
4131 pieces of code locally from the git history.
4133 * When logging a message about a unit systemd will now always
4134 log the unit name in the message meta data.
4136 * localectl will now also discover system locale data that is
4137 not stored in locale archives, but directly unpacked.
4139 * logind will no longer unconditionally use framebuffer
4140 devices as seat masters, i.e. as devices that are required
4141 to be existing before a seat is considered preset. Instead,
4142 it will now look for all devices that are tagged as
4143 "seat-master" in udev. By default framebuffer devices will
4144 be marked as such, but depending on local systems other
4145 devices might be marked as well. This may be used to
4146 integrate graphics cards using closed source drivers (such
4147 as NVidia ones) more nicely into logind. Note however, that
4148 we recommend using the open source NVidia drivers instead,
4149 and no udev rules for the closed-source drivers will be
4150 shipped from us upstream.
4152 Contributions from: Adam Williamson, Alessandro Crismani, Auke
4153 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
4154 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
4155 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
4156 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
4157 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
4158 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
4159 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
4160 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
4161 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
4162 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
4163 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
4168 * udev gained support for loading additional device properties
4169 from an indexed database that is keyed by vendor/product IDs
4170 and similar device identifiers. For the beginning this
4171 "hwdb" is populated with data from the well-known PCI and
4172 USB database, but also includes PNP, ACPI and OID data. In
4173 the longer run this indexed database shall grow into
4174 becoming the one central database for non-essential
4175 userspace device metadata. Previously, data from the PCI/USB
4176 database was only attached to select devices, since the
4177 lookup was a relatively expensive operation due to O(n) time
4178 complexity (with n being the number of entries in the
4179 database). Since this is now O(1), we decided to add in this
4180 data for all devices where this is available, by
4181 default. Note that the indexed database needs to be rebuilt
4182 when new data files are installed. To achieve this you need
4183 to update your packaging scripts to invoke "udevadm hwdb
4184 --update" after installation of hwdb data files. For
4185 RPM-based distributions we introduced the new
4186 %udev_hwdb_update macro for this purpose.
4188 * The Journal gained support for the "Message Catalog", an
4189 indexed database to link up additional information with
4190 journal entries. For further details please check:
4192 http://www.freedesktop.org/wiki/Software/systemd/catalog
4194 The indexed message catalog database also needs to be
4195 rebuilt after installation of message catalog files. Use
4196 "journalctl --update-catalog" for this. For RPM-based
4197 distributions we introduced the %journal_catalog_update
4198 macro for this purpose.
4200 * The Python Journal bindings gained support for the standard
4201 Python logging framework.
4203 * The Journal API gained new functions for checking whether
4204 the underlying file system of a journal file is capable of
4205 properly reporting file change notifications, or whether
4206 applications that want to reflect journal changes "live"
4207 need to recheck journal files continuously in appropriate
4210 * It is now possible to set the "age" field for tmpfiles
4211 entries to 0, indicating that files matching this entry
4212 shall always be removed when the directories are cleaned up.
4214 * coredumpctl gained a new "gdb" verb which invokes gdb
4215 right-away on the selected coredump.
4217 * There's now support for "hybrid sleep" on kernels that
4218 support this, in addition to "suspend" and "hibernate". Use
4219 "systemctl hybrid-sleep" to make use of this.
4221 * logind's HandleSuspendKey= setting (and related settings)
4222 now gained support for a new "lock" setting to simply
4223 request the screen lock on all local sessions, instead of
4224 actually executing a suspend or hibernation.
4226 * systemd will now mount the EFI variables file system by
4229 * Socket units now gained support for configuration of the
4230 SMACK security label.
4232 * timedatectl will now output the time of the last and next
4233 daylight saving change.
4235 * We dropped support for various legacy and distro-specific
4236 concepts, such as insserv, early-boot SysV services
4237 (i.e. those for non-standard runlevels such as 'b' or 'S')
4238 or ArchLinux /etc/rc.conf support. We recommend the
4239 distributions who still need support this to either continue
4240 to maintain the necessary patches downstream, or find a
4241 different solution. (Talk to us if you have questions!)
4243 * Various systemd components will now bypass PolicyKit checks
4244 for root and otherwise handle properly if PolicyKit is not
4245 found to be around. This should fix most issues for
4246 PolicyKit-less systems. Quite frankly this should have been
4247 this way since day one. It is absolutely our intention to
4248 make systemd work fine on PolicyKit-less systems, and we
4249 consider it a bug if something does not work as it should if
4250 PolicyKit is not around.
4252 * For embedded systems it is now possible to build udev and
4253 systemd without blkid and/or kmod support.
4255 * "systemctl switch-root" is now capable of switching root
4256 more than once. I.e. in addition to transitions from the
4257 initrd to the host OS it is now possible to transition to
4258 further OS images from the host. This is useful to implement
4259 offline updating tools.
4261 * Various other additions have been made to the RPM macros
4262 shipped with systemd. Use %udev_rules_update() after
4263 installing new udev rules files. %_udevhwdbdir,
4264 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
4265 %_sysctldir are now available which resolve to the right
4266 directories for packages to place various data files in.
4268 * journalctl gained the new --full switch (in addition to
4269 --all, to disable ellipsation for long messages.
4271 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
4272 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
4273 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
4274 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
4275 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
4276 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
4277 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
4278 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
4279 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
4283 * journalctl gained new --since= and --until= switches to
4284 filter by time. It also now supports nice filtering for
4285 units via --unit=/-u.
4287 * Type=oneshot services may use ExecReload= and do the
4290 * The journal daemon now supports time-based rotation and
4291 vacuuming, in addition to the usual disk-space based
4294 * The journal will now index the available field values for
4295 each field name. This enables clients to show pretty drop
4296 downs of available match values when filtering. The bash
4297 completion of journalctl has been updated
4298 accordingly. journalctl gained a new switch -F to list all
4299 values a certain field takes in the journal database.
4301 * More service events are now written as structured messages
4302 to the journal, and made recognizable via message IDs.
4304 * The timedated, localed and hostnamed mini-services which
4305 previously only provided support for changing time, locale
4306 and hostname settings from graphical DEs such as GNOME now
4307 also have a minimal (but very useful) text-based client
4308 utility each. This is probably the nicest way to changing
4309 these settings from the command line now, especially since
4310 it lists available options and is fully integrated with bash
4313 * There's now a new tool "systemd-coredumpctl" to list and
4314 extract coredumps from the journal.
4316 * We now install a README each in /var/log/ and
4317 /etc/rc.d/init.d explaining where the system logs and init
4318 scripts went. This hopefully should help folks who go to
4319 that dirs and look into the otherwise now empty void and
4320 scratch their heads.
4322 * When user-services are invoked (by systemd --user) the
4323 $MANAGERPID env var is set to the PID of systemd.
4325 * SIGRTMIN+24 when sent to a --user instance will now result
4326 in immediate termination of systemd.
4328 * gatewayd received numerous feature additions such as a
4329 "follow" mode, for live syncing and filtering.
4331 * browse.html now allows filtering and showing detailed
4332 information on specific entries. Keyboard navigation and
4333 mouse screen support has been added.
4335 * gatewayd/journalctl now supports HTML5/JSON
4336 Server-Sent-Events as output.
4338 * The SysV init script compatibility logic will now
4339 heuristically determine whether a script supports the
4340 "reload" verb, and only then make this available as
4343 * "systemctl status --follow" has been removed, use "journalctl
4346 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
4347 have been removed since they are hardly useful to be
4350 * And I'd like to take the opportunity to specifically mention
4351 Zbigniew for his great contributions. Zbigniew, you rock!
4353 Contributions from: Andrew Eikum, Christian Hesse, Colin
4354 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
4355 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
4356 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
4357 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
4358 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
4359 Jędrzejewski-Szmek, Сковорода Никита Андреевич
4363 * If /etc/vconsole.conf is non-existent or empty we will no
4364 longer load any console font or key map at boot by
4365 default. Instead the kernel defaults will be left
4366 intact. This is definitely the right thing to do, as no
4367 configuration should mean no configuration, and hard-coding
4368 font names that are different on all archs is probably a bad
4369 idea. Also, the kernel default key map and font should be
4370 good enough for most cases anyway, and mostly identical to
4371 the userspace fonts/key maps we previously overloaded them
4372 with. If distributions want to continue to default to a
4373 non-kernel font or key map they should ship a default
4374 /etc/vconsole.conf with the appropriate contents.
4376 Contributions from: Colin Walters, Daniel J Walsh, Dave
4377 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
4378 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
4382 * journalctl gained a new --cursor= switch to show entries
4383 starting from the specified location in the journal.
4385 * We now enforce a size limit on journal entry fields exported
4386 with "-o json" in journalctl. Fields larger than 4K will be
4387 assigned null. This can be turned off with --all.
4389 * An (optional) journal gateway daemon is now available as
4390 "systemd-journal-gatewayd.service". This service provides
4391 access to the journal via HTTP and JSON. This functionality
4392 will be used to implement live log synchronization in both
4393 pull and push modes, but has various other users too, such
4394 as easy log access for debugging of embedded devices. Right
4395 now it is already useful to retrieve the journal via HTTP:
4397 # systemctl start systemd-journal-gatewayd.service
4398 # wget http://localhost:19531/entries
4400 This will download the journal contents in a
4401 /var/log/messages compatible format. The same as JSON:
4403 # curl -H"Accept: application/json" http://localhost:19531/entries
4405 This service is also accessible via a web browser where a
4406 single static HTML5 app is served that uses the JSON logic
4407 to enable the user to do some basic browsing of the
4408 journal. This will be extended later on. Here's an example
4409 screenshot of this app in its current state:
4411 http://0pointer.de/public/journal-gatewayd
4413 Contributions from: Kay Sievers, Lennart Poettering, Robert
4414 Milasan, Tom Gundersen
4418 * The bash completion logic is now available for journalctl
4421 * We do not mount the "cpuset" controller anymore together with
4422 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
4423 started if no parameters are assigned to it. "cpuset" hence
4424 broke code that assumed it it could create "cpu" groups and
4427 * journalctl -f will now subscribe to terminal size changes,
4428 and line break accordingly.
4430 Contributions from: Dave Reisner, Kay Sievers, Lennart
4431 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
4435 * nspawn will now create a symlink /etc/localtime in the
4436 container environment, copying the host's timezone
4437 setting. Previously this has been done via a bind mount, but
4438 since symlinks cannot be bind mounted this has now been
4439 changed to create/update the appropriate symlink.
4441 * journalctl -n's line number argument is now optional, and
4442 will default to 10 if omitted.
4444 * journald will now log the maximum size the journal files may
4445 take up on disk. This is particularly useful if the default
4446 built-in logic of determining this parameter from the file
4447 system size is used. Use "systemctl status
4448 systemd-journald.service" to see this information.
4450 * The multi-seat X wrapper tool has been stripped down. As X
4451 is now capable of enumerating graphics devices via udev in a
4452 seat-aware way the wrapper is not strictly necessary
4453 anymore. A stripped down temporary stop-gap is still shipped
4454 until the upstream display managers have been updated to
4455 fully support the new X logic. Expect this wrapper to be
4456 removed entirely in one of the next releases.
4458 * HandleSleepKey= in logind.conf has been split up into
4459 HandleSuspendKey= and HandleHibernateKey=. The old setting
4460 is not available anymore. X11 and the kernel are
4461 distinguishing between these keys and we should too. This
4462 also means the inhibition lock for these keys has been split
4465 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
4466 Poettering, Lukas Nykryn, Václav Pavlín
4470 * Whenever a unit changes state we will now log this to the
4471 journal and show along the unit's own log output in
4474 * ConditionPathIsMountPoint= can now properly detect bind
4475 mount points too. (Previously, a bind mount of one file
4476 system to another place in the same file system could not be
4477 detected as mount, since they shared struct stat's st_dev
4480 * We will now mount the cgroup controllers cpu, cpuacct,
4481 cpuset and the controllers net_cls, net_prio together by
4484 * nspawn containers will now have a virtualized boot
4485 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
4486 over with a randomized ID at container initialization). This
4487 has the effect of making "journalctl -b" do the right thing
4490 * The JSON output journal serialization has been updated not
4491 to generate "endless" list objects anymore, but rather one
4492 JSON object per line. This is more in line how most JSON
4493 parsers expect JSON objects. The new output mode
4494 "json-pretty" has been added to provide similar output, but
4495 neatly aligned for readability by humans.
4497 * We dropped all explicit sync() invocations in the shutdown
4498 code. The kernel does this implicitly anyway in the kernel
4499 reboot() syscall. halt(8)'s -n option is now a compatibility
4502 * We now support virtualized reboot() in containers, as
4503 supported by newer kernels. We will fall back to exit() if
4504 CAP_SYS_REBOOT is not available to the container. Also,
4505 nspawn makes use of this now and will actually reboot the
4506 container if the containerized OS asks for that.
4508 * journalctl will only show local log output by default
4509 now. Use --merge (-m) to show remote log output, too.
4511 * libsystemd-journal gained the new sd_journal_get_usage()
4512 call to determine the current disk usage of all journal
4513 files. This is exposed in the new "journalctl --disk-usage"
4516 * journald gained a new configuration setting SplitMode= in
4517 journald.conf which may be used to control how user journals
4518 are split off. See journald.conf(5) for details.
4520 * A new condition type ConditionFileNotEmpty= has been added.
4522 * tmpfiles' "w" lines now support file globbing, to write
4523 multiple files at once.
4525 * We added Python bindings for the journal submission
4526 APIs. More Python APIs for a number of selected APIs will
4527 likely follow. Note that we intend to add native bindings
4528 only for the Python language, as we consider it common
4529 enough to deserve bindings shipped within systemd. There are
4530 various projects outside of systemd that provide bindings
4531 for languages such as PHP or Lua.
4533 * Many conditions will now resolve specifiers such as %i. In
4534 addition, PathChanged= and related directives of .path units
4535 now support specifiers as well.
4537 * There's now a new RPM macro definition for the system preset
4540 * journald will now warn if it ca not forward a message to the
4541 syslog daemon because its socket is full.
4543 * timedated will no longer write or process /etc/timezone,
4544 except on Debian. As we do not support late mounted /usr
4545 anymore /etc/localtime always being a symlink is now safe,
4546 and hence the information in /etc/timezone is not necessary
4549 * logind will now always reserve one VT for a text getty (VT6
4550 by default). Previously if more than 6 X sessions where
4551 started they took up all the VTs with auto-spawned gettys,
4552 so that no text gettys were available anymore.
4554 * udev will now automatically inform the btrfs kernel logic
4555 about btrfs RAID components showing up. This should make
4556 simple hotplug based btrfs RAID assembly work.
4558 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
4559 (but not for its children which will stay at the kernel
4560 default). This should allow setups with a lot more listening
4563 * systemd will now always pass the configured timezone to the
4564 kernel at boot. timedated will do the same when the timezone
4567 * logind's inhibition logic has been updated. By default,
4568 logind will now handle the lid switch, the power and sleep
4569 keys all the time, even in graphical sessions. If DEs want
4570 to handle these events on their own they should take the new
4571 handle-power-key, handle-sleep-key and handle-lid-switch
4572 inhibitors during their runtime. A simple way to achieve
4573 that is to invoke the DE wrapped in an invocation of:
4575 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
4577 * Access to unit operations is now checked via SELinux taking
4578 the unit file label and client process label into account.
4580 * systemd will now notify the administrator in the journal
4581 when he over-mounts a non-empty directory.
4583 * There are new specifiers that are resolved in unit files,
4584 for the host name (%H), the machine ID (%m) and the boot ID
4587 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
4588 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
4589 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
4590 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
4591 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
4592 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
4593 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
4597 * Support for reading structured kernel messages from
4598 /dev/kmsg has now been added and is enabled by default.
4600 * Support for reading kernel messages from /proc/kmsg has now
4601 been removed. If you want kernel messages in the journal
4602 make sure to run a recent kernel (>= 3.5) that supports
4603 reading structured messages from /dev/kmsg (see
4604 above). /proc/kmsg is now exclusive property of classic
4605 syslog daemons again.
4607 * The libudev API gained the new
4608 udev_device_new_from_device_id() call.
4610 * The logic for file system namespace (ReadOnlyDirectory=,
4611 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
4612 require pivot_root() anymore. This means fewer temporary
4613 directories are created below /tmp for this feature.
4615 * nspawn containers will now see and receive all submounts
4616 made on the host OS below the root file system of the
4619 * Forward Secure Sealing is now supported for Journal files,
4620 which provide cryptographical sealing of journal files so
4621 that attackers cannot alter log history anymore without this
4622 being detectable. Lennart will soon post a blog story about
4623 this explaining it in more detail.
4625 * There are two new service settings RestartPreventExitStatus=
4626 and SuccessExitStatus= which allow configuration of exit
4627 status (exit code or signal) which will be excepted from the
4628 restart logic, resp. consider successful.
4630 * journalctl gained the new --verify switch that can be used
4631 to check the integrity of the structure of journal files and
4632 (if Forward Secure Sealing is enabled) the contents of
4635 * nspawn containers will now be run with /dev/stdin, /dev/fd/
4636 and similar symlinks pre-created. This makes running shells
4637 as container init process a lot more fun.
4639 * The fstab support can now handle PARTUUID= and PARTLABEL=
4642 * A new ConditionHost= condition has been added to match
4643 against the hostname (with globs) and machine ID. This is
4644 useful for clusters where a single OS image is used to
4645 provision a large number of hosts which shall run slightly
4646 different sets of services.
4648 * Services which hit the restart limit will now be placed in a
4651 Contributions from: Bertram Poettering, Dave Reisner, Huang
4652 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
4653 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
4657 * When running in --user mode systemd will now become a
4658 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
4659 tree a lot more organized.
4661 * A new PartOf= unit dependency type has been introduced that
4662 may be used to group services in a natural way.
4664 * "systemctl enable" may now be used to enable instances of
4667 * journalctl now prints error log levels in red, and
4668 warning/notice log levels in bright white. It also supports
4669 filtering by log level now.
4671 * cgtop gained a new -n switch (similar to top), to configure
4672 the maximum number of iterations to run for. It also gained
4673 -b, to run in batch mode (accepting no input).
4675 * The suffix ".service" may now be omitted on most systemctl
4676 command lines involving service unit names.
4678 * There's a new bus call in logind to lock all sessions, as
4679 well as a loginctl verb for it "lock-sessions".
4681 * libsystemd-logind.so gained a new call sd_journal_perror()
4682 that works similar to libc perror() but logs to the journal
4683 and encodes structured information about the error number.
4685 * /etc/crypttab entries now understand the new keyfile-size=
4688 * shutdown(8) now can send a (configurable) wall message when
4689 a shutdown is cancelled.
4691 * The mount propagation mode for the root file system will now
4692 default to "shared", which is useful to make containers work
4693 nicely out-of-the-box so that they receive new mounts from
4694 the host. This can be undone locally by running "mount
4695 --make-rprivate /" if needed.
4697 * The prefdm.service file has been removed. Distributions
4698 should maintain this unit downstream if they intend to keep
4699 it around. However, we recommend writing normal unit files
4700 for display managers instead.
4702 * Since systemd is a crucial part of the OS we will now
4703 default to a number of compiler switches that improve
4704 security (hardening) such as read-only relocations, stack
4705 protection, and suchlike.
4707 * The TimeoutSec= setting for services is now split into
4708 TimeoutStartSec= and TimeoutStopSec= to allow configuration
4709 of individual time outs for the start and the stop phase of
4712 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
4713 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
4714 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
4715 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
4716 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
4717 Gundersen, Zbigniew Jędrzejewski-Szmek
4721 * The journal and id128 C APIs are now fully documented as man
4724 * Extra safety checks have been added when transitioning from
4725 the initial RAM disk to the main system to avoid accidental
4728 * /etc/crypttab entries now understand the new keyfile-offset=
4731 * systemctl -t can now be used to filter by unit load state.
4733 * The journal C API gained the new sd_journal_wait() call to
4734 make writing synchronous journal clients easier.
4736 * journalctl gained the new -D switch to show journals from a
4739 * journalctl now displays a special marker between log
4740 messages of two different boots.
4742 * The journal is now explicitly flushed to /var via a service
4743 systemd-journal-flush.service, rather than implicitly simply
4744 by seeing /var/log/journal to be writable.
4746 * journalctl (and the journal C APIs) can now match for much
4747 more complex expressions, with alternatives and
4750 * When transitioning from the initial RAM disk to the main
4751 system we will now kill all processes in a killing spree to
4752 ensure no processes stay around by accident.
4754 * Three new specifiers may be used in unit files: %u, %h, %s
4755 resolve to the user name, user home directory resp. user
4756 shell. This is useful for running systemd user instances.
4758 * We now automatically rotate journal files if their data
4759 object hash table gets a fill level > 75%. We also size the
4760 hash table based on the configured maximum file size. This
4761 together should lower hash collisions drastically and thus
4762 speed things up a bit.
4764 * journalctl gained the new "--header" switch to introspect
4765 header data of journal files.
4767 * A new setting SystemCallFilters= has been added to services
4768 which may be used to apply blacklists or whitelists to
4769 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
4771 * nspawn gained a new --link-journal= switch (and quicker: -j)
4772 to link the container journal with the host. This makes it
4773 very easy to centralize log viewing on the host for all
4774 guests while still keeping the journal files separated.
4776 * Many bugfixes and optimizations
4778 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
4779 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
4780 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
4785 * Several tools now understand kernel command line arguments,
4786 which are only read when run in an initial RAM disk. They
4787 usually follow closely their normal counterparts, but are
4790 * There's a new tool to analyze the readahead files that are
4791 automatically generated at boot. Use:
4793 /usr/lib/systemd/systemd-readahead analyze /.readahead
4795 * We now provide an early debug shell on tty9 if this enabled. Use:
4797 systemctl enable debug-shell.service
4799 * All plymouth related units have been moved into the Plymouth
4800 package. Please make sure to upgrade your Plymouth version
4803 * systemd-tmpfiles now supports getting passed the basename of
4804 a configuration file only, in which case it will look for it
4805 in all appropriate directories automatically.
4807 * udevadm info now takes a /dev or /sys path as argument, and
4808 does the right thing. Example:
4810 udevadm info /dev/sda
4811 udevadm info /sys/class/block/sda
4813 * systemctl now prints a warning if a unit is stopped but a
4814 unit that might trigger it continues to run. Example: a
4815 service is stopped but the socket that activates it is left
4818 * "systemctl status" will now mention if the log output was
4819 shortened due to rotation since a service has been started.
4821 * The journal API now exposes functions to determine the
4822 "cutoff" times due to rotation.
4824 * journald now understands SIGUSR1 and SIGUSR2 for triggering
4825 immediately flushing of runtime logs to /var if possible,
4826 resp. for triggering immediate rotation of the journal
4829 * It is now considered an error if a service is attempted to
4830 be stopped that is not loaded.
4832 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
4834 * systemd-analyze now supports Python 3
4836 * tmpfiles now supports cleaning up directories via aging
4837 where the first level dirs are always kept around but
4838 directories beneath it automatically aged. This is enabled
4839 by prefixing the age field with '~'.
4841 * Seat objects now expose CanGraphical, CanTTY properties
4842 which is required to deal with very fast bootups where the
4843 display manager might be running before the graphics drivers
4844 completed initialization.
4846 * Seat objects now expose a State property.
4848 * We now include RPM macros for service enabling/disabling
4849 based on the preset logic. We recommend RPM based
4850 distributions to make use of these macros if possible. This
4851 makes it simpler to reuse RPM spec files across
4854 * We now make sure that the collected systemd unit name is
4855 always valid when services log to the journal via
4858 * There's a new man page kernel-command-line(7) detailing all
4859 command line options we understand.
4861 * The fstab generator may now be disabled at boot by passing
4862 fstab=0 on the kernel command line.
4864 * A new kernel command line option modules-load= is now understood
4865 to load a specific kernel module statically, early at boot.
4867 * Unit names specified on the systemctl command line are now
4868 automatically escaped as needed. Also, if file system or
4869 device paths are specified they are automatically turned
4870 into the appropriate mount or device unit names. Example:
4872 systemctl status /home
4873 systemctl status /dev/sda
4875 * The SysVConsole= configuration option has been removed from
4876 system.conf parsing.
4878 * The SysV search path is no longer exported on the D-Bus
4881 * The Names= option is been removed from unit file parsing.
4883 * There's a new man page bootup(7) detailing the boot process.
4885 * Every unit and every generator we ship with systemd now
4886 comes with full documentation. The self-explanatory boot is
4889 * A couple of services gained "systemd-" prefixes in their
4890 name if they wrap systemd code, rather than only external
4891 code. Among them fsck@.service which is now
4892 systemd-fsck@.service.
4894 * The HaveWatchdog property has been removed from the D-Bus
4897 * systemd.confirm_spawn= on the kernel command line should now
4900 * There's a new man page crypttab(5) which details all options
4901 we actually understand.
4903 * systemd-nspawn gained a new --capability= switch to pass
4904 additional capabilities to the container.
4906 * timedated will now read known NTP implementation unit names
4907 from /usr/lib/systemd/ntp-units.d/*.list,
4908 systemd-timedated-ntp.target has been removed.
4910 * journalctl gained a new switch "-b" that lists log data of
4911 the current boot only.
4913 * The notify socket is in the abstract namespace again, in
4914 order to support daemons which chroot() at start-up.
4916 * There is a new Storage= configuration option for journald
4917 which allows configuration of where log data should go. This
4918 also provides a way to disable journal logging entirely, so
4919 that data collected is only forwarded to the console, the
4920 kernel log buffer or another syslog implementation.
4922 * Many bugfixes and optimizations
4924 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
4925 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
4926 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
4927 Shawn Landden, Tom Gundersen
4931 * "systemctl help <unit>" now shows the man page if one is
4934 * Several new man pages have been added.
4936 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
4937 MaxLevelConsole= can now be specified in
4938 journald.conf. These options allow reducing the amount of
4939 data stored on disk or forwarded by the log level.
4941 * TimerSlackNSec= can now be specified in system.conf for
4942 PID1. This allows system-wide power savings.
4944 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
4945 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
4950 * logind is now capable of (optionally) handling power and
4951 sleep keys as well as the lid switch.
4953 * journalctl now understands the syntax "journalctl
4954 /usr/bin/avahi-daemon" to get all log output of a specific
4957 * CapabilityBoundingSet= in system.conf now also influences
4958 the capability bound set of usermode helpers of the kernel.
4960 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
4961 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
4962 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
4963 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
4967 * Note that we skipped 139 releases here in order to set the
4968 new version to something that is greater than both udev's
4969 and systemd's most recent version number.
4971 * udev: all udev sources are merged into the systemd source tree now.
4972 All future udev development will happen in the systemd tree. It
4973 is still fully supported to use the udev daemon and tools without
4974 systemd running, like in initramfs or other init systems. Building
4975 udev though, will require the *build* of the systemd tree, but
4976 udev can be properly *run* without systemd.
4978 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
4979 should be used to create dead device nodes as workarounds for broken
4982 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
4983 no longer supported. udev_monitor_new_from_netlink() needs to be
4984 used to subscribe to events.
4986 * udev: when udevd is started by systemd, processes which are left
4987 behind by forking them off of udev rules, are unconditionally cleaned
4988 up and killed now after the event handling has finished. Services or
4989 daemons must be started as systemd services. Services can be
4990 pulled-in by udev to get started, but they can no longer be directly
4991 forked by udev rules.
4993 * udev: the daemon binary is called systemd-udevd now and installed
4994 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
4995 to adapt to that, create symlink, or rename the binary after building
4998 * libudev no longer provides these symbols:
4999 udev_monitor_from_socket()
5000 udev_queue_get_failed_list_entry()
5001 udev_get_{dev,sys,run}_path()
5002 The versions number was bumped and symbol versioning introduced.
5004 * systemd-loginctl and systemd-journalctl have been renamed
5005 to loginctl and journalctl to match systemctl.
5007 * The config files: /etc/systemd/systemd-logind.conf and
5008 /etc/systemd/systemd-journald.conf have been renamed to
5009 logind.conf and journald.conf. Package updates should rename
5010 the files to the new names on upgrade.
5012 * For almost all files the license is now LGPL2.1+, changed
5013 from the previous GPL2.0+. Exceptions are some minor stuff
5014 of udev (which will be changed to LGPL2.1 eventually, too),
5015 and the MIT licensed sd-daemon.[ch] library that is suitable
5016 to be used as drop-in files.
5018 * systemd and logind now handle system sleep states, in
5019 particular suspending and hibernating.
5021 * logind now implements a sleep/shutdown/idle inhibiting logic
5022 suitable for a variety of uses. Soonishly Lennart will blog
5023 about this in more detail.
5025 * var-run.mount and var-lock.mount are no longer provided
5026 (which prevously bind mounted these directories to their new
5027 places). Distributions which have not converted these
5028 directories to symlinks should consider stealing these files
5029 from git history and add them downstream.
5031 * We introduced the Documentation= field for units and added
5032 this to all our shipped units. This is useful to make it
5033 easier to explore the boot and the purpose of the various
5036 * All smaller setup units (such as
5037 systemd-vconsole-setup.service) now detect properly if they
5038 are run in a container and are skipped when
5039 appropriate. This guarantees an entirely noise-free boot in
5040 Linux container environments such as systemd-nspawn.
5042 * A framework for implementing offline system updates is now
5043 integrated, for details see:
5044 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
5046 * A new service type Type=idle is available now which helps us
5047 avoiding ugly interleaving of getty output and boot status
5050 * There's now a system-wide CapabilityBoundingSet= option to
5051 globally reduce the set of capabilities for the
5052 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
5053 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
5054 even CAP_NET_ADMIN system-wide for secure systems.
5056 * There are now system-wide DefaultLimitXXX= options to
5057 globally change the defaults of the various resource limits
5058 for all units started by PID 1.
5060 * Harald Hoyer's systemd test suite has been integrated into
5061 systemd which allows easy testing of systemd builds in qemu
5062 and nspawn. (This is really awesome! Ask us for details!)
5064 * The fstab parser is now implemented as generator, not inside
5067 * systemctl will now warn you if .mount units generated from
5068 /etc/fstab are out of date due to changes in fstab that
5069 have not been read by systemd yet.
5071 * systemd is now suitable for usage in initrds. Dracut has
5072 already been updated to make use of this. With this in place
5073 initrds get a slight bit faster but primarily are much
5074 easier to introspect and debug since "systemctl status" in
5075 the host system can be used to introspect initrd services,
5076 and the journal from the initrd is kept around too.
5078 * systemd-delta has been added, a tool to explore differences
5079 between user/admin configuration and vendor defaults.
5081 * PrivateTmp= now affects both /tmp and /var/tmp.
5083 * Boot time status messages are now much prettier and feature
5084 proper english language. Booting up systemd has never been
5087 * Read-ahead pack files now include the inode number of all
5088 files to pre-cache. When the inode changes the pre-caching
5089 is not attempted. This should be nicer to deal with updated
5090 packages which might result in changes of read-ahead
5093 * We now temporaritly lower the kernel's read_ahead_kb variable
5094 when collecting read-ahead data to ensure the kernel's
5095 built-in read-ahead does not add noise to our measurements
5096 of necessary blocks to pre-cache.
5098 * There's now RequiresMountsFor= to add automatic dependencies
5099 for all mounts necessary for a specific file system path.
5101 * MountAuto= and SwapAuto= have been removed from
5102 system.conf. Mounting file systems at boot has to take place
5105 * nspawn now learned a new switch --uuid= to set the machine
5106 ID on the command line.
5108 * nspawn now learned the -b switch to automatically search
5111 * vt102 is now the default TERM for serial TTYs, upgraded from
5114 * systemd-logind now works on VT-less systems.
5116 * The build tree has been reorganized. The individual
5117 components now have directories of their own.
5119 * A new condition type ConditionPathIsReadWrite= is now available.
5121 * nspawn learned the new -C switch to create cgroups for the
5122 container in other hierarchies.
5124 * We now have support for hardware watchdogs, configurable in
5127 * The scheduled shutdown logic now has a public API.
5129 * We now mount /tmp as tmpfs by default, but this can be
5130 masked and /etc/fstab can override it.
5132 * Since udisks does not make use of /media anymore we are not
5133 mounting a tmpfs on it anymore.
5135 * journalctl gained a new --local switch to only interleave
5136 locally generated journal files.
5138 * We can now load the IMA policy at boot automatically.
5140 * The GTK tools have been split off into a systemd-ui.
5142 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
5143 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
5144 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
5145 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
5146 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
5147 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
5148 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
5149 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
5150 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
5155 * This is mostly a bugfix release
5157 * Support optional initialization of the machine ID from the
5158 KVM or container configured UUID.
5160 * Support immediate reboots with "systemctl reboot -ff"
5162 * Show /etc/os-release data in systemd-analyze output
5164 * Many bugfixes for the journal, including endianness fixes and
5165 ensuring that disk space enforcement works
5167 * sd-login.h is C++ comptaible again
5169 * Extend the /etc/os-release format on request of the Debian
5172 * We now refuse non-UTF8 strings used in various configuration
5173 and unit files. This is done to ensure we do not pass invalid
5174 data over D-Bus or expose it elsewhere.
5176 * Register Mimo USB Screens as suitable for automatic seat
5179 * Read SELinux client context from journal clients in a race
5182 * Reorder configuration file lookup order. /etc now always
5183 overrides /run in order to allow the administrator to always
5184 and unconditionally override vendor supplied or
5185 automatically generated data.
5187 * The various user visible bits of the journal now have man
5188 pages. We still lack man pages for the journal API calls
5191 * We now ship all man pages in HTML format again in the
5194 Contributions from: Dave Reisner, Dirk Eibach, Frederic
5195 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
5196 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
5201 * This is mostly a bugfix release
5203 * systems lacking /etc/os-release are no longer supported.
5205 * Various functionality updates to libsystemd-login.so
5207 * Track class of PAM logins to distinguish greeters from
5210 Contributions from: Kay Sievers, Lennart Poettering, Michael
5215 * This is an important bugfix release for v41.
5217 * Building man pages is now optional which should be useful
5218 for those building systemd from git but unwilling to install
5221 * Watchdog support for supervising services is now usable. In
5222 a future release support for hardware watchdogs
5223 (i.e. /dev/watchdog) will be added building on this.
5225 * Service start rate limiting is now configurable and can be
5226 turned off per service. When a start rate limit is hit a
5227 reboot can automatically be triggered.
5229 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
5231 Contributions from: Benjamin Franzke, Bill Nottingham,
5232 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
5233 Schmidt, Michał Górny, Piotr Drąg
5237 * The systemd binary is installed /usr/lib/systemd/systemd now;
5238 An existing /sbin/init symlink needs to be adapted with the
5241 * The code that loads kernel modules has been ported to invoke
5242 libkmod directly, instead of modprobe. This means we do not
5243 support systems with module-init-tools anymore.
5245 * Watchdog support is now already useful, but still not
5248 * A new kernel command line option systemd.setenv= is
5249 understood to set system wide environment variables
5250 dynamically at boot.
5252 * We now limit the set of capabilities of systemd-journald.
5254 * We now set SIGPIPE to ignore by default, since it only is
5255 useful in shell pipelines, and has little use in general
5256 code. This can be disabled with IgnoreSIPIPE=no in unit
5259 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
5260 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
5265 * This is mostly a bugfix release
5267 * We now expose the reason why a service failed in the
5268 "Result" D-Bus property.
5270 * Rudimentary service watchdog support (will be completed over
5271 the next few releases.)
5273 * When systemd forks off in order execute some service we will
5274 now immediately changes its argv[0] to reflect which process
5275 it will execute. This is useful to minimize the time window
5276 with a generic argv[0], which makes bootcharts more useful
5278 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
5279 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
5280 Mike Kazantsev, Ray Strode
5284 * This is mostly a test release, but incorporates many
5287 * New systemd-cgtop tool to show control groups by their
5290 * Linking against libacl for ACLs is optional again. If
5291 disabled, support tracking device access for active logins
5292 goes becomes unavailable, and so does access to the user
5293 journals by the respective users.
5295 * If a group "adm" exists, journal files are automatically
5296 owned by them, thus allow members of this group full access
5297 to the system journal as well as all user journals.
5299 * The journal now stores the SELinux context of the logging
5300 client for all entries.
5302 * Add C++ inclusion guards to all public headers
5304 * New output mode "cat" in the journal to print only text
5305 messages, without any meta data like date or time.
5307 * Include tiny X server wrapper as a temporary stop-gap to
5308 teach XOrg udev display enumeration. This is used by display
5309 managers such as gdm, and will go away as soon as XOrg
5310 learned native udev hotplugging for display devices.
5312 * Add new systemd-cat tool for executing arbitrary programs
5313 with STDERR/STDOUT connected to the journal. Can also act as
5314 BSD logger replacement, and does so by default.
5316 * Optionally store all locally generated coredumps in the
5317 journal along with meta data.
5319 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
5320 writing short strings to files (for usage for /sys), and for
5321 creating symlinks, character and block device nodes.
5323 * New unit file option ControlGroupPersistent= to make cgroups
5324 persistent, following the mechanisms outlined in
5325 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
5327 * Support multiple local RTCs in a sane way
5329 * No longer monopolize IO when replaying readahead data on
5330 rotating disks, since we might starve non-file-system IO to
5331 death, since fanotify() will not see accesses done by blkid,
5334 * Do not show kernel threads in systemd-cgls anymore, unless
5335 requested with new -k switch.
5337 Contributions from: Dan Horák, Kay Sievers, Lennart
5338 Poettering, Michal Schmidt
5342 * This is mostly a test release, but incorporates many
5345 * The git repository moved to:
5346 git://anongit.freedesktop.org/systemd/systemd
5347 ssh://git.freedesktop.org/git/systemd/systemd
5349 * First release with the journal
5350 http://0pointer.de/blog/projects/the-journal.html
5352 * The journal replaces both systemd-kmsg-syslogd and
5353 systemd-stdout-bridge.
5355 * New sd_pid_get_unit() API call in libsystemd-logind
5357 * Many systemadm clean-ups
5359 * Introduce remote-fs-pre.target which is ordered before all
5360 remote mounts and may be used to start services before all
5363 * Added Mageia support
5365 * Add bash completion for systemd-loginctl
5367 * Actively monitor PID file creation for daemons which exit in
5368 the parent process before having finished writing the PID
5369 file in the daemon process. Daemons which do this need to be
5370 fixed (i.e. PID file creation must have finished before the
5371 parent exits), but we now react a bit more gracefully to them.
5373 * Add colourful boot output, mimicking the well-known output
5374 of existing distributions.
5376 * New option PassCredentials= for socket units, for
5377 compatibility with a recent kernel ABI breakage.
5379 * /etc/rc.local is now hooked in via a generator binary, and
5380 thus will no longer act as synchronization point during
5383 * systemctl list-unit-files now supports --root=.
5385 * systemd-tmpfiles now understands two new commands: z, Z for
5386 relabelling files according to the SELinux database. This is
5387 useful to apply SELinux labels to specific files in /sys,
5390 * Output of SysV services is now forwarded to both the console
5391 and the journal by default, not only just the console.
5393 * New man pages for all APIs from libsystemd-login.
5395 * The build tree got reorganized and a the build system is a
5396 lot more modular allowing embedded setups to specifically
5397 select the components of systemd they are interested in.
5399 * Support for Linux systems lacking the kernel VT subsystem is
5402 * configure's --with-rootdir= got renamed to
5403 --with-rootprefix= to follow the naming used by udev and
5406 * Unless specified otherwise we will now install to /usr instead
5407 of /usr/local by default.
5409 * Processes with '@' in argv[0][0] are now excluded from the
5410 final shut-down killing spree, following the logic explained
5412 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
5414 * All processes remaining in a service cgroup when we enter
5415 the START or START_PRE states are now killed with
5416 SIGKILL. That means it is no longer possible to spawn
5417 background processes from ExecStart= lines (which was never
5418 supported anyway, and bad style).
5420 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
5421 reloading of units together.
5423 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
5424 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
5425 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
5426 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
5427 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek