1 systemd System and Service Manager
5 * The python-systemd code has been removed from the systemd repository.
6 A new repository has been created which accommodates the code from
7 now on, and we kindly ask distributions to create a separate package
8 for this: https://github.com/systemd/python-systemd
10 * The systemd daemon will now reload its main configuration
11 (/etc/systemd/system.conf) on daemon-reload.
13 * sd-dhcp now exposes vendor specific extensions via
14 sd_dhcp_lease_get_vendor_specific().
16 * systemd-networkd gained a number of new configuration options.
18 - A new boolean configuration option for TAP devices called
19 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
20 device, thus allowing to send and receive GSO packets.
22 - A new tunnel configuration option called 'CopyDSCP='.
23 If enabled, the DSCP field of ip6 tunnels is copied into the
26 - A set of boolean bridge configuration options were added.
27 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
28 and 'UnicastFlood=' are now parsed by networkd and applied to the
29 respective bridge link device via the respective IFLA_BRPORT_*
32 - A new string configuration option to override the hostname sent
33 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
34 is true, networkd will use the configured hostname instead of the
35 system hostname when sending DHCP requests.
37 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
38 networkd will configure the IPv6 flow-label of the tunnel device
41 - The 'macvtap' virtual network devices are now supported, similar to
42 the already supported 'macvlan' devices.
44 * systemd-resolved now implements RFC5452 to improve resilience against
45 cache poisoning. Additionally, source port randomization is enabled
46 by default to further protect against DNS spoofing attacks.
48 * nss-mymachines now supports translating UIDs and GIDs of running
49 containers with user-namespaces enabled. If a container 'foo'
50 translates a host uid 'UID' to the container uid 'TUID', then
51 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
52 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
53 mapped as 'vg-foo-TGID'.
55 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
56 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov, Jan
57 Alexander Steffens (heftig), Johan Ouwerkerk, Jose Carlos Venegas Munoz,
58 Kay Sievers, Lennart Poettering, Lidong Zhong, Martin Pitt, Michael
59 Biebl, Michael Olbrich, Michal Schmidt, Mike Gilbert, Namhyung Kim, Nick
60 Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo, Susant
61 Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
62 Umut Tezduyar Lindskog, Vito Caputo, Vivenzio Pagliari, Zbigniew
69 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
70 There are no known issues with current sysfs, and udev does not need
71 or should be used to work around such bugs.
73 * udev does no longer enable USB HID power management. Several reports
74 indicate, that some devices cannot handle that setting.
76 * The udev accelerometer helper was removed. The functionality
77 is now fully included in iio-sensor-proxy. But this means,
78 older iio-sensor-proxy versions will no longer provide
79 accelerometer/orientation data with this systemd version.
80 Please upgrade iio-sensor-proxy to version 1.0.
82 * networkd gained a new configuration option IPv6PrivacyExtensions=
83 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
84 for Stateless Address") on selected networks.
86 * For the sake of fewer build-time dependencies and less code in the
87 main repository, the python bindings are about to be removed in the
88 next release. A new repository has been created which accommodates
89 the code from now on, and we kindly ask distributions to create a
90 separate package for this. The removal will take place in v223.
92 https://github.com/systemd/python-systemd
94 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
95 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
96 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
97 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
98 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
99 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
100 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
101 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
102 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
103 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
105 -- Berlin, 2015-07-07
109 * The sd-bus.h and sd-event.h APIs have now been declared
110 stable and have been added to the official interface of
111 libsystemd.so. sd-bus implements an alternative D-Bus client
112 library, that is relatively easy to use, very efficient and
113 supports both classic D-Bus as well as kdbus as transport
114 backend. sd-event is a generic event loop abstraction that
115 is built around Linux epoll, but adds features such as event
116 prioritization or efficient timer handling. Both APIs are good
117 choices for C programs looking for a bus and/or event loop
118 implementation that is minimal and does not have to be
119 portable to other kernels.
121 * kdbus support is no longer compile-time optional. It is now
122 always built-in. However, it can still be disabled at
123 runtime using the kdbus=0 kernel command line setting, and
124 that setting may be changed to default to off, by specifying
125 --disable-kdbus at build-time. Note though that the kernel
126 command line setting has no effect if the kdbus.ko kernel
127 module is not installed, in which case kdbus is (obviously)
128 also disabled. We encourage all downstream distributions to
129 begin testing kdbus by adding it to the kernel images in the
130 development distributions, and leaving kdbus support in
133 * The minimal required util-linux version has been bumped to
136 * Support for chkconfig (--enable-chkconfig) was removed in
137 favor of calling an abstraction tool
138 /lib/systemd/systemd-sysv-install. This needs to be
139 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
140 in README for details.
142 * If there's a systemd unit and a SysV init script for the
143 same service name, and the user executes "systemctl enable"
144 for it (or a related call), then this will now enable both
145 (or execute the related operation on both), not just the
148 * The libudev API documentation has been converted from gtkdoc
151 * gudev has been removed from the systemd tree, it is now an
154 * The systemd-cgtop tool learnt a new --raw switch to generate
155 "raw" (machine parsable) output.
157 * networkd's IPForwarding= .network file setting learnt the
158 new setting "kernel", which ensures that networkd does not
159 change the IP forwarding sysctl from the default kernel
162 * The systemd-logind bus API now exposes a new boolean
163 property "Docked" that reports whether logind considers the
164 system "docked", i.e. connected to a docking station or not.
166 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
167 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
168 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
169 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
170 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
171 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
172 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
173 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
174 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
175 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
176 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
177 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
178 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
179 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
180 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
181 Fink, Zbigniew Jędrzejewski-Szmek
183 -- Berlin, 2015-06-19
187 * The gudev library has been extracted into a separate repository
188 available at: https://git.gnome.org/browse/libgudev/
189 It is now managed as part of the Gnome project. Distributions
190 are recommended to pass --disable-gudev to systemd and use
191 gudev from the Gnome project instead. gudev is still included
192 in systemd, for now. It will be removed soon, though. Please
193 also see the announcement-thread on systemd-devel:
194 http://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
196 * systemd now exposes a CPUUsageNSec= property for each
197 service unit on the bus, that contains the overall consumed
198 CPU time of a service (the sum of what each process of the
199 service consumed). This value is only available if
200 CPUAccounting= is turned on for a service, and is then shown
201 in the "systemctl status" output.
203 * Support for configuring alternative mappings of the old SysV
204 runlevels to systemd targets has been removed. They are now
205 hardcoded in a way that runlevels 2, 3, 4 all map to
206 multi-user.target and 5 to graphical.target (which
207 previously was already the default behaviour).
209 * The auto-mounter logic gained support for mount point
210 expiry, using a new TimeoutIdleSec= setting in .automount
211 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
213 * The EFI System Partition (ESP) as mounted to /boot by
214 systemd-efi-boot-generator will now be unmounted
215 automatically after 2 minutes of not being used. This should
216 minimize the risk of ESP corruptions.
218 * New /etc/fstab options x-systemd.requires= and
219 x-systemd.requires-mounts-for= are now supported to express
220 additional dependencies for mounts. This is useful for
221 journalling file systems that support external journal
222 devices or overlay file systems that require underlying file
223 systems to be mounted.
225 * systemd does not support direct live-upgrades (via systemctl
226 daemon-reexec) from versions older than v44 anymore. As no
227 distribution we are aware of shipped such old versions in a
228 stable release this should not be problematic.
230 * When systemd forks off a new per-connection service instance
231 it will now set the $REMOTE_ADDR environment variable to the
232 remote IP address, and $REMOTE_PORT environment variable to
233 the remote IP port. This behaviour is similar to the
234 corresponding environment variables defined by CGI.
236 * systemd-networkd gained support for uplink failure
237 detection. The BindCarrier= option allows binding interface
238 configuration dynamically to the link sense of other
239 interfaces. This is useful to achieve behaviour like in
242 * systemd-networkd gained support for configuring the DHCP
243 client identifier to use when requesting leases.
245 * systemd-networkd now has a per-network UseNTP= option to
246 configure whether NTP server information acquired via DHCP
247 is passed on to services like systemd-timesyncd.
249 * systemd-networkd gained support for vti6 tunnels.
251 * Note that systemd-networkd manages the sysctl variable
252 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
253 it is configured for since v219. The variable controls IP
254 forwarding, and is a per-interface alternative to the global
255 /proc/sys/net/ipv[46]/ip_forward. This setting is
256 configurable in the IPForward= option, which defaults to
257 "no". This means if networkd is used for an interface it is
258 no longer sufficient to set the global sysctl option to turn
259 on IP forwarding! Instead, the .network file option
260 IPForward= needs to be turned on! Note that the
261 implementation of this behaviour was broken in v219 and has
264 * Many bonding and vxlan options are now configurable in
267 * systemd-nspawn gained a new --property= setting to set unit
268 properties for the container scope. This is useful for
269 setting resource parameters (e.g "CPUShares=500") on
270 containers started from the command line.
272 * systemd-nspawn gained a new --private-users= switch to make
273 use of user namespacing available on recent Linux kernels.
275 * systemd-nspawn may now be called as part of a shell pipeline
276 in which case the pipes used for stdin and stdout are passed
277 directly to the process invoked in the container, without
278 indirection via a pseudo tty.
280 * systemd-nspawn gained a new switch to control the UNIX
281 signal to use when killing the init process of the container
284 * systemd-nspawn gained a new --overlay= switch for mounting
285 overlay file systems into the container using the new kernel
288 * When a container image is imported via systemd-importd and
289 the host file system is not btrfs, a loopback block device
290 file is created in /var/lib/machines.raw with a btrfs file
291 system inside. It is then mounted to /var/lib/machines to
292 enable btrfs features for container management. The loopback
293 file and btrfs file system is grown as needed when container
294 images are imported via systemd-importd.
296 * systemd-machined/systemd-importd gained support for btrfs
297 quota, to enforce container disk space limits on disk. This
298 is exposed in "machinectl set-limit".
300 * systemd-importd now can import containers from local .tar,
301 .raw and .qcow2 images, and export them to .tar and .raw. It
302 can also import dkr v2 images now from the network (on top
305 * systemd-importd gained support for verifying downloaded
306 images with gpg2 (previously only gpg1 was supported).
308 * systemd-machined, systemd-logind, systemd: most bus calls
309 are now accessible to unprivileged processes via
310 PolicyKit. Also, systemd-logind will now allow users to kill
311 their own sessions without further privileges or
314 * systemd-shutdownd has been removed. This service was
315 previously responsible for implementing scheduled shutdowns
316 as exposed in /usr/bin/shutdown's time parameter. This
317 functionality has now been moved into systemd-logind and is
318 accessible via a bus interface.
320 * "systemctl reboot" gained a new switch --firmware-setup that
321 can be used to reboot into the EFI firmware setup, if that
322 is available. systemd-logind now exposes an API on the bus
323 to trigger such reboots, in case graphical desktop UIs want
324 to cover this functionality.
326 * "systemctl enable", "systemctl disable" and "systemctl mask"
327 now support a new "--now" switch. If specified the units
328 that are enabled will also be started, and the ones
329 disabled/masked also stopped.
331 * The Gummiboot EFI boot loader tool has been merged into
332 systemd, and renamed to "systemd-boot". The bootctl tool has been
333 updated to support systemd-boot.
335 * An EFI kernel stub has been added that may be used to create
336 kernel EFI binaries that contain not only the actual kernel,
337 but also an initrd, boot splash, command line and OS release
338 information. This combined binary can then be signed as a
339 single image, so that the firmware can verify it all in one
340 step. systemd-boot has special support for EFI binaries created
341 like this and can extract OS release information from them
342 and show them in the boot menu. This functionality is useful
343 to implement cryptographically verified boot schemes.
345 * Optional support has been added to systemd-fsck to pass
346 fsck's progress report to an AF_UNIX socket in the file
349 * udev will no longer create device symlinks for all block
350 devices by default. A blacklist for excluding special block
351 devices from this logic has been turned into a whitelist
352 that requires picking block devices explicitly that require
355 * A new (currently still internal) API sd-device.h has been
356 added to libsystemd. This modernized API is supposed to
357 replace libudev eventually. In fact, already much of libudev
358 is now just a wrapper around sd-device.h.
360 * A new hwdb database for storing metadata about pointing
361 stick devices has been added.
363 * systemd-tmpfiles gained support for setting file attributes
364 similar to the "chattr" tool with new 'h' and 'H' lines.
366 * systemd-journald will no longer unconditionally set the
367 btrfs NOCOW flag on new journal files. This is instead done
368 with tmpfiles snippet using the new 'h' line type. This
369 allows easy disabling of this logic, by masking the
370 journal-nocow.conf tmpfiles file.
372 * systemd-journald will now translate audit message types to
373 human readable identifiers when writing them to the
374 journal. This should improve readability of audit messages.
376 * The LUKS logic gained support for the offset= and skip=
377 options in /etc/crypttab, as previously implemented by
380 * /usr/lib/os-release gained a new optional field VARIANT= for
381 distributions that support multiple variants (such as a
382 desktop edition, a server edition, ...)
384 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
385 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
386 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
387 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
388 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
389 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
390 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
391 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
392 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
393 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
394 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
395 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
396 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
397 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
398 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
399 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
400 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
401 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
402 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
403 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
404 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
405 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
406 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
407 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
408 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
409 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
410 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
412 -- Berlin, 2015-05-22
416 * Introduce a new API "sd-hwdb.h" for querying the hardware
417 metadata database. With this minimal interface one can query
418 and enumerate the udev hwdb, decoupled from the old libudev
419 library. libudev's interface for this is now only a wrapper
420 around sd-hwdb. A new tool systemd-hwdb has been added to
421 interface with and update the database.
423 * When any of systemd's tools copies files (for example due to
424 tmpfiles' C lines) a btrfs reflink will attempted first,
425 before bytewise copying is done.
427 * systemd-nspawn gained a new --ephemeral switch. When
428 specified a btrfs snapshot is taken of the container's root
429 directory, and immediately removed when the container
430 terminates again. Thus, a container can be started whose
431 changes never alter the container's root directory, and are
432 lost on container termination. This switch can also be used
433 for starting a container off the root file system of the
434 host without affecting the host OS. This switch is only
435 available on btrfs file systems.
437 * systemd-nspawn gained a new --template= switch. It takes the
438 path to a container tree to use as template for the tree
439 specified via --directory=, should that directory be
440 missing. This allows instantiating containers dynamically,
441 on first run. This switch is only available on btrfs file
444 * When a .mount unit refers to a mount point on which multiple
445 mounts are stacked, and the .mount unit is stopped all of
446 the stacked mount points will now be unmounted until no
449 * systemd now has an explicit notion of supported and
450 unsupported unit types. Jobs enqueued for unsupported unit
451 types will now fail with an "unsupported" error code. More
452 specifically .swap, .automount and .device units are not
453 supported in containers, .busname units are not supported on
454 non-kdbus systems. .swap and .automount are also not
455 supported if their respective kernel compile time options
458 * machinectl gained support for two new "copy-from" and
459 "copy-to" commands for copying files from a running
460 container to the host or vice versa.
462 * machinectl gained support for a new "bind" command to bind
463 mount host directories into local containers. This is
464 currently only supported for nspawn containers.
466 * networkd gained support for configuring bridge forwarding
467 database entries (fdb) from .network files.
469 * A new tiny daemon "systemd-importd" has been added that can
470 download container images in tar, raw, qcow2 or dkr formats,
471 and make them available locally in /var/lib/machines, so
472 that they can run as nspawn containers. The daemon can GPG
473 verify the downloads (not supported for dkr, since it has no
474 provisions for verifying downloads). It will transparently
475 decompress bz2, xz, gzip compressed downloads if necessary,
476 and restore sparse files on disk. The daemon uses privilege
477 separation to ensure the actual download logic runs with
478 fewer privileges than the daemon itself. machinectl has
479 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
480 make the functionality of importd available to the
481 user. With this in place the Fedora and Ubuntu "Cloud"
482 images can be downloaded and booted as containers unmodified
483 (the Fedora images lack the appropriate GPG signature files
484 currently, so they cannot be verified, but this will change
485 soon, hopefully). Note that downloading images is currently
486 only fully supported on btrfs.
488 * machinectl is now able to list container images found in
489 /var/lib/machines, along with some metadata about sizes of
490 disk and similar. If the directory is located on btrfs and
491 quota is enabled, this includes quota display. A new command
492 "image-status" has been added that shows additional
493 information about images.
495 * machinectl is now able to clone container images
496 efficiently, if the underlying file system (btrfs) supports
497 it, with the new "machinectl list-images" command. It also
498 gained commands for renaming and removing images, as well as
499 marking them read-only or read-write (supported also on
500 legacy file systems).
502 * networkd gained support for collecting LLDP network
503 announcements, from hardware that supports this. This is
504 shown in networkctl output.
506 * systemd-run gained support for a new -t (--pty) switch for
507 invoking a binary on a pty whose input and output is
508 connected to the invoking terminal. This allows executing
509 processes as system services while interactively
510 communicating with them via the terminal. Most interestingly
511 this is supported across container boundaries. Invoking
512 "systemd-run -t /bin/bash" is an alternative to running a
513 full login session, the difference being that the former
514 will not register a session, nor go through the PAM session
517 * tmpfiles gained support for a new "v" line type for creating
518 btrfs subvolumes. If the underlying file system is a legacy
519 file system, this automatically degrades to creating a
520 normal directory. Among others /var/lib/machines is now
521 created like this at boot, should it be missing.
523 * The directory /var/lib/containers/ has been deprecated and
524 been replaced by /var/lib/machines. The term "machines" has
525 been used in the systemd context as generic term for both
526 VMs and containers, and hence appears more appropriate for
527 this, as the directory can also contain raw images bootable
530 * systemd-nspawn when invoked with -M but without --directory=
531 or --image= is now capable of searching for the container
532 root directory, subvolume or disk image automatically, in
533 /var/lib/machines. systemd-nspawn@.service has been updated
534 to make use of this, thus allowing it to be used for raw
537 * A new machines.target unit has been introduced that is
538 supposed to group all containers/VMs invoked as services on
539 the system. systemd-nspawn@.service has been updated to
542 * machinectl gained a new "start" command, for invoking a
543 container as a service. "machinectl start foo" is mostly
544 equivalent to "systemctl start systemd-nspawn@foo.service",
545 but handles escaping in a nicer way.
547 * systemd-nspawn will now mount most of the cgroupfs tree
548 read-only into each container, with the exception of the
549 container's own subtree in the name=systemd hierarchy.
551 * journald now sets the special FS_NOCOW file flag for its
552 journal files. This should improve performance on btrfs, by
553 avoiding heavy fragmentation when journald's write-pattern
554 is used on COW file systems. It degrades btrfs' data
555 integrity guarantees for the files to the same levels as for
556 ext3/ext4 however. This should be OK though as journald does
557 its own data integrity checks and all its objects are
558 checksummed on disk. Also, journald should handle btrfs disk
559 full events a lot more gracefully now, by processing SIGBUS
560 errors, and not relying on fallocate() anymore.
562 * When journald detects that journal files it is writing to
563 have been deleted it will immediately start new journal
566 * systemd now provides a way to store file descriptors
567 per-service in PID 1.This is useful for daemons to ensure
568 that fds they require are not lost during a daemon
569 restart. The fds are passed to the daemon on the next
570 invocation in the same way socket activation fds are
571 passed. This is now used by journald to ensure that the
572 various sockets connected to all the system's stdout/stderr
573 are not lost when journald is restarted. File descriptors
574 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
575 an extension to sd_notify(). Note that a limit is enforced
576 on the number of fds a service can store in PID 1, and it
577 defaults to 0, so that no fds may be stored, unless this is
578 explicitly turned on.
580 * The default TERM variable to use for units connected to a
581 terminal, when no other value is explicitly is set is now
582 vt220 rather than vt102. This should be fairly safe still,
583 but allows PgUp/PgDn work.
585 * The /etc/crypttab option header= as known from Debian is now
588 * "loginctl user-status" and "loginctl session-status" will
589 now show the last 10 lines of log messages of the
590 user/session following the status output. Similar,
591 "machinectl status" will show the last 10 log lines
592 associated with a virtual machine or container
593 service. (Note that this is usually not the log messages
594 done in the VM/container itself, but simply what the
595 container manager logs. For nspawn this includes all console
598 * "loginctl session-status" without further argument will now
599 show the status of the session of the caller. Similar,
600 "lock-session", "unlock-session", "activate",
601 "enable-linger", "disable-linger" may now be called without
602 session/user parameter in which case they apply to the
603 caller's session/user.
605 * An X11 session scriptlet is now shipped that uploads
606 $DISPLAY and $XAUTHORITY into the environment of the systemd
607 --user daemon if a session begins. This should improve
608 compatibility with X11 enabled applications run as systemd
611 * Generators are now subject to masking via /etc and /run, the
612 same way as unit files.
614 * networkd .network files gained support for configuring
615 per-link IPv4/IPv6 packet forwarding as well as IPv4
616 masquerading. This is by default turned on for veth links to
617 containers, as registered by systemd-nspawn. This means that
618 nspawn containers run with --network-veth will now get
619 automatic routed access to the host's networks without any
620 further configuration or setup, as long as networkd runs on
623 * systemd-nspawn gained the --port= (-p) switch to expose TCP
624 or UDP posts of a container on the host. With this in place
625 it is possible to run containers with private veth links
626 (--network-veth), and have their functionality exposed on
627 the host as if their services were running directly on the
630 * systemd-nspawn's --network-veth switch now gained a short
631 version "-n", since with the changes above it is now truly
632 useful out-of-the-box. The systemd-nspawn@.service has been
633 updated to make use of it too by default.
635 * systemd-nspawn will now maintain a per-image R/W lock, to
636 ensure that the same image is not started more than once
637 writable. (It's OK to run an image multiple times
638 simultaneously in read-only mode.)
640 * systemd-nspawn's --image= option is now capable of
641 dissecting and booting MBR and GPT disk images that contain
642 only a single active Linux partition. Previously it
643 supported only GPT disk images with proper GPT type
644 IDs. This allows running cloud images from major
645 distributions directly with systemd-nspawn, without
648 * In addition to collecting mouse dpi data in the udev
649 hardware database, there's now support for collecting angle
650 information for mouse scroll wheels. The database is
651 supposed to guarantee similar scrolling behavior on mice
652 that it knows about. There's also support for collecting
653 information about Touchpad types.
655 * udev's input_id built-in will now also collect touch screen
656 dimension data and attach it to probed devices.
658 * /etc/os-release gained support for a Distribution Privacy
661 * networkd gained support for creating "ipvlan", "gretap",
662 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
664 * systemd-tmpfiles gained support for "a" lines for setting
667 * systemd-nspawn will now mount /tmp in the container to
668 tmpfs, automatically.
670 * systemd now exposes the memory.usage_in_bytes cgroup
671 attribute and shows it for each service in the "systemctl
672 status" output, if available.
674 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
675 immediate reboot is triggered. This useful if shutdown is
676 hung and is unable to complete, to expedite the
677 operation. Note that this kind of reboot will still unmount
678 all file systems, and hence should not result in fsck being
681 * A .device unit for an optical block device will now be
682 considered active only when a medium is in the drive. Also,
683 mount units are now bound to their backing devices thus
684 triggering automatic unmounting when devices become
685 unavailable. With this in place systemd will now
686 automatically unmount left-over mounts when a CD-ROM is
687 ejected or an USB stick is yanked from the system.
689 * networkd-wait-online now has support for waiting for
690 specific interfaces only (with globbing), and for giving up
691 after a configurable timeout.
693 * networkd now exits when idle. It will be automatically
694 restarted as soon as interfaces show up, are removed or
695 change state. networkd will stay around as long as there is
696 at least one DHCP state machine or similar around, that keep
699 * networkd may now configure IPv6 link-local addressing in
700 addition to IPv4 link-local addressing.
702 * The IPv6 "token" for use in SLAAC may now be configured for
703 each .network interface in networkd.
705 * Routes configured with networkd may now be assigned a scope
708 * networkd's [Match] sections now support globbing and lists
709 of multiple space-separated matches per item.
711 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
712 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
713 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
714 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
715 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
716 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
717 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
718 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
719 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
720 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
721 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
722 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
723 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
724 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
725 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
726 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
727 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
728 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
729 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
730 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
731 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
732 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
733 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
734 Hoffmann, Zbigniew Jędrzejewski-Szmek
736 -- Berlin, 2015-02-16
740 * When querying unit file enablement status (for example via
741 "systemctl is-enabled"), a new state "indirect" is now known
742 which indicates that a unit might not be enabled itself, but
743 another unit listed in its Alias= setting might be.
745 * Similar to the various existing ConditionXYZ= settings for
746 units there are now matching AssertXYZ= settings. While
747 failing conditions cause a unit to be skipped, but its job
748 to succeed, failing assertions declared like this will cause
749 a unit start operation and its job to fail.
751 * hostnamed now knows a new chassis type "embedded".
753 * systemctl gained a new "edit" command. When used on a unit
754 file this allows extending unit files with .d/ drop-in
755 configuration snippets or editing the full file (after
756 copying it from /usr/lib to /etc). This will invoke the
757 user's editor (as configured with $EDITOR), and reload the
758 modified configuration after editing.
760 * "systemctl status" now shows the suggested enablement state
761 for a unit, as declared in the (usually vendor-supplied)
764 * nss-myhostname will now resolve the single-label host name
765 "gateway" to the locally configured default IP routing
766 gateways, ordered by their metrics. This assigns a stable
767 name to the used gateways, regardless which ones are
768 currently configured. Note that the name will only be
769 resolved after all other name sources (if nss-myhostname is
770 configured properly) and should hence not negatively impact
771 systems that use the single-label host name "gateway" in
774 * systemd-inhibit now allows filtering by mode when listing
777 * Scope and service units gained a new "Delegate" boolean
778 property, which when set allows processes running inside the
779 unit to further partition resources. This is primarily
780 useful for systemd user instances as well as container
783 * journald will now pick up audit messages directly from
784 the kernel, and log them like any other log message. The
785 audit fields are split up and fully indexed. This means that
786 journalctl in many ways is now a (nicer!) alternative to
787 ausearch, the traditional audit client. Note that this
788 implements only a minimal audit client, if you want the
789 special audit modes like reboot-on-log-overflow, please use
790 the traditional auditd instead, which can be used in
791 parallel to journald.
793 * The ConditionSecurity= unit file option now understands the
794 special string "audit" to check whether auditing is
797 * journalctl gained two new commands --vacuum-size= and
798 --vacuum-time= to delete old journal files until the
799 remaining ones take up no more the specified size on disk,
800 or are not older than the specified time.
802 * A new, native PPPoE library has been added to sd-network,
803 systemd's library of light-weight networking protocols. This
804 library will be used in a future version of networkd to
805 enable PPPoE communication without an external pppd daemon.
807 * The busctl tool now understands a new "capture" verb that
808 works similar to "monitor", but writes a packet capture
809 trace to STDOUT that can be redirected to a file which is
810 compatible with libcap's capture file format. This can then
811 be loaded in Wireshark and similar tools to inspect bus
814 * The busctl tool now understands a new "tree" verb that shows
815 the object trees of a specific service on the bus, or of all
818 * The busctl tool now understands a new "introspect" verb that
819 shows all interfaces and members of objects on the bus,
820 including their signature and values. This is particularly
821 useful to get more information about bus objects shown by
822 the new "busctl tree" command.
824 * The busctl tool now understands new verbs "call",
825 "set-property" and "get-property" for invoking bus method
826 calls, setting and getting bus object properties in a
829 * busctl gained a new --augment-creds= argument that controls
830 whether the tool shall augment credential information it
831 gets from the bus with data from /proc, in a possibly
834 * nspawn's --link-journal= switch gained two new values
835 "try-guest" and "try-host" that work like "guest" and
836 "host", but do not fail if the host has no persistent
837 journalling enabled. -j is now equivalent to
838 --link-journal=try-guest.
840 * macvlan network devices created by nspawn will now have
841 stable MAC addresses.
843 * A new SmackProcessLabel= unit setting has been added, which
844 controls the SMACK security label processes forked off by
845 the respective unit shall use.
847 * If compiled with --enable-xkbcommon, systemd-localed will
848 verify x11 keymap settings by compiling the given keymap. It
849 will spew out warnings if the compilation fails. This
850 requires libxkbcommon to be installed.
852 * When a coredump is collected a larger number of metadata
853 fields is now collected and included in the journal records
854 created for it. More specifically control group membership,
855 environment variables, memory maps, working directory,
856 chroot directory, /proc/$PID/status, and a list of open file
857 descriptors is now stored in the log entry.
859 * The udev hwdb now contains DPI information for mice. For
862 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
864 * All systemd programs that read standalone configuration
865 files in /etc now also support a corresponding series of
866 .conf.d configuration directories in /etc/, /run/,
867 /usr/local/lib/, /usr/lib/, and (if configured with
868 --enable-split-usr) /lib/. In particular, the following
869 configuration files now have corresponding configuration
870 directories: system.conf user.conf, logind.conf,
871 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
872 resolved.conf, timesyncd.conf, journal-remote.conf, and
873 journal-upload.conf. Note that distributions should use the
874 configuration directories in /usr/lib/; the directories in
875 /etc/ are reserved for the system administrator.
877 * systemd-rfkill will no longer take the rfkill device name
878 into account when storing rfkill state on disk, as the name
879 might be dynamically assigned and not stable. Instead, the
880 ID_PATH udev variable combined with the rfkill type (wlan,
881 bluetooth, ...) is used.
883 * A new service systemd-machine-id-commit.service has been
884 added. When used on systems where /etc is read-only during
885 boot, and /etc/machine-id is not initialized (but an empty
886 file), this service will copy the temporary machine ID
887 created as replacement into /etc after the system is fully
888 booted up. This is useful for systems that are freshly
889 installed with a non-initialized machine ID, but should get
890 a fixed machine ID for subsequent boots.
892 * networkd's .netdev files now provide a large set of
893 configuration parameters for VXLAN devices. Similar, the
894 bridge port cost parameter is now configurable in .network
895 files. There's also new support for configuring IP source
896 routing. networkd .link files gained support for a new
897 OriginalName= match that is useful to match against the
898 original interface name the kernel assigned. .network files
899 may include MTU= and MACAddress= fields for altering the MTU
900 and MAC address while being connected to a specific network
903 * The LUKS logic gained supported for configuring
904 UUID-specific key files. There's also new support for naming
905 LUKS device from the kernel command line, using the new
908 * Timer units may now be transiently created via the bus API
909 (this was previously already available for scope and service
910 units). In addition it is now possible to create multiple
911 transient units at the same time with a single bus call. The
912 "systemd-run" tool has been updated to make use of this for
913 running commands on a specified time, in at(1)-style.
915 * tmpfiles gained support for "t" lines, for assigning
916 extended attributes to files. Among other uses this may be
917 used to assign SMACK labels to files.
919 Contributions from: Alin Rauta, Alison Chaiken, Andrej
920 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
921 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
922 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
923 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
924 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
925 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
926 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
927 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
928 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
929 Michael Biebl, Michael Chapman, Michael Marineau, Michal
930 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
931 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
932 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
933 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
934 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
935 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
936 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
938 -- Berlin, 2014-12-10
942 * journalctl gained the new options -t/--identifier= to match
943 on the syslog identifier (aka "tag"), as well as --utc to
944 show log timestamps in the UTC timezone. journalctl now also
945 accepts -n/--lines=all to disable line capping in a pager.
947 * journalctl gained a new switch, --flush, that synchronously
948 flushes logs from /run/log/journal to /var/log/journal if
949 persistent storage is enabled. systemd-journal-flush.service
950 now waits until the operation is complete.
952 * Services can notify the manager before they start a reload
953 (by sending RELOADING=1) or shutdown (by sending
954 STOPPING=1). This allows the manager to track and show the
955 internal state of daemons and closes a race condition when
956 the process is still running but has closed its D-Bus
959 * Services with Type=oneshot do not have to have any ExecStart
962 * User units are now loaded also from
963 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
964 /run/systemd/user directory that was already previously
965 supported, but is under the control of the user.
967 * Job timeouts (i.e. time-outs on the time a job that is
968 queued stays in the run queue) can now optionally result in
969 immediate reboot or power-off actions (JobTimeoutAction= and
970 JobTimeoutRebootArgument=). This is useful on ".target"
971 units, to limit the maximum time a target remains
972 undispatched in the run queue, and to trigger an emergency
973 operation in such a case. This is now used by default to
974 turn off the system if boot-up (as defined by everything in
975 basic.target) hangs and does not complete for at least
976 15min. Also, if power-off or reboot hang for at least 30min
977 an immediate power-off/reboot operation is triggered. This
978 functionality is particularly useful to increase reliability
979 on embedded devices, but also on laptops which might
980 accidentally get powered on when carried in a backpack and
981 whose boot stays stuck in a hard disk encryption passphrase
984 * systemd-logind can be configured to also handle lid switch
985 events even when the machine is docked or multiple displays
986 are attached (HandleLidSwitchDocked= option).
988 * A helper binary and a service have been added which can be
989 used to resume from hibernation in the initramfs. A
990 generator will parse the resume= option on the kernel
991 command line to trigger resume.
993 * A user console daemon systemd-consoled has been
994 added. Currently, it is a preview, and will so far open a
995 single terminal on each session of the user marked as
996 Desktop=systemd-console.
998 * Route metrics can be specified for DHCP routes added by
1001 * The SELinux context of socket-activated services can be set
1002 from the information provided by the networking stack
1003 (SELinuxContextFromNet= option).
1005 * Userspace firmware loading support has been removed and
1006 the minimum supported kernel version is thus bumped to 3.7.
1008 * Timeout for udev workers has been increased from 1 to 3
1009 minutes, but a warning will be printed after 1 minute to
1010 help diagnose kernel modules that take a long time to load.
1012 * Udev rules can now remove tags on devices with TAG-="foobar".
1014 * systemd's readahead implementation has been removed. In many
1015 circumstances it didn't give expected benefits even for
1016 rotational disk drives and was becoming less relevant in the
1017 age of SSDs. As none of the developers has been using
1018 rotating media anymore, and nobody stepped up to actively
1019 maintain this component of systemd it has now been removed.
1021 * Swap units can use Options= to specify discard options.
1022 Discard options specified for swaps in /etc/fstab are now
1025 * Docker containers are now detected as a separate type of
1028 * The Password Agent protocol gained support for queries where
1029 the user input is shown, useful e.g. for user names.
1030 systemd-ask-password gained a new --echo option to turn that
1033 * The default sysctl.d/ snippets will now set:
1035 net.core.default_qdisc = fq_codel
1037 This selects Fair Queuing Controlled Delay as the default
1038 queuing discipline for network interfaces. fq_codel helps
1039 fight the network bufferbloat problem. It is believed to be
1040 a good default with no tuning required for most workloads.
1041 Downstream distributions may override this choice. On 10Gbit
1042 servers that do not do forwarding, "fq" may perform better.
1043 Systems without a good clocksource should use "pfifo_fast".
1045 * If kdbus is enabled during build a new option BusPolicy= is
1046 available for service units, that allows locking all service
1047 processes into a stricter bus policy, in order to limit
1048 access to various bus services, or even hide most of them
1049 from the service's view entirely.
1051 * networkctl will now show the .network and .link file
1052 networkd has applied to a specific interface.
1054 * sd-login gained a new API call sd_session_get_desktop() to
1055 query which desktop environment has been selected for a
1058 * UNIX utmp support is now compile-time optional to support
1059 legacy-free systems.
1061 * systemctl gained two new commands "add-wants" and
1062 "add-requires" for pulling in units from specific targets
1065 * If the word "rescue" is specified on the kernel command line
1066 the system will now boot into rescue mode (aka
1067 rescue.target), which was previously available only by
1068 specifying "1" or "systemd.unit=rescue.target" on the kernel
1069 command line. This new kernel command line option nicely
1070 mirrors the already existing "emergency" kernel command line
1073 * New kernel command line options mount.usr=, mount.usrflags=,
1074 mount.usrfstype= have been added that match root=, rootflags=,
1075 rootfstype= but allow mounting a specific file system to
1078 * The $NOTIFY_SOCKET is now also passed to control processes of
1079 services, not only the main process.
1081 * This version reenables support for fsck's -l switch. This
1082 means at least version v2.25 of util-linux is required for
1083 operation, otherwise dead-locks on device nodes may
1084 occur. Again: you need to update util-linux to at least
1085 v2.25 when updating systemd to v217.
1087 * The "multi-seat-x" tool has been removed from systemd, as
1088 its functionality has been integrated into X servers 1.16,
1089 and the tool is hence redundant. It is recommended to update
1090 display managers invoking this tool to simply invoke X
1091 directly from now on, again.
1093 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
1094 message flag has been added for all of systemd's PolicyKit
1095 authenticated method calls has been added. In particular
1096 this now allows optional interactive authorization via
1097 PolicyKit for many of PID1's privileged operations such as
1098 unit file enabling and disabling.
1100 * "udevadm hwdb --update" learnt a new switch "--usr" for
1101 placing the rebuilt hardware database in /usr instead of
1102 /etc. When used only hardware database entries stored in
1103 /usr will be used, and any user database entries in /etc are
1104 ignored. This functionality is useful for vendors to ship a
1105 pre-built database on systems where local configuration is
1106 unnecessary or unlikely.
1108 * Calendar time specifications in .timer units now also
1109 understand the strings "semi-annually", "quarterly" and
1110 "minutely" as shortcuts (in addition to the preexisting
1111 "anually", "hourly", ...).
1113 * systemd-tmpfiles will now correctly create files in /dev
1114 at boot which are marked for creation only at boot. It is
1115 recommended to always create static device nodes with 'c!'
1116 and 'b!', so that they are created only at boot and not
1117 overwritten at runtime.
1119 * When the watchdog logic is used for a service (WatchdogSec=)
1120 and the watchdog timeout is hit the service will now be
1121 terminated with SIGABRT (instead of just SIGTERM), in order
1122 to make sure a proper coredump and backtrace is
1123 generated. This ensures that hanging services will result in
1124 similar coredump/backtrace behaviour as services that hit a
1127 Contributions from: Andreas Henriksson, Andrei Borzenkov,
1128 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
1129 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
1130 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
1131 Herrmann, David Sommerseth, David Strauss, Emil Renner
1132 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
1133 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
1134 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
1135 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
1136 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
1137 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
1138 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
1139 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
1140 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
1141 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
1142 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
1143 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
1144 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
1145 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
1146 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
1147 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
1150 -- Berlin, 2014-10-28
1154 * timedated no longer reads NTP implementation unit names from
1155 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
1156 implementations should add a
1158 Conflicts=systemd-timesyncd.service
1160 to their unit files to take over and replace systemd's NTP
1161 default functionality.
1163 * systemd-sysusers gained a new line type "r" for configuring
1164 which UID/GID ranges to allocate system users/groups
1165 from. Lines of type "u" may now add an additional column
1166 that specifies the home directory for the system user to be
1167 created. Also, systemd-sysusers may now optionally read user
1168 information from STDIN instead of a file. This is useful for
1169 invoking it from RPM preinst scriptlets that need to create
1170 users before the first RPM file is installed since these
1171 files might need to be owned by them. A new
1172 %sysusers_create_inline RPM macro has been introduced to do
1173 just that. systemd-sysusers now updates the shadow files as
1174 well as the user/group databases, which should enhance
1175 compatibility with certain tools like grpck.
1177 * A number of bus APIs of PID 1 now optionally consult
1178 PolicyKit to permit access for otherwise unprivileged
1179 clients under certain conditions. Note that this currently
1180 doesn't support interactive authentication yet, but this is
1181 expected to be added eventually, too.
1183 * /etc/machine-info now has new fields for configuring the
1184 deployment environment of the machine, as well as the
1185 location of the machine. hostnamectl has been updated with
1186 new command to update these fields.
1188 * systemd-timesyncd has been updated to automatically acquire
1189 NTP server information from systemd-networkd, which might
1190 have been discovered via DHCP.
1192 * systemd-resolved now includes a caching DNS stub resolver
1193 and a complete LLMNR name resolution implementation. A new
1194 NSS module "nss-resolve" has been added which can be used
1195 instead of glibc's own "nss-dns" to resolve hostnames via
1196 systemd-resolved. Hostnames, addresses and arbitrary RRs may
1197 be resolved via systemd-resolved D-Bus APIs. In contrast to
1198 the glibc internal resolver systemd-resolved is aware of
1199 multi-homed system, and keeps DNS server and caches separate
1200 and per-interface. Queries are sent simultaneously on all
1201 interfaces that have DNS servers configured, in order to
1202 properly handle VPNs and local LANs which might resolve
1203 separate sets of domain names. systemd-resolved may acquire
1204 DNS server information from systemd-networkd automatically,
1205 which in turn might have discovered them via DHCP. A tool
1206 "systemd-resolve-host" has been added that may be used to
1207 query the DNS logic in resolved. systemd-resolved implements
1208 IDNA and automatically uses IDNA or UTF-8 encoding depending
1209 on whether classic DNS or LLMNR is used as transport. In the
1210 next releases we intend to add a DNSSEC and mDNS/DNS-SD
1211 implementation to systemd-resolved.
1213 * A new NSS module nss-mymachines has been added, that
1214 automatically resolves the names of all local registered
1215 containers to their respective IP addresses.
1217 * A new client tool "networkctl" for systemd-networkd has been
1218 added. It currently is entirely passive and will query
1219 networking configuration from udev, rtnetlink and networkd,
1220 and present it to the user in a very friendly
1221 way. Eventually, we hope to extend it to become a full
1222 control utility for networkd.
1224 * .socket units gained a new DeferAcceptSec= setting that
1225 controls the kernels' TCP_DEFER_ACCEPT sockopt for
1226 TCP. Similar, support for controlling TCP keep-alive
1227 settings has been added (KeepAliveTimeSec=,
1228 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
1229 turning off Nagle's algorithm on TCP has been added
1232 * logind learned a new session type "web", for use in projects
1233 like Cockpit which register web clients as PAM sessions.
1235 * timer units with at least one OnCalendar= setting will now
1236 be started only after timer-sync.target has been
1237 reached. This way they will not elapse before the system
1238 clock has been corrected by a local NTP client or
1239 similar. This is particular useful on RTC-less embedded
1240 machines, that come up with an invalid system clock.
1242 * systemd-nspawn's --network-veth= switch should now result in
1243 stable MAC addresses for both the outer and the inner side
1246 * systemd-nspawn gained a new --volatile= switch for running
1247 container instances with /etc or /var unpopulated.
1249 * The kdbus client code has been updated to use the new Linux
1250 3.17 memfd subsystem instead of the old kdbus-specific one.
1252 * systemd-networkd's DHCP client and server now support
1253 FORCERENEW. There are also new configuration options to
1254 configure the vendor client identifier and broadcast mode
1257 * systemd will no longer inform the kernel about the current
1258 timezone, as this is necessarily incorrect and racy as the
1259 kernel has no understanding of DST and similar
1260 concepts. This hence means FAT timestamps will be always
1261 considered UTC, similar to what Android is already
1262 doing. Also, when the RTC is configured to the local time
1263 (rather than UTC) systemd will never synchronize back to it,
1264 as this might confuse Windows at a later boot.
1266 * systemd-analyze gained a new command "verify" for offline
1267 validation of unit files.
1269 * systemd-networkd gained support for a couple of additional
1270 settings for bonding networking setups. Also, the metric for
1271 statically configured routes may now be configured. For
1272 network interfaces where this is appropriate the peer IP
1273 address may now be configured.
1275 * systemd-networkd's DHCP client will no longer request
1276 broadcasting by default, as this tripped up some networks.
1277 For hardware where broadcast is required the feature should
1278 be switched back on using RequestBroadcast=yes.
1280 * systemd-networkd will now set up IPv4LL addresses (when
1281 enabled) even if DHCP is configured successfully.
1283 * udev will now default to respect network device names given
1284 by the kernel when the kernel indicates that these are
1285 predictable. This behavior can be tweaked by changing
1286 NamePolicy= in the relevant .link file.
1288 * A new library systemd-terminal has been added that
1289 implements full TTY stream parsing and rendering. This
1290 library is supposed to be used later on for implementing a
1291 full userspace VT subsystem, replacing the current kernel
1294 * A new tool systemd-journal-upload has been added to push
1295 journal data to a remote system running
1296 systemd-journal-remote.
1298 * journald will no longer forward all local data to another
1299 running syslog daemon. This change has been made because
1300 rsyslog (which appears to be the most commonly used syslog
1301 implementation these days) no longer makes use of this, and
1302 instead pulls the data out of the journal on its own. Since
1303 forwarding the messages to a non-existent syslog server is
1304 more expensive than we assumed we have now turned this
1305 off. If you run a syslog server that is not a recent rsyslog
1306 version, you have to turn this option on again
1307 (ForwardToSyslog= in journald.conf).
1309 * journald now optionally supports the LZ4 compressor for
1310 larger journal fields. This compressor should perform much
1311 better than XZ which was the previous default.
1313 * machinectl now shows the IP addresses of local containers,
1314 if it knows them, plus the interface name of the container.
1316 * A new tool "systemd-escape" has been added that makes it
1317 easy to escape strings to build unit names and similar.
1319 * sd_notify() messages may now include a new ERRNO= field
1320 which is parsed and collected by systemd and shown among the
1321 "systemctl status" output for a service.
1323 * A new component "systemd-firstboot" has been added that
1324 queries the most basic systemd information (timezone,
1325 hostname, root password) interactively on first
1326 boot. Alternatively it may also be used to provision these
1327 things offline on OS images installed into directories.
1329 * The default sysctl.d/ snippets will now set
1331 net.ipv4.conf.default.promote_secondaries=1
1333 This has the benefit of no flushing secondary IP addresses
1334 when primary addresses are removed.
1336 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
1337 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
1338 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
1339 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
1340 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
1341 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
1342 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
1343 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
1344 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
1345 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
1346 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
1347 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
1348 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
1349 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
1350 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
1352 -- Berlin, 2014-08-19
1356 * A new tool systemd-sysusers has been added. This tool
1357 creates system users and groups in /etc/passwd and
1358 /etc/group, based on static declarative system user/group
1359 definitions in /usr/lib/sysusers.d/. This is useful to
1360 enable factory resets and volatile systems that boot up with
1361 an empty /etc directory, and thus need system users and
1362 groups created during early boot. systemd now also ships
1363 with two default sysusers.d/ files for the most basic
1364 users and groups systemd and the core operating system
1367 * A new tmpfiles snippet has been added that rebuilds the
1368 essential files in /etc on boot, should they be missing.
1370 * A directive for ensuring automatic clean-up of
1371 /var/cache/man/ has been removed from the default
1372 configuration. This line should now be shipped by the man
1373 implementation. The necessary change has been made to the
1374 man-db implementation. Note that you need to update your man
1375 implementation to one that ships this line, otherwise no
1376 automatic clean-up of /var/cache/man will take place.
1378 * A new condition ConditionNeedsUpdate= has been added that
1379 may conditionalize services to only run when /etc or /var
1380 are "older" than the vendor operating system resources in
1381 /usr. This is useful for reconstructing or updating /etc
1382 after an offline update of /usr or a factory reset, on the
1383 next reboot. Services that want to run once after such an
1384 update or reset should use this condition and order
1385 themselves before the new systemd-update-done.service, which
1386 will mark the two directories as fully updated. A number of
1387 service files have been added making use of this, to rebuild
1388 the udev hardware database, the journald message catalog and
1389 dynamic loader cache (ldconfig). The systemd-sysusers tool
1390 described above also makes use of this now. With this in
1391 place it is now possible to start up a minimal operating
1392 system with /etc empty cleanly. For more information on the
1393 concepts involved see this recent blog story:
1395 http://0pointer.de/blog/projects/stateless.html
1397 * A new system group "input" has been introduced, and all
1398 input device nodes get this group assigned. This is useful
1399 for system-level software to get access to input devices. It
1400 complements what is already done for "audio" and "video".
1402 * systemd-networkd learnt minimal DHCPv4 server support in
1403 addition to the existing DHCPv4 client support. It also
1404 learnt DHCPv6 client and IPv6 Router Solicitation client
1405 support. The DHCPv4 client gained support for static routes
1406 passed in from the server. Note that the [DHCPv4] section
1407 known in older systemd-networkd versions has been renamed to
1408 [DHCP] and is now also used by the DHCPv6 client. Existing
1409 .network files using settings of this section should be
1410 updated, though compatibility is maintained. Optionally, the
1411 client hostname may now be sent to the DHCP server.
1413 * networkd gained support for vxlan virtual networks as well
1414 as tun/tap and dummy devices.
1416 * networkd gained support for automatic allocation of address
1417 ranges for interfaces from a system-wide pool of
1418 addresses. This is useful for dynamically managing a large
1419 number of interfaces with a single network configuration
1420 file. In particular this is useful to easily assign
1421 appropriate IP addresses to the veth links of a large number
1422 of nspawn instances.
1424 * RPM macros for processing sysusers, sysctl and binfmt
1425 drop-in snippets at package installation time have been
1428 * The /etc/os-release file should now be placed in
1429 /usr/lib/os-release. The old location is automatically
1430 created as symlink. /usr/lib is the more appropriate
1431 location of this file, since it shall actually describe the
1432 vendor operating system shipped in /usr, and not the
1433 configuration stored in /etc.
1435 * .mount units gained a new boolean SloppyOptions= setting
1436 that maps to mount(8)'s -s option which enables permissive
1437 parsing of unknown mount options.
1439 * tmpfiles learnt a new "L+" directive which creates a symlink
1440 but (unlike "L") deletes a pre-existing file first, should
1441 it already exist and not already be the correct
1442 symlink. Similar, "b+", "c+" and "p+" directives have been
1443 added as well, which create block and character devices, as
1444 well as fifos in the filesystem, possibly removing any
1445 pre-existing files of different types.
1447 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
1448 'argument' field (which so far specified the source to
1449 symlink/copy the files from) is now optional. If omitted the
1450 same file os copied from /usr/share/factory/ suffixed by the
1451 full destination path. This is useful for populating /etc
1452 with essential files, by copying them from vendor defaults
1453 shipped in /usr/share/factory/etc.
1455 * A new command "systemctl preset-all" has been added that
1456 applies the service preset settings to all installed unit
1457 files. A new switch --preset-mode= has been added that
1458 controls whether only enable or only disable operations
1461 * A new command "systemctl is-system-running" has been added
1462 that allows checking the overall state of the system, for
1463 example whether it is fully up and running.
1465 * When the system boots up with an empty /etc, the equivalent
1466 to "systemctl preset-all" is executed during early boot, to
1467 make sure all default services are enabled after a factory
1470 * systemd now contains a minimal preset file that enables the
1471 most basic services systemd ships by default.
1473 * Unit files' [Install] section gained a new DefaultInstance=
1474 field for defining the default instance to create if a
1475 template unit is enabled with no instance specified.
1477 * A new passive target cryptsetup-pre.target has been added
1478 that may be used by services that need to make they run and
1479 finish before the first LUKS cryptographic device is set up.
1481 * The /dev/loop-control and /dev/btrfs-control device nodes
1482 are now owned by the "disk" group by default, opening up
1483 access to this group.
1485 * systemd-coredump will now automatically generate a
1486 stack trace of all core dumps taking place on the system,
1487 based on elfutils' libdw library. This stack trace is logged
1490 * systemd-coredump may now optionally store coredumps directly
1491 on disk (in /var/lib/systemd/coredump, possibly compressed),
1492 instead of storing them unconditionally in the journal. This
1493 mode is the new default. A new configuration file
1494 /etc/systemd/coredump.conf has been added to configure this
1495 and other parameters of systemd-coredump.
1497 * coredumpctl gained a new "info" verb to show details about a
1498 specific coredump. A new switch "-1" has also been added
1499 that makes sure to only show information about the most
1500 recent entry instead of all entries. Also, as the tool is
1501 generally useful now the "systemd-" prefix of the binary
1502 name has been removed. Distributions that want to maintain
1503 compatibility with the old name should add a symlink from
1504 the old name to the new name.
1506 * journald's SplitMode= now defaults to "uid". This makes sure
1507 that unprivileged users can access their own coredumps with
1508 coredumpctl without restrictions.
1510 * New kernel command line options "systemd.wants=" (for
1511 pulling an additional unit during boot), "systemd.mask="
1512 (for masking a specific unit for the boot), and
1513 "systemd.debug-shell" (for enabling the debug shell on tty9)
1514 have been added. This is implemented in the new generator
1515 "systemd-debug-generator".
1517 * systemd-nspawn will now by default filter a couple of
1518 syscalls for containers, among them those required for
1519 kernel module loading, direct x86 IO port access, swap
1520 management, and kexec. Most importantly though
1521 open_by_handle_at() is now prohibited for containers,
1522 closing a hole similar to a recently discussed vulnerability
1523 in docker regarding access to files on file hierarchies the
1524 container should normally not have access to. Note that for
1525 nspawn we generally make no security claims anyway (and
1526 this is explicitly documented in the man page), so this is
1527 just a fix for one of the most obvious problems.
1529 * A new man page file-hierarchy(7) has been added that
1530 contains a minimized, modernized version of the file system
1531 layout systemd expects, similar in style to the FHS
1532 specification or hier(5). A new tool systemd-path(1) has
1533 been added to query many of these paths for the local
1536 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
1537 longer done. Since the directory now has a per-user size
1538 limit, and is cleaned on logout this appears unnecessary,
1539 in particular since this now brings the lifecycle of this
1540 directory closer in line with how IPC objects are handled.
1542 * systemd.pc now exports a number of additional directories,
1543 including $libdir (which is useful to identify the library
1544 path for the primary architecture of the system), and a
1545 couple of drop-in directories.
1547 * udev's predictable network interface names now use the dev_port
1548 sysfs attribute, introduced in linux 3.15 instead of dev_id to
1549 distinguish between ports of the same PCI function. dev_id should
1550 only be used for ports using the same HW address, hence the need
1553 * machined has been updated to export the OS version of a
1554 container (read from /etc/os-release and
1555 /usr/lib/os-release) on the bus. This is now shown in
1556 "machinectl status" for a machine.
1558 * A new service setting RestartForceExitStatus= has been
1559 added. If configured to a set of exit signals or process
1560 return values, the service will be restarted when the main
1561 daemon process exits with any of them, regardless of the
1564 * systemctl's -H switch for connecting to remote systemd
1565 machines has been extended so that it may be used to
1566 directly connect to a specific container on the
1567 host. "systemctl -H root@foobar:waldi" will now connect as
1568 user "root" to host "foobar", and then proceed directly to
1569 the container named "waldi". Note that currently you have to
1570 authenticate as user "root" for this to work, as entering
1571 containers is a privileged operation.
1573 Contributions from: Andreas Henriksson, Benjamin Steinwender,
1574 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
1575 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
1576 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
1577 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
1578 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
1579 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
1580 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
1581 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
1582 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
1583 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
1584 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
1586 -- Berlin, 2014-07-03
1590 * As an experimental feature, udev now tries to lock the
1591 disk device node (flock(LOCK_SH|LOCK_NB)) while it
1592 executes events for the disk or any of its partitions.
1593 Applications like partitioning programs can lock the
1594 disk device node (flock(LOCK_EX)) and claim temporary
1595 device ownership that way; udev will entirely skip all event
1596 handling for this disk and its partitions. If the disk
1597 was opened for writing, the close will trigger a partition
1598 table rescan in udev's "watch" facility, and if needed
1599 synthesize "change" events for the disk and all its partitions.
1600 This is now unconditionally enabled, and if it turns out to
1601 cause major problems, we might turn it on only for specific
1602 devices, or might need to disable it entirely. Device Mapper
1603 devices are excluded from this logic.
1605 * We temporarily dropped the "-l" switch for fsck invocations,
1606 since they collide with the flock() logic above. util-linux
1607 upstream has been changed already to avoid this conflict,
1608 and we will readd "-l" as soon as util-linux with this
1609 change has been released.
1611 * The dependency on libattr has been removed. Since a long
1612 time, the extended attribute calls have moved to glibc, and
1613 libattr is thus unnecessary.
1615 * Virtualization detection works without priviliges now. This
1616 means the systemd-detect-virt binary no longer requires
1617 CAP_SYS_PTRACE file capabilities, and our daemons can run
1618 with fewer privileges.
1620 * systemd-networkd now runs under its own "systemd-network"
1621 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
1622 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
1623 loses the ability to write to files owned by root this way.
1625 * Similar, systemd-resolved now runs under its own
1626 "systemd-resolve" user with no capabilities remaining.
1628 * Similar, systemd-bus-proxyd now runs under its own
1629 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
1631 * systemd-networkd gained support for setting up "veth"
1632 virtual ethernet devices for container connectivity, as well
1633 as GRE and VTI tunnels.
1635 * systemd-networkd will no longer automatically attempt to
1636 manually load kernel modules necessary for certain tunnel
1637 transports. Instead, it is assumed the kernel loads them
1638 automatically when required. This only works correctly on
1639 very new kernels. On older kernels, please consider adding
1640 the kernel modules to /etc/modules-load.d/ as a work-around.
1642 * The resolv.conf file systemd-resolved generates has been
1643 moved to /run/systemd/resolve/. If you have a symlink from
1644 /etc/resolv.conf, it might be necessary to correct it.
1646 * Two new service settings, ProtectHome= and ProtectSystem=,
1647 have been added. When enabled, they will make the user data
1648 (such as /home) inaccessible or read-only and the system
1649 (such as /usr) read-only, for specific services. This allows
1650 very light-weight per-service sandboxing to avoid
1651 modifications of user data or system files from
1652 services. These two new switches have been enabled for all
1653 of systemd's long-running services, where appropriate.
1655 * Socket units gained new SocketUser= and SocketGroup=
1656 settings to set the owner user and group of AF_UNIX sockets
1657 and FIFOs in the file system.
1659 * Socket units gained a new RemoveOnStop= setting. If enabled,
1660 all FIFOS and sockets in the file system will be removed
1661 when the specific socket unit is stopped.
1663 * Socket units gained a new Symlinks= setting. It takes a list
1664 of symlinks to create to file system sockets or FIFOs
1665 created by the specific Unix sockets. This is useful to
1666 manage symlinks to socket nodes with the same life-cycle as
1669 * The /dev/log socket and /dev/initctl FIFO have been moved to
1670 /run, and have been replaced by symlinks. This allows
1671 connecting to these facilities even if PrivateDevices=yes is
1672 used for a service (which makes /dev/log itself unavailable,
1673 but /run is left). This also has the benefit of ensuring
1674 that /dev only contains device nodes, directories and
1675 symlinks, and nothing else.
1677 * sd-daemon gained two new calls sd_pid_notify() and
1678 sd_pid_notifyf(). They are similar to sd_notify() and
1679 sd_notifyf(), but allow overriding of the source PID of
1680 notification messages if permissions permit this. This is
1681 useful to send notify messages on behalf of a different
1682 process (for example, the parent process). The
1683 systemd-notify tool has been updated to make use of this
1684 when sending messages (so that notification messages now
1685 originate from the shell script invoking systemd-notify and
1686 not the systemd-notify process itself. This should minimize
1687 a race where systemd fails to associate notification
1688 messages to services when the originating process already
1691 * A new "on-abnormal" setting for Restart= has been added. If
1692 set, it will result in automatic restarts on all "abnormal"
1693 reasons for a process to exit, which includes unclean
1694 signals, core dumps, timeouts and watchdog timeouts, but
1695 does not include clean and unclean exit codes or clean
1696 signals. Restart=on-abnormal is an alternative for
1697 Restart=on-failure for services that shall be able to
1698 terminate and avoid restarts on certain errors, by
1699 indicating so with an unclean exit code. Restart=on-failure
1700 or Restart=on-abnormal is now the recommended setting for
1701 all long-running services.
1703 * If the InaccessibleDirectories= service setting points to a
1704 mount point (or if there are any submounts contained within
1705 it), it is now attempted to completely unmount it, to make
1706 the file systems truly unavailable for the respective
1709 * The ReadOnlyDirectories= service setting and
1710 systemd-nspawn's --read-only parameter are now recursively
1711 applied to all submounts, too.
1713 * Mount units may now be created transiently via the bus APIs.
1715 * The support for SysV and LSB init scripts has been removed
1716 from the systemd daemon itself. Instead, it is now
1717 implemented as a generator that creates native systemd units
1718 from these scripts when needed. This enables us to remove a
1719 substantial amount of legacy code from PID 1, following the
1720 fact that many distributions only ship a very small number
1721 of LSB/SysV init scripts nowadays.
1723 * Privileged Xen (dom0) domains are not considered
1724 virtualization anymore by the virtualization detection
1725 logic. After all, they generally have unrestricted access to
1726 the hardware and usually are used to manage the unprivileged
1729 * systemd-tmpfiles gained a new "C" line type, for copying
1730 files or entire directories.
1732 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
1733 lines. So far, they have been non-globbing versions of the
1734 latter, and have thus been redundant. In future, it is
1735 recommended to only use "z". "m" has hence been removed
1736 from the documentation, even though it stays supported.
1738 * A tmpfiles snippet to recreate the most basic structure in
1739 /var has been added. This is enough to create the /var/run →
1740 /run symlink and create a couple of structural
1741 directories. This allows systems to boot up with an empty or
1742 volatile /var. Of course, while with this change, the core OS
1743 now is capable with dealing with a volatile /var, not all
1744 user services are ready for it. However, we hope that sooner
1745 or later, many service daemons will be changed upstream so
1746 that they are able to automatically create their necessary
1747 directories in /var at boot, should they be missing. This is
1748 the first step to allow state-less systems that only require
1749 the vendor image for /usr to boot.
1751 * systemd-nspawn has gained a new --tmpfs= switch to mount an
1752 empty tmpfs instance to a specific directory. This is
1753 particularly useful for making use of the automatic
1754 reconstruction of /var (see above), by passing --tmpfs=/var.
1756 * Access modes specified in tmpfiles snippets may now be
1757 prefixed with "~", which indicates that they shall be masked
1758 by whether the existing file or directory is currently
1759 writable, readable or executable at all. Also, if specified,
1760 the sgid/suid/sticky bits will be masked for all
1763 * A new passive target unit "network-pre.target" has been
1764 added which is useful for services that shall run before any
1765 network is configured, for example firewall scripts.
1767 * The "floppy" group that previously owned the /dev/fd*
1768 devices is no longer used. The "disk" group is now used
1769 instead. Distributions should probably deprecate usage of
1772 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
1773 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
1774 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
1775 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
1776 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
1777 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
1780 -- Berlin, 2014-06-11
1784 * A new "systemd-timesyncd" daemon has been added for
1785 synchronizing the system clock across the network. It
1786 implements an SNTP client. In contrast to NTP
1787 implementations such as chrony or the NTP reference server,
1788 this only implements a client side, and does not bother with
1789 the full NTP complexity, focusing only on querying time from
1790 one remote server and synchronizing the local clock to
1791 it. Unless you intend to serve NTP to networked clients or
1792 want to connect to local hardware clocks, this simple NTP
1793 client should be more than appropriate for most
1794 installations. The daemon runs with minimal privileges, and
1795 has been hooked up with networkd to only operate when
1796 network connectivity is available. The daemon saves the
1797 current clock to disk every time a new NTP sync has been
1798 acquired, and uses this to possibly correct the system clock
1799 early at bootup, in order to accommodate for systems that
1800 lack an RTC such as the Raspberry Pi and embedded devices,
1801 and to make sure that time monotonically progresses on these
1802 systems, even if it is not always correct. To make use of
1803 this daemon, a new system user and group "systemd-timesync"
1804 needs to be created on installation of systemd.
1806 * The queue "seqnum" interface of libudev has been disabled, as
1807 it was generally incompatible with device namespacing as
1808 sequence numbers of devices go "missing" if the devices are
1809 part of a different namespace.
1811 * "systemctl list-timers" and "systemctl list-sockets" gained
1812 a --recursive switch for showing units of these types also
1813 for all local containers, similar in style to the already
1814 supported --recursive switch for "systemctl list-units".
1816 * A new RebootArgument= setting has been added for service
1817 units, which may be used to specify a kernel reboot argument
1818 to use when triggering reboots with StartLimitAction=.
1820 * A new FailureAction= setting has been added for service
1821 units which may be used to specify an operation to trigger
1822 when a service fails. This works similarly to
1823 StartLimitAction=, but unlike it, controls what is done
1824 immediately rather than only after several attempts to
1825 restart the service in question.
1827 * hostnamed got updated to also expose the kernel name,
1828 release, and version on the bus. This is useful for
1829 executing commands like hostnamectl with the -H switch.
1830 systemd-analyze makes use of this to properly display
1831 details when running non-locally.
1833 * The bootchart tool can now show cgroup information in the
1834 graphs it generates.
1836 * The CFS CPU quota cgroup attribute is now exposed for
1837 services. The new CPUQuota= switch has been added for this
1838 which takes a percentage value. Setting this will have the
1839 result that a service may never get more CPU time than the
1840 specified percentage, even if the machine is otherwise idle.
1842 * systemd-networkd learned IPIP and SIT tunnel support.
1844 * LSB init scripts exposing a dependency on $network will now
1845 get a dependency on network-online.target rather than simply
1846 network.target. This should bring LSB handling closer to
1847 what it was on SysV systems.
1849 * A new fsck.repair= kernel option has been added to control
1850 how fsck shall deal with unclean file systems at boot.
1852 * The (.ini) configuration file parser will now silently
1853 ignore sections whose name begins with "X-". This may be
1854 used to maintain application-specific extension sections in unit
1857 * machined gained a new API to query the IP addresses of
1858 registered containers. "machinectl status" has been updated
1859 to show these addresses in its output.
1861 * A new call sd_uid_get_display() has been added to the
1862 sd-login APIs for querying the "primary" session of a
1863 user. The "primary" session of the user is elected from the
1864 user's sessions and generally a graphical session is
1865 preferred over a text one.
1867 * A minimal systemd-resolved daemon has been added. It
1868 currently simply acts as a companion to systemd-networkd and
1869 manages resolv.conf based on per-interface DNS
1870 configuration, possibly supplied via DHCP. In the long run
1871 we hope to extend this into a local DNSSEC enabled DNS and
1874 * The systemd-networkd-wait-online tool is now enabled by
1875 default. It will delay network-online.target until a network
1876 connection has been configured. The tool primarily integrates
1877 with networkd, but will also make a best effort to make sense
1878 of network configuration performed in some other way.
1880 * Two new service options StartupCPUShares= and
1881 StartupBlockIOWeight= have been added that work similarly to
1882 CPUShares= and BlockIOWeight= however only apply during
1883 system startup. This is useful to prioritize certain services
1884 differently during bootup than during normal runtime.
1886 * hostnamed has been changed to prefer the statically
1887 configured hostname in /etc/hostname (unless set to
1888 'localhost' or empty) over any dynamic one supplied by
1889 dhcp. With this change, the rules for picking the hostname
1890 match more closely the rules of other configuration settings
1891 where the local administrator's configuration in /etc always
1892 overrides any other settings.
1894 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
1895 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
1896 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
1897 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
1898 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
1899 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
1900 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
1901 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
1902 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
1903 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
1904 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
1905 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
1906 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
1907 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
1908 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
1909 Lindskog, WaLyong Cho, Will Woods, Zbigniew
1912 -- Beijing, 2014-05-28
1916 * When restoring the screen brightness at boot, stay away from
1917 the darkest setting or from the lowest 5% of the available
1918 range, depending on which is the larger value of both. This
1919 should effectively protect the user from rebooting into a
1920 black screen, should the brightness have been set to minimum
1923 * sd-login gained a new sd_machine_get_class() call to
1924 determine the class ("vm" or "container") of a machine
1925 registered with machined.
1927 * sd-login gained new calls
1928 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
1929 to query the identity of the peer of a local AF_UNIX
1930 connection. They operate similarly to their sd_pid_get_xyz()
1933 * PID 1 will now maintain a system-wide system state engine
1934 with the states "starting", "running", "degraded",
1935 "maintenance", "stopping". These states are bound to system
1936 startup, normal runtime, runtime with at least one failed
1937 service, rescue/emergency mode and system shutdown. This
1938 state is shown in the "systemctl status" output when no unit
1939 name is passed. It is useful to determine system state, in
1940 particularly when doing so for many systems or containers at
1943 * A new command "list-machines" has been added to "systemctl"
1944 that lists all local OS containers and shows their system
1945 state (see above), if systemd runs inside of them.
1947 * systemctl gained a new "-r" switch to recursively enumerate
1948 units on all local containers, when used with the
1949 "list-unit" command (which is the default one that is
1950 executed when no parameters are specified).
1952 * The GPT automatic partition discovery logic will now honour
1953 two GPT partition flags: one may be set on a partition to
1954 cause it to be mounted read-only, and the other may be set
1955 on a partition to ignore it during automatic discovery.
1957 * Two new GPT type UUIDs have been added for automatic root
1958 partition discovery, for 32-bit and 64-bit ARM. This is not
1959 particularly useful for discovering the root directory on
1960 these architectures during bare-metal boots (since UEFI is
1961 not common there), but still very useful to allow booting of
1962 ARM disk images in nspawn with the -i option.
1964 * MAC addresses of interfaces created with nspawn's
1965 --network-interface= switch will now be generated from the
1966 machine name, and thus be stable between multiple invocations
1969 * logind will now automatically remove all IPC objects owned
1970 by a user if she or he fully logs out. This makes sure that
1971 users who are logged out cannot continue to consume IPC
1972 resources. This covers SysV memory, semaphores and message
1973 queues as well as POSIX shared memory and message
1974 queues. Traditionally, SysV and POSIX IPC had no life-cycle
1975 limits. With this functionality, that is corrected. This may
1976 be turned off by using the RemoveIPC= switch of logind.conf.
1978 * The systemd-machine-id-setup and tmpfiles tools gained a
1979 --root= switch to operate on a specific root directory,
1982 * journald can now forward logged messages to the TTYs of all
1983 logged in users ("wall"). This is the default for all
1984 emergency messages now.
1986 * A new tool systemd-journal-remote has been added to stream
1987 journal log messages across the network.
1989 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
1990 controller trees are mounted into it. Note that the
1991 directories mounted beneath it are not read-only. This is a
1992 security measure and is particularly useful because glibc
1993 actually includes a search logic to pick any tmpfs it can
1994 find to implement shm_open() if /dev/shm is not available
1995 (which it might very well be in namespaced setups).
1997 * machinectl gained a new "poweroff" command to cleanly power
1998 down a local OS container.
2000 * The PrivateDevices= unit file setting will now also drop the
2001 CAP_MKNOD capability from the capability bound set, and
2002 imply DevicePolicy=closed.
2004 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
2005 comprehensively on all long-running systemd services where
2006 this is appropriate.
2008 * systemd-udevd will now run in a disassociated mount
2009 namespace. To mount directories from udev rules, make sure to
2010 pull in mount units via SYSTEMD_WANTS properties.
2012 * The kdbus support gained support for uploading policy into
2013 the kernel. sd-bus gained support for creating "monitoring"
2014 connections that can eavesdrop into all bus communication
2015 for debugging purposes.
2017 * Timestamps may now be specified in seconds since the UNIX
2018 epoch Jan 1st, 1970 by specifying "@" followed by the value
2021 * Native tcpwrap support in systemd has been removed. tcpwrap
2022 is old code, not really maintained anymore and has serious
2023 shortcomings, and better options such as firewalls
2024 exist. For setups that require tcpwrap usage, please
2025 consider invoking your socket-activated service via tcpd,
2026 like on traditional inetd.
2028 * A new system.conf configuration option
2029 DefaultTimerAccuracySec= has been added that controls the
2030 default AccuracySec= setting of .timer units.
2032 * Timer units gained a new WakeSystem= switch. If enabled,
2033 timers configured this way will cause the system to resume
2034 from system suspend (if the system supports that, which most
2037 * Timer units gained a new Persistent= switch. If enabled,
2038 timers configured this way will save to disk when they have
2039 been last triggered. This information is then used on next
2040 reboot to possible execute overdue timer events, that
2041 could not take place because the system was powered off.
2042 This enables simple anacron-like behaviour for timer units.
2044 * systemctl's "list-timers" will now also list the time a
2045 timer unit was last triggered in addition to the next time
2046 it will be triggered.
2048 * systemd-networkd will now assign predictable IPv4LL
2049 addresses to its local interfaces.
2051 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
2052 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
2053 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
2054 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
2055 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
2056 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
2057 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
2058 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
2061 -- Berlin, 2014-03-25
2065 * A new unit file setting RestrictAddressFamilies= has been
2066 added to restrict which socket address families unit
2067 processes gain access to. This takes address family names
2068 like "AF_INET" or "AF_UNIX", and is useful to minimize the
2069 attack surface of services via exotic protocol stacks. This
2070 is built on seccomp system call filters.
2072 * Two new unit file settings RuntimeDirectory= and
2073 RuntimeDirectoryMode= have been added that may be used to
2074 manage a per-daemon runtime directories below /run. This is
2075 an alternative for setting up directory permissions with
2076 tmpfiles snippets, and has the advantage that the runtime
2077 directory's lifetime is bound to the daemon runtime and that
2078 the daemon starts up with an empty directory each time. This
2079 is particularly useful when writing services that drop
2080 privileges using the User= or Group= setting.
2082 * The DeviceAllow= unit setting now supports globbing for
2083 matching against device group names.
2085 * The systemd configuration file system.conf gained new
2086 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
2087 DefaultMemoryAccounting= to globally turn on/off accounting
2088 for specific resources (cgroups) for all units. These
2089 settings may still be overridden individually in each unit
2092 * systemd-gpt-auto-generator is now able to discover /srv and
2093 root partitions in addition to /home and swap partitions. It
2094 also supports LUKS-encrypted partitions now. With this in
2095 place, automatic discovery of partitions to mount following
2096 the Discoverable Partitions Specification
2097 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
2098 is now a lot more complete. This allows booting without
2099 /etc/fstab and without root= on the kernel command line on
2100 systems prepared appropriately.
2102 * systemd-nspawn gained a new --image= switch which allows
2103 booting up disk images and Linux installations on any block
2104 device that follow the Discoverable Partitions Specification
2105 (see above). This means that installations made with
2106 appropriately updated installers may now be started and
2107 deployed using container managers, completely
2108 unmodified. (We hope that libvirt-lxc will add support for
2109 this feature soon, too.)
2111 * systemd-nspawn gained a new --network-macvlan= setting to
2112 set up a private macvlan interface for the
2113 container. Similarly, systemd-networkd gained a new
2114 Kind=macvlan setting in .netdev files.
2116 * systemd-networkd now supports configuring local addresses
2119 * A new tool systemd-network-wait-online has been added to
2120 synchronously wait for network connectivity using
2123 * The sd-bus.h bus API gained a new sd_bus_track object for
2124 tracking the life-cycle of bus peers. Note that sd-bus.h is
2125 still not a public API though (unless you specify
2126 --enable-kdbus on the configure command line, which however
2127 voids your warranty and you get no API stability guarantee).
2129 * The $XDG_RUNTIME_DIR runtime directories for each user are
2130 now individual tmpfs instances, which has the benefit of
2131 introducing separate pools for each user, with individual
2132 size limits, and thus making sure that unprivileged clients
2133 can no longer negatively impact the system or other users by
2134 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
2135 RuntimeDirectorySize= has been introduced that allows
2136 controlling the default size limit for all users. It
2137 defaults to 10% of the available physical memory. This is no
2138 replacement for quotas on tmpfs though (which the kernel
2139 still does not support), as /dev/shm and /tmp are still
2140 shared resources used by both the system and unprivileged
2143 * logind will now automatically turn off automatic suspending
2144 on laptop lid close when more than one display is
2145 connected. This was previously expected to be implemented
2146 individually in desktop environments (such as GNOME),
2147 however has been added to logind now, in order to fix a
2148 boot-time race where a desktop environment might not have
2149 been started yet and thus not been able to take an inhibitor
2150 lock at the time where logind already suspends the system
2151 due to a closed lid.
2153 * logind will now wait at least 30s after each system
2154 suspend/resume cycle, and 3min after system boot before
2155 suspending the system due to a closed laptop lid. This
2156 should give USB docking stations and similar enough time to
2157 be probed and configured after system resume and boot in
2158 order to then act as suspend blocker.
2160 * systemd-run gained a new --property= setting which allows
2161 initialization of resource control properties (and others)
2162 for the created scope or service unit. Example: "systemd-run
2163 --property=BlockIOWeight=10 updatedb" may be used to run
2164 updatedb at a low block IO scheduling weight.
2166 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
2167 now also work in --scope mode.
2169 * When systemd is compiled with kdbus support, basic support
2170 for enforced policies is now in place. (Note that enabling
2171 kdbus still voids your warranty and no API compatibility
2174 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
2175 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
2176 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
2177 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
2178 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
2179 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
2180 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
2181 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
2182 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
2183 Zbigniew Jędrzejewski-Szmek
2185 -- Berlin, 2014-03-12
2189 * systemd will now relabel /dev after loading the SMACK policy
2190 according to SMACK rules.
2192 * A new unit file option AppArmorProfile= has been added to
2193 set the AppArmor profile for the processes of a unit.
2195 * A new condition check ConditionArchitecture= has been added
2196 to conditionalize units based on the system architecture, as
2197 reported by uname()'s "machine" field.
2199 * systemd-networkd now supports matching on the system
2200 virtualization, architecture, kernel command line, host name
2203 * logind is now a lot more aggressive when suspending the
2204 machine due to a closed laptop lid. Instead of acting only
2205 on the lid close action, it will continuously watch the lid
2206 status and act on it. This is useful for laptops where the
2207 power button is on the outside of the chassis so that it can
2208 be reached without opening the lid (such as the Lenovo
2209 Yoga). On those machines, logind will now immediately
2210 re-suspend the machine if the power button has been
2211 accidentally pressed while the laptop was suspended and in a
2212 backpack or similar.
2214 * logind will now watch SW_DOCK switches and inhibit reaction
2215 to the lid switch if it is pressed. This means that logind
2216 will not suspend the machine anymore if the lid is closed
2217 and the system is docked, if the laptop supports SW_DOCK
2218 notifications via the input layer. Note that ACPI docking
2219 stations do not generate this currently. Also note that this
2220 logic is usually not fully sufficient and Desktop
2221 Environments should take a lid switch inhibitor lock when an
2222 external display is connected, as systemd will not watch
2225 * nspawn will now make use of the devices cgroup controller by
2226 default, and only permit creation of and access to the usual
2227 API device nodes like /dev/null or /dev/random, as well as
2228 access to (but not creation of) the pty devices.
2230 * We will now ship a default .network file for
2231 systemd-networkd that automatically configures DHCP for
2232 network interfaces created by nspawn's --network-veth or
2233 --network-bridge= switches.
2235 * systemd will now understand the usual M, K, G, T suffixes
2236 according to SI conventions (i.e. to the base 1000) when
2237 referring to throughput and hardware metrics. It will stay
2238 with IEC conventions (i.e. to the base 1024) for software
2239 metrics, according to what is customary according to
2240 Wikipedia. We explicitly document which base applies for
2241 each configuration option.
2243 * The DeviceAllow= setting in unit files now supports a syntax
2244 to whitelist an entire group of devices node majors at once,
2245 based on the /proc/devices listing. For example, with the
2246 string "char-pts", it is now possible to whitelist all
2247 current and future pseudo-TTYs at once.
2249 * sd-event learned a new "post" event source. Event sources of
2250 this type are triggered by the dispatching of any event
2251 source of a type that is not "post". This is useful for
2252 implementing clean-up and check event sources that are
2253 triggered by other work being done in the program.
2255 * systemd-networkd is no longer statically enabled, but uses
2256 the usual [Install] sections so that it can be
2257 enabled/disabled using systemctl. It still is enabled by
2260 * When creating a veth interface pair with systemd-nspawn, the
2261 host side will now be prefixed with "vb-" if
2262 --network-bridge= is used, and with "ve-" if --network-veth
2263 is used. This way, it is easy to distinguish these cases on
2264 the host, for example to apply different configuration to
2265 them with systemd-networkd.
2267 * The compatibility libraries for libsystemd-journal.so,
2268 libsystem-id128.so, libsystemd-login.so and
2269 libsystemd-daemon.so do not make use of IFUNC
2270 anymore. Instead, we now build libsystemd.so multiple times
2271 under these alternative names. This means that the footprint
2272 is drastically increased, but given that these are
2273 transitional compatibility libraries, this should not matter
2274 much. This change has been made necessary to support the ARM
2275 platform for these compatibility libraries, as the ARM
2276 toolchain is not really at the same level as the toolchain
2277 for other architectures like x86 and does not support
2278 IFUNC. Please make sure to use --enable-compat-libs only
2279 during a transitional period!
2281 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
2282 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
2283 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
2284 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
2285 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
2286 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
2287 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
2288 Zbigniew Jędrzejewski-Szmek
2290 -- Berlin, 2014-02-24
2294 * A new component "systemd-networkd" has been added that can
2295 be used to configure local network interfaces statically or
2296 via DHCP. It is capable of bringing up bridges, VLANs, and
2297 bonding. Currently, no hook-ups for interactive network
2298 configuration are provided. Use this for your initrd,
2299 container, embedded, or server setup if you need a simple,
2300 yet powerful, network configuration solution. This
2301 configuration subsystem is quite nifty, as it allows wildcard
2302 hotplug matching in interfaces. For example, with a single
2303 configuration snippet, you can configure that all Ethernet
2304 interfaces showing up are automatically added to a bridge,
2305 or similar. It supports link-sensing and more.
2307 * A new tool "systemd-socket-proxyd" has been added which can
2308 act as a bidirectional proxy for TCP sockets. This is
2309 useful for adding socket activation support to services that
2310 do not actually support socket activation, including virtual
2311 machines and the like.
2313 * Add a new tool to save/restore rfkill state on
2316 * Save/restore state of keyboard backlights in addition to
2317 display backlights on shutdown/boot.
2319 * udev learned a new SECLABEL{} construct to label device
2320 nodes with a specific security label when they appear. For
2321 now, only SECLABEL{selinux} is supported, but the syntax is
2322 prepared for additional security frameworks.
2324 * udev gained a new scheme to configure link-level attributes
2325 from files in /etc/systemd/network/*.link. These files can
2326 match against MAC address, device path, driver name and type,
2327 and will apply attributes like the naming policy, link speed,
2328 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
2329 address assignment policy (randomized, ...).
2331 * The configuration of network interface naming rules for
2332 "permanent interface names" has changed: a new NamePolicy=
2333 setting in the [Link] section of .link files determines the
2334 priority of possible naming schemes (onboard, slot, mac,
2335 path). The default value of this setting is determined by
2336 /usr/lib/net/links/99-default.link. Old
2337 80-net-name-slot.rules udev configuration file has been
2338 removed, so local configuration overriding this file should
2339 be adapated to override 99-default.link instead.
2341 * When the User= switch is used in a unit file, also
2342 initialize $SHELL= based on the user database entry.
2344 * systemd no longer depends on libdbus. All communication is
2345 now done with sd-bus, systemd's low-level bus library
2348 * kdbus support has been added to PID 1 itself. When kdbus is
2349 enabled, this causes PID 1 to set up the system bus and
2350 enable support for a new ".busname" unit type that
2351 encapsulates bus name activation on kdbus. It works a little
2352 bit like ".socket" units, except for bus names. A new
2353 generator has been added that converts classic dbus1 service
2354 activation files automatically into native systemd .busname
2357 * sd-bus: add a light-weight vtable implementation that allows
2358 defining objects on the bus with a simple static const
2359 vtable array of its methods, signals and properties.
2361 * systemd will not generate or install static dbus
2362 introspection data anymore to /usr/share/dbus-1/interfaces,
2363 as the precise format of these files is unclear, and
2364 nothing makes use of it.
2366 * A proxy daemon is now provided to proxy clients connecting
2367 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
2368 compatibility with classic D-Bus.
2370 * A bus driver implementation has been added that supports the
2371 classic D-Bus bus driver calls on kdbus, also for
2372 compatibility purposes.
2374 * A new API "sd-event.h" has been added that implements a
2375 minimal event loop API built around epoll. It provides a
2376 couple of features that direct epoll usage is lacking:
2377 prioritization of events, scales to large numbers of timer
2378 events, per-event timer slack (accuracy), system-wide
2379 coalescing of timer events, exit handlers, watchdog
2380 supervision support using systemd's sd_notify() API, child
2383 * A new API "sd-rntl.h" has been added that provides an API
2384 around the route netlink interface of the kernel, similar in
2385 style to "sd-bus.h".
2387 * A new API "sd-dhcp-client.h" has been added that provides a
2388 small DHCPv4 client-side implementation. This is used by
2391 * There is a new kernel command line option
2392 "systemd.restore_state=0|1". When set to "0", none of the
2393 systemd tools will restore saved runtime state to hardware
2394 devices. More specifically, the rfkill and backlight states
2397 * The FsckPassNo= compatibility option in mount/service units
2398 has been removed. The fstab generator will now add the
2399 necessary dependencies automatically, and does not require
2400 PID1's support for that anymore.
2402 * journalctl gained a new switch, --list-boots, that lists
2403 recent boots with their times and boot IDs.
2405 * The various tools like systemctl, loginctl, timedatectl,
2406 busctl, systemd-run, ... have gained a new switch "-M" to
2407 connect to a specific, local OS container (as direct
2408 connection, without requiring SSH). This works on any
2409 container that is registered with machined, such as those
2410 created by libvirt-lxc or nspawn.
2412 * systemd-run and systemd-analyze also gained support for "-H"
2413 to connect to remote hosts via SSH. This is particularly
2414 useful for systemd-run because it enables queuing of jobs
2415 onto remote systems.
2417 * machinectl gained a new command "login" to open a getty
2418 login in any local container. This works with any container
2419 that is registered with machined (such as those created by
2420 libvirt-lxc or nspawn), and which runs systemd inside.
2422 * machinectl gained a new "reboot" command that may be used to
2423 trigger a reboot on a specific container that is registered
2424 with machined. This works on any container that runs an init
2425 system of some kind.
2427 * systemctl gained a new "list-timers" command to print a nice
2428 listing of installed timer units with the times they elapse
2431 * Alternative reboot() parameters may now be specified on the
2432 "systemctl reboot" command line and are passed to the
2433 reboot() system call.
2435 * systemctl gained a new --job-mode= switch to configure the
2436 mode to queue a job with. This is a more generic version of
2437 --fail, --irreversible, and --ignore-dependencies, which are
2438 still available but not advertised anymore.
2440 * /etc/systemd/system.conf gained new settings to configure
2441 various default timeouts of units, as well as the default
2442 start limit interval and burst. These may still be overridden
2445 * PID1 will now export on the bus profile data of the security
2446 policy upload process (such as the SELinux policy upload to
2449 * journald: when forwarding logs to the console, include
2450 timestamps (following the setting in
2451 /sys/module/printk/parameters/time).
2453 * OnCalendar= in timer units now understands the special
2454 strings "yearly" and "annually". (Both are equivalent)
2456 * The accuracy of timer units is now configurable with the new
2457 AccuracySec= setting. It defaults to 1min.
2459 * A new dependency type JoinsNamespaceOf= has been added that
2460 allows running two services within the same /tmp and network
2461 namespace, if PrivateNetwork= or PrivateTmp= are used.
2463 * A new command "cat" has been added to systemctl. It outputs
2464 the original unit file of a unit, and concatenates the
2465 contents of additional "drop-in" unit file snippets, so that
2466 the full configuration is shown.
2468 * systemctl now supports globbing on the various "list-xyz"
2469 commands, like "list-units" or "list-sockets", as well as on
2470 those commands which take multiple unit names.
2472 * journalctl's --unit= switch gained support for globbing.
2474 * All systemd daemons now make use of the watchdog logic so
2475 that systemd automatically notices when they hang.
2477 * If the $container_ttys environment variable is set,
2478 getty-generator will automatically spawn a getty for each
2479 listed tty. This is useful for container managers to request
2480 login gettys to be spawned on as many ttys as needed.
2482 * %h, %s, %U specifier support is not available anymore when
2483 used in unit files for PID 1. This is because NSS calls are
2484 not safe from PID 1. They stay available for --user
2485 instances of systemd, and as special case for the root user.
2487 * loginctl gained a new "--no-legend" switch to turn off output
2490 * The "sd-login.h" API gained three new calls:
2491 sd_session_is_remote(), sd_session_get_remote_user(),
2492 sd_session_get_remote_host() to query information about
2495 * The udev hardware database now also carries vendor/product
2496 information of SDIO devices.
2498 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
2499 determine whether watchdog notifications are requested by
2502 * Socket-activated per-connection services now include a
2503 short description of the connection parameters in the
2506 * tmpfiles gained a new "--boot" option. When this is not used,
2507 only lines where the command character is not suffixed with
2508 "!" are executed. When this option is specified, those
2509 options are executed too. This partitions tmpfiles
2510 directives into those that can be safely executed at any
2511 time, and those which should be run only at boot (for
2512 example, a line that creates /run/nologin).
2514 * A new API "sd-resolve.h" has been added which provides a simple
2515 asynchronous wrapper around glibc NSS host name resolution
2516 calls, such as getaddrinfo(). In contrast to glibc's
2517 getaddrinfo_a(), it does not use signals. In contrast to most
2518 other asynchronous name resolution libraries, this one does
2519 not reimplement DNS, but reuses NSS, so that alternate
2520 host name resolution systems continue to work, such as mDNS,
2521 LDAP, etc. This API is based on libasyncns, but it has been
2522 cleaned up for inclusion in systemd.
2524 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
2525 "sd-daemon.h" are no longer found in individual libraries
2526 libsystemd-journal.so, libsystemd-login.so,
2527 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
2528 merged them into a single library, libsystemd.so, which
2529 provides all symbols. The reason for this is cyclic
2530 dependencies, as these libraries tend to use each other's
2531 symbols. So far, we have managed to workaround that by linking
2532 a copy of a good part of our code into each of these
2533 libraries again and again, which, however, makes certain
2534 things hard to do, like sharing static variables. Also, it
2535 substantially increases footprint. With this change, there
2536 is only one library for the basic APIs systemd
2537 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
2538 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
2539 library as well, however are subject to the --enable-kdbus
2540 switch (see below). Note that "sd-dhcp-client.h" is not part
2541 of this library (this is because it only consumes, never
2542 provides, services of/to other APIs). To make the transition
2543 easy from the separate libraries to the unified one, we
2544 provide the --enable-compat-libs compile-time switch which
2545 will generate stub libraries that are compatible with the
2546 old ones but redirect all calls to the new one.
2548 * All of the kdbus logic and the new APIs "sd-bus.h",
2549 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
2550 and "sd-utf8.h" are compile-time optional via the
2551 "--enable-kdbus" switch, and they are not compiled in by
2552 default. To make use of kdbus, you have to explicitly enable
2553 the switch. Note however, that neither the kernel nor the
2554 userspace API for all of this is considered stable yet. We
2555 want to maintain the freedom to still change the APIs for
2556 now. By specifying this build-time switch, you acknowledge
2557 that you are aware of the instability of the current
2560 * Also, note that while kdbus is pretty much complete,
2561 it lacks one thing: proper policy support. This means you
2562 can build a fully working system with all features; however,
2563 it will be highly insecure. Policy support will be added in
2564 one of the next releases, at the same time that we will
2565 declare the APIs stable.
2567 * When the kernel command line argument "kdbus" is specified,
2568 systemd will automatically load the kdbus.ko kernel module. At
2569 this stage of development, it is only useful for testing kdbus
2570 and should not be used in production. Note: if "--enable-kdbus"
2571 is specified, and the kdbus.ko kernel module is available, and
2572 "kdbus" is added to the kernel command line, the entire system
2573 runs with kdbus instead of dbus-daemon, with the above mentioned
2574 problem of missing the system policy enforcement. Also a future
2575 version of kdbus.ko or a newer systemd will not be compatible with
2576 each other, and will unlikely be able to boot the machine if only
2577 one of them is updated.
2579 * systemctl gained a new "import-environment" command which
2580 uploads the caller's environment (or parts thereof) into the
2581 service manager so that it is inherited by services started
2582 by the manager. This is useful to upload variables like
2583 $DISPLAY into the user service manager.
2585 * A new PrivateDevices= switch has been added to service units
2586 which allows running a service with a namespaced /dev
2587 directory that does not contain any device nodes for
2588 physical devices. More specifically, it only includes devices
2589 such as /dev/null, /dev/urandom, and /dev/zero which are API
2592 * logind has been extended to support behaviour like VT
2593 switching on seats that do not support a VT. This makes
2594 multi-session available on seats that are not the first seat
2595 (seat0), and on systems where kernel support for VTs has
2596 been disabled at compile-time.
2598 * If a process holds a delay lock for system sleep or shutdown
2599 and fails to release it in time, we will now log its
2600 identity. This makes it easier to identify processes that
2601 cause slow suspends or power-offs.
2603 * When parsing /etc/crypttab, support for a new key-slot=
2604 option as supported by Debian is added. It allows indicating
2605 which LUKS slot to use on disk, speeding up key loading.
2607 * The sd_journald_sendv() API call has been checked and
2608 officially declared to be async-signal-safe so that it may
2609 be invoked from signal handlers for logging purposes.
2611 * Boot-time status output is now enabled automatically after a
2612 short timeout if boot does not progress, in order to give
2613 the user an indication what she or he is waiting for.
2615 * The boot-time output has been improved to show how much time
2616 remains until jobs expire.
2618 * The KillMode= switch in service units gained a new possible
2619 value "mixed". If set, and the unit is shut down, then the
2620 initial SIGTERM signal is sent only to the main daemon
2621 process, while the following SIGKILL signal is sent to
2622 all remaining processes of the service.
2624 * When a scope unit is registered, a new property "Controller"
2625 may be set. If set to a valid bus name, systemd will send a
2626 RequestStop() signal to this name when it would like to shut
2627 down the scope. This may be used to hook manager logic into
2628 the shutdown logic of scope units. Also, scope units may now
2629 be put in a special "abandoned" state, in which case the
2630 manager process which created them takes no further
2631 responsibilities for it.
2633 * When reading unit files, systemd will now verify
2634 the access mode of these files, and warn about certain
2635 suspicious combinations. This has been added to make it
2636 easier to track down packaging bugs where unit files are
2637 marked executable or world-writable.
2639 * systemd-nspawn gained a new "--setenv=" switch to set
2640 container-wide environment variables. The similar option in
2641 systemd-activate was renamed from "--environment=" to
2642 "--setenv=" for consistency.
2644 * systemd-nspawn has been updated to create a new kdbus domain
2645 for each container that is invoked, thus allowing each
2646 container to have its own set of system and user buses,
2647 independent of the host.
2649 * systemd-nspawn gained a new --drop-capability= switch to run
2650 the container with less capabilities than the default. Both
2651 --drop-capability= and --capability= now take the special
2652 string "all" for dropping or keeping all capabilities.
2654 * systemd-nspawn gained new switches for executing containers
2655 with specific SELinux labels set.
2657 * systemd-nspawn gained a new --quiet switch to not generate
2658 any additional output but the container's own console
2661 * systemd-nspawn gained a new --share-system switch to run a
2662 container without PID namespacing enabled.
2664 * systemd-nspawn gained a new --register= switch to control
2665 whether the container is registered with systemd-machined or
2666 not. This is useful for containers that do not run full
2667 OS images, but only specific apps.
2669 * systemd-nspawn gained a new --keep-unit which may be used
2670 when invoked as the only program from a service unit, and
2671 results in registration of the unit service itself in
2672 systemd-machined, instead of a newly opened scope unit.
2674 * systemd-nspawn gained a new --network-interface= switch for
2675 moving arbitrary interfaces to the container. The new
2676 --network-veth switch creates a virtual Ethernet connection
2677 between host and container. The new --network-bridge=
2678 switch then allows assigning the host side of this virtual
2679 Ethernet connection to a bridge device.
2681 * systemd-nspawn gained a new --personality= switch for
2682 setting the kernel personality for the container. This is
2683 useful when running a 32-bit container on a 64-bit host. A
2684 similar option Personality= is now also available for service
2687 * logind will now also track a "Desktop" identifier for each
2688 session which encodes the desktop environment of it. This is
2689 useful for desktop environments that want to identify
2690 multiple running sessions of itself easily.
2692 * A new SELinuxContext= setting for service units has been
2693 added that allows setting a specific SELinux execution
2694 context for a service.
2696 * Most systemd client tools will now honour $SYSTEMD_LESS for
2697 settings of the "less" pager. By default, these tools will
2698 override $LESS to allow certain operations to work, such as
2699 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
2700 influence this logic.
2702 * systemd's "seccomp" hook-up has been changed to make use of
2703 the libseccomp library instead of using its own
2704 implementation. This has benefits for portability among
2707 * For usage together with SystemCallFilter=, a new
2708 SystemCallErrorNumber= setting has been introduced that
2709 allows configuration of a system error number to be returned
2710 on filtered system calls, instead of immediately killing the
2711 process. Also, SystemCallArchitectures= has been added to
2712 limit access to system calls of a particular architecture
2713 (in order to turn off support for unused secondary
2714 architectures). There is also a global
2715 SystemCallArchitectures= setting in system.conf now to turn
2716 off support for non-native system calls system-wide.
2718 * systemd requires a kernel with a working name_to_handle_at(),
2719 please see the kernel config requirements in the README file.
2721 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
2722 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
2723 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
2724 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
2725 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
2726 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
2727 Elia Pinto, Florian Weimer, George McCollister, Goffredo
2728 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
2729 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
2730 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
2731 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
2732 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
2733 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
2734 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
2735 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
2736 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
2737 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
2738 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
2739 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
2740 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
2741 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
2742 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
2743 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
2744 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
2746 -- Berlin, 2014-02-20
2750 * logind has gained support for facilitating privileged input
2751 and drm device access for unprivileged clients. This work is
2752 useful to allow Wayland display servers (and similar
2753 programs, such as kmscon) to run under the user's ID and
2754 access input and drm devices which are normally
2755 protected. When this is used (and the kernel is new enough)
2756 logind will "mute" IO on the file descriptors passed to
2757 Wayland as long as it is in the background and "unmute" it
2758 if it returns into the foreground. This allows secure
2759 session switching without allowing background sessions to
2760 eavesdrop on input and display data. This also introduces
2761 session switching support if VT support is turned off in the
2762 kernel, and on seats that are not seat0.
2764 * A new kernel command line option luks.options= is understood
2765 now which allows specifying LUKS options for usage for LUKS
2766 encrypted partitions specified with luks.uuid=.
2768 * tmpfiles.d(5) snippets may now use specifier expansion in
2769 path names. More specifically %m, %b, %H, %v, are now
2770 replaced by the local machine id, boot id, hostname, and
2771 kernel version number.
2773 * A new tmpfiles.d(5) command "m" has been introduced which
2774 may be used to change the owner/group/access mode of a file
2775 or directory if it exists, but do nothing if it does not.
2777 * This release removes high-level support for the
2778 MemorySoftLimit= cgroup setting. The underlying kernel
2779 cgroup attribute memory.soft_limit= is currently badly
2780 designed and likely to be removed from the kernel API in its
2781 current form, hence we should not expose it for now.
2783 * The memory.use_hierarchy cgroup attribute is now enabled for
2784 all cgroups systemd creates in the memory cgroup
2785 hierarchy. This option is likely to be come the built-in
2786 default in the kernel anyway, and the non-hierarchical mode
2787 never made much sense in the intrinsically hierarchical
2790 * A new field _SYSTEMD_SLICE= is logged along with all journal
2791 messages containing the slice a message was generated
2792 from. This is useful to allow easy per-customer filtering of
2793 logs among other things.
2795 * systemd-journald will no longer adjust the group of journal
2796 files it creates to the "systemd-journal" group. Instead we
2797 rely on the journal directory to be owned by the
2798 "systemd-journal" group, and its setgid bit set, so that the
2799 kernel file system layer will automatically enforce that
2800 journal files inherit this group assignment. The reason for
2801 this change is that we cannot allow NSS look-ups from
2802 journald which would be necessary to resolve
2803 "systemd-journal" to a numeric GID, because this might
2804 create deadlocks if NSS involves synchronous queries to
2805 other daemons (such as nscd, or sssd) which in turn are
2806 logging clients of journald and might block on it, which
2807 would then dead lock. A tmpfiles.d(5) snippet included in
2808 systemd will make sure the setgid bit and group are
2809 properly set on the journal directory if it exists on every
2810 boot. However, we recommend adjusting it manually after
2811 upgrades too (or from RPM scriptlets), so that the change is
2812 not delayed until next reboot.
2814 * Backlight and random seed files in /var/lib/ have moved into
2815 the /var/lib/systemd/ directory, in order to centralize all
2816 systemd generated files in one directory.
2818 * Boot time performance measurements (as displayed by
2819 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
2820 performance information if that's available to determine how
2821 much time BIOS and boot loader initialization required. With
2822 a sufficiently new BIOS you hence no longer need to boot
2823 with Gummiboot to get access to such information.
2825 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
2826 Cristian Rodríguez, Dave Reisner, David Herrmann, David
2827 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
2828 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
2829 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
2830 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
2831 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
2833 -- Berlin, 2013-10-02
2837 * The Restart= option for services now understands a new
2838 on-watchdog setting, which will restart the service
2839 automatically if the service stops sending out watchdog keep
2840 alive messages (as configured with WatchdogSec=).
2842 * The getty generator (which is responsible for bringing up a
2843 getty on configured serial consoles) will no longer only
2844 start a getty on the primary kernel console but on all
2845 others, too. This makes the order in which console= is
2846 specified on the kernel command line less important.
2848 * libsystemd-logind gained a new sd_session_get_vt() call to
2849 retrieve the VT number of a session.
2851 * If the option "tries=0" is set for an entry of /etc/crypttab
2852 its passphrase is queried indefinitely instead of any
2853 maximum number of tries.
2855 * If a service with a configure PID file terminates its PID
2856 file will now be removed automatically if it still exists
2857 afterwards. This should put an end to stale PID files.
2859 * systemd-run will now also take relative binary path names
2860 for execution and no longer insists on absolute paths.
2862 * InaccessibleDirectories= and ReadOnlyDirectories= now take
2863 paths that are optionally prefixed with "-" to indicate that
2864 it should not be considered a failure if they do not exist.
2866 * journalctl -o (and similar commands) now understands a new
2867 output mode "short-precise", it is similar to "short" but
2868 shows timestamps with usec accuracy.
2870 * The option "discard" (as known from Debian) is now
2871 synonymous to "allow-discards" in /etc/crypttab. In fact,
2872 "discard" is preferred now (since it is easier to remember
2875 * Some licensing clean-ups were made, so that more code is now
2876 LGPL-2.1 licensed than before.
2878 * A minimal tool to save/restore the display backlight
2879 brightness across reboots has been added. It will store the
2880 backlight setting as late as possible at shutdown, and
2881 restore it as early as possible during reboot.
2883 * A logic to automatically discover and enable home and swap
2884 partitions on GPT disks has been added. With this in place
2885 /etc/fstab becomes optional for many setups as systemd can
2886 discover certain partitions located on the root disk
2887 automatically. Home partitions are recognized under their
2888 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
2889 partitions are recognized under their GPT type ID
2890 0657fd6da4ab43c484e50933c84b4f4f.
2892 * systemd will no longer pass any environment from the kernel
2893 or initrd to system services. If you want to set an
2894 environment for all services, do so via the kernel command
2895 line systemd.setenv= assignment.
2897 * The systemd-sysctl tool no longer natively reads the file
2898 /etc/sysctl.conf. If desired, the file should be symlinked
2899 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
2900 legacy support by a symlink rather than built-in code, it
2901 also makes the otherwise hidden order of application of the
2902 different files visible. (Note that this partly reverts to a
2903 pre-198 application order of sysctl knobs!)
2905 * The "systemctl set-log-level" and "systemctl dump" commands
2906 have been moved to systemd-analyze.
2908 * systemd-run learned the new --remain-after-exit switch,
2909 which causes the scope unit not to be cleaned up
2910 automatically after the process terminated.
2912 * tmpfiles learned a new --exclude-prefix= switch to exclude
2913 certain paths from operation.
2915 * journald will now automatically flush all messages to disk
2916 as soon as a message at the log level CRIT, ALERT or EMERG
2919 Contributions from: Andrew Cook, Brandon Philips, Christian
2920 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
2921 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
2922 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
2923 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
2924 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
2925 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
2926 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
2927 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
2928 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
2929 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
2930 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
2931 William Giokas, Zbigniew Jędrzejewski-Szmek
2933 -- Berlin, 2013-09-13
2937 * The documentation has been updated to cover the various new
2938 concepts introduced with 205.
2940 * Unit files now understand the new %v specifier which
2941 resolves to the kernel version string as returned by "uname
2944 * systemctl now supports filtering the unit list output by
2945 load state, active state and sub state, using the new
2948 * "systemctl status" will now show the results of the
2949 condition checks (like ConditionPathExists= and similar) of
2950 the last start attempts of the unit. They are also logged to
2953 * "journalctl -b" may now be used to look for boot output of a
2954 specific boot. Try "journalctl -b -1" for the previous boot,
2955 but the syntax is substantially more powerful.
2957 * "journalctl --show-cursor" has been added which prints the
2958 cursor string the last shown log line. This may then be used
2959 with the new "journalctl --after-cursor=" switch to continue
2960 browsing logs from that point on.
2962 * "journalctl --force" may now be used to force regeneration
2965 * Creation of "dead" device nodes has been moved from udev
2966 into kmod and tmpfiles. Previously, udev would read the kmod
2967 databases to pre-generate dead device nodes based on meta
2968 information contained in kernel modules, so that these would
2969 be auto-loaded on access rather then at boot. As this
2970 does not really have much to do with the exposing actual
2971 kernel devices to userspace this has always been slightly
2972 alien in the udev codebase. Following the new scheme kmod
2973 will now generate a runtime snippet for tmpfiles from the
2974 module meta information and it now is tmpfiles' job to the
2975 create the nodes. This also allows overriding access and
2976 other parameters for the nodes using the usual tmpfiles
2977 facilities. As side effect this allows us to remove the
2978 CAP_SYS_MKNOD capability bit from udevd entirely.
2980 * logind's device ACLs may now be applied to these "dead"
2981 devices nodes too, thus finally allowing managed access to
2982 devices such as /dev/snd/sequencer whithout loading the
2983 backing module right-away.
2985 * A new RPM macro has been added that may be used to apply
2986 tmpfiles configuration during package installation.
2988 * systemd-detect-virt and ConditionVirtualization= now can
2989 detect User-Mode-Linux machines (UML).
2991 * journald will now implicitly log the effective capabilities
2992 set of processes in the message metadata.
2994 * systemd-cryptsetup has gained support for TrueCrypt volumes.
2996 * The initrd interface has been simplified (more specifically,
2997 support for passing performance data via environment
2998 variables and fsck results via files in /run has been
2999 removed). These features were non-essential, and are
3000 nowadays available in a much nicer way by having systemd in
3001 the initrd serialize its state and have the hosts systemd
3002 deserialize it again.
3004 * The udev "keymap" data files and tools to apply keyboard
3005 specific mappings of scan to key codes, and force-release
3006 scan code lists have been entirely replaced by a udev
3007 "keyboard" builtin and a hwdb data file.
3009 * systemd will now honour the kernel's "quiet" command line
3010 argument also during late shutdown, resulting in a
3011 completely silent shutdown when used.
3013 * There's now an option to control the SO_REUSEPORT socket
3014 option in .socket units.
3016 * Instance units will now automatically get a per-template
3017 subslice of system.slice unless something else is explicitly
3018 configured. For example, instances of sshd@.service will now
3019 implicitly be placed in system-sshd.slice rather than
3020 system.slice as before.
3022 * Test coverage support may now be enabled at build time.
3024 Contributions from: Dave Reisner, Frederic Crozat, Harald
3025 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
3026 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
3027 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
3028 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
3029 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
3030 Giokas, Zbigniew Jędrzejewski-Szmek
3032 -- Berlin, 2013-07-23
3036 * Two new unit types have been introduced:
3038 Scope units are very similar to service units, however, are
3039 created out of pre-existing processes -- instead of PID 1
3040 forking off the processes. By using scope units it is
3041 possible for system services and applications to group their
3042 own child processes (worker processes) in a powerful way
3043 which then maybe used to organize them, or kill them
3044 together, or apply resource limits on them.
3046 Slice units may be used to partition system resources in an
3047 hierarchical fashion and then assign other units to them. By
3048 default there are now three slices: system.slice (for all
3049 system services), user.slice (for all user sessions),
3050 machine.slice (for VMs and containers).
3052 Slices and scopes have been introduced primarily in
3053 context of the work to move cgroup handling to a
3054 single-writer scheme, where only PID 1
3055 creates/removes/manages cgroups.
3057 * There's a new concept of "transient" units. In contrast to
3058 normal units these units are created via an API at runtime,
3059 not from configuration from disk. More specifically this
3060 means it is now possible to run arbitrary programs as
3061 independent services, with all execution parameters passed
3062 in via bus APIs rather than read from disk. Transient units
3063 make systemd substantially more dynamic then it ever was,
3064 and useful as a general batch manager.
3066 * logind has been updated to make use of scope and slice units
3067 for managing user sessions. As a user logs in he will get
3068 his own private slice unit, to which all sessions are added
3069 as scope units. We also added support for automatically
3070 adding an instance of user@.service for the user into the
3071 slice. Effectively logind will no longer create cgroup
3072 hierarchies on its own now, it will defer entirely to PID 1
3073 for this by means of scope, service and slice units. Since
3074 user sessions this way become entities managed by PID 1
3075 the output of "systemctl" is now a lot more comprehensive.
3077 * A new mini-daemon "systemd-machined" has been added which
3078 may be used by virtualization managers to register local
3079 VMs/containers. nspawn has been updated accordingly, and
3080 libvirt will be updated shortly. machined will collect a bit
3081 of meta information about the VMs/containers, and assign
3082 them their own scope unit (see above). The collected
3083 meta-data is then made available via the "machinectl" tool,
3084 and exposed in "ps" and similar tools. machined/machinectl
3085 is compile-time optional.
3087 * As discussed earlier, the low-level cgroup configuration
3088 options ControlGroup=, ControlGroupModify=,
3089 ControlGroupPersistent=, ControlGroupAttribute= have been
3090 removed. Please use high-level attribute settings instead as
3091 well as slice units.
3093 * A new bus call SetUnitProperties() has been added to alter
3094 various runtime parameters of a unit. This is primarily
3095 useful to alter cgroup parameters dynamically in a nice way,
3096 but will be extended later on to make more properties
3097 modifiable at runtime. systemctl gained a new set-properties
3098 command that wraps this call.
3100 * A new tool "systemd-run" has been added which can be used to
3101 run arbitrary command lines as transient services or scopes,
3102 while configuring a number of settings via the command
3103 line. This tool is currently very basic, however already
3104 very useful. We plan to extend this tool to even allow
3105 queuing of execution jobs with time triggers from the
3106 command line, similar in fashion to "at".
3108 * nspawn will now inform the user explicitly that kernels with
3109 audit enabled break containers, and suggest the user to turn
3112 * Support for detecting the IMA and AppArmor security
3113 frameworks with ConditionSecurity= has been added.
3115 * journalctl gained a new "-k" switch for showing only kernel
3116 messages, mimicking dmesg output; in addition to "--user"
3117 and "--system" switches for showing only user's own logs
3120 * systemd-delta can now show information about drop-in
3121 snippets extending unit files.
3123 * libsystemd-bus has been substantially updated but is still
3124 not available as public API.
3126 * systemd will now look for the "debug" argument on the kernel
3127 command line and enable debug logging, similar to what
3128 "systemd.log_level=debug" already did before.
3130 * "systemctl set-default", "systemctl get-default" has been
3131 added to configure the default.target symlink, which
3132 controls what to boot into by default.
3134 * "systemctl set-log-level" has been added as a convenient
3135 way to raise and lower systemd logging threshold.
3137 * "systemd-analyze plot" will now show the time the various
3138 generators needed for execution, as well as information
3139 about the unit file loading.
3141 * libsystemd-journal gained a new sd_journal_open_files() call
3142 for opening specific journal files. journactl also gained a
3143 new switch to expose this new functionality. Previously we
3144 only supported opening all files from a directory, or all
3145 files from the system, as opening individual files only is
3146 racy due to journal file rotation.
3148 * systemd gained the new DefaultEnvironment= setting in
3149 /etc/systemd/system.conf to set environment variables for
3152 * If a privileged process logs a journal message with the
3153 OBJECT_PID= field set, then journald will automatically
3154 augment this with additional OBJECT_UID=, OBJECT_GID=,
3155 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
3156 system services want to log events about specific client
3157 processes. journactl/systemctl has been updated to make use
3158 of this information if all log messages regarding a specific
3161 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
3162 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
3163 Reisner, David Coppa, David King, David Strauss, Eelco
3164 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
3165 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
3166 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
3167 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
3168 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
3169 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
3170 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
3171 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
3172 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
3173 Łukasz Stelmach, 장동준
3177 * The Python bindings gained some minimal support for the APIs
3178 exposed by libsystemd-logind.
3180 * ConditionSecurity= gained support for detecting SMACK. Since
3181 this condition already supports SELinux and AppArmor we only
3182 miss IMA for this. Patches welcome!
3184 Contributions from: Karol Lewandowski, Lennart Poettering,
3185 Zbigniew Jędrzejewski-Szmek
3189 * systemd-nspawn will now create /etc/resolv.conf if
3190 necessary, before bind-mounting the host's file onto it.
3192 * systemd-nspawn will now store meta information about a
3193 container on the container's cgroup as extended attribute
3194 fields, including the root directory.
3196 * The cgroup hierarchy has been reworked in many ways. All
3197 objects any of the components systemd creates in the cgroup
3198 tree are now suffixed. More specifically, user sessions are
3199 now placed in cgroups suffixed with ".session", users in
3200 cgroups suffixed with ".user", and nspawn containers in
3201 cgroups suffixed with ".nspawn". Furthermore, all cgroup
3202 names are now escaped in a simple scheme to avoid collision
3203 of userspace object names with kernel filenames. This work
3204 is preparation for making these objects relocatable in the
3205 cgroup tree, in order to allow easy resource partitioning of
3206 these objects without causing naming conflicts.
3208 * systemctl list-dependencies gained the new switches
3209 --plain, --reverse, --after and --before.
3211 * systemd-inhibit now shows the process name of processes that
3212 have taken an inhibitor lock.
3214 * nss-myhostname will now also resolve "localhost"
3215 implicitly. This makes /etc/hosts an optional file and
3216 nicely handles that on IPv6 ::1 maps to both "localhost" and
3219 * libsystemd-logind.so gained a new call
3220 sd_get_machine_names() to enumerate running containers and
3221 VMs (currently only supported by very new libvirt and
3222 nspawn). sd_login_monitor can now be used to watch
3223 VMs/containers coming and going.
3225 * .include is not allowed recursively anymore, and only in
3226 unit files. Usually it is better to use drop-in snippets in
3227 .d/*.conf anyway, as introduced with systemd 198.
3229 * systemd-analyze gained a new "critical-chain" command that
3230 determines the slowest chain of units run during system
3231 boot-up. It is very useful for tracking down where
3232 optimizing boot time is the most beneficial.
3234 * systemd will no longer allow manipulating service paths in
3235 the name=systemd:/system cgroup tree using ControlGroup= in
3236 units. (But is still fine with it in all other dirs.)
3238 * There's a new systemd-nspawn@.service service file that may
3239 be used to easily run nspawn containers as system
3240 services. With the container's root directory in
3241 /var/lib/container/foobar it is now sufficient to run
3242 "systemctl start systemd-nspawn@foobar.service" to boot it.
3244 * systemd-cgls gained a new parameter "--machine" to list only
3245 the processes within a certain container.
3247 * ConditionSecurity= now can check for "apparmor". We still
3248 are lacking checks for SMACK and IMA for this condition
3249 check though. Patches welcome!
3251 * A new configuration file /etc/systemd/sleep.conf has been
3252 added that may be used to configure which kernel operation
3253 systemd is supposed to execute when "suspend", "hibernate"
3254 or "hybrid-sleep" is requested. This makes the new kernel
3255 "freeze" state accessible to the user.
3257 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
3258 the passed argument if applicable.
3260 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
3261 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
3262 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
3263 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
3264 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
3265 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
3266 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
3271 * The output of 'systemctl list-jobs' got some polishing. The
3272 '--type=' argument may now be passed more than once. A new
3273 command 'systemctl list-sockets' has been added which shows
3274 a list of kernel sockets systemd is listening on with the
3275 socket units they belong to, plus the units these socket
3278 * The experimental libsystemd-bus library got substantial
3279 updates to work in conjunction with the (also experimental)
3280 kdbus kernel project. It works well enough to exchange
3281 messages with some sophistication. Note that kdbus is not
3282 ready yet, and the library is mostly an elaborate test case
3283 for now, and not installable.
3285 * systemd gained a new unit 'systemd-static-nodes.service'
3286 that generates static device nodes earlier during boot, and
3287 can run in conjunction with udev.
3289 * libsystemd-login gained a new call sd_pid_get_user_unit()
3290 to retrieve the user systemd unit a process is running
3291 in. This is useful for systems where systemd is used as
3294 * systemd-nspawn now places all containers in the new /machine
3295 top-level cgroup directory in the name=systemd
3296 hierarchy. libvirt will soon do the same, so that we get a
3297 uniform separation of /system, /user and /machine for system
3298 services, user processes and containers/virtual
3299 machines. This new cgroup hierarchy is also useful to stick
3300 stable names to specific container instances, which can be
3301 recognized later this way (this name may be controlled
3302 via systemd-nspawn's new -M switch). libsystemd-login also
3303 gained a new call sd_pid_get_machine_name() to retrieve the
3304 name of the container/VM a specific process belongs to.
3306 * bootchart can now store its data in the journal.
3308 * libsystemd-journal gained a new call
3309 sd_journal_add_conjunction() for AND expressions to the
3310 matching logic. This can be used to express more complex
3311 logical expressions.
3313 * journactl can now take multiple --unit= and --user-unit=
3316 * The cryptsetup logic now understands the "luks.key=" kernel
3317 command line switch for specifying a file to read the
3318 decryption key from. Also, if a configured key file is not
3319 found the tool will now automatically fall back to prompting
3322 * Python systemd.journal module was updated to wrap recently
3323 added functions from libsystemd-journal. The interface was
3324 changed to bring the low level interface in s.j._Reader
3325 closer to the C API, and the high level interface in
3326 s.j.Reader was updated to wrap and convert all data about
3329 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
3330 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
3331 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
3332 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
3333 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
3334 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3338 * journalctl --update-catalog now understands a new --root=
3339 option to operate on catalogs found in a different root
3342 * During shutdown after systemd has terminated all running
3343 services a final killing loop kills all remaining left-over
3344 processes. We will now print the name of these processes
3345 when we send SIGKILL to them, since this usually indicates a
3348 * If /etc/crypttab refers to password files stored on
3349 configured mount points automatic dependencies will now be
3350 generated to ensure the specific mount is established first
3351 before the key file is attempted to be read.
3353 * 'systemctl status' will now show information about the
3354 network sockets a socket unit is listening on.
3356 * 'systemctl status' will also shown information about any
3357 drop-in configuration file for units. (Drop-In configuration
3358 files in this context are files such as
3359 /etc/systemd/systemd/foobar.service.d/*.conf)
3361 * systemd-cgtop now optionally shows summed up CPU times of
3362 cgroups. Press '%' while running cgtop to switch between
3363 percentage and absolute mode. This is useful to determine
3364 which cgroups use up the most CPU time over the entire
3365 runtime of the system. systemd-cgtop has also been updated
3366 to be 'pipeable' for processing with further shell tools.
3368 * 'hostnamectl set-hostname' will now allow setting of FQDN
3371 * The formatting and parsing of time span values has been
3372 changed. The parser now understands fractional expressions
3373 such as "5.5h". The formatter will now output fractional
3374 expressions for all time spans under 1min, i.e. "5.123456s"
3375 rather than "5s 123ms 456us". For time spans under 1s
3376 millisecond values are shown, for those under 1ms
3377 microsecond values are shown. This should greatly improve
3378 all time-related output of systemd.
3380 * libsystemd-login and libsystemd-journal gained new
3381 functions for querying the poll() events mask and poll()
3382 timeout value for integration into arbitrary event
3385 * localectl gained the ability to list available X11 keymaps
3386 (models, layouts, variants, options).
3388 * 'systemd-analyze dot' gained the ability to filter for
3389 specific units via shell-style globs, to create smaller,
3390 more useful graphs. I.e. it is now possible to create simple
3391 graphs of all the dependencies between only target units, or
3392 of all units that Avahi has dependencies with.
3394 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
3395 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
3396 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
3397 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
3398 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
3399 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
3400 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
3404 * The boot-time readahead implementation for rotating media
3405 will now read the read-ahead data in multiple passes which
3406 consist of all read requests made in equidistant time
3407 intervals. This means instead of strictly reading read-ahead
3408 data in its physical order on disk we now try to find a
3409 middle ground between physical and access time order.
3411 * /etc/os-release files gained a new BUILD_ID= field for usage
3412 on operating systems that provide continuous builds of OS
3415 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
3416 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
3417 William Douglas, Zbigniew Jędrzejewski-Szmek
3421 * systemd-python gained an API exposing libsystemd-daemon.
3423 * The SMACK setup logic gained support for uploading CIPSO
3426 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
3427 ReadOnlyDirectories= and InaccessibleDirectories= has
3428 changed. The private /tmp and /var/tmp directories are now
3429 shared by all processes of a service (which means
3430 ExecStartPre= may now leave data in /tmp that ExecStart= of
3431 the same service can still access). When a service is
3432 stopped its temporary directories are immediately deleted
3433 (normal clean-up with tmpfiles is still done in addition to
3436 * By default, systemd will now set a couple of sysctl
3437 variables in the kernel: the safe sysrq options are turned
3438 on, IP route verification is turned on, and source routing
3439 disabled. The recently added hardlink and softlink
3440 protection of the kernel is turned on. These settings should
3441 be reasonably safe, and good defaults for all new systems.
3443 * The predictable network naming logic may now be turned off
3444 with a new kernel command line switch: net.ifnames=0.
3446 * A new libsystemd-bus module has been added that implements a
3447 pretty complete D-Bus client library. For details see:
3449 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
3451 * journald will now explicitly flush the journal files to disk
3452 at the latest 5min after each write. The file will then also
3453 be marked offline until the next write. This should increase
3454 reliability in case of a crash. The synchronization delay
3455 can be configured via SyncIntervalSec= in journald.conf.
3457 * There's a new remote-fs-setup.target unit that can be used
3458 to pull in specific services when at least one remote file
3459 system is to be mounted.
3461 * There are new targets timers.target and paths.target as
3462 canonical targets to pull user timer and path units in
3463 from. This complements sockets.target with a similar
3464 purpose for socket units.
3466 * libudev gained a new call udev_device_set_attribute_value()
3467 to set sysfs attributes of a device.
3469 * The udev daemon now sets the default number of worker
3470 processes executed in parallel based on the number of available
3471 CPUs instead of the amount of available RAM. This is supposed
3472 to provide a more reliable default and limit a too aggressive
3473 paralellism for setups with 1000s of devices connected.
3475 Contributions from: Auke Kok, Colin Walters, Cristian
3476 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
3477 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
3478 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
3479 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
3480 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
3481 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
3482 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
3483 Zbigniew Jędrzejewski-Szmek
3487 * Configuration of unit files may now be extended via drop-in
3488 files without having to edit/override the unit files
3489 themselves. More specifically, if the administrator wants to
3490 change one value for a service file foobar.service he can
3491 now do so by dropping in a configuration snippet into
3492 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
3493 will load all these snippets and apply them on top of the
3494 main unit configuration file, possibly extending or
3495 overriding its settings. Using these drop-in snippets is
3496 generally nicer than the two earlier options for changing
3497 unit files locally: copying the files from
3498 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
3499 them there; or creating a new file in /etc/systemd/system/
3500 that incorporates the original one via ".include". Drop-in
3501 snippets into these .d/ directories can be placed in any
3502 directory systemd looks for units in, and the usual
3503 overriding semantics between /usr/lib, /etc and /run apply
3506 * Most unit file settings which take lists of items can now be
3507 reset by assigning the empty string to them. For example,
3508 normally, settings such as Environment=FOO=BAR append a new
3509 environment variable assignment to the environment block,
3510 each time they are used. By assigning Environment= the empty
3511 string the environment block can be reset to empty. This is
3512 particularly useful with the .d/*.conf drop-in snippets
3513 mentioned above, since this adds the ability to reset list
3514 settings from vendor unit files via these drop-ins.
3516 * systemctl gained a new "list-dependencies" command for
3517 listing the dependencies of a unit recursively.
3519 * Inhibitors are now honored and listed by "systemctl
3520 suspend", "systemctl poweroff" (and similar) too, not only
3521 GNOME. These commands will also list active sessions by
3524 * Resource limits (as exposed by the various control group
3525 controllers) can now be controlled dynamically at runtime
3526 for all units. More specifically, you can now use a command
3527 like "systemctl set-cgroup-attr foobar.service cpu.shares
3528 2000" to alter the CPU shares a specific service gets. These
3529 settings are stored persistently on disk, and thus allow the
3530 administrator to easily adjust the resource usage of
3531 services with a few simple commands. This dynamic resource
3532 management logic is also available to other programs via the
3533 bus. Almost any kernel cgroup attribute and controller is
3536 * systemd-vconsole-setup will now copy all font settings to
3537 all allocated VTs, where it previously applied them only to
3540 * libsystemd-login gained the new sd_session_get_tty() API
3543 * This release drops support for a few legacy or
3544 distribution-specific LSB facility names when parsing init
3545 scripts: $x-display-manager, $mail-transfer-agent,
3546 $mail-transport-agent, $mail-transfer-agent, $smtp,
3547 $null. Also, the mail-transfer-agent.target unit backing
3548 this has been removed. Distributions which want to retain
3549 compatibility with this should carry the burden for
3550 supporting this themselves and patch support for these back
3551 in, if they really need to. Also, the facilities $syslog and
3552 $local_fs are now ignored, since systemd does not support
3553 early-boot LSB init scripts anymore, and these facilities
3554 are implied anyway for normal services. syslog.target has
3557 * There are new bus calls on PID1's Manager object for
3558 cancelling jobs, and removing snapshot units. Previously,
3559 both calls were only available on the Job and Snapshot
3562 * systemd-journal-gatewayd gained SSL support.
3564 * The various "environment" files, such as /etc/locale.conf
3565 now support continuation lines with a backslash ("\") as
3566 last character in the line, similarly in style (but different)
3567 to how this is supported in shells.
3569 * For normal user processes the _SYSTEMD_USER_UNIT= field is
3570 now implicitly appended to every log entry logged. systemctl
3571 has been updated to filter by this field when operating on a
3572 user systemd instance.
3574 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
3575 CAP_AUDIT_CONTROL capabilities to the capabilities set for
3576 the container. This makes it easier to boot unmodified
3577 Fedora systems in a container, which however still requires
3578 audit=0 to be passed on the kernel command line. Auditing in
3579 kernel and userspace is unfortunately still too broken in
3580 context of containers, hence we recommend compiling it out
3581 of the kernel or using audit=0. Hopefully this will be fixed
3582 one day for good in the kernel.
3584 * nspawn gained the new --bind= and --bind-ro= parameters to
3585 bind mount specific directories from the host into the
3588 * nspawn will now mount its own devpts file system instance
3589 into the container, in order not to leak pty devices from
3590 the host into the container.
3592 * systemd will now read the firmware boot time performance
3593 information from the EFI variables, if the used boot loader
3594 supports this, and takes it into account for boot performance
3595 analysis via "systemd-analyze". This is currently supported
3596 only in conjunction with Gummiboot, but could be supported
3597 by other boot loaders too. For details see:
3599 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
3601 * A new generator has been added that automatically mounts the
3602 EFI System Partition (ESP) to /boot, if that directory
3603 exists, is empty, and no other file system has been
3604 configured to be mounted there.
3606 * logind will now send out PrepareForSleep(false) out
3607 unconditionally, after coming back from suspend. This may be
3608 used by applications as asynchronous notification for
3609 system resume events.
3611 * "systemctl unlock-sessions" has been added, that allows
3612 unlocking the screens of all user sessions at once, similar
3613 to how "systemctl lock-sessions" already locked all users
3614 sessions. This is backed by a new D-Bus call UnlockSessions().
3616 * "loginctl seat-status" will now show the master device of a
3617 seat. (i.e. the device of a seat that needs to be around for
3618 the seat to be considered available, usually the graphics
3621 * tmpfiles gained a new "X" line type, that allows
3622 configuration of files and directories (with wildcards) that
3623 shall be excluded from automatic cleanup ("aging").
3625 * udev default rules set the device node permissions now only
3626 at "add" events, and do not change them any longer with a
3627 later "change" event.
3629 * The log messages for lid events and power/sleep keypresses
3630 now carry a message ID.
3632 * We now have a substantially larger unit test suite, but this
3633 continues to be work in progress.
3635 * udevadm hwdb gained a new --root= parameter to change the
3636 root directory to operate relative to.
3638 * logind will now issue a background sync() request to the kernel
3639 early at shutdown, so that dirty buffers are flushed to disk early
3640 instead of at the last moment, in order to optimize shutdown
3643 * A new bootctl tool has been added that is an interface for
3644 certain boot loader operations. This is currently a preview
3645 and is likely to be extended into a small mechanism daemon
3646 like timedated, localed, hostnamed, and can be used by
3647 graphical UIs to enumerate available boot options, and
3648 request boot into firmware operations.
3650 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
3651 the rest of the package. It also has been updated to work
3652 correctly in initrds.
3654 * Policykit previously has been runtime optional, and is now
3655 also compile time optional via a configure switch.
3657 * systemd-analyze has been reimplemented in C. Also "systemctl
3658 dot" has moved into systemd-analyze.
3660 * "systemctl status" with no further parameters will now print
3661 the status of all active or failed units.
3663 * Operations such as "systemctl start" can now be executed
3664 with a new mode "--irreversible" which may be used to queue
3665 operations that cannot accidentally be reversed by a later
3666 job queuing. This is by default used to make shutdown
3667 requests more robust.
3669 * The Python API of systemd now gained a new module for
3670 reading journal files.
3672 * A new tool kernel-install has been added that can install
3673 kernel images according to the Boot Loader Specification:
3675 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
3677 * Boot time console output has been improved to provide
3678 animated boot time output for hanging jobs.
3680 * A new tool systemd-activate has been added which can be used
3681 to test socket activation with, directly from the command
3682 line. This should make it much easier to test and debug
3683 socket activation in daemons.
3685 * journalctl gained a new "--reverse" (or -r) option to show
3686 journal output in reverse order (i.e. newest line first).
3688 * journalctl gained a new "--pager-end" (or -e) option to jump
3689 to immediately jump to the end of the journal in the
3690 pager. This is only supported in conjunction with "less".
3692 * journalctl gained a new "--user-unit=" option, that works
3693 similarly to "--unit=" but filters for user units rather than
3696 * A number of unit files to ease adoption of systemd in
3697 initrds has been added. This moves some minimal logic from
3698 the various initrd implementations into systemd proper.
3700 * The journal files are now owned by a new group
3701 "systemd-journal", which exists specifically to allow access
3702 to the journal, and nothing else. Previously, we used the
3703 "adm" group for that, which however possibly covers more
3704 than just journal/log file access. This new group is now
3705 already used by systemd-journal-gatewayd to ensure this
3706 daemon gets access to the journal files and as little else
3707 as possible. Note that "make install" will also set FS ACLs
3708 up for /var/log/journal to give "adm" and "wheel" read
3709 access to it, in addition to "systemd-journal" which owns
3710 the journal files. We recommend that packaging scripts also
3711 add read access to "adm" + "wheel" to /var/log/journal, and
3712 all existing/future journal files. To normal users and
3713 administrators little changes, however packagers need to
3714 ensure to create the "systemd-journal" system group at
3715 package installation time.
3717 * The systemd-journal-gatewayd now runs as unprivileged user
3718 systemd-journal-gateway:systemd-journal-gateway. Packaging
3719 scripts need to create these system user/group at
3722 * timedated now exposes a new boolean property CanNTP that
3723 indicates whether a local NTP service is available or not.
3725 * systemd-detect-virt will now also detect xen PVs
3727 * The pstore file system is now mounted by default, if it is
3730 * In addition to the SELinux and IMA policies we will now also
3731 load SMACK policies at early boot.
3733 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
3734 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
3735 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
3736 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
3737 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
3738 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
3739 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
3740 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
3741 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
3742 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
3743 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
3744 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
3745 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
3746 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
3750 * Timer units now support calendar time events in addition to
3751 monotonic time events. That means you can now trigger a unit
3752 based on a calendar time specification such as "Thu,Fri
3753 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
3754 or fifth day of any month of the year 2013, given that it is
3755 a thursday or friday. This brings timer event support
3756 considerably closer to cron's capabilities. For details on
3757 the supported calendar time specification language see
3760 * udev now supports a number of different naming policies for
3761 network interfaces for predictable names, and a combination
3762 of these policies is now the default. Please see this wiki
3763 document for details:
3765 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
3767 * Auke Kok's bootchart implementation has been added to the
3768 systemd tree. It is an optional component that can graph the
3769 boot in quite some detail. It is one of the best bootchart
3770 implementations around and minimal in its code and
3773 * nss-myhostname has been integrated into the systemd source
3774 tree. nss-myhostname guarantees that the local hostname
3775 always stays resolvable via NSS. It has been a weak
3776 requirement of systemd-hostnamed since a long time, and
3777 since its code is actually trivial we decided to just
3778 include it in systemd's source tree. It can be turned off
3779 with a configure switch.
3781 * The read-ahead logic is now capable of properly detecting
3782 whether a btrfs file system is on SSD or rotating media, in
3783 order to optimize the read-ahead scheme. Previously, it was
3784 only capable of detecting this on traditional file systems
3787 * In udev, additional device properties are now read from the
3788 IAB in addition to the OUI database. Also, Bluetooth company
3789 identities are attached to the devices as well.
3791 * In service files %U may be used as specifier that is
3792 replaced by the configured user name of the service.
3794 * nspawn may now be invoked without a controlling TTY. This
3795 makes it suitable for invocation as its own service. This
3796 may be used to set up a simple containerized server system
3797 using only core OS tools.
3799 * systemd and nspawn can now accept socket file descriptors
3800 when they are started for socket activation. This enables
3801 implementation of socket activated nspawn
3802 containers. i.e. think about autospawning an entire OS image
3803 when the first SSH or HTTP connection is received. We expect
3804 that similar functionality will also be added to libvirt-lxc
3807 * journalctl will now suppress ANSI color codes when
3808 presenting log data.
3810 * systemctl will no longer show control group information for
3811 a unit if a the control group is empty anyway.
3813 * logind can now automatically suspend/hibernate/shutdown the
3816 * /etc/machine-info and hostnamed now also expose the chassis
3817 type of the system. This can be used to determine whether
3818 the local system is a laptop, desktop, handset or
3819 tablet. This information may either be configured by the
3820 user/vendor or is automatically determined from ACPI and DMI
3821 information if possible.
3823 * A number of PolicyKit actions are now bound together with
3824 "imply" rules. This should simplify creating UIs because
3825 many actions will now authenticate similar ones as well.
3827 * Unit files learnt a new condition ConditionACPower= which
3828 may be used to conditionalize a unit depending on whether an
3829 AC power source is connected or not, of whether the system
3830 is running on battery power.
3832 * systemctl gained a new "is-failed" verb that may be used in
3833 shell scripts and suchlike to check whether a specific unit
3834 is in the "failed" state.
3836 * The EnvironmentFile= setting in unit files now supports file
3837 globbing, and can hence be used to easily read a number of
3838 environment files at once.
3840 * systemd will no longer detect and recognize specific
3841 distributions. All distribution-specific #ifdeffery has been
3842 removed, systemd is now fully generic and
3843 distribution-agnostic. Effectively, not too much is lost as
3844 a lot of the code is still accessible via explicit configure
3845 switches. However, support for some distribution specific
3846 legacy configuration file formats has been dropped. We
3847 recommend distributions to simply adopt the configuration
3848 files everybody else uses now and convert the old
3849 configuration from packaging scripts. Most distributions
3850 already did that. If that's not possible or desirable,
3851 distributions are welcome to forward port the specific
3852 pieces of code locally from the git history.
3854 * When logging a message about a unit systemd will now always
3855 log the unit name in the message meta data.
3857 * localectl will now also discover system locale data that is
3858 not stored in locale archives, but directly unpacked.
3860 * logind will no longer unconditionally use framebuffer
3861 devices as seat masters, i.e. as devices that are required
3862 to be existing before a seat is considered preset. Instead,
3863 it will now look for all devices that are tagged as
3864 "seat-master" in udev. By default framebuffer devices will
3865 be marked as such, but depending on local systems other
3866 devices might be marked as well. This may be used to
3867 integrate graphics cards using closed source drivers (such
3868 as NVidia ones) more nicely into logind. Note however, that
3869 we recommend using the open source NVidia drivers instead,
3870 and no udev rules for the closed-source drivers will be
3871 shipped from us upstream.
3873 Contributions from: Adam Williamson, Alessandro Crismani, Auke
3874 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
3875 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
3876 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
3877 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
3878 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
3879 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
3880 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
3881 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
3882 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
3883 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
3884 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
3889 * udev gained support for loading additional device properties
3890 from an indexed database that is keyed by vendor/product IDs
3891 and similar device identifiers. For the beginning this
3892 "hwdb" is populated with data from the well-known PCI and
3893 USB database, but also includes PNP, ACPI and OID data. In
3894 the longer run this indexed database shall grow into
3895 becoming the one central database for non-essential
3896 userspace device metadata. Previously, data from the PCI/USB
3897 database was only attached to select devices, since the
3898 lookup was a relatively expensive operation due to O(n) time
3899 complexity (with n being the number of entries in the
3900 database). Since this is now O(1), we decided to add in this
3901 data for all devices where this is available, by
3902 default. Note that the indexed database needs to be rebuilt
3903 when new data files are installed. To achieve this you need
3904 to update your packaging scripts to invoke "udevadm hwdb
3905 --update" after installation of hwdb data files. For
3906 RPM-based distributions we introduced the new
3907 %udev_hwdb_update macro for this purpose.
3909 * The Journal gained support for the "Message Catalog", an
3910 indexed database to link up additional information with
3911 journal entries. For further details please check:
3913 http://www.freedesktop.org/wiki/Software/systemd/catalog
3915 The indexed message catalog database also needs to be
3916 rebuilt after installation of message catalog files. Use
3917 "journalctl --update-catalog" for this. For RPM-based
3918 distributions we introduced the %journal_catalog_update
3919 macro for this purpose.
3921 * The Python Journal bindings gained support for the standard
3922 Python logging framework.
3924 * The Journal API gained new functions for checking whether
3925 the underlying file system of a journal file is capable of
3926 properly reporting file change notifications, or whether
3927 applications that want to reflect journal changes "live"
3928 need to recheck journal files continuously in appropriate
3931 * It is now possible to set the "age" field for tmpfiles
3932 entries to 0, indicating that files matching this entry
3933 shall always be removed when the directories are cleaned up.
3935 * coredumpctl gained a new "gdb" verb which invokes gdb
3936 right-away on the selected coredump.
3938 * There's now support for "hybrid sleep" on kernels that
3939 support this, in addition to "suspend" and "hibernate". Use
3940 "systemctl hybrid-sleep" to make use of this.
3942 * logind's HandleSuspendKey= setting (and related settings)
3943 now gained support for a new "lock" setting to simply
3944 request the screen lock on all local sessions, instead of
3945 actually executing a suspend or hibernation.
3947 * systemd will now mount the EFI variables file system by
3950 * Socket units now gained support for configuration of the
3951 SMACK security label.
3953 * timedatectl will now output the time of the last and next
3954 daylight saving change.
3956 * We dropped support for various legacy and distro-specific
3957 concepts, such as insserv, early-boot SysV services
3958 (i.e. those for non-standard runlevels such as 'b' or 'S')
3959 or ArchLinux /etc/rc.conf support. We recommend the
3960 distributions who still need support this to either continue
3961 to maintain the necessary patches downstream, or find a
3962 different solution. (Talk to us if you have questions!)
3964 * Various systemd components will now bypass PolicyKit checks
3965 for root and otherwise handle properly if PolicyKit is not
3966 found to be around. This should fix most issues for
3967 PolicyKit-less systems. Quite frankly this should have been
3968 this way since day one. It is absolutely our intention to
3969 make systemd work fine on PolicyKit-less systems, and we
3970 consider it a bug if something does not work as it should if
3971 PolicyKit is not around.
3973 * For embedded systems it is now possible to build udev and
3974 systemd without blkid and/or kmod support.
3976 * "systemctl switch-root" is now capable of switching root
3977 more than once. I.e. in addition to transitions from the
3978 initrd to the host OS it is now possible to transition to
3979 further OS images from the host. This is useful to implement
3980 offline updating tools.
3982 * Various other additions have been made to the RPM macros
3983 shipped with systemd. Use %udev_rules_update() after
3984 installing new udev rules files. %_udevhwdbdir,
3985 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
3986 %_sysctldir are now available which resolve to the right
3987 directories for packages to place various data files in.
3989 * journalctl gained the new --full switch (in addition to
3990 --all, to disable ellipsation for long messages.
3992 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
3993 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
3994 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
3995 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
3996 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
3997 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
3998 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
3999 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
4000 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
4004 * journalctl gained new --since= and --until= switches to
4005 filter by time. It also now supports nice filtering for
4006 units via --unit=/-u.
4008 * Type=oneshot services may use ExecReload= and do the
4011 * The journal daemon now supports time-based rotation and
4012 vacuuming, in addition to the usual disk-space based
4015 * The journal will now index the available field values for
4016 each field name. This enables clients to show pretty drop
4017 downs of available match values when filtering. The bash
4018 completion of journalctl has been updated
4019 accordingly. journalctl gained a new switch -F to list all
4020 values a certain field takes in the journal database.
4022 * More service events are now written as structured messages
4023 to the journal, and made recognizable via message IDs.
4025 * The timedated, localed and hostnamed mini-services which
4026 previously only provided support for changing time, locale
4027 and hostname settings from graphical DEs such as GNOME now
4028 also have a minimal (but very useful) text-based client
4029 utility each. This is probably the nicest way to changing
4030 these settings from the command line now, especially since
4031 it lists available options and is fully integrated with bash
4034 * There's now a new tool "systemd-coredumpctl" to list and
4035 extract coredumps from the journal.
4037 * We now install a README each in /var/log/ and
4038 /etc/rc.d/init.d explaining where the system logs and init
4039 scripts went. This hopefully should help folks who go to
4040 that dirs and look into the otherwise now empty void and
4041 scratch their heads.
4043 * When user-services are invoked (by systemd --user) the
4044 $MANAGERPID env var is set to the PID of systemd.
4046 * SIGRTMIN+24 when sent to a --user instance will now result
4047 in immediate termination of systemd.
4049 * gatewayd received numerous feature additions such as a
4050 "follow" mode, for live syncing and filtering.
4052 * browse.html now allows filtering and showing detailed
4053 information on specific entries. Keyboard navigation and
4054 mouse screen support has been added.
4056 * gatewayd/journalctl now supports HTML5/JSON
4057 Server-Sent-Events as output.
4059 * The SysV init script compatibility logic will now
4060 heuristically determine whether a script supports the
4061 "reload" verb, and only then make this available as
4064 * "systemctl status --follow" has been removed, use "journalctl
4067 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
4068 have been removed since they are hardly useful to be
4071 * And I'd like to take the opportunity to specifically mention
4072 Zbigniew for his great contributions. Zbigniew, you rock!
4074 Contributions from: Andrew Eikum, Christian Hesse, Colin
4075 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
4076 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
4077 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
4078 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
4079 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
4080 Jędrzejewski-Szmek, Сковорода Никита Андреевич
4084 * If /etc/vconsole.conf is non-existent or empty we will no
4085 longer load any console font or key map at boot by
4086 default. Instead the kernel defaults will be left
4087 intact. This is definitely the right thing to do, as no
4088 configuration should mean no configuration, and hard-coding
4089 font names that are different on all archs is probably a bad
4090 idea. Also, the kernel default key map and font should be
4091 good enough for most cases anyway, and mostly identical to
4092 the userspace fonts/key maps we previously overloaded them
4093 with. If distributions want to continue to default to a
4094 non-kernel font or key map they should ship a default
4095 /etc/vconsole.conf with the appropriate contents.
4097 Contributions from: Colin Walters, Daniel J Walsh, Dave
4098 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
4099 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
4103 * journalctl gained a new --cursor= switch to show entries
4104 starting from the specified location in the journal.
4106 * We now enforce a size limit on journal entry fields exported
4107 with "-o json" in journalctl. Fields larger than 4K will be
4108 assigned null. This can be turned off with --all.
4110 * An (optional) journal gateway daemon is now available as
4111 "systemd-journal-gatewayd.service". This service provides
4112 access to the journal via HTTP and JSON. This functionality
4113 will be used to implement live log synchronization in both
4114 pull and push modes, but has various other users too, such
4115 as easy log access for debugging of embedded devices. Right
4116 now it is already useful to retrieve the journal via HTTP:
4118 # systemctl start systemd-journal-gatewayd.service
4119 # wget http://localhost:19531/entries
4121 This will download the journal contents in a
4122 /var/log/messages compatible format. The same as JSON:
4124 # curl -H"Accept: application/json" http://localhost:19531/entries
4126 This service is also accessible via a web browser where a
4127 single static HTML5 app is served that uses the JSON logic
4128 to enable the user to do some basic browsing of the
4129 journal. This will be extended later on. Here's an example
4130 screenshot of this app in its current state:
4132 http://0pointer.de/public/journal-gatewayd
4134 Contributions from: Kay Sievers, Lennart Poettering, Robert
4135 Milasan, Tom Gundersen
4139 * The bash completion logic is now available for journalctl
4142 * We do not mount the "cpuset" controller anymore together with
4143 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
4144 started if no parameters are assigned to it. "cpuset" hence
4145 broke code that assumed it it could create "cpu" groups and
4148 * journalctl -f will now subscribe to terminal size changes,
4149 and line break accordingly.
4151 Contributions from: Dave Reisner, Kay Sievers, Lennart
4152 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
4156 * nspawn will now create a symlink /etc/localtime in the
4157 container environment, copying the host's timezone
4158 setting. Previously this has been done via a bind mount, but
4159 since symlinks cannot be bind mounted this has now been
4160 changed to create/update the appropriate symlink.
4162 * journalctl -n's line number argument is now optional, and
4163 will default to 10 if omitted.
4165 * journald will now log the maximum size the journal files may
4166 take up on disk. This is particularly useful if the default
4167 built-in logic of determining this parameter from the file
4168 system size is used. Use "systemctl status
4169 systemd-journald.service" to see this information.
4171 * The multi-seat X wrapper tool has been stripped down. As X
4172 is now capable of enumerating graphics devices via udev in a
4173 seat-aware way the wrapper is not strictly necessary
4174 anymore. A stripped down temporary stop-gap is still shipped
4175 until the upstream display managers have been updated to
4176 fully support the new X logic. Expect this wrapper to be
4177 removed entirely in one of the next releases.
4179 * HandleSleepKey= in logind.conf has been split up into
4180 HandleSuspendKey= and HandleHibernateKey=. The old setting
4181 is not available anymore. X11 and the kernel are
4182 distinguishing between these keys and we should too. This
4183 also means the inhibition lock for these keys has been split
4186 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
4187 Poettering, Lukas Nykryn, Václav Pavlín
4191 * Whenever a unit changes state we will now log this to the
4192 journal and show along the unit's own log output in
4195 * ConditionPathIsMountPoint= can now properly detect bind
4196 mount points too. (Previously, a bind mount of one file
4197 system to another place in the same file system could not be
4198 detected as mount, since they shared struct stat's st_dev
4201 * We will now mount the cgroup controllers cpu, cpuacct,
4202 cpuset and the controllers net_cls, net_prio together by
4205 * nspawn containers will now have a virtualized boot
4206 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
4207 over with a randomized ID at container initialization). This
4208 has the effect of making "journalctl -b" do the right thing
4211 * The JSON output journal serialization has been updated not
4212 to generate "endless" list objects anymore, but rather one
4213 JSON object per line. This is more in line how most JSON
4214 parsers expect JSON objects. The new output mode
4215 "json-pretty" has been added to provide similar output, but
4216 neatly aligned for readability by humans.
4218 * We dropped all explicit sync() invocations in the shutdown
4219 code. The kernel does this implicitly anyway in the kernel
4220 reboot() syscall. halt(8)'s -n option is now a compatibility
4223 * We now support virtualized reboot() in containers, as
4224 supported by newer kernels. We will fall back to exit() if
4225 CAP_SYS_REBOOT is not available to the container. Also,
4226 nspawn makes use of this now and will actually reboot the
4227 container if the containerized OS asks for that.
4229 * journalctl will only show local log output by default
4230 now. Use --merge (-m) to show remote log output, too.
4232 * libsystemd-journal gained the new sd_journal_get_usage()
4233 call to determine the current disk usage of all journal
4234 files. This is exposed in the new "journalctl --disk-usage"
4237 * journald gained a new configuration setting SplitMode= in
4238 journald.conf which may be used to control how user journals
4239 are split off. See journald.conf(5) for details.
4241 * A new condition type ConditionFileNotEmpty= has been added.
4243 * tmpfiles' "w" lines now support file globbing, to write
4244 multiple files at once.
4246 * We added Python bindings for the journal submission
4247 APIs. More Python APIs for a number of selected APIs will
4248 likely follow. Note that we intend to add native bindings
4249 only for the Python language, as we consider it common
4250 enough to deserve bindings shipped within systemd. There are
4251 various projects outside of systemd that provide bindings
4252 for languages such as PHP or Lua.
4254 * Many conditions will now resolve specifiers such as %i. In
4255 addition, PathChanged= and related directives of .path units
4256 now support specifiers as well.
4258 * There's now a new RPM macro definition for the system preset
4261 * journald will now warn if it ca not forward a message to the
4262 syslog daemon because its socket is full.
4264 * timedated will no longer write or process /etc/timezone,
4265 except on Debian. As we do not support late mounted /usr
4266 anymore /etc/localtime always being a symlink is now safe,
4267 and hence the information in /etc/timezone is not necessary
4270 * logind will now always reserve one VT for a text getty (VT6
4271 by default). Previously if more than 6 X sessions where
4272 started they took up all the VTs with auto-spawned gettys,
4273 so that no text gettys were available anymore.
4275 * udev will now automatically inform the btrfs kernel logic
4276 about btrfs RAID components showing up. This should make
4277 simple hotplug based btrfs RAID assembly work.
4279 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
4280 (but not for its children which will stay at the kernel
4281 default). This should allow setups with a lot more listening
4284 * systemd will now always pass the configured timezone to the
4285 kernel at boot. timedated will do the same when the timezone
4288 * logind's inhibition logic has been updated. By default,
4289 logind will now handle the lid switch, the power and sleep
4290 keys all the time, even in graphical sessions. If DEs want
4291 to handle these events on their own they should take the new
4292 handle-power-key, handle-sleep-key and handle-lid-switch
4293 inhibitors during their runtime. A simple way to achieve
4294 that is to invoke the DE wrapped in an invocation of:
4296 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
4298 * Access to unit operations is now checked via SELinux taking
4299 the unit file label and client process label into account.
4301 * systemd will now notify the administrator in the journal
4302 when he over-mounts a non-empty directory.
4304 * There are new specifiers that are resolved in unit files,
4305 for the host name (%H), the machine ID (%m) and the boot ID
4308 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
4309 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
4310 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
4311 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
4312 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
4313 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
4314 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
4318 * Support for reading structured kernel messages from
4319 /dev/kmsg has now been added and is enabled by default.
4321 * Support for reading kernel messages from /proc/kmsg has now
4322 been removed. If you want kernel messages in the journal
4323 make sure to run a recent kernel (>= 3.5) that supports
4324 reading structured messages from /dev/kmsg (see
4325 above). /proc/kmsg is now exclusive property of classic
4326 syslog daemons again.
4328 * The libudev API gained the new
4329 udev_device_new_from_device_id() call.
4331 * The logic for file system namespace (ReadOnlyDirectory=,
4332 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
4333 require pivot_root() anymore. This means fewer temporary
4334 directories are created below /tmp for this feature.
4336 * nspawn containers will now see and receive all submounts
4337 made on the host OS below the root file system of the
4340 * Forward Secure Sealing is now supported for Journal files,
4341 which provide cryptographical sealing of journal files so
4342 that attackers cannot alter log history anymore without this
4343 being detectable. Lennart will soon post a blog story about
4344 this explaining it in more detail.
4346 * There are two new service settings RestartPreventExitStatus=
4347 and SuccessExitStatus= which allow configuration of exit
4348 status (exit code or signal) which will be excepted from the
4349 restart logic, resp. consider successful.
4351 * journalctl gained the new --verify switch that can be used
4352 to check the integrity of the structure of journal files and
4353 (if Forward Secure Sealing is enabled) the contents of
4356 * nspawn containers will now be run with /dev/stdin, /dev/fd/
4357 and similar symlinks pre-created. This makes running shells
4358 as container init process a lot more fun.
4360 * The fstab support can now handle PARTUUID= and PARTLABEL=
4363 * A new ConditionHost= condition has been added to match
4364 against the hostname (with globs) and machine ID. This is
4365 useful for clusters where a single OS image is used to
4366 provision a large number of hosts which shall run slightly
4367 different sets of services.
4369 * Services which hit the restart limit will now be placed in a
4372 Contributions from: Bertram Poettering, Dave Reisner, Huang
4373 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
4374 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
4378 * When running in --user mode systemd will now become a
4379 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
4380 tree a lot more organized.
4382 * A new PartOf= unit dependency type has been introduced that
4383 may be used to group services in a natural way.
4385 * "systemctl enable" may now be used to enable instances of
4388 * journalctl now prints error log levels in red, and
4389 warning/notice log levels in bright white. It also supports
4390 filtering by log level now.
4392 * cgtop gained a new -n switch (similar to top), to configure
4393 the maximum number of iterations to run for. It also gained
4394 -b, to run in batch mode (accepting no input).
4396 * The suffix ".service" may now be omitted on most systemctl
4397 command lines involving service unit names.
4399 * There's a new bus call in logind to lock all sessions, as
4400 well as a loginctl verb for it "lock-sessions".
4402 * libsystemd-logind.so gained a new call sd_journal_perror()
4403 that works similar to libc perror() but logs to the journal
4404 and encodes structured information about the error number.
4406 * /etc/crypttab entries now understand the new keyfile-size=
4409 * shutdown(8) now can send a (configurable) wall message when
4410 a shutdown is cancelled.
4412 * The mount propagation mode for the root file system will now
4413 default to "shared", which is useful to make containers work
4414 nicely out-of-the-box so that they receive new mounts from
4415 the host. This can be undone locally by running "mount
4416 --make-rprivate /" if needed.
4418 * The prefdm.service file has been removed. Distributions
4419 should maintain this unit downstream if they intend to keep
4420 it around. However, we recommend writing normal unit files
4421 for display managers instead.
4423 * Since systemd is a crucial part of the OS we will now
4424 default to a number of compiler switches that improve
4425 security (hardening) such as read-only relocations, stack
4426 protection, and suchlike.
4428 * The TimeoutSec= setting for services is now split into
4429 TimeoutStartSec= and TimeoutStopSec= to allow configuration
4430 of individual time outs for the start and the stop phase of
4433 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
4434 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
4435 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
4436 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
4437 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
4438 Gundersen, Zbigniew Jędrzejewski-Szmek
4442 * The journal and id128 C APIs are now fully documented as man
4445 * Extra safety checks have been added when transitioning from
4446 the initial RAM disk to the main system to avoid accidental
4449 * /etc/crypttab entries now understand the new keyfile-offset=
4452 * systemctl -t can now be used to filter by unit load state.
4454 * The journal C API gained the new sd_journal_wait() call to
4455 make writing synchronous journal clients easier.
4457 * journalctl gained the new -D switch to show journals from a
4460 * journalctl now displays a special marker between log
4461 messages of two different boots.
4463 * The journal is now explicitly flushed to /var via a service
4464 systemd-journal-flush.service, rather than implicitly simply
4465 by seeing /var/log/journal to be writable.
4467 * journalctl (and the journal C APIs) can now match for much
4468 more complex expressions, with alternatives and
4471 * When transitioning from the initial RAM disk to the main
4472 system we will now kill all processes in a killing spree to
4473 ensure no processes stay around by accident.
4475 * Three new specifiers may be used in unit files: %u, %h, %s
4476 resolve to the user name, user home directory resp. user
4477 shell. This is useful for running systemd user instances.
4479 * We now automatically rotate journal files if their data
4480 object hash table gets a fill level > 75%. We also size the
4481 hash table based on the configured maximum file size. This
4482 together should lower hash collisions drastically and thus
4483 speed things up a bit.
4485 * journalctl gained the new "--header" switch to introspect
4486 header data of journal files.
4488 * A new setting SystemCallFilters= has been added to services
4489 which may be used to apply blacklists or whitelists to
4490 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
4492 * nspawn gained a new --link-journal= switch (and quicker: -j)
4493 to link the container journal with the host. This makes it
4494 very easy to centralize log viewing on the host for all
4495 guests while still keeping the journal files separated.
4497 * Many bugfixes and optimizations
4499 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
4500 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
4501 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
4506 * Several tools now understand kernel command line arguments,
4507 which are only read when run in an initial RAM disk. They
4508 usually follow closely their normal counterparts, but are
4511 * There's a new tool to analyze the readahead files that are
4512 automatically generated at boot. Use:
4514 /usr/lib/systemd/systemd-readahead analyze /.readahead
4516 * We now provide an early debug shell on tty9 if this enabled. Use:
4518 systemctl enable debug-shell.service
4520 * All plymouth related units have been moved into the Plymouth
4521 package. Please make sure to upgrade your Plymouth version
4524 * systemd-tmpfiles now supports getting passed the basename of
4525 a configuration file only, in which case it will look for it
4526 in all appropriate directories automatically.
4528 * udevadm info now takes a /dev or /sys path as argument, and
4529 does the right thing. Example:
4531 udevadm info /dev/sda
4532 udevadm info /sys/class/block/sda
4534 * systemctl now prints a warning if a unit is stopped but a
4535 unit that might trigger it continues to run. Example: a
4536 service is stopped but the socket that activates it is left
4539 * "systemctl status" will now mention if the log output was
4540 shortened due to rotation since a service has been started.
4542 * The journal API now exposes functions to determine the
4543 "cutoff" times due to rotation.
4545 * journald now understands SIGUSR1 and SIGUSR2 for triggering
4546 immediately flushing of runtime logs to /var if possible,
4547 resp. for triggering immediate rotation of the journal
4550 * It is now considered an error if a service is attempted to
4551 be stopped that is not loaded.
4553 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
4555 * systemd-analyze now supports Python 3
4557 * tmpfiles now supports cleaning up directories via aging
4558 where the first level dirs are always kept around but
4559 directories beneath it automatically aged. This is enabled
4560 by prefixing the age field with '~'.
4562 * Seat objects now expose CanGraphical, CanTTY properties
4563 which is required to deal with very fast bootups where the
4564 display manager might be running before the graphics drivers
4565 completed initialization.
4567 * Seat objects now expose a State property.
4569 * We now include RPM macros for service enabling/disabling
4570 based on the preset logic. We recommend RPM based
4571 distributions to make use of these macros if possible. This
4572 makes it simpler to reuse RPM spec files across
4575 * We now make sure that the collected systemd unit name is
4576 always valid when services log to the journal via
4579 * There's a new man page kernel-command-line(7) detailing all
4580 command line options we understand.
4582 * The fstab generator may now be disabled at boot by passing
4583 fstab=0 on the kernel command line.
4585 * A new kernel command line option modules-load= is now understood
4586 to load a specific kernel module statically, early at boot.
4588 * Unit names specified on the systemctl command line are now
4589 automatically escaped as needed. Also, if file system or
4590 device paths are specified they are automatically turned
4591 into the appropriate mount or device unit names. Example:
4593 systemctl status /home
4594 systemctl status /dev/sda
4596 * The SysVConsole= configuration option has been removed from
4597 system.conf parsing.
4599 * The SysV search path is no longer exported on the D-Bus
4602 * The Names= option is been removed from unit file parsing.
4604 * There's a new man page bootup(7) detailing the boot process.
4606 * Every unit and every generator we ship with systemd now
4607 comes with full documentation. The self-explanatory boot is
4610 * A couple of services gained "systemd-" prefixes in their
4611 name if they wrap systemd code, rather than only external
4612 code. Among them fsck@.service which is now
4613 systemd-fsck@.service.
4615 * The HaveWatchdog property has been removed from the D-Bus
4618 * systemd.confirm_spawn= on the kernel command line should now
4621 * There's a new man page crypttab(5) which details all options
4622 we actually understand.
4624 * systemd-nspawn gained a new --capability= switch to pass
4625 additional capabilities to the container.
4627 * timedated will now read known NTP implementation unit names
4628 from /usr/lib/systemd/ntp-units.d/*.list,
4629 systemd-timedated-ntp.target has been removed.
4631 * journalctl gained a new switch "-b" that lists log data of
4632 the current boot only.
4634 * The notify socket is in the abstract namespace again, in
4635 order to support daemons which chroot() at start-up.
4637 * There is a new Storage= configuration option for journald
4638 which allows configuration of where log data should go. This
4639 also provides a way to disable journal logging entirely, so
4640 that data collected is only forwarded to the console, the
4641 kernel log buffer or another syslog implementation.
4643 * Many bugfixes and optimizations
4645 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
4646 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
4647 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
4648 Shawn Landden, Tom Gundersen
4652 * "systemctl help <unit>" now shows the man page if one is
4655 * Several new man pages have been added.
4657 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
4658 MaxLevelConsole= can now be specified in
4659 journald.conf. These options allow reducing the amount of
4660 data stored on disk or forwarded by the log level.
4662 * TimerSlackNSec= can now be specified in system.conf for
4663 PID1. This allows system-wide power savings.
4665 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
4666 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
4671 * logind is now capable of (optionally) handling power and
4672 sleep keys as well as the lid switch.
4674 * journalctl now understands the syntax "journalctl
4675 /usr/bin/avahi-daemon" to get all log output of a specific
4678 * CapabilityBoundingSet= in system.conf now also influences
4679 the capability bound set of usermode helpers of the kernel.
4681 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
4682 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
4683 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
4684 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
4688 * Note that we skipped 139 releases here in order to set the
4689 new version to something that is greater than both udev's
4690 and systemd's most recent version number.
4692 * udev: all udev sources are merged into the systemd source tree now.
4693 All future udev development will happen in the systemd tree. It
4694 is still fully supported to use the udev daemon and tools without
4695 systemd running, like in initramfs or other init systems. Building
4696 udev though, will require the *build* of the systemd tree, but
4697 udev can be properly *run* without systemd.
4699 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
4700 should be used to create dead device nodes as workarounds for broken
4703 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
4704 no longer supported. udev_monitor_new_from_netlink() needs to be
4705 used to subscribe to events.
4707 * udev: when udevd is started by systemd, processes which are left
4708 behind by forking them off of udev rules, are unconditionally cleaned
4709 up and killed now after the event handling has finished. Services or
4710 daemons must be started as systemd services. Services can be
4711 pulled-in by udev to get started, but they can no longer be directly
4712 forked by udev rules.
4714 * udev: the daemon binary is called systemd-udevd now and installed
4715 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
4716 to adapt to that, create symlink, or rename the binary after building
4719 * libudev no longer provides these symbols:
4720 udev_monitor_from_socket()
4721 udev_queue_get_failed_list_entry()
4722 udev_get_{dev,sys,run}_path()
4723 The versions number was bumped and symbol versioning introduced.
4725 * systemd-loginctl and systemd-journalctl have been renamed
4726 to loginctl and journalctl to match systemctl.
4728 * The config files: /etc/systemd/systemd-logind.conf and
4729 /etc/systemd/systemd-journald.conf have been renamed to
4730 logind.conf and journald.conf. Package updates should rename
4731 the files to the new names on upgrade.
4733 * For almost all files the license is now LGPL2.1+, changed
4734 from the previous GPL2.0+. Exceptions are some minor stuff
4735 of udev (which will be changed to LGPL2.1 eventually, too),
4736 and the MIT licensed sd-daemon.[ch] library that is suitable
4737 to be used as drop-in files.
4739 * systemd and logind now handle system sleep states, in
4740 particular suspending and hibernating.
4742 * logind now implements a sleep/shutdown/idle inhibiting logic
4743 suitable for a variety of uses. Soonishly Lennart will blog
4744 about this in more detail.
4746 * var-run.mount and var-lock.mount are no longer provided
4747 (which prevously bind mounted these directories to their new
4748 places). Distributions which have not converted these
4749 directories to symlinks should consider stealing these files
4750 from git history and add them downstream.
4752 * We introduced the Documentation= field for units and added
4753 this to all our shipped units. This is useful to make it
4754 easier to explore the boot and the purpose of the various
4757 * All smaller setup units (such as
4758 systemd-vconsole-setup.service) now detect properly if they
4759 are run in a container and are skipped when
4760 appropriate. This guarantees an entirely noise-free boot in
4761 Linux container environments such as systemd-nspawn.
4763 * A framework for implementing offline system updates is now
4764 integrated, for details see:
4765 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
4767 * A new service type Type=idle is available now which helps us
4768 avoiding ugly interleaving of getty output and boot status
4771 * There's now a system-wide CapabilityBoundingSet= option to
4772 globally reduce the set of capabilities for the
4773 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
4774 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
4775 even CAP_NET_ADMIN system-wide for secure systems.
4777 * There are now system-wide DefaultLimitXXX= options to
4778 globally change the defaults of the various resource limits
4779 for all units started by PID 1.
4781 * Harald Hoyer's systemd test suite has been integrated into
4782 systemd which allows easy testing of systemd builds in qemu
4783 and nspawn. (This is really awesome! Ask us for details!)
4785 * The fstab parser is now implemented as generator, not inside
4788 * systemctl will now warn you if .mount units generated from
4789 /etc/fstab are out of date due to changes in fstab that
4790 have not been read by systemd yet.
4792 * systemd is now suitable for usage in initrds. Dracut has
4793 already been updated to make use of this. With this in place
4794 initrds get a slight bit faster but primarily are much
4795 easier to introspect and debug since "systemctl status" in
4796 the host system can be used to introspect initrd services,
4797 and the journal from the initrd is kept around too.
4799 * systemd-delta has been added, a tool to explore differences
4800 between user/admin configuration and vendor defaults.
4802 * PrivateTmp= now affects both /tmp and /var/tmp.
4804 * Boot time status messages are now much prettier and feature
4805 proper english language. Booting up systemd has never been
4808 * Read-ahead pack files now include the inode number of all
4809 files to pre-cache. When the inode changes the pre-caching
4810 is not attempted. This should be nicer to deal with updated
4811 packages which might result in changes of read-ahead
4814 * We now temporaritly lower the kernel's read_ahead_kb variable
4815 when collecting read-ahead data to ensure the kernel's
4816 built-in read-ahead does not add noise to our measurements
4817 of necessary blocks to pre-cache.
4819 * There's now RequiresMountsFor= to add automatic dependencies
4820 for all mounts necessary for a specific file system path.
4822 * MountAuto= and SwapAuto= have been removed from
4823 system.conf. Mounting file systems at boot has to take place
4826 * nspawn now learned a new switch --uuid= to set the machine
4827 ID on the command line.
4829 * nspawn now learned the -b switch to automatically search
4832 * vt102 is now the default TERM for serial TTYs, upgraded from
4835 * systemd-logind now works on VT-less systems.
4837 * The build tree has been reorganized. The individual
4838 components now have directories of their own.
4840 * A new condition type ConditionPathIsReadWrite= is now available.
4842 * nspawn learned the new -C switch to create cgroups for the
4843 container in other hierarchies.
4845 * We now have support for hardware watchdogs, configurable in
4848 * The scheduled shutdown logic now has a public API.
4850 * We now mount /tmp as tmpfs by default, but this can be
4851 masked and /etc/fstab can override it.
4853 * Since udisks does not make use of /media anymore we are not
4854 mounting a tmpfs on it anymore.
4856 * journalctl gained a new --local switch to only interleave
4857 locally generated journal files.
4859 * We can now load the IMA policy at boot automatically.
4861 * The GTK tools have been split off into a systemd-ui.
4863 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
4864 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
4865 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
4866 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
4867 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
4868 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
4869 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
4870 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
4871 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
4876 * This is mostly a bugfix release
4878 * Support optional initialization of the machine ID from the
4879 KVM or container configured UUID.
4881 * Support immediate reboots with "systemctl reboot -ff"
4883 * Show /etc/os-release data in systemd-analyze output
4885 * Many bugfixes for the journal, including endianness fixes and
4886 ensuring that disk space enforcement works
4888 * sd-login.h is C++ comptaible again
4890 * Extend the /etc/os-release format on request of the Debian
4893 * We now refuse non-UTF8 strings used in various configuration
4894 and unit files. This is done to ensure we do not pass invalid
4895 data over D-Bus or expose it elsewhere.
4897 * Register Mimo USB Screens as suitable for automatic seat
4900 * Read SELinux client context from journal clients in a race
4903 * Reorder configuration file lookup order. /etc now always
4904 overrides /run in order to allow the administrator to always
4905 and unconditionally override vendor supplied or
4906 automatically generated data.
4908 * The various user visible bits of the journal now have man
4909 pages. We still lack man pages for the journal API calls
4912 * We now ship all man pages in HTML format again in the
4915 Contributions from: Dave Reisner, Dirk Eibach, Frederic
4916 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
4917 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
4922 * This is mostly a bugfix release
4924 * systems lacking /etc/os-release are no longer supported.
4926 * Various functionality updates to libsystemd-login.so
4928 * Track class of PAM logins to distinguish greeters from
4931 Contributions from: Kay Sievers, Lennart Poettering, Michael
4936 * This is an important bugfix release for v41.
4938 * Building man pages is now optional which should be useful
4939 for those building systemd from git but unwilling to install
4942 * Watchdog support for supervising services is now usable. In
4943 a future release support for hardware watchdogs
4944 (i.e. /dev/watchdog) will be added building on this.
4946 * Service start rate limiting is now configurable and can be
4947 turned off per service. When a start rate limit is hit a
4948 reboot can automatically be triggered.
4950 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
4952 Contributions from: Benjamin Franzke, Bill Nottingham,
4953 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
4954 Schmidt, Michał Górny, Piotr Drąg
4958 * The systemd binary is installed /usr/lib/systemd/systemd now;
4959 An existing /sbin/init symlink needs to be adapted with the
4962 * The code that loads kernel modules has been ported to invoke
4963 libkmod directly, instead of modprobe. This means we do not
4964 support systems with module-init-tools anymore.
4966 * Watchdog support is now already useful, but still not
4969 * A new kernel command line option systemd.setenv= is
4970 understood to set system wide environment variables
4971 dynamically at boot.
4973 * We now limit the set of capabilities of systemd-journald.
4975 * We now set SIGPIPE to ignore by default, since it only is
4976 useful in shell pipelines, and has little use in general
4977 code. This can be disabled with IgnoreSIPIPE=no in unit
4980 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
4981 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
4986 * This is mostly a bugfix release
4988 * We now expose the reason why a service failed in the
4989 "Result" D-Bus property.
4991 * Rudimentary service watchdog support (will be completed over
4992 the next few releases.)
4994 * When systemd forks off in order execute some service we will
4995 now immediately changes its argv[0] to reflect which process
4996 it will execute. This is useful to minimize the time window
4997 with a generic argv[0], which makes bootcharts more useful
4999 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
5000 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
5001 Mike Kazantsev, Ray Strode
5005 * This is mostly a test release, but incorporates many
5008 * New systemd-cgtop tool to show control groups by their
5011 * Linking against libacl for ACLs is optional again. If
5012 disabled, support tracking device access for active logins
5013 goes becomes unavailable, and so does access to the user
5014 journals by the respective users.
5016 * If a group "adm" exists, journal files are automatically
5017 owned by them, thus allow members of this group full access
5018 to the system journal as well as all user journals.
5020 * The journal now stores the SELinux context of the logging
5021 client for all entries.
5023 * Add C++ inclusion guards to all public headers
5025 * New output mode "cat" in the journal to print only text
5026 messages, without any meta data like date or time.
5028 * Include tiny X server wrapper as a temporary stop-gap to
5029 teach XOrg udev display enumeration. This is used by display
5030 managers such as gdm, and will go away as soon as XOrg
5031 learned native udev hotplugging for display devices.
5033 * Add new systemd-cat tool for executing arbitrary programs
5034 with STDERR/STDOUT connected to the journal. Can also act as
5035 BSD logger replacement, and does so by default.
5037 * Optionally store all locally generated coredumps in the
5038 journal along with meta data.
5040 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
5041 writing short strings to files (for usage for /sys), and for
5042 creating symlinks, character and block device nodes.
5044 * New unit file option ControlGroupPersistent= to make cgroups
5045 persistent, following the mechanisms outlined in
5046 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
5048 * Support multiple local RTCs in a sane way
5050 * No longer monopolize IO when replaying readahead data on
5051 rotating disks, since we might starve non-file-system IO to
5052 death, since fanotify() will not see accesses done by blkid,
5055 * Do not show kernel threads in systemd-cgls anymore, unless
5056 requested with new -k switch.
5058 Contributions from: Dan Horák, Kay Sievers, Lennart
5059 Poettering, Michal Schmidt
5063 * This is mostly a test release, but incorporates many
5066 * The git repository moved to:
5067 git://anongit.freedesktop.org/systemd/systemd
5068 ssh://git.freedesktop.org/git/systemd/systemd
5070 * First release with the journal
5071 http://0pointer.de/blog/projects/the-journal.html
5073 * The journal replaces both systemd-kmsg-syslogd and
5074 systemd-stdout-bridge.
5076 * New sd_pid_get_unit() API call in libsystemd-logind
5078 * Many systemadm clean-ups
5080 * Introduce remote-fs-pre.target which is ordered before all
5081 remote mounts and may be used to start services before all
5084 * Added Mageia support
5086 * Add bash completion for systemd-loginctl
5088 * Actively monitor PID file creation for daemons which exit in
5089 the parent process before having finished writing the PID
5090 file in the daemon process. Daemons which do this need to be
5091 fixed (i.e. PID file creation must have finished before the
5092 parent exits), but we now react a bit more gracefully to them.
5094 * Add colourful boot output, mimicking the well-known output
5095 of existing distributions.
5097 * New option PassCredentials= for socket units, for
5098 compatibility with a recent kernel ABI breakage.
5100 * /etc/rc.local is now hooked in via a generator binary, and
5101 thus will no longer act as synchronization point during
5104 * systemctl list-unit-files now supports --root=.
5106 * systemd-tmpfiles now understands two new commands: z, Z for
5107 relabelling files according to the SELinux database. This is
5108 useful to apply SELinux labels to specific files in /sys,
5111 * Output of SysV services is now forwarded to both the console
5112 and the journal by default, not only just the console.
5114 * New man pages for all APIs from libsystemd-login.
5116 * The build tree got reorganized and a the build system is a
5117 lot more modular allowing embedded setups to specifically
5118 select the components of systemd they are interested in.
5120 * Support for Linux systems lacking the kernel VT subsystem is
5123 * configure's --with-rootdir= got renamed to
5124 --with-rootprefix= to follow the naming used by udev and
5127 * Unless specified otherwise we will now install to /usr instead
5128 of /usr/local by default.
5130 * Processes with '@' in argv[0][0] are now excluded from the
5131 final shut-down killing spree, following the logic explained
5133 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
5135 * All processes remaining in a service cgroup when we enter
5136 the START or START_PRE states are now killed with
5137 SIGKILL. That means it is no longer possible to spawn
5138 background processes from ExecStart= lines (which was never
5139 supported anyway, and bad style).
5141 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
5142 reloading of units together.
5144 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
5145 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
5146 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
5147 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
5148 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek