1 systemd System and Service Manager
5 * machinectl gained a new verb 'shell' which opens a fresh shell on the
6 target machine. It is similar to 'login', but spawns the shell
7 directly. The pseudo machine '.host' now refers to the local host and
8 is used by default. Hence, 'machinectl shell' can be used as
9 replacement for 'su' which spawns the session as a fresh systemd
12 * systemd-networkd learned to cope with private-zone DHCP options and
13 allows other programs to query the values.
15 Contributions from: Alastair Hughes, Alex Crawford, Daniel Mack, David
16 Herrmann, Dimitri John Ledkov, Eric Kostrowski, Evgeny Vereshchagin,
17 Felipe Sateler, HATAYAMA Daisuke, Jan Pokorný, Jan Synacek, Johnny
18 Robeson, Karel Zak, Kay Sievers, Kefeng Wang, Lennart Poettering, Major
19 Hayden, Marcel Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt,
20 Matt Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim, Nicolas
21 Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer, reverendhomer,
22 Richard Maw, Ronny Chevalier, Seth Jennings, Stef Walter, Susant Sahani,
23 Thomas Blume, Thomas Hindoe Paaboel Andersen, Thomas Meyer, Tom
24 Gundersen, Vincent Batts, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
30 * The systemd-efi-boot-generator functionality was merged into
31 systemd-gpt-auto-generator.
33 * systemd-networkd now supports Group Policy for vxlan devices. It can
34 be enabled via the new boolean configuration option called
35 'GroupPolicyExtension='.
37 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
38 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
39 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
45 * The python-systemd code has been removed from the systemd repository.
46 A new repository has been created which accommodates the code from
47 now on, and we kindly ask distributions to create a separate package
48 for this: https://github.com/systemd/python-systemd
50 * The systemd daemon will now reload its main configuration
51 (/etc/systemd/system.conf) on daemon-reload.
53 * sd-dhcp now exposes vendor specific extensions via
54 sd_dhcp_lease_get_vendor_specific().
56 * systemd-networkd gained a number of new configuration options.
58 - A new boolean configuration option for TAP devices called
59 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
60 device, thus allowing to send and receive GSO packets.
62 - A new tunnel configuration option called 'CopyDSCP='.
63 If enabled, the DSCP field of ip6 tunnels is copied into the
66 - A set of boolean bridge configuration options were added.
67 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
68 and 'UnicastFlood=' are now parsed by networkd and applied to the
69 respective bridge link device via the respective IFLA_BRPORT_*
72 - A new string configuration option to override the hostname sent
73 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
74 is true, networkd will use the configured hostname instead of the
75 system hostname when sending DHCP requests.
77 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
78 networkd will configure the IPv6 flow-label of the tunnel device
81 - The 'macvtap' virtual network devices are now supported, similar to
82 the already supported 'macvlan' devices.
84 * systemd-resolved now implements RFC5452 to improve resilience against
85 cache poisoning. Additionally, source port randomization is enabled
86 by default to further protect against DNS spoofing attacks.
88 * nss-mymachines now supports translating UIDs and GIDs of running
89 containers with user-namespaces enabled. If a container 'foo'
90 translates a host uid 'UID' to the container uid 'TUID', then
91 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
92 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
93 mapped as 'vg-foo-TGID'.
95 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
96 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
97 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
98 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
99 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
100 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
101 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
102 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
103 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
104 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
106 -- Berlin, 2015-07-29
110 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
111 There are no known issues with current sysfs, and udev does not need
112 or should be used to work around such bugs.
114 * udev does no longer enable USB HID power management. Several reports
115 indicate, that some devices cannot handle that setting.
117 * The udev accelerometer helper was removed. The functionality
118 is now fully included in iio-sensor-proxy. But this means,
119 older iio-sensor-proxy versions will no longer provide
120 accelerometer/orientation data with this systemd version.
121 Please upgrade iio-sensor-proxy to version 1.0.
123 * networkd gained a new configuration option IPv6PrivacyExtensions=
124 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
125 for Stateless Address") on selected networks.
127 * For the sake of fewer build-time dependencies and less code in the
128 main repository, the python bindings are about to be removed in the
129 next release. A new repository has been created which accommodates
130 the code from now on, and we kindly ask distributions to create a
131 separate package for this. The removal will take place in v223.
133 https://github.com/systemd/python-systemd
135 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
136 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
137 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
138 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
139 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
140 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
141 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
142 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
143 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
144 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
146 -- Berlin, 2015-07-07
150 * The sd-bus.h and sd-event.h APIs have now been declared
151 stable and have been added to the official interface of
152 libsystemd.so. sd-bus implements an alternative D-Bus client
153 library, that is relatively easy to use, very efficient and
154 supports both classic D-Bus as well as kdbus as transport
155 backend. sd-event is a generic event loop abstraction that
156 is built around Linux epoll, but adds features such as event
157 prioritization or efficient timer handling. Both APIs are good
158 choices for C programs looking for a bus and/or event loop
159 implementation that is minimal and does not have to be
160 portable to other kernels.
162 * kdbus support is no longer compile-time optional. It is now
163 always built-in. However, it can still be disabled at
164 runtime using the kdbus=0 kernel command line setting, and
165 that setting may be changed to default to off, by specifying
166 --disable-kdbus at build-time. Note though that the kernel
167 command line setting has no effect if the kdbus.ko kernel
168 module is not installed, in which case kdbus is (obviously)
169 also disabled. We encourage all downstream distributions to
170 begin testing kdbus by adding it to the kernel images in the
171 development distributions, and leaving kdbus support in
174 * The minimal required util-linux version has been bumped to
177 * Support for chkconfig (--enable-chkconfig) was removed in
178 favor of calling an abstraction tool
179 /lib/systemd/systemd-sysv-install. This needs to be
180 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
181 in README for details.
183 * If there's a systemd unit and a SysV init script for the
184 same service name, and the user executes "systemctl enable"
185 for it (or a related call), then this will now enable both
186 (or execute the related operation on both), not just the
189 * The libudev API documentation has been converted from gtkdoc
192 * gudev has been removed from the systemd tree, it is now an
195 * The systemd-cgtop tool learnt a new --raw switch to generate
196 "raw" (machine parsable) output.
198 * networkd's IPForwarding= .network file setting learnt the
199 new setting "kernel", which ensures that networkd does not
200 change the IP forwarding sysctl from the default kernel
203 * The systemd-logind bus API now exposes a new boolean
204 property "Docked" that reports whether logind considers the
205 system "docked", i.e. connected to a docking station or not.
207 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
208 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
209 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
210 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
211 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
212 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
213 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
214 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
215 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
216 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
217 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
218 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
219 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
220 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
221 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
222 Fink, Zbigniew Jędrzejewski-Szmek
224 -- Berlin, 2015-06-19
228 * The gudev library has been extracted into a separate repository
229 available at: https://git.gnome.org/browse/libgudev/
230 It is now managed as part of the Gnome project. Distributions
231 are recommended to pass --disable-gudev to systemd and use
232 gudev from the Gnome project instead. gudev is still included
233 in systemd, for now. It will be removed soon, though. Please
234 also see the announcement-thread on systemd-devel:
235 http://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
237 * systemd now exposes a CPUUsageNSec= property for each
238 service unit on the bus, that contains the overall consumed
239 CPU time of a service (the sum of what each process of the
240 service consumed). This value is only available if
241 CPUAccounting= is turned on for a service, and is then shown
242 in the "systemctl status" output.
244 * Support for configuring alternative mappings of the old SysV
245 runlevels to systemd targets has been removed. They are now
246 hardcoded in a way that runlevels 2, 3, 4 all map to
247 multi-user.target and 5 to graphical.target (which
248 previously was already the default behaviour).
250 * The auto-mounter logic gained support for mount point
251 expiry, using a new TimeoutIdleSec= setting in .automount
252 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
254 * The EFI System Partition (ESP) as mounted to /boot by
255 systemd-efi-boot-generator will now be unmounted
256 automatically after 2 minutes of not being used. This should
257 minimize the risk of ESP corruptions.
259 * New /etc/fstab options x-systemd.requires= and
260 x-systemd.requires-mounts-for= are now supported to express
261 additional dependencies for mounts. This is useful for
262 journalling file systems that support external journal
263 devices or overlay file systems that require underlying file
264 systems to be mounted.
266 * systemd does not support direct live-upgrades (via systemctl
267 daemon-reexec) from versions older than v44 anymore. As no
268 distribution we are aware of shipped such old versions in a
269 stable release this should not be problematic.
271 * When systemd forks off a new per-connection service instance
272 it will now set the $REMOTE_ADDR environment variable to the
273 remote IP address, and $REMOTE_PORT environment variable to
274 the remote IP port. This behaviour is similar to the
275 corresponding environment variables defined by CGI.
277 * systemd-networkd gained support for uplink failure
278 detection. The BindCarrier= option allows binding interface
279 configuration dynamically to the link sense of other
280 interfaces. This is useful to achieve behaviour like in
283 * systemd-networkd gained support for configuring the DHCP
284 client identifier to use when requesting leases.
286 * systemd-networkd now has a per-network UseNTP= option to
287 configure whether NTP server information acquired via DHCP
288 is passed on to services like systemd-timesyncd.
290 * systemd-networkd gained support for vti6 tunnels.
292 * Note that systemd-networkd manages the sysctl variable
293 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
294 it is configured for since v219. The variable controls IP
295 forwarding, and is a per-interface alternative to the global
296 /proc/sys/net/ipv[46]/ip_forward. This setting is
297 configurable in the IPForward= option, which defaults to
298 "no". This means if networkd is used for an interface it is
299 no longer sufficient to set the global sysctl option to turn
300 on IP forwarding! Instead, the .network file option
301 IPForward= needs to be turned on! Note that the
302 implementation of this behaviour was broken in v219 and has
305 * Many bonding and vxlan options are now configurable in
308 * systemd-nspawn gained a new --property= setting to set unit
309 properties for the container scope. This is useful for
310 setting resource parameters (e.g "CPUShares=500") on
311 containers started from the command line.
313 * systemd-nspawn gained a new --private-users= switch to make
314 use of user namespacing available on recent Linux kernels.
316 * systemd-nspawn may now be called as part of a shell pipeline
317 in which case the pipes used for stdin and stdout are passed
318 directly to the process invoked in the container, without
319 indirection via a pseudo tty.
321 * systemd-nspawn gained a new switch to control the UNIX
322 signal to use when killing the init process of the container
325 * systemd-nspawn gained a new --overlay= switch for mounting
326 overlay file systems into the container using the new kernel
329 * When a container image is imported via systemd-importd and
330 the host file system is not btrfs, a loopback block device
331 file is created in /var/lib/machines.raw with a btrfs file
332 system inside. It is then mounted to /var/lib/machines to
333 enable btrfs features for container management. The loopback
334 file and btrfs file system is grown as needed when container
335 images are imported via systemd-importd.
337 * systemd-machined/systemd-importd gained support for btrfs
338 quota, to enforce container disk space limits on disk. This
339 is exposed in "machinectl set-limit".
341 * systemd-importd now can import containers from local .tar,
342 .raw and .qcow2 images, and export them to .tar and .raw. It
343 can also import dkr v2 images now from the network (on top
346 * systemd-importd gained support for verifying downloaded
347 images with gpg2 (previously only gpg1 was supported).
349 * systemd-machined, systemd-logind, systemd: most bus calls
350 are now accessible to unprivileged processes via
351 PolicyKit. Also, systemd-logind will now allow users to kill
352 their own sessions without further privileges or
355 * systemd-shutdownd has been removed. This service was
356 previously responsible for implementing scheduled shutdowns
357 as exposed in /usr/bin/shutdown's time parameter. This
358 functionality has now been moved into systemd-logind and is
359 accessible via a bus interface.
361 * "systemctl reboot" gained a new switch --firmware-setup that
362 can be used to reboot into the EFI firmware setup, if that
363 is available. systemd-logind now exposes an API on the bus
364 to trigger such reboots, in case graphical desktop UIs want
365 to cover this functionality.
367 * "systemctl enable", "systemctl disable" and "systemctl mask"
368 now support a new "--now" switch. If specified the units
369 that are enabled will also be started, and the ones
370 disabled/masked also stopped.
372 * The Gummiboot EFI boot loader tool has been merged into
373 systemd, and renamed to "systemd-boot". The bootctl tool has been
374 updated to support systemd-boot.
376 * An EFI kernel stub has been added that may be used to create
377 kernel EFI binaries that contain not only the actual kernel,
378 but also an initrd, boot splash, command line and OS release
379 information. This combined binary can then be signed as a
380 single image, so that the firmware can verify it all in one
381 step. systemd-boot has special support for EFI binaries created
382 like this and can extract OS release information from them
383 and show them in the boot menu. This functionality is useful
384 to implement cryptographically verified boot schemes.
386 * Optional support has been added to systemd-fsck to pass
387 fsck's progress report to an AF_UNIX socket in the file
390 * udev will no longer create device symlinks for all block
391 devices by default. A blacklist for excluding special block
392 devices from this logic has been turned into a whitelist
393 that requires picking block devices explicitly that require
396 * A new (currently still internal) API sd-device.h has been
397 added to libsystemd. This modernized API is supposed to
398 replace libudev eventually. In fact, already much of libudev
399 is now just a wrapper around sd-device.h.
401 * A new hwdb database for storing metadata about pointing
402 stick devices has been added.
404 * systemd-tmpfiles gained support for setting file attributes
405 similar to the "chattr" tool with new 'h' and 'H' lines.
407 * systemd-journald will no longer unconditionally set the
408 btrfs NOCOW flag on new journal files. This is instead done
409 with tmpfiles snippet using the new 'h' line type. This
410 allows easy disabling of this logic, by masking the
411 journal-nocow.conf tmpfiles file.
413 * systemd-journald will now translate audit message types to
414 human readable identifiers when writing them to the
415 journal. This should improve readability of audit messages.
417 * The LUKS logic gained support for the offset= and skip=
418 options in /etc/crypttab, as previously implemented by
421 * /usr/lib/os-release gained a new optional field VARIANT= for
422 distributions that support multiple variants (such as a
423 desktop edition, a server edition, ...)
425 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
426 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
427 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
428 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
429 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
430 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
431 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
432 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
433 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
434 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
435 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
436 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
437 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
438 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
439 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
440 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
441 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
442 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
443 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
444 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
445 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
446 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
447 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
448 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
449 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
450 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
451 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
453 -- Berlin, 2015-05-22
457 * Introduce a new API "sd-hwdb.h" for querying the hardware
458 metadata database. With this minimal interface one can query
459 and enumerate the udev hwdb, decoupled from the old libudev
460 library. libudev's interface for this is now only a wrapper
461 around sd-hwdb. A new tool systemd-hwdb has been added to
462 interface with and update the database.
464 * When any of systemd's tools copies files (for example due to
465 tmpfiles' C lines) a btrfs reflink will attempted first,
466 before bytewise copying is done.
468 * systemd-nspawn gained a new --ephemeral switch. When
469 specified a btrfs snapshot is taken of the container's root
470 directory, and immediately removed when the container
471 terminates again. Thus, a container can be started whose
472 changes never alter the container's root directory, and are
473 lost on container termination. This switch can also be used
474 for starting a container off the root file system of the
475 host without affecting the host OS. This switch is only
476 available on btrfs file systems.
478 * systemd-nspawn gained a new --template= switch. It takes the
479 path to a container tree to use as template for the tree
480 specified via --directory=, should that directory be
481 missing. This allows instantiating containers dynamically,
482 on first run. This switch is only available on btrfs file
485 * When a .mount unit refers to a mount point on which multiple
486 mounts are stacked, and the .mount unit is stopped all of
487 the stacked mount points will now be unmounted until no
490 * systemd now has an explicit notion of supported and
491 unsupported unit types. Jobs enqueued for unsupported unit
492 types will now fail with an "unsupported" error code. More
493 specifically .swap, .automount and .device units are not
494 supported in containers, .busname units are not supported on
495 non-kdbus systems. .swap and .automount are also not
496 supported if their respective kernel compile time options
499 * machinectl gained support for two new "copy-from" and
500 "copy-to" commands for copying files from a running
501 container to the host or vice versa.
503 * machinectl gained support for a new "bind" command to bind
504 mount host directories into local containers. This is
505 currently only supported for nspawn containers.
507 * networkd gained support for configuring bridge forwarding
508 database entries (fdb) from .network files.
510 * A new tiny daemon "systemd-importd" has been added that can
511 download container images in tar, raw, qcow2 or dkr formats,
512 and make them available locally in /var/lib/machines, so
513 that they can run as nspawn containers. The daemon can GPG
514 verify the downloads (not supported for dkr, since it has no
515 provisions for verifying downloads). It will transparently
516 decompress bz2, xz, gzip compressed downloads if necessary,
517 and restore sparse files on disk. The daemon uses privilege
518 separation to ensure the actual download logic runs with
519 fewer privileges than the daemon itself. machinectl has
520 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
521 make the functionality of importd available to the
522 user. With this in place the Fedora and Ubuntu "Cloud"
523 images can be downloaded and booted as containers unmodified
524 (the Fedora images lack the appropriate GPG signature files
525 currently, so they cannot be verified, but this will change
526 soon, hopefully). Note that downloading images is currently
527 only fully supported on btrfs.
529 * machinectl is now able to list container images found in
530 /var/lib/machines, along with some metadata about sizes of
531 disk and similar. If the directory is located on btrfs and
532 quota is enabled, this includes quota display. A new command
533 "image-status" has been added that shows additional
534 information about images.
536 * machinectl is now able to clone container images
537 efficiently, if the underlying file system (btrfs) supports
538 it, with the new "machinectl list-images" command. It also
539 gained commands for renaming and removing images, as well as
540 marking them read-only or read-write (supported also on
541 legacy file systems).
543 * networkd gained support for collecting LLDP network
544 announcements, from hardware that supports this. This is
545 shown in networkctl output.
547 * systemd-run gained support for a new -t (--pty) switch for
548 invoking a binary on a pty whose input and output is
549 connected to the invoking terminal. This allows executing
550 processes as system services while interactively
551 communicating with them via the terminal. Most interestingly
552 this is supported across container boundaries. Invoking
553 "systemd-run -t /bin/bash" is an alternative to running a
554 full login session, the difference being that the former
555 will not register a session, nor go through the PAM session
558 * tmpfiles gained support for a new "v" line type for creating
559 btrfs subvolumes. If the underlying file system is a legacy
560 file system, this automatically degrades to creating a
561 normal directory. Among others /var/lib/machines is now
562 created like this at boot, should it be missing.
564 * The directory /var/lib/containers/ has been deprecated and
565 been replaced by /var/lib/machines. The term "machines" has
566 been used in the systemd context as generic term for both
567 VMs and containers, and hence appears more appropriate for
568 this, as the directory can also contain raw images bootable
571 * systemd-nspawn when invoked with -M but without --directory=
572 or --image= is now capable of searching for the container
573 root directory, subvolume or disk image automatically, in
574 /var/lib/machines. systemd-nspawn@.service has been updated
575 to make use of this, thus allowing it to be used for raw
578 * A new machines.target unit has been introduced that is
579 supposed to group all containers/VMs invoked as services on
580 the system. systemd-nspawn@.service has been updated to
583 * machinectl gained a new "start" command, for invoking a
584 container as a service. "machinectl start foo" is mostly
585 equivalent to "systemctl start systemd-nspawn@foo.service",
586 but handles escaping in a nicer way.
588 * systemd-nspawn will now mount most of the cgroupfs tree
589 read-only into each container, with the exception of the
590 container's own subtree in the name=systemd hierarchy.
592 * journald now sets the special FS_NOCOW file flag for its
593 journal files. This should improve performance on btrfs, by
594 avoiding heavy fragmentation when journald's write-pattern
595 is used on COW file systems. It degrades btrfs' data
596 integrity guarantees for the files to the same levels as for
597 ext3/ext4 however. This should be OK though as journald does
598 its own data integrity checks and all its objects are
599 checksummed on disk. Also, journald should handle btrfs disk
600 full events a lot more gracefully now, by processing SIGBUS
601 errors, and not relying on fallocate() anymore.
603 * When journald detects that journal files it is writing to
604 have been deleted it will immediately start new journal
607 * systemd now provides a way to store file descriptors
608 per-service in PID 1.This is useful for daemons to ensure
609 that fds they require are not lost during a daemon
610 restart. The fds are passed to the daemon on the next
611 invocation in the same way socket activation fds are
612 passed. This is now used by journald to ensure that the
613 various sockets connected to all the system's stdout/stderr
614 are not lost when journald is restarted. File descriptors
615 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
616 an extension to sd_notify(). Note that a limit is enforced
617 on the number of fds a service can store in PID 1, and it
618 defaults to 0, so that no fds may be stored, unless this is
619 explicitly turned on.
621 * The default TERM variable to use for units connected to a
622 terminal, when no other value is explicitly is set is now
623 vt220 rather than vt102. This should be fairly safe still,
624 but allows PgUp/PgDn work.
626 * The /etc/crypttab option header= as known from Debian is now
629 * "loginctl user-status" and "loginctl session-status" will
630 now show the last 10 lines of log messages of the
631 user/session following the status output. Similar,
632 "machinectl status" will show the last 10 log lines
633 associated with a virtual machine or container
634 service. (Note that this is usually not the log messages
635 done in the VM/container itself, but simply what the
636 container manager logs. For nspawn this includes all console
639 * "loginctl session-status" without further argument will now
640 show the status of the session of the caller. Similar,
641 "lock-session", "unlock-session", "activate",
642 "enable-linger", "disable-linger" may now be called without
643 session/user parameter in which case they apply to the
644 caller's session/user.
646 * An X11 session scriptlet is now shipped that uploads
647 $DISPLAY and $XAUTHORITY into the environment of the systemd
648 --user daemon if a session begins. This should improve
649 compatibility with X11 enabled applications run as systemd
652 * Generators are now subject to masking via /etc and /run, the
653 same way as unit files.
655 * networkd .network files gained support for configuring
656 per-link IPv4/IPv6 packet forwarding as well as IPv4
657 masquerading. This is by default turned on for veth links to
658 containers, as registered by systemd-nspawn. This means that
659 nspawn containers run with --network-veth will now get
660 automatic routed access to the host's networks without any
661 further configuration or setup, as long as networkd runs on
664 * systemd-nspawn gained the --port= (-p) switch to expose TCP
665 or UDP posts of a container on the host. With this in place
666 it is possible to run containers with private veth links
667 (--network-veth), and have their functionality exposed on
668 the host as if their services were running directly on the
671 * systemd-nspawn's --network-veth switch now gained a short
672 version "-n", since with the changes above it is now truly
673 useful out-of-the-box. The systemd-nspawn@.service has been
674 updated to make use of it too by default.
676 * systemd-nspawn will now maintain a per-image R/W lock, to
677 ensure that the same image is not started more than once
678 writable. (It's OK to run an image multiple times
679 simultaneously in read-only mode.)
681 * systemd-nspawn's --image= option is now capable of
682 dissecting and booting MBR and GPT disk images that contain
683 only a single active Linux partition. Previously it
684 supported only GPT disk images with proper GPT type
685 IDs. This allows running cloud images from major
686 distributions directly with systemd-nspawn, without
689 * In addition to collecting mouse dpi data in the udev
690 hardware database, there's now support for collecting angle
691 information for mouse scroll wheels. The database is
692 supposed to guarantee similar scrolling behavior on mice
693 that it knows about. There's also support for collecting
694 information about Touchpad types.
696 * udev's input_id built-in will now also collect touch screen
697 dimension data and attach it to probed devices.
699 * /etc/os-release gained support for a Distribution Privacy
702 * networkd gained support for creating "ipvlan", "gretap",
703 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
705 * systemd-tmpfiles gained support for "a" lines for setting
708 * systemd-nspawn will now mount /tmp in the container to
709 tmpfs, automatically.
711 * systemd now exposes the memory.usage_in_bytes cgroup
712 attribute and shows it for each service in the "systemctl
713 status" output, if available.
715 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
716 immediate reboot is triggered. This useful if shutdown is
717 hung and is unable to complete, to expedite the
718 operation. Note that this kind of reboot will still unmount
719 all file systems, and hence should not result in fsck being
722 * A .device unit for an optical block device will now be
723 considered active only when a medium is in the drive. Also,
724 mount units are now bound to their backing devices thus
725 triggering automatic unmounting when devices become
726 unavailable. With this in place systemd will now
727 automatically unmount left-over mounts when a CD-ROM is
728 ejected or an USB stick is yanked from the system.
730 * networkd-wait-online now has support for waiting for
731 specific interfaces only (with globbing), and for giving up
732 after a configurable timeout.
734 * networkd now exits when idle. It will be automatically
735 restarted as soon as interfaces show up, are removed or
736 change state. networkd will stay around as long as there is
737 at least one DHCP state machine or similar around, that keep
740 * networkd may now configure IPv6 link-local addressing in
741 addition to IPv4 link-local addressing.
743 * The IPv6 "token" for use in SLAAC may now be configured for
744 each .network interface in networkd.
746 * Routes configured with networkd may now be assigned a scope
749 * networkd's [Match] sections now support globbing and lists
750 of multiple space-separated matches per item.
752 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
753 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
754 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
755 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
756 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
757 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
758 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
759 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
760 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
761 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
762 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
763 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
764 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
765 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
766 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
767 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
768 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
769 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
770 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
771 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
772 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
773 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
774 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
775 Hoffmann, Zbigniew Jędrzejewski-Szmek
777 -- Berlin, 2015-02-16
781 * When querying unit file enablement status (for example via
782 "systemctl is-enabled"), a new state "indirect" is now known
783 which indicates that a unit might not be enabled itself, but
784 another unit listed in its Also= setting might be.
786 * Similar to the various existing ConditionXYZ= settings for
787 units there are now matching AssertXYZ= settings. While
788 failing conditions cause a unit to be skipped, but its job
789 to succeed, failing assertions declared like this will cause
790 a unit start operation and its job to fail.
792 * hostnamed now knows a new chassis type "embedded".
794 * systemctl gained a new "edit" command. When used on a unit
795 file this allows extending unit files with .d/ drop-in
796 configuration snippets or editing the full file (after
797 copying it from /usr/lib to /etc). This will invoke the
798 user's editor (as configured with $EDITOR), and reload the
799 modified configuration after editing.
801 * "systemctl status" now shows the suggested enablement state
802 for a unit, as declared in the (usually vendor-supplied)
805 * nss-myhostname will now resolve the single-label host name
806 "gateway" to the locally configured default IP routing
807 gateways, ordered by their metrics. This assigns a stable
808 name to the used gateways, regardless which ones are
809 currently configured. Note that the name will only be
810 resolved after all other name sources (if nss-myhostname is
811 configured properly) and should hence not negatively impact
812 systems that use the single-label host name "gateway" in
815 * systemd-inhibit now allows filtering by mode when listing
818 * Scope and service units gained a new "Delegate" boolean
819 property, which when set allows processes running inside the
820 unit to further partition resources. This is primarily
821 useful for systemd user instances as well as container
824 * journald will now pick up audit messages directly from
825 the kernel, and log them like any other log message. The
826 audit fields are split up and fully indexed. This means that
827 journalctl in many ways is now a (nicer!) alternative to
828 ausearch, the traditional audit client. Note that this
829 implements only a minimal audit client, if you want the
830 special audit modes like reboot-on-log-overflow, please use
831 the traditional auditd instead, which can be used in
832 parallel to journald.
834 * The ConditionSecurity= unit file option now understands the
835 special string "audit" to check whether auditing is
838 * journalctl gained two new commands --vacuum-size= and
839 --vacuum-time= to delete old journal files until the
840 remaining ones take up no more the specified size on disk,
841 or are not older than the specified time.
843 * A new, native PPPoE library has been added to sd-network,
844 systemd's library of light-weight networking protocols. This
845 library will be used in a future version of networkd to
846 enable PPPoE communication without an external pppd daemon.
848 * The busctl tool now understands a new "capture" verb that
849 works similar to "monitor", but writes a packet capture
850 trace to STDOUT that can be redirected to a file which is
851 compatible with libcap's capture file format. This can then
852 be loaded in Wireshark and similar tools to inspect bus
855 * The busctl tool now understands a new "tree" verb that shows
856 the object trees of a specific service on the bus, or of all
859 * The busctl tool now understands a new "introspect" verb that
860 shows all interfaces and members of objects on the bus,
861 including their signature and values. This is particularly
862 useful to get more information about bus objects shown by
863 the new "busctl tree" command.
865 * The busctl tool now understands new verbs "call",
866 "set-property" and "get-property" for invoking bus method
867 calls, setting and getting bus object properties in a
870 * busctl gained a new --augment-creds= argument that controls
871 whether the tool shall augment credential information it
872 gets from the bus with data from /proc, in a possibly
875 * nspawn's --link-journal= switch gained two new values
876 "try-guest" and "try-host" that work like "guest" and
877 "host", but do not fail if the host has no persistent
878 journalling enabled. -j is now equivalent to
879 --link-journal=try-guest.
881 * macvlan network devices created by nspawn will now have
882 stable MAC addresses.
884 * A new SmackProcessLabel= unit setting has been added, which
885 controls the SMACK security label processes forked off by
886 the respective unit shall use.
888 * If compiled with --enable-xkbcommon, systemd-localed will
889 verify x11 keymap settings by compiling the given keymap. It
890 will spew out warnings if the compilation fails. This
891 requires libxkbcommon to be installed.
893 * When a coredump is collected a larger number of metadata
894 fields is now collected and included in the journal records
895 created for it. More specifically control group membership,
896 environment variables, memory maps, working directory,
897 chroot directory, /proc/$PID/status, and a list of open file
898 descriptors is now stored in the log entry.
900 * The udev hwdb now contains DPI information for mice. For
903 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
905 * All systemd programs that read standalone configuration
906 files in /etc now also support a corresponding series of
907 .conf.d configuration directories in /etc/, /run/,
908 /usr/local/lib/, /usr/lib/, and (if configured with
909 --enable-split-usr) /lib/. In particular, the following
910 configuration files now have corresponding configuration
911 directories: system.conf user.conf, logind.conf,
912 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
913 resolved.conf, timesyncd.conf, journal-remote.conf, and
914 journal-upload.conf. Note that distributions should use the
915 configuration directories in /usr/lib/; the directories in
916 /etc/ are reserved for the system administrator.
918 * systemd-rfkill will no longer take the rfkill device name
919 into account when storing rfkill state on disk, as the name
920 might be dynamically assigned and not stable. Instead, the
921 ID_PATH udev variable combined with the rfkill type (wlan,
922 bluetooth, ...) is used.
924 * A new service systemd-machine-id-commit.service has been
925 added. When used on systems where /etc is read-only during
926 boot, and /etc/machine-id is not initialized (but an empty
927 file), this service will copy the temporary machine ID
928 created as replacement into /etc after the system is fully
929 booted up. This is useful for systems that are freshly
930 installed with a non-initialized machine ID, but should get
931 a fixed machine ID for subsequent boots.
933 * networkd's .netdev files now provide a large set of
934 configuration parameters for VXLAN devices. Similar, the
935 bridge port cost parameter is now configurable in .network
936 files. There's also new support for configuring IP source
937 routing. networkd .link files gained support for a new
938 OriginalName= match that is useful to match against the
939 original interface name the kernel assigned. .network files
940 may include MTU= and MACAddress= fields for altering the MTU
941 and MAC address while being connected to a specific network
944 * The LUKS logic gained supported for configuring
945 UUID-specific key files. There's also new support for naming
946 LUKS device from the kernel command line, using the new
949 * Timer units may now be transiently created via the bus API
950 (this was previously already available for scope and service
951 units). In addition it is now possible to create multiple
952 transient units at the same time with a single bus call. The
953 "systemd-run" tool has been updated to make use of this for
954 running commands on a specified time, in at(1)-style.
956 * tmpfiles gained support for "t" lines, for assigning
957 extended attributes to files. Among other uses this may be
958 used to assign SMACK labels to files.
960 Contributions from: Alin Rauta, Alison Chaiken, Andrej
961 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
962 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
963 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
964 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
965 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
966 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
967 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
968 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
969 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
970 Michael Biebl, Michael Chapman, Michael Marineau, Michal
971 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
972 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
973 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
974 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
975 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
976 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
977 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
979 -- Berlin, 2014-12-10
983 * journalctl gained the new options -t/--identifier= to match
984 on the syslog identifier (aka "tag"), as well as --utc to
985 show log timestamps in the UTC timezone. journalctl now also
986 accepts -n/--lines=all to disable line capping in a pager.
988 * journalctl gained a new switch, --flush, that synchronously
989 flushes logs from /run/log/journal to /var/log/journal if
990 persistent storage is enabled. systemd-journal-flush.service
991 now waits until the operation is complete.
993 * Services can notify the manager before they start a reload
994 (by sending RELOADING=1) or shutdown (by sending
995 STOPPING=1). This allows the manager to track and show the
996 internal state of daemons and closes a race condition when
997 the process is still running but has closed its D-Bus
1000 * Services with Type=oneshot do not have to have any ExecStart
1003 * User units are now loaded also from
1004 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
1005 /run/systemd/user directory that was already previously
1006 supported, but is under the control of the user.
1008 * Job timeouts (i.e. time-outs on the time a job that is
1009 queued stays in the run queue) can now optionally result in
1010 immediate reboot or power-off actions (JobTimeoutAction= and
1011 JobTimeoutRebootArgument=). This is useful on ".target"
1012 units, to limit the maximum time a target remains
1013 undispatched in the run queue, and to trigger an emergency
1014 operation in such a case. This is now used by default to
1015 turn off the system if boot-up (as defined by everything in
1016 basic.target) hangs and does not complete for at least
1017 15min. Also, if power-off or reboot hang for at least 30min
1018 an immediate power-off/reboot operation is triggered. This
1019 functionality is particularly useful to increase reliability
1020 on embedded devices, but also on laptops which might
1021 accidentally get powered on when carried in a backpack and
1022 whose boot stays stuck in a hard disk encryption passphrase
1025 * systemd-logind can be configured to also handle lid switch
1026 events even when the machine is docked or multiple displays
1027 are attached (HandleLidSwitchDocked= option).
1029 * A helper binary and a service have been added which can be
1030 used to resume from hibernation in the initramfs. A
1031 generator will parse the resume= option on the kernel
1032 command line to trigger resume.
1034 * A user console daemon systemd-consoled has been
1035 added. Currently, it is a preview, and will so far open a
1036 single terminal on each session of the user marked as
1037 Desktop=systemd-console.
1039 * Route metrics can be specified for DHCP routes added by
1042 * The SELinux context of socket-activated services can be set
1043 from the information provided by the networking stack
1044 (SELinuxContextFromNet= option).
1046 * Userspace firmware loading support has been removed and
1047 the minimum supported kernel version is thus bumped to 3.7.
1049 * Timeout for udev workers has been increased from 1 to 3
1050 minutes, but a warning will be printed after 1 minute to
1051 help diagnose kernel modules that take a long time to load.
1053 * Udev rules can now remove tags on devices with TAG-="foobar".
1055 * systemd's readahead implementation has been removed. In many
1056 circumstances it didn't give expected benefits even for
1057 rotational disk drives and was becoming less relevant in the
1058 age of SSDs. As none of the developers has been using
1059 rotating media anymore, and nobody stepped up to actively
1060 maintain this component of systemd it has now been removed.
1062 * Swap units can use Options= to specify discard options.
1063 Discard options specified for swaps in /etc/fstab are now
1066 * Docker containers are now detected as a separate type of
1069 * The Password Agent protocol gained support for queries where
1070 the user input is shown, useful e.g. for user names.
1071 systemd-ask-password gained a new --echo option to turn that
1074 * The default sysctl.d/ snippets will now set:
1076 net.core.default_qdisc = fq_codel
1078 This selects Fair Queuing Controlled Delay as the default
1079 queuing discipline for network interfaces. fq_codel helps
1080 fight the network bufferbloat problem. It is believed to be
1081 a good default with no tuning required for most workloads.
1082 Downstream distributions may override this choice. On 10Gbit
1083 servers that do not do forwarding, "fq" may perform better.
1084 Systems without a good clocksource should use "pfifo_fast".
1086 * If kdbus is enabled during build a new option BusPolicy= is
1087 available for service units, that allows locking all service
1088 processes into a stricter bus policy, in order to limit
1089 access to various bus services, or even hide most of them
1090 from the service's view entirely.
1092 * networkctl will now show the .network and .link file
1093 networkd has applied to a specific interface.
1095 * sd-login gained a new API call sd_session_get_desktop() to
1096 query which desktop environment has been selected for a
1099 * UNIX utmp support is now compile-time optional to support
1100 legacy-free systems.
1102 * systemctl gained two new commands "add-wants" and
1103 "add-requires" for pulling in units from specific targets
1106 * If the word "rescue" is specified on the kernel command line
1107 the system will now boot into rescue mode (aka
1108 rescue.target), which was previously available only by
1109 specifying "1" or "systemd.unit=rescue.target" on the kernel
1110 command line. This new kernel command line option nicely
1111 mirrors the already existing "emergency" kernel command line
1114 * New kernel command line options mount.usr=, mount.usrflags=,
1115 mount.usrfstype= have been added that match root=, rootflags=,
1116 rootfstype= but allow mounting a specific file system to
1119 * The $NOTIFY_SOCKET is now also passed to control processes of
1120 services, not only the main process.
1122 * This version reenables support for fsck's -l switch. This
1123 means at least version v2.25 of util-linux is required for
1124 operation, otherwise dead-locks on device nodes may
1125 occur. Again: you need to update util-linux to at least
1126 v2.25 when updating systemd to v217.
1128 * The "multi-seat-x" tool has been removed from systemd, as
1129 its functionality has been integrated into X servers 1.16,
1130 and the tool is hence redundant. It is recommended to update
1131 display managers invoking this tool to simply invoke X
1132 directly from now on, again.
1134 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
1135 message flag has been added for all of systemd's PolicyKit
1136 authenticated method calls has been added. In particular
1137 this now allows optional interactive authorization via
1138 PolicyKit for many of PID1's privileged operations such as
1139 unit file enabling and disabling.
1141 * "udevadm hwdb --update" learnt a new switch "--usr" for
1142 placing the rebuilt hardware database in /usr instead of
1143 /etc. When used only hardware database entries stored in
1144 /usr will be used, and any user database entries in /etc are
1145 ignored. This functionality is useful for vendors to ship a
1146 pre-built database on systems where local configuration is
1147 unnecessary or unlikely.
1149 * Calendar time specifications in .timer units now also
1150 understand the strings "semi-annually", "quarterly" and
1151 "minutely" as shortcuts (in addition to the preexisting
1152 "anually", "hourly", ...).
1154 * systemd-tmpfiles will now correctly create files in /dev
1155 at boot which are marked for creation only at boot. It is
1156 recommended to always create static device nodes with 'c!'
1157 and 'b!', so that they are created only at boot and not
1158 overwritten at runtime.
1160 * When the watchdog logic is used for a service (WatchdogSec=)
1161 and the watchdog timeout is hit the service will now be
1162 terminated with SIGABRT (instead of just SIGTERM), in order
1163 to make sure a proper coredump and backtrace is
1164 generated. This ensures that hanging services will result in
1165 similar coredump/backtrace behaviour as services that hit a
1168 Contributions from: Andreas Henriksson, Andrei Borzenkov,
1169 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
1170 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
1171 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
1172 Herrmann, David Sommerseth, David Strauss, Emil Renner
1173 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
1174 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
1175 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
1176 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
1177 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
1178 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
1179 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
1180 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
1181 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
1182 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
1183 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
1184 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
1185 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
1186 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
1187 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
1188 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
1191 -- Berlin, 2014-10-28
1195 * timedated no longer reads NTP implementation unit names from
1196 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
1197 implementations should add a
1199 Conflicts=systemd-timesyncd.service
1201 to their unit files to take over and replace systemd's NTP
1202 default functionality.
1204 * systemd-sysusers gained a new line type "r" for configuring
1205 which UID/GID ranges to allocate system users/groups
1206 from. Lines of type "u" may now add an additional column
1207 that specifies the home directory for the system user to be
1208 created. Also, systemd-sysusers may now optionally read user
1209 information from STDIN instead of a file. This is useful for
1210 invoking it from RPM preinst scriptlets that need to create
1211 users before the first RPM file is installed since these
1212 files might need to be owned by them. A new
1213 %sysusers_create_inline RPM macro has been introduced to do
1214 just that. systemd-sysusers now updates the shadow files as
1215 well as the user/group databases, which should enhance
1216 compatibility with certain tools like grpck.
1218 * A number of bus APIs of PID 1 now optionally consult
1219 PolicyKit to permit access for otherwise unprivileged
1220 clients under certain conditions. Note that this currently
1221 doesn't support interactive authentication yet, but this is
1222 expected to be added eventually, too.
1224 * /etc/machine-info now has new fields for configuring the
1225 deployment environment of the machine, as well as the
1226 location of the machine. hostnamectl has been updated with
1227 new command to update these fields.
1229 * systemd-timesyncd has been updated to automatically acquire
1230 NTP server information from systemd-networkd, which might
1231 have been discovered via DHCP.
1233 * systemd-resolved now includes a caching DNS stub resolver
1234 and a complete LLMNR name resolution implementation. A new
1235 NSS module "nss-resolve" has been added which can be used
1236 instead of glibc's own "nss-dns" to resolve hostnames via
1237 systemd-resolved. Hostnames, addresses and arbitrary RRs may
1238 be resolved via systemd-resolved D-Bus APIs. In contrast to
1239 the glibc internal resolver systemd-resolved is aware of
1240 multi-homed system, and keeps DNS server and caches separate
1241 and per-interface. Queries are sent simultaneously on all
1242 interfaces that have DNS servers configured, in order to
1243 properly handle VPNs and local LANs which might resolve
1244 separate sets of domain names. systemd-resolved may acquire
1245 DNS server information from systemd-networkd automatically,
1246 which in turn might have discovered them via DHCP. A tool
1247 "systemd-resolve-host" has been added that may be used to
1248 query the DNS logic in resolved. systemd-resolved implements
1249 IDNA and automatically uses IDNA or UTF-8 encoding depending
1250 on whether classic DNS or LLMNR is used as transport. In the
1251 next releases we intend to add a DNSSEC and mDNS/DNS-SD
1252 implementation to systemd-resolved.
1254 * A new NSS module nss-mymachines has been added, that
1255 automatically resolves the names of all local registered
1256 containers to their respective IP addresses.
1258 * A new client tool "networkctl" for systemd-networkd has been
1259 added. It currently is entirely passive and will query
1260 networking configuration from udev, rtnetlink and networkd,
1261 and present it to the user in a very friendly
1262 way. Eventually, we hope to extend it to become a full
1263 control utility for networkd.
1265 * .socket units gained a new DeferAcceptSec= setting that
1266 controls the kernels' TCP_DEFER_ACCEPT sockopt for
1267 TCP. Similar, support for controlling TCP keep-alive
1268 settings has been added (KeepAliveTimeSec=,
1269 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
1270 turning off Nagle's algorithm on TCP has been added
1273 * logind learned a new session type "web", for use in projects
1274 like Cockpit which register web clients as PAM sessions.
1276 * timer units with at least one OnCalendar= setting will now
1277 be started only after timer-sync.target has been
1278 reached. This way they will not elapse before the system
1279 clock has been corrected by a local NTP client or
1280 similar. This is particular useful on RTC-less embedded
1281 machines, that come up with an invalid system clock.
1283 * systemd-nspawn's --network-veth= switch should now result in
1284 stable MAC addresses for both the outer and the inner side
1287 * systemd-nspawn gained a new --volatile= switch for running
1288 container instances with /etc or /var unpopulated.
1290 * The kdbus client code has been updated to use the new Linux
1291 3.17 memfd subsystem instead of the old kdbus-specific one.
1293 * systemd-networkd's DHCP client and server now support
1294 FORCERENEW. There are also new configuration options to
1295 configure the vendor client identifier and broadcast mode
1298 * systemd will no longer inform the kernel about the current
1299 timezone, as this is necessarily incorrect and racy as the
1300 kernel has no understanding of DST and similar
1301 concepts. This hence means FAT timestamps will be always
1302 considered UTC, similar to what Android is already
1303 doing. Also, when the RTC is configured to the local time
1304 (rather than UTC) systemd will never synchronize back to it,
1305 as this might confuse Windows at a later boot.
1307 * systemd-analyze gained a new command "verify" for offline
1308 validation of unit files.
1310 * systemd-networkd gained support for a couple of additional
1311 settings for bonding networking setups. Also, the metric for
1312 statically configured routes may now be configured. For
1313 network interfaces where this is appropriate the peer IP
1314 address may now be configured.
1316 * systemd-networkd's DHCP client will no longer request
1317 broadcasting by default, as this tripped up some networks.
1318 For hardware where broadcast is required the feature should
1319 be switched back on using RequestBroadcast=yes.
1321 * systemd-networkd will now set up IPv4LL addresses (when
1322 enabled) even if DHCP is configured successfully.
1324 * udev will now default to respect network device names given
1325 by the kernel when the kernel indicates that these are
1326 predictable. This behavior can be tweaked by changing
1327 NamePolicy= in the relevant .link file.
1329 * A new library systemd-terminal has been added that
1330 implements full TTY stream parsing and rendering. This
1331 library is supposed to be used later on for implementing a
1332 full userspace VT subsystem, replacing the current kernel
1335 * A new tool systemd-journal-upload has been added to push
1336 journal data to a remote system running
1337 systemd-journal-remote.
1339 * journald will no longer forward all local data to another
1340 running syslog daemon. This change has been made because
1341 rsyslog (which appears to be the most commonly used syslog
1342 implementation these days) no longer makes use of this, and
1343 instead pulls the data out of the journal on its own. Since
1344 forwarding the messages to a non-existent syslog server is
1345 more expensive than we assumed we have now turned this
1346 off. If you run a syslog server that is not a recent rsyslog
1347 version, you have to turn this option on again
1348 (ForwardToSyslog= in journald.conf).
1350 * journald now optionally supports the LZ4 compressor for
1351 larger journal fields. This compressor should perform much
1352 better than XZ which was the previous default.
1354 * machinectl now shows the IP addresses of local containers,
1355 if it knows them, plus the interface name of the container.
1357 * A new tool "systemd-escape" has been added that makes it
1358 easy to escape strings to build unit names and similar.
1360 * sd_notify() messages may now include a new ERRNO= field
1361 which is parsed and collected by systemd and shown among the
1362 "systemctl status" output for a service.
1364 * A new component "systemd-firstboot" has been added that
1365 queries the most basic systemd information (timezone,
1366 hostname, root password) interactively on first
1367 boot. Alternatively it may also be used to provision these
1368 things offline on OS images installed into directories.
1370 * The default sysctl.d/ snippets will now set
1372 net.ipv4.conf.default.promote_secondaries=1
1374 This has the benefit of no flushing secondary IP addresses
1375 when primary addresses are removed.
1377 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
1378 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
1379 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
1380 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
1381 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
1382 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
1383 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
1384 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
1385 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
1386 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
1387 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
1388 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
1389 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
1390 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
1391 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
1393 -- Berlin, 2014-08-19
1397 * A new tool systemd-sysusers has been added. This tool
1398 creates system users and groups in /etc/passwd and
1399 /etc/group, based on static declarative system user/group
1400 definitions in /usr/lib/sysusers.d/. This is useful to
1401 enable factory resets and volatile systems that boot up with
1402 an empty /etc directory, and thus need system users and
1403 groups created during early boot. systemd now also ships
1404 with two default sysusers.d/ files for the most basic
1405 users and groups systemd and the core operating system
1408 * A new tmpfiles snippet has been added that rebuilds the
1409 essential files in /etc on boot, should they be missing.
1411 * A directive for ensuring automatic clean-up of
1412 /var/cache/man/ has been removed from the default
1413 configuration. This line should now be shipped by the man
1414 implementation. The necessary change has been made to the
1415 man-db implementation. Note that you need to update your man
1416 implementation to one that ships this line, otherwise no
1417 automatic clean-up of /var/cache/man will take place.
1419 * A new condition ConditionNeedsUpdate= has been added that
1420 may conditionalize services to only run when /etc or /var
1421 are "older" than the vendor operating system resources in
1422 /usr. This is useful for reconstructing or updating /etc
1423 after an offline update of /usr or a factory reset, on the
1424 next reboot. Services that want to run once after such an
1425 update or reset should use this condition and order
1426 themselves before the new systemd-update-done.service, which
1427 will mark the two directories as fully updated. A number of
1428 service files have been added making use of this, to rebuild
1429 the udev hardware database, the journald message catalog and
1430 dynamic loader cache (ldconfig). The systemd-sysusers tool
1431 described above also makes use of this now. With this in
1432 place it is now possible to start up a minimal operating
1433 system with /etc empty cleanly. For more information on the
1434 concepts involved see this recent blog story:
1436 http://0pointer.de/blog/projects/stateless.html
1438 * A new system group "input" has been introduced, and all
1439 input device nodes get this group assigned. This is useful
1440 for system-level software to get access to input devices. It
1441 complements what is already done for "audio" and "video".
1443 * systemd-networkd learnt minimal DHCPv4 server support in
1444 addition to the existing DHCPv4 client support. It also
1445 learnt DHCPv6 client and IPv6 Router Solicitation client
1446 support. The DHCPv4 client gained support for static routes
1447 passed in from the server. Note that the [DHCPv4] section
1448 known in older systemd-networkd versions has been renamed to
1449 [DHCP] and is now also used by the DHCPv6 client. Existing
1450 .network files using settings of this section should be
1451 updated, though compatibility is maintained. Optionally, the
1452 client hostname may now be sent to the DHCP server.
1454 * networkd gained support for vxlan virtual networks as well
1455 as tun/tap and dummy devices.
1457 * networkd gained support for automatic allocation of address
1458 ranges for interfaces from a system-wide pool of
1459 addresses. This is useful for dynamically managing a large
1460 number of interfaces with a single network configuration
1461 file. In particular this is useful to easily assign
1462 appropriate IP addresses to the veth links of a large number
1463 of nspawn instances.
1465 * RPM macros for processing sysusers, sysctl and binfmt
1466 drop-in snippets at package installation time have been
1469 * The /etc/os-release file should now be placed in
1470 /usr/lib/os-release. The old location is automatically
1471 created as symlink. /usr/lib is the more appropriate
1472 location of this file, since it shall actually describe the
1473 vendor operating system shipped in /usr, and not the
1474 configuration stored in /etc.
1476 * .mount units gained a new boolean SloppyOptions= setting
1477 that maps to mount(8)'s -s option which enables permissive
1478 parsing of unknown mount options.
1480 * tmpfiles learnt a new "L+" directive which creates a symlink
1481 but (unlike "L") deletes a pre-existing file first, should
1482 it already exist and not already be the correct
1483 symlink. Similar, "b+", "c+" and "p+" directives have been
1484 added as well, which create block and character devices, as
1485 well as fifos in the filesystem, possibly removing any
1486 pre-existing files of different types.
1488 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
1489 'argument' field (which so far specified the source to
1490 symlink/copy the files from) is now optional. If omitted the
1491 same file os copied from /usr/share/factory/ suffixed by the
1492 full destination path. This is useful for populating /etc
1493 with essential files, by copying them from vendor defaults
1494 shipped in /usr/share/factory/etc.
1496 * A new command "systemctl preset-all" has been added that
1497 applies the service preset settings to all installed unit
1498 files. A new switch --preset-mode= has been added that
1499 controls whether only enable or only disable operations
1502 * A new command "systemctl is-system-running" has been added
1503 that allows checking the overall state of the system, for
1504 example whether it is fully up and running.
1506 * When the system boots up with an empty /etc, the equivalent
1507 to "systemctl preset-all" is executed during early boot, to
1508 make sure all default services are enabled after a factory
1511 * systemd now contains a minimal preset file that enables the
1512 most basic services systemd ships by default.
1514 * Unit files' [Install] section gained a new DefaultInstance=
1515 field for defining the default instance to create if a
1516 template unit is enabled with no instance specified.
1518 * A new passive target cryptsetup-pre.target has been added
1519 that may be used by services that need to make they run and
1520 finish before the first LUKS cryptographic device is set up.
1522 * The /dev/loop-control and /dev/btrfs-control device nodes
1523 are now owned by the "disk" group by default, opening up
1524 access to this group.
1526 * systemd-coredump will now automatically generate a
1527 stack trace of all core dumps taking place on the system,
1528 based on elfutils' libdw library. This stack trace is logged
1531 * systemd-coredump may now optionally store coredumps directly
1532 on disk (in /var/lib/systemd/coredump, possibly compressed),
1533 instead of storing them unconditionally in the journal. This
1534 mode is the new default. A new configuration file
1535 /etc/systemd/coredump.conf has been added to configure this
1536 and other parameters of systemd-coredump.
1538 * coredumpctl gained a new "info" verb to show details about a
1539 specific coredump. A new switch "-1" has also been added
1540 that makes sure to only show information about the most
1541 recent entry instead of all entries. Also, as the tool is
1542 generally useful now the "systemd-" prefix of the binary
1543 name has been removed. Distributions that want to maintain
1544 compatibility with the old name should add a symlink from
1545 the old name to the new name.
1547 * journald's SplitMode= now defaults to "uid". This makes sure
1548 that unprivileged users can access their own coredumps with
1549 coredumpctl without restrictions.
1551 * New kernel command line options "systemd.wants=" (for
1552 pulling an additional unit during boot), "systemd.mask="
1553 (for masking a specific unit for the boot), and
1554 "systemd.debug-shell" (for enabling the debug shell on tty9)
1555 have been added. This is implemented in the new generator
1556 "systemd-debug-generator".
1558 * systemd-nspawn will now by default filter a couple of
1559 syscalls for containers, among them those required for
1560 kernel module loading, direct x86 IO port access, swap
1561 management, and kexec. Most importantly though
1562 open_by_handle_at() is now prohibited for containers,
1563 closing a hole similar to a recently discussed vulnerability
1564 in docker regarding access to files on file hierarchies the
1565 container should normally not have access to. Note that for
1566 nspawn we generally make no security claims anyway (and
1567 this is explicitly documented in the man page), so this is
1568 just a fix for one of the most obvious problems.
1570 * A new man page file-hierarchy(7) has been added that
1571 contains a minimized, modernized version of the file system
1572 layout systemd expects, similar in style to the FHS
1573 specification or hier(5). A new tool systemd-path(1) has
1574 been added to query many of these paths for the local
1577 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
1578 longer done. Since the directory now has a per-user size
1579 limit, and is cleaned on logout this appears unnecessary,
1580 in particular since this now brings the lifecycle of this
1581 directory closer in line with how IPC objects are handled.
1583 * systemd.pc now exports a number of additional directories,
1584 including $libdir (which is useful to identify the library
1585 path for the primary architecture of the system), and a
1586 couple of drop-in directories.
1588 * udev's predictable network interface names now use the dev_port
1589 sysfs attribute, introduced in linux 3.15 instead of dev_id to
1590 distinguish between ports of the same PCI function. dev_id should
1591 only be used for ports using the same HW address, hence the need
1594 * machined has been updated to export the OS version of a
1595 container (read from /etc/os-release and
1596 /usr/lib/os-release) on the bus. This is now shown in
1597 "machinectl status" for a machine.
1599 * A new service setting RestartForceExitStatus= has been
1600 added. If configured to a set of exit signals or process
1601 return values, the service will be restarted when the main
1602 daemon process exits with any of them, regardless of the
1605 * systemctl's -H switch for connecting to remote systemd
1606 machines has been extended so that it may be used to
1607 directly connect to a specific container on the
1608 host. "systemctl -H root@foobar:waldi" will now connect as
1609 user "root" to host "foobar", and then proceed directly to
1610 the container named "waldi". Note that currently you have to
1611 authenticate as user "root" for this to work, as entering
1612 containers is a privileged operation.
1614 Contributions from: Andreas Henriksson, Benjamin Steinwender,
1615 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
1616 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
1617 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
1618 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
1619 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
1620 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
1621 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
1622 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
1623 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
1624 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
1625 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
1627 -- Berlin, 2014-07-03
1631 * As an experimental feature, udev now tries to lock the
1632 disk device node (flock(LOCK_SH|LOCK_NB)) while it
1633 executes events for the disk or any of its partitions.
1634 Applications like partitioning programs can lock the
1635 disk device node (flock(LOCK_EX)) and claim temporary
1636 device ownership that way; udev will entirely skip all event
1637 handling for this disk and its partitions. If the disk
1638 was opened for writing, the close will trigger a partition
1639 table rescan in udev's "watch" facility, and if needed
1640 synthesize "change" events for the disk and all its partitions.
1641 This is now unconditionally enabled, and if it turns out to
1642 cause major problems, we might turn it on only for specific
1643 devices, or might need to disable it entirely. Device Mapper
1644 devices are excluded from this logic.
1646 * We temporarily dropped the "-l" switch for fsck invocations,
1647 since they collide with the flock() logic above. util-linux
1648 upstream has been changed already to avoid this conflict,
1649 and we will readd "-l" as soon as util-linux with this
1650 change has been released.
1652 * The dependency on libattr has been removed. Since a long
1653 time, the extended attribute calls have moved to glibc, and
1654 libattr is thus unnecessary.
1656 * Virtualization detection works without priviliges now. This
1657 means the systemd-detect-virt binary no longer requires
1658 CAP_SYS_PTRACE file capabilities, and our daemons can run
1659 with fewer privileges.
1661 * systemd-networkd now runs under its own "systemd-network"
1662 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
1663 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
1664 loses the ability to write to files owned by root this way.
1666 * Similar, systemd-resolved now runs under its own
1667 "systemd-resolve" user with no capabilities remaining.
1669 * Similar, systemd-bus-proxyd now runs under its own
1670 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
1672 * systemd-networkd gained support for setting up "veth"
1673 virtual ethernet devices for container connectivity, as well
1674 as GRE and VTI tunnels.
1676 * systemd-networkd will no longer automatically attempt to
1677 manually load kernel modules necessary for certain tunnel
1678 transports. Instead, it is assumed the kernel loads them
1679 automatically when required. This only works correctly on
1680 very new kernels. On older kernels, please consider adding
1681 the kernel modules to /etc/modules-load.d/ as a work-around.
1683 * The resolv.conf file systemd-resolved generates has been
1684 moved to /run/systemd/resolve/. If you have a symlink from
1685 /etc/resolv.conf, it might be necessary to correct it.
1687 * Two new service settings, ProtectHome= and ProtectSystem=,
1688 have been added. When enabled, they will make the user data
1689 (such as /home) inaccessible or read-only and the system
1690 (such as /usr) read-only, for specific services. This allows
1691 very light-weight per-service sandboxing to avoid
1692 modifications of user data or system files from
1693 services. These two new switches have been enabled for all
1694 of systemd's long-running services, where appropriate.
1696 * Socket units gained new SocketUser= and SocketGroup=
1697 settings to set the owner user and group of AF_UNIX sockets
1698 and FIFOs in the file system.
1700 * Socket units gained a new RemoveOnStop= setting. If enabled,
1701 all FIFOS and sockets in the file system will be removed
1702 when the specific socket unit is stopped.
1704 * Socket units gained a new Symlinks= setting. It takes a list
1705 of symlinks to create to file system sockets or FIFOs
1706 created by the specific Unix sockets. This is useful to
1707 manage symlinks to socket nodes with the same life-cycle as
1710 * The /dev/log socket and /dev/initctl FIFO have been moved to
1711 /run, and have been replaced by symlinks. This allows
1712 connecting to these facilities even if PrivateDevices=yes is
1713 used for a service (which makes /dev/log itself unavailable,
1714 but /run is left). This also has the benefit of ensuring
1715 that /dev only contains device nodes, directories and
1716 symlinks, and nothing else.
1718 * sd-daemon gained two new calls sd_pid_notify() and
1719 sd_pid_notifyf(). They are similar to sd_notify() and
1720 sd_notifyf(), but allow overriding of the source PID of
1721 notification messages if permissions permit this. This is
1722 useful to send notify messages on behalf of a different
1723 process (for example, the parent process). The
1724 systemd-notify tool has been updated to make use of this
1725 when sending messages (so that notification messages now
1726 originate from the shell script invoking systemd-notify and
1727 not the systemd-notify process itself. This should minimize
1728 a race where systemd fails to associate notification
1729 messages to services when the originating process already
1732 * A new "on-abnormal" setting for Restart= has been added. If
1733 set, it will result in automatic restarts on all "abnormal"
1734 reasons for a process to exit, which includes unclean
1735 signals, core dumps, timeouts and watchdog timeouts, but
1736 does not include clean and unclean exit codes or clean
1737 signals. Restart=on-abnormal is an alternative for
1738 Restart=on-failure for services that shall be able to
1739 terminate and avoid restarts on certain errors, by
1740 indicating so with an unclean exit code. Restart=on-failure
1741 or Restart=on-abnormal is now the recommended setting for
1742 all long-running services.
1744 * If the InaccessibleDirectories= service setting points to a
1745 mount point (or if there are any submounts contained within
1746 it), it is now attempted to completely unmount it, to make
1747 the file systems truly unavailable for the respective
1750 * The ReadOnlyDirectories= service setting and
1751 systemd-nspawn's --read-only parameter are now recursively
1752 applied to all submounts, too.
1754 * Mount units may now be created transiently via the bus APIs.
1756 * The support for SysV and LSB init scripts has been removed
1757 from the systemd daemon itself. Instead, it is now
1758 implemented as a generator that creates native systemd units
1759 from these scripts when needed. This enables us to remove a
1760 substantial amount of legacy code from PID 1, following the
1761 fact that many distributions only ship a very small number
1762 of LSB/SysV init scripts nowadays.
1764 * Privileged Xen (dom0) domains are not considered
1765 virtualization anymore by the virtualization detection
1766 logic. After all, they generally have unrestricted access to
1767 the hardware and usually are used to manage the unprivileged
1770 * systemd-tmpfiles gained a new "C" line type, for copying
1771 files or entire directories.
1773 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
1774 lines. So far, they have been non-globbing versions of the
1775 latter, and have thus been redundant. In future, it is
1776 recommended to only use "z". "m" has hence been removed
1777 from the documentation, even though it stays supported.
1779 * A tmpfiles snippet to recreate the most basic structure in
1780 /var has been added. This is enough to create the /var/run →
1781 /run symlink and create a couple of structural
1782 directories. This allows systems to boot up with an empty or
1783 volatile /var. Of course, while with this change, the core OS
1784 now is capable with dealing with a volatile /var, not all
1785 user services are ready for it. However, we hope that sooner
1786 or later, many service daemons will be changed upstream so
1787 that they are able to automatically create their necessary
1788 directories in /var at boot, should they be missing. This is
1789 the first step to allow state-less systems that only require
1790 the vendor image for /usr to boot.
1792 * systemd-nspawn has gained a new --tmpfs= switch to mount an
1793 empty tmpfs instance to a specific directory. This is
1794 particularly useful for making use of the automatic
1795 reconstruction of /var (see above), by passing --tmpfs=/var.
1797 * Access modes specified in tmpfiles snippets may now be
1798 prefixed with "~", which indicates that they shall be masked
1799 by whether the existing file or directory is currently
1800 writable, readable or executable at all. Also, if specified,
1801 the sgid/suid/sticky bits will be masked for all
1804 * A new passive target unit "network-pre.target" has been
1805 added which is useful for services that shall run before any
1806 network is configured, for example firewall scripts.
1808 * The "floppy" group that previously owned the /dev/fd*
1809 devices is no longer used. The "disk" group is now used
1810 instead. Distributions should probably deprecate usage of
1813 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
1814 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
1815 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
1816 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
1817 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
1818 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
1821 -- Berlin, 2014-06-11
1825 * A new "systemd-timesyncd" daemon has been added for
1826 synchronizing the system clock across the network. It
1827 implements an SNTP client. In contrast to NTP
1828 implementations such as chrony or the NTP reference server,
1829 this only implements a client side, and does not bother with
1830 the full NTP complexity, focusing only on querying time from
1831 one remote server and synchronizing the local clock to
1832 it. Unless you intend to serve NTP to networked clients or
1833 want to connect to local hardware clocks, this simple NTP
1834 client should be more than appropriate for most
1835 installations. The daemon runs with minimal privileges, and
1836 has been hooked up with networkd to only operate when
1837 network connectivity is available. The daemon saves the
1838 current clock to disk every time a new NTP sync has been
1839 acquired, and uses this to possibly correct the system clock
1840 early at bootup, in order to accommodate for systems that
1841 lack an RTC such as the Raspberry Pi and embedded devices,
1842 and to make sure that time monotonically progresses on these
1843 systems, even if it is not always correct. To make use of
1844 this daemon, a new system user and group "systemd-timesync"
1845 needs to be created on installation of systemd.
1847 * The queue "seqnum" interface of libudev has been disabled, as
1848 it was generally incompatible with device namespacing as
1849 sequence numbers of devices go "missing" if the devices are
1850 part of a different namespace.
1852 * "systemctl list-timers" and "systemctl list-sockets" gained
1853 a --recursive switch for showing units of these types also
1854 for all local containers, similar in style to the already
1855 supported --recursive switch for "systemctl list-units".
1857 * A new RebootArgument= setting has been added for service
1858 units, which may be used to specify a kernel reboot argument
1859 to use when triggering reboots with StartLimitAction=.
1861 * A new FailureAction= setting has been added for service
1862 units which may be used to specify an operation to trigger
1863 when a service fails. This works similarly to
1864 StartLimitAction=, but unlike it, controls what is done
1865 immediately rather than only after several attempts to
1866 restart the service in question.
1868 * hostnamed got updated to also expose the kernel name,
1869 release, and version on the bus. This is useful for
1870 executing commands like hostnamectl with the -H switch.
1871 systemd-analyze makes use of this to properly display
1872 details when running non-locally.
1874 * The bootchart tool can now show cgroup information in the
1875 graphs it generates.
1877 * The CFS CPU quota cgroup attribute is now exposed for
1878 services. The new CPUQuota= switch has been added for this
1879 which takes a percentage value. Setting this will have the
1880 result that a service may never get more CPU time than the
1881 specified percentage, even if the machine is otherwise idle.
1883 * systemd-networkd learned IPIP and SIT tunnel support.
1885 * LSB init scripts exposing a dependency on $network will now
1886 get a dependency on network-online.target rather than simply
1887 network.target. This should bring LSB handling closer to
1888 what it was on SysV systems.
1890 * A new fsck.repair= kernel option has been added to control
1891 how fsck shall deal with unclean file systems at boot.
1893 * The (.ini) configuration file parser will now silently
1894 ignore sections whose name begins with "X-". This may be
1895 used to maintain application-specific extension sections in unit
1898 * machined gained a new API to query the IP addresses of
1899 registered containers. "machinectl status" has been updated
1900 to show these addresses in its output.
1902 * A new call sd_uid_get_display() has been added to the
1903 sd-login APIs for querying the "primary" session of a
1904 user. The "primary" session of the user is elected from the
1905 user's sessions and generally a graphical session is
1906 preferred over a text one.
1908 * A minimal systemd-resolved daemon has been added. It
1909 currently simply acts as a companion to systemd-networkd and
1910 manages resolv.conf based on per-interface DNS
1911 configuration, possibly supplied via DHCP. In the long run
1912 we hope to extend this into a local DNSSEC enabled DNS and
1915 * The systemd-networkd-wait-online tool is now enabled by
1916 default. It will delay network-online.target until a network
1917 connection has been configured. The tool primarily integrates
1918 with networkd, but will also make a best effort to make sense
1919 of network configuration performed in some other way.
1921 * Two new service options StartupCPUShares= and
1922 StartupBlockIOWeight= have been added that work similarly to
1923 CPUShares= and BlockIOWeight= however only apply during
1924 system startup. This is useful to prioritize certain services
1925 differently during bootup than during normal runtime.
1927 * hostnamed has been changed to prefer the statically
1928 configured hostname in /etc/hostname (unless set to
1929 'localhost' or empty) over any dynamic one supplied by
1930 dhcp. With this change, the rules for picking the hostname
1931 match more closely the rules of other configuration settings
1932 where the local administrator's configuration in /etc always
1933 overrides any other settings.
1935 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
1936 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
1937 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
1938 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
1939 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
1940 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
1941 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
1942 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
1943 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
1944 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
1945 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
1946 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
1947 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
1948 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
1949 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
1950 Lindskog, WaLyong Cho, Will Woods, Zbigniew
1953 -- Beijing, 2014-05-28
1957 * When restoring the screen brightness at boot, stay away from
1958 the darkest setting or from the lowest 5% of the available
1959 range, depending on which is the larger value of both. This
1960 should effectively protect the user from rebooting into a
1961 black screen, should the brightness have been set to minimum
1964 * sd-login gained a new sd_machine_get_class() call to
1965 determine the class ("vm" or "container") of a machine
1966 registered with machined.
1968 * sd-login gained new calls
1969 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
1970 to query the identity of the peer of a local AF_UNIX
1971 connection. They operate similarly to their sd_pid_get_xyz()
1974 * PID 1 will now maintain a system-wide system state engine
1975 with the states "starting", "running", "degraded",
1976 "maintenance", "stopping". These states are bound to system
1977 startup, normal runtime, runtime with at least one failed
1978 service, rescue/emergency mode and system shutdown. This
1979 state is shown in the "systemctl status" output when no unit
1980 name is passed. It is useful to determine system state, in
1981 particularly when doing so for many systems or containers at
1984 * A new command "list-machines" has been added to "systemctl"
1985 that lists all local OS containers and shows their system
1986 state (see above), if systemd runs inside of them.
1988 * systemctl gained a new "-r" switch to recursively enumerate
1989 units on all local containers, when used with the
1990 "list-unit" command (which is the default one that is
1991 executed when no parameters are specified).
1993 * The GPT automatic partition discovery logic will now honour
1994 two GPT partition flags: one may be set on a partition to
1995 cause it to be mounted read-only, and the other may be set
1996 on a partition to ignore it during automatic discovery.
1998 * Two new GPT type UUIDs have been added for automatic root
1999 partition discovery, for 32-bit and 64-bit ARM. This is not
2000 particularly useful for discovering the root directory on
2001 these architectures during bare-metal boots (since UEFI is
2002 not common there), but still very useful to allow booting of
2003 ARM disk images in nspawn with the -i option.
2005 * MAC addresses of interfaces created with nspawn's
2006 --network-interface= switch will now be generated from the
2007 machine name, and thus be stable between multiple invocations
2010 * logind will now automatically remove all IPC objects owned
2011 by a user if she or he fully logs out. This makes sure that
2012 users who are logged out cannot continue to consume IPC
2013 resources. This covers SysV memory, semaphores and message
2014 queues as well as POSIX shared memory and message
2015 queues. Traditionally, SysV and POSIX IPC had no life-cycle
2016 limits. With this functionality, that is corrected. This may
2017 be turned off by using the RemoveIPC= switch of logind.conf.
2019 * The systemd-machine-id-setup and tmpfiles tools gained a
2020 --root= switch to operate on a specific root directory,
2023 * journald can now forward logged messages to the TTYs of all
2024 logged in users ("wall"). This is the default for all
2025 emergency messages now.
2027 * A new tool systemd-journal-remote has been added to stream
2028 journal log messages across the network.
2030 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
2031 controller trees are mounted into it. Note that the
2032 directories mounted beneath it are not read-only. This is a
2033 security measure and is particularly useful because glibc
2034 actually includes a search logic to pick any tmpfs it can
2035 find to implement shm_open() if /dev/shm is not available
2036 (which it might very well be in namespaced setups).
2038 * machinectl gained a new "poweroff" command to cleanly power
2039 down a local OS container.
2041 * The PrivateDevices= unit file setting will now also drop the
2042 CAP_MKNOD capability from the capability bound set, and
2043 imply DevicePolicy=closed.
2045 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
2046 comprehensively on all long-running systemd services where
2047 this is appropriate.
2049 * systemd-udevd will now run in a disassociated mount
2050 namespace. To mount directories from udev rules, make sure to
2051 pull in mount units via SYSTEMD_WANTS properties.
2053 * The kdbus support gained support for uploading policy into
2054 the kernel. sd-bus gained support for creating "monitoring"
2055 connections that can eavesdrop into all bus communication
2056 for debugging purposes.
2058 * Timestamps may now be specified in seconds since the UNIX
2059 epoch Jan 1st, 1970 by specifying "@" followed by the value
2062 * Native tcpwrap support in systemd has been removed. tcpwrap
2063 is old code, not really maintained anymore and has serious
2064 shortcomings, and better options such as firewalls
2065 exist. For setups that require tcpwrap usage, please
2066 consider invoking your socket-activated service via tcpd,
2067 like on traditional inetd.
2069 * A new system.conf configuration option
2070 DefaultTimerAccuracySec= has been added that controls the
2071 default AccuracySec= setting of .timer units.
2073 * Timer units gained a new WakeSystem= switch. If enabled,
2074 timers configured this way will cause the system to resume
2075 from system suspend (if the system supports that, which most
2078 * Timer units gained a new Persistent= switch. If enabled,
2079 timers configured this way will save to disk when they have
2080 been last triggered. This information is then used on next
2081 reboot to possible execute overdue timer events, that
2082 could not take place because the system was powered off.
2083 This enables simple anacron-like behaviour for timer units.
2085 * systemctl's "list-timers" will now also list the time a
2086 timer unit was last triggered in addition to the next time
2087 it will be triggered.
2089 * systemd-networkd will now assign predictable IPv4LL
2090 addresses to its local interfaces.
2092 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
2093 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
2094 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
2095 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
2096 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
2097 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
2098 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
2099 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
2102 -- Berlin, 2014-03-25
2106 * A new unit file setting RestrictAddressFamilies= has been
2107 added to restrict which socket address families unit
2108 processes gain access to. This takes address family names
2109 like "AF_INET" or "AF_UNIX", and is useful to minimize the
2110 attack surface of services via exotic protocol stacks. This
2111 is built on seccomp system call filters.
2113 * Two new unit file settings RuntimeDirectory= and
2114 RuntimeDirectoryMode= have been added that may be used to
2115 manage a per-daemon runtime directories below /run. This is
2116 an alternative for setting up directory permissions with
2117 tmpfiles snippets, and has the advantage that the runtime
2118 directory's lifetime is bound to the daemon runtime and that
2119 the daemon starts up with an empty directory each time. This
2120 is particularly useful when writing services that drop
2121 privileges using the User= or Group= setting.
2123 * The DeviceAllow= unit setting now supports globbing for
2124 matching against device group names.
2126 * The systemd configuration file system.conf gained new
2127 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
2128 DefaultMemoryAccounting= to globally turn on/off accounting
2129 for specific resources (cgroups) for all units. These
2130 settings may still be overridden individually in each unit
2133 * systemd-gpt-auto-generator is now able to discover /srv and
2134 root partitions in addition to /home and swap partitions. It
2135 also supports LUKS-encrypted partitions now. With this in
2136 place, automatic discovery of partitions to mount following
2137 the Discoverable Partitions Specification
2138 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
2139 is now a lot more complete. This allows booting without
2140 /etc/fstab and without root= on the kernel command line on
2141 systems prepared appropriately.
2143 * systemd-nspawn gained a new --image= switch which allows
2144 booting up disk images and Linux installations on any block
2145 device that follow the Discoverable Partitions Specification
2146 (see above). This means that installations made with
2147 appropriately updated installers may now be started and
2148 deployed using container managers, completely
2149 unmodified. (We hope that libvirt-lxc will add support for
2150 this feature soon, too.)
2152 * systemd-nspawn gained a new --network-macvlan= setting to
2153 set up a private macvlan interface for the
2154 container. Similarly, systemd-networkd gained a new
2155 Kind=macvlan setting in .netdev files.
2157 * systemd-networkd now supports configuring local addresses
2160 * A new tool systemd-network-wait-online has been added to
2161 synchronously wait for network connectivity using
2164 * The sd-bus.h bus API gained a new sd_bus_track object for
2165 tracking the life-cycle of bus peers. Note that sd-bus.h is
2166 still not a public API though (unless you specify
2167 --enable-kdbus on the configure command line, which however
2168 voids your warranty and you get no API stability guarantee).
2170 * The $XDG_RUNTIME_DIR runtime directories for each user are
2171 now individual tmpfs instances, which has the benefit of
2172 introducing separate pools for each user, with individual
2173 size limits, and thus making sure that unprivileged clients
2174 can no longer negatively impact the system or other users by
2175 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
2176 RuntimeDirectorySize= has been introduced that allows
2177 controlling the default size limit for all users. It
2178 defaults to 10% of the available physical memory. This is no
2179 replacement for quotas on tmpfs though (which the kernel
2180 still does not support), as /dev/shm and /tmp are still
2181 shared resources used by both the system and unprivileged
2184 * logind will now automatically turn off automatic suspending
2185 on laptop lid close when more than one display is
2186 connected. This was previously expected to be implemented
2187 individually in desktop environments (such as GNOME),
2188 however has been added to logind now, in order to fix a
2189 boot-time race where a desktop environment might not have
2190 been started yet and thus not been able to take an inhibitor
2191 lock at the time where logind already suspends the system
2192 due to a closed lid.
2194 * logind will now wait at least 30s after each system
2195 suspend/resume cycle, and 3min after system boot before
2196 suspending the system due to a closed laptop lid. This
2197 should give USB docking stations and similar enough time to
2198 be probed and configured after system resume and boot in
2199 order to then act as suspend blocker.
2201 * systemd-run gained a new --property= setting which allows
2202 initialization of resource control properties (and others)
2203 for the created scope or service unit. Example: "systemd-run
2204 --property=BlockIOWeight=10 updatedb" may be used to run
2205 updatedb at a low block IO scheduling weight.
2207 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
2208 now also work in --scope mode.
2210 * When systemd is compiled with kdbus support, basic support
2211 for enforced policies is now in place. (Note that enabling
2212 kdbus still voids your warranty and no API compatibility
2215 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
2216 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
2217 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
2218 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
2219 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
2220 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
2221 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
2222 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
2223 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
2224 Zbigniew Jędrzejewski-Szmek
2226 -- Berlin, 2014-03-12
2230 * systemd will now relabel /dev after loading the SMACK policy
2231 according to SMACK rules.
2233 * A new unit file option AppArmorProfile= has been added to
2234 set the AppArmor profile for the processes of a unit.
2236 * A new condition check ConditionArchitecture= has been added
2237 to conditionalize units based on the system architecture, as
2238 reported by uname()'s "machine" field.
2240 * systemd-networkd now supports matching on the system
2241 virtualization, architecture, kernel command line, host name
2244 * logind is now a lot more aggressive when suspending the
2245 machine due to a closed laptop lid. Instead of acting only
2246 on the lid close action, it will continuously watch the lid
2247 status and act on it. This is useful for laptops where the
2248 power button is on the outside of the chassis so that it can
2249 be reached without opening the lid (such as the Lenovo
2250 Yoga). On those machines, logind will now immediately
2251 re-suspend the machine if the power button has been
2252 accidentally pressed while the laptop was suspended and in a
2253 backpack or similar.
2255 * logind will now watch SW_DOCK switches and inhibit reaction
2256 to the lid switch if it is pressed. This means that logind
2257 will not suspend the machine anymore if the lid is closed
2258 and the system is docked, if the laptop supports SW_DOCK
2259 notifications via the input layer. Note that ACPI docking
2260 stations do not generate this currently. Also note that this
2261 logic is usually not fully sufficient and Desktop
2262 Environments should take a lid switch inhibitor lock when an
2263 external display is connected, as systemd will not watch
2266 * nspawn will now make use of the devices cgroup controller by
2267 default, and only permit creation of and access to the usual
2268 API device nodes like /dev/null or /dev/random, as well as
2269 access to (but not creation of) the pty devices.
2271 * We will now ship a default .network file for
2272 systemd-networkd that automatically configures DHCP for
2273 network interfaces created by nspawn's --network-veth or
2274 --network-bridge= switches.
2276 * systemd will now understand the usual M, K, G, T suffixes
2277 according to SI conventions (i.e. to the base 1000) when
2278 referring to throughput and hardware metrics. It will stay
2279 with IEC conventions (i.e. to the base 1024) for software
2280 metrics, according to what is customary according to
2281 Wikipedia. We explicitly document which base applies for
2282 each configuration option.
2284 * The DeviceAllow= setting in unit files now supports a syntax
2285 to whitelist an entire group of devices node majors at once,
2286 based on the /proc/devices listing. For example, with the
2287 string "char-pts", it is now possible to whitelist all
2288 current and future pseudo-TTYs at once.
2290 * sd-event learned a new "post" event source. Event sources of
2291 this type are triggered by the dispatching of any event
2292 source of a type that is not "post". This is useful for
2293 implementing clean-up and check event sources that are
2294 triggered by other work being done in the program.
2296 * systemd-networkd is no longer statically enabled, but uses
2297 the usual [Install] sections so that it can be
2298 enabled/disabled using systemctl. It still is enabled by
2301 * When creating a veth interface pair with systemd-nspawn, the
2302 host side will now be prefixed with "vb-" if
2303 --network-bridge= is used, and with "ve-" if --network-veth
2304 is used. This way, it is easy to distinguish these cases on
2305 the host, for example to apply different configuration to
2306 them with systemd-networkd.
2308 * The compatibility libraries for libsystemd-journal.so,
2309 libsystem-id128.so, libsystemd-login.so and
2310 libsystemd-daemon.so do not make use of IFUNC
2311 anymore. Instead, we now build libsystemd.so multiple times
2312 under these alternative names. This means that the footprint
2313 is drastically increased, but given that these are
2314 transitional compatibility libraries, this should not matter
2315 much. This change has been made necessary to support the ARM
2316 platform for these compatibility libraries, as the ARM
2317 toolchain is not really at the same level as the toolchain
2318 for other architectures like x86 and does not support
2319 IFUNC. Please make sure to use --enable-compat-libs only
2320 during a transitional period!
2322 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
2323 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
2324 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
2325 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
2326 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
2327 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
2328 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
2329 Zbigniew Jędrzejewski-Szmek
2331 -- Berlin, 2014-02-24
2335 * A new component "systemd-networkd" has been added that can
2336 be used to configure local network interfaces statically or
2337 via DHCP. It is capable of bringing up bridges, VLANs, and
2338 bonding. Currently, no hook-ups for interactive network
2339 configuration are provided. Use this for your initrd,
2340 container, embedded, or server setup if you need a simple,
2341 yet powerful, network configuration solution. This
2342 configuration subsystem is quite nifty, as it allows wildcard
2343 hotplug matching in interfaces. For example, with a single
2344 configuration snippet, you can configure that all Ethernet
2345 interfaces showing up are automatically added to a bridge,
2346 or similar. It supports link-sensing and more.
2348 * A new tool "systemd-socket-proxyd" has been added which can
2349 act as a bidirectional proxy for TCP sockets. This is
2350 useful for adding socket activation support to services that
2351 do not actually support socket activation, including virtual
2352 machines and the like.
2354 * Add a new tool to save/restore rfkill state on
2357 * Save/restore state of keyboard backlights in addition to
2358 display backlights on shutdown/boot.
2360 * udev learned a new SECLABEL{} construct to label device
2361 nodes with a specific security label when they appear. For
2362 now, only SECLABEL{selinux} is supported, but the syntax is
2363 prepared for additional security frameworks.
2365 * udev gained a new scheme to configure link-level attributes
2366 from files in /etc/systemd/network/*.link. These files can
2367 match against MAC address, device path, driver name and type,
2368 and will apply attributes like the naming policy, link speed,
2369 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
2370 address assignment policy (randomized, ...).
2372 * The configuration of network interface naming rules for
2373 "permanent interface names" has changed: a new NamePolicy=
2374 setting in the [Link] section of .link files determines the
2375 priority of possible naming schemes (onboard, slot, mac,
2376 path). The default value of this setting is determined by
2377 /usr/lib/net/links/99-default.link. Old
2378 80-net-name-slot.rules udev configuration file has been
2379 removed, so local configuration overriding this file should
2380 be adapated to override 99-default.link instead.
2382 * When the User= switch is used in a unit file, also
2383 initialize $SHELL= based on the user database entry.
2385 * systemd no longer depends on libdbus. All communication is
2386 now done with sd-bus, systemd's low-level bus library
2389 * kdbus support has been added to PID 1 itself. When kdbus is
2390 enabled, this causes PID 1 to set up the system bus and
2391 enable support for a new ".busname" unit type that
2392 encapsulates bus name activation on kdbus. It works a little
2393 bit like ".socket" units, except for bus names. A new
2394 generator has been added that converts classic dbus1 service
2395 activation files automatically into native systemd .busname
2398 * sd-bus: add a light-weight vtable implementation that allows
2399 defining objects on the bus with a simple static const
2400 vtable array of its methods, signals and properties.
2402 * systemd will not generate or install static dbus
2403 introspection data anymore to /usr/share/dbus-1/interfaces,
2404 as the precise format of these files is unclear, and
2405 nothing makes use of it.
2407 * A proxy daemon is now provided to proxy clients connecting
2408 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
2409 compatibility with classic D-Bus.
2411 * A bus driver implementation has been added that supports the
2412 classic D-Bus bus driver calls on kdbus, also for
2413 compatibility purposes.
2415 * A new API "sd-event.h" has been added that implements a
2416 minimal event loop API built around epoll. It provides a
2417 couple of features that direct epoll usage is lacking:
2418 prioritization of events, scales to large numbers of timer
2419 events, per-event timer slack (accuracy), system-wide
2420 coalescing of timer events, exit handlers, watchdog
2421 supervision support using systemd's sd_notify() API, child
2424 * A new API "sd-rntl.h" has been added that provides an API
2425 around the route netlink interface of the kernel, similar in
2426 style to "sd-bus.h".
2428 * A new API "sd-dhcp-client.h" has been added that provides a
2429 small DHCPv4 client-side implementation. This is used by
2432 * There is a new kernel command line option
2433 "systemd.restore_state=0|1". When set to "0", none of the
2434 systemd tools will restore saved runtime state to hardware
2435 devices. More specifically, the rfkill and backlight states
2438 * The FsckPassNo= compatibility option in mount/service units
2439 has been removed. The fstab generator will now add the
2440 necessary dependencies automatically, and does not require
2441 PID1's support for that anymore.
2443 * journalctl gained a new switch, --list-boots, that lists
2444 recent boots with their times and boot IDs.
2446 * The various tools like systemctl, loginctl, timedatectl,
2447 busctl, systemd-run, ... have gained a new switch "-M" to
2448 connect to a specific, local OS container (as direct
2449 connection, without requiring SSH). This works on any
2450 container that is registered with machined, such as those
2451 created by libvirt-lxc or nspawn.
2453 * systemd-run and systemd-analyze also gained support for "-H"
2454 to connect to remote hosts via SSH. This is particularly
2455 useful for systemd-run because it enables queuing of jobs
2456 onto remote systems.
2458 * machinectl gained a new command "login" to open a getty
2459 login in any local container. This works with any container
2460 that is registered with machined (such as those created by
2461 libvirt-lxc or nspawn), and which runs systemd inside.
2463 * machinectl gained a new "reboot" command that may be used to
2464 trigger a reboot on a specific container that is registered
2465 with machined. This works on any container that runs an init
2466 system of some kind.
2468 * systemctl gained a new "list-timers" command to print a nice
2469 listing of installed timer units with the times they elapse
2472 * Alternative reboot() parameters may now be specified on the
2473 "systemctl reboot" command line and are passed to the
2474 reboot() system call.
2476 * systemctl gained a new --job-mode= switch to configure the
2477 mode to queue a job with. This is a more generic version of
2478 --fail, --irreversible, and --ignore-dependencies, which are
2479 still available but not advertised anymore.
2481 * /etc/systemd/system.conf gained new settings to configure
2482 various default timeouts of units, as well as the default
2483 start limit interval and burst. These may still be overridden
2486 * PID1 will now export on the bus profile data of the security
2487 policy upload process (such as the SELinux policy upload to
2490 * journald: when forwarding logs to the console, include
2491 timestamps (following the setting in
2492 /sys/module/printk/parameters/time).
2494 * OnCalendar= in timer units now understands the special
2495 strings "yearly" and "annually". (Both are equivalent)
2497 * The accuracy of timer units is now configurable with the new
2498 AccuracySec= setting. It defaults to 1min.
2500 * A new dependency type JoinsNamespaceOf= has been added that
2501 allows running two services within the same /tmp and network
2502 namespace, if PrivateNetwork= or PrivateTmp= are used.
2504 * A new command "cat" has been added to systemctl. It outputs
2505 the original unit file of a unit, and concatenates the
2506 contents of additional "drop-in" unit file snippets, so that
2507 the full configuration is shown.
2509 * systemctl now supports globbing on the various "list-xyz"
2510 commands, like "list-units" or "list-sockets", as well as on
2511 those commands which take multiple unit names.
2513 * journalctl's --unit= switch gained support for globbing.
2515 * All systemd daemons now make use of the watchdog logic so
2516 that systemd automatically notices when they hang.
2518 * If the $container_ttys environment variable is set,
2519 getty-generator will automatically spawn a getty for each
2520 listed tty. This is useful for container managers to request
2521 login gettys to be spawned on as many ttys as needed.
2523 * %h, %s, %U specifier support is not available anymore when
2524 used in unit files for PID 1. This is because NSS calls are
2525 not safe from PID 1. They stay available for --user
2526 instances of systemd, and as special case for the root user.
2528 * loginctl gained a new "--no-legend" switch to turn off output
2531 * The "sd-login.h" API gained three new calls:
2532 sd_session_is_remote(), sd_session_get_remote_user(),
2533 sd_session_get_remote_host() to query information about
2536 * The udev hardware database now also carries vendor/product
2537 information of SDIO devices.
2539 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
2540 determine whether watchdog notifications are requested by
2543 * Socket-activated per-connection services now include a
2544 short description of the connection parameters in the
2547 * tmpfiles gained a new "--boot" option. When this is not used,
2548 only lines where the command character is not suffixed with
2549 "!" are executed. When this option is specified, those
2550 options are executed too. This partitions tmpfiles
2551 directives into those that can be safely executed at any
2552 time, and those which should be run only at boot (for
2553 example, a line that creates /run/nologin).
2555 * A new API "sd-resolve.h" has been added which provides a simple
2556 asynchronous wrapper around glibc NSS host name resolution
2557 calls, such as getaddrinfo(). In contrast to glibc's
2558 getaddrinfo_a(), it does not use signals. In contrast to most
2559 other asynchronous name resolution libraries, this one does
2560 not reimplement DNS, but reuses NSS, so that alternate
2561 host name resolution systems continue to work, such as mDNS,
2562 LDAP, etc. This API is based on libasyncns, but it has been
2563 cleaned up for inclusion in systemd.
2565 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
2566 "sd-daemon.h" are no longer found in individual libraries
2567 libsystemd-journal.so, libsystemd-login.so,
2568 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
2569 merged them into a single library, libsystemd.so, which
2570 provides all symbols. The reason for this is cyclic
2571 dependencies, as these libraries tend to use each other's
2572 symbols. So far, we have managed to workaround that by linking
2573 a copy of a good part of our code into each of these
2574 libraries again and again, which, however, makes certain
2575 things hard to do, like sharing static variables. Also, it
2576 substantially increases footprint. With this change, there
2577 is only one library for the basic APIs systemd
2578 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
2579 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
2580 library as well, however are subject to the --enable-kdbus
2581 switch (see below). Note that "sd-dhcp-client.h" is not part
2582 of this library (this is because it only consumes, never
2583 provides, services of/to other APIs). To make the transition
2584 easy from the separate libraries to the unified one, we
2585 provide the --enable-compat-libs compile-time switch which
2586 will generate stub libraries that are compatible with the
2587 old ones but redirect all calls to the new one.
2589 * All of the kdbus logic and the new APIs "sd-bus.h",
2590 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
2591 and "sd-utf8.h" are compile-time optional via the
2592 "--enable-kdbus" switch, and they are not compiled in by
2593 default. To make use of kdbus, you have to explicitly enable
2594 the switch. Note however, that neither the kernel nor the
2595 userspace API for all of this is considered stable yet. We
2596 want to maintain the freedom to still change the APIs for
2597 now. By specifying this build-time switch, you acknowledge
2598 that you are aware of the instability of the current
2601 * Also, note that while kdbus is pretty much complete,
2602 it lacks one thing: proper policy support. This means you
2603 can build a fully working system with all features; however,
2604 it will be highly insecure. Policy support will be added in
2605 one of the next releases, at the same time that we will
2606 declare the APIs stable.
2608 * When the kernel command line argument "kdbus" is specified,
2609 systemd will automatically load the kdbus.ko kernel module. At
2610 this stage of development, it is only useful for testing kdbus
2611 and should not be used in production. Note: if "--enable-kdbus"
2612 is specified, and the kdbus.ko kernel module is available, and
2613 "kdbus" is added to the kernel command line, the entire system
2614 runs with kdbus instead of dbus-daemon, with the above mentioned
2615 problem of missing the system policy enforcement. Also a future
2616 version of kdbus.ko or a newer systemd will not be compatible with
2617 each other, and will unlikely be able to boot the machine if only
2618 one of them is updated.
2620 * systemctl gained a new "import-environment" command which
2621 uploads the caller's environment (or parts thereof) into the
2622 service manager so that it is inherited by services started
2623 by the manager. This is useful to upload variables like
2624 $DISPLAY into the user service manager.
2626 * A new PrivateDevices= switch has been added to service units
2627 which allows running a service with a namespaced /dev
2628 directory that does not contain any device nodes for
2629 physical devices. More specifically, it only includes devices
2630 such as /dev/null, /dev/urandom, and /dev/zero which are API
2633 * logind has been extended to support behaviour like VT
2634 switching on seats that do not support a VT. This makes
2635 multi-session available on seats that are not the first seat
2636 (seat0), and on systems where kernel support for VTs has
2637 been disabled at compile-time.
2639 * If a process holds a delay lock for system sleep or shutdown
2640 and fails to release it in time, we will now log its
2641 identity. This makes it easier to identify processes that
2642 cause slow suspends or power-offs.
2644 * When parsing /etc/crypttab, support for a new key-slot=
2645 option as supported by Debian is added. It allows indicating
2646 which LUKS slot to use on disk, speeding up key loading.
2648 * The sd_journald_sendv() API call has been checked and
2649 officially declared to be async-signal-safe so that it may
2650 be invoked from signal handlers for logging purposes.
2652 * Boot-time status output is now enabled automatically after a
2653 short timeout if boot does not progress, in order to give
2654 the user an indication what she or he is waiting for.
2656 * The boot-time output has been improved to show how much time
2657 remains until jobs expire.
2659 * The KillMode= switch in service units gained a new possible
2660 value "mixed". If set, and the unit is shut down, then the
2661 initial SIGTERM signal is sent only to the main daemon
2662 process, while the following SIGKILL signal is sent to
2663 all remaining processes of the service.
2665 * When a scope unit is registered, a new property "Controller"
2666 may be set. If set to a valid bus name, systemd will send a
2667 RequestStop() signal to this name when it would like to shut
2668 down the scope. This may be used to hook manager logic into
2669 the shutdown logic of scope units. Also, scope units may now
2670 be put in a special "abandoned" state, in which case the
2671 manager process which created them takes no further
2672 responsibilities for it.
2674 * When reading unit files, systemd will now verify
2675 the access mode of these files, and warn about certain
2676 suspicious combinations. This has been added to make it
2677 easier to track down packaging bugs where unit files are
2678 marked executable or world-writable.
2680 * systemd-nspawn gained a new "--setenv=" switch to set
2681 container-wide environment variables. The similar option in
2682 systemd-activate was renamed from "--environment=" to
2683 "--setenv=" for consistency.
2685 * systemd-nspawn has been updated to create a new kdbus domain
2686 for each container that is invoked, thus allowing each
2687 container to have its own set of system and user buses,
2688 independent of the host.
2690 * systemd-nspawn gained a new --drop-capability= switch to run
2691 the container with less capabilities than the default. Both
2692 --drop-capability= and --capability= now take the special
2693 string "all" for dropping or keeping all capabilities.
2695 * systemd-nspawn gained new switches for executing containers
2696 with specific SELinux labels set.
2698 * systemd-nspawn gained a new --quiet switch to not generate
2699 any additional output but the container's own console
2702 * systemd-nspawn gained a new --share-system switch to run a
2703 container without PID namespacing enabled.
2705 * systemd-nspawn gained a new --register= switch to control
2706 whether the container is registered with systemd-machined or
2707 not. This is useful for containers that do not run full
2708 OS images, but only specific apps.
2710 * systemd-nspawn gained a new --keep-unit which may be used
2711 when invoked as the only program from a service unit, and
2712 results in registration of the unit service itself in
2713 systemd-machined, instead of a newly opened scope unit.
2715 * systemd-nspawn gained a new --network-interface= switch for
2716 moving arbitrary interfaces to the container. The new
2717 --network-veth switch creates a virtual Ethernet connection
2718 between host and container. The new --network-bridge=
2719 switch then allows assigning the host side of this virtual
2720 Ethernet connection to a bridge device.
2722 * systemd-nspawn gained a new --personality= switch for
2723 setting the kernel personality for the container. This is
2724 useful when running a 32-bit container on a 64-bit host. A
2725 similar option Personality= is now also available for service
2728 * logind will now also track a "Desktop" identifier for each
2729 session which encodes the desktop environment of it. This is
2730 useful for desktop environments that want to identify
2731 multiple running sessions of itself easily.
2733 * A new SELinuxContext= setting for service units has been
2734 added that allows setting a specific SELinux execution
2735 context for a service.
2737 * Most systemd client tools will now honour $SYSTEMD_LESS for
2738 settings of the "less" pager. By default, these tools will
2739 override $LESS to allow certain operations to work, such as
2740 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
2741 influence this logic.
2743 * systemd's "seccomp" hook-up has been changed to make use of
2744 the libseccomp library instead of using its own
2745 implementation. This has benefits for portability among
2748 * For usage together with SystemCallFilter=, a new
2749 SystemCallErrorNumber= setting has been introduced that
2750 allows configuration of a system error number to be returned
2751 on filtered system calls, instead of immediately killing the
2752 process. Also, SystemCallArchitectures= has been added to
2753 limit access to system calls of a particular architecture
2754 (in order to turn off support for unused secondary
2755 architectures). There is also a global
2756 SystemCallArchitectures= setting in system.conf now to turn
2757 off support for non-native system calls system-wide.
2759 * systemd requires a kernel with a working name_to_handle_at(),
2760 please see the kernel config requirements in the README file.
2762 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
2763 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
2764 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
2765 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
2766 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
2767 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
2768 Elia Pinto, Florian Weimer, George McCollister, Goffredo
2769 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
2770 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
2771 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
2772 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
2773 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
2774 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
2775 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
2776 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
2777 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
2778 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
2779 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
2780 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
2781 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
2782 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
2783 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
2784 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
2785 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
2787 -- Berlin, 2014-02-20
2791 * logind has gained support for facilitating privileged input
2792 and drm device access for unprivileged clients. This work is
2793 useful to allow Wayland display servers (and similar
2794 programs, such as kmscon) to run under the user's ID and
2795 access input and drm devices which are normally
2796 protected. When this is used (and the kernel is new enough)
2797 logind will "mute" IO on the file descriptors passed to
2798 Wayland as long as it is in the background and "unmute" it
2799 if it returns into the foreground. This allows secure
2800 session switching without allowing background sessions to
2801 eavesdrop on input and display data. This also introduces
2802 session switching support if VT support is turned off in the
2803 kernel, and on seats that are not seat0.
2805 * A new kernel command line option luks.options= is understood
2806 now which allows specifying LUKS options for usage for LUKS
2807 encrypted partitions specified with luks.uuid=.
2809 * tmpfiles.d(5) snippets may now use specifier expansion in
2810 path names. More specifically %m, %b, %H, %v, are now
2811 replaced by the local machine id, boot id, hostname, and
2812 kernel version number.
2814 * A new tmpfiles.d(5) command "m" has been introduced which
2815 may be used to change the owner/group/access mode of a file
2816 or directory if it exists, but do nothing if it does not.
2818 * This release removes high-level support for the
2819 MemorySoftLimit= cgroup setting. The underlying kernel
2820 cgroup attribute memory.soft_limit= is currently badly
2821 designed and likely to be removed from the kernel API in its
2822 current form, hence we should not expose it for now.
2824 * The memory.use_hierarchy cgroup attribute is now enabled for
2825 all cgroups systemd creates in the memory cgroup
2826 hierarchy. This option is likely to be come the built-in
2827 default in the kernel anyway, and the non-hierarchical mode
2828 never made much sense in the intrinsically hierarchical
2831 * A new field _SYSTEMD_SLICE= is logged along with all journal
2832 messages containing the slice a message was generated
2833 from. This is useful to allow easy per-customer filtering of
2834 logs among other things.
2836 * systemd-journald will no longer adjust the group of journal
2837 files it creates to the "systemd-journal" group. Instead we
2838 rely on the journal directory to be owned by the
2839 "systemd-journal" group, and its setgid bit set, so that the
2840 kernel file system layer will automatically enforce that
2841 journal files inherit this group assignment. The reason for
2842 this change is that we cannot allow NSS look-ups from
2843 journald which would be necessary to resolve
2844 "systemd-journal" to a numeric GID, because this might
2845 create deadlocks if NSS involves synchronous queries to
2846 other daemons (such as nscd, or sssd) which in turn are
2847 logging clients of journald and might block on it, which
2848 would then dead lock. A tmpfiles.d(5) snippet included in
2849 systemd will make sure the setgid bit and group are
2850 properly set on the journal directory if it exists on every
2851 boot. However, we recommend adjusting it manually after
2852 upgrades too (or from RPM scriptlets), so that the change is
2853 not delayed until next reboot.
2855 * Backlight and random seed files in /var/lib/ have moved into
2856 the /var/lib/systemd/ directory, in order to centralize all
2857 systemd generated files in one directory.
2859 * Boot time performance measurements (as displayed by
2860 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
2861 performance information if that's available to determine how
2862 much time BIOS and boot loader initialization required. With
2863 a sufficiently new BIOS you hence no longer need to boot
2864 with Gummiboot to get access to such information.
2866 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
2867 Cristian Rodríguez, Dave Reisner, David Herrmann, David
2868 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
2869 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
2870 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
2871 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
2872 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
2874 -- Berlin, 2013-10-02
2878 * The Restart= option for services now understands a new
2879 on-watchdog setting, which will restart the service
2880 automatically if the service stops sending out watchdog keep
2881 alive messages (as configured with WatchdogSec=).
2883 * The getty generator (which is responsible for bringing up a
2884 getty on configured serial consoles) will no longer only
2885 start a getty on the primary kernel console but on all
2886 others, too. This makes the order in which console= is
2887 specified on the kernel command line less important.
2889 * libsystemd-logind gained a new sd_session_get_vt() call to
2890 retrieve the VT number of a session.
2892 * If the option "tries=0" is set for an entry of /etc/crypttab
2893 its passphrase is queried indefinitely instead of any
2894 maximum number of tries.
2896 * If a service with a configure PID file terminates its PID
2897 file will now be removed automatically if it still exists
2898 afterwards. This should put an end to stale PID files.
2900 * systemd-run will now also take relative binary path names
2901 for execution and no longer insists on absolute paths.
2903 * InaccessibleDirectories= and ReadOnlyDirectories= now take
2904 paths that are optionally prefixed with "-" to indicate that
2905 it should not be considered a failure if they do not exist.
2907 * journalctl -o (and similar commands) now understands a new
2908 output mode "short-precise", it is similar to "short" but
2909 shows timestamps with usec accuracy.
2911 * The option "discard" (as known from Debian) is now
2912 synonymous to "allow-discards" in /etc/crypttab. In fact,
2913 "discard" is preferred now (since it is easier to remember
2916 * Some licensing clean-ups were made, so that more code is now
2917 LGPL-2.1 licensed than before.
2919 * A minimal tool to save/restore the display backlight
2920 brightness across reboots has been added. It will store the
2921 backlight setting as late as possible at shutdown, and
2922 restore it as early as possible during reboot.
2924 * A logic to automatically discover and enable home and swap
2925 partitions on GPT disks has been added. With this in place
2926 /etc/fstab becomes optional for many setups as systemd can
2927 discover certain partitions located on the root disk
2928 automatically. Home partitions are recognized under their
2929 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
2930 partitions are recognized under their GPT type ID
2931 0657fd6da4ab43c484e50933c84b4f4f.
2933 * systemd will no longer pass any environment from the kernel
2934 or initrd to system services. If you want to set an
2935 environment for all services, do so via the kernel command
2936 line systemd.setenv= assignment.
2938 * The systemd-sysctl tool no longer natively reads the file
2939 /etc/sysctl.conf. If desired, the file should be symlinked
2940 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
2941 legacy support by a symlink rather than built-in code, it
2942 also makes the otherwise hidden order of application of the
2943 different files visible. (Note that this partly reverts to a
2944 pre-198 application order of sysctl knobs!)
2946 * The "systemctl set-log-level" and "systemctl dump" commands
2947 have been moved to systemd-analyze.
2949 * systemd-run learned the new --remain-after-exit switch,
2950 which causes the scope unit not to be cleaned up
2951 automatically after the process terminated.
2953 * tmpfiles learned a new --exclude-prefix= switch to exclude
2954 certain paths from operation.
2956 * journald will now automatically flush all messages to disk
2957 as soon as a message at the log level CRIT, ALERT or EMERG
2960 Contributions from: Andrew Cook, Brandon Philips, Christian
2961 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
2962 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
2963 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
2964 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
2965 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
2966 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
2967 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
2968 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
2969 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
2970 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
2971 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
2972 William Giokas, Zbigniew Jędrzejewski-Szmek
2974 -- Berlin, 2013-09-13
2978 * The documentation has been updated to cover the various new
2979 concepts introduced with 205.
2981 * Unit files now understand the new %v specifier which
2982 resolves to the kernel version string as returned by "uname
2985 * systemctl now supports filtering the unit list output by
2986 load state, active state and sub state, using the new
2989 * "systemctl status" will now show the results of the
2990 condition checks (like ConditionPathExists= and similar) of
2991 the last start attempts of the unit. They are also logged to
2994 * "journalctl -b" may now be used to look for boot output of a
2995 specific boot. Try "journalctl -b -1" for the previous boot,
2996 but the syntax is substantially more powerful.
2998 * "journalctl --show-cursor" has been added which prints the
2999 cursor string the last shown log line. This may then be used
3000 with the new "journalctl --after-cursor=" switch to continue
3001 browsing logs from that point on.
3003 * "journalctl --force" may now be used to force regeneration
3006 * Creation of "dead" device nodes has been moved from udev
3007 into kmod and tmpfiles. Previously, udev would read the kmod
3008 databases to pre-generate dead device nodes based on meta
3009 information contained in kernel modules, so that these would
3010 be auto-loaded on access rather then at boot. As this
3011 does not really have much to do with the exposing actual
3012 kernel devices to userspace this has always been slightly
3013 alien in the udev codebase. Following the new scheme kmod
3014 will now generate a runtime snippet for tmpfiles from the
3015 module meta information and it now is tmpfiles' job to the
3016 create the nodes. This also allows overriding access and
3017 other parameters for the nodes using the usual tmpfiles
3018 facilities. As side effect this allows us to remove the
3019 CAP_SYS_MKNOD capability bit from udevd entirely.
3021 * logind's device ACLs may now be applied to these "dead"
3022 devices nodes too, thus finally allowing managed access to
3023 devices such as /dev/snd/sequencer whithout loading the
3024 backing module right-away.
3026 * A new RPM macro has been added that may be used to apply
3027 tmpfiles configuration during package installation.
3029 * systemd-detect-virt and ConditionVirtualization= now can
3030 detect User-Mode-Linux machines (UML).
3032 * journald will now implicitly log the effective capabilities
3033 set of processes in the message metadata.
3035 * systemd-cryptsetup has gained support for TrueCrypt volumes.
3037 * The initrd interface has been simplified (more specifically,
3038 support for passing performance data via environment
3039 variables and fsck results via files in /run has been
3040 removed). These features were non-essential, and are
3041 nowadays available in a much nicer way by having systemd in
3042 the initrd serialize its state and have the hosts systemd
3043 deserialize it again.
3045 * The udev "keymap" data files and tools to apply keyboard
3046 specific mappings of scan to key codes, and force-release
3047 scan code lists have been entirely replaced by a udev
3048 "keyboard" builtin and a hwdb data file.
3050 * systemd will now honour the kernel's "quiet" command line
3051 argument also during late shutdown, resulting in a
3052 completely silent shutdown when used.
3054 * There's now an option to control the SO_REUSEPORT socket
3055 option in .socket units.
3057 * Instance units will now automatically get a per-template
3058 subslice of system.slice unless something else is explicitly
3059 configured. For example, instances of sshd@.service will now
3060 implicitly be placed in system-sshd.slice rather than
3061 system.slice as before.
3063 * Test coverage support may now be enabled at build time.
3065 Contributions from: Dave Reisner, Frederic Crozat, Harald
3066 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
3067 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
3068 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
3069 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
3070 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
3071 Giokas, Zbigniew Jędrzejewski-Szmek
3073 -- Berlin, 2013-07-23
3077 * Two new unit types have been introduced:
3079 Scope units are very similar to service units, however, are
3080 created out of pre-existing processes -- instead of PID 1
3081 forking off the processes. By using scope units it is
3082 possible for system services and applications to group their
3083 own child processes (worker processes) in a powerful way
3084 which then maybe used to organize them, or kill them
3085 together, or apply resource limits on them.
3087 Slice units may be used to partition system resources in an
3088 hierarchical fashion and then assign other units to them. By
3089 default there are now three slices: system.slice (for all
3090 system services), user.slice (for all user sessions),
3091 machine.slice (for VMs and containers).
3093 Slices and scopes have been introduced primarily in
3094 context of the work to move cgroup handling to a
3095 single-writer scheme, where only PID 1
3096 creates/removes/manages cgroups.
3098 * There's a new concept of "transient" units. In contrast to
3099 normal units these units are created via an API at runtime,
3100 not from configuration from disk. More specifically this
3101 means it is now possible to run arbitrary programs as
3102 independent services, with all execution parameters passed
3103 in via bus APIs rather than read from disk. Transient units
3104 make systemd substantially more dynamic then it ever was,
3105 and useful as a general batch manager.
3107 * logind has been updated to make use of scope and slice units
3108 for managing user sessions. As a user logs in he will get
3109 his own private slice unit, to which all sessions are added
3110 as scope units. We also added support for automatically
3111 adding an instance of user@.service for the user into the
3112 slice. Effectively logind will no longer create cgroup
3113 hierarchies on its own now, it will defer entirely to PID 1
3114 for this by means of scope, service and slice units. Since
3115 user sessions this way become entities managed by PID 1
3116 the output of "systemctl" is now a lot more comprehensive.
3118 * A new mini-daemon "systemd-machined" has been added which
3119 may be used by virtualization managers to register local
3120 VMs/containers. nspawn has been updated accordingly, and
3121 libvirt will be updated shortly. machined will collect a bit
3122 of meta information about the VMs/containers, and assign
3123 them their own scope unit (see above). The collected
3124 meta-data is then made available via the "machinectl" tool,
3125 and exposed in "ps" and similar tools. machined/machinectl
3126 is compile-time optional.
3128 * As discussed earlier, the low-level cgroup configuration
3129 options ControlGroup=, ControlGroupModify=,
3130 ControlGroupPersistent=, ControlGroupAttribute= have been
3131 removed. Please use high-level attribute settings instead as
3132 well as slice units.
3134 * A new bus call SetUnitProperties() has been added to alter
3135 various runtime parameters of a unit. This is primarily
3136 useful to alter cgroup parameters dynamically in a nice way,
3137 but will be extended later on to make more properties
3138 modifiable at runtime. systemctl gained a new set-properties
3139 command that wraps this call.
3141 * A new tool "systemd-run" has been added which can be used to
3142 run arbitrary command lines as transient services or scopes,
3143 while configuring a number of settings via the command
3144 line. This tool is currently very basic, however already
3145 very useful. We plan to extend this tool to even allow
3146 queuing of execution jobs with time triggers from the
3147 command line, similar in fashion to "at".
3149 * nspawn will now inform the user explicitly that kernels with
3150 audit enabled break containers, and suggest the user to turn
3153 * Support for detecting the IMA and AppArmor security
3154 frameworks with ConditionSecurity= has been added.
3156 * journalctl gained a new "-k" switch for showing only kernel
3157 messages, mimicking dmesg output; in addition to "--user"
3158 and "--system" switches for showing only user's own logs
3161 * systemd-delta can now show information about drop-in
3162 snippets extending unit files.
3164 * libsystemd-bus has been substantially updated but is still
3165 not available as public API.
3167 * systemd will now look for the "debug" argument on the kernel
3168 command line and enable debug logging, similar to what
3169 "systemd.log_level=debug" already did before.
3171 * "systemctl set-default", "systemctl get-default" has been
3172 added to configure the default.target symlink, which
3173 controls what to boot into by default.
3175 * "systemctl set-log-level" has been added as a convenient
3176 way to raise and lower systemd logging threshold.
3178 * "systemd-analyze plot" will now show the time the various
3179 generators needed for execution, as well as information
3180 about the unit file loading.
3182 * libsystemd-journal gained a new sd_journal_open_files() call
3183 for opening specific journal files. journactl also gained a
3184 new switch to expose this new functionality. Previously we
3185 only supported opening all files from a directory, or all
3186 files from the system, as opening individual files only is
3187 racy due to journal file rotation.
3189 * systemd gained the new DefaultEnvironment= setting in
3190 /etc/systemd/system.conf to set environment variables for
3193 * If a privileged process logs a journal message with the
3194 OBJECT_PID= field set, then journald will automatically
3195 augment this with additional OBJECT_UID=, OBJECT_GID=,
3196 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
3197 system services want to log events about specific client
3198 processes. journactl/systemctl has been updated to make use
3199 of this information if all log messages regarding a specific
3202 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
3203 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
3204 Reisner, David Coppa, David King, David Strauss, Eelco
3205 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
3206 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
3207 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
3208 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
3209 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
3210 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
3211 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
3212 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
3213 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
3214 Łukasz Stelmach, 장동준
3218 * The Python bindings gained some minimal support for the APIs
3219 exposed by libsystemd-logind.
3221 * ConditionSecurity= gained support for detecting SMACK. Since
3222 this condition already supports SELinux and AppArmor we only
3223 miss IMA for this. Patches welcome!
3225 Contributions from: Karol Lewandowski, Lennart Poettering,
3226 Zbigniew Jędrzejewski-Szmek
3230 * systemd-nspawn will now create /etc/resolv.conf if
3231 necessary, before bind-mounting the host's file onto it.
3233 * systemd-nspawn will now store meta information about a
3234 container on the container's cgroup as extended attribute
3235 fields, including the root directory.
3237 * The cgroup hierarchy has been reworked in many ways. All
3238 objects any of the components systemd creates in the cgroup
3239 tree are now suffixed. More specifically, user sessions are
3240 now placed in cgroups suffixed with ".session", users in
3241 cgroups suffixed with ".user", and nspawn containers in
3242 cgroups suffixed with ".nspawn". Furthermore, all cgroup
3243 names are now escaped in a simple scheme to avoid collision
3244 of userspace object names with kernel filenames. This work
3245 is preparation for making these objects relocatable in the
3246 cgroup tree, in order to allow easy resource partitioning of
3247 these objects without causing naming conflicts.
3249 * systemctl list-dependencies gained the new switches
3250 --plain, --reverse, --after and --before.
3252 * systemd-inhibit now shows the process name of processes that
3253 have taken an inhibitor lock.
3255 * nss-myhostname will now also resolve "localhost"
3256 implicitly. This makes /etc/hosts an optional file and
3257 nicely handles that on IPv6 ::1 maps to both "localhost" and
3260 * libsystemd-logind.so gained a new call
3261 sd_get_machine_names() to enumerate running containers and
3262 VMs (currently only supported by very new libvirt and
3263 nspawn). sd_login_monitor can now be used to watch
3264 VMs/containers coming and going.
3266 * .include is not allowed recursively anymore, and only in
3267 unit files. Usually it is better to use drop-in snippets in
3268 .d/*.conf anyway, as introduced with systemd 198.
3270 * systemd-analyze gained a new "critical-chain" command that
3271 determines the slowest chain of units run during system
3272 boot-up. It is very useful for tracking down where
3273 optimizing boot time is the most beneficial.
3275 * systemd will no longer allow manipulating service paths in
3276 the name=systemd:/system cgroup tree using ControlGroup= in
3277 units. (But is still fine with it in all other dirs.)
3279 * There's a new systemd-nspawn@.service service file that may
3280 be used to easily run nspawn containers as system
3281 services. With the container's root directory in
3282 /var/lib/container/foobar it is now sufficient to run
3283 "systemctl start systemd-nspawn@foobar.service" to boot it.
3285 * systemd-cgls gained a new parameter "--machine" to list only
3286 the processes within a certain container.
3288 * ConditionSecurity= now can check for "apparmor". We still
3289 are lacking checks for SMACK and IMA for this condition
3290 check though. Patches welcome!
3292 * A new configuration file /etc/systemd/sleep.conf has been
3293 added that may be used to configure which kernel operation
3294 systemd is supposed to execute when "suspend", "hibernate"
3295 or "hybrid-sleep" is requested. This makes the new kernel
3296 "freeze" state accessible to the user.
3298 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
3299 the passed argument if applicable.
3301 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
3302 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
3303 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
3304 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
3305 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
3306 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
3307 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
3312 * The output of 'systemctl list-jobs' got some polishing. The
3313 '--type=' argument may now be passed more than once. A new
3314 command 'systemctl list-sockets' has been added which shows
3315 a list of kernel sockets systemd is listening on with the
3316 socket units they belong to, plus the units these socket
3319 * The experimental libsystemd-bus library got substantial
3320 updates to work in conjunction with the (also experimental)
3321 kdbus kernel project. It works well enough to exchange
3322 messages with some sophistication. Note that kdbus is not
3323 ready yet, and the library is mostly an elaborate test case
3324 for now, and not installable.
3326 * systemd gained a new unit 'systemd-static-nodes.service'
3327 that generates static device nodes earlier during boot, and
3328 can run in conjunction with udev.
3330 * libsystemd-login gained a new call sd_pid_get_user_unit()
3331 to retrieve the user systemd unit a process is running
3332 in. This is useful for systems where systemd is used as
3335 * systemd-nspawn now places all containers in the new /machine
3336 top-level cgroup directory in the name=systemd
3337 hierarchy. libvirt will soon do the same, so that we get a
3338 uniform separation of /system, /user and /machine for system
3339 services, user processes and containers/virtual
3340 machines. This new cgroup hierarchy is also useful to stick
3341 stable names to specific container instances, which can be
3342 recognized later this way (this name may be controlled
3343 via systemd-nspawn's new -M switch). libsystemd-login also
3344 gained a new call sd_pid_get_machine_name() to retrieve the
3345 name of the container/VM a specific process belongs to.
3347 * bootchart can now store its data in the journal.
3349 * libsystemd-journal gained a new call
3350 sd_journal_add_conjunction() for AND expressions to the
3351 matching logic. This can be used to express more complex
3352 logical expressions.
3354 * journactl can now take multiple --unit= and --user-unit=
3357 * The cryptsetup logic now understands the "luks.key=" kernel
3358 command line switch for specifying a file to read the
3359 decryption key from. Also, if a configured key file is not
3360 found the tool will now automatically fall back to prompting
3363 * Python systemd.journal module was updated to wrap recently
3364 added functions from libsystemd-journal. The interface was
3365 changed to bring the low level interface in s.j._Reader
3366 closer to the C API, and the high level interface in
3367 s.j.Reader was updated to wrap and convert all data about
3370 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
3371 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
3372 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
3373 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
3374 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
3375 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3379 * journalctl --update-catalog now understands a new --root=
3380 option to operate on catalogs found in a different root
3383 * During shutdown after systemd has terminated all running
3384 services a final killing loop kills all remaining left-over
3385 processes. We will now print the name of these processes
3386 when we send SIGKILL to them, since this usually indicates a
3389 * If /etc/crypttab refers to password files stored on
3390 configured mount points automatic dependencies will now be
3391 generated to ensure the specific mount is established first
3392 before the key file is attempted to be read.
3394 * 'systemctl status' will now show information about the
3395 network sockets a socket unit is listening on.
3397 * 'systemctl status' will also shown information about any
3398 drop-in configuration file for units. (Drop-In configuration
3399 files in this context are files such as
3400 /etc/systemd/systemd/foobar.service.d/*.conf)
3402 * systemd-cgtop now optionally shows summed up CPU times of
3403 cgroups. Press '%' while running cgtop to switch between
3404 percentage and absolute mode. This is useful to determine
3405 which cgroups use up the most CPU time over the entire
3406 runtime of the system. systemd-cgtop has also been updated
3407 to be 'pipeable' for processing with further shell tools.
3409 * 'hostnamectl set-hostname' will now allow setting of FQDN
3412 * The formatting and parsing of time span values has been
3413 changed. The parser now understands fractional expressions
3414 such as "5.5h". The formatter will now output fractional
3415 expressions for all time spans under 1min, i.e. "5.123456s"
3416 rather than "5s 123ms 456us". For time spans under 1s
3417 millisecond values are shown, for those under 1ms
3418 microsecond values are shown. This should greatly improve
3419 all time-related output of systemd.
3421 * libsystemd-login and libsystemd-journal gained new
3422 functions for querying the poll() events mask and poll()
3423 timeout value for integration into arbitrary event
3426 * localectl gained the ability to list available X11 keymaps
3427 (models, layouts, variants, options).
3429 * 'systemd-analyze dot' gained the ability to filter for
3430 specific units via shell-style globs, to create smaller,
3431 more useful graphs. I.e. it is now possible to create simple
3432 graphs of all the dependencies between only target units, or
3433 of all units that Avahi has dependencies with.
3435 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
3436 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
3437 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
3438 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
3439 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
3440 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
3441 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
3445 * The boot-time readahead implementation for rotating media
3446 will now read the read-ahead data in multiple passes which
3447 consist of all read requests made in equidistant time
3448 intervals. This means instead of strictly reading read-ahead
3449 data in its physical order on disk we now try to find a
3450 middle ground between physical and access time order.
3452 * /etc/os-release files gained a new BUILD_ID= field for usage
3453 on operating systems that provide continuous builds of OS
3456 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
3457 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
3458 William Douglas, Zbigniew Jędrzejewski-Szmek
3462 * systemd-python gained an API exposing libsystemd-daemon.
3464 * The SMACK setup logic gained support for uploading CIPSO
3467 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
3468 ReadOnlyDirectories= and InaccessibleDirectories= has
3469 changed. The private /tmp and /var/tmp directories are now
3470 shared by all processes of a service (which means
3471 ExecStartPre= may now leave data in /tmp that ExecStart= of
3472 the same service can still access). When a service is
3473 stopped its temporary directories are immediately deleted
3474 (normal clean-up with tmpfiles is still done in addition to
3477 * By default, systemd will now set a couple of sysctl
3478 variables in the kernel: the safe sysrq options are turned
3479 on, IP route verification is turned on, and source routing
3480 disabled. The recently added hardlink and softlink
3481 protection of the kernel is turned on. These settings should
3482 be reasonably safe, and good defaults for all new systems.
3484 * The predictable network naming logic may now be turned off
3485 with a new kernel command line switch: net.ifnames=0.
3487 * A new libsystemd-bus module has been added that implements a
3488 pretty complete D-Bus client library. For details see:
3490 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
3492 * journald will now explicitly flush the journal files to disk
3493 at the latest 5min after each write. The file will then also
3494 be marked offline until the next write. This should increase
3495 reliability in case of a crash. The synchronization delay
3496 can be configured via SyncIntervalSec= in journald.conf.
3498 * There's a new remote-fs-setup.target unit that can be used
3499 to pull in specific services when at least one remote file
3500 system is to be mounted.
3502 * There are new targets timers.target and paths.target as
3503 canonical targets to pull user timer and path units in
3504 from. This complements sockets.target with a similar
3505 purpose for socket units.
3507 * libudev gained a new call udev_device_set_attribute_value()
3508 to set sysfs attributes of a device.
3510 * The udev daemon now sets the default number of worker
3511 processes executed in parallel based on the number of available
3512 CPUs instead of the amount of available RAM. This is supposed
3513 to provide a more reliable default and limit a too aggressive
3514 paralellism for setups with 1000s of devices connected.
3516 Contributions from: Auke Kok, Colin Walters, Cristian
3517 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
3518 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
3519 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
3520 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
3521 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
3522 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
3523 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
3524 Zbigniew Jędrzejewski-Szmek
3528 * Configuration of unit files may now be extended via drop-in
3529 files without having to edit/override the unit files
3530 themselves. More specifically, if the administrator wants to
3531 change one value for a service file foobar.service he can
3532 now do so by dropping in a configuration snippet into
3533 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
3534 will load all these snippets and apply them on top of the
3535 main unit configuration file, possibly extending or
3536 overriding its settings. Using these drop-in snippets is
3537 generally nicer than the two earlier options for changing
3538 unit files locally: copying the files from
3539 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
3540 them there; or creating a new file in /etc/systemd/system/
3541 that incorporates the original one via ".include". Drop-in
3542 snippets into these .d/ directories can be placed in any
3543 directory systemd looks for units in, and the usual
3544 overriding semantics between /usr/lib, /etc and /run apply
3547 * Most unit file settings which take lists of items can now be
3548 reset by assigning the empty string to them. For example,
3549 normally, settings such as Environment=FOO=BAR append a new
3550 environment variable assignment to the environment block,
3551 each time they are used. By assigning Environment= the empty
3552 string the environment block can be reset to empty. This is
3553 particularly useful with the .d/*.conf drop-in snippets
3554 mentioned above, since this adds the ability to reset list
3555 settings from vendor unit files via these drop-ins.
3557 * systemctl gained a new "list-dependencies" command for
3558 listing the dependencies of a unit recursively.
3560 * Inhibitors are now honored and listed by "systemctl
3561 suspend", "systemctl poweroff" (and similar) too, not only
3562 GNOME. These commands will also list active sessions by
3565 * Resource limits (as exposed by the various control group
3566 controllers) can now be controlled dynamically at runtime
3567 for all units. More specifically, you can now use a command
3568 like "systemctl set-cgroup-attr foobar.service cpu.shares
3569 2000" to alter the CPU shares a specific service gets. These
3570 settings are stored persistently on disk, and thus allow the
3571 administrator to easily adjust the resource usage of
3572 services with a few simple commands. This dynamic resource
3573 management logic is also available to other programs via the
3574 bus. Almost any kernel cgroup attribute and controller is
3577 * systemd-vconsole-setup will now copy all font settings to
3578 all allocated VTs, where it previously applied them only to
3581 * libsystemd-login gained the new sd_session_get_tty() API
3584 * This release drops support for a few legacy or
3585 distribution-specific LSB facility names when parsing init
3586 scripts: $x-display-manager, $mail-transfer-agent,
3587 $mail-transport-agent, $mail-transfer-agent, $smtp,
3588 $null. Also, the mail-transfer-agent.target unit backing
3589 this has been removed. Distributions which want to retain
3590 compatibility with this should carry the burden for
3591 supporting this themselves and patch support for these back
3592 in, if they really need to. Also, the facilities $syslog and
3593 $local_fs are now ignored, since systemd does not support
3594 early-boot LSB init scripts anymore, and these facilities
3595 are implied anyway for normal services. syslog.target has
3598 * There are new bus calls on PID1's Manager object for
3599 cancelling jobs, and removing snapshot units. Previously,
3600 both calls were only available on the Job and Snapshot
3603 * systemd-journal-gatewayd gained SSL support.
3605 * The various "environment" files, such as /etc/locale.conf
3606 now support continuation lines with a backslash ("\") as
3607 last character in the line, similarly in style (but different)
3608 to how this is supported in shells.
3610 * For normal user processes the _SYSTEMD_USER_UNIT= field is
3611 now implicitly appended to every log entry logged. systemctl
3612 has been updated to filter by this field when operating on a
3613 user systemd instance.
3615 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
3616 CAP_AUDIT_CONTROL capabilities to the capabilities set for
3617 the container. This makes it easier to boot unmodified
3618 Fedora systems in a container, which however still requires
3619 audit=0 to be passed on the kernel command line. Auditing in
3620 kernel and userspace is unfortunately still too broken in
3621 context of containers, hence we recommend compiling it out
3622 of the kernel or using audit=0. Hopefully this will be fixed
3623 one day for good in the kernel.
3625 * nspawn gained the new --bind= and --bind-ro= parameters to
3626 bind mount specific directories from the host into the
3629 * nspawn will now mount its own devpts file system instance
3630 into the container, in order not to leak pty devices from
3631 the host into the container.
3633 * systemd will now read the firmware boot time performance
3634 information from the EFI variables, if the used boot loader
3635 supports this, and takes it into account for boot performance
3636 analysis via "systemd-analyze". This is currently supported
3637 only in conjunction with Gummiboot, but could be supported
3638 by other boot loaders too. For details see:
3640 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
3642 * A new generator has been added that automatically mounts the
3643 EFI System Partition (ESP) to /boot, if that directory
3644 exists, is empty, and no other file system has been
3645 configured to be mounted there.
3647 * logind will now send out PrepareForSleep(false) out
3648 unconditionally, after coming back from suspend. This may be
3649 used by applications as asynchronous notification for
3650 system resume events.
3652 * "systemctl unlock-sessions" has been added, that allows
3653 unlocking the screens of all user sessions at once, similar
3654 to how "systemctl lock-sessions" already locked all users
3655 sessions. This is backed by a new D-Bus call UnlockSessions().
3657 * "loginctl seat-status" will now show the master device of a
3658 seat. (i.e. the device of a seat that needs to be around for
3659 the seat to be considered available, usually the graphics
3662 * tmpfiles gained a new "X" line type, that allows
3663 configuration of files and directories (with wildcards) that
3664 shall be excluded from automatic cleanup ("aging").
3666 * udev default rules set the device node permissions now only
3667 at "add" events, and do not change them any longer with a
3668 later "change" event.
3670 * The log messages for lid events and power/sleep keypresses
3671 now carry a message ID.
3673 * We now have a substantially larger unit test suite, but this
3674 continues to be work in progress.
3676 * udevadm hwdb gained a new --root= parameter to change the
3677 root directory to operate relative to.
3679 * logind will now issue a background sync() request to the kernel
3680 early at shutdown, so that dirty buffers are flushed to disk early
3681 instead of at the last moment, in order to optimize shutdown
3684 * A new bootctl tool has been added that is an interface for
3685 certain boot loader operations. This is currently a preview
3686 and is likely to be extended into a small mechanism daemon
3687 like timedated, localed, hostnamed, and can be used by
3688 graphical UIs to enumerate available boot options, and
3689 request boot into firmware operations.
3691 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
3692 the rest of the package. It also has been updated to work
3693 correctly in initrds.
3695 * Policykit previously has been runtime optional, and is now
3696 also compile time optional via a configure switch.
3698 * systemd-analyze has been reimplemented in C. Also "systemctl
3699 dot" has moved into systemd-analyze.
3701 * "systemctl status" with no further parameters will now print
3702 the status of all active or failed units.
3704 * Operations such as "systemctl start" can now be executed
3705 with a new mode "--irreversible" which may be used to queue
3706 operations that cannot accidentally be reversed by a later
3707 job queuing. This is by default used to make shutdown
3708 requests more robust.
3710 * The Python API of systemd now gained a new module for
3711 reading journal files.
3713 * A new tool kernel-install has been added that can install
3714 kernel images according to the Boot Loader Specification:
3716 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
3718 * Boot time console output has been improved to provide
3719 animated boot time output for hanging jobs.
3721 * A new tool systemd-activate has been added which can be used
3722 to test socket activation with, directly from the command
3723 line. This should make it much easier to test and debug
3724 socket activation in daemons.
3726 * journalctl gained a new "--reverse" (or -r) option to show
3727 journal output in reverse order (i.e. newest line first).
3729 * journalctl gained a new "--pager-end" (or -e) option to jump
3730 to immediately jump to the end of the journal in the
3731 pager. This is only supported in conjunction with "less".
3733 * journalctl gained a new "--user-unit=" option, that works
3734 similarly to "--unit=" but filters for user units rather than
3737 * A number of unit files to ease adoption of systemd in
3738 initrds has been added. This moves some minimal logic from
3739 the various initrd implementations into systemd proper.
3741 * The journal files are now owned by a new group
3742 "systemd-journal", which exists specifically to allow access
3743 to the journal, and nothing else. Previously, we used the
3744 "adm" group for that, which however possibly covers more
3745 than just journal/log file access. This new group is now
3746 already used by systemd-journal-gatewayd to ensure this
3747 daemon gets access to the journal files and as little else
3748 as possible. Note that "make install" will also set FS ACLs
3749 up for /var/log/journal to give "adm" and "wheel" read
3750 access to it, in addition to "systemd-journal" which owns
3751 the journal files. We recommend that packaging scripts also
3752 add read access to "adm" + "wheel" to /var/log/journal, and
3753 all existing/future journal files. To normal users and
3754 administrators little changes, however packagers need to
3755 ensure to create the "systemd-journal" system group at
3756 package installation time.
3758 * The systemd-journal-gatewayd now runs as unprivileged user
3759 systemd-journal-gateway:systemd-journal-gateway. Packaging
3760 scripts need to create these system user/group at
3763 * timedated now exposes a new boolean property CanNTP that
3764 indicates whether a local NTP service is available or not.
3766 * systemd-detect-virt will now also detect xen PVs
3768 * The pstore file system is now mounted by default, if it is
3771 * In addition to the SELinux and IMA policies we will now also
3772 load SMACK policies at early boot.
3774 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
3775 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
3776 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
3777 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
3778 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
3779 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
3780 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
3781 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
3782 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
3783 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
3784 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
3785 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
3786 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
3787 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
3791 * Timer units now support calendar time events in addition to
3792 monotonic time events. That means you can now trigger a unit
3793 based on a calendar time specification such as "Thu,Fri
3794 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
3795 or fifth day of any month of the year 2013, given that it is
3796 a thursday or friday. This brings timer event support
3797 considerably closer to cron's capabilities. For details on
3798 the supported calendar time specification language see
3801 * udev now supports a number of different naming policies for
3802 network interfaces for predictable names, and a combination
3803 of these policies is now the default. Please see this wiki
3804 document for details:
3806 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
3808 * Auke Kok's bootchart implementation has been added to the
3809 systemd tree. It is an optional component that can graph the
3810 boot in quite some detail. It is one of the best bootchart
3811 implementations around and minimal in its code and
3814 * nss-myhostname has been integrated into the systemd source
3815 tree. nss-myhostname guarantees that the local hostname
3816 always stays resolvable via NSS. It has been a weak
3817 requirement of systemd-hostnamed since a long time, and
3818 since its code is actually trivial we decided to just
3819 include it in systemd's source tree. It can be turned off
3820 with a configure switch.
3822 * The read-ahead logic is now capable of properly detecting
3823 whether a btrfs file system is on SSD or rotating media, in
3824 order to optimize the read-ahead scheme. Previously, it was
3825 only capable of detecting this on traditional file systems
3828 * In udev, additional device properties are now read from the
3829 IAB in addition to the OUI database. Also, Bluetooth company
3830 identities are attached to the devices as well.
3832 * In service files %U may be used as specifier that is
3833 replaced by the configured user name of the service.
3835 * nspawn may now be invoked without a controlling TTY. This
3836 makes it suitable for invocation as its own service. This
3837 may be used to set up a simple containerized server system
3838 using only core OS tools.
3840 * systemd and nspawn can now accept socket file descriptors
3841 when they are started for socket activation. This enables
3842 implementation of socket activated nspawn
3843 containers. i.e. think about autospawning an entire OS image
3844 when the first SSH or HTTP connection is received. We expect
3845 that similar functionality will also be added to libvirt-lxc
3848 * journalctl will now suppress ANSI color codes when
3849 presenting log data.
3851 * systemctl will no longer show control group information for
3852 a unit if a the control group is empty anyway.
3854 * logind can now automatically suspend/hibernate/shutdown the
3857 * /etc/machine-info and hostnamed now also expose the chassis
3858 type of the system. This can be used to determine whether
3859 the local system is a laptop, desktop, handset or
3860 tablet. This information may either be configured by the
3861 user/vendor or is automatically determined from ACPI and DMI
3862 information if possible.
3864 * A number of PolicyKit actions are now bound together with
3865 "imply" rules. This should simplify creating UIs because
3866 many actions will now authenticate similar ones as well.
3868 * Unit files learnt a new condition ConditionACPower= which
3869 may be used to conditionalize a unit depending on whether an
3870 AC power source is connected or not, of whether the system
3871 is running on battery power.
3873 * systemctl gained a new "is-failed" verb that may be used in
3874 shell scripts and suchlike to check whether a specific unit
3875 is in the "failed" state.
3877 * The EnvironmentFile= setting in unit files now supports file
3878 globbing, and can hence be used to easily read a number of
3879 environment files at once.
3881 * systemd will no longer detect and recognize specific
3882 distributions. All distribution-specific #ifdeffery has been
3883 removed, systemd is now fully generic and
3884 distribution-agnostic. Effectively, not too much is lost as
3885 a lot of the code is still accessible via explicit configure
3886 switches. However, support for some distribution specific
3887 legacy configuration file formats has been dropped. We
3888 recommend distributions to simply adopt the configuration
3889 files everybody else uses now and convert the old
3890 configuration from packaging scripts. Most distributions
3891 already did that. If that's not possible or desirable,
3892 distributions are welcome to forward port the specific
3893 pieces of code locally from the git history.
3895 * When logging a message about a unit systemd will now always
3896 log the unit name in the message meta data.
3898 * localectl will now also discover system locale data that is
3899 not stored in locale archives, but directly unpacked.
3901 * logind will no longer unconditionally use framebuffer
3902 devices as seat masters, i.e. as devices that are required
3903 to be existing before a seat is considered preset. Instead,
3904 it will now look for all devices that are tagged as
3905 "seat-master" in udev. By default framebuffer devices will
3906 be marked as such, but depending on local systems other
3907 devices might be marked as well. This may be used to
3908 integrate graphics cards using closed source drivers (such
3909 as NVidia ones) more nicely into logind. Note however, that
3910 we recommend using the open source NVidia drivers instead,
3911 and no udev rules for the closed-source drivers will be
3912 shipped from us upstream.
3914 Contributions from: Adam Williamson, Alessandro Crismani, Auke
3915 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
3916 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
3917 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
3918 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
3919 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
3920 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
3921 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
3922 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
3923 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
3924 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
3925 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
3930 * udev gained support for loading additional device properties
3931 from an indexed database that is keyed by vendor/product IDs
3932 and similar device identifiers. For the beginning this
3933 "hwdb" is populated with data from the well-known PCI and
3934 USB database, but also includes PNP, ACPI and OID data. In
3935 the longer run this indexed database shall grow into
3936 becoming the one central database for non-essential
3937 userspace device metadata. Previously, data from the PCI/USB
3938 database was only attached to select devices, since the
3939 lookup was a relatively expensive operation due to O(n) time
3940 complexity (with n being the number of entries in the
3941 database). Since this is now O(1), we decided to add in this
3942 data for all devices where this is available, by
3943 default. Note that the indexed database needs to be rebuilt
3944 when new data files are installed. To achieve this you need
3945 to update your packaging scripts to invoke "udevadm hwdb
3946 --update" after installation of hwdb data files. For
3947 RPM-based distributions we introduced the new
3948 %udev_hwdb_update macro for this purpose.
3950 * The Journal gained support for the "Message Catalog", an
3951 indexed database to link up additional information with
3952 journal entries. For further details please check:
3954 http://www.freedesktop.org/wiki/Software/systemd/catalog
3956 The indexed message catalog database also needs to be
3957 rebuilt after installation of message catalog files. Use
3958 "journalctl --update-catalog" for this. For RPM-based
3959 distributions we introduced the %journal_catalog_update
3960 macro for this purpose.
3962 * The Python Journal bindings gained support for the standard
3963 Python logging framework.
3965 * The Journal API gained new functions for checking whether
3966 the underlying file system of a journal file is capable of
3967 properly reporting file change notifications, or whether
3968 applications that want to reflect journal changes "live"
3969 need to recheck journal files continuously in appropriate
3972 * It is now possible to set the "age" field for tmpfiles
3973 entries to 0, indicating that files matching this entry
3974 shall always be removed when the directories are cleaned up.
3976 * coredumpctl gained a new "gdb" verb which invokes gdb
3977 right-away on the selected coredump.
3979 * There's now support for "hybrid sleep" on kernels that
3980 support this, in addition to "suspend" and "hibernate". Use
3981 "systemctl hybrid-sleep" to make use of this.
3983 * logind's HandleSuspendKey= setting (and related settings)
3984 now gained support for a new "lock" setting to simply
3985 request the screen lock on all local sessions, instead of
3986 actually executing a suspend or hibernation.
3988 * systemd will now mount the EFI variables file system by
3991 * Socket units now gained support for configuration of the
3992 SMACK security label.
3994 * timedatectl will now output the time of the last and next
3995 daylight saving change.
3997 * We dropped support for various legacy and distro-specific
3998 concepts, such as insserv, early-boot SysV services
3999 (i.e. those for non-standard runlevels such as 'b' or 'S')
4000 or ArchLinux /etc/rc.conf support. We recommend the
4001 distributions who still need support this to either continue
4002 to maintain the necessary patches downstream, or find a
4003 different solution. (Talk to us if you have questions!)
4005 * Various systemd components will now bypass PolicyKit checks
4006 for root and otherwise handle properly if PolicyKit is not
4007 found to be around. This should fix most issues for
4008 PolicyKit-less systems. Quite frankly this should have been
4009 this way since day one. It is absolutely our intention to
4010 make systemd work fine on PolicyKit-less systems, and we
4011 consider it a bug if something does not work as it should if
4012 PolicyKit is not around.
4014 * For embedded systems it is now possible to build udev and
4015 systemd without blkid and/or kmod support.
4017 * "systemctl switch-root" is now capable of switching root
4018 more than once. I.e. in addition to transitions from the
4019 initrd to the host OS it is now possible to transition to
4020 further OS images from the host. This is useful to implement
4021 offline updating tools.
4023 * Various other additions have been made to the RPM macros
4024 shipped with systemd. Use %udev_rules_update() after
4025 installing new udev rules files. %_udevhwdbdir,
4026 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
4027 %_sysctldir are now available which resolve to the right
4028 directories for packages to place various data files in.
4030 * journalctl gained the new --full switch (in addition to
4031 --all, to disable ellipsation for long messages.
4033 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
4034 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
4035 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
4036 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
4037 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
4038 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
4039 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
4040 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
4041 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
4045 * journalctl gained new --since= and --until= switches to
4046 filter by time. It also now supports nice filtering for
4047 units via --unit=/-u.
4049 * Type=oneshot services may use ExecReload= and do the
4052 * The journal daemon now supports time-based rotation and
4053 vacuuming, in addition to the usual disk-space based
4056 * The journal will now index the available field values for
4057 each field name. This enables clients to show pretty drop
4058 downs of available match values when filtering. The bash
4059 completion of journalctl has been updated
4060 accordingly. journalctl gained a new switch -F to list all
4061 values a certain field takes in the journal database.
4063 * More service events are now written as structured messages
4064 to the journal, and made recognizable via message IDs.
4066 * The timedated, localed and hostnamed mini-services which
4067 previously only provided support for changing time, locale
4068 and hostname settings from graphical DEs such as GNOME now
4069 also have a minimal (but very useful) text-based client
4070 utility each. This is probably the nicest way to changing
4071 these settings from the command line now, especially since
4072 it lists available options and is fully integrated with bash
4075 * There's now a new tool "systemd-coredumpctl" to list and
4076 extract coredumps from the journal.
4078 * We now install a README each in /var/log/ and
4079 /etc/rc.d/init.d explaining where the system logs and init
4080 scripts went. This hopefully should help folks who go to
4081 that dirs and look into the otherwise now empty void and
4082 scratch their heads.
4084 * When user-services are invoked (by systemd --user) the
4085 $MANAGERPID env var is set to the PID of systemd.
4087 * SIGRTMIN+24 when sent to a --user instance will now result
4088 in immediate termination of systemd.
4090 * gatewayd received numerous feature additions such as a
4091 "follow" mode, for live syncing and filtering.
4093 * browse.html now allows filtering and showing detailed
4094 information on specific entries. Keyboard navigation and
4095 mouse screen support has been added.
4097 * gatewayd/journalctl now supports HTML5/JSON
4098 Server-Sent-Events as output.
4100 * The SysV init script compatibility logic will now
4101 heuristically determine whether a script supports the
4102 "reload" verb, and only then make this available as
4105 * "systemctl status --follow" has been removed, use "journalctl
4108 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
4109 have been removed since they are hardly useful to be
4112 * And I'd like to take the opportunity to specifically mention
4113 Zbigniew for his great contributions. Zbigniew, you rock!
4115 Contributions from: Andrew Eikum, Christian Hesse, Colin
4116 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
4117 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
4118 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
4119 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
4120 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
4121 Jędrzejewski-Szmek, Сковорода Никита Андреевич
4125 * If /etc/vconsole.conf is non-existent or empty we will no
4126 longer load any console font or key map at boot by
4127 default. Instead the kernel defaults will be left
4128 intact. This is definitely the right thing to do, as no
4129 configuration should mean no configuration, and hard-coding
4130 font names that are different on all archs is probably a bad
4131 idea. Also, the kernel default key map and font should be
4132 good enough for most cases anyway, and mostly identical to
4133 the userspace fonts/key maps we previously overloaded them
4134 with. If distributions want to continue to default to a
4135 non-kernel font or key map they should ship a default
4136 /etc/vconsole.conf with the appropriate contents.
4138 Contributions from: Colin Walters, Daniel J Walsh, Dave
4139 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
4140 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
4144 * journalctl gained a new --cursor= switch to show entries
4145 starting from the specified location in the journal.
4147 * We now enforce a size limit on journal entry fields exported
4148 with "-o json" in journalctl. Fields larger than 4K will be
4149 assigned null. This can be turned off with --all.
4151 * An (optional) journal gateway daemon is now available as
4152 "systemd-journal-gatewayd.service". This service provides
4153 access to the journal via HTTP and JSON. This functionality
4154 will be used to implement live log synchronization in both
4155 pull and push modes, but has various other users too, such
4156 as easy log access for debugging of embedded devices. Right
4157 now it is already useful to retrieve the journal via HTTP:
4159 # systemctl start systemd-journal-gatewayd.service
4160 # wget http://localhost:19531/entries
4162 This will download the journal contents in a
4163 /var/log/messages compatible format. The same as JSON:
4165 # curl -H"Accept: application/json" http://localhost:19531/entries
4167 This service is also accessible via a web browser where a
4168 single static HTML5 app is served that uses the JSON logic
4169 to enable the user to do some basic browsing of the
4170 journal. This will be extended later on. Here's an example
4171 screenshot of this app in its current state:
4173 http://0pointer.de/public/journal-gatewayd
4175 Contributions from: Kay Sievers, Lennart Poettering, Robert
4176 Milasan, Tom Gundersen
4180 * The bash completion logic is now available for journalctl
4183 * We do not mount the "cpuset" controller anymore together with
4184 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
4185 started if no parameters are assigned to it. "cpuset" hence
4186 broke code that assumed it it could create "cpu" groups and
4189 * journalctl -f will now subscribe to terminal size changes,
4190 and line break accordingly.
4192 Contributions from: Dave Reisner, Kay Sievers, Lennart
4193 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
4197 * nspawn will now create a symlink /etc/localtime in the
4198 container environment, copying the host's timezone
4199 setting. Previously this has been done via a bind mount, but
4200 since symlinks cannot be bind mounted this has now been
4201 changed to create/update the appropriate symlink.
4203 * journalctl -n's line number argument is now optional, and
4204 will default to 10 if omitted.
4206 * journald will now log the maximum size the journal files may
4207 take up on disk. This is particularly useful if the default
4208 built-in logic of determining this parameter from the file
4209 system size is used. Use "systemctl status
4210 systemd-journald.service" to see this information.
4212 * The multi-seat X wrapper tool has been stripped down. As X
4213 is now capable of enumerating graphics devices via udev in a
4214 seat-aware way the wrapper is not strictly necessary
4215 anymore. A stripped down temporary stop-gap is still shipped
4216 until the upstream display managers have been updated to
4217 fully support the new X logic. Expect this wrapper to be
4218 removed entirely in one of the next releases.
4220 * HandleSleepKey= in logind.conf has been split up into
4221 HandleSuspendKey= and HandleHibernateKey=. The old setting
4222 is not available anymore. X11 and the kernel are
4223 distinguishing between these keys and we should too. This
4224 also means the inhibition lock for these keys has been split
4227 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
4228 Poettering, Lukas Nykryn, Václav Pavlín
4232 * Whenever a unit changes state we will now log this to the
4233 journal and show along the unit's own log output in
4236 * ConditionPathIsMountPoint= can now properly detect bind
4237 mount points too. (Previously, a bind mount of one file
4238 system to another place in the same file system could not be
4239 detected as mount, since they shared struct stat's st_dev
4242 * We will now mount the cgroup controllers cpu, cpuacct,
4243 cpuset and the controllers net_cls, net_prio together by
4246 * nspawn containers will now have a virtualized boot
4247 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
4248 over with a randomized ID at container initialization). This
4249 has the effect of making "journalctl -b" do the right thing
4252 * The JSON output journal serialization has been updated not
4253 to generate "endless" list objects anymore, but rather one
4254 JSON object per line. This is more in line how most JSON
4255 parsers expect JSON objects. The new output mode
4256 "json-pretty" has been added to provide similar output, but
4257 neatly aligned for readability by humans.
4259 * We dropped all explicit sync() invocations in the shutdown
4260 code. The kernel does this implicitly anyway in the kernel
4261 reboot() syscall. halt(8)'s -n option is now a compatibility
4264 * We now support virtualized reboot() in containers, as
4265 supported by newer kernels. We will fall back to exit() if
4266 CAP_SYS_REBOOT is not available to the container. Also,
4267 nspawn makes use of this now and will actually reboot the
4268 container if the containerized OS asks for that.
4270 * journalctl will only show local log output by default
4271 now. Use --merge (-m) to show remote log output, too.
4273 * libsystemd-journal gained the new sd_journal_get_usage()
4274 call to determine the current disk usage of all journal
4275 files. This is exposed in the new "journalctl --disk-usage"
4278 * journald gained a new configuration setting SplitMode= in
4279 journald.conf which may be used to control how user journals
4280 are split off. See journald.conf(5) for details.
4282 * A new condition type ConditionFileNotEmpty= has been added.
4284 * tmpfiles' "w" lines now support file globbing, to write
4285 multiple files at once.
4287 * We added Python bindings for the journal submission
4288 APIs. More Python APIs for a number of selected APIs will
4289 likely follow. Note that we intend to add native bindings
4290 only for the Python language, as we consider it common
4291 enough to deserve bindings shipped within systemd. There are
4292 various projects outside of systemd that provide bindings
4293 for languages such as PHP or Lua.
4295 * Many conditions will now resolve specifiers such as %i. In
4296 addition, PathChanged= and related directives of .path units
4297 now support specifiers as well.
4299 * There's now a new RPM macro definition for the system preset
4302 * journald will now warn if it ca not forward a message to the
4303 syslog daemon because its socket is full.
4305 * timedated will no longer write or process /etc/timezone,
4306 except on Debian. As we do not support late mounted /usr
4307 anymore /etc/localtime always being a symlink is now safe,
4308 and hence the information in /etc/timezone is not necessary
4311 * logind will now always reserve one VT for a text getty (VT6
4312 by default). Previously if more than 6 X sessions where
4313 started they took up all the VTs with auto-spawned gettys,
4314 so that no text gettys were available anymore.
4316 * udev will now automatically inform the btrfs kernel logic
4317 about btrfs RAID components showing up. This should make
4318 simple hotplug based btrfs RAID assembly work.
4320 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
4321 (but not for its children which will stay at the kernel
4322 default). This should allow setups with a lot more listening
4325 * systemd will now always pass the configured timezone to the
4326 kernel at boot. timedated will do the same when the timezone
4329 * logind's inhibition logic has been updated. By default,
4330 logind will now handle the lid switch, the power and sleep
4331 keys all the time, even in graphical sessions. If DEs want
4332 to handle these events on their own they should take the new
4333 handle-power-key, handle-sleep-key and handle-lid-switch
4334 inhibitors during their runtime. A simple way to achieve
4335 that is to invoke the DE wrapped in an invocation of:
4337 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
4339 * Access to unit operations is now checked via SELinux taking
4340 the unit file label and client process label into account.
4342 * systemd will now notify the administrator in the journal
4343 when he over-mounts a non-empty directory.
4345 * There are new specifiers that are resolved in unit files,
4346 for the host name (%H), the machine ID (%m) and the boot ID
4349 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
4350 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
4351 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
4352 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
4353 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
4354 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
4355 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
4359 * Support for reading structured kernel messages from
4360 /dev/kmsg has now been added and is enabled by default.
4362 * Support for reading kernel messages from /proc/kmsg has now
4363 been removed. If you want kernel messages in the journal
4364 make sure to run a recent kernel (>= 3.5) that supports
4365 reading structured messages from /dev/kmsg (see
4366 above). /proc/kmsg is now exclusive property of classic
4367 syslog daemons again.
4369 * The libudev API gained the new
4370 udev_device_new_from_device_id() call.
4372 * The logic for file system namespace (ReadOnlyDirectory=,
4373 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
4374 require pivot_root() anymore. This means fewer temporary
4375 directories are created below /tmp for this feature.
4377 * nspawn containers will now see and receive all submounts
4378 made on the host OS below the root file system of the
4381 * Forward Secure Sealing is now supported for Journal files,
4382 which provide cryptographical sealing of journal files so
4383 that attackers cannot alter log history anymore without this
4384 being detectable. Lennart will soon post a blog story about
4385 this explaining it in more detail.
4387 * There are two new service settings RestartPreventExitStatus=
4388 and SuccessExitStatus= which allow configuration of exit
4389 status (exit code or signal) which will be excepted from the
4390 restart logic, resp. consider successful.
4392 * journalctl gained the new --verify switch that can be used
4393 to check the integrity of the structure of journal files and
4394 (if Forward Secure Sealing is enabled) the contents of
4397 * nspawn containers will now be run with /dev/stdin, /dev/fd/
4398 and similar symlinks pre-created. This makes running shells
4399 as container init process a lot more fun.
4401 * The fstab support can now handle PARTUUID= and PARTLABEL=
4404 * A new ConditionHost= condition has been added to match
4405 against the hostname (with globs) and machine ID. This is
4406 useful for clusters where a single OS image is used to
4407 provision a large number of hosts which shall run slightly
4408 different sets of services.
4410 * Services which hit the restart limit will now be placed in a
4413 Contributions from: Bertram Poettering, Dave Reisner, Huang
4414 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
4415 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
4419 * When running in --user mode systemd will now become a
4420 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
4421 tree a lot more organized.
4423 * A new PartOf= unit dependency type has been introduced that
4424 may be used to group services in a natural way.
4426 * "systemctl enable" may now be used to enable instances of
4429 * journalctl now prints error log levels in red, and
4430 warning/notice log levels in bright white. It also supports
4431 filtering by log level now.
4433 * cgtop gained a new -n switch (similar to top), to configure
4434 the maximum number of iterations to run for. It also gained
4435 -b, to run in batch mode (accepting no input).
4437 * The suffix ".service" may now be omitted on most systemctl
4438 command lines involving service unit names.
4440 * There's a new bus call in logind to lock all sessions, as
4441 well as a loginctl verb for it "lock-sessions".
4443 * libsystemd-logind.so gained a new call sd_journal_perror()
4444 that works similar to libc perror() but logs to the journal
4445 and encodes structured information about the error number.
4447 * /etc/crypttab entries now understand the new keyfile-size=
4450 * shutdown(8) now can send a (configurable) wall message when
4451 a shutdown is cancelled.
4453 * The mount propagation mode for the root file system will now
4454 default to "shared", which is useful to make containers work
4455 nicely out-of-the-box so that they receive new mounts from
4456 the host. This can be undone locally by running "mount
4457 --make-rprivate /" if needed.
4459 * The prefdm.service file has been removed. Distributions
4460 should maintain this unit downstream if they intend to keep
4461 it around. However, we recommend writing normal unit files
4462 for display managers instead.
4464 * Since systemd is a crucial part of the OS we will now
4465 default to a number of compiler switches that improve
4466 security (hardening) such as read-only relocations, stack
4467 protection, and suchlike.
4469 * The TimeoutSec= setting for services is now split into
4470 TimeoutStartSec= and TimeoutStopSec= to allow configuration
4471 of individual time outs for the start and the stop phase of
4474 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
4475 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
4476 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
4477 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
4478 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
4479 Gundersen, Zbigniew Jędrzejewski-Szmek
4483 * The journal and id128 C APIs are now fully documented as man
4486 * Extra safety checks have been added when transitioning from
4487 the initial RAM disk to the main system to avoid accidental
4490 * /etc/crypttab entries now understand the new keyfile-offset=
4493 * systemctl -t can now be used to filter by unit load state.
4495 * The journal C API gained the new sd_journal_wait() call to
4496 make writing synchronous journal clients easier.
4498 * journalctl gained the new -D switch to show journals from a
4501 * journalctl now displays a special marker between log
4502 messages of two different boots.
4504 * The journal is now explicitly flushed to /var via a service
4505 systemd-journal-flush.service, rather than implicitly simply
4506 by seeing /var/log/journal to be writable.
4508 * journalctl (and the journal C APIs) can now match for much
4509 more complex expressions, with alternatives and
4512 * When transitioning from the initial RAM disk to the main
4513 system we will now kill all processes in a killing spree to
4514 ensure no processes stay around by accident.
4516 * Three new specifiers may be used in unit files: %u, %h, %s
4517 resolve to the user name, user home directory resp. user
4518 shell. This is useful for running systemd user instances.
4520 * We now automatically rotate journal files if their data
4521 object hash table gets a fill level > 75%. We also size the
4522 hash table based on the configured maximum file size. This
4523 together should lower hash collisions drastically and thus
4524 speed things up a bit.
4526 * journalctl gained the new "--header" switch to introspect
4527 header data of journal files.
4529 * A new setting SystemCallFilters= has been added to services
4530 which may be used to apply blacklists or whitelists to
4531 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
4533 * nspawn gained a new --link-journal= switch (and quicker: -j)
4534 to link the container journal with the host. This makes it
4535 very easy to centralize log viewing on the host for all
4536 guests while still keeping the journal files separated.
4538 * Many bugfixes and optimizations
4540 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
4541 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
4542 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
4547 * Several tools now understand kernel command line arguments,
4548 which are only read when run in an initial RAM disk. They
4549 usually follow closely their normal counterparts, but are
4552 * There's a new tool to analyze the readahead files that are
4553 automatically generated at boot. Use:
4555 /usr/lib/systemd/systemd-readahead analyze /.readahead
4557 * We now provide an early debug shell on tty9 if this enabled. Use:
4559 systemctl enable debug-shell.service
4561 * All plymouth related units have been moved into the Plymouth
4562 package. Please make sure to upgrade your Plymouth version
4565 * systemd-tmpfiles now supports getting passed the basename of
4566 a configuration file only, in which case it will look for it
4567 in all appropriate directories automatically.
4569 * udevadm info now takes a /dev or /sys path as argument, and
4570 does the right thing. Example:
4572 udevadm info /dev/sda
4573 udevadm info /sys/class/block/sda
4575 * systemctl now prints a warning if a unit is stopped but a
4576 unit that might trigger it continues to run. Example: a
4577 service is stopped but the socket that activates it is left
4580 * "systemctl status" will now mention if the log output was
4581 shortened due to rotation since a service has been started.
4583 * The journal API now exposes functions to determine the
4584 "cutoff" times due to rotation.
4586 * journald now understands SIGUSR1 and SIGUSR2 for triggering
4587 immediately flushing of runtime logs to /var if possible,
4588 resp. for triggering immediate rotation of the journal
4591 * It is now considered an error if a service is attempted to
4592 be stopped that is not loaded.
4594 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
4596 * systemd-analyze now supports Python 3
4598 * tmpfiles now supports cleaning up directories via aging
4599 where the first level dirs are always kept around but
4600 directories beneath it automatically aged. This is enabled
4601 by prefixing the age field with '~'.
4603 * Seat objects now expose CanGraphical, CanTTY properties
4604 which is required to deal with very fast bootups where the
4605 display manager might be running before the graphics drivers
4606 completed initialization.
4608 * Seat objects now expose a State property.
4610 * We now include RPM macros for service enabling/disabling
4611 based on the preset logic. We recommend RPM based
4612 distributions to make use of these macros if possible. This
4613 makes it simpler to reuse RPM spec files across
4616 * We now make sure that the collected systemd unit name is
4617 always valid when services log to the journal via
4620 * There's a new man page kernel-command-line(7) detailing all
4621 command line options we understand.
4623 * The fstab generator may now be disabled at boot by passing
4624 fstab=0 on the kernel command line.
4626 * A new kernel command line option modules-load= is now understood
4627 to load a specific kernel module statically, early at boot.
4629 * Unit names specified on the systemctl command line are now
4630 automatically escaped as needed. Also, if file system or
4631 device paths are specified they are automatically turned
4632 into the appropriate mount or device unit names. Example:
4634 systemctl status /home
4635 systemctl status /dev/sda
4637 * The SysVConsole= configuration option has been removed from
4638 system.conf parsing.
4640 * The SysV search path is no longer exported on the D-Bus
4643 * The Names= option is been removed from unit file parsing.
4645 * There's a new man page bootup(7) detailing the boot process.
4647 * Every unit and every generator we ship with systemd now
4648 comes with full documentation. The self-explanatory boot is
4651 * A couple of services gained "systemd-" prefixes in their
4652 name if they wrap systemd code, rather than only external
4653 code. Among them fsck@.service which is now
4654 systemd-fsck@.service.
4656 * The HaveWatchdog property has been removed from the D-Bus
4659 * systemd.confirm_spawn= on the kernel command line should now
4662 * There's a new man page crypttab(5) which details all options
4663 we actually understand.
4665 * systemd-nspawn gained a new --capability= switch to pass
4666 additional capabilities to the container.
4668 * timedated will now read known NTP implementation unit names
4669 from /usr/lib/systemd/ntp-units.d/*.list,
4670 systemd-timedated-ntp.target has been removed.
4672 * journalctl gained a new switch "-b" that lists log data of
4673 the current boot only.
4675 * The notify socket is in the abstract namespace again, in
4676 order to support daemons which chroot() at start-up.
4678 * There is a new Storage= configuration option for journald
4679 which allows configuration of where log data should go. This
4680 also provides a way to disable journal logging entirely, so
4681 that data collected is only forwarded to the console, the
4682 kernel log buffer or another syslog implementation.
4684 * Many bugfixes and optimizations
4686 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
4687 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
4688 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
4689 Shawn Landden, Tom Gundersen
4693 * "systemctl help <unit>" now shows the man page if one is
4696 * Several new man pages have been added.
4698 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
4699 MaxLevelConsole= can now be specified in
4700 journald.conf. These options allow reducing the amount of
4701 data stored on disk or forwarded by the log level.
4703 * TimerSlackNSec= can now be specified in system.conf for
4704 PID1. This allows system-wide power savings.
4706 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
4707 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
4712 * logind is now capable of (optionally) handling power and
4713 sleep keys as well as the lid switch.
4715 * journalctl now understands the syntax "journalctl
4716 /usr/bin/avahi-daemon" to get all log output of a specific
4719 * CapabilityBoundingSet= in system.conf now also influences
4720 the capability bound set of usermode helpers of the kernel.
4722 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
4723 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
4724 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
4725 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
4729 * Note that we skipped 139 releases here in order to set the
4730 new version to something that is greater than both udev's
4731 and systemd's most recent version number.
4733 * udev: all udev sources are merged into the systemd source tree now.
4734 All future udev development will happen in the systemd tree. It
4735 is still fully supported to use the udev daemon and tools without
4736 systemd running, like in initramfs or other init systems. Building
4737 udev though, will require the *build* of the systemd tree, but
4738 udev can be properly *run* without systemd.
4740 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
4741 should be used to create dead device nodes as workarounds for broken
4744 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
4745 no longer supported. udev_monitor_new_from_netlink() needs to be
4746 used to subscribe to events.
4748 * udev: when udevd is started by systemd, processes which are left
4749 behind by forking them off of udev rules, are unconditionally cleaned
4750 up and killed now after the event handling has finished. Services or
4751 daemons must be started as systemd services. Services can be
4752 pulled-in by udev to get started, but they can no longer be directly
4753 forked by udev rules.
4755 * udev: the daemon binary is called systemd-udevd now and installed
4756 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
4757 to adapt to that, create symlink, or rename the binary after building
4760 * libudev no longer provides these symbols:
4761 udev_monitor_from_socket()
4762 udev_queue_get_failed_list_entry()
4763 udev_get_{dev,sys,run}_path()
4764 The versions number was bumped and symbol versioning introduced.
4766 * systemd-loginctl and systemd-journalctl have been renamed
4767 to loginctl and journalctl to match systemctl.
4769 * The config files: /etc/systemd/systemd-logind.conf and
4770 /etc/systemd/systemd-journald.conf have been renamed to
4771 logind.conf and journald.conf. Package updates should rename
4772 the files to the new names on upgrade.
4774 * For almost all files the license is now LGPL2.1+, changed
4775 from the previous GPL2.0+. Exceptions are some minor stuff
4776 of udev (which will be changed to LGPL2.1 eventually, too),
4777 and the MIT licensed sd-daemon.[ch] library that is suitable
4778 to be used as drop-in files.
4780 * systemd and logind now handle system sleep states, in
4781 particular suspending and hibernating.
4783 * logind now implements a sleep/shutdown/idle inhibiting logic
4784 suitable for a variety of uses. Soonishly Lennart will blog
4785 about this in more detail.
4787 * var-run.mount and var-lock.mount are no longer provided
4788 (which prevously bind mounted these directories to their new
4789 places). Distributions which have not converted these
4790 directories to symlinks should consider stealing these files
4791 from git history and add them downstream.
4793 * We introduced the Documentation= field for units and added
4794 this to all our shipped units. This is useful to make it
4795 easier to explore the boot and the purpose of the various
4798 * All smaller setup units (such as
4799 systemd-vconsole-setup.service) now detect properly if they
4800 are run in a container and are skipped when
4801 appropriate. This guarantees an entirely noise-free boot in
4802 Linux container environments such as systemd-nspawn.
4804 * A framework for implementing offline system updates is now
4805 integrated, for details see:
4806 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
4808 * A new service type Type=idle is available now which helps us
4809 avoiding ugly interleaving of getty output and boot status
4812 * There's now a system-wide CapabilityBoundingSet= option to
4813 globally reduce the set of capabilities for the
4814 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
4815 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
4816 even CAP_NET_ADMIN system-wide for secure systems.
4818 * There are now system-wide DefaultLimitXXX= options to
4819 globally change the defaults of the various resource limits
4820 for all units started by PID 1.
4822 * Harald Hoyer's systemd test suite has been integrated into
4823 systemd which allows easy testing of systemd builds in qemu
4824 and nspawn. (This is really awesome! Ask us for details!)
4826 * The fstab parser is now implemented as generator, not inside
4829 * systemctl will now warn you if .mount units generated from
4830 /etc/fstab are out of date due to changes in fstab that
4831 have not been read by systemd yet.
4833 * systemd is now suitable for usage in initrds. Dracut has
4834 already been updated to make use of this. With this in place
4835 initrds get a slight bit faster but primarily are much
4836 easier to introspect and debug since "systemctl status" in
4837 the host system can be used to introspect initrd services,
4838 and the journal from the initrd is kept around too.
4840 * systemd-delta has been added, a tool to explore differences
4841 between user/admin configuration and vendor defaults.
4843 * PrivateTmp= now affects both /tmp and /var/tmp.
4845 * Boot time status messages are now much prettier and feature
4846 proper english language. Booting up systemd has never been
4849 * Read-ahead pack files now include the inode number of all
4850 files to pre-cache. When the inode changes the pre-caching
4851 is not attempted. This should be nicer to deal with updated
4852 packages which might result in changes of read-ahead
4855 * We now temporaritly lower the kernel's read_ahead_kb variable
4856 when collecting read-ahead data to ensure the kernel's
4857 built-in read-ahead does not add noise to our measurements
4858 of necessary blocks to pre-cache.
4860 * There's now RequiresMountsFor= to add automatic dependencies
4861 for all mounts necessary for a specific file system path.
4863 * MountAuto= and SwapAuto= have been removed from
4864 system.conf. Mounting file systems at boot has to take place
4867 * nspawn now learned a new switch --uuid= to set the machine
4868 ID on the command line.
4870 * nspawn now learned the -b switch to automatically search
4873 * vt102 is now the default TERM for serial TTYs, upgraded from
4876 * systemd-logind now works on VT-less systems.
4878 * The build tree has been reorganized. The individual
4879 components now have directories of their own.
4881 * A new condition type ConditionPathIsReadWrite= is now available.
4883 * nspawn learned the new -C switch to create cgroups for the
4884 container in other hierarchies.
4886 * We now have support for hardware watchdogs, configurable in
4889 * The scheduled shutdown logic now has a public API.
4891 * We now mount /tmp as tmpfs by default, but this can be
4892 masked and /etc/fstab can override it.
4894 * Since udisks does not make use of /media anymore we are not
4895 mounting a tmpfs on it anymore.
4897 * journalctl gained a new --local switch to only interleave
4898 locally generated journal files.
4900 * We can now load the IMA policy at boot automatically.
4902 * The GTK tools have been split off into a systemd-ui.
4904 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
4905 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
4906 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
4907 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
4908 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
4909 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
4910 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
4911 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
4912 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
4917 * This is mostly a bugfix release
4919 * Support optional initialization of the machine ID from the
4920 KVM or container configured UUID.
4922 * Support immediate reboots with "systemctl reboot -ff"
4924 * Show /etc/os-release data in systemd-analyze output
4926 * Many bugfixes for the journal, including endianness fixes and
4927 ensuring that disk space enforcement works
4929 * sd-login.h is C++ comptaible again
4931 * Extend the /etc/os-release format on request of the Debian
4934 * We now refuse non-UTF8 strings used in various configuration
4935 and unit files. This is done to ensure we do not pass invalid
4936 data over D-Bus or expose it elsewhere.
4938 * Register Mimo USB Screens as suitable for automatic seat
4941 * Read SELinux client context from journal clients in a race
4944 * Reorder configuration file lookup order. /etc now always
4945 overrides /run in order to allow the administrator to always
4946 and unconditionally override vendor supplied or
4947 automatically generated data.
4949 * The various user visible bits of the journal now have man
4950 pages. We still lack man pages for the journal API calls
4953 * We now ship all man pages in HTML format again in the
4956 Contributions from: Dave Reisner, Dirk Eibach, Frederic
4957 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
4958 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
4963 * This is mostly a bugfix release
4965 * systems lacking /etc/os-release are no longer supported.
4967 * Various functionality updates to libsystemd-login.so
4969 * Track class of PAM logins to distinguish greeters from
4972 Contributions from: Kay Sievers, Lennart Poettering, Michael
4977 * This is an important bugfix release for v41.
4979 * Building man pages is now optional which should be useful
4980 for those building systemd from git but unwilling to install
4983 * Watchdog support for supervising services is now usable. In
4984 a future release support for hardware watchdogs
4985 (i.e. /dev/watchdog) will be added building on this.
4987 * Service start rate limiting is now configurable and can be
4988 turned off per service. When a start rate limit is hit a
4989 reboot can automatically be triggered.
4991 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
4993 Contributions from: Benjamin Franzke, Bill Nottingham,
4994 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
4995 Schmidt, Michał Górny, Piotr Drąg
4999 * The systemd binary is installed /usr/lib/systemd/systemd now;
5000 An existing /sbin/init symlink needs to be adapted with the
5003 * The code that loads kernel modules has been ported to invoke
5004 libkmod directly, instead of modprobe. This means we do not
5005 support systems with module-init-tools anymore.
5007 * Watchdog support is now already useful, but still not
5010 * A new kernel command line option systemd.setenv= is
5011 understood to set system wide environment variables
5012 dynamically at boot.
5014 * We now limit the set of capabilities of systemd-journald.
5016 * We now set SIGPIPE to ignore by default, since it only is
5017 useful in shell pipelines, and has little use in general
5018 code. This can be disabled with IgnoreSIPIPE=no in unit
5021 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
5022 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
5027 * This is mostly a bugfix release
5029 * We now expose the reason why a service failed in the
5030 "Result" D-Bus property.
5032 * Rudimentary service watchdog support (will be completed over
5033 the next few releases.)
5035 * When systemd forks off in order execute some service we will
5036 now immediately changes its argv[0] to reflect which process
5037 it will execute. This is useful to minimize the time window
5038 with a generic argv[0], which makes bootcharts more useful
5040 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
5041 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
5042 Mike Kazantsev, Ray Strode
5046 * This is mostly a test release, but incorporates many
5049 * New systemd-cgtop tool to show control groups by their
5052 * Linking against libacl for ACLs is optional again. If
5053 disabled, support tracking device access for active logins
5054 goes becomes unavailable, and so does access to the user
5055 journals by the respective users.
5057 * If a group "adm" exists, journal files are automatically
5058 owned by them, thus allow members of this group full access
5059 to the system journal as well as all user journals.
5061 * The journal now stores the SELinux context of the logging
5062 client for all entries.
5064 * Add C++ inclusion guards to all public headers
5066 * New output mode "cat" in the journal to print only text
5067 messages, without any meta data like date or time.
5069 * Include tiny X server wrapper as a temporary stop-gap to
5070 teach XOrg udev display enumeration. This is used by display
5071 managers such as gdm, and will go away as soon as XOrg
5072 learned native udev hotplugging for display devices.
5074 * Add new systemd-cat tool for executing arbitrary programs
5075 with STDERR/STDOUT connected to the journal. Can also act as
5076 BSD logger replacement, and does so by default.
5078 * Optionally store all locally generated coredumps in the
5079 journal along with meta data.
5081 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
5082 writing short strings to files (for usage for /sys), and for
5083 creating symlinks, character and block device nodes.
5085 * New unit file option ControlGroupPersistent= to make cgroups
5086 persistent, following the mechanisms outlined in
5087 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
5089 * Support multiple local RTCs in a sane way
5091 * No longer monopolize IO when replaying readahead data on
5092 rotating disks, since we might starve non-file-system IO to
5093 death, since fanotify() will not see accesses done by blkid,
5096 * Do not show kernel threads in systemd-cgls anymore, unless
5097 requested with new -k switch.
5099 Contributions from: Dan Horák, Kay Sievers, Lennart
5100 Poettering, Michal Schmidt
5104 * This is mostly a test release, but incorporates many
5107 * The git repository moved to:
5108 git://anongit.freedesktop.org/systemd/systemd
5109 ssh://git.freedesktop.org/git/systemd/systemd
5111 * First release with the journal
5112 http://0pointer.de/blog/projects/the-journal.html
5114 * The journal replaces both systemd-kmsg-syslogd and
5115 systemd-stdout-bridge.
5117 * New sd_pid_get_unit() API call in libsystemd-logind
5119 * Many systemadm clean-ups
5121 * Introduce remote-fs-pre.target which is ordered before all
5122 remote mounts and may be used to start services before all
5125 * Added Mageia support
5127 * Add bash completion for systemd-loginctl
5129 * Actively monitor PID file creation for daemons which exit in
5130 the parent process before having finished writing the PID
5131 file in the daemon process. Daemons which do this need to be
5132 fixed (i.e. PID file creation must have finished before the
5133 parent exits), but we now react a bit more gracefully to them.
5135 * Add colourful boot output, mimicking the well-known output
5136 of existing distributions.
5138 * New option PassCredentials= for socket units, for
5139 compatibility with a recent kernel ABI breakage.
5141 * /etc/rc.local is now hooked in via a generator binary, and
5142 thus will no longer act as synchronization point during
5145 * systemctl list-unit-files now supports --root=.
5147 * systemd-tmpfiles now understands two new commands: z, Z for
5148 relabelling files according to the SELinux database. This is
5149 useful to apply SELinux labels to specific files in /sys,
5152 * Output of SysV services is now forwarded to both the console
5153 and the journal by default, not only just the console.
5155 * New man pages for all APIs from libsystemd-login.
5157 * The build tree got reorganized and a the build system is a
5158 lot more modular allowing embedded setups to specifically
5159 select the components of systemd they are interested in.
5161 * Support for Linux systems lacking the kernel VT subsystem is
5164 * configure's --with-rootdir= got renamed to
5165 --with-rootprefix= to follow the naming used by udev and
5168 * Unless specified otherwise we will now install to /usr instead
5169 of /usr/local by default.
5171 * Processes with '@' in argv[0][0] are now excluded from the
5172 final shut-down killing spree, following the logic explained
5174 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
5176 * All processes remaining in a service cgroup when we enter
5177 the START or START_PRE states are now killed with
5178 SIGKILL. That means it is no longer possible to spawn
5179 background processes from ExecStart= lines (which was never
5180 supported anyway, and bad style).
5182 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
5183 reloading of units together.
5185 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
5186 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
5187 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
5188 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
5189 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek