1 systemd System and Service Manager
5 * The systemd-efi-boot-generator functionality was merged into
6 systemd-gpt-auto-generator.
8 * systemd-networkd now supports Group Policy for vxlan devices. It can
9 be enabled via the new boolean configuration option called
10 'GroupPolicyExtension='.
12 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
13 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
14 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
20 * The python-systemd code has been removed from the systemd repository.
21 A new repository has been created which accommodates the code from
22 now on, and we kindly ask distributions to create a separate package
23 for this: https://github.com/systemd/python-systemd
25 * The systemd daemon will now reload its main configuration
26 (/etc/systemd/system.conf) on daemon-reload.
28 * sd-dhcp now exposes vendor specific extensions via
29 sd_dhcp_lease_get_vendor_specific().
31 * systemd-networkd gained a number of new configuration options.
33 - A new boolean configuration option for TAP devices called
34 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
35 device, thus allowing to send and receive GSO packets.
37 - A new tunnel configuration option called 'CopyDSCP='.
38 If enabled, the DSCP field of ip6 tunnels is copied into the
41 - A set of boolean bridge configuration options were added.
42 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
43 and 'UnicastFlood=' are now parsed by networkd and applied to the
44 respective bridge link device via the respective IFLA_BRPORT_*
47 - A new string configuration option to override the hostname sent
48 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
49 is true, networkd will use the configured hostname instead of the
50 system hostname when sending DHCP requests.
52 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
53 networkd will configure the IPv6 flow-label of the tunnel device
56 - The 'macvtap' virtual network devices are now supported, similar to
57 the already supported 'macvlan' devices.
59 * systemd-resolved now implements RFC5452 to improve resilience against
60 cache poisoning. Additionally, source port randomization is enabled
61 by default to further protect against DNS spoofing attacks.
63 * nss-mymachines now supports translating UIDs and GIDs of running
64 containers with user-namespaces enabled. If a container 'foo'
65 translates a host uid 'UID' to the container uid 'TUID', then
66 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
67 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
68 mapped as 'vg-foo-TGID'.
70 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
71 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
72 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
73 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
74 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
75 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
76 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
77 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
78 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
79 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
85 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
86 There are no known issues with current sysfs, and udev does not need
87 or should be used to work around such bugs.
89 * udev does no longer enable USB HID power management. Several reports
90 indicate, that some devices cannot handle that setting.
92 * The udev accelerometer helper was removed. The functionality
93 is now fully included in iio-sensor-proxy. But this means,
94 older iio-sensor-proxy versions will no longer provide
95 accelerometer/orientation data with this systemd version.
96 Please upgrade iio-sensor-proxy to version 1.0.
98 * networkd gained a new configuration option IPv6PrivacyExtensions=
99 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
100 for Stateless Address") on selected networks.
102 * For the sake of fewer build-time dependencies and less code in the
103 main repository, the python bindings are about to be removed in the
104 next release. A new repository has been created which accommodates
105 the code from now on, and we kindly ask distributions to create a
106 separate package for this. The removal will take place in v223.
108 https://github.com/systemd/python-systemd
110 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
111 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
112 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
113 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
114 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
115 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
116 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
117 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
118 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
119 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
121 -- Berlin, 2015-07-07
125 * The sd-bus.h and sd-event.h APIs have now been declared
126 stable and have been added to the official interface of
127 libsystemd.so. sd-bus implements an alternative D-Bus client
128 library, that is relatively easy to use, very efficient and
129 supports both classic D-Bus as well as kdbus as transport
130 backend. sd-event is a generic event loop abstraction that
131 is built around Linux epoll, but adds features such as event
132 prioritization or efficient timer handling. Both APIs are good
133 choices for C programs looking for a bus and/or event loop
134 implementation that is minimal and does not have to be
135 portable to other kernels.
137 * kdbus support is no longer compile-time optional. It is now
138 always built-in. However, it can still be disabled at
139 runtime using the kdbus=0 kernel command line setting, and
140 that setting may be changed to default to off, by specifying
141 --disable-kdbus at build-time. Note though that the kernel
142 command line setting has no effect if the kdbus.ko kernel
143 module is not installed, in which case kdbus is (obviously)
144 also disabled. We encourage all downstream distributions to
145 begin testing kdbus by adding it to the kernel images in the
146 development distributions, and leaving kdbus support in
149 * The minimal required util-linux version has been bumped to
152 * Support for chkconfig (--enable-chkconfig) was removed in
153 favor of calling an abstraction tool
154 /lib/systemd/systemd-sysv-install. This needs to be
155 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
156 in README for details.
158 * If there's a systemd unit and a SysV init script for the
159 same service name, and the user executes "systemctl enable"
160 for it (or a related call), then this will now enable both
161 (or execute the related operation on both), not just the
164 * The libudev API documentation has been converted from gtkdoc
167 * gudev has been removed from the systemd tree, it is now an
170 * The systemd-cgtop tool learnt a new --raw switch to generate
171 "raw" (machine parsable) output.
173 * networkd's IPForwarding= .network file setting learnt the
174 new setting "kernel", which ensures that networkd does not
175 change the IP forwarding sysctl from the default kernel
178 * The systemd-logind bus API now exposes a new boolean
179 property "Docked" that reports whether logind considers the
180 system "docked", i.e. connected to a docking station or not.
182 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
183 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
184 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
185 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
186 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
187 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
188 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
189 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
190 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
191 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
192 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
193 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
194 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
195 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
196 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
197 Fink, Zbigniew Jędrzejewski-Szmek
199 -- Berlin, 2015-06-19
203 * The gudev library has been extracted into a separate repository
204 available at: https://git.gnome.org/browse/libgudev/
205 It is now managed as part of the Gnome project. Distributions
206 are recommended to pass --disable-gudev to systemd and use
207 gudev from the Gnome project instead. gudev is still included
208 in systemd, for now. It will be removed soon, though. Please
209 also see the announcement-thread on systemd-devel:
210 http://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
212 * systemd now exposes a CPUUsageNSec= property for each
213 service unit on the bus, that contains the overall consumed
214 CPU time of a service (the sum of what each process of the
215 service consumed). This value is only available if
216 CPUAccounting= is turned on for a service, and is then shown
217 in the "systemctl status" output.
219 * Support for configuring alternative mappings of the old SysV
220 runlevels to systemd targets has been removed. They are now
221 hardcoded in a way that runlevels 2, 3, 4 all map to
222 multi-user.target and 5 to graphical.target (which
223 previously was already the default behaviour).
225 * The auto-mounter logic gained support for mount point
226 expiry, using a new TimeoutIdleSec= setting in .automount
227 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
229 * The EFI System Partition (ESP) as mounted to /boot by
230 systemd-efi-boot-generator will now be unmounted
231 automatically after 2 minutes of not being used. This should
232 minimize the risk of ESP corruptions.
234 * New /etc/fstab options x-systemd.requires= and
235 x-systemd.requires-mounts-for= are now supported to express
236 additional dependencies for mounts. This is useful for
237 journalling file systems that support external journal
238 devices or overlay file systems that require underlying file
239 systems to be mounted.
241 * systemd does not support direct live-upgrades (via systemctl
242 daemon-reexec) from versions older than v44 anymore. As no
243 distribution we are aware of shipped such old versions in a
244 stable release this should not be problematic.
246 * When systemd forks off a new per-connection service instance
247 it will now set the $REMOTE_ADDR environment variable to the
248 remote IP address, and $REMOTE_PORT environment variable to
249 the remote IP port. This behaviour is similar to the
250 corresponding environment variables defined by CGI.
252 * systemd-networkd gained support for uplink failure
253 detection. The BindCarrier= option allows binding interface
254 configuration dynamically to the link sense of other
255 interfaces. This is useful to achieve behaviour like in
258 * systemd-networkd gained support for configuring the DHCP
259 client identifier to use when requesting leases.
261 * systemd-networkd now has a per-network UseNTP= option to
262 configure whether NTP server information acquired via DHCP
263 is passed on to services like systemd-timesyncd.
265 * systemd-networkd gained support for vti6 tunnels.
267 * Note that systemd-networkd manages the sysctl variable
268 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
269 it is configured for since v219. The variable controls IP
270 forwarding, and is a per-interface alternative to the global
271 /proc/sys/net/ipv[46]/ip_forward. This setting is
272 configurable in the IPForward= option, which defaults to
273 "no". This means if networkd is used for an interface it is
274 no longer sufficient to set the global sysctl option to turn
275 on IP forwarding! Instead, the .network file option
276 IPForward= needs to be turned on! Note that the
277 implementation of this behaviour was broken in v219 and has
280 * Many bonding and vxlan options are now configurable in
283 * systemd-nspawn gained a new --property= setting to set unit
284 properties for the container scope. This is useful for
285 setting resource parameters (e.g "CPUShares=500") on
286 containers started from the command line.
288 * systemd-nspawn gained a new --private-users= switch to make
289 use of user namespacing available on recent Linux kernels.
291 * systemd-nspawn may now be called as part of a shell pipeline
292 in which case the pipes used for stdin and stdout are passed
293 directly to the process invoked in the container, without
294 indirection via a pseudo tty.
296 * systemd-nspawn gained a new switch to control the UNIX
297 signal to use when killing the init process of the container
300 * systemd-nspawn gained a new --overlay= switch for mounting
301 overlay file systems into the container using the new kernel
304 * When a container image is imported via systemd-importd and
305 the host file system is not btrfs, a loopback block device
306 file is created in /var/lib/machines.raw with a btrfs file
307 system inside. It is then mounted to /var/lib/machines to
308 enable btrfs features for container management. The loopback
309 file and btrfs file system is grown as needed when container
310 images are imported via systemd-importd.
312 * systemd-machined/systemd-importd gained support for btrfs
313 quota, to enforce container disk space limits on disk. This
314 is exposed in "machinectl set-limit".
316 * systemd-importd now can import containers from local .tar,
317 .raw and .qcow2 images, and export them to .tar and .raw. It
318 can also import dkr v2 images now from the network (on top
321 * systemd-importd gained support for verifying downloaded
322 images with gpg2 (previously only gpg1 was supported).
324 * systemd-machined, systemd-logind, systemd: most bus calls
325 are now accessible to unprivileged processes via
326 PolicyKit. Also, systemd-logind will now allow users to kill
327 their own sessions without further privileges or
330 * systemd-shutdownd has been removed. This service was
331 previously responsible for implementing scheduled shutdowns
332 as exposed in /usr/bin/shutdown's time parameter. This
333 functionality has now been moved into systemd-logind and is
334 accessible via a bus interface.
336 * "systemctl reboot" gained a new switch --firmware-setup that
337 can be used to reboot into the EFI firmware setup, if that
338 is available. systemd-logind now exposes an API on the bus
339 to trigger such reboots, in case graphical desktop UIs want
340 to cover this functionality.
342 * "systemctl enable", "systemctl disable" and "systemctl mask"
343 now support a new "--now" switch. If specified the units
344 that are enabled will also be started, and the ones
345 disabled/masked also stopped.
347 * The Gummiboot EFI boot loader tool has been merged into
348 systemd, and renamed to "systemd-boot". The bootctl tool has been
349 updated to support systemd-boot.
351 * An EFI kernel stub has been added that may be used to create
352 kernel EFI binaries that contain not only the actual kernel,
353 but also an initrd, boot splash, command line and OS release
354 information. This combined binary can then be signed as a
355 single image, so that the firmware can verify it all in one
356 step. systemd-boot has special support for EFI binaries created
357 like this and can extract OS release information from them
358 and show them in the boot menu. This functionality is useful
359 to implement cryptographically verified boot schemes.
361 * Optional support has been added to systemd-fsck to pass
362 fsck's progress report to an AF_UNIX socket in the file
365 * udev will no longer create device symlinks for all block
366 devices by default. A blacklist for excluding special block
367 devices from this logic has been turned into a whitelist
368 that requires picking block devices explicitly that require
371 * A new (currently still internal) API sd-device.h has been
372 added to libsystemd. This modernized API is supposed to
373 replace libudev eventually. In fact, already much of libudev
374 is now just a wrapper around sd-device.h.
376 * A new hwdb database for storing metadata about pointing
377 stick devices has been added.
379 * systemd-tmpfiles gained support for setting file attributes
380 similar to the "chattr" tool with new 'h' and 'H' lines.
382 * systemd-journald will no longer unconditionally set the
383 btrfs NOCOW flag on new journal files. This is instead done
384 with tmpfiles snippet using the new 'h' line type. This
385 allows easy disabling of this logic, by masking the
386 journal-nocow.conf tmpfiles file.
388 * systemd-journald will now translate audit message types to
389 human readable identifiers when writing them to the
390 journal. This should improve readability of audit messages.
392 * The LUKS logic gained support for the offset= and skip=
393 options in /etc/crypttab, as previously implemented by
396 * /usr/lib/os-release gained a new optional field VARIANT= for
397 distributions that support multiple variants (such as a
398 desktop edition, a server edition, ...)
400 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
401 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
402 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
403 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
404 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
405 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
406 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
407 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
408 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
409 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
410 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
411 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
412 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
413 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
414 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
415 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
416 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
417 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
418 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
419 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
420 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
421 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
422 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
423 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
424 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
425 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
426 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
428 -- Berlin, 2015-05-22
432 * Introduce a new API "sd-hwdb.h" for querying the hardware
433 metadata database. With this minimal interface one can query
434 and enumerate the udev hwdb, decoupled from the old libudev
435 library. libudev's interface for this is now only a wrapper
436 around sd-hwdb. A new tool systemd-hwdb has been added to
437 interface with and update the database.
439 * When any of systemd's tools copies files (for example due to
440 tmpfiles' C lines) a btrfs reflink will attempted first,
441 before bytewise copying is done.
443 * systemd-nspawn gained a new --ephemeral switch. When
444 specified a btrfs snapshot is taken of the container's root
445 directory, and immediately removed when the container
446 terminates again. Thus, a container can be started whose
447 changes never alter the container's root directory, and are
448 lost on container termination. This switch can also be used
449 for starting a container off the root file system of the
450 host without affecting the host OS. This switch is only
451 available on btrfs file systems.
453 * systemd-nspawn gained a new --template= switch. It takes the
454 path to a container tree to use as template for the tree
455 specified via --directory=, should that directory be
456 missing. This allows instantiating containers dynamically,
457 on first run. This switch is only available on btrfs file
460 * When a .mount unit refers to a mount point on which multiple
461 mounts are stacked, and the .mount unit is stopped all of
462 the stacked mount points will now be unmounted until no
465 * systemd now has an explicit notion of supported and
466 unsupported unit types. Jobs enqueued for unsupported unit
467 types will now fail with an "unsupported" error code. More
468 specifically .swap, .automount and .device units are not
469 supported in containers, .busname units are not supported on
470 non-kdbus systems. .swap and .automount are also not
471 supported if their respective kernel compile time options
474 * machinectl gained support for two new "copy-from" and
475 "copy-to" commands for copying files from a running
476 container to the host or vice versa.
478 * machinectl gained support for a new "bind" command to bind
479 mount host directories into local containers. This is
480 currently only supported for nspawn containers.
482 * networkd gained support for configuring bridge forwarding
483 database entries (fdb) from .network files.
485 * A new tiny daemon "systemd-importd" has been added that can
486 download container images in tar, raw, qcow2 or dkr formats,
487 and make them available locally in /var/lib/machines, so
488 that they can run as nspawn containers. The daemon can GPG
489 verify the downloads (not supported for dkr, since it has no
490 provisions for verifying downloads). It will transparently
491 decompress bz2, xz, gzip compressed downloads if necessary,
492 and restore sparse files on disk. The daemon uses privilege
493 separation to ensure the actual download logic runs with
494 fewer privileges than the daemon itself. machinectl has
495 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
496 make the functionality of importd available to the
497 user. With this in place the Fedora and Ubuntu "Cloud"
498 images can be downloaded and booted as containers unmodified
499 (the Fedora images lack the appropriate GPG signature files
500 currently, so they cannot be verified, but this will change
501 soon, hopefully). Note that downloading images is currently
502 only fully supported on btrfs.
504 * machinectl is now able to list container images found in
505 /var/lib/machines, along with some metadata about sizes of
506 disk and similar. If the directory is located on btrfs and
507 quota is enabled, this includes quota display. A new command
508 "image-status" has been added that shows additional
509 information about images.
511 * machinectl is now able to clone container images
512 efficiently, if the underlying file system (btrfs) supports
513 it, with the new "machinectl list-images" command. It also
514 gained commands for renaming and removing images, as well as
515 marking them read-only or read-write (supported also on
516 legacy file systems).
518 * networkd gained support for collecting LLDP network
519 announcements, from hardware that supports this. This is
520 shown in networkctl output.
522 * systemd-run gained support for a new -t (--pty) switch for
523 invoking a binary on a pty whose input and output is
524 connected to the invoking terminal. This allows executing
525 processes as system services while interactively
526 communicating with them via the terminal. Most interestingly
527 this is supported across container boundaries. Invoking
528 "systemd-run -t /bin/bash" is an alternative to running a
529 full login session, the difference being that the former
530 will not register a session, nor go through the PAM session
533 * tmpfiles gained support for a new "v" line type for creating
534 btrfs subvolumes. If the underlying file system is a legacy
535 file system, this automatically degrades to creating a
536 normal directory. Among others /var/lib/machines is now
537 created like this at boot, should it be missing.
539 * The directory /var/lib/containers/ has been deprecated and
540 been replaced by /var/lib/machines. The term "machines" has
541 been used in the systemd context as generic term for both
542 VMs and containers, and hence appears more appropriate for
543 this, as the directory can also contain raw images bootable
546 * systemd-nspawn when invoked with -M but without --directory=
547 or --image= is now capable of searching for the container
548 root directory, subvolume or disk image automatically, in
549 /var/lib/machines. systemd-nspawn@.service has been updated
550 to make use of this, thus allowing it to be used for raw
553 * A new machines.target unit has been introduced that is
554 supposed to group all containers/VMs invoked as services on
555 the system. systemd-nspawn@.service has been updated to
558 * machinectl gained a new "start" command, for invoking a
559 container as a service. "machinectl start foo" is mostly
560 equivalent to "systemctl start systemd-nspawn@foo.service",
561 but handles escaping in a nicer way.
563 * systemd-nspawn will now mount most of the cgroupfs tree
564 read-only into each container, with the exception of the
565 container's own subtree in the name=systemd hierarchy.
567 * journald now sets the special FS_NOCOW file flag for its
568 journal files. This should improve performance on btrfs, by
569 avoiding heavy fragmentation when journald's write-pattern
570 is used on COW file systems. It degrades btrfs' data
571 integrity guarantees for the files to the same levels as for
572 ext3/ext4 however. This should be OK though as journald does
573 its own data integrity checks and all its objects are
574 checksummed on disk. Also, journald should handle btrfs disk
575 full events a lot more gracefully now, by processing SIGBUS
576 errors, and not relying on fallocate() anymore.
578 * When journald detects that journal files it is writing to
579 have been deleted it will immediately start new journal
582 * systemd now provides a way to store file descriptors
583 per-service in PID 1.This is useful for daemons to ensure
584 that fds they require are not lost during a daemon
585 restart. The fds are passed to the daemon on the next
586 invocation in the same way socket activation fds are
587 passed. This is now used by journald to ensure that the
588 various sockets connected to all the system's stdout/stderr
589 are not lost when journald is restarted. File descriptors
590 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
591 an extension to sd_notify(). Note that a limit is enforced
592 on the number of fds a service can store in PID 1, and it
593 defaults to 0, so that no fds may be stored, unless this is
594 explicitly turned on.
596 * The default TERM variable to use for units connected to a
597 terminal, when no other value is explicitly is set is now
598 vt220 rather than vt102. This should be fairly safe still,
599 but allows PgUp/PgDn work.
601 * The /etc/crypttab option header= as known from Debian is now
604 * "loginctl user-status" and "loginctl session-status" will
605 now show the last 10 lines of log messages of the
606 user/session following the status output. Similar,
607 "machinectl status" will show the last 10 log lines
608 associated with a virtual machine or container
609 service. (Note that this is usually not the log messages
610 done in the VM/container itself, but simply what the
611 container manager logs. For nspawn this includes all console
614 * "loginctl session-status" without further argument will now
615 show the status of the session of the caller. Similar,
616 "lock-session", "unlock-session", "activate",
617 "enable-linger", "disable-linger" may now be called without
618 session/user parameter in which case they apply to the
619 caller's session/user.
621 * An X11 session scriptlet is now shipped that uploads
622 $DISPLAY and $XAUTHORITY into the environment of the systemd
623 --user daemon if a session begins. This should improve
624 compatibility with X11 enabled applications run as systemd
627 * Generators are now subject to masking via /etc and /run, the
628 same way as unit files.
630 * networkd .network files gained support for configuring
631 per-link IPv4/IPv6 packet forwarding as well as IPv4
632 masquerading. This is by default turned on for veth links to
633 containers, as registered by systemd-nspawn. This means that
634 nspawn containers run with --network-veth will now get
635 automatic routed access to the host's networks without any
636 further configuration or setup, as long as networkd runs on
639 * systemd-nspawn gained the --port= (-p) switch to expose TCP
640 or UDP posts of a container on the host. With this in place
641 it is possible to run containers with private veth links
642 (--network-veth), and have their functionality exposed on
643 the host as if their services were running directly on the
646 * systemd-nspawn's --network-veth switch now gained a short
647 version "-n", since with the changes above it is now truly
648 useful out-of-the-box. The systemd-nspawn@.service has been
649 updated to make use of it too by default.
651 * systemd-nspawn will now maintain a per-image R/W lock, to
652 ensure that the same image is not started more than once
653 writable. (It's OK to run an image multiple times
654 simultaneously in read-only mode.)
656 * systemd-nspawn's --image= option is now capable of
657 dissecting and booting MBR and GPT disk images that contain
658 only a single active Linux partition. Previously it
659 supported only GPT disk images with proper GPT type
660 IDs. This allows running cloud images from major
661 distributions directly with systemd-nspawn, without
664 * In addition to collecting mouse dpi data in the udev
665 hardware database, there's now support for collecting angle
666 information for mouse scroll wheels. The database is
667 supposed to guarantee similar scrolling behavior on mice
668 that it knows about. There's also support for collecting
669 information about Touchpad types.
671 * udev's input_id built-in will now also collect touch screen
672 dimension data and attach it to probed devices.
674 * /etc/os-release gained support for a Distribution Privacy
677 * networkd gained support for creating "ipvlan", "gretap",
678 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
680 * systemd-tmpfiles gained support for "a" lines for setting
683 * systemd-nspawn will now mount /tmp in the container to
684 tmpfs, automatically.
686 * systemd now exposes the memory.usage_in_bytes cgroup
687 attribute and shows it for each service in the "systemctl
688 status" output, if available.
690 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
691 immediate reboot is triggered. This useful if shutdown is
692 hung and is unable to complete, to expedite the
693 operation. Note that this kind of reboot will still unmount
694 all file systems, and hence should not result in fsck being
697 * A .device unit for an optical block device will now be
698 considered active only when a medium is in the drive. Also,
699 mount units are now bound to their backing devices thus
700 triggering automatic unmounting when devices become
701 unavailable. With this in place systemd will now
702 automatically unmount left-over mounts when a CD-ROM is
703 ejected or an USB stick is yanked from the system.
705 * networkd-wait-online now has support for waiting for
706 specific interfaces only (with globbing), and for giving up
707 after a configurable timeout.
709 * networkd now exits when idle. It will be automatically
710 restarted as soon as interfaces show up, are removed or
711 change state. networkd will stay around as long as there is
712 at least one DHCP state machine or similar around, that keep
715 * networkd may now configure IPv6 link-local addressing in
716 addition to IPv4 link-local addressing.
718 * The IPv6 "token" for use in SLAAC may now be configured for
719 each .network interface in networkd.
721 * Routes configured with networkd may now be assigned a scope
724 * networkd's [Match] sections now support globbing and lists
725 of multiple space-separated matches per item.
727 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
728 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
729 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
730 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
731 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
732 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
733 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
734 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
735 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
736 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
737 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
738 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
739 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
740 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
741 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
742 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
743 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
744 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
745 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
746 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
747 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
748 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
749 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
750 Hoffmann, Zbigniew Jędrzejewski-Szmek
752 -- Berlin, 2015-02-16
756 * When querying unit file enablement status (for example via
757 "systemctl is-enabled"), a new state "indirect" is now known
758 which indicates that a unit might not be enabled itself, but
759 another unit listed in its Alias= setting might be.
761 * Similar to the various existing ConditionXYZ= settings for
762 units there are now matching AssertXYZ= settings. While
763 failing conditions cause a unit to be skipped, but its job
764 to succeed, failing assertions declared like this will cause
765 a unit start operation and its job to fail.
767 * hostnamed now knows a new chassis type "embedded".
769 * systemctl gained a new "edit" command. When used on a unit
770 file this allows extending unit files with .d/ drop-in
771 configuration snippets or editing the full file (after
772 copying it from /usr/lib to /etc). This will invoke the
773 user's editor (as configured with $EDITOR), and reload the
774 modified configuration after editing.
776 * "systemctl status" now shows the suggested enablement state
777 for a unit, as declared in the (usually vendor-supplied)
780 * nss-myhostname will now resolve the single-label host name
781 "gateway" to the locally configured default IP routing
782 gateways, ordered by their metrics. This assigns a stable
783 name to the used gateways, regardless which ones are
784 currently configured. Note that the name will only be
785 resolved after all other name sources (if nss-myhostname is
786 configured properly) and should hence not negatively impact
787 systems that use the single-label host name "gateway" in
790 * systemd-inhibit now allows filtering by mode when listing
793 * Scope and service units gained a new "Delegate" boolean
794 property, which when set allows processes running inside the
795 unit to further partition resources. This is primarily
796 useful for systemd user instances as well as container
799 * journald will now pick up audit messages directly from
800 the kernel, and log them like any other log message. The
801 audit fields are split up and fully indexed. This means that
802 journalctl in many ways is now a (nicer!) alternative to
803 ausearch, the traditional audit client. Note that this
804 implements only a minimal audit client, if you want the
805 special audit modes like reboot-on-log-overflow, please use
806 the traditional auditd instead, which can be used in
807 parallel to journald.
809 * The ConditionSecurity= unit file option now understands the
810 special string "audit" to check whether auditing is
813 * journalctl gained two new commands --vacuum-size= and
814 --vacuum-time= to delete old journal files until the
815 remaining ones take up no more the specified size on disk,
816 or are not older than the specified time.
818 * A new, native PPPoE library has been added to sd-network,
819 systemd's library of light-weight networking protocols. This
820 library will be used in a future version of networkd to
821 enable PPPoE communication without an external pppd daemon.
823 * The busctl tool now understands a new "capture" verb that
824 works similar to "monitor", but writes a packet capture
825 trace to STDOUT that can be redirected to a file which is
826 compatible with libcap's capture file format. This can then
827 be loaded in Wireshark and similar tools to inspect bus
830 * The busctl tool now understands a new "tree" verb that shows
831 the object trees of a specific service on the bus, or of all
834 * The busctl tool now understands a new "introspect" verb that
835 shows all interfaces and members of objects on the bus,
836 including their signature and values. This is particularly
837 useful to get more information about bus objects shown by
838 the new "busctl tree" command.
840 * The busctl tool now understands new verbs "call",
841 "set-property" and "get-property" for invoking bus method
842 calls, setting and getting bus object properties in a
845 * busctl gained a new --augment-creds= argument that controls
846 whether the tool shall augment credential information it
847 gets from the bus with data from /proc, in a possibly
850 * nspawn's --link-journal= switch gained two new values
851 "try-guest" and "try-host" that work like "guest" and
852 "host", but do not fail if the host has no persistent
853 journalling enabled. -j is now equivalent to
854 --link-journal=try-guest.
856 * macvlan network devices created by nspawn will now have
857 stable MAC addresses.
859 * A new SmackProcessLabel= unit setting has been added, which
860 controls the SMACK security label processes forked off by
861 the respective unit shall use.
863 * If compiled with --enable-xkbcommon, systemd-localed will
864 verify x11 keymap settings by compiling the given keymap. It
865 will spew out warnings if the compilation fails. This
866 requires libxkbcommon to be installed.
868 * When a coredump is collected a larger number of metadata
869 fields is now collected and included in the journal records
870 created for it. More specifically control group membership,
871 environment variables, memory maps, working directory,
872 chroot directory, /proc/$PID/status, and a list of open file
873 descriptors is now stored in the log entry.
875 * The udev hwdb now contains DPI information for mice. For
878 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
880 * All systemd programs that read standalone configuration
881 files in /etc now also support a corresponding series of
882 .conf.d configuration directories in /etc/, /run/,
883 /usr/local/lib/, /usr/lib/, and (if configured with
884 --enable-split-usr) /lib/. In particular, the following
885 configuration files now have corresponding configuration
886 directories: system.conf user.conf, logind.conf,
887 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
888 resolved.conf, timesyncd.conf, journal-remote.conf, and
889 journal-upload.conf. Note that distributions should use the
890 configuration directories in /usr/lib/; the directories in
891 /etc/ are reserved for the system administrator.
893 * systemd-rfkill will no longer take the rfkill device name
894 into account when storing rfkill state on disk, as the name
895 might be dynamically assigned and not stable. Instead, the
896 ID_PATH udev variable combined with the rfkill type (wlan,
897 bluetooth, ...) is used.
899 * A new service systemd-machine-id-commit.service has been
900 added. When used on systems where /etc is read-only during
901 boot, and /etc/machine-id is not initialized (but an empty
902 file), this service will copy the temporary machine ID
903 created as replacement into /etc after the system is fully
904 booted up. This is useful for systems that are freshly
905 installed with a non-initialized machine ID, but should get
906 a fixed machine ID for subsequent boots.
908 * networkd's .netdev files now provide a large set of
909 configuration parameters for VXLAN devices. Similar, the
910 bridge port cost parameter is now configurable in .network
911 files. There's also new support for configuring IP source
912 routing. networkd .link files gained support for a new
913 OriginalName= match that is useful to match against the
914 original interface name the kernel assigned. .network files
915 may include MTU= and MACAddress= fields for altering the MTU
916 and MAC address while being connected to a specific network
919 * The LUKS logic gained supported for configuring
920 UUID-specific key files. There's also new support for naming
921 LUKS device from the kernel command line, using the new
924 * Timer units may now be transiently created via the bus API
925 (this was previously already available for scope and service
926 units). In addition it is now possible to create multiple
927 transient units at the same time with a single bus call. The
928 "systemd-run" tool has been updated to make use of this for
929 running commands on a specified time, in at(1)-style.
931 * tmpfiles gained support for "t" lines, for assigning
932 extended attributes to files. Among other uses this may be
933 used to assign SMACK labels to files.
935 Contributions from: Alin Rauta, Alison Chaiken, Andrej
936 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
937 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
938 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
939 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
940 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
941 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
942 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
943 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
944 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
945 Michael Biebl, Michael Chapman, Michael Marineau, Michal
946 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
947 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
948 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
949 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
950 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
951 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
952 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
954 -- Berlin, 2014-12-10
958 * journalctl gained the new options -t/--identifier= to match
959 on the syslog identifier (aka "tag"), as well as --utc to
960 show log timestamps in the UTC timezone. journalctl now also
961 accepts -n/--lines=all to disable line capping in a pager.
963 * journalctl gained a new switch, --flush, that synchronously
964 flushes logs from /run/log/journal to /var/log/journal if
965 persistent storage is enabled. systemd-journal-flush.service
966 now waits until the operation is complete.
968 * Services can notify the manager before they start a reload
969 (by sending RELOADING=1) or shutdown (by sending
970 STOPPING=1). This allows the manager to track and show the
971 internal state of daemons and closes a race condition when
972 the process is still running but has closed its D-Bus
975 * Services with Type=oneshot do not have to have any ExecStart
978 * User units are now loaded also from
979 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
980 /run/systemd/user directory that was already previously
981 supported, but is under the control of the user.
983 * Job timeouts (i.e. time-outs on the time a job that is
984 queued stays in the run queue) can now optionally result in
985 immediate reboot or power-off actions (JobTimeoutAction= and
986 JobTimeoutRebootArgument=). This is useful on ".target"
987 units, to limit the maximum time a target remains
988 undispatched in the run queue, and to trigger an emergency
989 operation in such a case. This is now used by default to
990 turn off the system if boot-up (as defined by everything in
991 basic.target) hangs and does not complete for at least
992 15min. Also, if power-off or reboot hang for at least 30min
993 an immediate power-off/reboot operation is triggered. This
994 functionality is particularly useful to increase reliability
995 on embedded devices, but also on laptops which might
996 accidentally get powered on when carried in a backpack and
997 whose boot stays stuck in a hard disk encryption passphrase
1000 * systemd-logind can be configured to also handle lid switch
1001 events even when the machine is docked or multiple displays
1002 are attached (HandleLidSwitchDocked= option).
1004 * A helper binary and a service have been added which can be
1005 used to resume from hibernation in the initramfs. A
1006 generator will parse the resume= option on the kernel
1007 command line to trigger resume.
1009 * A user console daemon systemd-consoled has been
1010 added. Currently, it is a preview, and will so far open a
1011 single terminal on each session of the user marked as
1012 Desktop=systemd-console.
1014 * Route metrics can be specified for DHCP routes added by
1017 * The SELinux context of socket-activated services can be set
1018 from the information provided by the networking stack
1019 (SELinuxContextFromNet= option).
1021 * Userspace firmware loading support has been removed and
1022 the minimum supported kernel version is thus bumped to 3.7.
1024 * Timeout for udev workers has been increased from 1 to 3
1025 minutes, but a warning will be printed after 1 minute to
1026 help diagnose kernel modules that take a long time to load.
1028 * Udev rules can now remove tags on devices with TAG-="foobar".
1030 * systemd's readahead implementation has been removed. In many
1031 circumstances it didn't give expected benefits even for
1032 rotational disk drives and was becoming less relevant in the
1033 age of SSDs. As none of the developers has been using
1034 rotating media anymore, and nobody stepped up to actively
1035 maintain this component of systemd it has now been removed.
1037 * Swap units can use Options= to specify discard options.
1038 Discard options specified for swaps in /etc/fstab are now
1041 * Docker containers are now detected as a separate type of
1044 * The Password Agent protocol gained support for queries where
1045 the user input is shown, useful e.g. for user names.
1046 systemd-ask-password gained a new --echo option to turn that
1049 * The default sysctl.d/ snippets will now set:
1051 net.core.default_qdisc = fq_codel
1053 This selects Fair Queuing Controlled Delay as the default
1054 queuing discipline for network interfaces. fq_codel helps
1055 fight the network bufferbloat problem. It is believed to be
1056 a good default with no tuning required for most workloads.
1057 Downstream distributions may override this choice. On 10Gbit
1058 servers that do not do forwarding, "fq" may perform better.
1059 Systems without a good clocksource should use "pfifo_fast".
1061 * If kdbus is enabled during build a new option BusPolicy= is
1062 available for service units, that allows locking all service
1063 processes into a stricter bus policy, in order to limit
1064 access to various bus services, or even hide most of them
1065 from the service's view entirely.
1067 * networkctl will now show the .network and .link file
1068 networkd has applied to a specific interface.
1070 * sd-login gained a new API call sd_session_get_desktop() to
1071 query which desktop environment has been selected for a
1074 * UNIX utmp support is now compile-time optional to support
1075 legacy-free systems.
1077 * systemctl gained two new commands "add-wants" and
1078 "add-requires" for pulling in units from specific targets
1081 * If the word "rescue" is specified on the kernel command line
1082 the system will now boot into rescue mode (aka
1083 rescue.target), which was previously available only by
1084 specifying "1" or "systemd.unit=rescue.target" on the kernel
1085 command line. This new kernel command line option nicely
1086 mirrors the already existing "emergency" kernel command line
1089 * New kernel command line options mount.usr=, mount.usrflags=,
1090 mount.usrfstype= have been added that match root=, rootflags=,
1091 rootfstype= but allow mounting a specific file system to
1094 * The $NOTIFY_SOCKET is now also passed to control processes of
1095 services, not only the main process.
1097 * This version reenables support for fsck's -l switch. This
1098 means at least version v2.25 of util-linux is required for
1099 operation, otherwise dead-locks on device nodes may
1100 occur. Again: you need to update util-linux to at least
1101 v2.25 when updating systemd to v217.
1103 * The "multi-seat-x" tool has been removed from systemd, as
1104 its functionality has been integrated into X servers 1.16,
1105 and the tool is hence redundant. It is recommended to update
1106 display managers invoking this tool to simply invoke X
1107 directly from now on, again.
1109 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
1110 message flag has been added for all of systemd's PolicyKit
1111 authenticated method calls has been added. In particular
1112 this now allows optional interactive authorization via
1113 PolicyKit for many of PID1's privileged operations such as
1114 unit file enabling and disabling.
1116 * "udevadm hwdb --update" learnt a new switch "--usr" for
1117 placing the rebuilt hardware database in /usr instead of
1118 /etc. When used only hardware database entries stored in
1119 /usr will be used, and any user database entries in /etc are
1120 ignored. This functionality is useful for vendors to ship a
1121 pre-built database on systems where local configuration is
1122 unnecessary or unlikely.
1124 * Calendar time specifications in .timer units now also
1125 understand the strings "semi-annually", "quarterly" and
1126 "minutely" as shortcuts (in addition to the preexisting
1127 "anually", "hourly", ...).
1129 * systemd-tmpfiles will now correctly create files in /dev
1130 at boot which are marked for creation only at boot. It is
1131 recommended to always create static device nodes with 'c!'
1132 and 'b!', so that they are created only at boot and not
1133 overwritten at runtime.
1135 * When the watchdog logic is used for a service (WatchdogSec=)
1136 and the watchdog timeout is hit the service will now be
1137 terminated with SIGABRT (instead of just SIGTERM), in order
1138 to make sure a proper coredump and backtrace is
1139 generated. This ensures that hanging services will result in
1140 similar coredump/backtrace behaviour as services that hit a
1143 Contributions from: Andreas Henriksson, Andrei Borzenkov,
1144 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
1145 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
1146 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
1147 Herrmann, David Sommerseth, David Strauss, Emil Renner
1148 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
1149 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
1150 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
1151 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
1152 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
1153 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
1154 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
1155 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
1156 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
1157 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
1158 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
1159 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
1160 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
1161 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
1162 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
1163 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
1166 -- Berlin, 2014-10-28
1170 * timedated no longer reads NTP implementation unit names from
1171 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
1172 implementations should add a
1174 Conflicts=systemd-timesyncd.service
1176 to their unit files to take over and replace systemd's NTP
1177 default functionality.
1179 * systemd-sysusers gained a new line type "r" for configuring
1180 which UID/GID ranges to allocate system users/groups
1181 from. Lines of type "u" may now add an additional column
1182 that specifies the home directory for the system user to be
1183 created. Also, systemd-sysusers may now optionally read user
1184 information from STDIN instead of a file. This is useful for
1185 invoking it from RPM preinst scriptlets that need to create
1186 users before the first RPM file is installed since these
1187 files might need to be owned by them. A new
1188 %sysusers_create_inline RPM macro has been introduced to do
1189 just that. systemd-sysusers now updates the shadow files as
1190 well as the user/group databases, which should enhance
1191 compatibility with certain tools like grpck.
1193 * A number of bus APIs of PID 1 now optionally consult
1194 PolicyKit to permit access for otherwise unprivileged
1195 clients under certain conditions. Note that this currently
1196 doesn't support interactive authentication yet, but this is
1197 expected to be added eventually, too.
1199 * /etc/machine-info now has new fields for configuring the
1200 deployment environment of the machine, as well as the
1201 location of the machine. hostnamectl has been updated with
1202 new command to update these fields.
1204 * systemd-timesyncd has been updated to automatically acquire
1205 NTP server information from systemd-networkd, which might
1206 have been discovered via DHCP.
1208 * systemd-resolved now includes a caching DNS stub resolver
1209 and a complete LLMNR name resolution implementation. A new
1210 NSS module "nss-resolve" has been added which can be used
1211 instead of glibc's own "nss-dns" to resolve hostnames via
1212 systemd-resolved. Hostnames, addresses and arbitrary RRs may
1213 be resolved via systemd-resolved D-Bus APIs. In contrast to
1214 the glibc internal resolver systemd-resolved is aware of
1215 multi-homed system, and keeps DNS server and caches separate
1216 and per-interface. Queries are sent simultaneously on all
1217 interfaces that have DNS servers configured, in order to
1218 properly handle VPNs and local LANs which might resolve
1219 separate sets of domain names. systemd-resolved may acquire
1220 DNS server information from systemd-networkd automatically,
1221 which in turn might have discovered them via DHCP. A tool
1222 "systemd-resolve-host" has been added that may be used to
1223 query the DNS logic in resolved. systemd-resolved implements
1224 IDNA and automatically uses IDNA or UTF-8 encoding depending
1225 on whether classic DNS or LLMNR is used as transport. In the
1226 next releases we intend to add a DNSSEC and mDNS/DNS-SD
1227 implementation to systemd-resolved.
1229 * A new NSS module nss-mymachines has been added, that
1230 automatically resolves the names of all local registered
1231 containers to their respective IP addresses.
1233 * A new client tool "networkctl" for systemd-networkd has been
1234 added. It currently is entirely passive and will query
1235 networking configuration from udev, rtnetlink and networkd,
1236 and present it to the user in a very friendly
1237 way. Eventually, we hope to extend it to become a full
1238 control utility for networkd.
1240 * .socket units gained a new DeferAcceptSec= setting that
1241 controls the kernels' TCP_DEFER_ACCEPT sockopt for
1242 TCP. Similar, support for controlling TCP keep-alive
1243 settings has been added (KeepAliveTimeSec=,
1244 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
1245 turning off Nagle's algorithm on TCP has been added
1248 * logind learned a new session type "web", for use in projects
1249 like Cockpit which register web clients as PAM sessions.
1251 * timer units with at least one OnCalendar= setting will now
1252 be started only after timer-sync.target has been
1253 reached. This way they will not elapse before the system
1254 clock has been corrected by a local NTP client or
1255 similar. This is particular useful on RTC-less embedded
1256 machines, that come up with an invalid system clock.
1258 * systemd-nspawn's --network-veth= switch should now result in
1259 stable MAC addresses for both the outer and the inner side
1262 * systemd-nspawn gained a new --volatile= switch for running
1263 container instances with /etc or /var unpopulated.
1265 * The kdbus client code has been updated to use the new Linux
1266 3.17 memfd subsystem instead of the old kdbus-specific one.
1268 * systemd-networkd's DHCP client and server now support
1269 FORCERENEW. There are also new configuration options to
1270 configure the vendor client identifier and broadcast mode
1273 * systemd will no longer inform the kernel about the current
1274 timezone, as this is necessarily incorrect and racy as the
1275 kernel has no understanding of DST and similar
1276 concepts. This hence means FAT timestamps will be always
1277 considered UTC, similar to what Android is already
1278 doing. Also, when the RTC is configured to the local time
1279 (rather than UTC) systemd will never synchronize back to it,
1280 as this might confuse Windows at a later boot.
1282 * systemd-analyze gained a new command "verify" for offline
1283 validation of unit files.
1285 * systemd-networkd gained support for a couple of additional
1286 settings for bonding networking setups. Also, the metric for
1287 statically configured routes may now be configured. For
1288 network interfaces where this is appropriate the peer IP
1289 address may now be configured.
1291 * systemd-networkd's DHCP client will no longer request
1292 broadcasting by default, as this tripped up some networks.
1293 For hardware where broadcast is required the feature should
1294 be switched back on using RequestBroadcast=yes.
1296 * systemd-networkd will now set up IPv4LL addresses (when
1297 enabled) even if DHCP is configured successfully.
1299 * udev will now default to respect network device names given
1300 by the kernel when the kernel indicates that these are
1301 predictable. This behavior can be tweaked by changing
1302 NamePolicy= in the relevant .link file.
1304 * A new library systemd-terminal has been added that
1305 implements full TTY stream parsing and rendering. This
1306 library is supposed to be used later on for implementing a
1307 full userspace VT subsystem, replacing the current kernel
1310 * A new tool systemd-journal-upload has been added to push
1311 journal data to a remote system running
1312 systemd-journal-remote.
1314 * journald will no longer forward all local data to another
1315 running syslog daemon. This change has been made because
1316 rsyslog (which appears to be the most commonly used syslog
1317 implementation these days) no longer makes use of this, and
1318 instead pulls the data out of the journal on its own. Since
1319 forwarding the messages to a non-existent syslog server is
1320 more expensive than we assumed we have now turned this
1321 off. If you run a syslog server that is not a recent rsyslog
1322 version, you have to turn this option on again
1323 (ForwardToSyslog= in journald.conf).
1325 * journald now optionally supports the LZ4 compressor for
1326 larger journal fields. This compressor should perform much
1327 better than XZ which was the previous default.
1329 * machinectl now shows the IP addresses of local containers,
1330 if it knows them, plus the interface name of the container.
1332 * A new tool "systemd-escape" has been added that makes it
1333 easy to escape strings to build unit names and similar.
1335 * sd_notify() messages may now include a new ERRNO= field
1336 which is parsed and collected by systemd and shown among the
1337 "systemctl status" output for a service.
1339 * A new component "systemd-firstboot" has been added that
1340 queries the most basic systemd information (timezone,
1341 hostname, root password) interactively on first
1342 boot. Alternatively it may also be used to provision these
1343 things offline on OS images installed into directories.
1345 * The default sysctl.d/ snippets will now set
1347 net.ipv4.conf.default.promote_secondaries=1
1349 This has the benefit of no flushing secondary IP addresses
1350 when primary addresses are removed.
1352 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
1353 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
1354 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
1355 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
1356 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
1357 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
1358 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
1359 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
1360 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
1361 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
1362 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
1363 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
1364 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
1365 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
1366 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
1368 -- Berlin, 2014-08-19
1372 * A new tool systemd-sysusers has been added. This tool
1373 creates system users and groups in /etc/passwd and
1374 /etc/group, based on static declarative system user/group
1375 definitions in /usr/lib/sysusers.d/. This is useful to
1376 enable factory resets and volatile systems that boot up with
1377 an empty /etc directory, and thus need system users and
1378 groups created during early boot. systemd now also ships
1379 with two default sysusers.d/ files for the most basic
1380 users and groups systemd and the core operating system
1383 * A new tmpfiles snippet has been added that rebuilds the
1384 essential files in /etc on boot, should they be missing.
1386 * A directive for ensuring automatic clean-up of
1387 /var/cache/man/ has been removed from the default
1388 configuration. This line should now be shipped by the man
1389 implementation. The necessary change has been made to the
1390 man-db implementation. Note that you need to update your man
1391 implementation to one that ships this line, otherwise no
1392 automatic clean-up of /var/cache/man will take place.
1394 * A new condition ConditionNeedsUpdate= has been added that
1395 may conditionalize services to only run when /etc or /var
1396 are "older" than the vendor operating system resources in
1397 /usr. This is useful for reconstructing or updating /etc
1398 after an offline update of /usr or a factory reset, on the
1399 next reboot. Services that want to run once after such an
1400 update or reset should use this condition and order
1401 themselves before the new systemd-update-done.service, which
1402 will mark the two directories as fully updated. A number of
1403 service files have been added making use of this, to rebuild
1404 the udev hardware database, the journald message catalog and
1405 dynamic loader cache (ldconfig). The systemd-sysusers tool
1406 described above also makes use of this now. With this in
1407 place it is now possible to start up a minimal operating
1408 system with /etc empty cleanly. For more information on the
1409 concepts involved see this recent blog story:
1411 http://0pointer.de/blog/projects/stateless.html
1413 * A new system group "input" has been introduced, and all
1414 input device nodes get this group assigned. This is useful
1415 for system-level software to get access to input devices. It
1416 complements what is already done for "audio" and "video".
1418 * systemd-networkd learnt minimal DHCPv4 server support in
1419 addition to the existing DHCPv4 client support. It also
1420 learnt DHCPv6 client and IPv6 Router Solicitation client
1421 support. The DHCPv4 client gained support for static routes
1422 passed in from the server. Note that the [DHCPv4] section
1423 known in older systemd-networkd versions has been renamed to
1424 [DHCP] and is now also used by the DHCPv6 client. Existing
1425 .network files using settings of this section should be
1426 updated, though compatibility is maintained. Optionally, the
1427 client hostname may now be sent to the DHCP server.
1429 * networkd gained support for vxlan virtual networks as well
1430 as tun/tap and dummy devices.
1432 * networkd gained support for automatic allocation of address
1433 ranges for interfaces from a system-wide pool of
1434 addresses. This is useful for dynamically managing a large
1435 number of interfaces with a single network configuration
1436 file. In particular this is useful to easily assign
1437 appropriate IP addresses to the veth links of a large number
1438 of nspawn instances.
1440 * RPM macros for processing sysusers, sysctl and binfmt
1441 drop-in snippets at package installation time have been
1444 * The /etc/os-release file should now be placed in
1445 /usr/lib/os-release. The old location is automatically
1446 created as symlink. /usr/lib is the more appropriate
1447 location of this file, since it shall actually describe the
1448 vendor operating system shipped in /usr, and not the
1449 configuration stored in /etc.
1451 * .mount units gained a new boolean SloppyOptions= setting
1452 that maps to mount(8)'s -s option which enables permissive
1453 parsing of unknown mount options.
1455 * tmpfiles learnt a new "L+" directive which creates a symlink
1456 but (unlike "L") deletes a pre-existing file first, should
1457 it already exist and not already be the correct
1458 symlink. Similar, "b+", "c+" and "p+" directives have been
1459 added as well, which create block and character devices, as
1460 well as fifos in the filesystem, possibly removing any
1461 pre-existing files of different types.
1463 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
1464 'argument' field (which so far specified the source to
1465 symlink/copy the files from) is now optional. If omitted the
1466 same file os copied from /usr/share/factory/ suffixed by the
1467 full destination path. This is useful for populating /etc
1468 with essential files, by copying them from vendor defaults
1469 shipped in /usr/share/factory/etc.
1471 * A new command "systemctl preset-all" has been added that
1472 applies the service preset settings to all installed unit
1473 files. A new switch --preset-mode= has been added that
1474 controls whether only enable or only disable operations
1477 * A new command "systemctl is-system-running" has been added
1478 that allows checking the overall state of the system, for
1479 example whether it is fully up and running.
1481 * When the system boots up with an empty /etc, the equivalent
1482 to "systemctl preset-all" is executed during early boot, to
1483 make sure all default services are enabled after a factory
1486 * systemd now contains a minimal preset file that enables the
1487 most basic services systemd ships by default.
1489 * Unit files' [Install] section gained a new DefaultInstance=
1490 field for defining the default instance to create if a
1491 template unit is enabled with no instance specified.
1493 * A new passive target cryptsetup-pre.target has been added
1494 that may be used by services that need to make they run and
1495 finish before the first LUKS cryptographic device is set up.
1497 * The /dev/loop-control and /dev/btrfs-control device nodes
1498 are now owned by the "disk" group by default, opening up
1499 access to this group.
1501 * systemd-coredump will now automatically generate a
1502 stack trace of all core dumps taking place on the system,
1503 based on elfutils' libdw library. This stack trace is logged
1506 * systemd-coredump may now optionally store coredumps directly
1507 on disk (in /var/lib/systemd/coredump, possibly compressed),
1508 instead of storing them unconditionally in the journal. This
1509 mode is the new default. A new configuration file
1510 /etc/systemd/coredump.conf has been added to configure this
1511 and other parameters of systemd-coredump.
1513 * coredumpctl gained a new "info" verb to show details about a
1514 specific coredump. A new switch "-1" has also been added
1515 that makes sure to only show information about the most
1516 recent entry instead of all entries. Also, as the tool is
1517 generally useful now the "systemd-" prefix of the binary
1518 name has been removed. Distributions that want to maintain
1519 compatibility with the old name should add a symlink from
1520 the old name to the new name.
1522 * journald's SplitMode= now defaults to "uid". This makes sure
1523 that unprivileged users can access their own coredumps with
1524 coredumpctl without restrictions.
1526 * New kernel command line options "systemd.wants=" (for
1527 pulling an additional unit during boot), "systemd.mask="
1528 (for masking a specific unit for the boot), and
1529 "systemd.debug-shell" (for enabling the debug shell on tty9)
1530 have been added. This is implemented in the new generator
1531 "systemd-debug-generator".
1533 * systemd-nspawn will now by default filter a couple of
1534 syscalls for containers, among them those required for
1535 kernel module loading, direct x86 IO port access, swap
1536 management, and kexec. Most importantly though
1537 open_by_handle_at() is now prohibited for containers,
1538 closing a hole similar to a recently discussed vulnerability
1539 in docker regarding access to files on file hierarchies the
1540 container should normally not have access to. Note that for
1541 nspawn we generally make no security claims anyway (and
1542 this is explicitly documented in the man page), so this is
1543 just a fix for one of the most obvious problems.
1545 * A new man page file-hierarchy(7) has been added that
1546 contains a minimized, modernized version of the file system
1547 layout systemd expects, similar in style to the FHS
1548 specification or hier(5). A new tool systemd-path(1) has
1549 been added to query many of these paths for the local
1552 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
1553 longer done. Since the directory now has a per-user size
1554 limit, and is cleaned on logout this appears unnecessary,
1555 in particular since this now brings the lifecycle of this
1556 directory closer in line with how IPC objects are handled.
1558 * systemd.pc now exports a number of additional directories,
1559 including $libdir (which is useful to identify the library
1560 path for the primary architecture of the system), and a
1561 couple of drop-in directories.
1563 * udev's predictable network interface names now use the dev_port
1564 sysfs attribute, introduced in linux 3.15 instead of dev_id to
1565 distinguish between ports of the same PCI function. dev_id should
1566 only be used for ports using the same HW address, hence the need
1569 * machined has been updated to export the OS version of a
1570 container (read from /etc/os-release and
1571 /usr/lib/os-release) on the bus. This is now shown in
1572 "machinectl status" for a machine.
1574 * A new service setting RestartForceExitStatus= has been
1575 added. If configured to a set of exit signals or process
1576 return values, the service will be restarted when the main
1577 daemon process exits with any of them, regardless of the
1580 * systemctl's -H switch for connecting to remote systemd
1581 machines has been extended so that it may be used to
1582 directly connect to a specific container on the
1583 host. "systemctl -H root@foobar:waldi" will now connect as
1584 user "root" to host "foobar", and then proceed directly to
1585 the container named "waldi". Note that currently you have to
1586 authenticate as user "root" for this to work, as entering
1587 containers is a privileged operation.
1589 Contributions from: Andreas Henriksson, Benjamin Steinwender,
1590 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
1591 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
1592 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
1593 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
1594 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
1595 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
1596 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
1597 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
1598 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
1599 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
1600 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
1602 -- Berlin, 2014-07-03
1606 * As an experimental feature, udev now tries to lock the
1607 disk device node (flock(LOCK_SH|LOCK_NB)) while it
1608 executes events for the disk or any of its partitions.
1609 Applications like partitioning programs can lock the
1610 disk device node (flock(LOCK_EX)) and claim temporary
1611 device ownership that way; udev will entirely skip all event
1612 handling for this disk and its partitions. If the disk
1613 was opened for writing, the close will trigger a partition
1614 table rescan in udev's "watch" facility, and if needed
1615 synthesize "change" events for the disk and all its partitions.
1616 This is now unconditionally enabled, and if it turns out to
1617 cause major problems, we might turn it on only for specific
1618 devices, or might need to disable it entirely. Device Mapper
1619 devices are excluded from this logic.
1621 * We temporarily dropped the "-l" switch for fsck invocations,
1622 since they collide with the flock() logic above. util-linux
1623 upstream has been changed already to avoid this conflict,
1624 and we will readd "-l" as soon as util-linux with this
1625 change has been released.
1627 * The dependency on libattr has been removed. Since a long
1628 time, the extended attribute calls have moved to glibc, and
1629 libattr is thus unnecessary.
1631 * Virtualization detection works without priviliges now. This
1632 means the systemd-detect-virt binary no longer requires
1633 CAP_SYS_PTRACE file capabilities, and our daemons can run
1634 with fewer privileges.
1636 * systemd-networkd now runs under its own "systemd-network"
1637 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
1638 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
1639 loses the ability to write to files owned by root this way.
1641 * Similar, systemd-resolved now runs under its own
1642 "systemd-resolve" user with no capabilities remaining.
1644 * Similar, systemd-bus-proxyd now runs under its own
1645 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
1647 * systemd-networkd gained support for setting up "veth"
1648 virtual ethernet devices for container connectivity, as well
1649 as GRE and VTI tunnels.
1651 * systemd-networkd will no longer automatically attempt to
1652 manually load kernel modules necessary for certain tunnel
1653 transports. Instead, it is assumed the kernel loads them
1654 automatically when required. This only works correctly on
1655 very new kernels. On older kernels, please consider adding
1656 the kernel modules to /etc/modules-load.d/ as a work-around.
1658 * The resolv.conf file systemd-resolved generates has been
1659 moved to /run/systemd/resolve/. If you have a symlink from
1660 /etc/resolv.conf, it might be necessary to correct it.
1662 * Two new service settings, ProtectHome= and ProtectSystem=,
1663 have been added. When enabled, they will make the user data
1664 (such as /home) inaccessible or read-only and the system
1665 (such as /usr) read-only, for specific services. This allows
1666 very light-weight per-service sandboxing to avoid
1667 modifications of user data or system files from
1668 services. These two new switches have been enabled for all
1669 of systemd's long-running services, where appropriate.
1671 * Socket units gained new SocketUser= and SocketGroup=
1672 settings to set the owner user and group of AF_UNIX sockets
1673 and FIFOs in the file system.
1675 * Socket units gained a new RemoveOnStop= setting. If enabled,
1676 all FIFOS and sockets in the file system will be removed
1677 when the specific socket unit is stopped.
1679 * Socket units gained a new Symlinks= setting. It takes a list
1680 of symlinks to create to file system sockets or FIFOs
1681 created by the specific Unix sockets. This is useful to
1682 manage symlinks to socket nodes with the same life-cycle as
1685 * The /dev/log socket and /dev/initctl FIFO have been moved to
1686 /run, and have been replaced by symlinks. This allows
1687 connecting to these facilities even if PrivateDevices=yes is
1688 used for a service (which makes /dev/log itself unavailable,
1689 but /run is left). This also has the benefit of ensuring
1690 that /dev only contains device nodes, directories and
1691 symlinks, and nothing else.
1693 * sd-daemon gained two new calls sd_pid_notify() and
1694 sd_pid_notifyf(). They are similar to sd_notify() and
1695 sd_notifyf(), but allow overriding of the source PID of
1696 notification messages if permissions permit this. This is
1697 useful to send notify messages on behalf of a different
1698 process (for example, the parent process). The
1699 systemd-notify tool has been updated to make use of this
1700 when sending messages (so that notification messages now
1701 originate from the shell script invoking systemd-notify and
1702 not the systemd-notify process itself. This should minimize
1703 a race where systemd fails to associate notification
1704 messages to services when the originating process already
1707 * A new "on-abnormal" setting for Restart= has been added. If
1708 set, it will result in automatic restarts on all "abnormal"
1709 reasons for a process to exit, which includes unclean
1710 signals, core dumps, timeouts and watchdog timeouts, but
1711 does not include clean and unclean exit codes or clean
1712 signals. Restart=on-abnormal is an alternative for
1713 Restart=on-failure for services that shall be able to
1714 terminate and avoid restarts on certain errors, by
1715 indicating so with an unclean exit code. Restart=on-failure
1716 or Restart=on-abnormal is now the recommended setting for
1717 all long-running services.
1719 * If the InaccessibleDirectories= service setting points to a
1720 mount point (or if there are any submounts contained within
1721 it), it is now attempted to completely unmount it, to make
1722 the file systems truly unavailable for the respective
1725 * The ReadOnlyDirectories= service setting and
1726 systemd-nspawn's --read-only parameter are now recursively
1727 applied to all submounts, too.
1729 * Mount units may now be created transiently via the bus APIs.
1731 * The support for SysV and LSB init scripts has been removed
1732 from the systemd daemon itself. Instead, it is now
1733 implemented as a generator that creates native systemd units
1734 from these scripts when needed. This enables us to remove a
1735 substantial amount of legacy code from PID 1, following the
1736 fact that many distributions only ship a very small number
1737 of LSB/SysV init scripts nowadays.
1739 * Privileged Xen (dom0) domains are not considered
1740 virtualization anymore by the virtualization detection
1741 logic. After all, they generally have unrestricted access to
1742 the hardware and usually are used to manage the unprivileged
1745 * systemd-tmpfiles gained a new "C" line type, for copying
1746 files or entire directories.
1748 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
1749 lines. So far, they have been non-globbing versions of the
1750 latter, and have thus been redundant. In future, it is
1751 recommended to only use "z". "m" has hence been removed
1752 from the documentation, even though it stays supported.
1754 * A tmpfiles snippet to recreate the most basic structure in
1755 /var has been added. This is enough to create the /var/run →
1756 /run symlink and create a couple of structural
1757 directories. This allows systems to boot up with an empty or
1758 volatile /var. Of course, while with this change, the core OS
1759 now is capable with dealing with a volatile /var, not all
1760 user services are ready for it. However, we hope that sooner
1761 or later, many service daemons will be changed upstream so
1762 that they are able to automatically create their necessary
1763 directories in /var at boot, should they be missing. This is
1764 the first step to allow state-less systems that only require
1765 the vendor image for /usr to boot.
1767 * systemd-nspawn has gained a new --tmpfs= switch to mount an
1768 empty tmpfs instance to a specific directory. This is
1769 particularly useful for making use of the automatic
1770 reconstruction of /var (see above), by passing --tmpfs=/var.
1772 * Access modes specified in tmpfiles snippets may now be
1773 prefixed with "~", which indicates that they shall be masked
1774 by whether the existing file or directory is currently
1775 writable, readable or executable at all. Also, if specified,
1776 the sgid/suid/sticky bits will be masked for all
1779 * A new passive target unit "network-pre.target" has been
1780 added which is useful for services that shall run before any
1781 network is configured, for example firewall scripts.
1783 * The "floppy" group that previously owned the /dev/fd*
1784 devices is no longer used. The "disk" group is now used
1785 instead. Distributions should probably deprecate usage of
1788 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
1789 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
1790 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
1791 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
1792 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
1793 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
1796 -- Berlin, 2014-06-11
1800 * A new "systemd-timesyncd" daemon has been added for
1801 synchronizing the system clock across the network. It
1802 implements an SNTP client. In contrast to NTP
1803 implementations such as chrony or the NTP reference server,
1804 this only implements a client side, and does not bother with
1805 the full NTP complexity, focusing only on querying time from
1806 one remote server and synchronizing the local clock to
1807 it. Unless you intend to serve NTP to networked clients or
1808 want to connect to local hardware clocks, this simple NTP
1809 client should be more than appropriate for most
1810 installations. The daemon runs with minimal privileges, and
1811 has been hooked up with networkd to only operate when
1812 network connectivity is available. The daemon saves the
1813 current clock to disk every time a new NTP sync has been
1814 acquired, and uses this to possibly correct the system clock
1815 early at bootup, in order to accommodate for systems that
1816 lack an RTC such as the Raspberry Pi and embedded devices,
1817 and to make sure that time monotonically progresses on these
1818 systems, even if it is not always correct. To make use of
1819 this daemon, a new system user and group "systemd-timesync"
1820 needs to be created on installation of systemd.
1822 * The queue "seqnum" interface of libudev has been disabled, as
1823 it was generally incompatible with device namespacing as
1824 sequence numbers of devices go "missing" if the devices are
1825 part of a different namespace.
1827 * "systemctl list-timers" and "systemctl list-sockets" gained
1828 a --recursive switch for showing units of these types also
1829 for all local containers, similar in style to the already
1830 supported --recursive switch for "systemctl list-units".
1832 * A new RebootArgument= setting has been added for service
1833 units, which may be used to specify a kernel reboot argument
1834 to use when triggering reboots with StartLimitAction=.
1836 * A new FailureAction= setting has been added for service
1837 units which may be used to specify an operation to trigger
1838 when a service fails. This works similarly to
1839 StartLimitAction=, but unlike it, controls what is done
1840 immediately rather than only after several attempts to
1841 restart the service in question.
1843 * hostnamed got updated to also expose the kernel name,
1844 release, and version on the bus. This is useful for
1845 executing commands like hostnamectl with the -H switch.
1846 systemd-analyze makes use of this to properly display
1847 details when running non-locally.
1849 * The bootchart tool can now show cgroup information in the
1850 graphs it generates.
1852 * The CFS CPU quota cgroup attribute is now exposed for
1853 services. The new CPUQuota= switch has been added for this
1854 which takes a percentage value. Setting this will have the
1855 result that a service may never get more CPU time than the
1856 specified percentage, even if the machine is otherwise idle.
1858 * systemd-networkd learned IPIP and SIT tunnel support.
1860 * LSB init scripts exposing a dependency on $network will now
1861 get a dependency on network-online.target rather than simply
1862 network.target. This should bring LSB handling closer to
1863 what it was on SysV systems.
1865 * A new fsck.repair= kernel option has been added to control
1866 how fsck shall deal with unclean file systems at boot.
1868 * The (.ini) configuration file parser will now silently
1869 ignore sections whose name begins with "X-". This may be
1870 used to maintain application-specific extension sections in unit
1873 * machined gained a new API to query the IP addresses of
1874 registered containers. "machinectl status" has been updated
1875 to show these addresses in its output.
1877 * A new call sd_uid_get_display() has been added to the
1878 sd-login APIs for querying the "primary" session of a
1879 user. The "primary" session of the user is elected from the
1880 user's sessions and generally a graphical session is
1881 preferred over a text one.
1883 * A minimal systemd-resolved daemon has been added. It
1884 currently simply acts as a companion to systemd-networkd and
1885 manages resolv.conf based on per-interface DNS
1886 configuration, possibly supplied via DHCP. In the long run
1887 we hope to extend this into a local DNSSEC enabled DNS and
1890 * The systemd-networkd-wait-online tool is now enabled by
1891 default. It will delay network-online.target until a network
1892 connection has been configured. The tool primarily integrates
1893 with networkd, but will also make a best effort to make sense
1894 of network configuration performed in some other way.
1896 * Two new service options StartupCPUShares= and
1897 StartupBlockIOWeight= have been added that work similarly to
1898 CPUShares= and BlockIOWeight= however only apply during
1899 system startup. This is useful to prioritize certain services
1900 differently during bootup than during normal runtime.
1902 * hostnamed has been changed to prefer the statically
1903 configured hostname in /etc/hostname (unless set to
1904 'localhost' or empty) over any dynamic one supplied by
1905 dhcp. With this change, the rules for picking the hostname
1906 match more closely the rules of other configuration settings
1907 where the local administrator's configuration in /etc always
1908 overrides any other settings.
1910 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
1911 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
1912 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
1913 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
1914 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
1915 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
1916 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
1917 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
1918 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
1919 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
1920 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
1921 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
1922 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
1923 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
1924 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
1925 Lindskog, WaLyong Cho, Will Woods, Zbigniew
1928 -- Beijing, 2014-05-28
1932 * When restoring the screen brightness at boot, stay away from
1933 the darkest setting or from the lowest 5% of the available
1934 range, depending on which is the larger value of both. This
1935 should effectively protect the user from rebooting into a
1936 black screen, should the brightness have been set to minimum
1939 * sd-login gained a new sd_machine_get_class() call to
1940 determine the class ("vm" or "container") of a machine
1941 registered with machined.
1943 * sd-login gained new calls
1944 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
1945 to query the identity of the peer of a local AF_UNIX
1946 connection. They operate similarly to their sd_pid_get_xyz()
1949 * PID 1 will now maintain a system-wide system state engine
1950 with the states "starting", "running", "degraded",
1951 "maintenance", "stopping". These states are bound to system
1952 startup, normal runtime, runtime with at least one failed
1953 service, rescue/emergency mode and system shutdown. This
1954 state is shown in the "systemctl status" output when no unit
1955 name is passed. It is useful to determine system state, in
1956 particularly when doing so for many systems or containers at
1959 * A new command "list-machines" has been added to "systemctl"
1960 that lists all local OS containers and shows their system
1961 state (see above), if systemd runs inside of them.
1963 * systemctl gained a new "-r" switch to recursively enumerate
1964 units on all local containers, when used with the
1965 "list-unit" command (which is the default one that is
1966 executed when no parameters are specified).
1968 * The GPT automatic partition discovery logic will now honour
1969 two GPT partition flags: one may be set on a partition to
1970 cause it to be mounted read-only, and the other may be set
1971 on a partition to ignore it during automatic discovery.
1973 * Two new GPT type UUIDs have been added for automatic root
1974 partition discovery, for 32-bit and 64-bit ARM. This is not
1975 particularly useful for discovering the root directory on
1976 these architectures during bare-metal boots (since UEFI is
1977 not common there), but still very useful to allow booting of
1978 ARM disk images in nspawn with the -i option.
1980 * MAC addresses of interfaces created with nspawn's
1981 --network-interface= switch will now be generated from the
1982 machine name, and thus be stable between multiple invocations
1985 * logind will now automatically remove all IPC objects owned
1986 by a user if she or he fully logs out. This makes sure that
1987 users who are logged out cannot continue to consume IPC
1988 resources. This covers SysV memory, semaphores and message
1989 queues as well as POSIX shared memory and message
1990 queues. Traditionally, SysV and POSIX IPC had no life-cycle
1991 limits. With this functionality, that is corrected. This may
1992 be turned off by using the RemoveIPC= switch of logind.conf.
1994 * The systemd-machine-id-setup and tmpfiles tools gained a
1995 --root= switch to operate on a specific root directory,
1998 * journald can now forward logged messages to the TTYs of all
1999 logged in users ("wall"). This is the default for all
2000 emergency messages now.
2002 * A new tool systemd-journal-remote has been added to stream
2003 journal log messages across the network.
2005 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
2006 controller trees are mounted into it. Note that the
2007 directories mounted beneath it are not read-only. This is a
2008 security measure and is particularly useful because glibc
2009 actually includes a search logic to pick any tmpfs it can
2010 find to implement shm_open() if /dev/shm is not available
2011 (which it might very well be in namespaced setups).
2013 * machinectl gained a new "poweroff" command to cleanly power
2014 down a local OS container.
2016 * The PrivateDevices= unit file setting will now also drop the
2017 CAP_MKNOD capability from the capability bound set, and
2018 imply DevicePolicy=closed.
2020 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
2021 comprehensively on all long-running systemd services where
2022 this is appropriate.
2024 * systemd-udevd will now run in a disassociated mount
2025 namespace. To mount directories from udev rules, make sure to
2026 pull in mount units via SYSTEMD_WANTS properties.
2028 * The kdbus support gained support for uploading policy into
2029 the kernel. sd-bus gained support for creating "monitoring"
2030 connections that can eavesdrop into all bus communication
2031 for debugging purposes.
2033 * Timestamps may now be specified in seconds since the UNIX
2034 epoch Jan 1st, 1970 by specifying "@" followed by the value
2037 * Native tcpwrap support in systemd has been removed. tcpwrap
2038 is old code, not really maintained anymore and has serious
2039 shortcomings, and better options such as firewalls
2040 exist. For setups that require tcpwrap usage, please
2041 consider invoking your socket-activated service via tcpd,
2042 like on traditional inetd.
2044 * A new system.conf configuration option
2045 DefaultTimerAccuracySec= has been added that controls the
2046 default AccuracySec= setting of .timer units.
2048 * Timer units gained a new WakeSystem= switch. If enabled,
2049 timers configured this way will cause the system to resume
2050 from system suspend (if the system supports that, which most
2053 * Timer units gained a new Persistent= switch. If enabled,
2054 timers configured this way will save to disk when they have
2055 been last triggered. This information is then used on next
2056 reboot to possible execute overdue timer events, that
2057 could not take place because the system was powered off.
2058 This enables simple anacron-like behaviour for timer units.
2060 * systemctl's "list-timers" will now also list the time a
2061 timer unit was last triggered in addition to the next time
2062 it will be triggered.
2064 * systemd-networkd will now assign predictable IPv4LL
2065 addresses to its local interfaces.
2067 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
2068 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
2069 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
2070 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
2071 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
2072 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
2073 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
2074 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
2077 -- Berlin, 2014-03-25
2081 * A new unit file setting RestrictAddressFamilies= has been
2082 added to restrict which socket address families unit
2083 processes gain access to. This takes address family names
2084 like "AF_INET" or "AF_UNIX", and is useful to minimize the
2085 attack surface of services via exotic protocol stacks. This
2086 is built on seccomp system call filters.
2088 * Two new unit file settings RuntimeDirectory= and
2089 RuntimeDirectoryMode= have been added that may be used to
2090 manage a per-daemon runtime directories below /run. This is
2091 an alternative for setting up directory permissions with
2092 tmpfiles snippets, and has the advantage that the runtime
2093 directory's lifetime is bound to the daemon runtime and that
2094 the daemon starts up with an empty directory each time. This
2095 is particularly useful when writing services that drop
2096 privileges using the User= or Group= setting.
2098 * The DeviceAllow= unit setting now supports globbing for
2099 matching against device group names.
2101 * The systemd configuration file system.conf gained new
2102 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
2103 DefaultMemoryAccounting= to globally turn on/off accounting
2104 for specific resources (cgroups) for all units. These
2105 settings may still be overridden individually in each unit
2108 * systemd-gpt-auto-generator is now able to discover /srv and
2109 root partitions in addition to /home and swap partitions. It
2110 also supports LUKS-encrypted partitions now. With this in
2111 place, automatic discovery of partitions to mount following
2112 the Discoverable Partitions Specification
2113 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
2114 is now a lot more complete. This allows booting without
2115 /etc/fstab and without root= on the kernel command line on
2116 systems prepared appropriately.
2118 * systemd-nspawn gained a new --image= switch which allows
2119 booting up disk images and Linux installations on any block
2120 device that follow the Discoverable Partitions Specification
2121 (see above). This means that installations made with
2122 appropriately updated installers may now be started and
2123 deployed using container managers, completely
2124 unmodified. (We hope that libvirt-lxc will add support for
2125 this feature soon, too.)
2127 * systemd-nspawn gained a new --network-macvlan= setting to
2128 set up a private macvlan interface for the
2129 container. Similarly, systemd-networkd gained a new
2130 Kind=macvlan setting in .netdev files.
2132 * systemd-networkd now supports configuring local addresses
2135 * A new tool systemd-network-wait-online has been added to
2136 synchronously wait for network connectivity using
2139 * The sd-bus.h bus API gained a new sd_bus_track object for
2140 tracking the life-cycle of bus peers. Note that sd-bus.h is
2141 still not a public API though (unless you specify
2142 --enable-kdbus on the configure command line, which however
2143 voids your warranty and you get no API stability guarantee).
2145 * The $XDG_RUNTIME_DIR runtime directories for each user are
2146 now individual tmpfs instances, which has the benefit of
2147 introducing separate pools for each user, with individual
2148 size limits, and thus making sure that unprivileged clients
2149 can no longer negatively impact the system or other users by
2150 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
2151 RuntimeDirectorySize= has been introduced that allows
2152 controlling the default size limit for all users. It
2153 defaults to 10% of the available physical memory. This is no
2154 replacement for quotas on tmpfs though (which the kernel
2155 still does not support), as /dev/shm and /tmp are still
2156 shared resources used by both the system and unprivileged
2159 * logind will now automatically turn off automatic suspending
2160 on laptop lid close when more than one display is
2161 connected. This was previously expected to be implemented
2162 individually in desktop environments (such as GNOME),
2163 however has been added to logind now, in order to fix a
2164 boot-time race where a desktop environment might not have
2165 been started yet and thus not been able to take an inhibitor
2166 lock at the time where logind already suspends the system
2167 due to a closed lid.
2169 * logind will now wait at least 30s after each system
2170 suspend/resume cycle, and 3min after system boot before
2171 suspending the system due to a closed laptop lid. This
2172 should give USB docking stations and similar enough time to
2173 be probed and configured after system resume and boot in
2174 order to then act as suspend blocker.
2176 * systemd-run gained a new --property= setting which allows
2177 initialization of resource control properties (and others)
2178 for the created scope or service unit. Example: "systemd-run
2179 --property=BlockIOWeight=10 updatedb" may be used to run
2180 updatedb at a low block IO scheduling weight.
2182 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
2183 now also work in --scope mode.
2185 * When systemd is compiled with kdbus support, basic support
2186 for enforced policies is now in place. (Note that enabling
2187 kdbus still voids your warranty and no API compatibility
2190 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
2191 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
2192 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
2193 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
2194 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
2195 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
2196 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
2197 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
2198 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
2199 Zbigniew Jędrzejewski-Szmek
2201 -- Berlin, 2014-03-12
2205 * systemd will now relabel /dev after loading the SMACK policy
2206 according to SMACK rules.
2208 * A new unit file option AppArmorProfile= has been added to
2209 set the AppArmor profile for the processes of a unit.
2211 * A new condition check ConditionArchitecture= has been added
2212 to conditionalize units based on the system architecture, as
2213 reported by uname()'s "machine" field.
2215 * systemd-networkd now supports matching on the system
2216 virtualization, architecture, kernel command line, host name
2219 * logind is now a lot more aggressive when suspending the
2220 machine due to a closed laptop lid. Instead of acting only
2221 on the lid close action, it will continuously watch the lid
2222 status and act on it. This is useful for laptops where the
2223 power button is on the outside of the chassis so that it can
2224 be reached without opening the lid (such as the Lenovo
2225 Yoga). On those machines, logind will now immediately
2226 re-suspend the machine if the power button has been
2227 accidentally pressed while the laptop was suspended and in a
2228 backpack or similar.
2230 * logind will now watch SW_DOCK switches and inhibit reaction
2231 to the lid switch if it is pressed. This means that logind
2232 will not suspend the machine anymore if the lid is closed
2233 and the system is docked, if the laptop supports SW_DOCK
2234 notifications via the input layer. Note that ACPI docking
2235 stations do not generate this currently. Also note that this
2236 logic is usually not fully sufficient and Desktop
2237 Environments should take a lid switch inhibitor lock when an
2238 external display is connected, as systemd will not watch
2241 * nspawn will now make use of the devices cgroup controller by
2242 default, and only permit creation of and access to the usual
2243 API device nodes like /dev/null or /dev/random, as well as
2244 access to (but not creation of) the pty devices.
2246 * We will now ship a default .network file for
2247 systemd-networkd that automatically configures DHCP for
2248 network interfaces created by nspawn's --network-veth or
2249 --network-bridge= switches.
2251 * systemd will now understand the usual M, K, G, T suffixes
2252 according to SI conventions (i.e. to the base 1000) when
2253 referring to throughput and hardware metrics. It will stay
2254 with IEC conventions (i.e. to the base 1024) for software
2255 metrics, according to what is customary according to
2256 Wikipedia. We explicitly document which base applies for
2257 each configuration option.
2259 * The DeviceAllow= setting in unit files now supports a syntax
2260 to whitelist an entire group of devices node majors at once,
2261 based on the /proc/devices listing. For example, with the
2262 string "char-pts", it is now possible to whitelist all
2263 current and future pseudo-TTYs at once.
2265 * sd-event learned a new "post" event source. Event sources of
2266 this type are triggered by the dispatching of any event
2267 source of a type that is not "post". This is useful for
2268 implementing clean-up and check event sources that are
2269 triggered by other work being done in the program.
2271 * systemd-networkd is no longer statically enabled, but uses
2272 the usual [Install] sections so that it can be
2273 enabled/disabled using systemctl. It still is enabled by
2276 * When creating a veth interface pair with systemd-nspawn, the
2277 host side will now be prefixed with "vb-" if
2278 --network-bridge= is used, and with "ve-" if --network-veth
2279 is used. This way, it is easy to distinguish these cases on
2280 the host, for example to apply different configuration to
2281 them with systemd-networkd.
2283 * The compatibility libraries for libsystemd-journal.so,
2284 libsystem-id128.so, libsystemd-login.so and
2285 libsystemd-daemon.so do not make use of IFUNC
2286 anymore. Instead, we now build libsystemd.so multiple times
2287 under these alternative names. This means that the footprint
2288 is drastically increased, but given that these are
2289 transitional compatibility libraries, this should not matter
2290 much. This change has been made necessary to support the ARM
2291 platform for these compatibility libraries, as the ARM
2292 toolchain is not really at the same level as the toolchain
2293 for other architectures like x86 and does not support
2294 IFUNC. Please make sure to use --enable-compat-libs only
2295 during a transitional period!
2297 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
2298 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
2299 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
2300 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
2301 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
2302 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
2303 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
2304 Zbigniew Jędrzejewski-Szmek
2306 -- Berlin, 2014-02-24
2310 * A new component "systemd-networkd" has been added that can
2311 be used to configure local network interfaces statically or
2312 via DHCP. It is capable of bringing up bridges, VLANs, and
2313 bonding. Currently, no hook-ups for interactive network
2314 configuration are provided. Use this for your initrd,
2315 container, embedded, or server setup if you need a simple,
2316 yet powerful, network configuration solution. This
2317 configuration subsystem is quite nifty, as it allows wildcard
2318 hotplug matching in interfaces. For example, with a single
2319 configuration snippet, you can configure that all Ethernet
2320 interfaces showing up are automatically added to a bridge,
2321 or similar. It supports link-sensing and more.
2323 * A new tool "systemd-socket-proxyd" has been added which can
2324 act as a bidirectional proxy for TCP sockets. This is
2325 useful for adding socket activation support to services that
2326 do not actually support socket activation, including virtual
2327 machines and the like.
2329 * Add a new tool to save/restore rfkill state on
2332 * Save/restore state of keyboard backlights in addition to
2333 display backlights on shutdown/boot.
2335 * udev learned a new SECLABEL{} construct to label device
2336 nodes with a specific security label when they appear. For
2337 now, only SECLABEL{selinux} is supported, but the syntax is
2338 prepared for additional security frameworks.
2340 * udev gained a new scheme to configure link-level attributes
2341 from files in /etc/systemd/network/*.link. These files can
2342 match against MAC address, device path, driver name and type,
2343 and will apply attributes like the naming policy, link speed,
2344 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
2345 address assignment policy (randomized, ...).
2347 * The configuration of network interface naming rules for
2348 "permanent interface names" has changed: a new NamePolicy=
2349 setting in the [Link] section of .link files determines the
2350 priority of possible naming schemes (onboard, slot, mac,
2351 path). The default value of this setting is determined by
2352 /usr/lib/net/links/99-default.link. Old
2353 80-net-name-slot.rules udev configuration file has been
2354 removed, so local configuration overriding this file should
2355 be adapated to override 99-default.link instead.
2357 * When the User= switch is used in a unit file, also
2358 initialize $SHELL= based on the user database entry.
2360 * systemd no longer depends on libdbus. All communication is
2361 now done with sd-bus, systemd's low-level bus library
2364 * kdbus support has been added to PID 1 itself. When kdbus is
2365 enabled, this causes PID 1 to set up the system bus and
2366 enable support for a new ".busname" unit type that
2367 encapsulates bus name activation on kdbus. It works a little
2368 bit like ".socket" units, except for bus names. A new
2369 generator has been added that converts classic dbus1 service
2370 activation files automatically into native systemd .busname
2373 * sd-bus: add a light-weight vtable implementation that allows
2374 defining objects on the bus with a simple static const
2375 vtable array of its methods, signals and properties.
2377 * systemd will not generate or install static dbus
2378 introspection data anymore to /usr/share/dbus-1/interfaces,
2379 as the precise format of these files is unclear, and
2380 nothing makes use of it.
2382 * A proxy daemon is now provided to proxy clients connecting
2383 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
2384 compatibility with classic D-Bus.
2386 * A bus driver implementation has been added that supports the
2387 classic D-Bus bus driver calls on kdbus, also for
2388 compatibility purposes.
2390 * A new API "sd-event.h" has been added that implements a
2391 minimal event loop API built around epoll. It provides a
2392 couple of features that direct epoll usage is lacking:
2393 prioritization of events, scales to large numbers of timer
2394 events, per-event timer slack (accuracy), system-wide
2395 coalescing of timer events, exit handlers, watchdog
2396 supervision support using systemd's sd_notify() API, child
2399 * A new API "sd-rntl.h" has been added that provides an API
2400 around the route netlink interface of the kernel, similar in
2401 style to "sd-bus.h".
2403 * A new API "sd-dhcp-client.h" has been added that provides a
2404 small DHCPv4 client-side implementation. This is used by
2407 * There is a new kernel command line option
2408 "systemd.restore_state=0|1". When set to "0", none of the
2409 systemd tools will restore saved runtime state to hardware
2410 devices. More specifically, the rfkill and backlight states
2413 * The FsckPassNo= compatibility option in mount/service units
2414 has been removed. The fstab generator will now add the
2415 necessary dependencies automatically, and does not require
2416 PID1's support for that anymore.
2418 * journalctl gained a new switch, --list-boots, that lists
2419 recent boots with their times and boot IDs.
2421 * The various tools like systemctl, loginctl, timedatectl,
2422 busctl, systemd-run, ... have gained a new switch "-M" to
2423 connect to a specific, local OS container (as direct
2424 connection, without requiring SSH). This works on any
2425 container that is registered with machined, such as those
2426 created by libvirt-lxc or nspawn.
2428 * systemd-run and systemd-analyze also gained support for "-H"
2429 to connect to remote hosts via SSH. This is particularly
2430 useful for systemd-run because it enables queuing of jobs
2431 onto remote systems.
2433 * machinectl gained a new command "login" to open a getty
2434 login in any local container. This works with any container
2435 that is registered with machined (such as those created by
2436 libvirt-lxc or nspawn), and which runs systemd inside.
2438 * machinectl gained a new "reboot" command that may be used to
2439 trigger a reboot on a specific container that is registered
2440 with machined. This works on any container that runs an init
2441 system of some kind.
2443 * systemctl gained a new "list-timers" command to print a nice
2444 listing of installed timer units with the times they elapse
2447 * Alternative reboot() parameters may now be specified on the
2448 "systemctl reboot" command line and are passed to the
2449 reboot() system call.
2451 * systemctl gained a new --job-mode= switch to configure the
2452 mode to queue a job with. This is a more generic version of
2453 --fail, --irreversible, and --ignore-dependencies, which are
2454 still available but not advertised anymore.
2456 * /etc/systemd/system.conf gained new settings to configure
2457 various default timeouts of units, as well as the default
2458 start limit interval and burst. These may still be overridden
2461 * PID1 will now export on the bus profile data of the security
2462 policy upload process (such as the SELinux policy upload to
2465 * journald: when forwarding logs to the console, include
2466 timestamps (following the setting in
2467 /sys/module/printk/parameters/time).
2469 * OnCalendar= in timer units now understands the special
2470 strings "yearly" and "annually". (Both are equivalent)
2472 * The accuracy of timer units is now configurable with the new
2473 AccuracySec= setting. It defaults to 1min.
2475 * A new dependency type JoinsNamespaceOf= has been added that
2476 allows running two services within the same /tmp and network
2477 namespace, if PrivateNetwork= or PrivateTmp= are used.
2479 * A new command "cat" has been added to systemctl. It outputs
2480 the original unit file of a unit, and concatenates the
2481 contents of additional "drop-in" unit file snippets, so that
2482 the full configuration is shown.
2484 * systemctl now supports globbing on the various "list-xyz"
2485 commands, like "list-units" or "list-sockets", as well as on
2486 those commands which take multiple unit names.
2488 * journalctl's --unit= switch gained support for globbing.
2490 * All systemd daemons now make use of the watchdog logic so
2491 that systemd automatically notices when they hang.
2493 * If the $container_ttys environment variable is set,
2494 getty-generator will automatically spawn a getty for each
2495 listed tty. This is useful for container managers to request
2496 login gettys to be spawned on as many ttys as needed.
2498 * %h, %s, %U specifier support is not available anymore when
2499 used in unit files for PID 1. This is because NSS calls are
2500 not safe from PID 1. They stay available for --user
2501 instances of systemd, and as special case for the root user.
2503 * loginctl gained a new "--no-legend" switch to turn off output
2506 * The "sd-login.h" API gained three new calls:
2507 sd_session_is_remote(), sd_session_get_remote_user(),
2508 sd_session_get_remote_host() to query information about
2511 * The udev hardware database now also carries vendor/product
2512 information of SDIO devices.
2514 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
2515 determine whether watchdog notifications are requested by
2518 * Socket-activated per-connection services now include a
2519 short description of the connection parameters in the
2522 * tmpfiles gained a new "--boot" option. When this is not used,
2523 only lines where the command character is not suffixed with
2524 "!" are executed. When this option is specified, those
2525 options are executed too. This partitions tmpfiles
2526 directives into those that can be safely executed at any
2527 time, and those which should be run only at boot (for
2528 example, a line that creates /run/nologin).
2530 * A new API "sd-resolve.h" has been added which provides a simple
2531 asynchronous wrapper around glibc NSS host name resolution
2532 calls, such as getaddrinfo(). In contrast to glibc's
2533 getaddrinfo_a(), it does not use signals. In contrast to most
2534 other asynchronous name resolution libraries, this one does
2535 not reimplement DNS, but reuses NSS, so that alternate
2536 host name resolution systems continue to work, such as mDNS,
2537 LDAP, etc. This API is based on libasyncns, but it has been
2538 cleaned up for inclusion in systemd.
2540 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
2541 "sd-daemon.h" are no longer found in individual libraries
2542 libsystemd-journal.so, libsystemd-login.so,
2543 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
2544 merged them into a single library, libsystemd.so, which
2545 provides all symbols. The reason for this is cyclic
2546 dependencies, as these libraries tend to use each other's
2547 symbols. So far, we have managed to workaround that by linking
2548 a copy of a good part of our code into each of these
2549 libraries again and again, which, however, makes certain
2550 things hard to do, like sharing static variables. Also, it
2551 substantially increases footprint. With this change, there
2552 is only one library for the basic APIs systemd
2553 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
2554 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
2555 library as well, however are subject to the --enable-kdbus
2556 switch (see below). Note that "sd-dhcp-client.h" is not part
2557 of this library (this is because it only consumes, never
2558 provides, services of/to other APIs). To make the transition
2559 easy from the separate libraries to the unified one, we
2560 provide the --enable-compat-libs compile-time switch which
2561 will generate stub libraries that are compatible with the
2562 old ones but redirect all calls to the new one.
2564 * All of the kdbus logic and the new APIs "sd-bus.h",
2565 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
2566 and "sd-utf8.h" are compile-time optional via the
2567 "--enable-kdbus" switch, and they are not compiled in by
2568 default. To make use of kdbus, you have to explicitly enable
2569 the switch. Note however, that neither the kernel nor the
2570 userspace API for all of this is considered stable yet. We
2571 want to maintain the freedom to still change the APIs for
2572 now. By specifying this build-time switch, you acknowledge
2573 that you are aware of the instability of the current
2576 * Also, note that while kdbus is pretty much complete,
2577 it lacks one thing: proper policy support. This means you
2578 can build a fully working system with all features; however,
2579 it will be highly insecure. Policy support will be added in
2580 one of the next releases, at the same time that we will
2581 declare the APIs stable.
2583 * When the kernel command line argument "kdbus" is specified,
2584 systemd will automatically load the kdbus.ko kernel module. At
2585 this stage of development, it is only useful for testing kdbus
2586 and should not be used in production. Note: if "--enable-kdbus"
2587 is specified, and the kdbus.ko kernel module is available, and
2588 "kdbus" is added to the kernel command line, the entire system
2589 runs with kdbus instead of dbus-daemon, with the above mentioned
2590 problem of missing the system policy enforcement. Also a future
2591 version of kdbus.ko or a newer systemd will not be compatible with
2592 each other, and will unlikely be able to boot the machine if only
2593 one of them is updated.
2595 * systemctl gained a new "import-environment" command which
2596 uploads the caller's environment (or parts thereof) into the
2597 service manager so that it is inherited by services started
2598 by the manager. This is useful to upload variables like
2599 $DISPLAY into the user service manager.
2601 * A new PrivateDevices= switch has been added to service units
2602 which allows running a service with a namespaced /dev
2603 directory that does not contain any device nodes for
2604 physical devices. More specifically, it only includes devices
2605 such as /dev/null, /dev/urandom, and /dev/zero which are API
2608 * logind has been extended to support behaviour like VT
2609 switching on seats that do not support a VT. This makes
2610 multi-session available on seats that are not the first seat
2611 (seat0), and on systems where kernel support for VTs has
2612 been disabled at compile-time.
2614 * If a process holds a delay lock for system sleep or shutdown
2615 and fails to release it in time, we will now log its
2616 identity. This makes it easier to identify processes that
2617 cause slow suspends or power-offs.
2619 * When parsing /etc/crypttab, support for a new key-slot=
2620 option as supported by Debian is added. It allows indicating
2621 which LUKS slot to use on disk, speeding up key loading.
2623 * The sd_journald_sendv() API call has been checked and
2624 officially declared to be async-signal-safe so that it may
2625 be invoked from signal handlers for logging purposes.
2627 * Boot-time status output is now enabled automatically after a
2628 short timeout if boot does not progress, in order to give
2629 the user an indication what she or he is waiting for.
2631 * The boot-time output has been improved to show how much time
2632 remains until jobs expire.
2634 * The KillMode= switch in service units gained a new possible
2635 value "mixed". If set, and the unit is shut down, then the
2636 initial SIGTERM signal is sent only to the main daemon
2637 process, while the following SIGKILL signal is sent to
2638 all remaining processes of the service.
2640 * When a scope unit is registered, a new property "Controller"
2641 may be set. If set to a valid bus name, systemd will send a
2642 RequestStop() signal to this name when it would like to shut
2643 down the scope. This may be used to hook manager logic into
2644 the shutdown logic of scope units. Also, scope units may now
2645 be put in a special "abandoned" state, in which case the
2646 manager process which created them takes no further
2647 responsibilities for it.
2649 * When reading unit files, systemd will now verify
2650 the access mode of these files, and warn about certain
2651 suspicious combinations. This has been added to make it
2652 easier to track down packaging bugs where unit files are
2653 marked executable or world-writable.
2655 * systemd-nspawn gained a new "--setenv=" switch to set
2656 container-wide environment variables. The similar option in
2657 systemd-activate was renamed from "--environment=" to
2658 "--setenv=" for consistency.
2660 * systemd-nspawn has been updated to create a new kdbus domain
2661 for each container that is invoked, thus allowing each
2662 container to have its own set of system and user buses,
2663 independent of the host.
2665 * systemd-nspawn gained a new --drop-capability= switch to run
2666 the container with less capabilities than the default. Both
2667 --drop-capability= and --capability= now take the special
2668 string "all" for dropping or keeping all capabilities.
2670 * systemd-nspawn gained new switches for executing containers
2671 with specific SELinux labels set.
2673 * systemd-nspawn gained a new --quiet switch to not generate
2674 any additional output but the container's own console
2677 * systemd-nspawn gained a new --share-system switch to run a
2678 container without PID namespacing enabled.
2680 * systemd-nspawn gained a new --register= switch to control
2681 whether the container is registered with systemd-machined or
2682 not. This is useful for containers that do not run full
2683 OS images, but only specific apps.
2685 * systemd-nspawn gained a new --keep-unit which may be used
2686 when invoked as the only program from a service unit, and
2687 results in registration of the unit service itself in
2688 systemd-machined, instead of a newly opened scope unit.
2690 * systemd-nspawn gained a new --network-interface= switch for
2691 moving arbitrary interfaces to the container. The new
2692 --network-veth switch creates a virtual Ethernet connection
2693 between host and container. The new --network-bridge=
2694 switch then allows assigning the host side of this virtual
2695 Ethernet connection to a bridge device.
2697 * systemd-nspawn gained a new --personality= switch for
2698 setting the kernel personality for the container. This is
2699 useful when running a 32-bit container on a 64-bit host. A
2700 similar option Personality= is now also available for service
2703 * logind will now also track a "Desktop" identifier for each
2704 session which encodes the desktop environment of it. This is
2705 useful for desktop environments that want to identify
2706 multiple running sessions of itself easily.
2708 * A new SELinuxContext= setting for service units has been
2709 added that allows setting a specific SELinux execution
2710 context for a service.
2712 * Most systemd client tools will now honour $SYSTEMD_LESS for
2713 settings of the "less" pager. By default, these tools will
2714 override $LESS to allow certain operations to work, such as
2715 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
2716 influence this logic.
2718 * systemd's "seccomp" hook-up has been changed to make use of
2719 the libseccomp library instead of using its own
2720 implementation. This has benefits for portability among
2723 * For usage together with SystemCallFilter=, a new
2724 SystemCallErrorNumber= setting has been introduced that
2725 allows configuration of a system error number to be returned
2726 on filtered system calls, instead of immediately killing the
2727 process. Also, SystemCallArchitectures= has been added to
2728 limit access to system calls of a particular architecture
2729 (in order to turn off support for unused secondary
2730 architectures). There is also a global
2731 SystemCallArchitectures= setting in system.conf now to turn
2732 off support for non-native system calls system-wide.
2734 * systemd requires a kernel with a working name_to_handle_at(),
2735 please see the kernel config requirements in the README file.
2737 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
2738 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
2739 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
2740 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
2741 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
2742 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
2743 Elia Pinto, Florian Weimer, George McCollister, Goffredo
2744 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
2745 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
2746 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
2747 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
2748 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
2749 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
2750 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
2751 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
2752 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
2753 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
2754 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
2755 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
2756 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
2757 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
2758 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
2759 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
2760 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
2762 -- Berlin, 2014-02-20
2766 * logind has gained support for facilitating privileged input
2767 and drm device access for unprivileged clients. This work is
2768 useful to allow Wayland display servers (and similar
2769 programs, such as kmscon) to run under the user's ID and
2770 access input and drm devices which are normally
2771 protected. When this is used (and the kernel is new enough)
2772 logind will "mute" IO on the file descriptors passed to
2773 Wayland as long as it is in the background and "unmute" it
2774 if it returns into the foreground. This allows secure
2775 session switching without allowing background sessions to
2776 eavesdrop on input and display data. This also introduces
2777 session switching support if VT support is turned off in the
2778 kernel, and on seats that are not seat0.
2780 * A new kernel command line option luks.options= is understood
2781 now which allows specifying LUKS options for usage for LUKS
2782 encrypted partitions specified with luks.uuid=.
2784 * tmpfiles.d(5) snippets may now use specifier expansion in
2785 path names. More specifically %m, %b, %H, %v, are now
2786 replaced by the local machine id, boot id, hostname, and
2787 kernel version number.
2789 * A new tmpfiles.d(5) command "m" has been introduced which
2790 may be used to change the owner/group/access mode of a file
2791 or directory if it exists, but do nothing if it does not.
2793 * This release removes high-level support for the
2794 MemorySoftLimit= cgroup setting. The underlying kernel
2795 cgroup attribute memory.soft_limit= is currently badly
2796 designed and likely to be removed from the kernel API in its
2797 current form, hence we should not expose it for now.
2799 * The memory.use_hierarchy cgroup attribute is now enabled for
2800 all cgroups systemd creates in the memory cgroup
2801 hierarchy. This option is likely to be come the built-in
2802 default in the kernel anyway, and the non-hierarchical mode
2803 never made much sense in the intrinsically hierarchical
2806 * A new field _SYSTEMD_SLICE= is logged along with all journal
2807 messages containing the slice a message was generated
2808 from. This is useful to allow easy per-customer filtering of
2809 logs among other things.
2811 * systemd-journald will no longer adjust the group of journal
2812 files it creates to the "systemd-journal" group. Instead we
2813 rely on the journal directory to be owned by the
2814 "systemd-journal" group, and its setgid bit set, so that the
2815 kernel file system layer will automatically enforce that
2816 journal files inherit this group assignment. The reason for
2817 this change is that we cannot allow NSS look-ups from
2818 journald which would be necessary to resolve
2819 "systemd-journal" to a numeric GID, because this might
2820 create deadlocks if NSS involves synchronous queries to
2821 other daemons (such as nscd, or sssd) which in turn are
2822 logging clients of journald and might block on it, which
2823 would then dead lock. A tmpfiles.d(5) snippet included in
2824 systemd will make sure the setgid bit and group are
2825 properly set on the journal directory if it exists on every
2826 boot. However, we recommend adjusting it manually after
2827 upgrades too (or from RPM scriptlets), so that the change is
2828 not delayed until next reboot.
2830 * Backlight and random seed files in /var/lib/ have moved into
2831 the /var/lib/systemd/ directory, in order to centralize all
2832 systemd generated files in one directory.
2834 * Boot time performance measurements (as displayed by
2835 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
2836 performance information if that's available to determine how
2837 much time BIOS and boot loader initialization required. With
2838 a sufficiently new BIOS you hence no longer need to boot
2839 with Gummiboot to get access to such information.
2841 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
2842 Cristian Rodríguez, Dave Reisner, David Herrmann, David
2843 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
2844 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
2845 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
2846 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
2847 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
2849 -- Berlin, 2013-10-02
2853 * The Restart= option for services now understands a new
2854 on-watchdog setting, which will restart the service
2855 automatically if the service stops sending out watchdog keep
2856 alive messages (as configured with WatchdogSec=).
2858 * The getty generator (which is responsible for bringing up a
2859 getty on configured serial consoles) will no longer only
2860 start a getty on the primary kernel console but on all
2861 others, too. This makes the order in which console= is
2862 specified on the kernel command line less important.
2864 * libsystemd-logind gained a new sd_session_get_vt() call to
2865 retrieve the VT number of a session.
2867 * If the option "tries=0" is set for an entry of /etc/crypttab
2868 its passphrase is queried indefinitely instead of any
2869 maximum number of tries.
2871 * If a service with a configure PID file terminates its PID
2872 file will now be removed automatically if it still exists
2873 afterwards. This should put an end to stale PID files.
2875 * systemd-run will now also take relative binary path names
2876 for execution and no longer insists on absolute paths.
2878 * InaccessibleDirectories= and ReadOnlyDirectories= now take
2879 paths that are optionally prefixed with "-" to indicate that
2880 it should not be considered a failure if they do not exist.
2882 * journalctl -o (and similar commands) now understands a new
2883 output mode "short-precise", it is similar to "short" but
2884 shows timestamps with usec accuracy.
2886 * The option "discard" (as known from Debian) is now
2887 synonymous to "allow-discards" in /etc/crypttab. In fact,
2888 "discard" is preferred now (since it is easier to remember
2891 * Some licensing clean-ups were made, so that more code is now
2892 LGPL-2.1 licensed than before.
2894 * A minimal tool to save/restore the display backlight
2895 brightness across reboots has been added. It will store the
2896 backlight setting as late as possible at shutdown, and
2897 restore it as early as possible during reboot.
2899 * A logic to automatically discover and enable home and swap
2900 partitions on GPT disks has been added. With this in place
2901 /etc/fstab becomes optional for many setups as systemd can
2902 discover certain partitions located on the root disk
2903 automatically. Home partitions are recognized under their
2904 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
2905 partitions are recognized under their GPT type ID
2906 0657fd6da4ab43c484e50933c84b4f4f.
2908 * systemd will no longer pass any environment from the kernel
2909 or initrd to system services. If you want to set an
2910 environment for all services, do so via the kernel command
2911 line systemd.setenv= assignment.
2913 * The systemd-sysctl tool no longer natively reads the file
2914 /etc/sysctl.conf. If desired, the file should be symlinked
2915 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
2916 legacy support by a symlink rather than built-in code, it
2917 also makes the otherwise hidden order of application of the
2918 different files visible. (Note that this partly reverts to a
2919 pre-198 application order of sysctl knobs!)
2921 * The "systemctl set-log-level" and "systemctl dump" commands
2922 have been moved to systemd-analyze.
2924 * systemd-run learned the new --remain-after-exit switch,
2925 which causes the scope unit not to be cleaned up
2926 automatically after the process terminated.
2928 * tmpfiles learned a new --exclude-prefix= switch to exclude
2929 certain paths from operation.
2931 * journald will now automatically flush all messages to disk
2932 as soon as a message at the log level CRIT, ALERT or EMERG
2935 Contributions from: Andrew Cook, Brandon Philips, Christian
2936 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
2937 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
2938 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
2939 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
2940 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
2941 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
2942 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
2943 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
2944 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
2945 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
2946 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
2947 William Giokas, Zbigniew Jędrzejewski-Szmek
2949 -- Berlin, 2013-09-13
2953 * The documentation has been updated to cover the various new
2954 concepts introduced with 205.
2956 * Unit files now understand the new %v specifier which
2957 resolves to the kernel version string as returned by "uname
2960 * systemctl now supports filtering the unit list output by
2961 load state, active state and sub state, using the new
2964 * "systemctl status" will now show the results of the
2965 condition checks (like ConditionPathExists= and similar) of
2966 the last start attempts of the unit. They are also logged to
2969 * "journalctl -b" may now be used to look for boot output of a
2970 specific boot. Try "journalctl -b -1" for the previous boot,
2971 but the syntax is substantially more powerful.
2973 * "journalctl --show-cursor" has been added which prints the
2974 cursor string the last shown log line. This may then be used
2975 with the new "journalctl --after-cursor=" switch to continue
2976 browsing logs from that point on.
2978 * "journalctl --force" may now be used to force regeneration
2981 * Creation of "dead" device nodes has been moved from udev
2982 into kmod and tmpfiles. Previously, udev would read the kmod
2983 databases to pre-generate dead device nodes based on meta
2984 information contained in kernel modules, so that these would
2985 be auto-loaded on access rather then at boot. As this
2986 does not really have much to do with the exposing actual
2987 kernel devices to userspace this has always been slightly
2988 alien in the udev codebase. Following the new scheme kmod
2989 will now generate a runtime snippet for tmpfiles from the
2990 module meta information and it now is tmpfiles' job to the
2991 create the nodes. This also allows overriding access and
2992 other parameters for the nodes using the usual tmpfiles
2993 facilities. As side effect this allows us to remove the
2994 CAP_SYS_MKNOD capability bit from udevd entirely.
2996 * logind's device ACLs may now be applied to these "dead"
2997 devices nodes too, thus finally allowing managed access to
2998 devices such as /dev/snd/sequencer whithout loading the
2999 backing module right-away.
3001 * A new RPM macro has been added that may be used to apply
3002 tmpfiles configuration during package installation.
3004 * systemd-detect-virt and ConditionVirtualization= now can
3005 detect User-Mode-Linux machines (UML).
3007 * journald will now implicitly log the effective capabilities
3008 set of processes in the message metadata.
3010 * systemd-cryptsetup has gained support for TrueCrypt volumes.
3012 * The initrd interface has been simplified (more specifically,
3013 support for passing performance data via environment
3014 variables and fsck results via files in /run has been
3015 removed). These features were non-essential, and are
3016 nowadays available in a much nicer way by having systemd in
3017 the initrd serialize its state and have the hosts systemd
3018 deserialize it again.
3020 * The udev "keymap" data files and tools to apply keyboard
3021 specific mappings of scan to key codes, and force-release
3022 scan code lists have been entirely replaced by a udev
3023 "keyboard" builtin and a hwdb data file.
3025 * systemd will now honour the kernel's "quiet" command line
3026 argument also during late shutdown, resulting in a
3027 completely silent shutdown when used.
3029 * There's now an option to control the SO_REUSEPORT socket
3030 option in .socket units.
3032 * Instance units will now automatically get a per-template
3033 subslice of system.slice unless something else is explicitly
3034 configured. For example, instances of sshd@.service will now
3035 implicitly be placed in system-sshd.slice rather than
3036 system.slice as before.
3038 * Test coverage support may now be enabled at build time.
3040 Contributions from: Dave Reisner, Frederic Crozat, Harald
3041 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
3042 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
3043 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
3044 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
3045 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
3046 Giokas, Zbigniew Jędrzejewski-Szmek
3048 -- Berlin, 2013-07-23
3052 * Two new unit types have been introduced:
3054 Scope units are very similar to service units, however, are
3055 created out of pre-existing processes -- instead of PID 1
3056 forking off the processes. By using scope units it is
3057 possible for system services and applications to group their
3058 own child processes (worker processes) in a powerful way
3059 which then maybe used to organize them, or kill them
3060 together, or apply resource limits on them.
3062 Slice units may be used to partition system resources in an
3063 hierarchical fashion and then assign other units to them. By
3064 default there are now three slices: system.slice (for all
3065 system services), user.slice (for all user sessions),
3066 machine.slice (for VMs and containers).
3068 Slices and scopes have been introduced primarily in
3069 context of the work to move cgroup handling to a
3070 single-writer scheme, where only PID 1
3071 creates/removes/manages cgroups.
3073 * There's a new concept of "transient" units. In contrast to
3074 normal units these units are created via an API at runtime,
3075 not from configuration from disk. More specifically this
3076 means it is now possible to run arbitrary programs as
3077 independent services, with all execution parameters passed
3078 in via bus APIs rather than read from disk. Transient units
3079 make systemd substantially more dynamic then it ever was,
3080 and useful as a general batch manager.
3082 * logind has been updated to make use of scope and slice units
3083 for managing user sessions. As a user logs in he will get
3084 his own private slice unit, to which all sessions are added
3085 as scope units. We also added support for automatically
3086 adding an instance of user@.service for the user into the
3087 slice. Effectively logind will no longer create cgroup
3088 hierarchies on its own now, it will defer entirely to PID 1
3089 for this by means of scope, service and slice units. Since
3090 user sessions this way become entities managed by PID 1
3091 the output of "systemctl" is now a lot more comprehensive.
3093 * A new mini-daemon "systemd-machined" has been added which
3094 may be used by virtualization managers to register local
3095 VMs/containers. nspawn has been updated accordingly, and
3096 libvirt will be updated shortly. machined will collect a bit
3097 of meta information about the VMs/containers, and assign
3098 them their own scope unit (see above). The collected
3099 meta-data is then made available via the "machinectl" tool,
3100 and exposed in "ps" and similar tools. machined/machinectl
3101 is compile-time optional.
3103 * As discussed earlier, the low-level cgroup configuration
3104 options ControlGroup=, ControlGroupModify=,
3105 ControlGroupPersistent=, ControlGroupAttribute= have been
3106 removed. Please use high-level attribute settings instead as
3107 well as slice units.
3109 * A new bus call SetUnitProperties() has been added to alter
3110 various runtime parameters of a unit. This is primarily
3111 useful to alter cgroup parameters dynamically in a nice way,
3112 but will be extended later on to make more properties
3113 modifiable at runtime. systemctl gained a new set-properties
3114 command that wraps this call.
3116 * A new tool "systemd-run" has been added which can be used to
3117 run arbitrary command lines as transient services or scopes,
3118 while configuring a number of settings via the command
3119 line. This tool is currently very basic, however already
3120 very useful. We plan to extend this tool to even allow
3121 queuing of execution jobs with time triggers from the
3122 command line, similar in fashion to "at".
3124 * nspawn will now inform the user explicitly that kernels with
3125 audit enabled break containers, and suggest the user to turn
3128 * Support for detecting the IMA and AppArmor security
3129 frameworks with ConditionSecurity= has been added.
3131 * journalctl gained a new "-k" switch for showing only kernel
3132 messages, mimicking dmesg output; in addition to "--user"
3133 and "--system" switches for showing only user's own logs
3136 * systemd-delta can now show information about drop-in
3137 snippets extending unit files.
3139 * libsystemd-bus has been substantially updated but is still
3140 not available as public API.
3142 * systemd will now look for the "debug" argument on the kernel
3143 command line and enable debug logging, similar to what
3144 "systemd.log_level=debug" already did before.
3146 * "systemctl set-default", "systemctl get-default" has been
3147 added to configure the default.target symlink, which
3148 controls what to boot into by default.
3150 * "systemctl set-log-level" has been added as a convenient
3151 way to raise and lower systemd logging threshold.
3153 * "systemd-analyze plot" will now show the time the various
3154 generators needed for execution, as well as information
3155 about the unit file loading.
3157 * libsystemd-journal gained a new sd_journal_open_files() call
3158 for opening specific journal files. journactl also gained a
3159 new switch to expose this new functionality. Previously we
3160 only supported opening all files from a directory, or all
3161 files from the system, as opening individual files only is
3162 racy due to journal file rotation.
3164 * systemd gained the new DefaultEnvironment= setting in
3165 /etc/systemd/system.conf to set environment variables for
3168 * If a privileged process logs a journal message with the
3169 OBJECT_PID= field set, then journald will automatically
3170 augment this with additional OBJECT_UID=, OBJECT_GID=,
3171 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
3172 system services want to log events about specific client
3173 processes. journactl/systemctl has been updated to make use
3174 of this information if all log messages regarding a specific
3177 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
3178 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
3179 Reisner, David Coppa, David King, David Strauss, Eelco
3180 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
3181 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
3182 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
3183 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
3184 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
3185 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
3186 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
3187 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
3188 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
3189 Łukasz Stelmach, 장동준
3193 * The Python bindings gained some minimal support for the APIs
3194 exposed by libsystemd-logind.
3196 * ConditionSecurity= gained support for detecting SMACK. Since
3197 this condition already supports SELinux and AppArmor we only
3198 miss IMA for this. Patches welcome!
3200 Contributions from: Karol Lewandowski, Lennart Poettering,
3201 Zbigniew Jędrzejewski-Szmek
3205 * systemd-nspawn will now create /etc/resolv.conf if
3206 necessary, before bind-mounting the host's file onto it.
3208 * systemd-nspawn will now store meta information about a
3209 container on the container's cgroup as extended attribute
3210 fields, including the root directory.
3212 * The cgroup hierarchy has been reworked in many ways. All
3213 objects any of the components systemd creates in the cgroup
3214 tree are now suffixed. More specifically, user sessions are
3215 now placed in cgroups suffixed with ".session", users in
3216 cgroups suffixed with ".user", and nspawn containers in
3217 cgroups suffixed with ".nspawn". Furthermore, all cgroup
3218 names are now escaped in a simple scheme to avoid collision
3219 of userspace object names with kernel filenames. This work
3220 is preparation for making these objects relocatable in the
3221 cgroup tree, in order to allow easy resource partitioning of
3222 these objects without causing naming conflicts.
3224 * systemctl list-dependencies gained the new switches
3225 --plain, --reverse, --after and --before.
3227 * systemd-inhibit now shows the process name of processes that
3228 have taken an inhibitor lock.
3230 * nss-myhostname will now also resolve "localhost"
3231 implicitly. This makes /etc/hosts an optional file and
3232 nicely handles that on IPv6 ::1 maps to both "localhost" and
3235 * libsystemd-logind.so gained a new call
3236 sd_get_machine_names() to enumerate running containers and
3237 VMs (currently only supported by very new libvirt and
3238 nspawn). sd_login_monitor can now be used to watch
3239 VMs/containers coming and going.
3241 * .include is not allowed recursively anymore, and only in
3242 unit files. Usually it is better to use drop-in snippets in
3243 .d/*.conf anyway, as introduced with systemd 198.
3245 * systemd-analyze gained a new "critical-chain" command that
3246 determines the slowest chain of units run during system
3247 boot-up. It is very useful for tracking down where
3248 optimizing boot time is the most beneficial.
3250 * systemd will no longer allow manipulating service paths in
3251 the name=systemd:/system cgroup tree using ControlGroup= in
3252 units. (But is still fine with it in all other dirs.)
3254 * There's a new systemd-nspawn@.service service file that may
3255 be used to easily run nspawn containers as system
3256 services. With the container's root directory in
3257 /var/lib/container/foobar it is now sufficient to run
3258 "systemctl start systemd-nspawn@foobar.service" to boot it.
3260 * systemd-cgls gained a new parameter "--machine" to list only
3261 the processes within a certain container.
3263 * ConditionSecurity= now can check for "apparmor". We still
3264 are lacking checks for SMACK and IMA for this condition
3265 check though. Patches welcome!
3267 * A new configuration file /etc/systemd/sleep.conf has been
3268 added that may be used to configure which kernel operation
3269 systemd is supposed to execute when "suspend", "hibernate"
3270 or "hybrid-sleep" is requested. This makes the new kernel
3271 "freeze" state accessible to the user.
3273 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
3274 the passed argument if applicable.
3276 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
3277 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
3278 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
3279 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
3280 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
3281 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
3282 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
3287 * The output of 'systemctl list-jobs' got some polishing. The
3288 '--type=' argument may now be passed more than once. A new
3289 command 'systemctl list-sockets' has been added which shows
3290 a list of kernel sockets systemd is listening on with the
3291 socket units they belong to, plus the units these socket
3294 * The experimental libsystemd-bus library got substantial
3295 updates to work in conjunction with the (also experimental)
3296 kdbus kernel project. It works well enough to exchange
3297 messages with some sophistication. Note that kdbus is not
3298 ready yet, and the library is mostly an elaborate test case
3299 for now, and not installable.
3301 * systemd gained a new unit 'systemd-static-nodes.service'
3302 that generates static device nodes earlier during boot, and
3303 can run in conjunction with udev.
3305 * libsystemd-login gained a new call sd_pid_get_user_unit()
3306 to retrieve the user systemd unit a process is running
3307 in. This is useful for systems where systemd is used as
3310 * systemd-nspawn now places all containers in the new /machine
3311 top-level cgroup directory in the name=systemd
3312 hierarchy. libvirt will soon do the same, so that we get a
3313 uniform separation of /system, /user and /machine for system
3314 services, user processes and containers/virtual
3315 machines. This new cgroup hierarchy is also useful to stick
3316 stable names to specific container instances, which can be
3317 recognized later this way (this name may be controlled
3318 via systemd-nspawn's new -M switch). libsystemd-login also
3319 gained a new call sd_pid_get_machine_name() to retrieve the
3320 name of the container/VM a specific process belongs to.
3322 * bootchart can now store its data in the journal.
3324 * libsystemd-journal gained a new call
3325 sd_journal_add_conjunction() for AND expressions to the
3326 matching logic. This can be used to express more complex
3327 logical expressions.
3329 * journactl can now take multiple --unit= and --user-unit=
3332 * The cryptsetup logic now understands the "luks.key=" kernel
3333 command line switch for specifying a file to read the
3334 decryption key from. Also, if a configured key file is not
3335 found the tool will now automatically fall back to prompting
3338 * Python systemd.journal module was updated to wrap recently
3339 added functions from libsystemd-journal. The interface was
3340 changed to bring the low level interface in s.j._Reader
3341 closer to the C API, and the high level interface in
3342 s.j.Reader was updated to wrap and convert all data about
3345 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
3346 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
3347 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
3348 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
3349 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
3350 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3354 * journalctl --update-catalog now understands a new --root=
3355 option to operate on catalogs found in a different root
3358 * During shutdown after systemd has terminated all running
3359 services a final killing loop kills all remaining left-over
3360 processes. We will now print the name of these processes
3361 when we send SIGKILL to them, since this usually indicates a
3364 * If /etc/crypttab refers to password files stored on
3365 configured mount points automatic dependencies will now be
3366 generated to ensure the specific mount is established first
3367 before the key file is attempted to be read.
3369 * 'systemctl status' will now show information about the
3370 network sockets a socket unit is listening on.
3372 * 'systemctl status' will also shown information about any
3373 drop-in configuration file for units. (Drop-In configuration
3374 files in this context are files such as
3375 /etc/systemd/systemd/foobar.service.d/*.conf)
3377 * systemd-cgtop now optionally shows summed up CPU times of
3378 cgroups. Press '%' while running cgtop to switch between
3379 percentage and absolute mode. This is useful to determine
3380 which cgroups use up the most CPU time over the entire
3381 runtime of the system. systemd-cgtop has also been updated
3382 to be 'pipeable' for processing with further shell tools.
3384 * 'hostnamectl set-hostname' will now allow setting of FQDN
3387 * The formatting and parsing of time span values has been
3388 changed. The parser now understands fractional expressions
3389 such as "5.5h". The formatter will now output fractional
3390 expressions for all time spans under 1min, i.e. "5.123456s"
3391 rather than "5s 123ms 456us". For time spans under 1s
3392 millisecond values are shown, for those under 1ms
3393 microsecond values are shown. This should greatly improve
3394 all time-related output of systemd.
3396 * libsystemd-login and libsystemd-journal gained new
3397 functions for querying the poll() events mask and poll()
3398 timeout value for integration into arbitrary event
3401 * localectl gained the ability to list available X11 keymaps
3402 (models, layouts, variants, options).
3404 * 'systemd-analyze dot' gained the ability to filter for
3405 specific units via shell-style globs, to create smaller,
3406 more useful graphs. I.e. it is now possible to create simple
3407 graphs of all the dependencies between only target units, or
3408 of all units that Avahi has dependencies with.
3410 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
3411 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
3412 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
3413 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
3414 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
3415 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
3416 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
3420 * The boot-time readahead implementation for rotating media
3421 will now read the read-ahead data in multiple passes which
3422 consist of all read requests made in equidistant time
3423 intervals. This means instead of strictly reading read-ahead
3424 data in its physical order on disk we now try to find a
3425 middle ground between physical and access time order.
3427 * /etc/os-release files gained a new BUILD_ID= field for usage
3428 on operating systems that provide continuous builds of OS
3431 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
3432 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
3433 William Douglas, Zbigniew Jędrzejewski-Szmek
3437 * systemd-python gained an API exposing libsystemd-daemon.
3439 * The SMACK setup logic gained support for uploading CIPSO
3442 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
3443 ReadOnlyDirectories= and InaccessibleDirectories= has
3444 changed. The private /tmp and /var/tmp directories are now
3445 shared by all processes of a service (which means
3446 ExecStartPre= may now leave data in /tmp that ExecStart= of
3447 the same service can still access). When a service is
3448 stopped its temporary directories are immediately deleted
3449 (normal clean-up with tmpfiles is still done in addition to
3452 * By default, systemd will now set a couple of sysctl
3453 variables in the kernel: the safe sysrq options are turned
3454 on, IP route verification is turned on, and source routing
3455 disabled. The recently added hardlink and softlink
3456 protection of the kernel is turned on. These settings should
3457 be reasonably safe, and good defaults for all new systems.
3459 * The predictable network naming logic may now be turned off
3460 with a new kernel command line switch: net.ifnames=0.
3462 * A new libsystemd-bus module has been added that implements a
3463 pretty complete D-Bus client library. For details see:
3465 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
3467 * journald will now explicitly flush the journal files to disk
3468 at the latest 5min after each write. The file will then also
3469 be marked offline until the next write. This should increase
3470 reliability in case of a crash. The synchronization delay
3471 can be configured via SyncIntervalSec= in journald.conf.
3473 * There's a new remote-fs-setup.target unit that can be used
3474 to pull in specific services when at least one remote file
3475 system is to be mounted.
3477 * There are new targets timers.target and paths.target as
3478 canonical targets to pull user timer and path units in
3479 from. This complements sockets.target with a similar
3480 purpose for socket units.
3482 * libudev gained a new call udev_device_set_attribute_value()
3483 to set sysfs attributes of a device.
3485 * The udev daemon now sets the default number of worker
3486 processes executed in parallel based on the number of available
3487 CPUs instead of the amount of available RAM. This is supposed
3488 to provide a more reliable default and limit a too aggressive
3489 paralellism for setups with 1000s of devices connected.
3491 Contributions from: Auke Kok, Colin Walters, Cristian
3492 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
3493 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
3494 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
3495 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
3496 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
3497 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
3498 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
3499 Zbigniew Jędrzejewski-Szmek
3503 * Configuration of unit files may now be extended via drop-in
3504 files without having to edit/override the unit files
3505 themselves. More specifically, if the administrator wants to
3506 change one value for a service file foobar.service he can
3507 now do so by dropping in a configuration snippet into
3508 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
3509 will load all these snippets and apply them on top of the
3510 main unit configuration file, possibly extending or
3511 overriding its settings. Using these drop-in snippets is
3512 generally nicer than the two earlier options for changing
3513 unit files locally: copying the files from
3514 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
3515 them there; or creating a new file in /etc/systemd/system/
3516 that incorporates the original one via ".include". Drop-in
3517 snippets into these .d/ directories can be placed in any
3518 directory systemd looks for units in, and the usual
3519 overriding semantics between /usr/lib, /etc and /run apply
3522 * Most unit file settings which take lists of items can now be
3523 reset by assigning the empty string to them. For example,
3524 normally, settings such as Environment=FOO=BAR append a new
3525 environment variable assignment to the environment block,
3526 each time they are used. By assigning Environment= the empty
3527 string the environment block can be reset to empty. This is
3528 particularly useful with the .d/*.conf drop-in snippets
3529 mentioned above, since this adds the ability to reset list
3530 settings from vendor unit files via these drop-ins.
3532 * systemctl gained a new "list-dependencies" command for
3533 listing the dependencies of a unit recursively.
3535 * Inhibitors are now honored and listed by "systemctl
3536 suspend", "systemctl poweroff" (and similar) too, not only
3537 GNOME. These commands will also list active sessions by
3540 * Resource limits (as exposed by the various control group
3541 controllers) can now be controlled dynamically at runtime
3542 for all units. More specifically, you can now use a command
3543 like "systemctl set-cgroup-attr foobar.service cpu.shares
3544 2000" to alter the CPU shares a specific service gets. These
3545 settings are stored persistently on disk, and thus allow the
3546 administrator to easily adjust the resource usage of
3547 services with a few simple commands. This dynamic resource
3548 management logic is also available to other programs via the
3549 bus. Almost any kernel cgroup attribute and controller is
3552 * systemd-vconsole-setup will now copy all font settings to
3553 all allocated VTs, where it previously applied them only to
3556 * libsystemd-login gained the new sd_session_get_tty() API
3559 * This release drops support for a few legacy or
3560 distribution-specific LSB facility names when parsing init
3561 scripts: $x-display-manager, $mail-transfer-agent,
3562 $mail-transport-agent, $mail-transfer-agent, $smtp,
3563 $null. Also, the mail-transfer-agent.target unit backing
3564 this has been removed. Distributions which want to retain
3565 compatibility with this should carry the burden for
3566 supporting this themselves and patch support for these back
3567 in, if they really need to. Also, the facilities $syslog and
3568 $local_fs are now ignored, since systemd does not support
3569 early-boot LSB init scripts anymore, and these facilities
3570 are implied anyway for normal services. syslog.target has
3573 * There are new bus calls on PID1's Manager object for
3574 cancelling jobs, and removing snapshot units. Previously,
3575 both calls were only available on the Job and Snapshot
3578 * systemd-journal-gatewayd gained SSL support.
3580 * The various "environment" files, such as /etc/locale.conf
3581 now support continuation lines with a backslash ("\") as
3582 last character in the line, similarly in style (but different)
3583 to how this is supported in shells.
3585 * For normal user processes the _SYSTEMD_USER_UNIT= field is
3586 now implicitly appended to every log entry logged. systemctl
3587 has been updated to filter by this field when operating on a
3588 user systemd instance.
3590 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
3591 CAP_AUDIT_CONTROL capabilities to the capabilities set for
3592 the container. This makes it easier to boot unmodified
3593 Fedora systems in a container, which however still requires
3594 audit=0 to be passed on the kernel command line. Auditing in
3595 kernel and userspace is unfortunately still too broken in
3596 context of containers, hence we recommend compiling it out
3597 of the kernel or using audit=0. Hopefully this will be fixed
3598 one day for good in the kernel.
3600 * nspawn gained the new --bind= and --bind-ro= parameters to
3601 bind mount specific directories from the host into the
3604 * nspawn will now mount its own devpts file system instance
3605 into the container, in order not to leak pty devices from
3606 the host into the container.
3608 * systemd will now read the firmware boot time performance
3609 information from the EFI variables, if the used boot loader
3610 supports this, and takes it into account for boot performance
3611 analysis via "systemd-analyze". This is currently supported
3612 only in conjunction with Gummiboot, but could be supported
3613 by other boot loaders too. For details see:
3615 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
3617 * A new generator has been added that automatically mounts the
3618 EFI System Partition (ESP) to /boot, if that directory
3619 exists, is empty, and no other file system has been
3620 configured to be mounted there.
3622 * logind will now send out PrepareForSleep(false) out
3623 unconditionally, after coming back from suspend. This may be
3624 used by applications as asynchronous notification for
3625 system resume events.
3627 * "systemctl unlock-sessions" has been added, that allows
3628 unlocking the screens of all user sessions at once, similar
3629 to how "systemctl lock-sessions" already locked all users
3630 sessions. This is backed by a new D-Bus call UnlockSessions().
3632 * "loginctl seat-status" will now show the master device of a
3633 seat. (i.e. the device of a seat that needs to be around for
3634 the seat to be considered available, usually the graphics
3637 * tmpfiles gained a new "X" line type, that allows
3638 configuration of files and directories (with wildcards) that
3639 shall be excluded from automatic cleanup ("aging").
3641 * udev default rules set the device node permissions now only
3642 at "add" events, and do not change them any longer with a
3643 later "change" event.
3645 * The log messages for lid events and power/sleep keypresses
3646 now carry a message ID.
3648 * We now have a substantially larger unit test suite, but this
3649 continues to be work in progress.
3651 * udevadm hwdb gained a new --root= parameter to change the
3652 root directory to operate relative to.
3654 * logind will now issue a background sync() request to the kernel
3655 early at shutdown, so that dirty buffers are flushed to disk early
3656 instead of at the last moment, in order to optimize shutdown
3659 * A new bootctl tool has been added that is an interface for
3660 certain boot loader operations. This is currently a preview
3661 and is likely to be extended into a small mechanism daemon
3662 like timedated, localed, hostnamed, and can be used by
3663 graphical UIs to enumerate available boot options, and
3664 request boot into firmware operations.
3666 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
3667 the rest of the package. It also has been updated to work
3668 correctly in initrds.
3670 * Policykit previously has been runtime optional, and is now
3671 also compile time optional via a configure switch.
3673 * systemd-analyze has been reimplemented in C. Also "systemctl
3674 dot" has moved into systemd-analyze.
3676 * "systemctl status" with no further parameters will now print
3677 the status of all active or failed units.
3679 * Operations such as "systemctl start" can now be executed
3680 with a new mode "--irreversible" which may be used to queue
3681 operations that cannot accidentally be reversed by a later
3682 job queuing. This is by default used to make shutdown
3683 requests more robust.
3685 * The Python API of systemd now gained a new module for
3686 reading journal files.
3688 * A new tool kernel-install has been added that can install
3689 kernel images according to the Boot Loader Specification:
3691 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
3693 * Boot time console output has been improved to provide
3694 animated boot time output for hanging jobs.
3696 * A new tool systemd-activate has been added which can be used
3697 to test socket activation with, directly from the command
3698 line. This should make it much easier to test and debug
3699 socket activation in daemons.
3701 * journalctl gained a new "--reverse" (or -r) option to show
3702 journal output in reverse order (i.e. newest line first).
3704 * journalctl gained a new "--pager-end" (or -e) option to jump
3705 to immediately jump to the end of the journal in the
3706 pager. This is only supported in conjunction with "less".
3708 * journalctl gained a new "--user-unit=" option, that works
3709 similarly to "--unit=" but filters for user units rather than
3712 * A number of unit files to ease adoption of systemd in
3713 initrds has been added. This moves some minimal logic from
3714 the various initrd implementations into systemd proper.
3716 * The journal files are now owned by a new group
3717 "systemd-journal", which exists specifically to allow access
3718 to the journal, and nothing else. Previously, we used the
3719 "adm" group for that, which however possibly covers more
3720 than just journal/log file access. This new group is now
3721 already used by systemd-journal-gatewayd to ensure this
3722 daemon gets access to the journal files and as little else
3723 as possible. Note that "make install" will also set FS ACLs
3724 up for /var/log/journal to give "adm" and "wheel" read
3725 access to it, in addition to "systemd-journal" which owns
3726 the journal files. We recommend that packaging scripts also
3727 add read access to "adm" + "wheel" to /var/log/journal, and
3728 all existing/future journal files. To normal users and
3729 administrators little changes, however packagers need to
3730 ensure to create the "systemd-journal" system group at
3731 package installation time.
3733 * The systemd-journal-gatewayd now runs as unprivileged user
3734 systemd-journal-gateway:systemd-journal-gateway. Packaging
3735 scripts need to create these system user/group at
3738 * timedated now exposes a new boolean property CanNTP that
3739 indicates whether a local NTP service is available or not.
3741 * systemd-detect-virt will now also detect xen PVs
3743 * The pstore file system is now mounted by default, if it is
3746 * In addition to the SELinux and IMA policies we will now also
3747 load SMACK policies at early boot.
3749 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
3750 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
3751 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
3752 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
3753 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
3754 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
3755 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
3756 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
3757 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
3758 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
3759 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
3760 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
3761 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
3762 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
3766 * Timer units now support calendar time events in addition to
3767 monotonic time events. That means you can now trigger a unit
3768 based on a calendar time specification such as "Thu,Fri
3769 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
3770 or fifth day of any month of the year 2013, given that it is
3771 a thursday or friday. This brings timer event support
3772 considerably closer to cron's capabilities. For details on
3773 the supported calendar time specification language see
3776 * udev now supports a number of different naming policies for
3777 network interfaces for predictable names, and a combination
3778 of these policies is now the default. Please see this wiki
3779 document for details:
3781 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
3783 * Auke Kok's bootchart implementation has been added to the
3784 systemd tree. It is an optional component that can graph the
3785 boot in quite some detail. It is one of the best bootchart
3786 implementations around and minimal in its code and
3789 * nss-myhostname has been integrated into the systemd source
3790 tree. nss-myhostname guarantees that the local hostname
3791 always stays resolvable via NSS. It has been a weak
3792 requirement of systemd-hostnamed since a long time, and
3793 since its code is actually trivial we decided to just
3794 include it in systemd's source tree. It can be turned off
3795 with a configure switch.
3797 * The read-ahead logic is now capable of properly detecting
3798 whether a btrfs file system is on SSD or rotating media, in
3799 order to optimize the read-ahead scheme. Previously, it was
3800 only capable of detecting this on traditional file systems
3803 * In udev, additional device properties are now read from the
3804 IAB in addition to the OUI database. Also, Bluetooth company
3805 identities are attached to the devices as well.
3807 * In service files %U may be used as specifier that is
3808 replaced by the configured user name of the service.
3810 * nspawn may now be invoked without a controlling TTY. This
3811 makes it suitable for invocation as its own service. This
3812 may be used to set up a simple containerized server system
3813 using only core OS tools.
3815 * systemd and nspawn can now accept socket file descriptors
3816 when they are started for socket activation. This enables
3817 implementation of socket activated nspawn
3818 containers. i.e. think about autospawning an entire OS image
3819 when the first SSH or HTTP connection is received. We expect
3820 that similar functionality will also be added to libvirt-lxc
3823 * journalctl will now suppress ANSI color codes when
3824 presenting log data.
3826 * systemctl will no longer show control group information for
3827 a unit if a the control group is empty anyway.
3829 * logind can now automatically suspend/hibernate/shutdown the
3832 * /etc/machine-info and hostnamed now also expose the chassis
3833 type of the system. This can be used to determine whether
3834 the local system is a laptop, desktop, handset or
3835 tablet. This information may either be configured by the
3836 user/vendor or is automatically determined from ACPI and DMI
3837 information if possible.
3839 * A number of PolicyKit actions are now bound together with
3840 "imply" rules. This should simplify creating UIs because
3841 many actions will now authenticate similar ones as well.
3843 * Unit files learnt a new condition ConditionACPower= which
3844 may be used to conditionalize a unit depending on whether an
3845 AC power source is connected or not, of whether the system
3846 is running on battery power.
3848 * systemctl gained a new "is-failed" verb that may be used in
3849 shell scripts and suchlike to check whether a specific unit
3850 is in the "failed" state.
3852 * The EnvironmentFile= setting in unit files now supports file
3853 globbing, and can hence be used to easily read a number of
3854 environment files at once.
3856 * systemd will no longer detect and recognize specific
3857 distributions. All distribution-specific #ifdeffery has been
3858 removed, systemd is now fully generic and
3859 distribution-agnostic. Effectively, not too much is lost as
3860 a lot of the code is still accessible via explicit configure
3861 switches. However, support for some distribution specific
3862 legacy configuration file formats has been dropped. We
3863 recommend distributions to simply adopt the configuration
3864 files everybody else uses now and convert the old
3865 configuration from packaging scripts. Most distributions
3866 already did that. If that's not possible or desirable,
3867 distributions are welcome to forward port the specific
3868 pieces of code locally from the git history.
3870 * When logging a message about a unit systemd will now always
3871 log the unit name in the message meta data.
3873 * localectl will now also discover system locale data that is
3874 not stored in locale archives, but directly unpacked.
3876 * logind will no longer unconditionally use framebuffer
3877 devices as seat masters, i.e. as devices that are required
3878 to be existing before a seat is considered preset. Instead,
3879 it will now look for all devices that are tagged as
3880 "seat-master" in udev. By default framebuffer devices will
3881 be marked as such, but depending on local systems other
3882 devices might be marked as well. This may be used to
3883 integrate graphics cards using closed source drivers (such
3884 as NVidia ones) more nicely into logind. Note however, that
3885 we recommend using the open source NVidia drivers instead,
3886 and no udev rules for the closed-source drivers will be
3887 shipped from us upstream.
3889 Contributions from: Adam Williamson, Alessandro Crismani, Auke
3890 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
3891 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
3892 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
3893 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
3894 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
3895 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
3896 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
3897 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
3898 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
3899 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
3900 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
3905 * udev gained support for loading additional device properties
3906 from an indexed database that is keyed by vendor/product IDs
3907 and similar device identifiers. For the beginning this
3908 "hwdb" is populated with data from the well-known PCI and
3909 USB database, but also includes PNP, ACPI and OID data. In
3910 the longer run this indexed database shall grow into
3911 becoming the one central database for non-essential
3912 userspace device metadata. Previously, data from the PCI/USB
3913 database was only attached to select devices, since the
3914 lookup was a relatively expensive operation due to O(n) time
3915 complexity (with n being the number of entries in the
3916 database). Since this is now O(1), we decided to add in this
3917 data for all devices where this is available, by
3918 default. Note that the indexed database needs to be rebuilt
3919 when new data files are installed. To achieve this you need
3920 to update your packaging scripts to invoke "udevadm hwdb
3921 --update" after installation of hwdb data files. For
3922 RPM-based distributions we introduced the new
3923 %udev_hwdb_update macro for this purpose.
3925 * The Journal gained support for the "Message Catalog", an
3926 indexed database to link up additional information with
3927 journal entries. For further details please check:
3929 http://www.freedesktop.org/wiki/Software/systemd/catalog
3931 The indexed message catalog database also needs to be
3932 rebuilt after installation of message catalog files. Use
3933 "journalctl --update-catalog" for this. For RPM-based
3934 distributions we introduced the %journal_catalog_update
3935 macro for this purpose.
3937 * The Python Journal bindings gained support for the standard
3938 Python logging framework.
3940 * The Journal API gained new functions for checking whether
3941 the underlying file system of a journal file is capable of
3942 properly reporting file change notifications, or whether
3943 applications that want to reflect journal changes "live"
3944 need to recheck journal files continuously in appropriate
3947 * It is now possible to set the "age" field for tmpfiles
3948 entries to 0, indicating that files matching this entry
3949 shall always be removed when the directories are cleaned up.
3951 * coredumpctl gained a new "gdb" verb which invokes gdb
3952 right-away on the selected coredump.
3954 * There's now support for "hybrid sleep" on kernels that
3955 support this, in addition to "suspend" and "hibernate". Use
3956 "systemctl hybrid-sleep" to make use of this.
3958 * logind's HandleSuspendKey= setting (and related settings)
3959 now gained support for a new "lock" setting to simply
3960 request the screen lock on all local sessions, instead of
3961 actually executing a suspend or hibernation.
3963 * systemd will now mount the EFI variables file system by
3966 * Socket units now gained support for configuration of the
3967 SMACK security label.
3969 * timedatectl will now output the time of the last and next
3970 daylight saving change.
3972 * We dropped support for various legacy and distro-specific
3973 concepts, such as insserv, early-boot SysV services
3974 (i.e. those for non-standard runlevels such as 'b' or 'S')
3975 or ArchLinux /etc/rc.conf support. We recommend the
3976 distributions who still need support this to either continue
3977 to maintain the necessary patches downstream, or find a
3978 different solution. (Talk to us if you have questions!)
3980 * Various systemd components will now bypass PolicyKit checks
3981 for root and otherwise handle properly if PolicyKit is not
3982 found to be around. This should fix most issues for
3983 PolicyKit-less systems. Quite frankly this should have been
3984 this way since day one. It is absolutely our intention to
3985 make systemd work fine on PolicyKit-less systems, and we
3986 consider it a bug if something does not work as it should if
3987 PolicyKit is not around.
3989 * For embedded systems it is now possible to build udev and
3990 systemd without blkid and/or kmod support.
3992 * "systemctl switch-root" is now capable of switching root
3993 more than once. I.e. in addition to transitions from the
3994 initrd to the host OS it is now possible to transition to
3995 further OS images from the host. This is useful to implement
3996 offline updating tools.
3998 * Various other additions have been made to the RPM macros
3999 shipped with systemd. Use %udev_rules_update() after
4000 installing new udev rules files. %_udevhwdbdir,
4001 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
4002 %_sysctldir are now available which resolve to the right
4003 directories for packages to place various data files in.
4005 * journalctl gained the new --full switch (in addition to
4006 --all, to disable ellipsation for long messages.
4008 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
4009 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
4010 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
4011 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
4012 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
4013 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
4014 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
4015 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
4016 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
4020 * journalctl gained new --since= and --until= switches to
4021 filter by time. It also now supports nice filtering for
4022 units via --unit=/-u.
4024 * Type=oneshot services may use ExecReload= and do the
4027 * The journal daemon now supports time-based rotation and
4028 vacuuming, in addition to the usual disk-space based
4031 * The journal will now index the available field values for
4032 each field name. This enables clients to show pretty drop
4033 downs of available match values when filtering. The bash
4034 completion of journalctl has been updated
4035 accordingly. journalctl gained a new switch -F to list all
4036 values a certain field takes in the journal database.
4038 * More service events are now written as structured messages
4039 to the journal, and made recognizable via message IDs.
4041 * The timedated, localed and hostnamed mini-services which
4042 previously only provided support for changing time, locale
4043 and hostname settings from graphical DEs such as GNOME now
4044 also have a minimal (but very useful) text-based client
4045 utility each. This is probably the nicest way to changing
4046 these settings from the command line now, especially since
4047 it lists available options and is fully integrated with bash
4050 * There's now a new tool "systemd-coredumpctl" to list and
4051 extract coredumps from the journal.
4053 * We now install a README each in /var/log/ and
4054 /etc/rc.d/init.d explaining where the system logs and init
4055 scripts went. This hopefully should help folks who go to
4056 that dirs and look into the otherwise now empty void and
4057 scratch their heads.
4059 * When user-services are invoked (by systemd --user) the
4060 $MANAGERPID env var is set to the PID of systemd.
4062 * SIGRTMIN+24 when sent to a --user instance will now result
4063 in immediate termination of systemd.
4065 * gatewayd received numerous feature additions such as a
4066 "follow" mode, for live syncing and filtering.
4068 * browse.html now allows filtering and showing detailed
4069 information on specific entries. Keyboard navigation and
4070 mouse screen support has been added.
4072 * gatewayd/journalctl now supports HTML5/JSON
4073 Server-Sent-Events as output.
4075 * The SysV init script compatibility logic will now
4076 heuristically determine whether a script supports the
4077 "reload" verb, and only then make this available as
4080 * "systemctl status --follow" has been removed, use "journalctl
4083 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
4084 have been removed since they are hardly useful to be
4087 * And I'd like to take the opportunity to specifically mention
4088 Zbigniew for his great contributions. Zbigniew, you rock!
4090 Contributions from: Andrew Eikum, Christian Hesse, Colin
4091 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
4092 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
4093 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
4094 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
4095 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
4096 Jędrzejewski-Szmek, Сковорода Никита Андреевич
4100 * If /etc/vconsole.conf is non-existent or empty we will no
4101 longer load any console font or key map at boot by
4102 default. Instead the kernel defaults will be left
4103 intact. This is definitely the right thing to do, as no
4104 configuration should mean no configuration, and hard-coding
4105 font names that are different on all archs is probably a bad
4106 idea. Also, the kernel default key map and font should be
4107 good enough for most cases anyway, and mostly identical to
4108 the userspace fonts/key maps we previously overloaded them
4109 with. If distributions want to continue to default to a
4110 non-kernel font or key map they should ship a default
4111 /etc/vconsole.conf with the appropriate contents.
4113 Contributions from: Colin Walters, Daniel J Walsh, Dave
4114 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
4115 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
4119 * journalctl gained a new --cursor= switch to show entries
4120 starting from the specified location in the journal.
4122 * We now enforce a size limit on journal entry fields exported
4123 with "-o json" in journalctl. Fields larger than 4K will be
4124 assigned null. This can be turned off with --all.
4126 * An (optional) journal gateway daemon is now available as
4127 "systemd-journal-gatewayd.service". This service provides
4128 access to the journal via HTTP and JSON. This functionality
4129 will be used to implement live log synchronization in both
4130 pull and push modes, but has various other users too, such
4131 as easy log access for debugging of embedded devices. Right
4132 now it is already useful to retrieve the journal via HTTP:
4134 # systemctl start systemd-journal-gatewayd.service
4135 # wget http://localhost:19531/entries
4137 This will download the journal contents in a
4138 /var/log/messages compatible format. The same as JSON:
4140 # curl -H"Accept: application/json" http://localhost:19531/entries
4142 This service is also accessible via a web browser where a
4143 single static HTML5 app is served that uses the JSON logic
4144 to enable the user to do some basic browsing of the
4145 journal. This will be extended later on. Here's an example
4146 screenshot of this app in its current state:
4148 http://0pointer.de/public/journal-gatewayd
4150 Contributions from: Kay Sievers, Lennart Poettering, Robert
4151 Milasan, Tom Gundersen
4155 * The bash completion logic is now available for journalctl
4158 * We do not mount the "cpuset" controller anymore together with
4159 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
4160 started if no parameters are assigned to it. "cpuset" hence
4161 broke code that assumed it it could create "cpu" groups and
4164 * journalctl -f will now subscribe to terminal size changes,
4165 and line break accordingly.
4167 Contributions from: Dave Reisner, Kay Sievers, Lennart
4168 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
4172 * nspawn will now create a symlink /etc/localtime in the
4173 container environment, copying the host's timezone
4174 setting. Previously this has been done via a bind mount, but
4175 since symlinks cannot be bind mounted this has now been
4176 changed to create/update the appropriate symlink.
4178 * journalctl -n's line number argument is now optional, and
4179 will default to 10 if omitted.
4181 * journald will now log the maximum size the journal files may
4182 take up on disk. This is particularly useful if the default
4183 built-in logic of determining this parameter from the file
4184 system size is used. Use "systemctl status
4185 systemd-journald.service" to see this information.
4187 * The multi-seat X wrapper tool has been stripped down. As X
4188 is now capable of enumerating graphics devices via udev in a
4189 seat-aware way the wrapper is not strictly necessary
4190 anymore. A stripped down temporary stop-gap is still shipped
4191 until the upstream display managers have been updated to
4192 fully support the new X logic. Expect this wrapper to be
4193 removed entirely in one of the next releases.
4195 * HandleSleepKey= in logind.conf has been split up into
4196 HandleSuspendKey= and HandleHibernateKey=. The old setting
4197 is not available anymore. X11 and the kernel are
4198 distinguishing between these keys and we should too. This
4199 also means the inhibition lock for these keys has been split
4202 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
4203 Poettering, Lukas Nykryn, Václav Pavlín
4207 * Whenever a unit changes state we will now log this to the
4208 journal and show along the unit's own log output in
4211 * ConditionPathIsMountPoint= can now properly detect bind
4212 mount points too. (Previously, a bind mount of one file
4213 system to another place in the same file system could not be
4214 detected as mount, since they shared struct stat's st_dev
4217 * We will now mount the cgroup controllers cpu, cpuacct,
4218 cpuset and the controllers net_cls, net_prio together by
4221 * nspawn containers will now have a virtualized boot
4222 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
4223 over with a randomized ID at container initialization). This
4224 has the effect of making "journalctl -b" do the right thing
4227 * The JSON output journal serialization has been updated not
4228 to generate "endless" list objects anymore, but rather one
4229 JSON object per line. This is more in line how most JSON
4230 parsers expect JSON objects. The new output mode
4231 "json-pretty" has been added to provide similar output, but
4232 neatly aligned for readability by humans.
4234 * We dropped all explicit sync() invocations in the shutdown
4235 code. The kernel does this implicitly anyway in the kernel
4236 reboot() syscall. halt(8)'s -n option is now a compatibility
4239 * We now support virtualized reboot() in containers, as
4240 supported by newer kernels. We will fall back to exit() if
4241 CAP_SYS_REBOOT is not available to the container. Also,
4242 nspawn makes use of this now and will actually reboot the
4243 container if the containerized OS asks for that.
4245 * journalctl will only show local log output by default
4246 now. Use --merge (-m) to show remote log output, too.
4248 * libsystemd-journal gained the new sd_journal_get_usage()
4249 call to determine the current disk usage of all journal
4250 files. This is exposed in the new "journalctl --disk-usage"
4253 * journald gained a new configuration setting SplitMode= in
4254 journald.conf which may be used to control how user journals
4255 are split off. See journald.conf(5) for details.
4257 * A new condition type ConditionFileNotEmpty= has been added.
4259 * tmpfiles' "w" lines now support file globbing, to write
4260 multiple files at once.
4262 * We added Python bindings for the journal submission
4263 APIs. More Python APIs for a number of selected APIs will
4264 likely follow. Note that we intend to add native bindings
4265 only for the Python language, as we consider it common
4266 enough to deserve bindings shipped within systemd. There are
4267 various projects outside of systemd that provide bindings
4268 for languages such as PHP or Lua.
4270 * Many conditions will now resolve specifiers such as %i. In
4271 addition, PathChanged= and related directives of .path units
4272 now support specifiers as well.
4274 * There's now a new RPM macro definition for the system preset
4277 * journald will now warn if it ca not forward a message to the
4278 syslog daemon because its socket is full.
4280 * timedated will no longer write or process /etc/timezone,
4281 except on Debian. As we do not support late mounted /usr
4282 anymore /etc/localtime always being a symlink is now safe,
4283 and hence the information in /etc/timezone is not necessary
4286 * logind will now always reserve one VT for a text getty (VT6
4287 by default). Previously if more than 6 X sessions where
4288 started they took up all the VTs with auto-spawned gettys,
4289 so that no text gettys were available anymore.
4291 * udev will now automatically inform the btrfs kernel logic
4292 about btrfs RAID components showing up. This should make
4293 simple hotplug based btrfs RAID assembly work.
4295 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
4296 (but not for its children which will stay at the kernel
4297 default). This should allow setups with a lot more listening
4300 * systemd will now always pass the configured timezone to the
4301 kernel at boot. timedated will do the same when the timezone
4304 * logind's inhibition logic has been updated. By default,
4305 logind will now handle the lid switch, the power and sleep
4306 keys all the time, even in graphical sessions. If DEs want
4307 to handle these events on their own they should take the new
4308 handle-power-key, handle-sleep-key and handle-lid-switch
4309 inhibitors during their runtime. A simple way to achieve
4310 that is to invoke the DE wrapped in an invocation of:
4312 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
4314 * Access to unit operations is now checked via SELinux taking
4315 the unit file label and client process label into account.
4317 * systemd will now notify the administrator in the journal
4318 when he over-mounts a non-empty directory.
4320 * There are new specifiers that are resolved in unit files,
4321 for the host name (%H), the machine ID (%m) and the boot ID
4324 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
4325 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
4326 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
4327 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
4328 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
4329 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
4330 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
4334 * Support for reading structured kernel messages from
4335 /dev/kmsg has now been added and is enabled by default.
4337 * Support for reading kernel messages from /proc/kmsg has now
4338 been removed. If you want kernel messages in the journal
4339 make sure to run a recent kernel (>= 3.5) that supports
4340 reading structured messages from /dev/kmsg (see
4341 above). /proc/kmsg is now exclusive property of classic
4342 syslog daemons again.
4344 * The libudev API gained the new
4345 udev_device_new_from_device_id() call.
4347 * The logic for file system namespace (ReadOnlyDirectory=,
4348 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
4349 require pivot_root() anymore. This means fewer temporary
4350 directories are created below /tmp for this feature.
4352 * nspawn containers will now see and receive all submounts
4353 made on the host OS below the root file system of the
4356 * Forward Secure Sealing is now supported for Journal files,
4357 which provide cryptographical sealing of journal files so
4358 that attackers cannot alter log history anymore without this
4359 being detectable. Lennart will soon post a blog story about
4360 this explaining it in more detail.
4362 * There are two new service settings RestartPreventExitStatus=
4363 and SuccessExitStatus= which allow configuration of exit
4364 status (exit code or signal) which will be excepted from the
4365 restart logic, resp. consider successful.
4367 * journalctl gained the new --verify switch that can be used
4368 to check the integrity of the structure of journal files and
4369 (if Forward Secure Sealing is enabled) the contents of
4372 * nspawn containers will now be run with /dev/stdin, /dev/fd/
4373 and similar symlinks pre-created. This makes running shells
4374 as container init process a lot more fun.
4376 * The fstab support can now handle PARTUUID= and PARTLABEL=
4379 * A new ConditionHost= condition has been added to match
4380 against the hostname (with globs) and machine ID. This is
4381 useful for clusters where a single OS image is used to
4382 provision a large number of hosts which shall run slightly
4383 different sets of services.
4385 * Services which hit the restart limit will now be placed in a
4388 Contributions from: Bertram Poettering, Dave Reisner, Huang
4389 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
4390 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
4394 * When running in --user mode systemd will now become a
4395 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
4396 tree a lot more organized.
4398 * A new PartOf= unit dependency type has been introduced that
4399 may be used to group services in a natural way.
4401 * "systemctl enable" may now be used to enable instances of
4404 * journalctl now prints error log levels in red, and
4405 warning/notice log levels in bright white. It also supports
4406 filtering by log level now.
4408 * cgtop gained a new -n switch (similar to top), to configure
4409 the maximum number of iterations to run for. It also gained
4410 -b, to run in batch mode (accepting no input).
4412 * The suffix ".service" may now be omitted on most systemctl
4413 command lines involving service unit names.
4415 * There's a new bus call in logind to lock all sessions, as
4416 well as a loginctl verb for it "lock-sessions".
4418 * libsystemd-logind.so gained a new call sd_journal_perror()
4419 that works similar to libc perror() but logs to the journal
4420 and encodes structured information about the error number.
4422 * /etc/crypttab entries now understand the new keyfile-size=
4425 * shutdown(8) now can send a (configurable) wall message when
4426 a shutdown is cancelled.
4428 * The mount propagation mode for the root file system will now
4429 default to "shared", which is useful to make containers work
4430 nicely out-of-the-box so that they receive new mounts from
4431 the host. This can be undone locally by running "mount
4432 --make-rprivate /" if needed.
4434 * The prefdm.service file has been removed. Distributions
4435 should maintain this unit downstream if they intend to keep
4436 it around. However, we recommend writing normal unit files
4437 for display managers instead.
4439 * Since systemd is a crucial part of the OS we will now
4440 default to a number of compiler switches that improve
4441 security (hardening) such as read-only relocations, stack
4442 protection, and suchlike.
4444 * The TimeoutSec= setting for services is now split into
4445 TimeoutStartSec= and TimeoutStopSec= to allow configuration
4446 of individual time outs for the start and the stop phase of
4449 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
4450 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
4451 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
4452 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
4453 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
4454 Gundersen, Zbigniew Jędrzejewski-Szmek
4458 * The journal and id128 C APIs are now fully documented as man
4461 * Extra safety checks have been added when transitioning from
4462 the initial RAM disk to the main system to avoid accidental
4465 * /etc/crypttab entries now understand the new keyfile-offset=
4468 * systemctl -t can now be used to filter by unit load state.
4470 * The journal C API gained the new sd_journal_wait() call to
4471 make writing synchronous journal clients easier.
4473 * journalctl gained the new -D switch to show journals from a
4476 * journalctl now displays a special marker between log
4477 messages of two different boots.
4479 * The journal is now explicitly flushed to /var via a service
4480 systemd-journal-flush.service, rather than implicitly simply
4481 by seeing /var/log/journal to be writable.
4483 * journalctl (and the journal C APIs) can now match for much
4484 more complex expressions, with alternatives and
4487 * When transitioning from the initial RAM disk to the main
4488 system we will now kill all processes in a killing spree to
4489 ensure no processes stay around by accident.
4491 * Three new specifiers may be used in unit files: %u, %h, %s
4492 resolve to the user name, user home directory resp. user
4493 shell. This is useful for running systemd user instances.
4495 * We now automatically rotate journal files if their data
4496 object hash table gets a fill level > 75%. We also size the
4497 hash table based on the configured maximum file size. This
4498 together should lower hash collisions drastically and thus
4499 speed things up a bit.
4501 * journalctl gained the new "--header" switch to introspect
4502 header data of journal files.
4504 * A new setting SystemCallFilters= has been added to services
4505 which may be used to apply blacklists or whitelists to
4506 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
4508 * nspawn gained a new --link-journal= switch (and quicker: -j)
4509 to link the container journal with the host. This makes it
4510 very easy to centralize log viewing on the host for all
4511 guests while still keeping the journal files separated.
4513 * Many bugfixes and optimizations
4515 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
4516 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
4517 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
4522 * Several tools now understand kernel command line arguments,
4523 which are only read when run in an initial RAM disk. They
4524 usually follow closely their normal counterparts, but are
4527 * There's a new tool to analyze the readahead files that are
4528 automatically generated at boot. Use:
4530 /usr/lib/systemd/systemd-readahead analyze /.readahead
4532 * We now provide an early debug shell on tty9 if this enabled. Use:
4534 systemctl enable debug-shell.service
4536 * All plymouth related units have been moved into the Plymouth
4537 package. Please make sure to upgrade your Plymouth version
4540 * systemd-tmpfiles now supports getting passed the basename of
4541 a configuration file only, in which case it will look for it
4542 in all appropriate directories automatically.
4544 * udevadm info now takes a /dev or /sys path as argument, and
4545 does the right thing. Example:
4547 udevadm info /dev/sda
4548 udevadm info /sys/class/block/sda
4550 * systemctl now prints a warning if a unit is stopped but a
4551 unit that might trigger it continues to run. Example: a
4552 service is stopped but the socket that activates it is left
4555 * "systemctl status" will now mention if the log output was
4556 shortened due to rotation since a service has been started.
4558 * The journal API now exposes functions to determine the
4559 "cutoff" times due to rotation.
4561 * journald now understands SIGUSR1 and SIGUSR2 for triggering
4562 immediately flushing of runtime logs to /var if possible,
4563 resp. for triggering immediate rotation of the journal
4566 * It is now considered an error if a service is attempted to
4567 be stopped that is not loaded.
4569 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
4571 * systemd-analyze now supports Python 3
4573 * tmpfiles now supports cleaning up directories via aging
4574 where the first level dirs are always kept around but
4575 directories beneath it automatically aged. This is enabled
4576 by prefixing the age field with '~'.
4578 * Seat objects now expose CanGraphical, CanTTY properties
4579 which is required to deal with very fast bootups where the
4580 display manager might be running before the graphics drivers
4581 completed initialization.
4583 * Seat objects now expose a State property.
4585 * We now include RPM macros for service enabling/disabling
4586 based on the preset logic. We recommend RPM based
4587 distributions to make use of these macros if possible. This
4588 makes it simpler to reuse RPM spec files across
4591 * We now make sure that the collected systemd unit name is
4592 always valid when services log to the journal via
4595 * There's a new man page kernel-command-line(7) detailing all
4596 command line options we understand.
4598 * The fstab generator may now be disabled at boot by passing
4599 fstab=0 on the kernel command line.
4601 * A new kernel command line option modules-load= is now understood
4602 to load a specific kernel module statically, early at boot.
4604 * Unit names specified on the systemctl command line are now
4605 automatically escaped as needed. Also, if file system or
4606 device paths are specified they are automatically turned
4607 into the appropriate mount or device unit names. Example:
4609 systemctl status /home
4610 systemctl status /dev/sda
4612 * The SysVConsole= configuration option has been removed from
4613 system.conf parsing.
4615 * The SysV search path is no longer exported on the D-Bus
4618 * The Names= option is been removed from unit file parsing.
4620 * There's a new man page bootup(7) detailing the boot process.
4622 * Every unit and every generator we ship with systemd now
4623 comes with full documentation. The self-explanatory boot is
4626 * A couple of services gained "systemd-" prefixes in their
4627 name if they wrap systemd code, rather than only external
4628 code. Among them fsck@.service which is now
4629 systemd-fsck@.service.
4631 * The HaveWatchdog property has been removed from the D-Bus
4634 * systemd.confirm_spawn= on the kernel command line should now
4637 * There's a new man page crypttab(5) which details all options
4638 we actually understand.
4640 * systemd-nspawn gained a new --capability= switch to pass
4641 additional capabilities to the container.
4643 * timedated will now read known NTP implementation unit names
4644 from /usr/lib/systemd/ntp-units.d/*.list,
4645 systemd-timedated-ntp.target has been removed.
4647 * journalctl gained a new switch "-b" that lists log data of
4648 the current boot only.
4650 * The notify socket is in the abstract namespace again, in
4651 order to support daemons which chroot() at start-up.
4653 * There is a new Storage= configuration option for journald
4654 which allows configuration of where log data should go. This
4655 also provides a way to disable journal logging entirely, so
4656 that data collected is only forwarded to the console, the
4657 kernel log buffer or another syslog implementation.
4659 * Many bugfixes and optimizations
4661 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
4662 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
4663 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
4664 Shawn Landden, Tom Gundersen
4668 * "systemctl help <unit>" now shows the man page if one is
4671 * Several new man pages have been added.
4673 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
4674 MaxLevelConsole= can now be specified in
4675 journald.conf. These options allow reducing the amount of
4676 data stored on disk or forwarded by the log level.
4678 * TimerSlackNSec= can now be specified in system.conf for
4679 PID1. This allows system-wide power savings.
4681 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
4682 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
4687 * logind is now capable of (optionally) handling power and
4688 sleep keys as well as the lid switch.
4690 * journalctl now understands the syntax "journalctl
4691 /usr/bin/avahi-daemon" to get all log output of a specific
4694 * CapabilityBoundingSet= in system.conf now also influences
4695 the capability bound set of usermode helpers of the kernel.
4697 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
4698 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
4699 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
4700 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
4704 * Note that we skipped 139 releases here in order to set the
4705 new version to something that is greater than both udev's
4706 and systemd's most recent version number.
4708 * udev: all udev sources are merged into the systemd source tree now.
4709 All future udev development will happen in the systemd tree. It
4710 is still fully supported to use the udev daemon and tools without
4711 systemd running, like in initramfs or other init systems. Building
4712 udev though, will require the *build* of the systemd tree, but
4713 udev can be properly *run* without systemd.
4715 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
4716 should be used to create dead device nodes as workarounds for broken
4719 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
4720 no longer supported. udev_monitor_new_from_netlink() needs to be
4721 used to subscribe to events.
4723 * udev: when udevd is started by systemd, processes which are left
4724 behind by forking them off of udev rules, are unconditionally cleaned
4725 up and killed now after the event handling has finished. Services or
4726 daemons must be started as systemd services. Services can be
4727 pulled-in by udev to get started, but they can no longer be directly
4728 forked by udev rules.
4730 * udev: the daemon binary is called systemd-udevd now and installed
4731 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
4732 to adapt to that, create symlink, or rename the binary after building
4735 * libudev no longer provides these symbols:
4736 udev_monitor_from_socket()
4737 udev_queue_get_failed_list_entry()
4738 udev_get_{dev,sys,run}_path()
4739 The versions number was bumped and symbol versioning introduced.
4741 * systemd-loginctl and systemd-journalctl have been renamed
4742 to loginctl and journalctl to match systemctl.
4744 * The config files: /etc/systemd/systemd-logind.conf and
4745 /etc/systemd/systemd-journald.conf have been renamed to
4746 logind.conf and journald.conf. Package updates should rename
4747 the files to the new names on upgrade.
4749 * For almost all files the license is now LGPL2.1+, changed
4750 from the previous GPL2.0+. Exceptions are some minor stuff
4751 of udev (which will be changed to LGPL2.1 eventually, too),
4752 and the MIT licensed sd-daemon.[ch] library that is suitable
4753 to be used as drop-in files.
4755 * systemd and logind now handle system sleep states, in
4756 particular suspending and hibernating.
4758 * logind now implements a sleep/shutdown/idle inhibiting logic
4759 suitable for a variety of uses. Soonishly Lennart will blog
4760 about this in more detail.
4762 * var-run.mount and var-lock.mount are no longer provided
4763 (which prevously bind mounted these directories to their new
4764 places). Distributions which have not converted these
4765 directories to symlinks should consider stealing these files
4766 from git history and add them downstream.
4768 * We introduced the Documentation= field for units and added
4769 this to all our shipped units. This is useful to make it
4770 easier to explore the boot and the purpose of the various
4773 * All smaller setup units (such as
4774 systemd-vconsole-setup.service) now detect properly if they
4775 are run in a container and are skipped when
4776 appropriate. This guarantees an entirely noise-free boot in
4777 Linux container environments such as systemd-nspawn.
4779 * A framework for implementing offline system updates is now
4780 integrated, for details see:
4781 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
4783 * A new service type Type=idle is available now which helps us
4784 avoiding ugly interleaving of getty output and boot status
4787 * There's now a system-wide CapabilityBoundingSet= option to
4788 globally reduce the set of capabilities for the
4789 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
4790 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
4791 even CAP_NET_ADMIN system-wide for secure systems.
4793 * There are now system-wide DefaultLimitXXX= options to
4794 globally change the defaults of the various resource limits
4795 for all units started by PID 1.
4797 * Harald Hoyer's systemd test suite has been integrated into
4798 systemd which allows easy testing of systemd builds in qemu
4799 and nspawn. (This is really awesome! Ask us for details!)
4801 * The fstab parser is now implemented as generator, not inside
4804 * systemctl will now warn you if .mount units generated from
4805 /etc/fstab are out of date due to changes in fstab that
4806 have not been read by systemd yet.
4808 * systemd is now suitable for usage in initrds. Dracut has
4809 already been updated to make use of this. With this in place
4810 initrds get a slight bit faster but primarily are much
4811 easier to introspect and debug since "systemctl status" in
4812 the host system can be used to introspect initrd services,
4813 and the journal from the initrd is kept around too.
4815 * systemd-delta has been added, a tool to explore differences
4816 between user/admin configuration and vendor defaults.
4818 * PrivateTmp= now affects both /tmp and /var/tmp.
4820 * Boot time status messages are now much prettier and feature
4821 proper english language. Booting up systemd has never been
4824 * Read-ahead pack files now include the inode number of all
4825 files to pre-cache. When the inode changes the pre-caching
4826 is not attempted. This should be nicer to deal with updated
4827 packages which might result in changes of read-ahead
4830 * We now temporaritly lower the kernel's read_ahead_kb variable
4831 when collecting read-ahead data to ensure the kernel's
4832 built-in read-ahead does not add noise to our measurements
4833 of necessary blocks to pre-cache.
4835 * There's now RequiresMountsFor= to add automatic dependencies
4836 for all mounts necessary for a specific file system path.
4838 * MountAuto= and SwapAuto= have been removed from
4839 system.conf. Mounting file systems at boot has to take place
4842 * nspawn now learned a new switch --uuid= to set the machine
4843 ID on the command line.
4845 * nspawn now learned the -b switch to automatically search
4848 * vt102 is now the default TERM for serial TTYs, upgraded from
4851 * systemd-logind now works on VT-less systems.
4853 * The build tree has been reorganized. The individual
4854 components now have directories of their own.
4856 * A new condition type ConditionPathIsReadWrite= is now available.
4858 * nspawn learned the new -C switch to create cgroups for the
4859 container in other hierarchies.
4861 * We now have support for hardware watchdogs, configurable in
4864 * The scheduled shutdown logic now has a public API.
4866 * We now mount /tmp as tmpfs by default, but this can be
4867 masked and /etc/fstab can override it.
4869 * Since udisks does not make use of /media anymore we are not
4870 mounting a tmpfs on it anymore.
4872 * journalctl gained a new --local switch to only interleave
4873 locally generated journal files.
4875 * We can now load the IMA policy at boot automatically.
4877 * The GTK tools have been split off into a systemd-ui.
4879 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
4880 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
4881 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
4882 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
4883 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
4884 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
4885 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
4886 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
4887 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
4892 * This is mostly a bugfix release
4894 * Support optional initialization of the machine ID from the
4895 KVM or container configured UUID.
4897 * Support immediate reboots with "systemctl reboot -ff"
4899 * Show /etc/os-release data in systemd-analyze output
4901 * Many bugfixes for the journal, including endianness fixes and
4902 ensuring that disk space enforcement works
4904 * sd-login.h is C++ comptaible again
4906 * Extend the /etc/os-release format on request of the Debian
4909 * We now refuse non-UTF8 strings used in various configuration
4910 and unit files. This is done to ensure we do not pass invalid
4911 data over D-Bus or expose it elsewhere.
4913 * Register Mimo USB Screens as suitable for automatic seat
4916 * Read SELinux client context from journal clients in a race
4919 * Reorder configuration file lookup order. /etc now always
4920 overrides /run in order to allow the administrator to always
4921 and unconditionally override vendor supplied or
4922 automatically generated data.
4924 * The various user visible bits of the journal now have man
4925 pages. We still lack man pages for the journal API calls
4928 * We now ship all man pages in HTML format again in the
4931 Contributions from: Dave Reisner, Dirk Eibach, Frederic
4932 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
4933 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
4938 * This is mostly a bugfix release
4940 * systems lacking /etc/os-release are no longer supported.
4942 * Various functionality updates to libsystemd-login.so
4944 * Track class of PAM logins to distinguish greeters from
4947 Contributions from: Kay Sievers, Lennart Poettering, Michael
4952 * This is an important bugfix release for v41.
4954 * Building man pages is now optional which should be useful
4955 for those building systemd from git but unwilling to install
4958 * Watchdog support for supervising services is now usable. In
4959 a future release support for hardware watchdogs
4960 (i.e. /dev/watchdog) will be added building on this.
4962 * Service start rate limiting is now configurable and can be
4963 turned off per service. When a start rate limit is hit a
4964 reboot can automatically be triggered.
4966 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
4968 Contributions from: Benjamin Franzke, Bill Nottingham,
4969 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
4970 Schmidt, Michał Górny, Piotr Drąg
4974 * The systemd binary is installed /usr/lib/systemd/systemd now;
4975 An existing /sbin/init symlink needs to be adapted with the
4978 * The code that loads kernel modules has been ported to invoke
4979 libkmod directly, instead of modprobe. This means we do not
4980 support systems with module-init-tools anymore.
4982 * Watchdog support is now already useful, but still not
4985 * A new kernel command line option systemd.setenv= is
4986 understood to set system wide environment variables
4987 dynamically at boot.
4989 * We now limit the set of capabilities of systemd-journald.
4991 * We now set SIGPIPE to ignore by default, since it only is
4992 useful in shell pipelines, and has little use in general
4993 code. This can be disabled with IgnoreSIPIPE=no in unit
4996 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
4997 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
5002 * This is mostly a bugfix release
5004 * We now expose the reason why a service failed in the
5005 "Result" D-Bus property.
5007 * Rudimentary service watchdog support (will be completed over
5008 the next few releases.)
5010 * When systemd forks off in order execute some service we will
5011 now immediately changes its argv[0] to reflect which process
5012 it will execute. This is useful to minimize the time window
5013 with a generic argv[0], which makes bootcharts more useful
5015 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
5016 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
5017 Mike Kazantsev, Ray Strode
5021 * This is mostly a test release, but incorporates many
5024 * New systemd-cgtop tool to show control groups by their
5027 * Linking against libacl for ACLs is optional again. If
5028 disabled, support tracking device access for active logins
5029 goes becomes unavailable, and so does access to the user
5030 journals by the respective users.
5032 * If a group "adm" exists, journal files are automatically
5033 owned by them, thus allow members of this group full access
5034 to the system journal as well as all user journals.
5036 * The journal now stores the SELinux context of the logging
5037 client for all entries.
5039 * Add C++ inclusion guards to all public headers
5041 * New output mode "cat" in the journal to print only text
5042 messages, without any meta data like date or time.
5044 * Include tiny X server wrapper as a temporary stop-gap to
5045 teach XOrg udev display enumeration. This is used by display
5046 managers such as gdm, and will go away as soon as XOrg
5047 learned native udev hotplugging for display devices.
5049 * Add new systemd-cat tool for executing arbitrary programs
5050 with STDERR/STDOUT connected to the journal. Can also act as
5051 BSD logger replacement, and does so by default.
5053 * Optionally store all locally generated coredumps in the
5054 journal along with meta data.
5056 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
5057 writing short strings to files (for usage for /sys), and for
5058 creating symlinks, character and block device nodes.
5060 * New unit file option ControlGroupPersistent= to make cgroups
5061 persistent, following the mechanisms outlined in
5062 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
5064 * Support multiple local RTCs in a sane way
5066 * No longer monopolize IO when replaying readahead data on
5067 rotating disks, since we might starve non-file-system IO to
5068 death, since fanotify() will not see accesses done by blkid,
5071 * Do not show kernel threads in systemd-cgls anymore, unless
5072 requested with new -k switch.
5074 Contributions from: Dan Horák, Kay Sievers, Lennart
5075 Poettering, Michal Schmidt
5079 * This is mostly a test release, but incorporates many
5082 * The git repository moved to:
5083 git://anongit.freedesktop.org/systemd/systemd
5084 ssh://git.freedesktop.org/git/systemd/systemd
5086 * First release with the journal
5087 http://0pointer.de/blog/projects/the-journal.html
5089 * The journal replaces both systemd-kmsg-syslogd and
5090 systemd-stdout-bridge.
5092 * New sd_pid_get_unit() API call in libsystemd-logind
5094 * Many systemadm clean-ups
5096 * Introduce remote-fs-pre.target which is ordered before all
5097 remote mounts and may be used to start services before all
5100 * Added Mageia support
5102 * Add bash completion for systemd-loginctl
5104 * Actively monitor PID file creation for daemons which exit in
5105 the parent process before having finished writing the PID
5106 file in the daemon process. Daemons which do this need to be
5107 fixed (i.e. PID file creation must have finished before the
5108 parent exits), but we now react a bit more gracefully to them.
5110 * Add colourful boot output, mimicking the well-known output
5111 of existing distributions.
5113 * New option PassCredentials= for socket units, for
5114 compatibility with a recent kernel ABI breakage.
5116 * /etc/rc.local is now hooked in via a generator binary, and
5117 thus will no longer act as synchronization point during
5120 * systemctl list-unit-files now supports --root=.
5122 * systemd-tmpfiles now understands two new commands: z, Z for
5123 relabelling files according to the SELinux database. This is
5124 useful to apply SELinux labels to specific files in /sys,
5127 * Output of SysV services is now forwarded to both the console
5128 and the journal by default, not only just the console.
5130 * New man pages for all APIs from libsystemd-login.
5132 * The build tree got reorganized and a the build system is a
5133 lot more modular allowing embedded setups to specifically
5134 select the components of systemd they are interested in.
5136 * Support for Linux systems lacking the kernel VT subsystem is
5139 * configure's --with-rootdir= got renamed to
5140 --with-rootprefix= to follow the naming used by udev and
5143 * Unless specified otherwise we will now install to /usr instead
5144 of /usr/local by default.
5146 * Processes with '@' in argv[0][0] are now excluded from the
5147 final shut-down killing spree, following the logic explained
5149 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
5151 * All processes remaining in a service cgroup when we enter
5152 the START or START_PRE states are now killed with
5153 SIGKILL. That means it is no longer possible to spawn
5154 background processes from ExecStart= lines (which was never
5155 supported anyway, and bad style).
5157 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
5158 reloading of units together.
5160 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
5161 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
5162 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
5163 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
5164 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek