1 systemd System and Service Manager
5 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
6 There are no known issues with current sysfs, and udev does not need
7 or should be used to work around such bugs.
9 * udev does no longer enable USB HID power management. Several reports
10 indicate, that some devices cannot handle that setting.
12 * The udev accelerometer helper was removed. The functionality
13 is now fully included in iio-sensor-proxy. But this means,
14 older iio-sensor-proxy versions will no longer provide
15 accelerometer/orientation data with this systemd version.
16 Please upgrade iio-sensor-proxy to version 1.0.
18 * networkd gained a new configuration option IPv6PrivacyExtensions=
19 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
20 for Stateless Address") on selected networks.
22 * For the sake of fewer build-time dependencies and less code in the
23 main repository, the python bindings are about to be removed in the
24 next release. A new repository has been created which accommodates
25 the code from now on, and we kindly ask distributions to create a
26 separate package for this. The removal will take place in v223.
28 https://github.com/systemd/python-systemd
30 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
31 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
32 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
33 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
34 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
35 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
36 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
37 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
38 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
39 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
45 * The sd-bus.h and sd-event.h APIs have now been declared
46 stable and have been added to the official interface of
47 libsystemd.so. sd-bus implements an alternative D-Bus client
48 library, that is relatively easy to use, very efficient and
49 supports both classic D-Bus as well as kdbus as transport
50 backend. sd-event is a generic event loop abstraction that
51 is built around Linux epoll, but adds features such as event
52 prioritization or efficient timer handling. Both APIs are good
53 choices for C programs looking for a bus and/or event loop
54 implementation that is minimal and does not have to be
55 portable to other kernels.
57 * kdbus support is no longer compile-time optional. It is now
58 always built-in. However, it can still be disabled at
59 runtime using the kdbus=0 kernel command line setting, and
60 that setting may be changed to default to off, by specifying
61 --disable-kdbus at build-time. Note though that the kernel
62 command line setting has no effect if the kdbus.ko kernel
63 module is not installed, in which case kdbus is (obviously)
64 also disabled. We encourage all downstream distributions to
65 begin testing kdbus by adding it to the kernel images in the
66 development distributions, and leaving kdbus support in
69 * The minimal required util-linux version has been bumped to
72 * Support for chkconfig (--enable-chkconfig) was removed in
73 favor of calling an abstraction tool
74 /lib/systemd/systemd-sysv-install. This needs to be
75 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
76 in README for details.
78 * If there's a systemd unit and a SysV init script for the
79 same service name, and the user executes "systemctl enable"
80 for it (or a related call), then this will now enable both
81 (or execute the related operation on both), not just the
84 * The libudev API documentation has been converted from gtkdoc
87 * gudev has been removed from the systemd tree, it is now an
90 * The systemd-cgtop tool learnt a new --raw switch to generate
91 "raw" (machine parsable) output.
93 * networkd's IPForwarding= .network file setting learnt the
94 new setting "kernel", which ensures that networkd does not
95 change the IP forwarding sysctl from the default kernel
98 * The systemd-logind bus API now exposes a new boolean
99 property "Docked" that reports whether logind considers the
100 system "docked", i.e. connected to a docking station or not.
102 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
103 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
104 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
105 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
106 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
107 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
108 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
109 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
110 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
111 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
112 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
113 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
114 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
115 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
116 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
117 Fink, Zbigniew Jędrzejewski-Szmek
119 -- Berlin, 2015-06-19
123 * The gudev library has been extracted into a separate repository
124 available at: https://git.gnome.org/browse/libgudev/
125 It is now managed as part of the Gnome project. Distributions
126 are recommended to pass --disable-gudev to systemd and use
127 gudev from the Gnome project instead. gudev is still included
128 in systemd, for now. It will be removed soon, though. Please
129 also see the announcement-thread on systemd-devel:
130 http://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
132 * systemd now exposes a CPUUsageNSec= property for each
133 service unit on the bus, that contains the overall consumed
134 CPU time of a service (the sum of what each process of the
135 service consumed). This value is only available if
136 CPUAccounting= is turned on for a service, and is then shown
137 in the "systemctl status" output.
139 * Support for configuring alternative mappings of the old SysV
140 runlevels to systemd targets has been removed. They are now
141 hardcoded in a way that runlevels 2, 3, 4 all map to
142 multi-user.target and 5 to graphical.target (which
143 previously was already the default behaviour).
145 * The auto-mounter logic gained support for mount point
146 expiry, using a new TimeoutIdleSec= setting in .automount
147 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
149 * The EFI System Partition (ESP) as mounted to /boot by
150 systemd-efi-boot-generator will now be unmounted
151 automatically after 2 minutes of not being used. This should
152 minimize the risk of ESP corruptions.
154 * New /etc/fstab options x-systemd.requires= and
155 x-systemd.requires-mounts-for= are now supported to express
156 additional dependencies for mounts. This is useful for
157 journalling file systems that support external journal
158 devices or overlay file systems that require underlying file
159 systems to be mounted.
161 * systemd does not support direct live-upgrades (via systemctl
162 daemon-reexec) from versions older than v44 anymore. As no
163 distribution we are aware of shipped such old versions in a
164 stable release this should not be problematic.
166 * When systemd forks off a new per-connection service instance
167 it will now set the $REMOTE_ADDR environment variable to the
168 remote IP address, and $REMOTE_PORT environment variable to
169 the remote IP port. This behaviour is similar to the
170 corresponding environment variables defined by CGI.
172 * systemd-networkd gained support for uplink failure
173 detection. The BindCarrier= option allows binding interface
174 configuration dynamically to the link sense of other
175 interfaces. This is useful to achieve behaviour like in
178 * systemd-networkd gained support for configuring the DHCP
179 client identifier to use when requesting leases.
181 * systemd-networkd now has a per-network UseNTP= option to
182 configure whether NTP server information acquired via DHCP
183 is passed on to services like systemd-timesyncd.
185 * systemd-networkd gained support for vti6 tunnels.
187 * Note that systemd-networkd manages the sysctl variable
188 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
189 it is configured for since v219. The variable controls IP
190 forwarding, and is a per-interface alternative to the global
191 /proc/sys/net/ipv[46]/ip_forward. This setting is
192 configurable in the IPForward= option, which defaults to
193 "no". This means if networkd is used for an interface it is
194 no longer sufficient to set the global sysctl option to turn
195 on IP forwarding! Instead, the .network file option
196 IPForward= needs to be turned on! Note that the
197 implementation of this behaviour was broken in v219 and has
200 * Many bonding and vxlan options are now configurable in
203 * systemd-nspawn gained a new --property= setting to set unit
204 properties for the container scope. This is useful for
205 setting resource parameters (e.g "CPUShares=500") on
206 containers started from the command line.
208 * systemd-nspawn gained a new --private-users= switch to make
209 use of user namespacing available on recent Linux kernels.
211 * systemd-nspawn may now be called as part of a shell pipeline
212 in which case the pipes used for stdin and stdout are passed
213 directly to the process invoked in the container, without
214 indirection via a pseudo tty.
216 * systemd-nspawn gained a new switch to control the UNIX
217 signal to use when killing the init process of the container
220 * systemd-nspawn gained a new --overlay= switch for mounting
221 overlay file systems into the container using the new kernel
224 * When a container image is imported via systemd-importd and
225 the host file system is not btrfs, a loopback block device
226 file is created in /var/lib/machines.raw with a btrfs file
227 system inside. It is then mounted to /var/lib/machines to
228 enable btrfs features for container management. The loopback
229 file and btrfs file system is grown as needed when container
230 images are imported via systemd-importd.
232 * systemd-machined/systemd-importd gained support for btrfs
233 quota, to enforce container disk space limits on disk. This
234 is exposed in "machinectl set-limit".
236 * systemd-importd now can import containers from local .tar,
237 .raw and .qcow2 images, and export them to .tar and .raw. It
238 can also import dkr v2 images now from the network (on top
241 * systemd-importd gained support for verifying downloaded
242 images with gpg2 (previously only gpg1 was supported).
244 * systemd-machined, systemd-logind, systemd: most bus calls
245 are now accessible to unprivileged processes via
246 PolicyKit. Also, systemd-logind will now allow users to kill
247 their own sessions without further privileges or
250 * systemd-shutdownd has been removed. This service was
251 previously responsible for implementing scheduled shutdowns
252 as exposed in /usr/bin/shutdown's time parameter. This
253 functionality has now been moved into systemd-logind and is
254 accessible via a bus interface.
256 * "systemctl reboot" gained a new switch --firmware-setup that
257 can be used to reboot into the EFI firmware setup, if that
258 is available. systemd-logind now exposes an API on the bus
259 to trigger such reboots, in case graphical desktop UIs want
260 to cover this functionality.
262 * "systemctl enable", "systemctl disable" and "systemctl mask"
263 now support a new "--now" switch. If specified the units
264 that are enabled will also be started, and the ones
265 disabled/masked also stopped.
267 * The Gummiboot EFI boot loader tool has been merged into
268 systemd, and renamed to "systemd-boot". The bootctl tool has been
269 updated to support systemd-boot.
271 * An EFI kernel stub has been added that may be used to create
272 kernel EFI binaries that contain not only the actual kernel,
273 but also an initrd, boot splash, command line and OS release
274 information. This combined binary can then be signed as a
275 single image, so that the firmware can verify it all in one
276 step. systemd-boot has special support for EFI binaries created
277 like this and can extract OS release information from them
278 and show them in the boot menu. This functionality is useful
279 to implement cryptographically verified boot schemes.
281 * Optional support has been added to systemd-fsck to pass
282 fsck's progress report to an AF_UNIX socket in the file
285 * udev will no longer create device symlinks for all block
286 devices by default. A blacklist for excluding special block
287 devices from this logic has been turned into a whitelist
288 that requires picking block devices explicitly that require
291 * A new (currently still internal) API sd-device.h has been
292 added to libsystemd. This modernized API is supposed to
293 replace libudev eventually. In fact, already much of libudev
294 is now just a wrapper around sd-device.h.
296 * A new hwdb database for storing metadata about pointing
297 stick devices has been added.
299 * systemd-tmpfiles gained support for setting file attributes
300 similar to the "chattr" tool with new 'h' and 'H' lines.
302 * systemd-journald will no longer unconditionally set the
303 btrfs NOCOW flag on new journal files. This is instead done
304 with tmpfiles snippet using the new 'h' line type. This
305 allows easy disabling of this logic, by masking the
306 journal-nocow.conf tmpfiles file.
308 * systemd-journald will now translate audit message types to
309 human readable identifiers when writing them to the
310 journal. This should improve readability of audit messages.
312 * The LUKS logic gained support for the offset= and skip=
313 options in /etc/crypttab, as previously implemented by
316 * /usr/lib/os-release gained a new optional field VARIANT= for
317 distributions that support multiple variants (such as a
318 desktop edition, a server edition, ...)
320 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
321 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
322 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
323 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
324 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
325 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
326 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
327 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
328 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
329 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
330 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
331 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
332 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
333 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
334 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
335 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
336 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
337 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
338 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
339 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
340 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
341 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
342 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
343 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
344 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
345 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
346 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
348 -- Berlin, 2015-05-22
352 * Introduce a new API "sd-hwdb.h" for querying the hardware
353 metadata database. With this minimal interface one can query
354 and enumerate the udev hwdb, decoupled from the old libudev
355 library. libudev's interface for this is now only a wrapper
356 around sd-hwdb. A new tool systemd-hwdb has been added to
357 interface with and update the database.
359 * When any of systemd's tools copies files (for example due to
360 tmpfiles' C lines) a btrfs reflink will attempted first,
361 before bytewise copying is done.
363 * systemd-nspawn gained a new --ephemeral switch. When
364 specified a btrfs snapshot is taken of the container's root
365 directory, and immediately removed when the container
366 terminates again. Thus, a container can be started whose
367 changes never alter the container's root directory, and are
368 lost on container termination. This switch can also be used
369 for starting a container off the root file system of the
370 host without affecting the host OS. This switch is only
371 available on btrfs file systems.
373 * systemd-nspawn gained a new --template= switch. It takes the
374 path to a container tree to use as template for the tree
375 specified via --directory=, should that directory be
376 missing. This allows instantiating containers dynamically,
377 on first run. This switch is only available on btrfs file
380 * When a .mount unit refers to a mount point on which multiple
381 mounts are stacked, and the .mount unit is stopped all of
382 the stacked mount points will now be unmounted until no
385 * systemd now has an explicit notion of supported and
386 unsupported unit types. Jobs enqueued for unsupported unit
387 types will now fail with an "unsupported" error code. More
388 specifically .swap, .automount and .device units are not
389 supported in containers, .busname units are not supported on
390 non-kdbus systems. .swap and .automount are also not
391 supported if their respective kernel compile time options
394 * machinectl gained support for two new "copy-from" and
395 "copy-to" commands for copying files from a running
396 container to the host or vice versa.
398 * machinectl gained support for a new "bind" command to bind
399 mount host directories into local containers. This is
400 currently only supported for nspawn containers.
402 * networkd gained support for configuring bridge forwarding
403 database entries (fdb) from .network files.
405 * A new tiny daemon "systemd-importd" has been added that can
406 download container images in tar, raw, qcow2 or dkr formats,
407 and make them available locally in /var/lib/machines, so
408 that they can run as nspawn containers. The daemon can GPG
409 verify the downloads (not supported for dkr, since it has no
410 provisions for verifying downloads). It will transparently
411 decompress bz2, xz, gzip compressed downloads if necessary,
412 and restore sparse files on disk. The daemon uses privilege
413 separation to ensure the actual download logic runs with
414 fewer privileges than the deamon itself. machinectl has
415 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
416 make the functionality of importd available to the
417 user. With this in place the Fedora and Ubuntu "Cloud"
418 images can be downloaded and booted as containers unmodified
419 (the Fedora images lack the appropriate GPG signature files
420 currently, so they cannot be verified, but this will change
421 soon, hopefully). Note that downloading images is currently
422 only fully supported on btrfs.
424 * machinectl is now able to list container images found in
425 /var/lib/machines, along with some metadata about sizes of
426 disk and similar. If the directory is located on btrfs and
427 quota is enabled, this includes quota display. A new command
428 "image-status" has been added that shows additional
429 information about images.
431 * machinectl is now able to clone container images
432 efficiently, if the underlying file system (btrfs) supports
433 it, with the new "machinectl list-images" command. It also
434 gained commands for renaming and removing images, as well as
435 marking them read-only or read-write (supported also on
436 legacy file systems).
438 * networkd gained support for collecting LLDP network
439 announcements, from hardware that supports this. This is
440 shown in networkctl output.
442 * systemd-run gained support for a new -t (--pty) switch for
443 invoking a binary on a pty whose input and output is
444 connected to the invoking terminal. This allows executing
445 processes as system services while interactively
446 communicating with them via the terminal. Most interestingly
447 this is supported across container boundaries. Invoking
448 "systemd-run -t /bin/bash" is an alternative to running a
449 full login session, the difference being that the former
450 will not register a session, nor go through the PAM session
453 * tmpfiles gained support for a new "v" line type for creating
454 btrfs subvolumes. If the underlying file system is a legacy
455 file system, this automatically degrades to creating a
456 normal directory. Among others /var/lib/machines is now
457 created like this at boot, should it be missing.
459 * The directory /var/lib/containers/ has been deprecated and
460 been replaced by /var/lib/machines. The term "machines" has
461 been used in the systemd context as generic term for both
462 VMs and containers, and hence appears more appropriate for
463 this, as the directory can also contain raw images bootable
466 * systemd-nspawn when invoked with -M but without --directory=
467 or --image= is now capable of searching for the container
468 root directory, subvolume or disk image automatically, in
469 /var/lib/machines. systemd-nspawn@.service has been updated
470 to make use of this, thus allowing it to be used for raw
473 * A new machines.target unit has been introduced that is
474 supposed to group all containers/VMs invoked as services on
475 the system. systemd-nspawn@.service has been updated to
478 * machinectl gained a new "start" command, for invoking a
479 container as a service. "machinectl start foo" is mostly
480 equivalent to "systemctl start systemd-nspawn@foo.service",
481 but handles escaping in a nicer way.
483 * systemd-nspawn will now mount most of the cgroupfs tree
484 read-only into each container, with the exception of the
485 container's own subtree in the name=systemd hierarchy.
487 * journald now sets the special FS_NOCOW file flag for its
488 journal files. This should improve performance on btrfs, by
489 avoiding heavy fragmentation when journald's write-pattern
490 is used on COW file systems. It degrades btrfs' data
491 integrity guarantees for the files to the same levels as for
492 ext3/ext4 however. This should be OK though as journald does
493 its own data integrity checks and all its objects are
494 checksummed on disk. Also, journald should handle btrfs disk
495 full events a lot more gracefully now, by processing SIGBUS
496 errors, and not relying on fallocate() anymore.
498 * When journald detects that journal files it is writing to
499 have been deleted it will immediately start new journal
502 * systemd now provides a way to store file descriptors
503 per-service in PID 1.This is useful for daemons to ensure
504 that fds they require are not lost during a daemon
505 restart. The fds are passed to the deamon on the next
506 invocation in the same way socket activation fds are
507 passed. This is now used by journald to ensure that the
508 various sockets connected to all the system's stdout/stderr
509 are not lost when journald is restarted. File descriptors
510 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
511 an extension to sd_notify(). Note that a limit is enforced
512 on the number of fds a service can store in PID 1, and it
513 defaults to 0, so that no fds may be stored, unless this is
514 explicitly turned on.
516 * The default TERM variable to use for units connected to a
517 terminal, when no other value is explicitly is set is now
518 vt220 rather than vt102. This should be fairly safe still,
519 but allows PgUp/PgDn work.
521 * The /etc/crypttab option header= as known from Debian is now
524 * "loginctl user-status" and "loginctl session-status" will
525 now show the last 10 lines of log messages of the
526 user/session following the status output. Similar,
527 "machinectl status" will show the last 10 log lines
528 associated with a virtual machine or container
529 service. (Note that this is usually not the log messages
530 done in the VM/container itself, but simply what the
531 container manager logs. For nspawn this includes all console
534 * "loginctl session-status" without further argument will now
535 show the status of the session of the caller. Similar,
536 "lock-session", "unlock-session", "activate",
537 "enable-linger", "disable-linger" may now be called without
538 session/user parameter in which case they apply to the
539 caller's session/user.
541 * An X11 session scriptlet is now shipped that uploads
542 $DISPLAY and $XAUTHORITY into the environment of the systemd
543 --user daemon if a session begins. This should improve
544 compatibility with X11 enabled applications run as systemd
547 * Generators are now subject to masking via /etc and /run, the
548 same way as unit files.
550 * networkd .network files gained support for configuring
551 per-link IPv4/IPv6 packet forwarding as well as IPv4
552 masquerading. This is by default turned on for veth links to
553 containers, as registered by systemd-nspawn. This means that
554 nspawn containers run with --network-veth will now get
555 automatic routed access to the host's networks without any
556 further configuration or setup, as long as networkd runs on
559 * systemd-nspawn gained the --port= (-p) switch to expose TCP
560 or UDP posts of a container on the host. With this in place
561 it is possible to run containers with private veth links
562 (--network-veth), and have their functionality exposed on
563 the host as if their services were running directly on the
566 * systemd-nspawn's --network-veth switch now gained a short
567 version "-n", since with the changes above it is now truly
568 useful out-of-the-box. The systemd-nspawn@.service has been
569 updated to make use of it too by default.
571 * systemd-nspawn will now maintain a per-image R/W lock, to
572 ensure that the same image is not started more than once
573 writable. (It's OK to run an image multiple times
574 simultaneously in read-only mode.)
576 * systemd-nspawn's --image= option is now capable of
577 dissecting and booting MBR and GPT disk images that contain
578 only a single active Linux partition. Previously it
579 supported only GPT disk images with proper GPT type
580 IDs. This allows running cloud images from major
581 distributions directly with systemd-nspawn, without
584 * In addition to collecting mouse dpi data in the udev
585 hardware database, there's now support for collecting angle
586 information for mouse scroll wheels. The database is
587 supposed to guarantee similar scrolling behavior on mice
588 that it knows about. There's also support for collecting
589 information about Touchpad types.
591 * udev's input_id built-in will now also collect touch screen
592 dimension data and attach it to probed devices.
594 * /etc/os-release gained support for a Distribution Privacy
597 * networkd gained support for creating "ipvlan", "gretap",
598 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
600 * systemd-tmpfiles gained support for "a" lines for setting
603 * systemd-nspawn will now mount /tmp in the container to
604 tmpfs, automatically.
606 * systemd now exposes the memory.usage_in_bytes cgroup
607 attribute and shows it for each service in the "systemctl
608 status" output, if available.
610 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
611 immediate reboot is triggered. This useful if shutdown is
612 hung and is unable to complete, to expedite the
613 operation. Note that this kind of reboot will still unmount
614 all file systems, and hence should not result in fsck being
617 * A .device unit for an optical block device will now be
618 considered active only when a medium is in the drive. Also,
619 mount units are now bound to their backing devices thus
620 triggering automatic unmounting when devices become
621 unavailable. With this in place systemd will now
622 automatically unmount left-over mounts when a CD-ROM is
623 ejected or an USB stick is yanked from the system.
625 * networkd-wait-online now has support for waiting for
626 specific interfaces only (with globbing), and for giving up
627 after a configurable timeout.
629 * networkd now exits when idle. It will be automatically
630 restarted as soon as interfaces show up, are removed or
631 change state. networkd will stay around as long as there is
632 at least one DHCP state machine or similar around, that keep
635 * networkd may now configure IPv6 link-local addressing in
636 addition to IPv4 link-local addressing.
638 * The IPv6 "token" for use in SLAAC may now be configured for
639 each .network interface in networkd.
641 * Routes configured with networkd may now be assigned a scope
644 * networkd's [Match] sections now support globbing and lists
645 of multiple space-separated matches per item.
647 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
648 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
649 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
650 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
651 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
652 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
653 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
654 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
655 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
656 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
657 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
658 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
659 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
660 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
661 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
662 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
663 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
664 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
665 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
666 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
667 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
668 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
669 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
670 Hoffmann, Zbigniew Jędrzejewski-Szmek
672 -- Berlin, 2015-02-16
676 * When querying unit file enablement status (for example via
677 "systemctl is-enabled"), a new state "indirect" is now known
678 which indicates that a unit might not be enabled itself, but
679 another unit listed in its Alias= setting might be.
681 * Similar to the various existing ConditionXYZ= settings for
682 units there are now matching AssertXYZ= settings. While
683 failing conditions cause a unit to be skipped, but its job
684 to succeed, failing assertions declared like this will cause
685 a unit start operation and its job to fail.
687 * hostnamed now knows a new chassis type "embedded".
689 * systemctl gained a new "edit" command. When used on a unit
690 file this allows extending unit files with .d/ drop-in
691 configuration snippets or editing the full file (after
692 copying it from /usr/lib to /etc). This will invoke the
693 user's editor (as configured with $EDITOR), and reload the
694 modified configuration after editing.
696 * "systemctl status" now shows the suggested enablement state
697 for a unit, as declared in the (usually vendor-supplied)
700 * nss-myhostname will now resolve the single-label host name
701 "gateway" to the locally configured default IP routing
702 gateways, ordered by their metrics. This assigns a stable
703 name to the used gateways, regardless which ones are
704 currently configured. Note that the name will only be
705 resolved after all other name sources (if nss-myhostname is
706 configured properly) and should hence not negatively impact
707 systems that use the single-label host name "gateway" in
710 * systemd-inhibit now allows filtering by mode when listing
713 * Scope and service units gained a new "Delegate" boolean
714 property, which when set allows processes running inside the
715 unit to further partition resources. This is primarily
716 useful for systemd user instances as well as container
719 * journald will now pick up audit messages directly from
720 the kernel, and log them like any other log message. The
721 audit fields are split up and fully indexed. This means that
722 journalctl in many ways is now a (nicer!) alternative to
723 ausearch, the traditional audit client. Note that this
724 implements only a minimal audit client, if you want the
725 special audit modes like reboot-on-log-overflow, please use
726 the traditional auditd instead, which can be used in
727 parallel to journald.
729 * The ConditionSecurity= unit file option now understands the
730 special string "audit" to check whether auditing is
733 * journalctl gained two new commands --vacuum-size= and
734 --vacuum-time= to delete old journal files until the
735 remaining ones take up no more the specified size on disk,
736 or are not older than the specified time.
738 * A new, native PPPoE library has been added to sd-network,
739 systemd's library of light-weight networking protocols. This
740 library will be used in a future version of networkd to
741 enable PPPoE communication without an external pppd daemon.
743 * The busctl tool now understands a new "capture" verb that
744 works similar to "monitor", but writes a packet capture
745 trace to STDOUT that can be redirected to a file which is
746 compatible with libcap's capture file format. This can then
747 be loaded in Wireshark and similar tools to inspect bus
750 * The busctl tool now understands a new "tree" verb that shows
751 the object trees of a specific service on the bus, or of all
754 * The busctl tool now understands a new "introspect" verb that
755 shows all interfaces and members of objects on the bus,
756 including their signature and values. This is particularly
757 useful to get more information about bus objects shown by
758 the new "busctl tree" command.
760 * The busctl tool now understands new verbs "call",
761 "set-property" and "get-property" for invoking bus method
762 calls, setting and getting bus object properties in a
765 * busctl gained a new --augment-creds= argument that controls
766 whether the tool shall augment credential information it
767 gets from the bus with data from /proc, in a possibly
770 * nspawn's --link-journal= switch gained two new values
771 "try-guest" and "try-host" that work like "guest" and
772 "host", but do not fail if the host has no persistent
773 journalling enabled. -j is now equivalent to
774 --link-journal=try-guest.
776 * macvlan network devices created by nspawn will now have
777 stable MAC addresses.
779 * A new SmackProcessLabel= unit setting has been added, which
780 controls the SMACK security label processes forked off by
781 the respective unit shall use.
783 * If compiled with --enable-xkbcommon, systemd-localed will
784 verify x11 keymap settings by compiling the given keymap. It
785 will spew out warnings if the compilation fails. This
786 requires libxkbcommon to be installed.
788 * When a coredump is collected a larger number of metadata
789 fields is now collected and included in the journal records
790 created for it. More specifically control group membership,
791 environment variables, memory maps, working directory,
792 chroot directory, /proc/$PID/status, and a list of open file
793 descriptors is now stored in the log entry.
795 * The udev hwdb now contains DPI information for mice. For
798 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
800 * All systemd programs that read standalone configuration
801 files in /etc now also support a corresponding series of
802 .conf.d configuration directories in /etc/, /run/,
803 /usr/local/lib/, /usr/lib/, and (if configured with
804 --enable-split-usr) /lib/. In particular, the following
805 configuration files now have corresponding configuration
806 directories: system.conf user.conf, logind.conf,
807 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
808 resolved.conf, timesyncd.conf, journal-remote.conf, and
809 journal-upload.conf. Note that distributions should use the
810 configuration directories in /usr/lib/; the directories in
811 /etc/ are reserved for the system administrator.
813 * systemd-rfkill will no longer take the rfkill device name
814 into account when storing rfkill state on disk, as the name
815 might be dynamically assigned and not stable. Instead, the
816 ID_PATH udev variable combined with the rfkill type (wlan,
817 bluetooth, ...) is used.
819 * A new service systemd-machine-id-commit.service has been
820 added. When used on systems where /etc is read-only during
821 boot, and /etc/machine-id is not initialized (but an empty
822 file), this service will copy the temporary machine ID
823 created as replacement into /etc after the system is fully
824 booted up. This is useful for systems that are freshly
825 installed with a non-initialized machine ID, but should get
826 a fixed machine ID for subsequent boots.
828 * networkd's .netdev files now provide a large set of
829 configuration parameters for VXLAN devices. Similar, the
830 bridge port cost parameter is now configurable in .network
831 files. There's also new support for configuring IP source
832 routing. networkd .link files gained support for a new
833 OriginalName= match that is useful to match against the
834 original interface name the kernel assigned. .network files
835 may include MTU= and MACAddress= fields for altering the MTU
836 and MAC address while being connected to a specific network
839 * The LUKS logic gained supported for configuring
840 UUID-specific key files. There's also new support for naming
841 LUKS device from the kernel command line, using the new
844 * Timer units may now be transiently created via the bus API
845 (this was previously already available for scope and service
846 units). In addition it is now possible to create multiple
847 transient units at the same time with a single bus call. The
848 "systemd-run" tool has been updated to make use of this for
849 running commands on a specified time, in at(1)-style.
851 * tmpfiles gained support for "t" lines, for assigning
852 extended attributes to files. Among other uses this may be
853 used to assign SMACK labels to files.
855 Contributions from: Alin Rauta, Alison Chaiken, Andrej
856 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
857 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
858 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
859 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
860 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
861 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
862 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
863 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
864 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
865 Michael Biebl, Michael Chapman, Michael Marineau, Michal
866 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
867 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
868 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
869 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
870 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
871 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
872 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
874 -- Berlin, 2014-12-10
878 * journalctl gained the new options -t/--identifier= to match
879 on the syslog identifier (aka "tag"), as well as --utc to
880 show log timestamps in the UTC timezone. journalctl now also
881 accepts -n/--lines=all to disable line capping in a pager.
883 * journalctl gained a new switch, --flush, that synchronously
884 flushes logs from /run/log/journal to /var/log/journal if
885 persistent storage is enabled. systemd-journal-flush.service
886 now waits until the operation is complete.
888 * Services can notify the manager before they start a reload
889 (by sending RELOADING=1) or shutdown (by sending
890 STOPPING=1). This allows the manager to track and show the
891 internal state of daemons and closes a race condition when
892 the process is still running but has closed its D-Bus
895 * Services with Type=oneshot do not have to have any ExecStart
898 * User units are now loaded also from
899 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
900 /run/systemd/user directory that was already previously
901 supported, but is under the control of the user.
903 * Job timeouts (i.e. time-outs on the time a job that is
904 queued stays in the run queue) can now optionally result in
905 immediate reboot or power-off actions (JobTimeoutAction= and
906 JobTimeoutRebootArgument=). This is useful on ".target"
907 units, to limit the maximum time a target remains
908 undispatched in the run queue, and to trigger an emergency
909 operation in such a case. This is now used by default to
910 turn off the system if boot-up (as defined by everything in
911 basic.target) hangs and does not complete for at least
912 15min. Also, if power-off or reboot hang for at least 30min
913 an immediate power-off/reboot operation is triggered. This
914 functionality is particularly useful to increase reliability
915 on embedded devices, but also on laptops which might
916 accidentally get powered on when carried in a backpack and
917 whose boot stays stuck in a hard disk encryption passphrase
920 * systemd-logind can be configured to also handle lid switch
921 events even when the machine is docked or multiple displays
922 are attached (HandleLidSwitchDocked= option).
924 * A helper binary and a service have been added which can be
925 used to resume from hibernation in the initramfs. A
926 generator will parse the resume= option on the kernel
927 command line to trigger resume.
929 * A user console daemon systemd-consoled has been
930 added. Currently, it is a preview, and will so far open a
931 single terminal on each session of the user marked as
932 Desktop=systemd-console.
934 * Route metrics can be specified for DHCP routes added by
937 * The SELinux context of socket-activated services can be set
938 from the information provided by the networking stack
939 (SELinuxContextFromNet= option).
941 * Userspace firmware loading support has been removed and
942 the minimum supported kernel version is thus bumped to 3.7.
944 * Timeout for udev workers has been increased from 1 to 3
945 minutes, but a warning will be printed after 1 minute to
946 help diagnose kernel modules that take a long time to load.
948 * Udev rules can now remove tags on devices with TAG-="foobar".
950 * systemd's readahead implementation has been removed. In many
951 circumstances it didn't give expected benefits even for
952 rotational disk drives and was becoming less relevant in the
953 age of SSDs. As none of the developers has been using
954 rotating media anymore, and nobody stepped up to actively
955 maintain this component of systemd it has now been removed.
957 * Swap units can use Options= to specify discard options.
958 Discard options specified for swaps in /etc/fstab are now
961 * Docker containers are now detected as a separate type of
964 * The Password Agent protocol gained support for queries where
965 the user input is shown, useful e.g. for user names.
966 systemd-ask-password gained a new --echo option to turn that
969 * The default sysctl.d/ snippets will now set:
971 net.core.default_qdisc = fq_codel
973 This selects Fair Queuing Controlled Delay as the default
974 queuing discipline for network interfaces. fq_codel helps
975 fight the network bufferbloat problem. It is believed to be
976 a good default with no tuning required for most workloads.
977 Downstream distributions may override this choice. On 10Gbit
978 servers that do not do forwarding, "fq" may perform better.
979 Systems without a good clocksource should use "pfifo_fast".
981 * If kdbus is enabled during build a new option BusPolicy= is
982 available for service units, that allows locking all service
983 processes into a stricter bus policy, in order to limit
984 access to various bus services, or even hide most of them
985 from the service's view entirely.
987 * networkctl will now show the .network and .link file
988 networkd has applied to a specific interface.
990 * sd-login gained a new API call sd_session_get_desktop() to
991 query which desktop environment has been selected for a
994 * UNIX utmp support is now compile-time optional to support
997 * systemctl gained two new commands "add-wants" and
998 "add-requires" for pulling in units from specific targets
1001 * If the word "rescue" is specified on the kernel command line
1002 the system will now boot into rescue mode (aka
1003 rescue.target), which was previously available only by
1004 specifying "1" or "systemd.unit=rescue.target" on the kernel
1005 command line. This new kernel command line option nicely
1006 mirrors the already existing "emergency" kernel command line
1009 * New kernel command line options mount.usr=, mount.usrflags=,
1010 mount.usrfstype= have been added that match root=, rootflags=,
1011 rootfstype= but allow mounting a specific file system to
1014 * The $NOTIFY_SOCKET is now also passed to control processes of
1015 services, not only the main process.
1017 * This version reenables support for fsck's -l switch. This
1018 means at least version v2.25 of util-linux is required for
1019 operation, otherwise dead-locks on device nodes may
1020 occur. Again: you need to update util-linux to at least
1021 v2.25 when updating systemd to v217.
1023 * The "multi-seat-x" tool has been removed from systemd, as
1024 its functionality has been integrated into X servers 1.16,
1025 and the tool is hence redundant. It is recommended to update
1026 display managers invoking this tool to simply invoke X
1027 directly from now on, again.
1029 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
1030 message flag has been added for all of systemd's PolicyKit
1031 authenticated method calls has been added. In particular
1032 this now allows optional interactive authorization via
1033 PolicyKit for many of PID1's privileged operations such as
1034 unit file enabling and disabling.
1036 * "udevadm hwdb --update" learnt a new switch "--usr" for
1037 placing the rebuilt hardware database in /usr instead of
1038 /etc. When used only hardware database entries stored in
1039 /usr will be used, and any user database entries in /etc are
1040 ignored. This functionality is useful for vendors to ship a
1041 pre-built database on systems where local configuration is
1042 unnecessary or unlikely.
1044 * Calendar time specifications in .timer units now also
1045 understand the strings "semi-annually", "quarterly" and
1046 "minutely" as shortcuts (in addition to the preexisting
1047 "anually", "hourly", ...).
1049 * systemd-tmpfiles will now correctly create files in /dev
1050 at boot which are marked for creation only at boot. It is
1051 recommended to always create static device nodes with 'c!'
1052 and 'b!', so that they are created only at boot and not
1053 overwritten at runtime.
1055 * When the watchdog logic is used for a service (WatchdogSec=)
1056 and the watchdog timeout is hit the service will now be
1057 terminated with SIGABRT (instead of just SIGTERM), in order
1058 to make sure a proper coredump and backtrace is
1059 generated. This ensures that hanging services will result in
1060 similar coredump/backtrace behaviour as services that hit a
1063 Contributions from: Andreas Henriksson, Andrei Borzenkov,
1064 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
1065 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
1066 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
1067 Herrmann, David Sommerseth, David Strauss, Emil Renner
1068 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
1069 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
1070 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
1071 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
1072 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
1073 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
1074 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
1075 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
1076 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
1077 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
1078 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
1079 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
1080 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
1081 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
1082 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
1083 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
1086 -- Berlin, 2014-10-28
1090 * timedated no longer reads NTP implementation unit names from
1091 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
1092 implementations should add a
1094 Conflicts=systemd-timesyncd.service
1096 to their unit files to take over and replace systemd's NTP
1097 default functionality.
1099 * systemd-sysusers gained a new line type "r" for configuring
1100 which UID/GID ranges to allocate system users/groups
1101 from. Lines of type "u" may now add an additional column
1102 that specifies the home directory for the system user to be
1103 created. Also, systemd-sysusers may now optionally read user
1104 information from STDIN instead of a file. This is useful for
1105 invoking it from RPM preinst scriptlets that need to create
1106 users before the first RPM file is installed since these
1107 files might need to be owned by them. A new
1108 %sysusers_create_inline RPM macro has been introduced to do
1109 just that. systemd-sysusers now updates the shadow files as
1110 well as the user/group databases, which should enhance
1111 compatibility with certain tools like grpck.
1113 * A number of bus APIs of PID 1 now optionally consult
1114 PolicyKit to permit access for otherwise unprivileged
1115 clients under certain conditions. Note that this currently
1116 doesn't support interactive authentication yet, but this is
1117 expected to be added eventually, too.
1119 * /etc/machine-info now has new fields for configuring the
1120 deployment environment of the machine, as well as the
1121 location of the machine. hostnamectl has been updated with
1122 new command to update these fields.
1124 * systemd-timesyncd has been updated to automatically acquire
1125 NTP server information from systemd-networkd, which might
1126 have been discovered via DHCP.
1128 * systemd-resolved now includes a caching DNS stub resolver
1129 and a complete LLMNR name resolution implementation. A new
1130 NSS module "nss-resolve" has been added which can be used
1131 instead of glibc's own "nss-dns" to resolve hostnames via
1132 systemd-resolved. Hostnames, addresses and arbitrary RRs may
1133 be resolved via systemd-resolved D-Bus APIs. In contrast to
1134 the glibc internal resolver systemd-resolved is aware of
1135 multi-homed system, and keeps DNS server and caches separate
1136 and per-interface. Queries are sent simultaneously on all
1137 interfaces that have DNS servers configured, in order to
1138 properly handle VPNs and local LANs which might resolve
1139 separate sets of domain names. systemd-resolved may acquire
1140 DNS server information from systemd-networkd automatically,
1141 which in turn might have discovered them via DHCP. A tool
1142 "systemd-resolve-host" has been added that may be used to
1143 query the DNS logic in resolved. systemd-resolved implements
1144 IDNA and automatically uses IDNA or UTF-8 encoding depending
1145 on whether classic DNS or LLMNR is used as transport. In the
1146 next releases we intend to add a DNSSEC and mDNS/DNS-SD
1147 implementation to systemd-resolved.
1149 * A new NSS module nss-mymachines has been added, that
1150 automatically resolves the names of all local registered
1151 containers to their respective IP addresses.
1153 * A new client tool "networkctl" for systemd-networkd has been
1154 added. It currently is entirely passive and will query
1155 networking configuration from udev, rtnetlink and networkd,
1156 and present it to the user in a very friendly
1157 way. Eventually, we hope to extend it to become a full
1158 control utility for networkd.
1160 * .socket units gained a new DeferAcceptSec= setting that
1161 controls the kernels' TCP_DEFER_ACCEPT sockopt for
1162 TCP. Similar, support for controlling TCP keep-alive
1163 settings has been added (KeepAliveTimeSec=,
1164 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
1165 turning off Nagle's algorithm on TCP has been added
1168 * logind learned a new session type "web", for use in projects
1169 like Cockpit which register web clients as PAM sessions.
1171 * timer units with at least one OnCalendar= setting will now
1172 be started only after timer-sync.target has been
1173 reached. This way they will not elapse before the system
1174 clock has been corrected by a local NTP client or
1175 similar. This is particular useful on RTC-less embedded
1176 machines, that come up with an invalid system clock.
1178 * systemd-nspawn's --network-veth= switch should now result in
1179 stable MAC addresses for both the outer and the inner side
1182 * systemd-nspawn gained a new --volatile= switch for running
1183 container instances with /etc or /var unpopulated.
1185 * The kdbus client code has been updated to use the new Linux
1186 3.17 memfd subsystem instead of the old kdbus-specific one.
1188 * systemd-networkd's DHCP client and server now support
1189 FORCERENEW. There are also new configuration options to
1190 configure the vendor client identifier and broadcast mode
1193 * systemd will no longer inform the kernel about the current
1194 timezone, as this is necessarily incorrect and racy as the
1195 kernel has no understanding of DST and similar
1196 concepts. This hence means FAT timestamps will be always
1197 considered UTC, similar to what Android is already
1198 doing. Also, when the RTC is configured to the local time
1199 (rather than UTC) systemd will never synchronize back to it,
1200 as this might confuse Windows at a later boot.
1202 * systemd-analyze gained a new command "verify" for offline
1203 validation of unit files.
1205 * systemd-networkd gained support for a couple of additional
1206 settings for bonding networking setups. Also, the metric for
1207 statically configured routes may now be configured. For
1208 network interfaces where this is appropriate the peer IP
1209 address may now be configured.
1211 * systemd-networkd's DHCP client will no longer request
1212 broadcasting by default, as this tripped up some networks.
1213 For hardware where broadcast is required the feature should
1214 be switched back on using RequestBroadcast=yes.
1216 * systemd-networkd will now set up IPv4LL addresses (when
1217 enabled) even if DHCP is configured successfully.
1219 * udev will now default to respect network device names given
1220 by the kernel when the kernel indicates that these are
1221 predictable. This behavior can be tweaked by changing
1222 NamePolicy= in the relevant .link file.
1224 * A new library systemd-terminal has been added that
1225 implements full TTY stream parsing and rendering. This
1226 library is supposed to be used later on for implementing a
1227 full userspace VT subsystem, replacing the current kernel
1230 * A new tool systemd-journal-upload has been added to push
1231 journal data to a remote system running
1232 systemd-journal-remote.
1234 * journald will no longer forward all local data to another
1235 running syslog daemon. This change has been made because
1236 rsyslog (which appears to be the most commonly used syslog
1237 implementation these days) no longer makes use of this, and
1238 instead pulls the data out of the journal on its own. Since
1239 forwarding the messages to a non-existent syslog server is
1240 more expensive than we assumed we have now turned this
1241 off. If you run a syslog server that is not a recent rsyslog
1242 version, you have to turn this option on again
1243 (ForwardToSyslog= in journald.conf).
1245 * journald now optionally supports the LZ4 compressor for
1246 larger journal fields. This compressor should perform much
1247 better than XZ which was the previous default.
1249 * machinectl now shows the IP addresses of local containers,
1250 if it knows them, plus the interface name of the container.
1252 * A new tool "systemd-escape" has been added that makes it
1253 easy to escape strings to build unit names and similar.
1255 * sd_notify() messages may now include a new ERRNO= field
1256 which is parsed and collected by systemd and shown among the
1257 "systemctl status" output for a service.
1259 * A new component "systemd-firstboot" has been added that
1260 queries the most basic systemd information (timezone,
1261 hostname, root password) interactively on first
1262 boot. Alternatively it may also be used to provision these
1263 things offline on OS images installed into directories.
1265 * The default sysctl.d/ snippets will now set
1267 net.ipv4.conf.default.promote_secondaries=1
1269 This has the benefit of no flushing secondary IP addresses
1270 when primary addresses are removed.
1272 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
1273 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
1274 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
1275 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
1276 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
1277 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
1278 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
1279 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
1280 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
1281 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
1282 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
1283 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
1284 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
1285 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
1286 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
1288 -- Berlin, 2014-08-19
1292 * A new tool systemd-sysusers has been added. This tool
1293 creates system users and groups in /etc/passwd and
1294 /etc/group, based on static declarative system user/group
1295 definitions in /usr/lib/sysusers.d/. This is useful to
1296 enable factory resets and volatile systems that boot up with
1297 an empty /etc directory, and thus need system users and
1298 groups created during early boot. systemd now also ships
1299 with two default sysusers.d/ files for the most basic
1300 users and groups systemd and the core operating system
1303 * A new tmpfiles snippet has been added that rebuilds the
1304 essential files in /etc on boot, should they be missing.
1306 * A directive for ensuring automatic clean-up of
1307 /var/cache/man/ has been removed from the default
1308 configuration. This line should now be shipped by the man
1309 implementation. The necessary change has been made to the
1310 man-db implementation. Note that you need to update your man
1311 implementation to one that ships this line, otherwise no
1312 automatic clean-up of /var/cache/man will take place.
1314 * A new condition ConditionNeedsUpdate= has been added that
1315 may conditionalize services to only run when /etc or /var
1316 are "older" than the vendor operating system resources in
1317 /usr. This is useful for reconstructing or updating /etc
1318 after an offline update of /usr or a factory reset, on the
1319 next reboot. Services that want to run once after such an
1320 update or reset should use this condition and order
1321 themselves before the new systemd-update-done.service, which
1322 will mark the two directories as fully updated. A number of
1323 service files have been added making use of this, to rebuild
1324 the udev hardware database, the journald message catalog and
1325 dynamic loader cache (ldconfig). The systemd-sysusers tool
1326 described above also makes use of this now. With this in
1327 place it is now possible to start up a minimal operating
1328 system with /etc empty cleanly. For more information on the
1329 concepts involved see this recent blog story:
1331 http://0pointer.de/blog/projects/stateless.html
1333 * A new system group "input" has been introduced, and all
1334 input device nodes get this group assigned. This is useful
1335 for system-level software to get access to input devices. It
1336 complements what is already done for "audio" and "video".
1338 * systemd-networkd learnt minimal DHCPv4 server support in
1339 addition to the existing DHCPv4 client support. It also
1340 learnt DHCPv6 client and IPv6 Router Solicitation client
1341 support. The DHCPv4 client gained support for static routes
1342 passed in from the server. Note that the [DHCPv4] section
1343 known in older systemd-networkd versions has been renamed to
1344 [DHCP] and is now also used by the DHCPv6 client. Existing
1345 .network files using settings of this section should be
1346 updated, though compatibility is maintained. Optionally, the
1347 client hostname may now be sent to the DHCP server.
1349 * networkd gained support for vxlan virtual networks as well
1350 as tun/tap and dummy devices.
1352 * networkd gained support for automatic allocation of address
1353 ranges for interfaces from a system-wide pool of
1354 addresses. This is useful for dynamically managing a large
1355 number of interfaces with a single network configuration
1356 file. In particular this is useful to easily assign
1357 appropriate IP addresses to the veth links of a large number
1358 of nspawn instances.
1360 * RPM macros for processing sysusers, sysctl and binfmt
1361 drop-in snippets at package installation time have been
1364 * The /etc/os-release file should now be placed in
1365 /usr/lib/os-release. The old location is automatically
1366 created as symlink. /usr/lib is the more appropriate
1367 location of this file, since it shall actually describe the
1368 vendor operating system shipped in /usr, and not the
1369 configuration stored in /etc.
1371 * .mount units gained a new boolean SloppyOptions= setting
1372 that maps to mount(8)'s -s option which enables permissive
1373 parsing of unknown mount options.
1375 * tmpfiles learnt a new "L+" directive which creates a symlink
1376 but (unlike "L") deletes a pre-existing file first, should
1377 it already exist and not already be the correct
1378 symlink. Similar, "b+", "c+" and "p+" directives have been
1379 added as well, which create block and character devices, as
1380 well as fifos in the filesystem, possibly removing any
1381 pre-existing files of different types.
1383 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
1384 'argument' field (which so far specified the source to
1385 symlink/copy the files from) is now optional. If omitted the
1386 same file os copied from /usr/share/factory/ suffixed by the
1387 full destination path. This is useful for populating /etc
1388 with essential files, by copying them from vendor defaults
1389 shipped in /usr/share/factory/etc.
1391 * A new command "systemctl preset-all" has been added that
1392 applies the service preset settings to all installed unit
1393 files. A new switch --preset-mode= has been added that
1394 controls whether only enable or only disable operations
1397 * A new command "systemctl is-system-running" has been added
1398 that allows checking the overall state of the system, for
1399 example whether it is fully up and running.
1401 * When the system boots up with an empty /etc, the equivalent
1402 to "systemctl preset-all" is executed during early boot, to
1403 make sure all default services are enabled after a factory
1406 * systemd now contains a minimal preset file that enables the
1407 most basic services systemd ships by default.
1409 * Unit files' [Install] section gained a new DefaultInstance=
1410 field for defining the default instance to create if a
1411 template unit is enabled with no instance specified.
1413 * A new passive target cryptsetup-pre.target has been added
1414 that may be used by services that need to make they run and
1415 finish before the first LUKS cryptographic device is set up.
1417 * The /dev/loop-control and /dev/btrfs-control device nodes
1418 are now owned by the "disk" group by default, opening up
1419 access to this group.
1421 * systemd-coredump will now automatically generate a
1422 stack trace of all core dumps taking place on the system,
1423 based on elfutils' libdw library. This stack trace is logged
1426 * systemd-coredump may now optionally store coredumps directly
1427 on disk (in /var/lib/systemd/coredump, possibly compressed),
1428 instead of storing them unconditionally in the journal. This
1429 mode is the new default. A new configuration file
1430 /etc/systemd/coredump.conf has been added to configure this
1431 and other parameters of systemd-coredump.
1433 * coredumpctl gained a new "info" verb to show details about a
1434 specific coredump. A new switch "-1" has also been added
1435 that makes sure to only show information about the most
1436 recent entry instead of all entries. Also, as the tool is
1437 generally useful now the "systemd-" prefix of the binary
1438 name has been removed. Distributions that want to maintain
1439 compatibility with the old name should add a symlink from
1440 the old name to the new name.
1442 * journald's SplitMode= now defaults to "uid". This makes sure
1443 that unprivileged users can access their own coredumps with
1444 coredumpctl without restrictions.
1446 * New kernel command line options "systemd.wants=" (for
1447 pulling an additional unit during boot), "systemd.mask="
1448 (for masking a specific unit for the boot), and
1449 "systemd.debug-shell" (for enabling the debug shell on tty9)
1450 have been added. This is implemented in the new generator
1451 "systemd-debug-generator".
1453 * systemd-nspawn will now by default filter a couple of
1454 syscalls for containers, among them those required for
1455 kernel module loading, direct x86 IO port access, swap
1456 management, and kexec. Most importantly though
1457 open_by_handle_at() is now prohibited for containers,
1458 closing a hole similar to a recently discussed vulnerability
1459 in docker regarding access to files on file hierarchies the
1460 container should normally not have access to. Note that for
1461 nspawn we generally make no security claims anyway (and
1462 this is explicitly documented in the man page), so this is
1463 just a fix for one of the most obvious problems.
1465 * A new man page file-hierarchy(7) has been added that
1466 contains a minimized, modernized version of the file system
1467 layout systemd expects, similar in style to the FHS
1468 specification or hier(5). A new tool systemd-path(1) has
1469 been added to query many of these paths for the local
1472 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
1473 longer done. Since the directory now has a per-user size
1474 limit, and is cleaned on logout this appears unnecessary,
1475 in particular since this now brings the lifecycle of this
1476 directory closer in line with how IPC objects are handled.
1478 * systemd.pc now exports a number of additional directories,
1479 including $libdir (which is useful to identify the library
1480 path for the primary architecture of the system), and a
1481 couple of drop-in directories.
1483 * udev's predictable network interface names now use the dev_port
1484 sysfs attribute, introduced in linux 3.15 instead of dev_id to
1485 distinguish between ports of the same PCI function. dev_id should
1486 only be used for ports using the same HW address, hence the need
1489 * machined has been updated to export the OS version of a
1490 container (read from /etc/os-release and
1491 /usr/lib/os-release) on the bus. This is now shown in
1492 "machinectl status" for a machine.
1494 * A new service setting RestartForceExitStatus= has been
1495 added. If configured to a set of exit signals or process
1496 return values, the service will be restarted when the main
1497 daemon process exits with any of them, regardless of the
1500 * systemctl's -H switch for connecting to remote systemd
1501 machines has been extended so that it may be used to
1502 directly connect to a specific container on the
1503 host. "systemctl -H root@foobar:waldi" will now connect as
1504 user "root" to host "foobar", and then proceed directly to
1505 the container named "waldi". Note that currently you have to
1506 authenticate as user "root" for this to work, as entering
1507 containers is a privileged operation.
1509 Contributions from: Andreas Henriksson, Benjamin Steinwender,
1510 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
1511 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
1512 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
1513 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
1514 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
1515 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
1516 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
1517 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
1518 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
1519 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
1520 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
1522 -- Berlin, 2014-07-03
1526 * As an experimental feature, udev now tries to lock the
1527 disk device node (flock(LOCK_SH|LOCK_NB)) while it
1528 executes events for the disk or any of its partitions.
1529 Applications like partitioning programs can lock the
1530 disk device node (flock(LOCK_EX)) and claim temporary
1531 device ownership that way; udev will entirely skip all event
1532 handling for this disk and its partitions. If the disk
1533 was opened for writing, the close will trigger a partition
1534 table rescan in udev's "watch" facility, and if needed
1535 synthesize "change" events for the disk and all its partitions.
1536 This is now unconditionally enabled, and if it turns out to
1537 cause major problems, we might turn it on only for specific
1538 devices, or might need to disable it entirely. Device Mapper
1539 devices are excluded from this logic.
1541 * We temporarily dropped the "-l" switch for fsck invocations,
1542 since they collide with the flock() logic above. util-linux
1543 upstream has been changed already to avoid this conflict,
1544 and we will readd "-l" as soon as util-linux with this
1545 change has been released.
1547 * The dependency on libattr has been removed. Since a long
1548 time, the extended attribute calls have moved to glibc, and
1549 libattr is thus unnecessary.
1551 * Virtualization detection works without priviliges now. This
1552 means the systemd-detect-virt binary no longer requires
1553 CAP_SYS_PTRACE file capabilities, and our daemons can run
1554 with fewer privileges.
1556 * systemd-networkd now runs under its own "systemd-network"
1557 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
1558 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
1559 loses the ability to write to files owned by root this way.
1561 * Similar, systemd-resolved now runs under its own
1562 "systemd-resolve" user with no capabilities remaining.
1564 * Similar, systemd-bus-proxyd now runs under its own
1565 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
1567 * systemd-networkd gained support for setting up "veth"
1568 virtual ethernet devices for container connectivity, as well
1569 as GRE and VTI tunnels.
1571 * systemd-networkd will no longer automatically attempt to
1572 manually load kernel modules necessary for certain tunnel
1573 transports. Instead, it is assumed the kernel loads them
1574 automatically when required. This only works correctly on
1575 very new kernels. On older kernels, please consider adding
1576 the kernel modules to /etc/modules-load.d/ as a work-around.
1578 * The resolv.conf file systemd-resolved generates has been
1579 moved to /run/systemd/resolve/. If you have a symlink from
1580 /etc/resolv.conf, it might be necessary to correct it.
1582 * Two new service settings, ProtectHome= and ProtectSystem=,
1583 have been added. When enabled, they will make the user data
1584 (such as /home) inaccessible or read-only and the system
1585 (such as /usr) read-only, for specific services. This allows
1586 very light-weight per-service sandboxing to avoid
1587 modifications of user data or system files from
1588 services. These two new switches have been enabled for all
1589 of systemd's long-running services, where appropriate.
1591 * Socket units gained new SocketUser= and SocketGroup=
1592 settings to set the owner user and group of AF_UNIX sockets
1593 and FIFOs in the file system.
1595 * Socket units gained a new RemoveOnStop= setting. If enabled,
1596 all FIFOS and sockets in the file system will be removed
1597 when the specific socket unit is stopped.
1599 * Socket units gained a new Symlinks= setting. It takes a list
1600 of symlinks to create to file system sockets or FIFOs
1601 created by the specific Unix sockets. This is useful to
1602 manage symlinks to socket nodes with the same life-cycle as
1605 * The /dev/log socket and /dev/initctl FIFO have been moved to
1606 /run, and have been replaced by symlinks. This allows
1607 connecting to these facilities even if PrivateDevices=yes is
1608 used for a service (which makes /dev/log itself unavailable,
1609 but /run is left). This also has the benefit of ensuring
1610 that /dev only contains device nodes, directories and
1611 symlinks, and nothing else.
1613 * sd-daemon gained two new calls sd_pid_notify() and
1614 sd_pid_notifyf(). They are similar to sd_notify() and
1615 sd_notifyf(), but allow overriding of the source PID of
1616 notification messages if permissions permit this. This is
1617 useful to send notify messages on behalf of a different
1618 process (for example, the parent process). The
1619 systemd-notify tool has been updated to make use of this
1620 when sending messages (so that notification messages now
1621 originate from the shell script invoking systemd-notify and
1622 not the systemd-notify process itself. This should minimize
1623 a race where systemd fails to associate notification
1624 messages to services when the originating process already
1627 * A new "on-abnormal" setting for Restart= has been added. If
1628 set, it will result in automatic restarts on all "abnormal"
1629 reasons for a process to exit, which includes unclean
1630 signals, core dumps, timeouts and watchdog timeouts, but
1631 does not include clean and unclean exit codes or clean
1632 signals. Restart=on-abnormal is an alternative for
1633 Restart=on-failure for services that shall be able to
1634 terminate and avoid restarts on certain errors, by
1635 indicating so with an unclean exit code. Restart=on-failure
1636 or Restart=on-abnormal is now the recommended setting for
1637 all long-running services.
1639 * If the InaccessibleDirectories= service setting points to a
1640 mount point (or if there are any submounts contained within
1641 it), it is now attempted to completely unmount it, to make
1642 the file systems truly unavailable for the respective
1645 * The ReadOnlyDirectories= service setting and
1646 systemd-nspawn's --read-only parameter are now recursively
1647 applied to all submounts, too.
1649 * Mount units may now be created transiently via the bus APIs.
1651 * The support for SysV and LSB init scripts has been removed
1652 from the systemd daemon itself. Instead, it is now
1653 implemented as a generator that creates native systemd units
1654 from these scripts when needed. This enables us to remove a
1655 substantial amount of legacy code from PID 1, following the
1656 fact that many distributions only ship a very small number
1657 of LSB/SysV init scripts nowadays.
1659 * Privileged Xen (dom0) domains are not considered
1660 virtualization anymore by the virtualization detection
1661 logic. After all, they generally have unrestricted access to
1662 the hardware and usually are used to manage the unprivileged
1665 * systemd-tmpfiles gained a new "C" line type, for copying
1666 files or entire directories.
1668 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
1669 lines. So far, they have been non-globbing versions of the
1670 latter, and have thus been redundant. In future, it is
1671 recommended to only use "z". "m" has hence been removed
1672 from the documentation, even though it stays supported.
1674 * A tmpfiles snippet to recreate the most basic structure in
1675 /var has been added. This is enough to create the /var/run →
1676 /run symlink and create a couple of structural
1677 directories. This allows systems to boot up with an empty or
1678 volatile /var. Of course, while with this change, the core OS
1679 now is capable with dealing with a volatile /var, not all
1680 user services are ready for it. However, we hope that sooner
1681 or later, many service daemons will be changed upstream so
1682 that they are able to automatically create their necessary
1683 directories in /var at boot, should they be missing. This is
1684 the first step to allow state-less systems that only require
1685 the vendor image for /usr to boot.
1687 * systemd-nspawn has gained a new --tmpfs= switch to mount an
1688 empty tmpfs instance to a specific directory. This is
1689 particularly useful for making use of the automatic
1690 reconstruction of /var (see above), by passing --tmpfs=/var.
1692 * Access modes specified in tmpfiles snippets may now be
1693 prefixed with "~", which indicates that they shall be masked
1694 by whether the existing file or directory is currently
1695 writable, readable or executable at all. Also, if specified,
1696 the sgid/suid/sticky bits will be masked for all
1699 * A new passive target unit "network-pre.target" has been
1700 added which is useful for services that shall run before any
1701 network is configured, for example firewall scripts.
1703 * The "floppy" group that previously owned the /dev/fd*
1704 devices is no longer used. The "disk" group is now used
1705 instead. Distributions should probably deprecate usage of
1708 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
1709 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
1710 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
1711 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
1712 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
1713 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
1716 -- Berlin, 2014-06-11
1720 * A new "systemd-timesyncd" daemon has been added for
1721 synchronizing the system clock across the network. It
1722 implements an SNTP client. In contrast to NTP
1723 implementations such as chrony or the NTP reference server,
1724 this only implements a client side, and does not bother with
1725 the full NTP complexity, focusing only on querying time from
1726 one remote server and synchronizing the local clock to
1727 it. Unless you intend to serve NTP to networked clients or
1728 want to connect to local hardware clocks, this simple NTP
1729 client should be more than appropriate for most
1730 installations. The daemon runs with minimal privileges, and
1731 has been hooked up with networkd to only operate when
1732 network connectivity is available. The daemon saves the
1733 current clock to disk every time a new NTP sync has been
1734 acquired, and uses this to possibly correct the system clock
1735 early at bootup, in order to accommodate for systems that
1736 lack an RTC such as the Raspberry Pi and embedded devices,
1737 and to make sure that time monotonically progresses on these
1738 systems, even if it is not always correct. To make use of
1739 this daemon, a new system user and group "systemd-timesync"
1740 needs to be created on installation of systemd.
1742 * The queue "seqnum" interface of libudev has been disabled, as
1743 it was generally incompatible with device namespacing as
1744 sequence numbers of devices go "missing" if the devices are
1745 part of a different namespace.
1747 * "systemctl list-timers" and "systemctl list-sockets" gained
1748 a --recursive switch for showing units of these types also
1749 for all local containers, similar in style to the already
1750 supported --recursive switch for "systemctl list-units".
1752 * A new RebootArgument= setting has been added for service
1753 units, which may be used to specify a kernel reboot argument
1754 to use when triggering reboots with StartLimitAction=.
1756 * A new FailureAction= setting has been added for service
1757 units which may be used to specify an operation to trigger
1758 when a service fails. This works similarly to
1759 StartLimitAction=, but unlike it, controls what is done
1760 immediately rather than only after several attempts to
1761 restart the service in question.
1763 * hostnamed got updated to also expose the kernel name,
1764 release, and version on the bus. This is useful for
1765 executing commands like hostnamectl with the -H switch.
1766 systemd-analyze makes use of this to properly display
1767 details when running non-locally.
1769 * The bootchart tool can now show cgroup information in the
1770 graphs it generates.
1772 * The CFS CPU quota cgroup attribute is now exposed for
1773 services. The new CPUQuota= switch has been added for this
1774 which takes a percentage value. Setting this will have the
1775 result that a service may never get more CPU time than the
1776 specified percentage, even if the machine is otherwise idle.
1778 * systemd-networkd learned IPIP and SIT tunnel support.
1780 * LSB init scripts exposing a dependency on $network will now
1781 get a dependency on network-online.target rather than simply
1782 network.target. This should bring LSB handling closer to
1783 what it was on SysV systems.
1785 * A new fsck.repair= kernel option has been added to control
1786 how fsck shall deal with unclean file systems at boot.
1788 * The (.ini) configuration file parser will now silently
1789 ignore sections whose name begins with "X-". This may be
1790 used to maintain application-specific extension sections in unit
1793 * machined gained a new API to query the IP addresses of
1794 registered containers. "machinectl status" has been updated
1795 to show these addresses in its output.
1797 * A new call sd_uid_get_display() has been added to the
1798 sd-login APIs for querying the "primary" session of a
1799 user. The "primary" session of the user is elected from the
1800 user's sessions and generally a graphical session is
1801 preferred over a text one.
1803 * A minimal systemd-resolved daemon has been added. It
1804 currently simply acts as a companion to systemd-networkd and
1805 manages resolv.conf based on per-interface DNS
1806 configuration, possibly supplied via DHCP. In the long run
1807 we hope to extend this into a local DNSSEC enabled DNS and
1810 * The systemd-networkd-wait-online tool is now enabled by
1811 default. It will delay network-online.target until a network
1812 connection has been configured. The tool primarily integrates
1813 with networkd, but will also make a best effort to make sense
1814 of network configuration performed in some other way.
1816 * Two new service options StartupCPUShares= and
1817 StartupBlockIOWeight= have been added that work similarly to
1818 CPUShares= and BlockIOWeight= however only apply during
1819 system startup. This is useful to prioritize certain services
1820 differently during bootup than during normal runtime.
1822 * hostnamed has been changed to prefer the statically
1823 configured hostname in /etc/hostname (unless set to
1824 'localhost' or empty) over any dynamic one supplied by
1825 dhcp. With this change, the rules for picking the hostname
1826 match more closely the rules of other configuration settings
1827 where the local administrator's configuration in /etc always
1828 overrides any other settings.
1830 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
1831 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
1832 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
1833 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
1834 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
1835 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
1836 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
1837 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
1838 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
1839 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
1840 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
1841 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
1842 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
1843 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
1844 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
1845 Lindskog, WaLyong Cho, Will Woods, Zbigniew
1848 -- Beijing, 2014-05-28
1852 * When restoring the screen brightness at boot, stay away from
1853 the darkest setting or from the lowest 5% of the available
1854 range, depending on which is the larger value of both. This
1855 should effectively protect the user from rebooting into a
1856 black screen, should the brightness have been set to minimum
1859 * sd-login gained a new sd_machine_get_class() call to
1860 determine the class ("vm" or "container") of a machine
1861 registered with machined.
1863 * sd-login gained new calls
1864 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
1865 to query the identity of the peer of a local AF_UNIX
1866 connection. They operate similarly to their sd_pid_get_xyz()
1869 * PID 1 will now maintain a system-wide system state engine
1870 with the states "starting", "running", "degraded",
1871 "maintenance", "stopping". These states are bound to system
1872 startup, normal runtime, runtime with at least one failed
1873 service, rescue/emergency mode and system shutdown. This
1874 state is shown in the "systemctl status" output when no unit
1875 name is passed. It is useful to determine system state, in
1876 particularly when doing so for many systems or containers at
1879 * A new command "list-machines" has been added to "systemctl"
1880 that lists all local OS containers and shows their system
1881 state (see above), if systemd runs inside of them.
1883 * systemctl gained a new "-r" switch to recursively enumerate
1884 units on all local containers, when used with the
1885 "list-unit" command (which is the default one that is
1886 executed when no parameters are specified).
1888 * The GPT automatic partition discovery logic will now honour
1889 two GPT partition flags: one may be set on a partition to
1890 cause it to be mounted read-only, and the other may be set
1891 on a partition to ignore it during automatic discovery.
1893 * Two new GPT type UUIDs have been added for automatic root
1894 partition discovery, for 32-bit and 64-bit ARM. This is not
1895 particularly useful for discovering the root directory on
1896 these architectures during bare-metal boots (since UEFI is
1897 not common there), but still very useful to allow booting of
1898 ARM disk images in nspawn with the -i option.
1900 * MAC addresses of interfaces created with nspawn's
1901 --network-interface= switch will now be generated from the
1902 machine name, and thus be stable between multiple invocations
1905 * logind will now automatically remove all IPC objects owned
1906 by a user if she or he fully logs out. This makes sure that
1907 users who are logged out cannot continue to consume IPC
1908 resources. This covers SysV memory, semaphores and message
1909 queues as well as POSIX shared memory and message
1910 queues. Traditionally, SysV and POSIX IPC had no life-cycle
1911 limits. With this functionality, that is corrected. This may
1912 be turned off by using the RemoveIPC= switch of logind.conf.
1914 * The systemd-machine-id-setup and tmpfiles tools gained a
1915 --root= switch to operate on a specific root directory,
1918 * journald can now forward logged messages to the TTYs of all
1919 logged in users ("wall"). This is the default for all
1920 emergency messages now.
1922 * A new tool systemd-journal-remote has been added to stream
1923 journal log messages across the network.
1925 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
1926 controller trees are mounted into it. Note that the
1927 directories mounted beneath it are not read-only. This is a
1928 security measure and is particularly useful because glibc
1929 actually includes a search logic to pick any tmpfs it can
1930 find to implement shm_open() if /dev/shm is not available
1931 (which it might very well be in namespaced setups).
1933 * machinectl gained a new "poweroff" command to cleanly power
1934 down a local OS container.
1936 * The PrivateDevices= unit file setting will now also drop the
1937 CAP_MKNOD capability from the capability bound set, and
1938 imply DevicePolicy=closed.
1940 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
1941 comprehensively on all long-running systemd services where
1942 this is appropriate.
1944 * systemd-udevd will now run in a disassociated mount
1945 namespace. To mount directories from udev rules, make sure to
1946 pull in mount units via SYSTEMD_WANTS properties.
1948 * The kdbus support gained support for uploading policy into
1949 the kernel. sd-bus gained support for creating "monitoring"
1950 connections that can eavesdrop into all bus communication
1951 for debugging purposes.
1953 * Timestamps may now be specified in seconds since the UNIX
1954 epoch Jan 1st, 1970 by specifying "@" followed by the value
1957 * Native tcpwrap support in systemd has been removed. tcpwrap
1958 is old code, not really maintained anymore and has serious
1959 shortcomings, and better options such as firewalls
1960 exist. For setups that require tcpwrap usage, please
1961 consider invoking your socket-activated service via tcpd,
1962 like on traditional inetd.
1964 * A new system.conf configuration option
1965 DefaultTimerAccuracySec= has been added that controls the
1966 default AccuracySec= setting of .timer units.
1968 * Timer units gained a new WakeSystem= switch. If enabled,
1969 timers configured this way will cause the system to resume
1970 from system suspend (if the system supports that, which most
1973 * Timer units gained a new Persistent= switch. If enabled,
1974 timers configured this way will save to disk when they have
1975 been last triggered. This information is then used on next
1976 reboot to possible execute overdue timer events, that
1977 could not take place because the system was powered off.
1978 This enables simple anacron-like behaviour for timer units.
1980 * systemctl's "list-timers" will now also list the time a
1981 timer unit was last triggered in addition to the next time
1982 it will be triggered.
1984 * systemd-networkd will now assign predictable IPv4LL
1985 addresses to its local interfaces.
1987 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
1988 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
1989 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
1990 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
1991 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
1992 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
1993 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
1994 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
1997 -- Berlin, 2014-03-25
2001 * A new unit file setting RestrictAddressFamilies= has been
2002 added to restrict which socket address families unit
2003 processes gain access to. This takes address family names
2004 like "AF_INET" or "AF_UNIX", and is useful to minimize the
2005 attack surface of services via exotic protocol stacks. This
2006 is built on seccomp system call filters.
2008 * Two new unit file settings RuntimeDirectory= and
2009 RuntimeDirectoryMode= have been added that may be used to
2010 manage a per-daemon runtime directories below /run. This is
2011 an alternative for setting up directory permissions with
2012 tmpfiles snippets, and has the advantage that the runtime
2013 directory's lifetime is bound to the daemon runtime and that
2014 the daemon starts up with an empty directory each time. This
2015 is particularly useful when writing services that drop
2016 privileges using the User= or Group= setting.
2018 * The DeviceAllow= unit setting now supports globbing for
2019 matching against device group names.
2021 * The systemd configuration file system.conf gained new
2022 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
2023 DefaultMemoryAccounting= to globally turn on/off accounting
2024 for specific resources (cgroups) for all units. These
2025 settings may still be overridden individually in each unit
2028 * systemd-gpt-auto-generator is now able to discover /srv and
2029 root partitions in addition to /home and swap partitions. It
2030 also supports LUKS-encrypted partitions now. With this in
2031 place, automatic discovery of partitions to mount following
2032 the Discoverable Partitions Specification
2033 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
2034 is now a lot more complete. This allows booting without
2035 /etc/fstab and without root= on the kernel command line on
2036 systems prepared appropriately.
2038 * systemd-nspawn gained a new --image= switch which allows
2039 booting up disk images and Linux installations on any block
2040 device that follow the Discoverable Partitions Specification
2041 (see above). This means that installations made with
2042 appropriately updated installers may now be started and
2043 deployed using container managers, completely
2044 unmodified. (We hope that libvirt-lxc will add support for
2045 this feature soon, too.)
2047 * systemd-nspawn gained a new --network-macvlan= setting to
2048 set up a private macvlan interface for the
2049 container. Similarly, systemd-networkd gained a new
2050 Kind=macvlan setting in .netdev files.
2052 * systemd-networkd now supports configuring local addresses
2055 * A new tool systemd-network-wait-online has been added to
2056 synchronously wait for network connectivity using
2059 * The sd-bus.h bus API gained a new sd_bus_track object for
2060 tracking the life-cycle of bus peers. Note that sd-bus.h is
2061 still not a public API though (unless you specify
2062 --enable-kdbus on the configure command line, which however
2063 voids your warranty and you get no API stability guarantee).
2065 * The $XDG_RUNTIME_DIR runtime directories for each user are
2066 now individual tmpfs instances, which has the benefit of
2067 introducing separate pools for each user, with individual
2068 size limits, and thus making sure that unprivileged clients
2069 can no longer negatively impact the system or other users by
2070 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
2071 RuntimeDirectorySize= has been introduced that allows
2072 controlling the default size limit for all users. It
2073 defaults to 10% of the available physical memory. This is no
2074 replacement for quotas on tmpfs though (which the kernel
2075 still does not support), as /dev/shm and /tmp are still
2076 shared resources used by both the system and unprivileged
2079 * logind will now automatically turn off automatic suspending
2080 on laptop lid close when more than one display is
2081 connected. This was previously expected to be implemented
2082 individually in desktop environments (such as GNOME),
2083 however has been added to logind now, in order to fix a
2084 boot-time race where a desktop environment might not have
2085 been started yet and thus not been able to take an inhibitor
2086 lock at the time where logind already suspends the system
2087 due to a closed lid.
2089 * logind will now wait at least 30s after each system
2090 suspend/resume cycle, and 3min after system boot before
2091 suspending the system due to a closed laptop lid. This
2092 should give USB docking stations and similar enough time to
2093 be probed and configured after system resume and boot in
2094 order to then act as suspend blocker.
2096 * systemd-run gained a new --property= setting which allows
2097 initialization of resource control properties (and others)
2098 for the created scope or service unit. Example: "systemd-run
2099 --property=BlockIOWeight=10 updatedb" may be used to run
2100 updatedb at a low block IO scheduling weight.
2102 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
2103 now also work in --scope mode.
2105 * When systemd is compiled with kdbus support, basic support
2106 for enforced policies is now in place. (Note that enabling
2107 kdbus still voids your warranty and no API compatibility
2110 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
2111 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
2112 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
2113 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
2114 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
2115 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
2116 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
2117 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
2118 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
2119 Zbigniew Jędrzejewski-Szmek
2121 -- Berlin, 2014-03-12
2125 * systemd will now relabel /dev after loading the SMACK policy
2126 according to SMACK rules.
2128 * A new unit file option AppArmorProfile= has been added to
2129 set the AppArmor profile for the processes of a unit.
2131 * A new condition check ConditionArchitecture= has been added
2132 to conditionalize units based on the system architecture, as
2133 reported by uname()'s "machine" field.
2135 * systemd-networkd now supports matching on the system
2136 virtualization, architecture, kernel command line, host name
2139 * logind is now a lot more aggressive when suspending the
2140 machine due to a closed laptop lid. Instead of acting only
2141 on the lid close action, it will continuously watch the lid
2142 status and act on it. This is useful for laptops where the
2143 power button is on the outside of the chassis so that it can
2144 be reached without opening the lid (such as the Lenovo
2145 Yoga). On those machines, logind will now immediately
2146 re-suspend the machine if the power button has been
2147 accidentally pressed while the laptop was suspended and in a
2148 backpack or similar.
2150 * logind will now watch SW_DOCK switches and inhibit reaction
2151 to the lid switch if it is pressed. This means that logind
2152 will not suspend the machine anymore if the lid is closed
2153 and the system is docked, if the laptop supports SW_DOCK
2154 notifications via the input layer. Note that ACPI docking
2155 stations do not generate this currently. Also note that this
2156 logic is usually not fully sufficient and Desktop
2157 Environments should take a lid switch inhibitor lock when an
2158 external display is connected, as systemd will not watch
2161 * nspawn will now make use of the devices cgroup controller by
2162 default, and only permit creation of and access to the usual
2163 API device nodes like /dev/null or /dev/random, as well as
2164 access to (but not creation of) the pty devices.
2166 * We will now ship a default .network file for
2167 systemd-networkd that automatically configures DHCP for
2168 network interfaces created by nspawn's --network-veth or
2169 --network-bridge= switches.
2171 * systemd will now understand the usual M, K, G, T suffixes
2172 according to SI conventions (i.e. to the base 1000) when
2173 referring to throughput and hardware metrics. It will stay
2174 with IEC conventions (i.e. to the base 1024) for software
2175 metrics, according to what is customary according to
2176 Wikipedia. We explicitly document which base applies for
2177 each configuration option.
2179 * The DeviceAllow= setting in unit files now supports a syntax
2180 to whitelist an entire group of devices node majors at once,
2181 based on the /proc/devices listing. For example, with the
2182 string "char-pts", it is now possible to whitelist all
2183 current and future pseudo-TTYs at once.
2185 * sd-event learned a new "post" event source. Event sources of
2186 this type are triggered by the dispatching of any event
2187 source of a type that is not "post". This is useful for
2188 implementing clean-up and check event sources that are
2189 triggered by other work being done in the program.
2191 * systemd-networkd is no longer statically enabled, but uses
2192 the usual [Install] sections so that it can be
2193 enabled/disabled using systemctl. It still is enabled by
2196 * When creating a veth interface pair with systemd-nspawn, the
2197 host side will now be prefixed with "vb-" if
2198 --network-bridge= is used, and with "ve-" if --network-veth
2199 is used. This way, it is easy to distinguish these cases on
2200 the host, for example to apply different configuration to
2201 them with systemd-networkd.
2203 * The compatibility libraries for libsystemd-journal.so,
2204 libsystem-id128.so, libsystemd-login.so and
2205 libsystemd-daemon.so do not make use of IFUNC
2206 anymore. Instead, we now build libsystemd.so multiple times
2207 under these alternative names. This means that the footprint
2208 is drastically increased, but given that these are
2209 transitional compatibility libraries, this should not matter
2210 much. This change has been made necessary to support the ARM
2211 platform for these compatibility libraries, as the ARM
2212 toolchain is not really at the same level as the toolchain
2213 for other architectures like x86 and does not support
2214 IFUNC. Please make sure to use --enable-compat-libs only
2215 during a transitional period!
2217 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
2218 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
2219 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
2220 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
2221 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
2222 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
2223 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
2224 Zbigniew Jędrzejewski-Szmek
2226 -- Berlin, 2014-02-24
2230 * A new component "systemd-networkd" has been added that can
2231 be used to configure local network interfaces statically or
2232 via DHCP. It is capable of bringing up bridges, VLANs, and
2233 bonding. Currently, no hook-ups for interactive network
2234 configuration are provided. Use this for your initrd,
2235 container, embedded, or server setup if you need a simple,
2236 yet powerful, network configuration solution. This
2237 configuration subsystem is quite nifty, as it allows wildcard
2238 hotplug matching in interfaces. For example, with a single
2239 configuration snippet, you can configure that all Ethernet
2240 interfaces showing up are automatically added to a bridge,
2241 or similar. It supports link-sensing and more.
2243 * A new tool "systemd-socket-proxyd" has been added which can
2244 act as a bidirectional proxy for TCP sockets. This is
2245 useful for adding socket activation support to services that
2246 do not actually support socket activation, including virtual
2247 machines and the like.
2249 * Add a new tool to save/restore rfkill state on
2252 * Save/restore state of keyboard backlights in addition to
2253 display backlights on shutdown/boot.
2255 * udev learned a new SECLABEL{} construct to label device
2256 nodes with a specific security label when they appear. For
2257 now, only SECLABEL{selinux} is supported, but the syntax is
2258 prepared for additional security frameworks.
2260 * udev gained a new scheme to configure link-level attributes
2261 from files in /etc/systemd/network/*.link. These files can
2262 match against MAC address, device path, driver name and type,
2263 and will apply attributes like the naming policy, link speed,
2264 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
2265 address assignment policy (randomized, ...).
2267 * The configuration of network interface naming rules for
2268 "permanent interface names" has changed: a new NamePolicy=
2269 setting in the [Link] section of .link files determines the
2270 priority of possible naming schemes (onboard, slot, mac,
2271 path). The default value of this setting is determined by
2272 /usr/lib/net/links/99-default.link. Old
2273 80-net-name-slot.rules udev configuration file has been
2274 removed, so local configuration overriding this file should
2275 be adapated to override 99-default.link instead.
2277 * When the User= switch is used in a unit file, also
2278 initialize $SHELL= based on the user database entry.
2280 * systemd no longer depends on libdbus. All communication is
2281 now done with sd-bus, systemd's low-level bus library
2284 * kdbus support has been added to PID 1 itself. When kdbus is
2285 enabled, this causes PID 1 to set up the system bus and
2286 enable support for a new ".busname" unit type that
2287 encapsulates bus name activation on kdbus. It works a little
2288 bit like ".socket" units, except for bus names. A new
2289 generator has been added that converts classic dbus1 service
2290 activation files automatically into native systemd .busname
2293 * sd-bus: add a light-weight vtable implementation that allows
2294 defining objects on the bus with a simple static const
2295 vtable array of its methods, signals and properties.
2297 * systemd will not generate or install static dbus
2298 introspection data anymore to /usr/share/dbus-1/interfaces,
2299 as the precise format of these files is unclear, and
2300 nothing makes use of it.
2302 * A proxy daemon is now provided to proxy clients connecting
2303 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
2304 compatibility with classic D-Bus.
2306 * A bus driver implementation has been added that supports the
2307 classic D-Bus bus driver calls on kdbus, also for
2308 compatibility purposes.
2310 * A new API "sd-event.h" has been added that implements a
2311 minimal event loop API built around epoll. It provides a
2312 couple of features that direct epoll usage is lacking:
2313 prioritization of events, scales to large numbers of timer
2314 events, per-event timer slack (accuracy), system-wide
2315 coalescing of timer events, exit handlers, watchdog
2316 supervision support using systemd's sd_notify() API, child
2319 * A new API "sd-rntl.h" has been added that provides an API
2320 around the route netlink interface of the kernel, similar in
2321 style to "sd-bus.h".
2323 * A new API "sd-dhcp-client.h" has been added that provides a
2324 small DHCPv4 client-side implementation. This is used by
2327 * There is a new kernel command line option
2328 "systemd.restore_state=0|1". When set to "0", none of the
2329 systemd tools will restore saved runtime state to hardware
2330 devices. More specifically, the rfkill and backlight states
2333 * The FsckPassNo= compatibility option in mount/service units
2334 has been removed. The fstab generator will now add the
2335 necessary dependencies automatically, and does not require
2336 PID1's support for that anymore.
2338 * journalctl gained a new switch, --list-boots, that lists
2339 recent boots with their times and boot IDs.
2341 * The various tools like systemctl, loginctl, timedatectl,
2342 busctl, systemd-run, ... have gained a new switch "-M" to
2343 connect to a specific, local OS container (as direct
2344 connection, without requiring SSH). This works on any
2345 container that is registered with machined, such as those
2346 created by libvirt-lxc or nspawn.
2348 * systemd-run and systemd-analyze also gained support for "-H"
2349 to connect to remote hosts via SSH. This is particularly
2350 useful for systemd-run because it enables queuing of jobs
2351 onto remote systems.
2353 * machinectl gained a new command "login" to open a getty
2354 login in any local container. This works with any container
2355 that is registered with machined (such as those created by
2356 libvirt-lxc or nspawn), and which runs systemd inside.
2358 * machinectl gained a new "reboot" command that may be used to
2359 trigger a reboot on a specific container that is registered
2360 with machined. This works on any container that runs an init
2361 system of some kind.
2363 * systemctl gained a new "list-timers" command to print a nice
2364 listing of installed timer units with the times they elapse
2367 * Alternative reboot() parameters may now be specified on the
2368 "systemctl reboot" command line and are passed to the
2369 reboot() system call.
2371 * systemctl gained a new --job-mode= switch to configure the
2372 mode to queue a job with. This is a more generic version of
2373 --fail, --irreversible, and --ignore-dependencies, which are
2374 still available but not advertised anymore.
2376 * /etc/systemd/system.conf gained new settings to configure
2377 various default timeouts of units, as well as the default
2378 start limit interval and burst. These may still be overridden
2381 * PID1 will now export on the bus profile data of the security
2382 policy upload process (such as the SELinux policy upload to
2385 * journald: when forwarding logs to the console, include
2386 timestamps (following the setting in
2387 /sys/module/printk/parameters/time).
2389 * OnCalendar= in timer units now understands the special
2390 strings "yearly" and "annually". (Both are equivalent)
2392 * The accuracy of timer units is now configurable with the new
2393 AccuracySec= setting. It defaults to 1min.
2395 * A new dependency type JoinsNamespaceOf= has been added that
2396 allows running two services within the same /tmp and network
2397 namespace, if PrivateNetwork= or PrivateTmp= are used.
2399 * A new command "cat" has been added to systemctl. It outputs
2400 the original unit file of a unit, and concatenates the
2401 contents of additional "drop-in" unit file snippets, so that
2402 the full configuration is shown.
2404 * systemctl now supports globbing on the various "list-xyz"
2405 commands, like "list-units" or "list-sockets", as well as on
2406 those commands which take multiple unit names.
2408 * journalctl's --unit= switch gained support for globbing.
2410 * All systemd daemons now make use of the watchdog logic so
2411 that systemd automatically notices when they hang.
2413 * If the $container_ttys environment variable is set,
2414 getty-generator will automatically spawn a getty for each
2415 listed tty. This is useful for container managers to request
2416 login gettys to be spawned on as many ttys as needed.
2418 * %h, %s, %U specifier support is not available anymore when
2419 used in unit files for PID 1. This is because NSS calls are
2420 not safe from PID 1. They stay available for --user
2421 instances of systemd, and as special case for the root user.
2423 * loginctl gained a new "--no-legend" switch to turn off output
2426 * The "sd-login.h" API gained three new calls:
2427 sd_session_is_remote(), sd_session_get_remote_user(),
2428 sd_session_get_remote_host() to query information about
2431 * The udev hardware database now also carries vendor/product
2432 information of SDIO devices.
2434 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
2435 determine whether watchdog notifications are requested by
2438 * Socket-activated per-connection services now include a
2439 short description of the connection parameters in the
2442 * tmpfiles gained a new "--boot" option. When this is not used,
2443 only lines where the command character is not suffixed with
2444 "!" are executed. When this option is specified, those
2445 options are executed too. This partitions tmpfiles
2446 directives into those that can be safely executed at any
2447 time, and those which should be run only at boot (for
2448 example, a line that creates /run/nologin).
2450 * A new API "sd-resolve.h" has been added which provides a simple
2451 asynchronous wrapper around glibc NSS host name resolution
2452 calls, such as getaddrinfo(). In contrast to glibc's
2453 getaddrinfo_a(), it does not use signals. In contrast to most
2454 other asynchronous name resolution libraries, this one does
2455 not reimplement DNS, but reuses NSS, so that alternate
2456 host name resolution systems continue to work, such as mDNS,
2457 LDAP, etc. This API is based on libasyncns, but it has been
2458 cleaned up for inclusion in systemd.
2460 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
2461 "sd-daemon.h" are no longer found in individual libraries
2462 libsystemd-journal.so, libsystemd-login.so,
2463 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
2464 merged them into a single library, libsystemd.so, which
2465 provides all symbols. The reason for this is cyclic
2466 dependencies, as these libraries tend to use each other's
2467 symbols. So far, we have managed to workaround that by linking
2468 a copy of a good part of our code into each of these
2469 libraries again and again, which, however, makes certain
2470 things hard to do, like sharing static variables. Also, it
2471 substantially increases footprint. With this change, there
2472 is only one library for the basic APIs systemd
2473 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
2474 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
2475 library as well, however are subject to the --enable-kdbus
2476 switch (see below). Note that "sd-dhcp-client.h" is not part
2477 of this library (this is because it only consumes, never
2478 provides, services of/to other APIs). To make the transition
2479 easy from the separate libraries to the unified one, we
2480 provide the --enable-compat-libs compile-time switch which
2481 will generate stub libraries that are compatible with the
2482 old ones but redirect all calls to the new one.
2484 * All of the kdbus logic and the new APIs "sd-bus.h",
2485 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
2486 and "sd-utf8.h" are compile-time optional via the
2487 "--enable-kdbus" switch, and they are not compiled in by
2488 default. To make use of kdbus, you have to explicitly enable
2489 the switch. Note however, that neither the kernel nor the
2490 userspace API for all of this is considered stable yet. We
2491 want to maintain the freedom to still change the APIs for
2492 now. By specifying this build-time switch, you acknowledge
2493 that you are aware of the instability of the current
2496 * Also, note that while kdbus is pretty much complete,
2497 it lacks one thing: proper policy support. This means you
2498 can build a fully working system with all features; however,
2499 it will be highly insecure. Policy support will be added in
2500 one of the next releases, at the same time that we will
2501 declare the APIs stable.
2503 * When the kernel command line argument "kdbus" is specified,
2504 systemd will automatically load the kdbus.ko kernel module. At
2505 this stage of development, it is only useful for testing kdbus
2506 and should not be used in production. Note: if "--enable-kdbus"
2507 is specified, and the kdbus.ko kernel module is available, and
2508 "kdbus" is added to the kernel command line, the entire system
2509 runs with kdbus instead of dbus-daemon, with the above mentioned
2510 problem of missing the system policy enforcement. Also a future
2511 version of kdbus.ko or a newer systemd will not be compatible with
2512 each other, and will unlikely be able to boot the machine if only
2513 one of them is updated.
2515 * systemctl gained a new "import-environment" command which
2516 uploads the caller's environment (or parts thereof) into the
2517 service manager so that it is inherited by services started
2518 by the manager. This is useful to upload variables like
2519 $DISPLAY into the user service manager.
2521 * A new PrivateDevices= switch has been added to service units
2522 which allows running a service with a namespaced /dev
2523 directory that does not contain any device nodes for
2524 physical devices. More specifically, it only includes devices
2525 such as /dev/null, /dev/urandom, and /dev/zero which are API
2528 * logind has been extended to support behaviour like VT
2529 switching on seats that do not support a VT. This makes
2530 multi-session available on seats that are not the first seat
2531 (seat0), and on systems where kernel support for VTs has
2532 been disabled at compile-time.
2534 * If a process holds a delay lock for system sleep or shutdown
2535 and fails to release it in time, we will now log its
2536 identity. This makes it easier to identify processes that
2537 cause slow suspends or power-offs.
2539 * When parsing /etc/crypttab, support for a new key-slot=
2540 option as supported by Debian is added. It allows indicating
2541 which LUKS slot to use on disk, speeding up key loading.
2543 * The sd_journald_sendv() API call has been checked and
2544 officially declared to be async-signal-safe so that it may
2545 be invoked from signal handlers for logging purposes.
2547 * Boot-time status output is now enabled automatically after a
2548 short timeout if boot does not progress, in order to give
2549 the user an indication what she or he is waiting for.
2551 * The boot-time output has been improved to show how much time
2552 remains until jobs expire.
2554 * The KillMode= switch in service units gained a new possible
2555 value "mixed". If set, and the unit is shut down, then the
2556 initial SIGTERM signal is sent only to the main daemon
2557 process, while the following SIGKILL signal is sent to
2558 all remaining processes of the service.
2560 * When a scope unit is registered, a new property "Controller"
2561 may be set. If set to a valid bus name, systemd will send a
2562 RequestStop() signal to this name when it would like to shut
2563 down the scope. This may be used to hook manager logic into
2564 the shutdown logic of scope units. Also, scope units may now
2565 be put in a special "abandoned" state, in which case the
2566 manager process which created them takes no further
2567 responsibilities for it.
2569 * When reading unit files, systemd will now verify
2570 the access mode of these files, and warn about certain
2571 suspicious combinations. This has been added to make it
2572 easier to track down packaging bugs where unit files are
2573 marked executable or world-writable.
2575 * systemd-nspawn gained a new "--setenv=" switch to set
2576 container-wide environment variables. The similar option in
2577 systemd-activate was renamed from "--environment=" to
2578 "--setenv=" for consistency.
2580 * systemd-nspawn has been updated to create a new kdbus domain
2581 for each container that is invoked, thus allowing each
2582 container to have its own set of system and user buses,
2583 independent of the host.
2585 * systemd-nspawn gained a new --drop-capability= switch to run
2586 the container with less capabilities than the default. Both
2587 --drop-capability= and --capability= now take the special
2588 string "all" for dropping or keeping all capabilities.
2590 * systemd-nspawn gained new switches for executing containers
2591 with specific SELinux labels set.
2593 * systemd-nspawn gained a new --quiet switch to not generate
2594 any additional output but the container's own console
2597 * systemd-nspawn gained a new --share-system switch to run a
2598 container without PID namespacing enabled.
2600 * systemd-nspawn gained a new --register= switch to control
2601 whether the container is registered with systemd-machined or
2602 not. This is useful for containers that do not run full
2603 OS images, but only specific apps.
2605 * systemd-nspawn gained a new --keep-unit which may be used
2606 when invoked as the only program from a service unit, and
2607 results in registration of the unit service itself in
2608 systemd-machined, instead of a newly opened scope unit.
2610 * systemd-nspawn gained a new --network-interface= switch for
2611 moving arbitrary interfaces to the container. The new
2612 --network-veth switch creates a virtual Ethernet connection
2613 between host and container. The new --network-bridge=
2614 switch then allows assigning the host side of this virtual
2615 Ethernet connection to a bridge device.
2617 * systemd-nspawn gained a new --personality= switch for
2618 setting the kernel personality for the container. This is
2619 useful when running a 32-bit container on a 64-bit host. A
2620 similar option Personality= is now also available for service
2623 * logind will now also track a "Desktop" identifier for each
2624 session which encodes the desktop environment of it. This is
2625 useful for desktop environments that want to identify
2626 multiple running sessions of itself easily.
2628 * A new SELinuxContext= setting for service units has been
2629 added that allows setting a specific SELinux execution
2630 context for a service.
2632 * Most systemd client tools will now honour $SYSTEMD_LESS for
2633 settings of the "less" pager. By default, these tools will
2634 override $LESS to allow certain operations to work, such as
2635 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
2636 influence this logic.
2638 * systemd's "seccomp" hook-up has been changed to make use of
2639 the libseccomp library instead of using its own
2640 implementation. This has benefits for portability among
2643 * For usage together with SystemCallFilter=, a new
2644 SystemCallErrorNumber= setting has been introduced that
2645 allows configuration of a system error number to be returned
2646 on filtered system calls, instead of immediately killing the
2647 process. Also, SystemCallArchitectures= has been added to
2648 limit access to system calls of a particular architecture
2649 (in order to turn off support for unused secondary
2650 architectures). There is also a global
2651 SystemCallArchitectures= setting in system.conf now to turn
2652 off support for non-native system calls system-wide.
2654 * systemd requires a kernel with a working name_to_handle_at(),
2655 please see the kernel config requirements in the README file.
2657 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
2658 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
2659 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
2660 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
2661 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
2662 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
2663 Elia Pinto, Florian Weimer, George McCollister, Goffredo
2664 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
2665 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
2666 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
2667 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
2668 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
2669 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
2670 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
2671 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
2672 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
2673 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
2674 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
2675 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
2676 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
2677 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
2678 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
2679 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
2680 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
2682 -- Berlin, 2014-02-20
2686 * logind has gained support for facilitating privileged input
2687 and drm device access for unprivileged clients. This work is
2688 useful to allow Wayland display servers (and similar
2689 programs, such as kmscon) to run under the user's ID and
2690 access input and drm devices which are normally
2691 protected. When this is used (and the kernel is new enough)
2692 logind will "mute" IO on the file descriptors passed to
2693 Wayland as long as it is in the background and "unmute" it
2694 if it returns into the foreground. This allows secure
2695 session switching without allowing background sessions to
2696 eavesdrop on input and display data. This also introduces
2697 session switching support if VT support is turned off in the
2698 kernel, and on seats that are not seat0.
2700 * A new kernel command line option luks.options= is understood
2701 now which allows specifying LUKS options for usage for LUKS
2702 encrypted partitions specified with luks.uuid=.
2704 * tmpfiles.d(5) snippets may now use specifier expansion in
2705 path names. More specifically %m, %b, %H, %v, are now
2706 replaced by the local machine id, boot id, hostname, and
2707 kernel version number.
2709 * A new tmpfiles.d(5) command "m" has been introduced which
2710 may be used to change the owner/group/access mode of a file
2711 or directory if it exists, but do nothing if it does not.
2713 * This release removes high-level support for the
2714 MemorySoftLimit= cgroup setting. The underlying kernel
2715 cgroup attribute memory.soft_limit= is currently badly
2716 designed and likely to be removed from the kernel API in its
2717 current form, hence we should not expose it for now.
2719 * The memory.use_hierarchy cgroup attribute is now enabled for
2720 all cgroups systemd creates in the memory cgroup
2721 hierarchy. This option is likely to be come the built-in
2722 default in the kernel anyway, and the non-hierarchical mode
2723 never made much sense in the intrinsically hierarchical
2726 * A new field _SYSTEMD_SLICE= is logged along with all journal
2727 messages containing the slice a message was generated
2728 from. This is useful to allow easy per-customer filtering of
2729 logs among other things.
2731 * systemd-journald will no longer adjust the group of journal
2732 files it creates to the "systemd-journal" group. Instead we
2733 rely on the journal directory to be owned by the
2734 "systemd-journal" group, and its setgid bit set, so that the
2735 kernel file system layer will automatically enforce that
2736 journal files inherit this group assignment. The reason for
2737 this change is that we cannot allow NSS look-ups from
2738 journald which would be necessary to resolve
2739 "systemd-journal" to a numeric GID, because this might
2740 create deadlocks if NSS involves synchronous queries to
2741 other daemons (such as nscd, or sssd) which in turn are
2742 logging clients of journald and might block on it, which
2743 would then dead lock. A tmpfiles.d(5) snippet included in
2744 systemd will make sure the setgid bit and group are
2745 properly set on the journal directory if it exists on every
2746 boot. However, we recommend adjusting it manually after
2747 upgrades too (or from RPM scriptlets), so that the change is
2748 not delayed until next reboot.
2750 * Backlight and random seed files in /var/lib/ have moved into
2751 the /var/lib/systemd/ directory, in order to centralize all
2752 systemd generated files in one directory.
2754 * Boot time performance measurements (as displayed by
2755 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
2756 performance information if that's available to determine how
2757 much time BIOS and boot loader initialization required. With
2758 a sufficiently new BIOS you hence no longer need to boot
2759 with Gummiboot to get access to such information.
2761 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
2762 Cristian Rodríguez, Dave Reisner, David Herrmann, David
2763 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
2764 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
2765 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
2766 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
2767 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
2769 -- Berlin, 2013-10-02
2773 * The Restart= option for services now understands a new
2774 on-watchdog setting, which will restart the service
2775 automatically if the service stops sending out watchdog keep
2776 alive messages (as configured with WatchdogSec=).
2778 * The getty generator (which is responsible for bringing up a
2779 getty on configured serial consoles) will no longer only
2780 start a getty on the primary kernel console but on all
2781 others, too. This makes the order in which console= is
2782 specified on the kernel command line less important.
2784 * libsystemd-logind gained a new sd_session_get_vt() call to
2785 retrieve the VT number of a session.
2787 * If the option "tries=0" is set for an entry of /etc/crypttab
2788 its passphrase is queried indefinitely instead of any
2789 maximum number of tries.
2791 * If a service with a configure PID file terminates its PID
2792 file will now be removed automatically if it still exists
2793 afterwards. This should put an end to stale PID files.
2795 * systemd-run will now also take relative binary path names
2796 for execution and no longer insists on absolute paths.
2798 * InaccessibleDirectories= and ReadOnlyDirectories= now take
2799 paths that are optionally prefixed with "-" to indicate that
2800 it should not be considered a failure if they do not exist.
2802 * journalctl -o (and similar commands) now understands a new
2803 output mode "short-precise", it is similar to "short" but
2804 shows timestamps with usec accuracy.
2806 * The option "discard" (as known from Debian) is now
2807 synonymous to "allow-discards" in /etc/crypttab. In fact,
2808 "discard" is preferred now (since it is easier to remember
2811 * Some licensing clean-ups were made, so that more code is now
2812 LGPL-2.1 licensed than before.
2814 * A minimal tool to save/restore the display backlight
2815 brightness across reboots has been added. It will store the
2816 backlight setting as late as possible at shutdown, and
2817 restore it as early as possible during reboot.
2819 * A logic to automatically discover and enable home and swap
2820 partitions on GPT disks has been added. With this in place
2821 /etc/fstab becomes optional for many setups as systemd can
2822 discover certain partitions located on the root disk
2823 automatically. Home partitions are recognized under their
2824 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
2825 partitions are recognized under their GPT type ID
2826 0657fd6da4ab43c484e50933c84b4f4f.
2828 * systemd will no longer pass any environment from the kernel
2829 or initrd to system services. If you want to set an
2830 environment for all services, do so via the kernel command
2831 line systemd.setenv= assignment.
2833 * The systemd-sysctl tool no longer natively reads the file
2834 /etc/sysctl.conf. If desired, the file should be symlinked
2835 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
2836 legacy support by a symlink rather than built-in code, it
2837 also makes the otherwise hidden order of application of the
2838 different files visible. (Note that this partly reverts to a
2839 pre-198 application order of sysctl knobs!)
2841 * The "systemctl set-log-level" and "systemctl dump" commands
2842 have been moved to systemd-analyze.
2844 * systemd-run learned the new --remain-after-exit switch,
2845 which causes the scope unit not to be cleaned up
2846 automatically after the process terminated.
2848 * tmpfiles learned a new --exclude-prefix= switch to exclude
2849 certain paths from operation.
2851 * journald will now automatically flush all messages to disk
2852 as soon as a message at the log level CRIT, ALERT or EMERG
2855 Contributions from: Andrew Cook, Brandon Philips, Christian
2856 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
2857 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
2858 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
2859 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
2860 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
2861 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
2862 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
2863 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
2864 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
2865 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
2866 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
2867 William Giokas, Zbigniew Jędrzejewski-Szmek
2869 -- Berlin, 2013-09-13
2873 * The documentation has been updated to cover the various new
2874 concepts introduced with 205.
2876 * Unit files now understand the new %v specifier which
2877 resolves to the kernel version string as returned by "uname
2880 * systemctl now supports filtering the unit list output by
2881 load state, active state and sub state, using the new
2884 * "systemctl status" will now show the results of the
2885 condition checks (like ConditionPathExists= and similar) of
2886 the last start attempts of the unit. They are also logged to
2889 * "journalctl -b" may now be used to look for boot output of a
2890 specific boot. Try "journalctl -b -1" for the previous boot,
2891 but the syntax is substantially more powerful.
2893 * "journalctl --show-cursor" has been added which prints the
2894 cursor string the last shown log line. This may then be used
2895 with the new "journalctl --after-cursor=" switch to continue
2896 browsing logs from that point on.
2898 * "journalctl --force" may now be used to force regeneration
2901 * Creation of "dead" device nodes has been moved from udev
2902 into kmod and tmpfiles. Previously, udev would read the kmod
2903 databases to pre-generate dead device nodes based on meta
2904 information contained in kernel modules, so that these would
2905 be auto-loaded on access rather then at boot. As this
2906 does not really have much to do with the exposing actual
2907 kernel devices to userspace this has always been slightly
2908 alien in the udev codebase. Following the new scheme kmod
2909 will now generate a runtime snippet for tmpfiles from the
2910 module meta information and it now is tmpfiles' job to the
2911 create the nodes. This also allows overriding access and
2912 other parameters for the nodes using the usual tmpfiles
2913 facilities. As side effect this allows us to remove the
2914 CAP_SYS_MKNOD capability bit from udevd entirely.
2916 * logind's device ACLs may now be applied to these "dead"
2917 devices nodes too, thus finally allowing managed access to
2918 devices such as /dev/snd/sequencer whithout loading the
2919 backing module right-away.
2921 * A new RPM macro has been added that may be used to apply
2922 tmpfiles configuration during package installation.
2924 * systemd-detect-virt and ConditionVirtualization= now can
2925 detect User-Mode-Linux machines (UML).
2927 * journald will now implicitly log the effective capabilities
2928 set of processes in the message metadata.
2930 * systemd-cryptsetup has gained support for TrueCrypt volumes.
2932 * The initrd interface has been simplified (more specifically,
2933 support for passing performance data via environment
2934 variables and fsck results via files in /run has been
2935 removed). These features were non-essential, and are
2936 nowadays available in a much nicer way by having systemd in
2937 the initrd serialize its state and have the hosts systemd
2938 deserialize it again.
2940 * The udev "keymap" data files and tools to apply keyboard
2941 specific mappings of scan to key codes, and force-release
2942 scan code lists have been entirely replaced by a udev
2943 "keyboard" builtin and a hwdb data file.
2945 * systemd will now honour the kernel's "quiet" command line
2946 argument also during late shutdown, resulting in a
2947 completely silent shutdown when used.
2949 * There's now an option to control the SO_REUSEPORT socket
2950 option in .socket units.
2952 * Instance units will now automatically get a per-template
2953 subslice of system.slice unless something else is explicitly
2954 configured. For example, instances of sshd@.service will now
2955 implicitly be placed in system-sshd.slice rather than
2956 system.slice as before.
2958 * Test coverage support may now be enabled at build time.
2960 Contributions from: Dave Reisner, Frederic Crozat, Harald
2961 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
2962 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
2963 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
2964 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
2965 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
2966 Giokas, Zbigniew Jędrzejewski-Szmek
2968 -- Berlin, 2013-07-23
2972 * Two new unit types have been introduced:
2974 Scope units are very similar to service units, however, are
2975 created out of pre-existing processes -- instead of PID 1
2976 forking off the processes. By using scope units it is
2977 possible for system services and applications to group their
2978 own child processes (worker processes) in a powerful way
2979 which then maybe used to organize them, or kill them
2980 together, or apply resource limits on them.
2982 Slice units may be used to partition system resources in an
2983 hierarchical fashion and then assign other units to them. By
2984 default there are now three slices: system.slice (for all
2985 system services), user.slice (for all user sessions),
2986 machine.slice (for VMs and containers).
2988 Slices and scopes have been introduced primarily in
2989 context of the work to move cgroup handling to a
2990 single-writer scheme, where only PID 1
2991 creates/removes/manages cgroups.
2993 * There's a new concept of "transient" units. In contrast to
2994 normal units these units are created via an API at runtime,
2995 not from configuration from disk. More specifically this
2996 means it is now possible to run arbitrary programs as
2997 independent services, with all execution parameters passed
2998 in via bus APIs rather than read from disk. Transient units
2999 make systemd substantially more dynamic then it ever was,
3000 and useful as a general batch manager.
3002 * logind has been updated to make use of scope and slice units
3003 for managing user sessions. As a user logs in he will get
3004 his own private slice unit, to which all sessions are added
3005 as scope units. We also added support for automatically
3006 adding an instance of user@.service for the user into the
3007 slice. Effectively logind will no longer create cgroup
3008 hierarchies on its own now, it will defer entirely to PID 1
3009 for this by means of scope, service and slice units. Since
3010 user sessions this way become entities managed by PID 1
3011 the output of "systemctl" is now a lot more comprehensive.
3013 * A new mini-daemon "systemd-machined" has been added which
3014 may be used by virtualization managers to register local
3015 VMs/containers. nspawn has been updated accordingly, and
3016 libvirt will be updated shortly. machined will collect a bit
3017 of meta information about the VMs/containers, and assign
3018 them their own scope unit (see above). The collected
3019 meta-data is then made available via the "machinectl" tool,
3020 and exposed in "ps" and similar tools. machined/machinectl
3021 is compile-time optional.
3023 * As discussed earlier, the low-level cgroup configuration
3024 options ControlGroup=, ControlGroupModify=,
3025 ControlGroupPersistent=, ControlGroupAttribute= have been
3026 removed. Please use high-level attribute settings instead as
3027 well as slice units.
3029 * A new bus call SetUnitProperties() has been added to alter
3030 various runtime parameters of a unit. This is primarily
3031 useful to alter cgroup parameters dynamically in a nice way,
3032 but will be extended later on to make more properties
3033 modifiable at runtime. systemctl gained a new set-properties
3034 command that wraps this call.
3036 * A new tool "systemd-run" has been added which can be used to
3037 run arbitrary command lines as transient services or scopes,
3038 while configuring a number of settings via the command
3039 line. This tool is currently very basic, however already
3040 very useful. We plan to extend this tool to even allow
3041 queuing of execution jobs with time triggers from the
3042 command line, similar in fashion to "at".
3044 * nspawn will now inform the user explicitly that kernels with
3045 audit enabled break containers, and suggest the user to turn
3048 * Support for detecting the IMA and AppArmor security
3049 frameworks with ConditionSecurity= has been added.
3051 * journalctl gained a new "-k" switch for showing only kernel
3052 messages, mimicking dmesg output; in addition to "--user"
3053 and "--system" switches for showing only user's own logs
3056 * systemd-delta can now show information about drop-in
3057 snippets extending unit files.
3059 * libsystemd-bus has been substantially updated but is still
3060 not available as public API.
3062 * systemd will now look for the "debug" argument on the kernel
3063 command line and enable debug logging, similar to what
3064 "systemd.log_level=debug" already did before.
3066 * "systemctl set-default", "systemctl get-default" has been
3067 added to configure the default.target symlink, which
3068 controls what to boot into by default.
3070 * "systemctl set-log-level" has been added as a convenient
3071 way to raise and lower systemd logging threshold.
3073 * "systemd-analyze plot" will now show the time the various
3074 generators needed for execution, as well as information
3075 about the unit file loading.
3077 * libsystemd-journal gained a new sd_journal_open_files() call
3078 for opening specific journal files. journactl also gained a
3079 new switch to expose this new functionality. Previously we
3080 only supported opening all files from a directory, or all
3081 files from the system, as opening individual files only is
3082 racy due to journal file rotation.
3084 * systemd gained the new DefaultEnvironment= setting in
3085 /etc/systemd/system.conf to set environment variables for
3088 * If a privileged process logs a journal message with the
3089 OBJECT_PID= field set, then journald will automatically
3090 augment this with additional OBJECT_UID=, OBJECT_GID=,
3091 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
3092 system services want to log events about specific client
3093 processes. journactl/systemctl has been updated to make use
3094 of this information if all log messages regarding a specific
3097 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
3098 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
3099 Reisner, David Coppa, David King, David Strauss, Eelco
3100 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
3101 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
3102 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
3103 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
3104 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
3105 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
3106 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
3107 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
3108 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
3109 Łukasz Stelmach, 장동준
3113 * The Python bindings gained some minimal support for the APIs
3114 exposed by libsystemd-logind.
3116 * ConditionSecurity= gained support for detecting SMACK. Since
3117 this condition already supports SELinux and AppArmor we only
3118 miss IMA for this. Patches welcome!
3120 Contributions from: Karol Lewandowski, Lennart Poettering,
3121 Zbigniew Jędrzejewski-Szmek
3125 * systemd-nspawn will now create /etc/resolv.conf if
3126 necessary, before bind-mounting the host's file onto it.
3128 * systemd-nspawn will now store meta information about a
3129 container on the container's cgroup as extended attribute
3130 fields, including the root directory.
3132 * The cgroup hierarchy has been reworked in many ways. All
3133 objects any of the components systemd creates in the cgroup
3134 tree are now suffixed. More specifically, user sessions are
3135 now placed in cgroups suffixed with ".session", users in
3136 cgroups suffixed with ".user", and nspawn containers in
3137 cgroups suffixed with ".nspawn". Furthermore, all cgroup
3138 names are now escaped in a simple scheme to avoid collision
3139 of userspace object names with kernel filenames. This work
3140 is preparation for making these objects relocatable in the
3141 cgroup tree, in order to allow easy resource partitioning of
3142 these objects without causing naming conflicts.
3144 * systemctl list-dependencies gained the new switches
3145 --plain, --reverse, --after and --before.
3147 * systemd-inhibit now shows the process name of processes that
3148 have taken an inhibitor lock.
3150 * nss-myhostname will now also resolve "localhost"
3151 implicitly. This makes /etc/hosts an optional file and
3152 nicely handles that on IPv6 ::1 maps to both "localhost" and
3155 * libsystemd-logind.so gained a new call
3156 sd_get_machine_names() to enumerate running containers and
3157 VMs (currently only supported by very new libvirt and
3158 nspawn). sd_login_monitor can now be used to watch
3159 VMs/containers coming and going.
3161 * .include is not allowed recursively anymore, and only in
3162 unit files. Usually it is better to use drop-in snippets in
3163 .d/*.conf anyway, as introduced with systemd 198.
3165 * systemd-analyze gained a new "critical-chain" command that
3166 determines the slowest chain of units run during system
3167 boot-up. It is very useful for tracking down where
3168 optimizing boot time is the most beneficial.
3170 * systemd will no longer allow manipulating service paths in
3171 the name=systemd:/system cgroup tree using ControlGroup= in
3172 units. (But is still fine with it in all other dirs.)
3174 * There's a new systemd-nspawn@.service service file that may
3175 be used to easily run nspawn containers as system
3176 services. With the container's root directory in
3177 /var/lib/container/foobar it is now sufficient to run
3178 "systemctl start systemd-nspawn@foobar.service" to boot it.
3180 * systemd-cgls gained a new parameter "--machine" to list only
3181 the processes within a certain container.
3183 * ConditionSecurity= now can check for "apparmor". We still
3184 are lacking checks for SMACK and IMA for this condition
3185 check though. Patches welcome!
3187 * A new configuration file /etc/systemd/sleep.conf has been
3188 added that may be used to configure which kernel operation
3189 systemd is supposed to execute when "suspend", "hibernate"
3190 or "hybrid-sleep" is requested. This makes the new kernel
3191 "freeze" state accessible to the user.
3193 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
3194 the passed argument if applicable.
3196 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
3197 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
3198 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
3199 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
3200 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
3201 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
3202 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
3207 * The output of 'systemctl list-jobs' got some polishing. The
3208 '--type=' argument may now be passed more than once. A new
3209 command 'systemctl list-sockets' has been added which shows
3210 a list of kernel sockets systemd is listening on with the
3211 socket units they belong to, plus the units these socket
3214 * The experimental libsystemd-bus library got substantial
3215 updates to work in conjunction with the (also experimental)
3216 kdbus kernel project. It works well enough to exchange
3217 messages with some sophistication. Note that kdbus is not
3218 ready yet, and the library is mostly an elaborate test case
3219 for now, and not installable.
3221 * systemd gained a new unit 'systemd-static-nodes.service'
3222 that generates static device nodes earlier during boot, and
3223 can run in conjunction with udev.
3225 * libsystemd-login gained a new call sd_pid_get_user_unit()
3226 to retrieve the user systemd unit a process is running
3227 in. This is useful for systems where systemd is used as
3230 * systemd-nspawn now places all containers in the new /machine
3231 top-level cgroup directory in the name=systemd
3232 hierarchy. libvirt will soon do the same, so that we get a
3233 uniform separation of /system, /user and /machine for system
3234 services, user processes and containers/virtual
3235 machines. This new cgroup hierarchy is also useful to stick
3236 stable names to specific container instances, which can be
3237 recognized later this way (this name may be controlled
3238 via systemd-nspawn's new -M switch). libsystemd-login also
3239 gained a new call sd_pid_get_machine_name() to retrieve the
3240 name of the container/VM a specific process belongs to.
3242 * bootchart can now store its data in the journal.
3244 * libsystemd-journal gained a new call
3245 sd_journal_add_conjunction() for AND expressions to the
3246 matching logic. This can be used to express more complex
3247 logical expressions.
3249 * journactl can now take multiple --unit= and --user-unit=
3252 * The cryptsetup logic now understands the "luks.key=" kernel
3253 command line switch for specifying a file to read the
3254 decryption key from. Also, if a configured key file is not
3255 found the tool will now automatically fall back to prompting
3258 * Python systemd.journal module was updated to wrap recently
3259 added functions from libsystemd-journal. The interface was
3260 changed to bring the low level interface in s.j._Reader
3261 closer to the C API, and the high level interface in
3262 s.j.Reader was updated to wrap and convert all data about
3265 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
3266 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
3267 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
3268 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
3269 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
3270 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3274 * journalctl --update-catalog now understands a new --root=
3275 option to operate on catalogs found in a different root
3278 * During shutdown after systemd has terminated all running
3279 services a final killing loop kills all remaining left-over
3280 processes. We will now print the name of these processes
3281 when we send SIGKILL to them, since this usually indicates a
3284 * If /etc/crypttab refers to password files stored on
3285 configured mount points automatic dependencies will now be
3286 generated to ensure the specific mount is established first
3287 before the key file is attempted to be read.
3289 * 'systemctl status' will now show information about the
3290 network sockets a socket unit is listening on.
3292 * 'systemctl status' will also shown information about any
3293 drop-in configuration file for units. (Drop-In configuration
3294 files in this context are files such as
3295 /etc/systemd/systemd/foobar.service.d/*.conf)
3297 * systemd-cgtop now optionally shows summed up CPU times of
3298 cgroups. Press '%' while running cgtop to switch between
3299 percentage and absolute mode. This is useful to determine
3300 which cgroups use up the most CPU time over the entire
3301 runtime of the system. systemd-cgtop has also been updated
3302 to be 'pipeable' for processing with further shell tools.
3304 * 'hostnamectl set-hostname' will now allow setting of FQDN
3307 * The formatting and parsing of time span values has been
3308 changed. The parser now understands fractional expressions
3309 such as "5.5h". The formatter will now output fractional
3310 expressions for all time spans under 1min, i.e. "5.123456s"
3311 rather than "5s 123ms 456us". For time spans under 1s
3312 millisecond values are shown, for those under 1ms
3313 microsecond values are shown. This should greatly improve
3314 all time-related output of systemd.
3316 * libsystemd-login and libsystemd-journal gained new
3317 functions for querying the poll() events mask and poll()
3318 timeout value for integration into arbitrary event
3321 * localectl gained the ability to list available X11 keymaps
3322 (models, layouts, variants, options).
3324 * 'systemd-analyze dot' gained the ability to filter for
3325 specific units via shell-style globs, to create smaller,
3326 more useful graphs. I.e. it is now possible to create simple
3327 graphs of all the dependencies between only target units, or
3328 of all units that Avahi has dependencies with.
3330 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
3331 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
3332 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
3333 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
3334 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
3335 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
3336 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
3340 * The boot-time readahead implementation for rotating media
3341 will now read the read-ahead data in multiple passes which
3342 consist of all read requests made in equidistant time
3343 intervals. This means instead of strictly reading read-ahead
3344 data in its physical order on disk we now try to find a
3345 middle ground between physical and access time order.
3347 * /etc/os-release files gained a new BUILD_ID= field for usage
3348 on operating systems that provide continuous builds of OS
3351 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
3352 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
3353 William Douglas, Zbigniew Jędrzejewski-Szmek
3357 * systemd-python gained an API exposing libsystemd-daemon.
3359 * The SMACK setup logic gained support for uploading CIPSO
3362 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
3363 ReadOnlyDirectories= and InaccessibleDirectories= has
3364 changed. The private /tmp and /var/tmp directories are now
3365 shared by all processes of a service (which means
3366 ExecStartPre= may now leave data in /tmp that ExecStart= of
3367 the same service can still access). When a service is
3368 stopped its temporary directories are immediately deleted
3369 (normal clean-up with tmpfiles is still done in addition to
3372 * By default, systemd will now set a couple of sysctl
3373 variables in the kernel: the safe sysrq options are turned
3374 on, IP route verification is turned on, and source routing
3375 disabled. The recently added hardlink and softlink
3376 protection of the kernel is turned on. These settings should
3377 be reasonably safe, and good defaults for all new systems.
3379 * The predictable network naming logic may now be turned off
3380 with a new kernel command line switch: net.ifnames=0.
3382 * A new libsystemd-bus module has been added that implements a
3383 pretty complete D-Bus client library. For details see:
3385 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
3387 * journald will now explicitly flush the journal files to disk
3388 at the latest 5min after each write. The file will then also
3389 be marked offline until the next write. This should increase
3390 reliability in case of a crash. The synchronization delay
3391 can be configured via SyncIntervalSec= in journald.conf.
3393 * There's a new remote-fs-setup.target unit that can be used
3394 to pull in specific services when at least one remote file
3395 system is to be mounted.
3397 * There are new targets timers.target and paths.target as
3398 canonical targets to pull user timer and path units in
3399 from. This complements sockets.target with a similar
3400 purpose for socket units.
3402 * libudev gained a new call udev_device_set_attribute_value()
3403 to set sysfs attributes of a device.
3405 * The udev daemon now sets the default number of worker
3406 processes executed in parallel based on the number of available
3407 CPUs instead of the amount of available RAM. This is supposed
3408 to provide a more reliable default and limit a too aggressive
3409 paralellism for setups with 1000s of devices connected.
3411 Contributions from: Auke Kok, Colin Walters, Cristian
3412 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
3413 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
3414 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
3415 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
3416 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
3417 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
3418 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
3419 Zbigniew Jędrzejewski-Szmek
3423 * Configuration of unit files may now be extended via drop-in
3424 files without having to edit/override the unit files
3425 themselves. More specifically, if the administrator wants to
3426 change one value for a service file foobar.service he can
3427 now do so by dropping in a configuration snippet into
3428 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
3429 will load all these snippets and apply them on top of the
3430 main unit configuration file, possibly extending or
3431 overriding its settings. Using these drop-in snippets is
3432 generally nicer than the two earlier options for changing
3433 unit files locally: copying the files from
3434 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
3435 them there; or creating a new file in /etc/systemd/system/
3436 that incorporates the original one via ".include". Drop-in
3437 snippets into these .d/ directories can be placed in any
3438 directory systemd looks for units in, and the usual
3439 overriding semantics between /usr/lib, /etc and /run apply
3442 * Most unit file settings which take lists of items can now be
3443 reset by assigning the empty string to them. For example,
3444 normally, settings such as Environment=FOO=BAR append a new
3445 environment variable assignment to the environment block,
3446 each time they are used. By assigning Environment= the empty
3447 string the environment block can be reset to empty. This is
3448 particularly useful with the .d/*.conf drop-in snippets
3449 mentioned above, since this adds the ability to reset list
3450 settings from vendor unit files via these drop-ins.
3452 * systemctl gained a new "list-dependencies" command for
3453 listing the dependencies of a unit recursively.
3455 * Inhibitors are now honored and listed by "systemctl
3456 suspend", "systemctl poweroff" (and similar) too, not only
3457 GNOME. These commands will also list active sessions by
3460 * Resource limits (as exposed by the various control group
3461 controllers) can now be controlled dynamically at runtime
3462 for all units. More specifically, you can now use a command
3463 like "systemctl set-cgroup-attr foobar.service cpu.shares
3464 2000" to alter the CPU shares a specific service gets. These
3465 settings are stored persistently on disk, and thus allow the
3466 administrator to easily adjust the resource usage of
3467 services with a few simple commands. This dynamic resource
3468 management logic is also available to other programs via the
3469 bus. Almost any kernel cgroup attribute and controller is
3472 * systemd-vconsole-setup will now copy all font settings to
3473 all allocated VTs, where it previously applied them only to
3476 * libsystemd-login gained the new sd_session_get_tty() API
3479 * This release drops support for a few legacy or
3480 distribution-specific LSB facility names when parsing init
3481 scripts: $x-display-manager, $mail-transfer-agent,
3482 $mail-transport-agent, $mail-transfer-agent, $smtp,
3483 $null. Also, the mail-transfer-agent.target unit backing
3484 this has been removed. Distributions which want to retain
3485 compatibility with this should carry the burden for
3486 supporting this themselves and patch support for these back
3487 in, if they really need to. Also, the facilities $syslog and
3488 $local_fs are now ignored, since systemd does not support
3489 early-boot LSB init scripts anymore, and these facilities
3490 are implied anyway for normal services. syslog.target has
3493 * There are new bus calls on PID1's Manager object for
3494 cancelling jobs, and removing snapshot units. Previously,
3495 both calls were only available on the Job and Snapshot
3498 * systemd-journal-gatewayd gained SSL support.
3500 * The various "environment" files, such as /etc/locale.conf
3501 now support continuation lines with a backslash ("\") as
3502 last character in the line, similarly in style (but different)
3503 to how this is supported in shells.
3505 * For normal user processes the _SYSTEMD_USER_UNIT= field is
3506 now implicitly appended to every log entry logged. systemctl
3507 has been updated to filter by this field when operating on a
3508 user systemd instance.
3510 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
3511 CAP_AUDIT_CONTROL capabilities to the capabilities set for
3512 the container. This makes it easier to boot unmodified
3513 Fedora systems in a container, which however still requires
3514 audit=0 to be passed on the kernel command line. Auditing in
3515 kernel and userspace is unfortunately still too broken in
3516 context of containers, hence we recommend compiling it out
3517 of the kernel or using audit=0. Hopefully this will be fixed
3518 one day for good in the kernel.
3520 * nspawn gained the new --bind= and --bind-ro= parameters to
3521 bind mount specific directories from the host into the
3524 * nspawn will now mount its own devpts file system instance
3525 into the container, in order not to leak pty devices from
3526 the host into the container.
3528 * systemd will now read the firmware boot time performance
3529 information from the EFI variables, if the used boot loader
3530 supports this, and takes it into account for boot performance
3531 analysis via "systemd-analyze". This is currently supported
3532 only in conjunction with Gummiboot, but could be supported
3533 by other boot loaders too. For details see:
3535 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
3537 * A new generator has been added that automatically mounts the
3538 EFI System Partition (ESP) to /boot, if that directory
3539 exists, is empty, and no other file system has been
3540 configured to be mounted there.
3542 * logind will now send out PrepareForSleep(false) out
3543 unconditionally, after coming back from suspend. This may be
3544 used by applications as asynchronous notification for
3545 system resume events.
3547 * "systemctl unlock-sessions" has been added, that allows
3548 unlocking the screens of all user sessions at once, similar
3549 to how "systemctl lock-sessions" already locked all users
3550 sessions. This is backed by a new D-Bus call UnlockSessions().
3552 * "loginctl seat-status" will now show the master device of a
3553 seat. (i.e. the device of a seat that needs to be around for
3554 the seat to be considered available, usually the graphics
3557 * tmpfiles gained a new "X" line type, that allows
3558 configuration of files and directories (with wildcards) that
3559 shall be excluded from automatic cleanup ("aging").
3561 * udev default rules set the device node permissions now only
3562 at "add" events, and do not change them any longer with a
3563 later "change" event.
3565 * The log messages for lid events and power/sleep keypresses
3566 now carry a message ID.
3568 * We now have a substantially larger unit test suite, but this
3569 continues to be work in progress.
3571 * udevadm hwdb gained a new --root= parameter to change the
3572 root directory to operate relative to.
3574 * logind will now issue a background sync() request to the kernel
3575 early at shutdown, so that dirty buffers are flushed to disk early
3576 instead of at the last moment, in order to optimize shutdown
3579 * A new bootctl tool has been added that is an interface for
3580 certain boot loader operations. This is currently a preview
3581 and is likely to be extended into a small mechanism daemon
3582 like timedated, localed, hostnamed, and can be used by
3583 graphical UIs to enumerate available boot options, and
3584 request boot into firmware operations.
3586 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
3587 the rest of the package. It also has been updated to work
3588 correctly in initrds.
3590 * Policykit previously has been runtime optional, and is now
3591 also compile time optional via a configure switch.
3593 * systemd-analyze has been reimplemented in C. Also "systemctl
3594 dot" has moved into systemd-analyze.
3596 * "systemctl status" with no further parameters will now print
3597 the status of all active or failed units.
3599 * Operations such as "systemctl start" can now be executed
3600 with a new mode "--irreversible" which may be used to queue
3601 operations that cannot accidentally be reversed by a later
3602 job queuing. This is by default used to make shutdown
3603 requests more robust.
3605 * The Python API of systemd now gained a new module for
3606 reading journal files.
3608 * A new tool kernel-install has been added that can install
3609 kernel images according to the Boot Loader Specification:
3611 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
3613 * Boot time console output has been improved to provide
3614 animated boot time output for hanging jobs.
3616 * A new tool systemd-activate has been added which can be used
3617 to test socket activation with, directly from the command
3618 line. This should make it much easier to test and debug
3619 socket activation in daemons.
3621 * journalctl gained a new "--reverse" (or -r) option to show
3622 journal output in reverse order (i.e. newest line first).
3624 * journalctl gained a new "--pager-end" (or -e) option to jump
3625 to immediately jump to the end of the journal in the
3626 pager. This is only supported in conjunction with "less".
3628 * journalctl gained a new "--user-unit=" option, that works
3629 similarly to "--unit=" but filters for user units rather than
3632 * A number of unit files to ease adoption of systemd in
3633 initrds has been added. This moves some minimal logic from
3634 the various initrd implementations into systemd proper.
3636 * The journal files are now owned by a new group
3637 "systemd-journal", which exists specifically to allow access
3638 to the journal, and nothing else. Previously, we used the
3639 "adm" group for that, which however possibly covers more
3640 than just journal/log file access. This new group is now
3641 already used by systemd-journal-gatewayd to ensure this
3642 daemon gets access to the journal files and as little else
3643 as possible. Note that "make install" will also set FS ACLs
3644 up for /var/log/journal to give "adm" and "wheel" read
3645 access to it, in addition to "systemd-journal" which owns
3646 the journal files. We recommend that packaging scripts also
3647 add read access to "adm" + "wheel" to /var/log/journal, and
3648 all existing/future journal files. To normal users and
3649 administrators little changes, however packagers need to
3650 ensure to create the "systemd-journal" system group at
3651 package installation time.
3653 * The systemd-journal-gatewayd now runs as unprivileged user
3654 systemd-journal-gateway:systemd-journal-gateway. Packaging
3655 scripts need to create these system user/group at
3658 * timedated now exposes a new boolean property CanNTP that
3659 indicates whether a local NTP service is available or not.
3661 * systemd-detect-virt will now also detect xen PVs
3663 * The pstore file system is now mounted by default, if it is
3666 * In addition to the SELinux and IMA policies we will now also
3667 load SMACK policies at early boot.
3669 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
3670 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
3671 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
3672 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
3673 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
3674 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
3675 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
3676 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
3677 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
3678 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
3679 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
3680 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
3681 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
3682 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
3686 * Timer units now support calendar time events in addition to
3687 monotonic time events. That means you can now trigger a unit
3688 based on a calendar time specification such as "Thu,Fri
3689 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
3690 or fifth day of any month of the year 2013, given that it is
3691 a thursday or friday. This brings timer event support
3692 considerably closer to cron's capabilities. For details on
3693 the supported calendar time specification language see
3696 * udev now supports a number of different naming policies for
3697 network interfaces for predictable names, and a combination
3698 of these policies is now the default. Please see this wiki
3699 document for details:
3701 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
3703 * Auke Kok's bootchart implementation has been added to the
3704 systemd tree. It is an optional component that can graph the
3705 boot in quite some detail. It is one of the best bootchart
3706 implementations around and minimal in its code and
3709 * nss-myhostname has been integrated into the systemd source
3710 tree. nss-myhostname guarantees that the local hostname
3711 always stays resolvable via NSS. It has been a weak
3712 requirement of systemd-hostnamed since a long time, and
3713 since its code is actually trivial we decided to just
3714 include it in systemd's source tree. It can be turned off
3715 with a configure switch.
3717 * The read-ahead logic is now capable of properly detecting
3718 whether a btrfs file system is on SSD or rotating media, in
3719 order to optimize the read-ahead scheme. Previously, it was
3720 only capable of detecting this on traditional file systems
3723 * In udev, additional device properties are now read from the
3724 IAB in addition to the OUI database. Also, Bluetooth company
3725 identities are attached to the devices as well.
3727 * In service files %U may be used as specifier that is
3728 replaced by the configured user name of the service.
3730 * nspawn may now be invoked without a controlling TTY. This
3731 makes it suitable for invocation as its own service. This
3732 may be used to set up a simple containerized server system
3733 using only core OS tools.
3735 * systemd and nspawn can now accept socket file descriptors
3736 when they are started for socket activation. This enables
3737 implementation of socket activated nspawn
3738 containers. i.e. think about autospawning an entire OS image
3739 when the first SSH or HTTP connection is received. We expect
3740 that similar functionality will also be added to libvirt-lxc
3743 * journalctl will now suppress ANSI color codes when
3744 presenting log data.
3746 * systemctl will no longer show control group information for
3747 a unit if a the control group is empty anyway.
3749 * logind can now automatically suspend/hibernate/shutdown the
3752 * /etc/machine-info and hostnamed now also expose the chassis
3753 type of the system. This can be used to determine whether
3754 the local system is a laptop, desktop, handset or
3755 tablet. This information may either be configured by the
3756 user/vendor or is automatically determined from ACPI and DMI
3757 information if possible.
3759 * A number of PolicyKit actions are now bound together with
3760 "imply" rules. This should simplify creating UIs because
3761 many actions will now authenticate similar ones as well.
3763 * Unit files learnt a new condition ConditionACPower= which
3764 may be used to conditionalize a unit depending on whether an
3765 AC power source is connected or not, of whether the system
3766 is running on battery power.
3768 * systemctl gained a new "is-failed" verb that may be used in
3769 shell scripts and suchlike to check whether a specific unit
3770 is in the "failed" state.
3772 * The EnvironmentFile= setting in unit files now supports file
3773 globbing, and can hence be used to easily read a number of
3774 environment files at once.
3776 * systemd will no longer detect and recognize specific
3777 distributions. All distribution-specific #ifdeffery has been
3778 removed, systemd is now fully generic and
3779 distribution-agnostic. Effectively, not too much is lost as
3780 a lot of the code is still accessible via explicit configure
3781 switches. However, support for some distribution specific
3782 legacy configuration file formats has been dropped. We
3783 recommend distributions to simply adopt the configuration
3784 files everybody else uses now and convert the old
3785 configuration from packaging scripts. Most distributions
3786 already did that. If that's not possible or desirable,
3787 distributions are welcome to forward port the specific
3788 pieces of code locally from the git history.
3790 * When logging a message about a unit systemd will now always
3791 log the unit name in the message meta data.
3793 * localectl will now also discover system locale data that is
3794 not stored in locale archives, but directly unpacked.
3796 * logind will no longer unconditionally use framebuffer
3797 devices as seat masters, i.e. as devices that are required
3798 to be existing before a seat is considered preset. Instead,
3799 it will now look for all devices that are tagged as
3800 "seat-master" in udev. By default framebuffer devices will
3801 be marked as such, but depending on local systems other
3802 devices might be marked as well. This may be used to
3803 integrate graphics cards using closed source drivers (such
3804 as NVidia ones) more nicely into logind. Note however, that
3805 we recommend using the open source NVidia drivers instead,
3806 and no udev rules for the closed-source drivers will be
3807 shipped from us upstream.
3809 Contributions from: Adam Williamson, Alessandro Crismani, Auke
3810 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
3811 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
3812 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
3813 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
3814 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
3815 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
3816 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
3817 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
3818 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
3819 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
3820 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
3825 * udev gained support for loading additional device properties
3826 from an indexed database that is keyed by vendor/product IDs
3827 and similar device identifiers. For the beginning this
3828 "hwdb" is populated with data from the well-known PCI and
3829 USB database, but also includes PNP, ACPI and OID data. In
3830 the longer run this indexed database shall grow into
3831 becoming the one central database for non-essential
3832 userspace device metadata. Previously, data from the PCI/USB
3833 database was only attached to select devices, since the
3834 lookup was a relatively expensive operation due to O(n) time
3835 complexity (with n being the number of entries in the
3836 database). Since this is now O(1), we decided to add in this
3837 data for all devices where this is available, by
3838 default. Note that the indexed database needs to be rebuilt
3839 when new data files are installed. To achieve this you need
3840 to update your packaging scripts to invoke "udevadm hwdb
3841 --update" after installation of hwdb data files. For
3842 RPM-based distributions we introduced the new
3843 %udev_hwdb_update macro for this purpose.
3845 * The Journal gained support for the "Message Catalog", an
3846 indexed database to link up additional information with
3847 journal entries. For further details please check:
3849 http://www.freedesktop.org/wiki/Software/systemd/catalog
3851 The indexed message catalog database also needs to be
3852 rebuilt after installation of message catalog files. Use
3853 "journalctl --update-catalog" for this. For RPM-based
3854 distributions we introduced the %journal_catalog_update
3855 macro for this purpose.
3857 * The Python Journal bindings gained support for the standard
3858 Python logging framework.
3860 * The Journal API gained new functions for checking whether
3861 the underlying file system of a journal file is capable of
3862 properly reporting file change notifications, or whether
3863 applications that want to reflect journal changes "live"
3864 need to recheck journal files continuously in appropriate
3867 * It is now possible to set the "age" field for tmpfiles
3868 entries to 0, indicating that files matching this entry
3869 shall always be removed when the directories are cleaned up.
3871 * coredumpctl gained a new "gdb" verb which invokes gdb
3872 right-away on the selected coredump.
3874 * There's now support for "hybrid sleep" on kernels that
3875 support this, in addition to "suspend" and "hibernate". Use
3876 "systemctl hybrid-sleep" to make use of this.
3878 * logind's HandleSuspendKey= setting (and related settings)
3879 now gained support for a new "lock" setting to simply
3880 request the screen lock on all local sessions, instead of
3881 actually executing a suspend or hibernation.
3883 * systemd will now mount the EFI variables file system by
3886 * Socket units now gained support for configuration of the
3887 SMACK security label.
3889 * timedatectl will now output the time of the last and next
3890 daylight saving change.
3892 * We dropped support for various legacy and distro-specific
3893 concepts, such as insserv, early-boot SysV services
3894 (i.e. those for non-standard runlevels such as 'b' or 'S')
3895 or ArchLinux /etc/rc.conf support. We recommend the
3896 distributions who still need support this to either continue
3897 to maintain the necessary patches downstream, or find a
3898 different solution. (Talk to us if you have questions!)
3900 * Various systemd components will now bypass PolicyKit checks
3901 for root and otherwise handle properly if PolicyKit is not
3902 found to be around. This should fix most issues for
3903 PolicyKit-less systems. Quite frankly this should have been
3904 this way since day one. It is absolutely our intention to
3905 make systemd work fine on PolicyKit-less systems, and we
3906 consider it a bug if something does not work as it should if
3907 PolicyKit is not around.
3909 * For embedded systems it is now possible to build udev and
3910 systemd without blkid and/or kmod support.
3912 * "systemctl switch-root" is now capable of switching root
3913 more than once. I.e. in addition to transitions from the
3914 initrd to the host OS it is now possible to transition to
3915 further OS images from the host. This is useful to implement
3916 offline updating tools.
3918 * Various other additions have been made to the RPM macros
3919 shipped with systemd. Use %udev_rules_update() after
3920 installing new udev rules files. %_udevhwdbdir,
3921 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
3922 %_sysctldir are now available which resolve to the right
3923 directories for packages to place various data files in.
3925 * journalctl gained the new --full switch (in addition to
3926 --all, to disable ellipsation for long messages.
3928 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
3929 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
3930 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
3931 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
3932 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
3933 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
3934 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
3935 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
3936 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
3940 * journalctl gained new --since= and --until= switches to
3941 filter by time. It also now supports nice filtering for
3942 units via --unit=/-u.
3944 * Type=oneshot services may use ExecReload= and do the
3947 * The journal daemon now supports time-based rotation and
3948 vacuuming, in addition to the usual disk-space based
3951 * The journal will now index the available field values for
3952 each field name. This enables clients to show pretty drop
3953 downs of available match values when filtering. The bash
3954 completion of journalctl has been updated
3955 accordingly. journalctl gained a new switch -F to list all
3956 values a certain field takes in the journal database.
3958 * More service events are now written as structured messages
3959 to the journal, and made recognizable via message IDs.
3961 * The timedated, localed and hostnamed mini-services which
3962 previously only provided support for changing time, locale
3963 and hostname settings from graphical DEs such as GNOME now
3964 also have a minimal (but very useful) text-based client
3965 utility each. This is probably the nicest way to changing
3966 these settings from the command line now, especially since
3967 it lists available options and is fully integrated with bash
3970 * There's now a new tool "systemd-coredumpctl" to list and
3971 extract coredumps from the journal.
3973 * We now install a README each in /var/log/ and
3974 /etc/rc.d/init.d explaining where the system logs and init
3975 scripts went. This hopefully should help folks who go to
3976 that dirs and look into the otherwise now empty void and
3977 scratch their heads.
3979 * When user-services are invoked (by systemd --user) the
3980 $MANAGERPID env var is set to the PID of systemd.
3982 * SIGRTMIN+24 when sent to a --user instance will now result
3983 in immediate termination of systemd.
3985 * gatewayd received numerous feature additions such as a
3986 "follow" mode, for live syncing and filtering.
3988 * browse.html now allows filtering and showing detailed
3989 information on specific entries. Keyboard navigation and
3990 mouse screen support has been added.
3992 * gatewayd/journalctl now supports HTML5/JSON
3993 Server-Sent-Events as output.
3995 * The SysV init script compatibility logic will now
3996 heuristically determine whether a script supports the
3997 "reload" verb, and only then make this available as
4000 * "systemctl status --follow" has been removed, use "journalctl
4003 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
4004 have been removed since they are hardly useful to be
4007 * And I'd like to take the opportunity to specifically mention
4008 Zbigniew for his great contributions. Zbigniew, you rock!
4010 Contributions from: Andrew Eikum, Christian Hesse, Colin
4011 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
4012 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
4013 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
4014 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
4015 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
4016 Jędrzejewski-Szmek, Сковорода Никита Андреевич
4020 * If /etc/vconsole.conf is non-existent or empty we will no
4021 longer load any console font or key map at boot by
4022 default. Instead the kernel defaults will be left
4023 intact. This is definitely the right thing to do, as no
4024 configuration should mean no configuration, and hard-coding
4025 font names that are different on all archs is probably a bad
4026 idea. Also, the kernel default key map and font should be
4027 good enough for most cases anyway, and mostly identical to
4028 the userspace fonts/key maps we previously overloaded them
4029 with. If distributions want to continue to default to a
4030 non-kernel font or key map they should ship a default
4031 /etc/vconsole.conf with the appropriate contents.
4033 Contributions from: Colin Walters, Daniel J Walsh, Dave
4034 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
4035 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
4039 * journalctl gained a new --cursor= switch to show entries
4040 starting from the specified location in the journal.
4042 * We now enforce a size limit on journal entry fields exported
4043 with "-o json" in journalctl. Fields larger than 4K will be
4044 assigned null. This can be turned off with --all.
4046 * An (optional) journal gateway daemon is now available as
4047 "systemd-journal-gatewayd.service". This service provides
4048 access to the journal via HTTP and JSON. This functionality
4049 will be used to implement live log synchronization in both
4050 pull and push modes, but has various other users too, such
4051 as easy log access for debugging of embedded devices. Right
4052 now it is already useful to retrieve the journal via HTTP:
4054 # systemctl start systemd-journal-gatewayd.service
4055 # wget http://localhost:19531/entries
4057 This will download the journal contents in a
4058 /var/log/messages compatible format. The same as JSON:
4060 # curl -H"Accept: application/json" http://localhost:19531/entries
4062 This service is also accessible via a web browser where a
4063 single static HTML5 app is served that uses the JSON logic
4064 to enable the user to do some basic browsing of the
4065 journal. This will be extended later on. Here's an example
4066 screenshot of this app in its current state:
4068 http://0pointer.de/public/journal-gatewayd
4070 Contributions from: Kay Sievers, Lennart Poettering, Robert
4071 Milasan, Tom Gundersen
4075 * The bash completion logic is now available for journalctl
4078 * We do not mount the "cpuset" controller anymore together with
4079 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
4080 started if no parameters are assigned to it. "cpuset" hence
4081 broke code that assumed it it could create "cpu" groups and
4084 * journalctl -f will now subscribe to terminal size changes,
4085 and line break accordingly.
4087 Contributions from: Dave Reisner, Kay Sievers, Lennart
4088 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
4092 * nspawn will now create a symlink /etc/localtime in the
4093 container environment, copying the host's timezone
4094 setting. Previously this has been done via a bind mount, but
4095 since symlinks cannot be bind mounted this has now been
4096 changed to create/update the appropriate symlink.
4098 * journalctl -n's line number argument is now optional, and
4099 will default to 10 if omitted.
4101 * journald will now log the maximum size the journal files may
4102 take up on disk. This is particularly useful if the default
4103 built-in logic of determining this parameter from the file
4104 system size is used. Use "systemctl status
4105 systemd-journald.service" to see this information.
4107 * The multi-seat X wrapper tool has been stripped down. As X
4108 is now capable of enumerating graphics devices via udev in a
4109 seat-aware way the wrapper is not strictly necessary
4110 anymore. A stripped down temporary stop-gap is still shipped
4111 until the upstream display managers have been updated to
4112 fully support the new X logic. Expect this wrapper to be
4113 removed entirely in one of the next releases.
4115 * HandleSleepKey= in logind.conf has been split up into
4116 HandleSuspendKey= and HandleHibernateKey=. The old setting
4117 is not available anymore. X11 and the kernel are
4118 distinguishing between these keys and we should too. This
4119 also means the inhibition lock for these keys has been split
4122 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
4123 Poettering, Lukas Nykryn, Václav Pavlín
4127 * Whenever a unit changes state we will now log this to the
4128 journal and show along the unit's own log output in
4131 * ConditionPathIsMountPoint= can now properly detect bind
4132 mount points too. (Previously, a bind mount of one file
4133 system to another place in the same file system could not be
4134 detected as mount, since they shared struct stat's st_dev
4137 * We will now mount the cgroup controllers cpu, cpuacct,
4138 cpuset and the controllers net_cls, net_prio together by
4141 * nspawn containers will now have a virtualized boot
4142 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
4143 over with a randomized ID at container initialization). This
4144 has the effect of making "journalctl -b" do the right thing
4147 * The JSON output journal serialization has been updated not
4148 to generate "endless" list objects anymore, but rather one
4149 JSON object per line. This is more in line how most JSON
4150 parsers expect JSON objects. The new output mode
4151 "json-pretty" has been added to provide similar output, but
4152 neatly aligned for readability by humans.
4154 * We dropped all explicit sync() invocations in the shutdown
4155 code. The kernel does this implicitly anyway in the kernel
4156 reboot() syscall. halt(8)'s -n option is now a compatibility
4159 * We now support virtualized reboot() in containers, as
4160 supported by newer kernels. We will fall back to exit() if
4161 CAP_SYS_REBOOT is not available to the container. Also,
4162 nspawn makes use of this now and will actually reboot the
4163 container if the containerized OS asks for that.
4165 * journalctl will only show local log output by default
4166 now. Use --merge (-m) to show remote log output, too.
4168 * libsystemd-journal gained the new sd_journal_get_usage()
4169 call to determine the current disk usage of all journal
4170 files. This is exposed in the new "journalctl --disk-usage"
4173 * journald gained a new configuration setting SplitMode= in
4174 journald.conf which may be used to control how user journals
4175 are split off. See journald.conf(5) for details.
4177 * A new condition type ConditionFileNotEmpty= has been added.
4179 * tmpfiles' "w" lines now support file globbing, to write
4180 multiple files at once.
4182 * We added Python bindings for the journal submission
4183 APIs. More Python APIs for a number of selected APIs will
4184 likely follow. Note that we intend to add native bindings
4185 only for the Python language, as we consider it common
4186 enough to deserve bindings shipped within systemd. There are
4187 various projects outside of systemd that provide bindings
4188 for languages such as PHP or Lua.
4190 * Many conditions will now resolve specifiers such as %i. In
4191 addition, PathChanged= and related directives of .path units
4192 now support specifiers as well.
4194 * There's now a new RPM macro definition for the system preset
4197 * journald will now warn if it ca not forward a message to the
4198 syslog daemon because its socket is full.
4200 * timedated will no longer write or process /etc/timezone,
4201 except on Debian. As we do not support late mounted /usr
4202 anymore /etc/localtime always being a symlink is now safe,
4203 and hence the information in /etc/timezone is not necessary
4206 * logind will now always reserve one VT for a text getty (VT6
4207 by default). Previously if more than 6 X sessions where
4208 started they took up all the VTs with auto-spawned gettys,
4209 so that no text gettys were available anymore.
4211 * udev will now automatically inform the btrfs kernel logic
4212 about btrfs RAID components showing up. This should make
4213 simple hotplug based btrfs RAID assembly work.
4215 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
4216 (but not for its children which will stay at the kernel
4217 default). This should allow setups with a lot more listening
4220 * systemd will now always pass the configured timezone to the
4221 kernel at boot. timedated will do the same when the timezone
4224 * logind's inhibition logic has been updated. By default,
4225 logind will now handle the lid switch, the power and sleep
4226 keys all the time, even in graphical sessions. If DEs want
4227 to handle these events on their own they should take the new
4228 handle-power-key, handle-sleep-key and handle-lid-switch
4229 inhibitors during their runtime. A simple way to achieve
4230 that is to invoke the DE wrapped in an invocation of:
4232 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
4234 * Access to unit operations is now checked via SELinux taking
4235 the unit file label and client process label into account.
4237 * systemd will now notify the administrator in the journal
4238 when he over-mounts a non-empty directory.
4240 * There are new specifiers that are resolved in unit files,
4241 for the host name (%H), the machine ID (%m) and the boot ID
4244 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
4245 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
4246 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
4247 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
4248 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
4249 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
4250 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
4254 * Support for reading structured kernel messages from
4255 /dev/kmsg has now been added and is enabled by default.
4257 * Support for reading kernel messages from /proc/kmsg has now
4258 been removed. If you want kernel messages in the journal
4259 make sure to run a recent kernel (>= 3.5) that supports
4260 reading structured messages from /dev/kmsg (see
4261 above). /proc/kmsg is now exclusive property of classic
4262 syslog daemons again.
4264 * The libudev API gained the new
4265 udev_device_new_from_device_id() call.
4267 * The logic for file system namespace (ReadOnlyDirectory=,
4268 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
4269 require pivot_root() anymore. This means fewer temporary
4270 directories are created below /tmp for this feature.
4272 * nspawn containers will now see and receive all submounts
4273 made on the host OS below the root file system of the
4276 * Forward Secure Sealing is now supported for Journal files,
4277 which provide cryptographical sealing of journal files so
4278 that attackers cannot alter log history anymore without this
4279 being detectable. Lennart will soon post a blog story about
4280 this explaining it in more detail.
4282 * There are two new service settings RestartPreventExitStatus=
4283 and SuccessExitStatus= which allow configuration of exit
4284 status (exit code or signal) which will be excepted from the
4285 restart logic, resp. consider successful.
4287 * journalctl gained the new --verify switch that can be used
4288 to check the integrity of the structure of journal files and
4289 (if Forward Secure Sealing is enabled) the contents of
4292 * nspawn containers will now be run with /dev/stdin, /dev/fd/
4293 and similar symlinks pre-created. This makes running shells
4294 as container init process a lot more fun.
4296 * The fstab support can now handle PARTUUID= and PARTLABEL=
4299 * A new ConditionHost= condition has been added to match
4300 against the hostname (with globs) and machine ID. This is
4301 useful for clusters where a single OS image is used to
4302 provision a large number of hosts which shall run slightly
4303 different sets of services.
4305 * Services which hit the restart limit will now be placed in a
4308 Contributions from: Bertram Poettering, Dave Reisner, Huang
4309 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
4310 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
4314 * When running in --user mode systemd will now become a
4315 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
4316 tree a lot more organized.
4318 * A new PartOf= unit dependency type has been introduced that
4319 may be used to group services in a natural way.
4321 * "systemctl enable" may now be used to enable instances of
4324 * journalctl now prints error log levels in red, and
4325 warning/notice log levels in bright white. It also supports
4326 filtering by log level now.
4328 * cgtop gained a new -n switch (similar to top), to configure
4329 the maximum number of iterations to run for. It also gained
4330 -b, to run in batch mode (accepting no input).
4332 * The suffix ".service" may now be omitted on most systemctl
4333 command lines involving service unit names.
4335 * There's a new bus call in logind to lock all sessions, as
4336 well as a loginctl verb for it "lock-sessions".
4338 * libsystemd-logind.so gained a new call sd_journal_perror()
4339 that works similar to libc perror() but logs to the journal
4340 and encodes structured information about the error number.
4342 * /etc/crypttab entries now understand the new keyfile-size=
4345 * shutdown(8) now can send a (configurable) wall message when
4346 a shutdown is cancelled.
4348 * The mount propagation mode for the root file system will now
4349 default to "shared", which is useful to make containers work
4350 nicely out-of-the-box so that they receive new mounts from
4351 the host. This can be undone locally by running "mount
4352 --make-rprivate /" if needed.
4354 * The prefdm.service file has been removed. Distributions
4355 should maintain this unit downstream if they intend to keep
4356 it around. However, we recommend writing normal unit files
4357 for display managers instead.
4359 * Since systemd is a crucial part of the OS we will now
4360 default to a number of compiler switches that improve
4361 security (hardening) such as read-only relocations, stack
4362 protection, and suchlike.
4364 * The TimeoutSec= setting for services is now split into
4365 TimeoutStartSec= and TimeoutStopSec= to allow configuration
4366 of individual time outs for the start and the stop phase of
4369 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
4370 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
4371 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
4372 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
4373 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
4374 Gundersen, Zbigniew Jędrzejewski-Szmek
4378 * The journal and id128 C APIs are now fully documented as man
4381 * Extra safety checks have been added when transitioning from
4382 the initial RAM disk to the main system to avoid accidental
4385 * /etc/crypttab entries now understand the new keyfile-offset=
4388 * systemctl -t can now be used to filter by unit load state.
4390 * The journal C API gained the new sd_journal_wait() call to
4391 make writing synchronous journal clients easier.
4393 * journalctl gained the new -D switch to show journals from a
4396 * journalctl now displays a special marker between log
4397 messages of two different boots.
4399 * The journal is now explicitly flushed to /var via a service
4400 systemd-journal-flush.service, rather than implicitly simply
4401 by seeing /var/log/journal to be writable.
4403 * journalctl (and the journal C APIs) can now match for much
4404 more complex expressions, with alternatives and
4407 * When transitioning from the initial RAM disk to the main
4408 system we will now kill all processes in a killing spree to
4409 ensure no processes stay around by accident.
4411 * Three new specifiers may be used in unit files: %u, %h, %s
4412 resolve to the user name, user home directory resp. user
4413 shell. This is useful for running systemd user instances.
4415 * We now automatically rotate journal files if their data
4416 object hash table gets a fill level > 75%. We also size the
4417 hash table based on the configured maximum file size. This
4418 together should lower hash collisions drastically and thus
4419 speed things up a bit.
4421 * journalctl gained the new "--header" switch to introspect
4422 header data of journal files.
4424 * A new setting SystemCallFilters= has been added to services
4425 which may be used to apply blacklists or whitelists to
4426 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
4428 * nspawn gained a new --link-journal= switch (and quicker: -j)
4429 to link the container journal with the host. This makes it
4430 very easy to centralize log viewing on the host for all
4431 guests while still keeping the journal files separated.
4433 * Many bugfixes and optimizations
4435 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
4436 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
4437 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
4442 * Several tools now understand kernel command line arguments,
4443 which are only read when run in an initial RAM disk. They
4444 usually follow closely their normal counterparts, but are
4447 * There's a new tool to analyze the readahead files that are
4448 automatically generated at boot. Use:
4450 /usr/lib/systemd/systemd-readahead analyze /.readahead
4452 * We now provide an early debug shell on tty9 if this enabled. Use:
4454 systemctl enable debug-shell.service
4456 * All plymouth related units have been moved into the Plymouth
4457 package. Please make sure to upgrade your Plymouth version
4460 * systemd-tmpfiles now supports getting passed the basename of
4461 a configuration file only, in which case it will look for it
4462 in all appropriate directories automatically.
4464 * udevadm info now takes a /dev or /sys path as argument, and
4465 does the right thing. Example:
4467 udevadm info /dev/sda
4468 udevadm info /sys/class/block/sda
4470 * systemctl now prints a warning if a unit is stopped but a
4471 unit that might trigger it continues to run. Example: a
4472 service is stopped but the socket that activates it is left
4475 * "systemctl status" will now mention if the log output was
4476 shortened due to rotation since a service has been started.
4478 * The journal API now exposes functions to determine the
4479 "cutoff" times due to rotation.
4481 * journald now understands SIGUSR1 and SIGUSR2 for triggering
4482 immediately flushing of runtime logs to /var if possible,
4483 resp. for triggering immediate rotation of the journal
4486 * It is now considered an error if a service is attempted to
4487 be stopped that is not loaded.
4489 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
4491 * systemd-analyze now supports Python 3
4493 * tmpfiles now supports cleaning up directories via aging
4494 where the first level dirs are always kept around but
4495 directories beneath it automatically aged. This is enabled
4496 by prefixing the age field with '~'.
4498 * Seat objects now expose CanGraphical, CanTTY properties
4499 which is required to deal with very fast bootups where the
4500 display manager might be running before the graphics drivers
4501 completed initialization.
4503 * Seat objects now expose a State property.
4505 * We now include RPM macros for service enabling/disabling
4506 based on the preset logic. We recommend RPM based
4507 distributions to make use of these macros if possible. This
4508 makes it simpler to reuse RPM spec files across
4511 * We now make sure that the collected systemd unit name is
4512 always valid when services log to the journal via
4515 * There's a new man page kernel-command-line(7) detailing all
4516 command line options we understand.
4518 * The fstab generator may now be disabled at boot by passing
4519 fstab=0 on the kernel command line.
4521 * A new kernel command line option modules-load= is now understood
4522 to load a specific kernel module statically, early at boot.
4524 * Unit names specified on the systemctl command line are now
4525 automatically escaped as needed. Also, if file system or
4526 device paths are specified they are automatically turned
4527 into the appropriate mount or device unit names. Example:
4529 systemctl status /home
4530 systemctl status /dev/sda
4532 * The SysVConsole= configuration option has been removed from
4533 system.conf parsing.
4535 * The SysV search path is no longer exported on the D-Bus
4538 * The Names= option is been removed from unit file parsing.
4540 * There's a new man page bootup(7) detailing the boot process.
4542 * Every unit and every generator we ship with systemd now
4543 comes with full documentation. The self-explanatory boot is
4546 * A couple of services gained "systemd-" prefixes in their
4547 name if they wrap systemd code, rather than only external
4548 code. Among them fsck@.service which is now
4549 systemd-fsck@.service.
4551 * The HaveWatchdog property has been removed from the D-Bus
4554 * systemd.confirm_spawn= on the kernel command line should now
4557 * There's a new man page crypttab(5) which details all options
4558 we actually understand.
4560 * systemd-nspawn gained a new --capability= switch to pass
4561 additional capabilities to the container.
4563 * timedated will now read known NTP implementation unit names
4564 from /usr/lib/systemd/ntp-units.d/*.list,
4565 systemd-timedated-ntp.target has been removed.
4567 * journalctl gained a new switch "-b" that lists log data of
4568 the current boot only.
4570 * The notify socket is in the abstract namespace again, in
4571 order to support daemons which chroot() at start-up.
4573 * There is a new Storage= configuration option for journald
4574 which allows configuration of where log data should go. This
4575 also provides a way to disable journal logging entirely, so
4576 that data collected is only forwarded to the console, the
4577 kernel log buffer or another syslog implementation.
4579 * Many bugfixes and optimizations
4581 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
4582 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
4583 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
4584 Shawn Landden, Tom Gundersen
4588 * "systemctl help <unit>" now shows the man page if one is
4591 * Several new man pages have been added.
4593 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
4594 MaxLevelConsole= can now be specified in
4595 journald.conf. These options allow reducing the amount of
4596 data stored on disk or forwarded by the log level.
4598 * TimerSlackNSec= can now be specified in system.conf for
4599 PID1. This allows system-wide power savings.
4601 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
4602 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
4607 * logind is now capable of (optionally) handling power and
4608 sleep keys as well as the lid switch.
4610 * journalctl now understands the syntax "journalctl
4611 /usr/bin/avahi-daemon" to get all log output of a specific
4614 * CapabilityBoundingSet= in system.conf now also influences
4615 the capability bound set of usermode helpers of the kernel.
4617 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
4618 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
4619 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
4620 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
4624 * Note that we skipped 139 releases here in order to set the
4625 new version to something that is greater than both udev's
4626 and systemd's most recent version number.
4628 * udev: all udev sources are merged into the systemd source tree now.
4629 All future udev development will happen in the systemd tree. It
4630 is still fully supported to use the udev daemon and tools without
4631 systemd running, like in initramfs or other init systems. Building
4632 udev though, will require the *build* of the systemd tree, but
4633 udev can be properly *run* without systemd.
4635 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
4636 should be used to create dead device nodes as workarounds for broken
4639 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
4640 no longer supported. udev_monitor_new_from_netlink() needs to be
4641 used to subscribe to events.
4643 * udev: when udevd is started by systemd, processes which are left
4644 behind by forking them off of udev rules, are unconditionally cleaned
4645 up and killed now after the event handling has finished. Services or
4646 daemons must be started as systemd services. Services can be
4647 pulled-in by udev to get started, but they can no longer be directly
4648 forked by udev rules.
4650 * udev: the daemon binary is called systemd-udevd now and installed
4651 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
4652 to adapt to that, create symlink, or rename the binary after building
4655 * libudev no longer provides these symbols:
4656 udev_monitor_from_socket()
4657 udev_queue_get_failed_list_entry()
4658 udev_get_{dev,sys,run}_path()
4659 The versions number was bumped and symbol versioning introduced.
4661 * systemd-loginctl and systemd-journalctl have been renamed
4662 to loginctl and journalctl to match systemctl.
4664 * The config files: /etc/systemd/systemd-logind.conf and
4665 /etc/systemd/systemd-journald.conf have been renamed to
4666 logind.conf and journald.conf. Package updates should rename
4667 the files to the new names on upgrade.
4669 * For almost all files the license is now LGPL2.1+, changed
4670 from the previous GPL2.0+. Exceptions are some minor stuff
4671 of udev (which will be changed to LGPL2.1 eventually, too),
4672 and the MIT licensed sd-daemon.[ch] library that is suitable
4673 to be used as drop-in files.
4675 * systemd and logind now handle system sleep states, in
4676 particular suspending and hibernating.
4678 * logind now implements a sleep/shutdown/idle inhibiting logic
4679 suitable for a variety of uses. Soonishly Lennart will blog
4680 about this in more detail.
4682 * var-run.mount and var-lock.mount are no longer provided
4683 (which prevously bind mounted these directories to their new
4684 places). Distributions which have not converted these
4685 directories to symlinks should consider stealing these files
4686 from git history and add them downstream.
4688 * We introduced the Documentation= field for units and added
4689 this to all our shipped units. This is useful to make it
4690 easier to explore the boot and the purpose of the various
4693 * All smaller setup units (such as
4694 systemd-vconsole-setup.service) now detect properly if they
4695 are run in a container and are skipped when
4696 appropriate. This guarantees an entirely noise-free boot in
4697 Linux container environments such as systemd-nspawn.
4699 * A framework for implementing offline system updates is now
4700 integrated, for details see:
4701 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
4703 * A new service type Type=idle is available now which helps us
4704 avoiding ugly interleaving of getty output and boot status
4707 * There's now a system-wide CapabilityBoundingSet= option to
4708 globally reduce the set of capabilities for the
4709 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
4710 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
4711 even CAP_NET_ADMIN system-wide for secure systems.
4713 * There are now system-wide DefaultLimitXXX= options to
4714 globally change the defaults of the various resource limits
4715 for all units started by PID 1.
4717 * Harald Hoyer's systemd test suite has been integrated into
4718 systemd which allows easy testing of systemd builds in qemu
4719 and nspawn. (This is really awesome! Ask us for details!)
4721 * The fstab parser is now implemented as generator, not inside
4724 * systemctl will now warn you if .mount units generated from
4725 /etc/fstab are out of date due to changes in fstab that
4726 have not been read by systemd yet.
4728 * systemd is now suitable for usage in initrds. Dracut has
4729 already been updated to make use of this. With this in place
4730 initrds get a slight bit faster but primarily are much
4731 easier to introspect and debug since "systemctl status" in
4732 the host system can be used to introspect initrd services,
4733 and the journal from the initrd is kept around too.
4735 * systemd-delta has been added, a tool to explore differences
4736 between user/admin configuration and vendor defaults.
4738 * PrivateTmp= now affects both /tmp and /var/tmp.
4740 * Boot time status messages are now much prettier and feature
4741 proper english language. Booting up systemd has never been
4744 * Read-ahead pack files now include the inode number of all
4745 files to pre-cache. When the inode changes the pre-caching
4746 is not attempted. This should be nicer to deal with updated
4747 packages which might result in changes of read-ahead
4750 * We now temporaritly lower the kernel's read_ahead_kb variable
4751 when collecting read-ahead data to ensure the kernel's
4752 built-in read-ahead does not add noise to our measurements
4753 of necessary blocks to pre-cache.
4755 * There's now RequiresMountsFor= to add automatic dependencies
4756 for all mounts necessary for a specific file system path.
4758 * MountAuto= and SwapAuto= have been removed from
4759 system.conf. Mounting file systems at boot has to take place
4762 * nspawn now learned a new switch --uuid= to set the machine
4763 ID on the command line.
4765 * nspawn now learned the -b switch to automatically search
4768 * vt102 is now the default TERM for serial TTYs, upgraded from
4771 * systemd-logind now works on VT-less systems.
4773 * The build tree has been reorganized. The individual
4774 components now have directories of their own.
4776 * A new condition type ConditionPathIsReadWrite= is now available.
4778 * nspawn learned the new -C switch to create cgroups for the
4779 container in other hierarchies.
4781 * We now have support for hardware watchdogs, configurable in
4784 * The scheduled shutdown logic now has a public API.
4786 * We now mount /tmp as tmpfs by default, but this can be
4787 masked and /etc/fstab can override it.
4789 * Since udisks does not make use of /media anymore we are not
4790 mounting a tmpfs on it anymore.
4792 * journalctl gained a new --local switch to only interleave
4793 locally generated journal files.
4795 * We can now load the IMA policy at boot automatically.
4797 * The GTK tools have been split off into a systemd-ui.
4799 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
4800 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
4801 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
4802 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
4803 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
4804 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
4805 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
4806 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
4807 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
4812 * This is mostly a bugfix release
4814 * Support optional initialization of the machine ID from the
4815 KVM or container configured UUID.
4817 * Support immediate reboots with "systemctl reboot -ff"
4819 * Show /etc/os-release data in systemd-analyze output
4821 * Many bugfixes for the journal, including endianness fixes and
4822 ensuring that disk space enforcement works
4824 * sd-login.h is C++ comptaible again
4826 * Extend the /etc/os-release format on request of the Debian
4829 * We now refuse non-UTF8 strings used in various configuration
4830 and unit files. This is done to ensure we do not pass invalid
4831 data over D-Bus or expose it elsewhere.
4833 * Register Mimo USB Screens as suitable for automatic seat
4836 * Read SELinux client context from journal clients in a race
4839 * Reorder configuration file lookup order. /etc now always
4840 overrides /run in order to allow the administrator to always
4841 and unconditionally override vendor supplied or
4842 automatically generated data.
4844 * The various user visible bits of the journal now have man
4845 pages. We still lack man pages for the journal API calls
4848 * We now ship all man pages in HTML format again in the
4851 Contributions from: Dave Reisner, Dirk Eibach, Frederic
4852 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
4853 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
4858 * This is mostly a bugfix release
4860 * systems lacking /etc/os-release are no longer supported.
4862 * Various functionality updates to libsystemd-login.so
4864 * Track class of PAM logins to distinguish greeters from
4867 Contributions from: Kay Sievers, Lennart Poettering, Michael
4872 * This is an important bugfix release for v41.
4874 * Building man pages is now optional which should be useful
4875 for those building systemd from git but unwilling to install
4878 * Watchdog support for supervising services is now usable. In
4879 a future release support for hardware watchdogs
4880 (i.e. /dev/watchdog) will be added building on this.
4882 * Service start rate limiting is now configurable and can be
4883 turned off per service. When a start rate limit is hit a
4884 reboot can automatically be triggered.
4886 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
4888 Contributions from: Benjamin Franzke, Bill Nottingham,
4889 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
4890 Schmidt, Michał Górny, Piotr Drąg
4894 * The systemd binary is installed /usr/lib/systemd/systemd now;
4895 An existing /sbin/init symlink needs to be adapted with the
4898 * The code that loads kernel modules has been ported to invoke
4899 libkmod directly, instead of modprobe. This means we do not
4900 support systems with module-init-tools anymore.
4902 * Watchdog support is now already useful, but still not
4905 * A new kernel command line option systemd.setenv= is
4906 understood to set system wide environment variables
4907 dynamically at boot.
4909 * We now limit the set of capabilities of systemd-journald.
4911 * We now set SIGPIPE to ignore by default, since it only is
4912 useful in shell pipelines, and has little use in general
4913 code. This can be disabled with IgnoreSIPIPE=no in unit
4916 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
4917 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
4922 * This is mostly a bugfix release
4924 * We now expose the reason why a service failed in the
4925 "Result" D-Bus property.
4927 * Rudimentary service watchdog support (will be completed over
4928 the next few releases.)
4930 * When systemd forks off in order execute some service we will
4931 now immediately changes its argv[0] to reflect which process
4932 it will execute. This is useful to minimize the time window
4933 with a generic argv[0], which makes bootcharts more useful
4935 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
4936 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
4937 Mike Kazantsev, Ray Strode
4941 * This is mostly a test release, but incorporates many
4944 * New systemd-cgtop tool to show control groups by their
4947 * Linking against libacl for ACLs is optional again. If
4948 disabled, support tracking device access for active logins
4949 goes becomes unavailable, and so does access to the user
4950 journals by the respective users.
4952 * If a group "adm" exists, journal files are automatically
4953 owned by them, thus allow members of this group full access
4954 to the system journal as well as all user journals.
4956 * The journal now stores the SELinux context of the logging
4957 client for all entries.
4959 * Add C++ inclusion guards to all public headers
4961 * New output mode "cat" in the journal to print only text
4962 messages, without any meta data like date or time.
4964 * Include tiny X server wrapper as a temporary stop-gap to
4965 teach XOrg udev display enumeration. This is used by display
4966 managers such as gdm, and will go away as soon as XOrg
4967 learned native udev hotplugging for display devices.
4969 * Add new systemd-cat tool for executing arbitrary programs
4970 with STDERR/STDOUT connected to the journal. Can also act as
4971 BSD logger replacement, and does so by default.
4973 * Optionally store all locally generated coredumps in the
4974 journal along with meta data.
4976 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
4977 writing short strings to files (for usage for /sys), and for
4978 creating symlinks, character and block device nodes.
4980 * New unit file option ControlGroupPersistent= to make cgroups
4981 persistent, following the mechanisms outlined in
4982 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
4984 * Support multiple local RTCs in a sane way
4986 * No longer monopolize IO when replaying readahead data on
4987 rotating disks, since we might starve non-file-system IO to
4988 death, since fanotify() will not see accesses done by blkid,
4991 * Do not show kernel threads in systemd-cgls anymore, unless
4992 requested with new -k switch.
4994 Contributions from: Dan Horák, Kay Sievers, Lennart
4995 Poettering, Michal Schmidt
4999 * This is mostly a test release, but incorporates many
5002 * The git repository moved to:
5003 git://anongit.freedesktop.org/systemd/systemd
5004 ssh://git.freedesktop.org/git/systemd/systemd
5006 * First release with the journal
5007 http://0pointer.de/blog/projects/the-journal.html
5009 * The journal replaces both systemd-kmsg-syslogd and
5010 systemd-stdout-bridge.
5012 * New sd_pid_get_unit() API call in libsystemd-logind
5014 * Many systemadm clean-ups
5016 * Introduce remote-fs-pre.target which is ordered before all
5017 remote mounts and may be used to start services before all
5020 * Added Mageia support
5022 * Add bash completion for systemd-loginctl
5024 * Actively monitor PID file creation for daemons which exit in
5025 the parent process before having finished writing the PID
5026 file in the daemon process. Daemons which do this need to be
5027 fixed (i.e. PID file creation must have finished before the
5028 parent exits), but we now react a bit more gracefully to them.
5030 * Add colourful boot output, mimicking the well-known output
5031 of existing distributions.
5033 * New option PassCredentials= for socket units, for
5034 compatibility with a recent kernel ABI breakage.
5036 * /etc/rc.local is now hooked in via a generator binary, and
5037 thus will no longer act as synchronization point during
5040 * systemctl list-unit-files now supports --root=.
5042 * systemd-tmpfiles now understands two new commands: z, Z for
5043 relabelling files according to the SELinux database. This is
5044 useful to apply SELinux labels to specific files in /sys,
5047 * Output of SysV services is now forwarded to both the console
5048 and the journal by default, not only just the console.
5050 * New man pages for all APIs from libsystemd-login.
5052 * The build tree got reorganized and a the build system is a
5053 lot more modular allowing embedded setups to specifically
5054 select the components of systemd they are interested in.
5056 * Support for Linux systems lacking the kernel VT subsystem is
5059 * configure's --with-rootdir= got renamed to
5060 --with-rootprefix= to follow the naming used by udev and
5063 * Unless specified otherwise we will now install to /usr instead
5064 of /usr/local by default.
5066 * Processes with '@' in argv[0][0] are now excluded from the
5067 final shut-down killing spree, following the logic explained
5069 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
5071 * All processes remaining in a service cgroup when we enter
5072 the START or START_PRE states are now killed with
5073 SIGKILL. That means it is no longer possible to spawn
5074 background processes from ExecStart= lines (which was never
5075 supported anyway, and bad style).
5077 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
5078 reloading of units together.
5080 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
5081 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
5082 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
5083 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
5084 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek