1 systemd System and Service Manager
4 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
5 There are no known issues with current sysfs, and udev does not need
6 or should be used to work around such bugs.
8 * udev does no longer enable USB HID power management. Several reports
9 indicate, that some devices cannot handle that setting.
11 * The udev accelerometer helper was removed. The functionality
12 is now fully included in iio-sensor-proxy. But this means,
13 older iio-sensor-proxy versions will no longer provide
14 accelerometer/orientation data with this systemd version.
15 Please upgrade iio-sensor-proxy to version 1.0.
17 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
18 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
19 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
20 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
21 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens,
22 Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
23 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
24 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
25 Tom Gundersen, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
31 * The sd-bus.h and sd-event.h APIs have now been declared
32 stable and have been added to the official interface of
33 libsystemd.so. sd-bus implements an alternative D-Bus client
34 library, that is relatively easy to use, very efficient and
35 supports both classic D-Bus as well as kdbus as transport
36 backend. sd-event is a generic event loop abstraction that
37 is built around Linux epoll, but adds features such as event
38 prioritization or efficient timer handling. Both APIs are good
39 choices for C programs looking for a bus and/or event loop
40 implementation that is minimal and does not have to be
41 portable to other kernels.
43 * kdbus support is no longer compile-time optional. It is now
44 always built-in. However, it can still be disabled at
45 runtime using the kdbus=0 kernel command line setting, and
46 that setting may be changed to default to off, by specifying
47 --disable-kdbus at build-time. Note though that the kernel
48 command line setting has no effect if the kdbus.ko kernel
49 module is not installed, in which case kdbus is (obviously)
50 also disabled. We encourage all downstream distributions to
51 begin testing kdbus by adding it to the kernel images in the
52 development distributions, and leaving kdbus support in
55 * The minimal required util-linux version has been bumped to
58 * Support for chkconfig (--enable-chkconfig) was removed in
59 favor of calling an abstraction tool
60 /lib/systemd/systemd-sysv-install. This needs to be
61 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
62 in README for details.
64 * If there's a systemd unit and a SysV init script for the
65 same service name, and the user executes "systemctl enable"
66 for it (or a related call), then this will now enable both
67 (or execute the related operation on both), not just the
70 * The libudev API documentation has been converted from gtkdoc
73 * gudev has been removed from the systemd tree, it is now an
76 * The systemd-cgtop tool learnt a new --raw switch to generate
77 "raw" (machine parsable) output.
79 * networkd's IPForwarding= .network file setting learnt the
80 new setting "kernel", which ensures that networkd does not
81 change the IP forwarding sysctl from the default kernel
84 * The systemd-logind bus API now exposes a new boolean
85 property "Docked" that reports whether logind considers the
86 system "docked", i.e. connected to a docking station or not.
88 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
89 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
90 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
91 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
92 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
93 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
94 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
95 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
96 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
97 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
98 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
99 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
100 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
101 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
102 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
103 Fink, Zbigniew Jędrzejewski-Szmek
105 -- Berlin, 2015-06-19
109 * The gudev library has been extracted into a separate repository
110 available at: https://git.gnome.org/browse/libgudev/
111 It is now managed as part of the Gnome project. Distributions
112 are recommended to pass --disable-gudev to systemd and use
113 gudev from the Gnome project instead. gudev is still included
114 in systemd, for now. It will be removed soon, though. Please
115 also see the announcement-thread on systemd-devel:
116 http://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
118 * systemd now exposes a CPUUsageNSec= property for each
119 service unit on the bus, that contains the overall consumed
120 CPU time of a service (the sum of what each process of the
121 service consumed). This value is only available if
122 CPUAccounting= is turned on for a service, and is then shown
123 in the "systemctl status" output.
125 * Support for configuring alternative mappings of the old SysV
126 runlevels to systemd targets has been removed. They are now
127 hardcoded in a way that runlevels 2, 3, 4 all map to
128 multi-user.target and 5 to graphical.target (which
129 previously was already the default behaviour).
131 * The auto-mounter logic gained support for mount point
132 expiry, using a new TimeoutIdleSec= setting in .automount
133 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
135 * The EFI System Partition (ESP) as mounted to /boot by
136 systemd-efi-boot-generator will now be unmounted
137 automatically after 2 minutes of not being used. This should
138 minimize the risk of ESP corruptions.
140 * New /etc/fstab options x-systemd.requires= and
141 x-systemd.requires-mounts-for= are now supported to express
142 additional dependencies for mounts. This is useful for
143 journalling file systems that support external journal
144 devices or overlay file systems that require underlying file
145 systems to be mounted.
147 * systemd does not support direct live-upgrades (via systemctl
148 daemon-reexec) from versions older than v44 anymore. As no
149 distribution we are aware of shipped such old versions in a
150 stable release this should not be problematic.
152 * When systemd forks off a new per-connection service instance
153 it will now set the $REMOTE_ADDR environment variable to the
154 remote IP address, and $REMOTE_PORT environment variable to
155 the remote IP port. This behaviour is similar to the
156 corresponding environment variables defined by CGI.
158 * systemd-networkd gained support for uplink failure
159 detection. The BindCarrier= option allows binding interface
160 configuration dynamically to the link sense of other
161 interfaces. This is useful to achieve behaviour like in
164 * systemd-networkd gained support for configuring the DHCP
165 client identifier to use when requesting leases.
167 * systemd-networkd now has a per-network UseNTP= option to
168 configure whether NTP server information acquired via DHCP
169 is passed on to services like systemd-timesyncd.
171 * systemd-networkd gained support for vti6 tunnels.
173 * Note that systemd-networkd manages the sysctl variable
174 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
175 it is configured for since v219. The variable controls IP
176 forwarding, and is a per-interface alternative to the global
177 /proc/sys/net/ipv[46]/ip_forward. This setting is
178 configurable in the IPForward= option, which defaults to
179 "no". This means if networkd is used for an interface it is
180 no longer sufficient to set the global sysctl option to turn
181 on IP forwarding! Instead, the .network file option
182 IPForward= needs to be turned on! Note that the
183 implementation of this behaviour was broken in v219 and has
186 * Many bonding and vxlan options are now configurable in
189 * systemd-nspawn gained a new --property= setting to set unit
190 properties for the container scope. This is useful for
191 setting resource parameters (e.g "CPUShares=500") on
192 containers started from the command line.
194 * systemd-nspawn gained a new --private-users= switch to make
195 use of user namespacing available on recent Linux kernels.
197 * systemd-nspawn may now be called as part of a shell pipeline
198 in which case the pipes used for stdin and stdout are passed
199 directly to the process invoked in the container, without
200 indirection via a pseudo tty.
202 * systemd-nspawn gained a new switch to control the UNIX
203 signal to use when killing the init process of the container
206 * systemd-nspawn gained a new --overlay= switch for mounting
207 overlay file systems into the container using the new kernel
210 * When a container image is imported via systemd-importd and
211 the host file system is not btrfs, a loopback block device
212 file is created in /var/lib/machines.raw with a btrfs file
213 system inside. It is then mounted to /var/lib/machines to
214 enable btrfs features for container management. The loopback
215 file and btrfs file system is grown as needed when container
216 images are imported via systemd-importd.
218 * systemd-machined/systemd-importd gained support for btrfs
219 quota, to enforce container disk space limits on disk. This
220 is exposed in "machinectl set-limit".
222 * systemd-importd now can import containers from local .tar,
223 .raw and .qcow2 images, and export them to .tar and .raw. It
224 can also import dkr v2 images now from the network (on top
227 * systemd-importd gained support for verifying downloaded
228 images with gpg2 (previously only gpg1 was supported).
230 * systemd-machined, systemd-logind, systemd: most bus calls
231 are now accessible to unprivileged processes via
232 PolicyKit. Also, systemd-logind will now allow users to kill
233 their own sessions without further privileges or
236 * systemd-shutdownd has been removed. This service was
237 previously responsible for implementing scheduled shutdowns
238 as exposed in /usr/bin/shutdown's time parameter. This
239 functionality has now been moved into systemd-logind and is
240 accessible via a bus interface.
242 * "systemctl reboot" gained a new switch --firmware-setup that
243 can be used to reboot into the EFI firmware setup, if that
244 is available. systemd-logind now exposes an API on the bus
245 to trigger such reboots, in case graphical desktop UIs want
246 to cover this functionality.
248 * "systemctl enable", "systemctl disable" and "systemctl mask"
249 now support a new "--now" switch. If specified the units
250 that are enabled will also be started, and the ones
251 disabled/masked also stopped.
253 * The Gummiboot EFI boot loader tool has been merged into
254 systemd, and renamed to "systemd-boot". The bootctl tool has been
255 updated to support systemd-boot.
257 * An EFI kernel stub has been added that may be used to create
258 kernel EFI binaries that contain not only the actual kernel,
259 but also an initrd, boot splash, command line and OS release
260 information. This combined binary can then be signed as a
261 single image, so that the firmware can verify it all in one
262 step. systemd-boot has special support for EFI binaries created
263 like this and can extract OS release information from them
264 and show them in the boot menu. This functionality is useful
265 to implement cryptographically verified boot schemes.
267 * Optional support has been added to systemd-fsck to pass
268 fsck's progress report to an AF_UNIX socket in the file
271 * udev will no longer create device symlinks for all block
272 devices by default. A blacklist for excluding special block
273 devices from this logic has been turned into a whitelist
274 that requires picking block devices explicitly that require
277 * A new (currently still internal) API sd-device.h has been
278 added to libsystemd. This modernized API is supposed to
279 replace libudev eventually. In fact, already much of libudev
280 is now just a wrapper around sd-device.h.
282 * A new hwdb database for storing metadata about pointing
283 stick devices has been added.
285 * systemd-tmpfiles gained support for setting file attributes
286 similar to the "chattr" tool with new 'h' and 'H' lines.
288 * systemd-journald will no longer unconditionally set the
289 btrfs NOCOW flag on new journal files. This is instead done
290 with tmpfiles snippet using the new 'h' line type. This
291 allows easy disabling of this logic, by masking the
292 journal-nocow.conf tmpfiles file.
294 * systemd-journald will now translate audit message types to
295 human readable identifiers when writing them to the
296 journal. This should improve readability of audit messages.
298 * The LUKS logic gained support for the offset= and skip=
299 options in /etc/crypttab, as previously implemented by
302 * /usr/lib/os-release gained a new optional field VARIANT= for
303 distributions that support multiple variants (such as a
304 desktop edition, a server edition, ...)
306 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
307 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
308 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
309 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
310 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
311 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
312 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
313 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
314 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
315 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
316 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
317 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
318 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
319 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
320 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
321 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
322 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
323 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
324 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
325 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
326 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
327 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
328 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
329 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
330 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
331 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
332 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
334 -- Berlin, 2015-05-22
338 * Introduce a new API "sd-hwdb.h" for querying the hardware
339 metadata database. With this minimal interface one can query
340 and enumerate the udev hwdb, decoupled from the old libudev
341 library. libudev's interface for this is now only a wrapper
342 around sd-hwdb. A new tool systemd-hwdb has been added to
343 interface with and update the database.
345 * When any of systemd's tools copies files (for example due to
346 tmpfiles' C lines) a btrfs reflink will attempted first,
347 before bytewise copying is done.
349 * systemd-nspawn gained a new --ephemeral switch. When
350 specified a btrfs snapshot is taken of the container's root
351 directory, and immediately removed when the container
352 terminates again. Thus, a container can be started whose
353 changes never alter the container's root directory, and are
354 lost on container termination. This switch can also be used
355 for starting a container off the root file system of the
356 host without affecting the host OS. This switch is only
357 available on btrfs file systems.
359 * systemd-nspawn gained a new --template= switch. It takes the
360 path to a container tree to use as template for the tree
361 specified via --directory=, should that directory be
362 missing. This allows instantiating containers dynamically,
363 on first run. This switch is only available on btrfs file
366 * When a .mount unit refers to a mount point on which multiple
367 mounts are stacked, and the .mount unit is stopped all of
368 the stacked mount points will now be unmounted until no
371 * systemd now has an explicit notion of supported and
372 unsupported unit types. Jobs enqueued for unsupported unit
373 types will now fail with an "unsupported" error code. More
374 specifically .swap, .automount and .device units are not
375 supported in containers, .busname units are not supported on
376 non-kdbus systems. .swap and .automount are also not
377 supported if their respective kernel compile time options
380 * machinectl gained support for two new "copy-from" and
381 "copy-to" commands for copying files from a running
382 container to the host or vice versa.
384 * machinectl gained support for a new "bind" command to bind
385 mount host directories into local containers. This is
386 currently only supported for nspawn containers.
388 * networkd gained support for configuring bridge forwarding
389 database entries (fdb) from .network files.
391 * A new tiny daemon "systemd-importd" has been added that can
392 download container images in tar, raw, qcow2 or dkr formats,
393 and make them available locally in /var/lib/machines, so
394 that they can run as nspawn containers. The daemon can GPG
395 verify the downloads (not supported for dkr, since it has no
396 provisions for verifying downloads). It will transparently
397 decompress bz2, xz, gzip compressed downloads if necessary,
398 and restore sparse files on disk. The daemon uses privilege
399 separation to ensure the actual download logic runs with
400 fewer privileges than the deamon itself. machinectl has
401 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
402 make the functionality of importd available to the
403 user. With this in place the Fedora and Ubuntu "Cloud"
404 images can be downloaded and booted as containers unmodified
405 (the Fedora images lack the appropriate GPG signature files
406 currently, so they cannot be verified, but this will change
407 soon, hopefully). Note that downloading images is currently
408 only fully supported on btrfs.
410 * machinectl is now able to list container images found in
411 /var/lib/machines, along with some metadata about sizes of
412 disk and similar. If the directory is located on btrfs and
413 quota is enabled, this includes quota display. A new command
414 "image-status" has been added that shows additional
415 information about images.
417 * machinectl is now able to clone container images
418 efficiently, if the underlying file system (btrfs) supports
419 it, with the new "machinectl list-images" command. It also
420 gained commands for renaming and removing images, as well as
421 marking them read-only or read-write (supported also on
422 legacy file systems).
424 * networkd gained support for collecting LLDP network
425 announcements, from hardware that supports this. This is
426 shown in networkctl output.
428 * systemd-run gained support for a new -t (--pty) switch for
429 invoking a binary on a pty whose input and output is
430 connected to the invoking terminal. This allows executing
431 processes as system services while interactively
432 communicating with them via the terminal. Most interestingly
433 this is supported across container boundaries. Invoking
434 "systemd-run -t /bin/bash" is an alternative to running a
435 full login session, the difference being that the former
436 will not register a session, nor go through the PAM session
439 * tmpfiles gained support for a new "v" line type for creating
440 btrfs subvolumes. If the underlying file system is a legacy
441 file system, this automatically degrades to creating a
442 normal directory. Among others /var/lib/machines is now
443 created like this at boot, should it be missing.
445 * The directory /var/lib/containers/ has been deprecated and
446 been replaced by /var/lib/machines. The term "machines" has
447 been used in the systemd context as generic term for both
448 VMs and containers, and hence appears more appropriate for
449 this, as the directory can also contain raw images bootable
452 * systemd-nspawn when invoked with -M but without --directory=
453 or --image= is now capable of searching for the container
454 root directory, subvolume or disk image automatically, in
455 /var/lib/machines. systemd-nspawn@.service has been updated
456 to make use of this, thus allowing it to be used for raw
459 * A new machines.target unit has been introduced that is
460 supposed to group all containers/VMs invoked as services on
461 the system. systemd-nspawn@.service has been updated to
464 * machinectl gained a new "start" command, for invoking a
465 container as a service. "machinectl start foo" is mostly
466 equivalent to "systemctl start systemd-nspawn@foo.service",
467 but handles escaping in a nicer way.
469 * systemd-nspawn will now mount most of the cgroupfs tree
470 read-only into each container, with the exception of the
471 container's own subtree in the name=systemd hierarchy.
473 * journald now sets the special FS_NOCOW file flag for its
474 journal files. This should improve performance on btrfs, by
475 avoiding heavy fragmentation when journald's write-pattern
476 is used on COW file systems. It degrades btrfs' data
477 integrity guarantees for the files to the same levels as for
478 ext3/ext4 however. This should be OK though as journald does
479 its own data integrity checks and all its objects are
480 checksummed on disk. Also, journald should handle btrfs disk
481 full events a lot more gracefully now, by processing SIGBUS
482 errors, and not relying on fallocate() anymore.
484 * When journald detects that journal files it is writing to
485 have been deleted it will immediately start new journal
488 * systemd now provides a way to store file descriptors
489 per-service in PID 1.This is useful for daemons to ensure
490 that fds they require are not lost during a daemon
491 restart. The fds are passed to the deamon on the next
492 invocation in the same way socket activation fds are
493 passed. This is now used by journald to ensure that the
494 various sockets connected to all the system's stdout/stderr
495 are not lost when journald is restarted. File descriptors
496 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
497 an extension to sd_notify(). Note that a limit is enforced
498 on the number of fds a service can store in PID 1, and it
499 defaults to 0, so that no fds may be stored, unless this is
500 explicitly turned on.
502 * The default TERM variable to use for units connected to a
503 terminal, when no other value is explicitly is set is now
504 vt220 rather than vt102. This should be fairly safe still,
505 but allows PgUp/PgDn work.
507 * The /etc/crypttab option header= as known from Debian is now
510 * "loginctl user-status" and "loginctl session-status" will
511 now show the last 10 lines of log messages of the
512 user/session following the status output. Similar,
513 "machinectl status" will show the last 10 log lines
514 associated with a virtual machine or container
515 service. (Note that this is usually not the log messages
516 done in the VM/container itself, but simply what the
517 container manager logs. For nspawn this includes all console
520 * "loginctl session-status" without further argument will now
521 show the status of the session of the caller. Similar,
522 "lock-session", "unlock-session", "activate",
523 "enable-linger", "disable-linger" may now be called without
524 session/user parameter in which case they apply to the
525 caller's session/user.
527 * An X11 session scriptlet is now shipped that uploads
528 $DISPLAY and $XAUTHORITY into the environment of the systemd
529 --user daemon if a session begins. This should improve
530 compatibility with X11 enabled applications run as systemd
533 * Generators are now subject to masking via /etc and /run, the
534 same way as unit files.
536 * networkd .network files gained support for configuring
537 per-link IPv4/IPv6 packet forwarding as well as IPv4
538 masquerading. This is by default turned on for veth links to
539 containers, as registered by systemd-nspawn. This means that
540 nspawn containers run with --network-veth will now get
541 automatic routed access to the host's networks without any
542 further configuration or setup, as long as networkd runs on
545 * systemd-nspawn gained the --port= (-p) switch to expose TCP
546 or UDP posts of a container on the host. With this in place
547 it is possible to run containers with private veth links
548 (--network-veth), and have their functionality exposed on
549 the host as if their services were running directly on the
552 * systemd-nspawn's --network-veth switch now gained a short
553 version "-n", since with the changes above it is now truly
554 useful out-of-the-box. The systemd-nspawn@.service has been
555 updated to make use of it too by default.
557 * systemd-nspawn will now maintain a per-image R/W lock, to
558 ensure that the same image is not started more than once
559 writable. (It's OK to run an image multiple times
560 simultaneously in read-only mode.)
562 * systemd-nspawn's --image= option is now capable of
563 dissecting and booting MBR and GPT disk images that contain
564 only a single active Linux partition. Previously it
565 supported only GPT disk images with proper GPT type
566 IDs. This allows running cloud images from major
567 distributions directly with systemd-nspawn, without
570 * In addition to collecting mouse dpi data in the udev
571 hardware database, there's now support for collecting angle
572 information for mouse scroll wheels. The database is
573 supposed to guarantee similar scrolling behavior on mice
574 that it knows about. There's also support for collecting
575 information about Touchpad types.
577 * udev's input_id built-in will now also collect touch screen
578 dimension data and attach it to probed devices.
580 * /etc/os-release gained support for a Distribution Privacy
583 * networkd gained support for creating "ipvlan", "gretap",
584 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
586 * systemd-tmpfiles gained support for "a" lines for setting
589 * systemd-nspawn will now mount /tmp in the container to
590 tmpfs, automatically.
592 * systemd now exposes the memory.usage_in_bytes cgroup
593 attribute and shows it for each service in the "systemctl
594 status" output, if available.
596 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
597 immediate reboot is triggered. This useful if shutdown is
598 hung and is unable to complete, to expedite the
599 operation. Note that this kind of reboot will still unmount
600 all file systems, and hence should not result in fsck being
603 * A .device unit for an optical block device will now be
604 considered active only when a medium is in the drive. Also,
605 mount units are now bound to their backing devices thus
606 triggering automatic unmounting when devices become
607 unavailable. With this in place systemd will now
608 automatically unmount left-over mounts when a CD-ROM is
609 ejected or an USB stick is yanked from the system.
611 * networkd-wait-online now has support for waiting for
612 specific interfaces only (with globbing), and for giving up
613 after a configurable timeout.
615 * networkd now exits when idle. It will be automatically
616 restarted as soon as interfaces show up, are removed or
617 change state. networkd will stay around as long as there is
618 at least one DHCP state machine or similar around, that keep
621 * networkd may now configure IPv6 link-local addressing in
622 addition to IPv4 link-local addressing.
624 * The IPv6 "token" for use in SLAAC may now be configured for
625 each .network interface in networkd.
627 * Routes configured with networkd may now be assigned a scope
630 * networkd's [Match] sections now support globbing and lists
631 of multiple space-separated matches per item.
633 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
634 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
635 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
636 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
637 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
638 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
639 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
640 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
641 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
642 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
643 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
644 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
645 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
646 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
647 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
648 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
649 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
650 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
651 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
652 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
653 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
654 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
655 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
656 Hoffmann, Zbigniew Jędrzejewski-Szmek
658 -- Berlin, 2015-02-16
662 * When querying unit file enablement status (for example via
663 "systemctl is-enabled"), a new state "indirect" is now known
664 which indicates that a unit might not be enabled itself, but
665 another unit listed in its Alias= setting might be.
667 * Similar to the various existing ConditionXYZ= settings for
668 units there are now matching AssertXYZ= settings. While
669 failing conditions cause a unit to be skipped, but its job
670 to succeed, failing assertions declared like this will cause
671 a unit start operation and its job to fail.
673 * hostnamed now knows a new chassis type "embedded".
675 * systemctl gained a new "edit" command. When used on a unit
676 file this allows extending unit files with .d/ drop-in
677 configuration snippets or editing the full file (after
678 copying it from /usr/lib to /etc). This will invoke the
679 user's editor (as configured with $EDITOR), and reload the
680 modified configuration after editing.
682 * "systemctl status" now shows the suggested enablement state
683 for a unit, as declared in the (usually vendor-supplied)
686 * nss-myhostname will now resolve the single-label host name
687 "gateway" to the locally configured default IP routing
688 gateways, ordered by their metrics. This assigns a stable
689 name to the used gateways, regardless which ones are
690 currently configured. Note that the name will only be
691 resolved after all other name sources (if nss-myhostname is
692 configured properly) and should hence not negatively impact
693 systems that use the single-label host name "gateway" in
696 * systemd-inhibit now allows filtering by mode when listing
699 * Scope and service units gained a new "Delegate" boolean
700 property, which when set allows processes running inside the
701 unit to further partition resources. This is primarily
702 useful for systemd user instances as well as container
705 * journald will now pick up audit messages directly from
706 the kernel, and log them like any other log message. The
707 audit fields are split up and fully indexed. This means that
708 journalctl in many ways is now a (nicer!) alternative to
709 ausearch, the traditional audit client. Note that this
710 implements only a minimal audit client, if you want the
711 special audit modes like reboot-on-log-overflow, please use
712 the traditional auditd instead, which can be used in
713 parallel to journald.
715 * The ConditionSecurity= unit file option now understands the
716 special string "audit" to check whether auditing is
719 * journalctl gained two new commands --vacuum-size= and
720 --vacuum-time= to delete old journal files until the
721 remaining ones take up no more the specified size on disk,
722 or are not older than the specified time.
724 * A new, native PPPoE library has been added to sd-network,
725 systemd's library of light-weight networking protocols. This
726 library will be used in a future version of networkd to
727 enable PPPoE communication without an external pppd daemon.
729 * The busctl tool now understands a new "capture" verb that
730 works similar to "monitor", but writes a packet capture
731 trace to STDOUT that can be redirected to a file which is
732 compatible with libcap's capture file format. This can then
733 be loaded in Wireshark and similar tools to inspect bus
736 * The busctl tool now understands a new "tree" verb that shows
737 the object trees of a specific service on the bus, or of all
740 * The busctl tool now understands a new "introspect" verb that
741 shows all interfaces and members of objects on the bus,
742 including their signature and values. This is particularly
743 useful to get more information about bus objects shown by
744 the new "busctl tree" command.
746 * The busctl tool now understands new verbs "call",
747 "set-property" and "get-property" for invoking bus method
748 calls, setting and getting bus object properties in a
751 * busctl gained a new --augment-creds= argument that controls
752 whether the tool shall augment credential information it
753 gets from the bus with data from /proc, in a possibly
756 * nspawn's --link-journal= switch gained two new values
757 "try-guest" and "try-host" that work like "guest" and
758 "host", but do not fail if the host has no persistent
759 journalling enabled. -j is now equivalent to
760 --link-journal=try-guest.
762 * macvlan network devices created by nspawn will now have
763 stable MAC addresses.
765 * A new SmackProcessLabel= unit setting has been added, which
766 controls the SMACK security label processes forked off by
767 the respective unit shall use.
769 * If compiled with --enable-xkbcommon, systemd-localed will
770 verify x11 keymap settings by compiling the given keymap. It
771 will spew out warnings if the compilation fails. This
772 requires libxkbcommon to be installed.
774 * When a coredump is collected a larger number of metadata
775 fields is now collected and included in the journal records
776 created for it. More specifically control group membership,
777 environment variables, memory maps, working directory,
778 chroot directory, /proc/$PID/status, and a list of open file
779 descriptors is now stored in the log entry.
781 * The udev hwdb now contains DPI information for mice. For
784 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
786 * All systemd programs that read standalone configuration
787 files in /etc now also support a corresponding series of
788 .conf.d configuration directories in /etc/, /run/,
789 /usr/local/lib/, /usr/lib/, and (if configured with
790 --enable-split-usr) /lib/. In particular, the following
791 configuration files now have corresponding configuration
792 directories: system.conf user.conf, logind.conf,
793 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
794 resolved.conf, timesyncd.conf, journal-remote.conf, and
795 journal-upload.conf. Note that distributions should use the
796 configuration directories in /usr/lib/; the directories in
797 /etc/ are reserved for the system administrator.
799 * systemd-rfkill will no longer take the rfkill device name
800 into account when storing rfkill state on disk, as the name
801 might be dynamically assigned and not stable. Instead, the
802 ID_PATH udev variable combined with the rfkill type (wlan,
803 bluetooth, ...) is used.
805 * A new service systemd-machine-id-commit.service has been
806 added. When used on systems where /etc is read-only during
807 boot, and /etc/machine-id is not initialized (but an empty
808 file), this service will copy the temporary machine ID
809 created as replacement into /etc after the system is fully
810 booted up. This is useful for systems that are freshly
811 installed with a non-initialized machine ID, but should get
812 a fixed machine ID for subsequent boots.
814 * networkd's .netdev files now provide a large set of
815 configuration parameters for VXLAN devices. Similar, the
816 bridge port cost parameter is now configurable in .network
817 files. There's also new support for configuring IP source
818 routing. networkd .link files gained support for a new
819 OriginalName= match that is useful to match against the
820 original interface name the kernel assigned. .network files
821 may include MTU= and MACAddress= fields for altering the MTU
822 and MAC address while being connected to a specific network
825 * The LUKS logic gained supported for configuring
826 UUID-specific key files. There's also new support for naming
827 LUKS device from the kernel command line, using the new
830 * Timer units may now be transiently created via the bus API
831 (this was previously already available for scope and service
832 units). In addition it is now possible to create multiple
833 transient units at the same time with a single bus call. The
834 "systemd-run" tool has been updated to make use of this for
835 running commands on a specified time, in at(1)-style.
837 * tmpfiles gained support for "t" lines, for assigning
838 extended attributes to files. Among other uses this may be
839 used to assign SMACK labels to files.
841 Contributions from: Alin Rauta, Alison Chaiken, Andrej
842 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
843 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
844 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
845 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
846 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
847 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
848 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
849 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
850 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
851 Michael Biebl, Michael Chapman, Michael Marineau, Michal
852 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
853 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
854 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
855 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
856 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
857 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
858 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
860 -- Berlin, 2014-12-10
864 * journalctl gained the new options -t/--identifier= to match
865 on the syslog identifier (aka "tag"), as well as --utc to
866 show log timestamps in the UTC timezone. journalctl now also
867 accepts -n/--lines=all to disable line capping in a pager.
869 * journalctl gained a new switch, --flush, that synchronously
870 flushes logs from /run/log/journal to /var/log/journal if
871 persistent storage is enabled. systemd-journal-flush.service
872 now waits until the operation is complete.
874 * Services can notify the manager before they start a reload
875 (by sending RELOADING=1) or shutdown (by sending
876 STOPPING=1). This allows the manager to track and show the
877 internal state of daemons and closes a race condition when
878 the process is still running but has closed its D-Bus
881 * Services with Type=oneshot do not have to have any ExecStart
884 * User units are now loaded also from
885 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
886 /run/systemd/user directory that was already previously
887 supported, but is under the control of the user.
889 * Job timeouts (i.e. time-outs on the time a job that is
890 queued stays in the run queue) can now optionally result in
891 immediate reboot or power-off actions (JobTimeoutAction= and
892 JobTimeoutRebootArgument=). This is useful on ".target"
893 units, to limit the maximum time a target remains
894 undispatched in the run queue, and to trigger an emergency
895 operation in such a case. This is now used by default to
896 turn off the system if boot-up (as defined by everything in
897 basic.target) hangs and does not complete for at least
898 15min. Also, if power-off or reboot hang for at least 30min
899 an immediate power-off/reboot operation is triggered. This
900 functionality is particularly useful to increase reliability
901 on embedded devices, but also on laptops which might
902 accidentally get powered on when carried in a backpack and
903 whose boot stays stuck in a hard disk encryption passphrase
906 * systemd-logind can be configured to also handle lid switch
907 events even when the machine is docked or multiple displays
908 are attached (HandleLidSwitchDocked= option).
910 * A helper binary and a service have been added which can be
911 used to resume from hibernation in the initramfs. A
912 generator will parse the resume= option on the kernel
913 command line to trigger resume.
915 * A user console daemon systemd-consoled has been
916 added. Currently, it is a preview, and will so far open a
917 single terminal on each session of the user marked as
918 Desktop=systemd-console.
920 * Route metrics can be specified for DHCP routes added by
923 * The SELinux context of socket-activated services can be set
924 from the information provided by the networking stack
925 (SELinuxContextFromNet= option).
927 * Userspace firmware loading support has been removed and
928 the minimum supported kernel version is thus bumped to 3.7.
930 * Timeout for udev workers has been increased from 1 to 3
931 minutes, but a warning will be printed after 1 minute to
932 help diagnose kernel modules that take a long time to load.
934 * Udev rules can now remove tags on devices with TAG-="foobar".
936 * systemd's readahead implementation has been removed. In many
937 circumstances it didn't give expected benefits even for
938 rotational disk drives and was becoming less relevant in the
939 age of SSDs. As none of the developers has been using
940 rotating media anymore, and nobody stepped up to actively
941 maintain this component of systemd it has now been removed.
943 * Swap units can use Options= to specify discard options.
944 Discard options specified for swaps in /etc/fstab are now
947 * Docker containers are now detected as a separate type of
950 * The Password Agent protocol gained support for queries where
951 the user input is shown, useful e.g. for user names.
952 systemd-ask-password gained a new --echo option to turn that
955 * The default sysctl.d/ snippets will now set:
957 net.core.default_qdisc = fq_codel
959 This selects Fair Queuing Controlled Delay as the default
960 queuing discipline for network interfaces. fq_codel helps
961 fight the network bufferbloat problem. It is believed to be
962 a good default with no tuning required for most workloads.
963 Downstream distributions may override this choice. On 10Gbit
964 servers that do not do forwarding, "fq" may perform better.
965 Systems without a good clocksource should use "pfifo_fast".
967 * If kdbus is enabled during build a new option BusPolicy= is
968 available for service units, that allows locking all service
969 processes into a stricter bus policy, in order to limit
970 access to various bus services, or even hide most of them
971 from the service's view entirely.
973 * networkctl will now show the .network and .link file
974 networkd has applied to a specific interface.
976 * sd-login gained a new API call sd_session_get_desktop() to
977 query which desktop environment has been selected for a
980 * UNIX utmp support is now compile-time optional to support
983 * systemctl gained two new commands "add-wants" and
984 "add-requires" for pulling in units from specific targets
987 * If the word "rescue" is specified on the kernel command line
988 the system will now boot into rescue mode (aka
989 rescue.target), which was previously available only by
990 specifying "1" or "systemd.unit=rescue.target" on the kernel
991 command line. This new kernel command line option nicely
992 mirrors the already existing "emergency" kernel command line
995 * New kernel command line options mount.usr=, mount.usrflags=,
996 mount.usrfstype= have been added that match root=, rootflags=,
997 rootfstype= but allow mounting a specific file system to
1000 * The $NOTIFY_SOCKET is now also passed to control processes of
1001 services, not only the main process.
1003 * This version reenables support for fsck's -l switch. This
1004 means at least version v2.25 of util-linux is required for
1005 operation, otherwise dead-locks on device nodes may
1006 occur. Again: you need to update util-linux to at least
1007 v2.25 when updating systemd to v217.
1009 * The "multi-seat-x" tool has been removed from systemd, as
1010 its functionality has been integrated into X servers 1.16,
1011 and the tool is hence redundant. It is recommended to update
1012 display managers invoking this tool to simply invoke X
1013 directly from now on, again.
1015 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
1016 message flag has been added for all of systemd's PolicyKit
1017 authenticated method calls has been added. In particular
1018 this now allows optional interactive authorization via
1019 PolicyKit for many of PID1's privileged operations such as
1020 unit file enabling and disabling.
1022 * "udevadm hwdb --update" learnt a new switch "--usr" for
1023 placing the rebuilt hardware database in /usr instead of
1024 /etc. When used only hardware database entries stored in
1025 /usr will be used, and any user database entries in /etc are
1026 ignored. This functionality is useful for vendors to ship a
1027 pre-built database on systems where local configuration is
1028 unnecessary or unlikely.
1030 * Calendar time specifications in .timer units now also
1031 understand the strings "semi-annually", "quarterly" and
1032 "minutely" as shortcuts (in addition to the preexisting
1033 "anually", "hourly", ...).
1035 * systemd-tmpfiles will now correctly create files in /dev
1036 at boot which are marked for creation only at boot. It is
1037 recommended to always create static device nodes with 'c!'
1038 and 'b!', so that they are created only at boot and not
1039 overwritten at runtime.
1041 * When the watchdog logic is used for a service (WatchdogSec=)
1042 and the watchdog timeout is hit the service will now be
1043 terminated with SIGABRT (instead of just SIGTERM), in order
1044 to make sure a proper coredump and backtrace is
1045 generated. This ensures that hanging services will result in
1046 similar coredump/backtrace behaviour as services that hit a
1049 Contributions from: Andreas Henriksson, Andrei Borzenkov,
1050 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
1051 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
1052 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
1053 Herrmann, David Sommerseth, David Strauss, Emil Renner
1054 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
1055 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
1056 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
1057 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
1058 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
1059 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
1060 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
1061 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
1062 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
1063 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
1064 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
1065 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
1066 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
1067 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
1068 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
1069 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
1072 -- Berlin, 2014-10-28
1076 * timedated no longer reads NTP implementation unit names from
1077 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
1078 implementations should add a
1080 Conflicts=systemd-timesyncd.service
1082 to their unit files to take over and replace systemd's NTP
1083 default functionality.
1085 * systemd-sysusers gained a new line type "r" for configuring
1086 which UID/GID ranges to allocate system users/groups
1087 from. Lines of type "u" may now add an additional column
1088 that specifies the home directory for the system user to be
1089 created. Also, systemd-sysusers may now optionally read user
1090 information from STDIN instead of a file. This is useful for
1091 invoking it from RPM preinst scriptlets that need to create
1092 users before the first RPM file is installed since these
1093 files might need to be owned by them. A new
1094 %sysusers_create_inline RPM macro has been introduced to do
1095 just that. systemd-sysusers now updates the shadow files as
1096 well as the user/group databases, which should enhance
1097 compatibility with certain tools like grpck.
1099 * A number of bus APIs of PID 1 now optionally consult
1100 PolicyKit to permit access for otherwise unprivileged
1101 clients under certain conditions. Note that this currently
1102 doesn't support interactive authentication yet, but this is
1103 expected to be added eventually, too.
1105 * /etc/machine-info now has new fields for configuring the
1106 deployment environment of the machine, as well as the
1107 location of the machine. hostnamectl has been updated with
1108 new command to update these fields.
1110 * systemd-timesyncd has been updated to automatically acquire
1111 NTP server information from systemd-networkd, which might
1112 have been discovered via DHCP.
1114 * systemd-resolved now includes a caching DNS stub resolver
1115 and a complete LLMNR name resolution implementation. A new
1116 NSS module "nss-resolve" has been added which can be used
1117 instead of glibc's own "nss-dns" to resolve hostnames via
1118 systemd-resolved. Hostnames, addresses and arbitrary RRs may
1119 be resolved via systemd-resolved D-Bus APIs. In contrast to
1120 the glibc internal resolver systemd-resolved is aware of
1121 multi-homed system, and keeps DNS server and caches separate
1122 and per-interface. Queries are sent simultaneously on all
1123 interfaces that have DNS servers configured, in order to
1124 properly handle VPNs and local LANs which might resolve
1125 separate sets of domain names. systemd-resolved may acquire
1126 DNS server information from systemd-networkd automatically,
1127 which in turn might have discovered them via DHCP. A tool
1128 "systemd-resolve-host" has been added that may be used to
1129 query the DNS logic in resolved. systemd-resolved implements
1130 IDNA and automatically uses IDNA or UTF-8 encoding depending
1131 on whether classic DNS or LLMNR is used as transport. In the
1132 next releases we intend to add a DNSSEC and mDNS/DNS-SD
1133 implementation to systemd-resolved.
1135 * A new NSS module nss-mymachines has been added, that
1136 automatically resolves the names of all local registered
1137 containers to their respective IP addresses.
1139 * A new client tool "networkctl" for systemd-networkd has been
1140 added. It currently is entirely passive and will query
1141 networking configuration from udev, rtnetlink and networkd,
1142 and present it to the user in a very friendly
1143 way. Eventually, we hope to extend it to become a full
1144 control utility for networkd.
1146 * .socket units gained a new DeferAcceptSec= setting that
1147 controls the kernels' TCP_DEFER_ACCEPT sockopt for
1148 TCP. Similar, support for controlling TCP keep-alive
1149 settings has been added (KeepAliveTimeSec=,
1150 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
1151 turning off Nagle's algorithm on TCP has been added
1154 * logind learned a new session type "web", for use in projects
1155 like Cockpit which register web clients as PAM sessions.
1157 * timer units with at least one OnCalendar= setting will now
1158 be started only after timer-sync.target has been
1159 reached. This way they will not elapse before the system
1160 clock has been corrected by a local NTP client or
1161 similar. This is particular useful on RTC-less embedded
1162 machines, that come up with an invalid system clock.
1164 * systemd-nspawn's --network-veth= switch should now result in
1165 stable MAC addresses for both the outer and the inner side
1168 * systemd-nspawn gained a new --volatile= switch for running
1169 container instances with /etc or /var unpopulated.
1171 * The kdbus client code has been updated to use the new Linux
1172 3.17 memfd subsystem instead of the old kdbus-specific one.
1174 * systemd-networkd's DHCP client and server now support
1175 FORCERENEW. There are also new configuration options to
1176 configure the vendor client identifier and broadcast mode
1179 * systemd will no longer inform the kernel about the current
1180 timezone, as this is necessarily incorrect and racy as the
1181 kernel has no understanding of DST and similar
1182 concepts. This hence means FAT timestamps will be always
1183 considered UTC, similar to what Android is already
1184 doing. Also, when the RTC is configured to the local time
1185 (rather than UTC) systemd will never synchronize back to it,
1186 as this might confuse Windows at a later boot.
1188 * systemd-analyze gained a new command "verify" for offline
1189 validation of unit files.
1191 * systemd-networkd gained support for a couple of additional
1192 settings for bonding networking setups. Also, the metric for
1193 statically configured routes may now be configured. For
1194 network interfaces where this is appropriate the peer IP
1195 address may now be configured.
1197 * systemd-networkd's DHCP client will no longer request
1198 broadcasting by default, as this tripped up some networks.
1199 For hardware where broadcast is required the feature should
1200 be switched back on using RequestBroadcast=yes.
1202 * systemd-networkd will now set up IPv4LL addresses (when
1203 enabled) even if DHCP is configured successfully.
1205 * udev will now default to respect network device names given
1206 by the kernel when the kernel indicates that these are
1207 predictable. This behavior can be tweaked by changing
1208 NamePolicy= in the relevant .link file.
1210 * A new library systemd-terminal has been added that
1211 implements full TTY stream parsing and rendering. This
1212 library is supposed to be used later on for implementing a
1213 full userspace VT subsystem, replacing the current kernel
1216 * A new tool systemd-journal-upload has been added to push
1217 journal data to a remote system running
1218 systemd-journal-remote.
1220 * journald will no longer forward all local data to another
1221 running syslog daemon. This change has been made because
1222 rsyslog (which appears to be the most commonly used syslog
1223 implementation these days) no longer makes use of this, and
1224 instead pulls the data out of the journal on its own. Since
1225 forwarding the messages to a non-existent syslog server is
1226 more expensive than we assumed we have now turned this
1227 off. If you run a syslog server that is not a recent rsyslog
1228 version, you have to turn this option on again
1229 (ForwardToSyslog= in journald.conf).
1231 * journald now optionally supports the LZ4 compressor for
1232 larger journal fields. This compressor should perform much
1233 better than XZ which was the previous default.
1235 * machinectl now shows the IP addresses of local containers,
1236 if it knows them, plus the interface name of the container.
1238 * A new tool "systemd-escape" has been added that makes it
1239 easy to escape strings to build unit names and similar.
1241 * sd_notify() messages may now include a new ERRNO= field
1242 which is parsed and collected by systemd and shown among the
1243 "systemctl status" output for a service.
1245 * A new component "systemd-firstboot" has been added that
1246 queries the most basic systemd information (timezone,
1247 hostname, root password) interactively on first
1248 boot. Alternatively it may also be used to provision these
1249 things offline on OS images installed into directories.
1251 * The default sysctl.d/ snippets will now set
1253 net.ipv4.conf.default.promote_secondaries=1
1255 This has the benefit of no flushing secondary IP addresses
1256 when primary addresses are removed.
1258 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
1259 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
1260 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
1261 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
1262 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
1263 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
1264 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
1265 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
1266 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
1267 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
1268 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
1269 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
1270 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
1271 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
1272 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
1274 -- Berlin, 2014-08-19
1278 * A new tool systemd-sysusers has been added. This tool
1279 creates system users and groups in /etc/passwd and
1280 /etc/group, based on static declarative system user/group
1281 definitions in /usr/lib/sysusers.d/. This is useful to
1282 enable factory resets and volatile systems that boot up with
1283 an empty /etc directory, and thus need system users and
1284 groups created during early boot. systemd now also ships
1285 with two default sysusers.d/ files for the most basic
1286 users and groups systemd and the core operating system
1289 * A new tmpfiles snippet has been added that rebuilds the
1290 essential files in /etc on boot, should they be missing.
1292 * A directive for ensuring automatic clean-up of
1293 /var/cache/man/ has been removed from the default
1294 configuration. This line should now be shipped by the man
1295 implementation. The necessary change has been made to the
1296 man-db implementation. Note that you need to update your man
1297 implementation to one that ships this line, otherwise no
1298 automatic clean-up of /var/cache/man will take place.
1300 * A new condition ConditionNeedsUpdate= has been added that
1301 may conditionalize services to only run when /etc or /var
1302 are "older" than the vendor operating system resources in
1303 /usr. This is useful for reconstructing or updating /etc
1304 after an offline update of /usr or a factory reset, on the
1305 next reboot. Services that want to run once after such an
1306 update or reset should use this condition and order
1307 themselves before the new systemd-update-done.service, which
1308 will mark the two directories as fully updated. A number of
1309 service files have been added making use of this, to rebuild
1310 the udev hardware database, the journald message catalog and
1311 dynamic loader cache (ldconfig). The systemd-sysusers tool
1312 described above also makes use of this now. With this in
1313 place it is now possible to start up a minimal operating
1314 system with /etc empty cleanly. For more information on the
1315 concepts involved see this recent blog story:
1317 http://0pointer.de/blog/projects/stateless.html
1319 * A new system group "input" has been introduced, and all
1320 input device nodes get this group assigned. This is useful
1321 for system-level software to get access to input devices. It
1322 complements what is already done for "audio" and "video".
1324 * systemd-networkd learnt minimal DHCPv4 server support in
1325 addition to the existing DHCPv4 client support. It also
1326 learnt DHCPv6 client and IPv6 Router Solicitation client
1327 support. The DHCPv4 client gained support for static routes
1328 passed in from the server. Note that the [DHCPv4] section
1329 known in older systemd-networkd versions has been renamed to
1330 [DHCP] and is now also used by the DHCPv6 client. Existing
1331 .network files using settings of this section should be
1332 updated, though compatibility is maintained. Optionally, the
1333 client hostname may now be sent to the DHCP server.
1335 * networkd gained support for vxlan virtual networks as well
1336 as tun/tap and dummy devices.
1338 * networkd gained support for automatic allocation of address
1339 ranges for interfaces from a system-wide pool of
1340 addresses. This is useful for dynamically managing a large
1341 number of interfaces with a single network configuration
1342 file. In particular this is useful to easily assign
1343 appropriate IP addresses to the veth links of a large number
1344 of nspawn instances.
1346 * RPM macros for processing sysusers, sysctl and binfmt
1347 drop-in snippets at package installation time have been
1350 * The /etc/os-release file should now be placed in
1351 /usr/lib/os-release. The old location is automatically
1352 created as symlink. /usr/lib is the more appropriate
1353 location of this file, since it shall actually describe the
1354 vendor operating system shipped in /usr, and not the
1355 configuration stored in /etc.
1357 * .mount units gained a new boolean SloppyOptions= setting
1358 that maps to mount(8)'s -s option which enables permissive
1359 parsing of unknown mount options.
1361 * tmpfiles learnt a new "L+" directive which creates a symlink
1362 but (unlike "L") deletes a pre-existing file first, should
1363 it already exist and not already be the correct
1364 symlink. Similar, "b+", "c+" and "p+" directives have been
1365 added as well, which create block and character devices, as
1366 well as fifos in the filesystem, possibly removing any
1367 pre-existing files of different types.
1369 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
1370 'argument' field (which so far specified the source to
1371 symlink/copy the files from) is now optional. If omitted the
1372 same file os copied from /usr/share/factory/ suffixed by the
1373 full destination path. This is useful for populating /etc
1374 with essential files, by copying them from vendor defaults
1375 shipped in /usr/share/factory/etc.
1377 * A new command "systemctl preset-all" has been added that
1378 applies the service preset settings to all installed unit
1379 files. A new switch --preset-mode= has been added that
1380 controls whether only enable or only disable operations
1383 * A new command "systemctl is-system-running" has been added
1384 that allows checking the overall state of the system, for
1385 example whether it is fully up and running.
1387 * When the system boots up with an empty /etc, the equivalent
1388 to "systemctl preset-all" is executed during early boot, to
1389 make sure all default services are enabled after a factory
1392 * systemd now contains a minimal preset file that enables the
1393 most basic services systemd ships by default.
1395 * Unit files' [Install] section gained a new DefaultInstance=
1396 field for defining the default instance to create if a
1397 template unit is enabled with no instance specified.
1399 * A new passive target cryptsetup-pre.target has been added
1400 that may be used by services that need to make they run and
1401 finish before the first LUKS cryptographic device is set up.
1403 * The /dev/loop-control and /dev/btrfs-control device nodes
1404 are now owned by the "disk" group by default, opening up
1405 access to this group.
1407 * systemd-coredump will now automatically generate a
1408 stack trace of all core dumps taking place on the system,
1409 based on elfutils' libdw library. This stack trace is logged
1412 * systemd-coredump may now optionally store coredumps directly
1413 on disk (in /var/lib/systemd/coredump, possibly compressed),
1414 instead of storing them unconditionally in the journal. This
1415 mode is the new default. A new configuration file
1416 /etc/systemd/coredump.conf has been added to configure this
1417 and other parameters of systemd-coredump.
1419 * coredumpctl gained a new "info" verb to show details about a
1420 specific coredump. A new switch "-1" has also been added
1421 that makes sure to only show information about the most
1422 recent entry instead of all entries. Also, as the tool is
1423 generally useful now the "systemd-" prefix of the binary
1424 name has been removed. Distributions that want to maintain
1425 compatibility with the old name should add a symlink from
1426 the old name to the new name.
1428 * journald's SplitMode= now defaults to "uid". This makes sure
1429 that unprivileged users can access their own coredumps with
1430 coredumpctl without restrictions.
1432 * New kernel command line options "systemd.wants=" (for
1433 pulling an additional unit during boot), "systemd.mask="
1434 (for masking a specific unit for the boot), and
1435 "systemd.debug-shell" (for enabling the debug shell on tty9)
1436 have been added. This is implemented in the new generator
1437 "systemd-debug-generator".
1439 * systemd-nspawn will now by default filter a couple of
1440 syscalls for containers, among them those required for
1441 kernel module loading, direct x86 IO port access, swap
1442 management, and kexec. Most importantly though
1443 open_by_handle_at() is now prohibited for containers,
1444 closing a hole similar to a recently discussed vulnerability
1445 in docker regarding access to files on file hierarchies the
1446 container should normally not have access to. Note that for
1447 nspawn we generally make no security claims anyway (and
1448 this is explicitly documented in the man page), so this is
1449 just a fix for one of the most obvious problems.
1451 * A new man page file-hierarchy(7) has been added that
1452 contains a minimized, modernized version of the file system
1453 layout systemd expects, similar in style to the FHS
1454 specification or hier(5). A new tool systemd-path(1) has
1455 been added to query many of these paths for the local
1458 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
1459 longer done. Since the directory now has a per-user size
1460 limit, and is cleaned on logout this appears unnecessary,
1461 in particular since this now brings the lifecycle of this
1462 directory closer in line with how IPC objects are handled.
1464 * systemd.pc now exports a number of additional directories,
1465 including $libdir (which is useful to identify the library
1466 path for the primary architecture of the system), and a
1467 couple of drop-in directories.
1469 * udev's predictable network interface names now use the dev_port
1470 sysfs attribute, introduced in linux 3.15 instead of dev_id to
1471 distinguish between ports of the same PCI function. dev_id should
1472 only be used for ports using the same HW address, hence the need
1475 * machined has been updated to export the OS version of a
1476 container (read from /etc/os-release and
1477 /usr/lib/os-release) on the bus. This is now shown in
1478 "machinectl status" for a machine.
1480 * A new service setting RestartForceExitStatus= has been
1481 added. If configured to a set of exit signals or process
1482 return values, the service will be restarted when the main
1483 daemon process exits with any of them, regardless of the
1486 * systemctl's -H switch for connecting to remote systemd
1487 machines has been extended so that it may be used to
1488 directly connect to a specific container on the
1489 host. "systemctl -H root@foobar:waldi" will now connect as
1490 user "root" to host "foobar", and then proceed directly to
1491 the container named "waldi". Note that currently you have to
1492 authenticate as user "root" for this to work, as entering
1493 containers is a privileged operation.
1495 Contributions from: Andreas Henriksson, Benjamin Steinwender,
1496 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
1497 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
1498 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
1499 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
1500 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
1501 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
1502 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
1503 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
1504 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
1505 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
1506 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
1508 -- Berlin, 2014-07-03
1512 * As an experimental feature, udev now tries to lock the
1513 disk device node (flock(LOCK_SH|LOCK_NB)) while it
1514 executes events for the disk or any of its partitions.
1515 Applications like partitioning programs can lock the
1516 disk device node (flock(LOCK_EX)) and claim temporary
1517 device ownership that way; udev will entirely skip all event
1518 handling for this disk and its partitions. If the disk
1519 was opened for writing, the close will trigger a partition
1520 table rescan in udev's "watch" facility, and if needed
1521 synthesize "change" events for the disk and all its partitions.
1522 This is now unconditionally enabled, and if it turns out to
1523 cause major problems, we might turn it on only for specific
1524 devices, or might need to disable it entirely. Device Mapper
1525 devices are excluded from this logic.
1527 * We temporarily dropped the "-l" switch for fsck invocations,
1528 since they collide with the flock() logic above. util-linux
1529 upstream has been changed already to avoid this conflict,
1530 and we will readd "-l" as soon as util-linux with this
1531 change has been released.
1533 * The dependency on libattr has been removed. Since a long
1534 time, the extended attribute calls have moved to glibc, and
1535 libattr is thus unnecessary.
1537 * Virtualization detection works without priviliges now. This
1538 means the systemd-detect-virt binary no longer requires
1539 CAP_SYS_PTRACE file capabilities, and our daemons can run
1540 with fewer privileges.
1542 * systemd-networkd now runs under its own "systemd-network"
1543 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
1544 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
1545 loses the ability to write to files owned by root this way.
1547 * Similar, systemd-resolved now runs under its own
1548 "systemd-resolve" user with no capabilities remaining.
1550 * Similar, systemd-bus-proxyd now runs under its own
1551 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
1553 * systemd-networkd gained support for setting up "veth"
1554 virtual ethernet devices for container connectivity, as well
1555 as GRE and VTI tunnels.
1557 * systemd-networkd will no longer automatically attempt to
1558 manually load kernel modules necessary for certain tunnel
1559 transports. Instead, it is assumed the kernel loads them
1560 automatically when required. This only works correctly on
1561 very new kernels. On older kernels, please consider adding
1562 the kernel modules to /etc/modules-load.d/ as a work-around.
1564 * The resolv.conf file systemd-resolved generates has been
1565 moved to /run/systemd/resolve/. If you have a symlink from
1566 /etc/resolv.conf, it might be necessary to correct it.
1568 * Two new service settings, ProtectHome= and ProtectSystem=,
1569 have been added. When enabled, they will make the user data
1570 (such as /home) inaccessible or read-only and the system
1571 (such as /usr) read-only, for specific services. This allows
1572 very light-weight per-service sandboxing to avoid
1573 modifications of user data or system files from
1574 services. These two new switches have been enabled for all
1575 of systemd's long-running services, where appropriate.
1577 * Socket units gained new SocketUser= and SocketGroup=
1578 settings to set the owner user and group of AF_UNIX sockets
1579 and FIFOs in the file system.
1581 * Socket units gained a new RemoveOnStop= setting. If enabled,
1582 all FIFOS and sockets in the file system will be removed
1583 when the specific socket unit is stopped.
1585 * Socket units gained a new Symlinks= setting. It takes a list
1586 of symlinks to create to file system sockets or FIFOs
1587 created by the specific Unix sockets. This is useful to
1588 manage symlinks to socket nodes with the same life-cycle as
1591 * The /dev/log socket and /dev/initctl FIFO have been moved to
1592 /run, and have been replaced by symlinks. This allows
1593 connecting to these facilities even if PrivateDevices=yes is
1594 used for a service (which makes /dev/log itself unavailable,
1595 but /run is left). This also has the benefit of ensuring
1596 that /dev only contains device nodes, directories and
1597 symlinks, and nothing else.
1599 * sd-daemon gained two new calls sd_pid_notify() and
1600 sd_pid_notifyf(). They are similar to sd_notify() and
1601 sd_notifyf(), but allow overriding of the source PID of
1602 notification messages if permissions permit this. This is
1603 useful to send notify messages on behalf of a different
1604 process (for example, the parent process). The
1605 systemd-notify tool has been updated to make use of this
1606 when sending messages (so that notification messages now
1607 originate from the shell script invoking systemd-notify and
1608 not the systemd-notify process itself. This should minimize
1609 a race where systemd fails to associate notification
1610 messages to services when the originating process already
1613 * A new "on-abnormal" setting for Restart= has been added. If
1614 set, it will result in automatic restarts on all "abnormal"
1615 reasons for a process to exit, which includes unclean
1616 signals, core dumps, timeouts and watchdog timeouts, but
1617 does not include clean and unclean exit codes or clean
1618 signals. Restart=on-abnormal is an alternative for
1619 Restart=on-failure for services that shall be able to
1620 terminate and avoid restarts on certain errors, by
1621 indicating so with an unclean exit code. Restart=on-failure
1622 or Restart=on-abnormal is now the recommended setting for
1623 all long-running services.
1625 * If the InaccessibleDirectories= service setting points to a
1626 mount point (or if there are any submounts contained within
1627 it), it is now attempted to completely unmount it, to make
1628 the file systems truly unavailable for the respective
1631 * The ReadOnlyDirectories= service setting and
1632 systemd-nspawn's --read-only parameter are now recursively
1633 applied to all submounts, too.
1635 * Mount units may now be created transiently via the bus APIs.
1637 * The support for SysV and LSB init scripts has been removed
1638 from the systemd daemon itself. Instead, it is now
1639 implemented as a generator that creates native systemd units
1640 from these scripts when needed. This enables us to remove a
1641 substantial amount of legacy code from PID 1, following the
1642 fact that many distributions only ship a very small number
1643 of LSB/SysV init scripts nowadays.
1645 * Privileged Xen (dom0) domains are not considered
1646 virtualization anymore by the virtualization detection
1647 logic. After all, they generally have unrestricted access to
1648 the hardware and usually are used to manage the unprivileged
1651 * systemd-tmpfiles gained a new "C" line type, for copying
1652 files or entire directories.
1654 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
1655 lines. So far, they have been non-globbing versions of the
1656 latter, and have thus been redundant. In future, it is
1657 recommended to only use "z". "m" has hence been removed
1658 from the documentation, even though it stays supported.
1660 * A tmpfiles snippet to recreate the most basic structure in
1661 /var has been added. This is enough to create the /var/run →
1662 /run symlink and create a couple of structural
1663 directories. This allows systems to boot up with an empty or
1664 volatile /var. Of course, while with this change, the core OS
1665 now is capable with dealing with a volatile /var, not all
1666 user services are ready for it. However, we hope that sooner
1667 or later, many service daemons will be changed upstream so
1668 that they are able to automatically create their necessary
1669 directories in /var at boot, should they be missing. This is
1670 the first step to allow state-less systems that only require
1671 the vendor image for /usr to boot.
1673 * systemd-nspawn has gained a new --tmpfs= switch to mount an
1674 empty tmpfs instance to a specific directory. This is
1675 particularly useful for making use of the automatic
1676 reconstruction of /var (see above), by passing --tmpfs=/var.
1678 * Access modes specified in tmpfiles snippets may now be
1679 prefixed with "~", which indicates that they shall be masked
1680 by whether the existing file or directory is currently
1681 writable, readable or executable at all. Also, if specified,
1682 the sgid/suid/sticky bits will be masked for all
1685 * A new passive target unit "network-pre.target" has been
1686 added which is useful for services that shall run before any
1687 network is configured, for example firewall scripts.
1689 * The "floppy" group that previously owned the /dev/fd*
1690 devices is no longer used. The "disk" group is now used
1691 instead. Distributions should probably deprecate usage of
1694 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
1695 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
1696 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
1697 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
1698 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
1699 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
1702 -- Berlin, 2014-06-11
1706 * A new "systemd-timesyncd" daemon has been added for
1707 synchronizing the system clock across the network. It
1708 implements an SNTP client. In contrast to NTP
1709 implementations such as chrony or the NTP reference server,
1710 this only implements a client side, and does not bother with
1711 the full NTP complexity, focusing only on querying time from
1712 one remote server and synchronizing the local clock to
1713 it. Unless you intend to serve NTP to networked clients or
1714 want to connect to local hardware clocks, this simple NTP
1715 client should be more than appropriate for most
1716 installations. The daemon runs with minimal privileges, and
1717 has been hooked up with networkd to only operate when
1718 network connectivity is available. The daemon saves the
1719 current clock to disk every time a new NTP sync has been
1720 acquired, and uses this to possibly correct the system clock
1721 early at bootup, in order to accommodate for systems that
1722 lack an RTC such as the Raspberry Pi and embedded devices,
1723 and to make sure that time monotonically progresses on these
1724 systems, even if it is not always correct. To make use of
1725 this daemon, a new system user and group "systemd-timesync"
1726 needs to be created on installation of systemd.
1728 * The queue "seqnum" interface of libudev has been disabled, as
1729 it was generally incompatible with device namespacing as
1730 sequence numbers of devices go "missing" if the devices are
1731 part of a different namespace.
1733 * "systemctl list-timers" and "systemctl list-sockets" gained
1734 a --recursive switch for showing units of these types also
1735 for all local containers, similar in style to the already
1736 supported --recursive switch for "systemctl list-units".
1738 * A new RebootArgument= setting has been added for service
1739 units, which may be used to specify a kernel reboot argument
1740 to use when triggering reboots with StartLimitAction=.
1742 * A new FailureAction= setting has been added for service
1743 units which may be used to specify an operation to trigger
1744 when a service fails. This works similarly to
1745 StartLimitAction=, but unlike it, controls what is done
1746 immediately rather than only after several attempts to
1747 restart the service in question.
1749 * hostnamed got updated to also expose the kernel name,
1750 release, and version on the bus. This is useful for
1751 executing commands like hostnamectl with the -H switch.
1752 systemd-analyze makes use of this to properly display
1753 details when running non-locally.
1755 * The bootchart tool can now show cgroup information in the
1756 graphs it generates.
1758 * The CFS CPU quota cgroup attribute is now exposed for
1759 services. The new CPUQuota= switch has been added for this
1760 which takes a percentage value. Setting this will have the
1761 result that a service may never get more CPU time than the
1762 specified percentage, even if the machine is otherwise idle.
1764 * systemd-networkd learned IPIP and SIT tunnel support.
1766 * LSB init scripts exposing a dependency on $network will now
1767 get a dependency on network-online.target rather than simply
1768 network.target. This should bring LSB handling closer to
1769 what it was on SysV systems.
1771 * A new fsck.repair= kernel option has been added to control
1772 how fsck shall deal with unclean file systems at boot.
1774 * The (.ini) configuration file parser will now silently
1775 ignore sections whose name begins with "X-". This may be
1776 used to maintain application-specific extension sections in unit
1779 * machined gained a new API to query the IP addresses of
1780 registered containers. "machinectl status" has been updated
1781 to show these addresses in its output.
1783 * A new call sd_uid_get_display() has been added to the
1784 sd-login APIs for querying the "primary" session of a
1785 user. The "primary" session of the user is elected from the
1786 user's sessions and generally a graphical session is
1787 preferred over a text one.
1789 * A minimal systemd-resolved daemon has been added. It
1790 currently simply acts as a companion to systemd-networkd and
1791 manages resolv.conf based on per-interface DNS
1792 configuration, possibly supplied via DHCP. In the long run
1793 we hope to extend this into a local DNSSEC enabled DNS and
1796 * The systemd-networkd-wait-online tool is now enabled by
1797 default. It will delay network-online.target until a network
1798 connection has been configured. The tool primarily integrates
1799 with networkd, but will also make a best effort to make sense
1800 of network configuration performed in some other way.
1802 * Two new service options StartupCPUShares= and
1803 StartupBlockIOWeight= have been added that work similarly to
1804 CPUShares= and BlockIOWeight= however only apply during
1805 system startup. This is useful to prioritize certain services
1806 differently during bootup than during normal runtime.
1808 * hostnamed has been changed to prefer the statically
1809 configured hostname in /etc/hostname (unless set to
1810 'localhost' or empty) over any dynamic one supplied by
1811 dhcp. With this change, the rules for picking the hostname
1812 match more closely the rules of other configuration settings
1813 where the local administrator's configuration in /etc always
1814 overrides any other settings.
1816 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
1817 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
1818 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
1819 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
1820 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
1821 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
1822 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
1823 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
1824 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
1825 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
1826 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
1827 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
1828 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
1829 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
1830 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
1831 Lindskog, WaLyong Cho, Will Woods, Zbigniew
1834 -- Beijing, 2014-05-28
1838 * When restoring the screen brightness at boot, stay away from
1839 the darkest setting or from the lowest 5% of the available
1840 range, depending on which is the larger value of both. This
1841 should effectively protect the user from rebooting into a
1842 black screen, should the brightness have been set to minimum
1845 * sd-login gained a new sd_machine_get_class() call to
1846 determine the class ("vm" or "container") of a machine
1847 registered with machined.
1849 * sd-login gained new calls
1850 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
1851 to query the identity of the peer of a local AF_UNIX
1852 connection. They operate similarly to their sd_pid_get_xyz()
1855 * PID 1 will now maintain a system-wide system state engine
1856 with the states "starting", "running", "degraded",
1857 "maintenance", "stopping". These states are bound to system
1858 startup, normal runtime, runtime with at least one failed
1859 service, rescue/emergency mode and system shutdown. This
1860 state is shown in the "systemctl status" output when no unit
1861 name is passed. It is useful to determine system state, in
1862 particularly when doing so for many systems or containers at
1865 * A new command "list-machines" has been added to "systemctl"
1866 that lists all local OS containers and shows their system
1867 state (see above), if systemd runs inside of them.
1869 * systemctl gained a new "-r" switch to recursively enumerate
1870 units on all local containers, when used with the
1871 "list-unit" command (which is the default one that is
1872 executed when no parameters are specified).
1874 * The GPT automatic partition discovery logic will now honour
1875 two GPT partition flags: one may be set on a partition to
1876 cause it to be mounted read-only, and the other may be set
1877 on a partition to ignore it during automatic discovery.
1879 * Two new GPT type UUIDs have been added for automatic root
1880 partition discovery, for 32-bit and 64-bit ARM. This is not
1881 particularly useful for discovering the root directory on
1882 these architectures during bare-metal boots (since UEFI is
1883 not common there), but still very useful to allow booting of
1884 ARM disk images in nspawn with the -i option.
1886 * MAC addresses of interfaces created with nspawn's
1887 --network-interface= switch will now be generated from the
1888 machine name, and thus be stable between multiple invocations
1891 * logind will now automatically remove all IPC objects owned
1892 by a user if she or he fully logs out. This makes sure that
1893 users who are logged out cannot continue to consume IPC
1894 resources. This covers SysV memory, semaphores and message
1895 queues as well as POSIX shared memory and message
1896 queues. Traditionally, SysV and POSIX IPC had no life-cycle
1897 limits. With this functionality, that is corrected. This may
1898 be turned off by using the RemoveIPC= switch of logind.conf.
1900 * The systemd-machine-id-setup and tmpfiles tools gained a
1901 --root= switch to operate on a specific root directory,
1904 * journald can now forward logged messages to the TTYs of all
1905 logged in users ("wall"). This is the default for all
1906 emergency messages now.
1908 * A new tool systemd-journal-remote has been added to stream
1909 journal log messages across the network.
1911 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
1912 controller trees are mounted into it. Note that the
1913 directories mounted beneath it are not read-only. This is a
1914 security measure and is particularly useful because glibc
1915 actually includes a search logic to pick any tmpfs it can
1916 find to implement shm_open() if /dev/shm is not available
1917 (which it might very well be in namespaced setups).
1919 * machinectl gained a new "poweroff" command to cleanly power
1920 down a local OS container.
1922 * The PrivateDevices= unit file setting will now also drop the
1923 CAP_MKNOD capability from the capability bound set, and
1924 imply DevicePolicy=closed.
1926 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
1927 comprehensively on all long-running systemd services where
1928 this is appropriate.
1930 * systemd-udevd will now run in a disassociated mount
1931 namespace. To mount directories from udev rules, make sure to
1932 pull in mount units via SYSTEMD_WANTS properties.
1934 * The kdbus support gained support for uploading policy into
1935 the kernel. sd-bus gained support for creating "monitoring"
1936 connections that can eavesdrop into all bus communication
1937 for debugging purposes.
1939 * Timestamps may now be specified in seconds since the UNIX
1940 epoch Jan 1st, 1970 by specifying "@" followed by the value
1943 * Native tcpwrap support in systemd has been removed. tcpwrap
1944 is old code, not really maintained anymore and has serious
1945 shortcomings, and better options such as firewalls
1946 exist. For setups that require tcpwrap usage, please
1947 consider invoking your socket-activated service via tcpd,
1948 like on traditional inetd.
1950 * A new system.conf configuration option
1951 DefaultTimerAccuracySec= has been added that controls the
1952 default AccuracySec= setting of .timer units.
1954 * Timer units gained a new WakeSystem= switch. If enabled,
1955 timers configured this way will cause the system to resume
1956 from system suspend (if the system supports that, which most
1959 * Timer units gained a new Persistent= switch. If enabled,
1960 timers configured this way will save to disk when they have
1961 been last triggered. This information is then used on next
1962 reboot to possible execute overdue timer events, that
1963 could not take place because the system was powered off.
1964 This enables simple anacron-like behaviour for timer units.
1966 * systemctl's "list-timers" will now also list the time a
1967 timer unit was last triggered in addition to the next time
1968 it will be triggered.
1970 * systemd-networkd will now assign predictable IPv4LL
1971 addresses to its local interfaces.
1973 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
1974 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
1975 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
1976 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
1977 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
1978 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
1979 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
1980 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
1983 -- Berlin, 2014-03-25
1987 * A new unit file setting RestrictAddressFamilies= has been
1988 added to restrict which socket address families unit
1989 processes gain access to. This takes address family names
1990 like "AF_INET" or "AF_UNIX", and is useful to minimize the
1991 attack surface of services via exotic protocol stacks. This
1992 is built on seccomp system call filters.
1994 * Two new unit file settings RuntimeDirectory= and
1995 RuntimeDirectoryMode= have been added that may be used to
1996 manage a per-daemon runtime directories below /run. This is
1997 an alternative for setting up directory permissions with
1998 tmpfiles snippets, and has the advantage that the runtime
1999 directory's lifetime is bound to the daemon runtime and that
2000 the daemon starts up with an empty directory each time. This
2001 is particularly useful when writing services that drop
2002 privileges using the User= or Group= setting.
2004 * The DeviceAllow= unit setting now supports globbing for
2005 matching against device group names.
2007 * The systemd configuration file system.conf gained new
2008 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
2009 DefaultMemoryAccounting= to globally turn on/off accounting
2010 for specific resources (cgroups) for all units. These
2011 settings may still be overridden individually in each unit
2014 * systemd-gpt-auto-generator is now able to discover /srv and
2015 root partitions in addition to /home and swap partitions. It
2016 also supports LUKS-encrypted partitions now. With this in
2017 place, automatic discovery of partitions to mount following
2018 the Discoverable Partitions Specification
2019 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
2020 is now a lot more complete. This allows booting without
2021 /etc/fstab and without root= on the kernel command line on
2022 systems prepared appropriately.
2024 * systemd-nspawn gained a new --image= switch which allows
2025 booting up disk images and Linux installations on any block
2026 device that follow the Discoverable Partitions Specification
2027 (see above). This means that installations made with
2028 appropriately updated installers may now be started and
2029 deployed using container managers, completely
2030 unmodified. (We hope that libvirt-lxc will add support for
2031 this feature soon, too.)
2033 * systemd-nspawn gained a new --network-macvlan= setting to
2034 set up a private macvlan interface for the
2035 container. Similarly, systemd-networkd gained a new
2036 Kind=macvlan setting in .netdev files.
2038 * systemd-networkd now supports configuring local addresses
2041 * A new tool systemd-network-wait-online has been added to
2042 synchronously wait for network connectivity using
2045 * The sd-bus.h bus API gained a new sd_bus_track object for
2046 tracking the life-cycle of bus peers. Note that sd-bus.h is
2047 still not a public API though (unless you specify
2048 --enable-kdbus on the configure command line, which however
2049 voids your warranty and you get no API stability guarantee).
2051 * The $XDG_RUNTIME_DIR runtime directories for each user are
2052 now individual tmpfs instances, which has the benefit of
2053 introducing separate pools for each user, with individual
2054 size limits, and thus making sure that unprivileged clients
2055 can no longer negatively impact the system or other users by
2056 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
2057 RuntimeDirectorySize= has been introduced that allows
2058 controlling the default size limit for all users. It
2059 defaults to 10% of the available physical memory. This is no
2060 replacement for quotas on tmpfs though (which the kernel
2061 still does not support), as /dev/shm and /tmp are still
2062 shared resources used by both the system and unprivileged
2065 * logind will now automatically turn off automatic suspending
2066 on laptop lid close when more than one display is
2067 connected. This was previously expected to be implemented
2068 individually in desktop environments (such as GNOME),
2069 however has been added to logind now, in order to fix a
2070 boot-time race where a desktop environment might not have
2071 been started yet and thus not been able to take an inhibitor
2072 lock at the time where logind already suspends the system
2073 due to a closed lid.
2075 * logind will now wait at least 30s after each system
2076 suspend/resume cycle, and 3min after system boot before
2077 suspending the system due to a closed laptop lid. This
2078 should give USB docking stations and similar enough time to
2079 be probed and configured after system resume and boot in
2080 order to then act as suspend blocker.
2082 * systemd-run gained a new --property= setting which allows
2083 initialization of resource control properties (and others)
2084 for the created scope or service unit. Example: "systemd-run
2085 --property=BlockIOWeight=10 updatedb" may be used to run
2086 updatedb at a low block IO scheduling weight.
2088 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
2089 now also work in --scope mode.
2091 * When systemd is compiled with kdbus support, basic support
2092 for enforced policies is now in place. (Note that enabling
2093 kdbus still voids your warranty and no API compatibility
2096 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
2097 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
2098 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
2099 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
2100 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
2101 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
2102 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
2103 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
2104 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
2105 Zbigniew Jędrzejewski-Szmek
2107 -- Berlin, 2014-03-12
2111 * systemd will now relabel /dev after loading the SMACK policy
2112 according to SMACK rules.
2114 * A new unit file option AppArmorProfile= has been added to
2115 set the AppArmor profile for the processes of a unit.
2117 * A new condition check ConditionArchitecture= has been added
2118 to conditionalize units based on the system architecture, as
2119 reported by uname()'s "machine" field.
2121 * systemd-networkd now supports matching on the system
2122 virtualization, architecture, kernel command line, host name
2125 * logind is now a lot more aggressive when suspending the
2126 machine due to a closed laptop lid. Instead of acting only
2127 on the lid close action, it will continuously watch the lid
2128 status and act on it. This is useful for laptops where the
2129 power button is on the outside of the chassis so that it can
2130 be reached without opening the lid (such as the Lenovo
2131 Yoga). On those machines, logind will now immediately
2132 re-suspend the machine if the power button has been
2133 accidentally pressed while the laptop was suspended and in a
2134 backpack or similar.
2136 * logind will now watch SW_DOCK switches and inhibit reaction
2137 to the lid switch if it is pressed. This means that logind
2138 will not suspend the machine anymore if the lid is closed
2139 and the system is docked, if the laptop supports SW_DOCK
2140 notifications via the input layer. Note that ACPI docking
2141 stations do not generate this currently. Also note that this
2142 logic is usually not fully sufficient and Desktop
2143 Environments should take a lid switch inhibitor lock when an
2144 external display is connected, as systemd will not watch
2147 * nspawn will now make use of the devices cgroup controller by
2148 default, and only permit creation of and access to the usual
2149 API device nodes like /dev/null or /dev/random, as well as
2150 access to (but not creation of) the pty devices.
2152 * We will now ship a default .network file for
2153 systemd-networkd that automatically configures DHCP for
2154 network interfaces created by nspawn's --network-veth or
2155 --network-bridge= switches.
2157 * systemd will now understand the usual M, K, G, T suffixes
2158 according to SI conventions (i.e. to the base 1000) when
2159 referring to throughput and hardware metrics. It will stay
2160 with IEC conventions (i.e. to the base 1024) for software
2161 metrics, according to what is customary according to
2162 Wikipedia. We explicitly document which base applies for
2163 each configuration option.
2165 * The DeviceAllow= setting in unit files now supports a syntax
2166 to whitelist an entire group of devices node majors at once,
2167 based on the /proc/devices listing. For example, with the
2168 string "char-pts", it is now possible to whitelist all
2169 current and future pseudo-TTYs at once.
2171 * sd-event learned a new "post" event source. Event sources of
2172 this type are triggered by the dispatching of any event
2173 source of a type that is not "post". This is useful for
2174 implementing clean-up and check event sources that are
2175 triggered by other work being done in the program.
2177 * systemd-networkd is no longer statically enabled, but uses
2178 the usual [Install] sections so that it can be
2179 enabled/disabled using systemctl. It still is enabled by
2182 * When creating a veth interface pair with systemd-nspawn, the
2183 host side will now be prefixed with "vb-" if
2184 --network-bridge= is used, and with "ve-" if --network-veth
2185 is used. This way, it is easy to distinguish these cases on
2186 the host, for example to apply different configuration to
2187 them with systemd-networkd.
2189 * The compatibility libraries for libsystemd-journal.so,
2190 libsystem-id128.so, libsystemd-login.so and
2191 libsystemd-daemon.so do not make use of IFUNC
2192 anymore. Instead, we now build libsystemd.so multiple times
2193 under these alternative names. This means that the footprint
2194 is drastically increased, but given that these are
2195 transitional compatibility libraries, this should not matter
2196 much. This change has been made necessary to support the ARM
2197 platform for these compatibility libraries, as the ARM
2198 toolchain is not really at the same level as the toolchain
2199 for other architectures like x86 and does not support
2200 IFUNC. Please make sure to use --enable-compat-libs only
2201 during a transitional period!
2203 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
2204 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
2205 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
2206 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
2207 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
2208 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
2209 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
2210 Zbigniew Jędrzejewski-Szmek
2212 -- Berlin, 2014-02-24
2216 * A new component "systemd-networkd" has been added that can
2217 be used to configure local network interfaces statically or
2218 via DHCP. It is capable of bringing up bridges, VLANs, and
2219 bonding. Currently, no hook-ups for interactive network
2220 configuration are provided. Use this for your initrd,
2221 container, embedded, or server setup if you need a simple,
2222 yet powerful, network configuration solution. This
2223 configuration subsystem is quite nifty, as it allows wildcard
2224 hotplug matching in interfaces. For example, with a single
2225 configuration snippet, you can configure that all Ethernet
2226 interfaces showing up are automatically added to a bridge,
2227 or similar. It supports link-sensing and more.
2229 * A new tool "systemd-socket-proxyd" has been added which can
2230 act as a bidirectional proxy for TCP sockets. This is
2231 useful for adding socket activation support to services that
2232 do not actually support socket activation, including virtual
2233 machines and the like.
2235 * Add a new tool to save/restore rfkill state on
2238 * Save/restore state of keyboard backlights in addition to
2239 display backlights on shutdown/boot.
2241 * udev learned a new SECLABEL{} construct to label device
2242 nodes with a specific security label when they appear. For
2243 now, only SECLABEL{selinux} is supported, but the syntax is
2244 prepared for additional security frameworks.
2246 * udev gained a new scheme to configure link-level attributes
2247 from files in /etc/systemd/network/*.link. These files can
2248 match against MAC address, device path, driver name and type,
2249 and will apply attributes like the naming policy, link speed,
2250 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
2251 address assignment policy (randomized, ...).
2253 * The configuration of network interface naming rules for
2254 "permanent interface names" has changed: a new NamePolicy=
2255 setting in the [Link] section of .link files determines the
2256 priority of possible naming schemes (onboard, slot, mac,
2257 path). The default value of this setting is determined by
2258 /usr/lib/net/links/99-default.link. Old
2259 80-net-name-slot.rules udev configuration file has been
2260 removed, so local configuration overriding this file should
2261 be adapated to override 99-default.link instead.
2263 * When the User= switch is used in a unit file, also
2264 initialize $SHELL= based on the user database entry.
2266 * systemd no longer depends on libdbus. All communication is
2267 now done with sd-bus, systemd's low-level bus library
2270 * kdbus support has been added to PID 1 itself. When kdbus is
2271 enabled, this causes PID 1 to set up the system bus and
2272 enable support for a new ".busname" unit type that
2273 encapsulates bus name activation on kdbus. It works a little
2274 bit like ".socket" units, except for bus names. A new
2275 generator has been added that converts classic dbus1 service
2276 activation files automatically into native systemd .busname
2279 * sd-bus: add a light-weight vtable implementation that allows
2280 defining objects on the bus with a simple static const
2281 vtable array of its methods, signals and properties.
2283 * systemd will not generate or install static dbus
2284 introspection data anymore to /usr/share/dbus-1/interfaces,
2285 as the precise format of these files is unclear, and
2286 nothing makes use of it.
2288 * A proxy daemon is now provided to proxy clients connecting
2289 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
2290 compatibility with classic D-Bus.
2292 * A bus driver implementation has been added that supports the
2293 classic D-Bus bus driver calls on kdbus, also for
2294 compatibility purposes.
2296 * A new API "sd-event.h" has been added that implements a
2297 minimal event loop API built around epoll. It provides a
2298 couple of features that direct epoll usage is lacking:
2299 prioritization of events, scales to large numbers of timer
2300 events, per-event timer slack (accuracy), system-wide
2301 coalescing of timer events, exit handlers, watchdog
2302 supervision support using systemd's sd_notify() API, child
2305 * A new API "sd-rntl.h" has been added that provides an API
2306 around the route netlink interface of the kernel, similar in
2307 style to "sd-bus.h".
2309 * A new API "sd-dhcp-client.h" has been added that provides a
2310 small DHCPv4 client-side implementation. This is used by
2313 * There is a new kernel command line option
2314 "systemd.restore_state=0|1". When set to "0", none of the
2315 systemd tools will restore saved runtime state to hardware
2316 devices. More specifically, the rfkill and backlight states
2319 * The FsckPassNo= compatibility option in mount/service units
2320 has been removed. The fstab generator will now add the
2321 necessary dependencies automatically, and does not require
2322 PID1's support for that anymore.
2324 * journalctl gained a new switch, --list-boots, that lists
2325 recent boots with their times and boot IDs.
2327 * The various tools like systemctl, loginctl, timedatectl,
2328 busctl, systemd-run, ... have gained a new switch "-M" to
2329 connect to a specific, local OS container (as direct
2330 connection, without requiring SSH). This works on any
2331 container that is registered with machined, such as those
2332 created by libvirt-lxc or nspawn.
2334 * systemd-run and systemd-analyze also gained support for "-H"
2335 to connect to remote hosts via SSH. This is particularly
2336 useful for systemd-run because it enables queuing of jobs
2337 onto remote systems.
2339 * machinectl gained a new command "login" to open a getty
2340 login in any local container. This works with any container
2341 that is registered with machined (such as those created by
2342 libvirt-lxc or nspawn), and which runs systemd inside.
2344 * machinectl gained a new "reboot" command that may be used to
2345 trigger a reboot on a specific container that is registered
2346 with machined. This works on any container that runs an init
2347 system of some kind.
2349 * systemctl gained a new "list-timers" command to print a nice
2350 listing of installed timer units with the times they elapse
2353 * Alternative reboot() parameters may now be specified on the
2354 "systemctl reboot" command line and are passed to the
2355 reboot() system call.
2357 * systemctl gained a new --job-mode= switch to configure the
2358 mode to queue a job with. This is a more generic version of
2359 --fail, --irreversible, and --ignore-dependencies, which are
2360 still available but not advertised anymore.
2362 * /etc/systemd/system.conf gained new settings to configure
2363 various default timeouts of units, as well as the default
2364 start limit interval and burst. These may still be overridden
2367 * PID1 will now export on the bus profile data of the security
2368 policy upload process (such as the SELinux policy upload to
2371 * journald: when forwarding logs to the console, include
2372 timestamps (following the setting in
2373 /sys/module/printk/parameters/time).
2375 * OnCalendar= in timer units now understands the special
2376 strings "yearly" and "annually". (Both are equivalent)
2378 * The accuracy of timer units is now configurable with the new
2379 AccuracySec= setting. It defaults to 1min.
2381 * A new dependency type JoinsNamespaceOf= has been added that
2382 allows running two services within the same /tmp and network
2383 namespace, if PrivateNetwork= or PrivateTmp= are used.
2385 * A new command "cat" has been added to systemctl. It outputs
2386 the original unit file of a unit, and concatenates the
2387 contents of additional "drop-in" unit file snippets, so that
2388 the full configuration is shown.
2390 * systemctl now supports globbing on the various "list-xyz"
2391 commands, like "list-units" or "list-sockets", as well as on
2392 those commands which take multiple unit names.
2394 * journalctl's --unit= switch gained support for globbing.
2396 * All systemd daemons now make use of the watchdog logic so
2397 that systemd automatically notices when they hang.
2399 * If the $container_ttys environment variable is set,
2400 getty-generator will automatically spawn a getty for each
2401 listed tty. This is useful for container managers to request
2402 login gettys to be spawned on as many ttys as needed.
2404 * %h, %s, %U specifier support is not available anymore when
2405 used in unit files for PID 1. This is because NSS calls are
2406 not safe from PID 1. They stay available for --user
2407 instances of systemd, and as special case for the root user.
2409 * loginctl gained a new "--no-legend" switch to turn off output
2412 * The "sd-login.h" API gained three new calls:
2413 sd_session_is_remote(), sd_session_get_remote_user(),
2414 sd_session_get_remote_host() to query information about
2417 * The udev hardware database now also carries vendor/product
2418 information of SDIO devices.
2420 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
2421 determine whether watchdog notifications are requested by
2424 * Socket-activated per-connection services now include a
2425 short description of the connection parameters in the
2428 * tmpfiles gained a new "--boot" option. When this is not used,
2429 only lines where the command character is not suffixed with
2430 "!" are executed. When this option is specified, those
2431 options are executed too. This partitions tmpfiles
2432 directives into those that can be safely executed at any
2433 time, and those which should be run only at boot (for
2434 example, a line that creates /run/nologin).
2436 * A new API "sd-resolve.h" has been added which provides a simple
2437 asynchronous wrapper around glibc NSS host name resolution
2438 calls, such as getaddrinfo(). In contrast to glibc's
2439 getaddrinfo_a(), it does not use signals. In contrast to most
2440 other asynchronous name resolution libraries, this one does
2441 not reimplement DNS, but reuses NSS, so that alternate
2442 host name resolution systems continue to work, such as mDNS,
2443 LDAP, etc. This API is based on libasyncns, but it has been
2444 cleaned up for inclusion in systemd.
2446 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
2447 "sd-daemon.h" are no longer found in individual libraries
2448 libsystemd-journal.so, libsystemd-login.so,
2449 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
2450 merged them into a single library, libsystemd.so, which
2451 provides all symbols. The reason for this is cyclic
2452 dependencies, as these libraries tend to use each other's
2453 symbols. So far, we have managed to workaround that by linking
2454 a copy of a good part of our code into each of these
2455 libraries again and again, which, however, makes certain
2456 things hard to do, like sharing static variables. Also, it
2457 substantially increases footprint. With this change, there
2458 is only one library for the basic APIs systemd
2459 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
2460 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
2461 library as well, however are subject to the --enable-kdbus
2462 switch (see below). Note that "sd-dhcp-client.h" is not part
2463 of this library (this is because it only consumes, never
2464 provides, services of/to other APIs). To make the transition
2465 easy from the separate libraries to the unified one, we
2466 provide the --enable-compat-libs compile-time switch which
2467 will generate stub libraries that are compatible with the
2468 old ones but redirect all calls to the new one.
2470 * All of the kdbus logic and the new APIs "sd-bus.h",
2471 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
2472 and "sd-utf8.h" are compile-time optional via the
2473 "--enable-kdbus" switch, and they are not compiled in by
2474 default. To make use of kdbus, you have to explicitly enable
2475 the switch. Note however, that neither the kernel nor the
2476 userspace API for all of this is considered stable yet. We
2477 want to maintain the freedom to still change the APIs for
2478 now. By specifying this build-time switch, you acknowledge
2479 that you are aware of the instability of the current
2482 * Also, note that while kdbus is pretty much complete,
2483 it lacks one thing: proper policy support. This means you
2484 can build a fully working system with all features; however,
2485 it will be highly insecure. Policy support will be added in
2486 one of the next releases, at the same time that we will
2487 declare the APIs stable.
2489 * When the kernel command line argument "kdbus" is specified,
2490 systemd will automatically load the kdbus.ko kernel module. At
2491 this stage of development, it is only useful for testing kdbus
2492 and should not be used in production. Note: if "--enable-kdbus"
2493 is specified, and the kdbus.ko kernel module is available, and
2494 "kdbus" is added to the kernel command line, the entire system
2495 runs with kdbus instead of dbus-daemon, with the above mentioned
2496 problem of missing the system policy enforcement. Also a future
2497 version of kdbus.ko or a newer systemd will not be compatible with
2498 each other, and will unlikely be able to boot the machine if only
2499 one of them is updated.
2501 * systemctl gained a new "import-environment" command which
2502 uploads the caller's environment (or parts thereof) into the
2503 service manager so that it is inherited by services started
2504 by the manager. This is useful to upload variables like
2505 $DISPLAY into the user service manager.
2507 * A new PrivateDevices= switch has been added to service units
2508 which allows running a service with a namespaced /dev
2509 directory that does not contain any device nodes for
2510 physical devices. More specifically, it only includes devices
2511 such as /dev/null, /dev/urandom, and /dev/zero which are API
2514 * logind has been extended to support behaviour like VT
2515 switching on seats that do not support a VT. This makes
2516 multi-session available on seats that are not the first seat
2517 (seat0), and on systems where kernel support for VTs has
2518 been disabled at compile-time.
2520 * If a process holds a delay lock for system sleep or shutdown
2521 and fails to release it in time, we will now log its
2522 identity. This makes it easier to identify processes that
2523 cause slow suspends or power-offs.
2525 * When parsing /etc/crypttab, support for a new key-slot=
2526 option as supported by Debian is added. It allows indicating
2527 which LUKS slot to use on disk, speeding up key loading.
2529 * The sd_journald_sendv() API call has been checked and
2530 officially declared to be async-signal-safe so that it may
2531 be invoked from signal handlers for logging purposes.
2533 * Boot-time status output is now enabled automatically after a
2534 short timeout if boot does not progress, in order to give
2535 the user an indication what she or he is waiting for.
2537 * The boot-time output has been improved to show how much time
2538 remains until jobs expire.
2540 * The KillMode= switch in service units gained a new possible
2541 value "mixed". If set, and the unit is shut down, then the
2542 initial SIGTERM signal is sent only to the main daemon
2543 process, while the following SIGKILL signal is sent to
2544 all remaining processes of the service.
2546 * When a scope unit is registered, a new property "Controller"
2547 may be set. If set to a valid bus name, systemd will send a
2548 RequestStop() signal to this name when it would like to shut
2549 down the scope. This may be used to hook manager logic into
2550 the shutdown logic of scope units. Also, scope units may now
2551 be put in a special "abandoned" state, in which case the
2552 manager process which created them takes no further
2553 responsibilities for it.
2555 * When reading unit files, systemd will now verify
2556 the access mode of these files, and warn about certain
2557 suspicious combinations. This has been added to make it
2558 easier to track down packaging bugs where unit files are
2559 marked executable or world-writable.
2561 * systemd-nspawn gained a new "--setenv=" switch to set
2562 container-wide environment variables. The similar option in
2563 systemd-activate was renamed from "--environment=" to
2564 "--setenv=" for consistency.
2566 * systemd-nspawn has been updated to create a new kdbus domain
2567 for each container that is invoked, thus allowing each
2568 container to have its own set of system and user buses,
2569 independent of the host.
2571 * systemd-nspawn gained a new --drop-capability= switch to run
2572 the container with less capabilities than the default. Both
2573 --drop-capability= and --capability= now take the special
2574 string "all" for dropping or keeping all capabilities.
2576 * systemd-nspawn gained new switches for executing containers
2577 with specific SELinux labels set.
2579 * systemd-nspawn gained a new --quiet switch to not generate
2580 any additional output but the container's own console
2583 * systemd-nspawn gained a new --share-system switch to run a
2584 container without PID namespacing enabled.
2586 * systemd-nspawn gained a new --register= switch to control
2587 whether the container is registered with systemd-machined or
2588 not. This is useful for containers that do not run full
2589 OS images, but only specific apps.
2591 * systemd-nspawn gained a new --keep-unit which may be used
2592 when invoked as the only program from a service unit, and
2593 results in registration of the unit service itself in
2594 systemd-machined, instead of a newly opened scope unit.
2596 * systemd-nspawn gained a new --network-interface= switch for
2597 moving arbitrary interfaces to the container. The new
2598 --network-veth switch creates a virtual Ethernet connection
2599 between host and container. The new --network-bridge=
2600 switch then allows assigning the host side of this virtual
2601 Ethernet connection to a bridge device.
2603 * systemd-nspawn gained a new --personality= switch for
2604 setting the kernel personality for the container. This is
2605 useful when running a 32-bit container on a 64-bit host. A
2606 similar option Personality= is now also available for service
2609 * logind will now also track a "Desktop" identifier for each
2610 session which encodes the desktop environment of it. This is
2611 useful for desktop environments that want to identify
2612 multiple running sessions of itself easily.
2614 * A new SELinuxContext= setting for service units has been
2615 added that allows setting a specific SELinux execution
2616 context for a service.
2618 * Most systemd client tools will now honour $SYSTEMD_LESS for
2619 settings of the "less" pager. By default, these tools will
2620 override $LESS to allow certain operations to work, such as
2621 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
2622 influence this logic.
2624 * systemd's "seccomp" hook-up has been changed to make use of
2625 the libseccomp library instead of using its own
2626 implementation. This has benefits for portability among
2629 * For usage together with SystemCallFilter=, a new
2630 SystemCallErrorNumber= setting has been introduced that
2631 allows configuration of a system error number to be returned
2632 on filtered system calls, instead of immediately killing the
2633 process. Also, SystemCallArchitectures= has been added to
2634 limit access to system calls of a particular architecture
2635 (in order to turn off support for unused secondary
2636 architectures). There is also a global
2637 SystemCallArchitectures= setting in system.conf now to turn
2638 off support for non-native system calls system-wide.
2640 * systemd requires a kernel with a working name_to_handle_at(),
2641 please see the kernel config requirements in the README file.
2643 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
2644 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
2645 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
2646 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
2647 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
2648 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
2649 Elia Pinto, Florian Weimer, George McCollister, Goffredo
2650 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
2651 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
2652 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
2653 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
2654 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
2655 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
2656 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
2657 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
2658 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
2659 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
2660 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
2661 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
2662 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
2663 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
2664 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
2665 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
2666 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
2668 -- Berlin, 2014-02-20
2672 * logind has gained support for facilitating privileged input
2673 and drm device access for unprivileged clients. This work is
2674 useful to allow Wayland display servers (and similar
2675 programs, such as kmscon) to run under the user's ID and
2676 access input and drm devices which are normally
2677 protected. When this is used (and the kernel is new enough)
2678 logind will "mute" IO on the file descriptors passed to
2679 Wayland as long as it is in the background and "unmute" it
2680 if it returns into the foreground. This allows secure
2681 session switching without allowing background sessions to
2682 eavesdrop on input and display data. This also introduces
2683 session switching support if VT support is turned off in the
2684 kernel, and on seats that are not seat0.
2686 * A new kernel command line option luks.options= is understood
2687 now which allows specifying LUKS options for usage for LUKS
2688 encrypted partitions specified with luks.uuid=.
2690 * tmpfiles.d(5) snippets may now use specifier expansion in
2691 path names. More specifically %m, %b, %H, %v, are now
2692 replaced by the local machine id, boot id, hostname, and
2693 kernel version number.
2695 * A new tmpfiles.d(5) command "m" has been introduced which
2696 may be used to change the owner/group/access mode of a file
2697 or directory if it exists, but do nothing if it does not.
2699 * This release removes high-level support for the
2700 MemorySoftLimit= cgroup setting. The underlying kernel
2701 cgroup attribute memory.soft_limit= is currently badly
2702 designed and likely to be removed from the kernel API in its
2703 current form, hence we should not expose it for now.
2705 * The memory.use_hierarchy cgroup attribute is now enabled for
2706 all cgroups systemd creates in the memory cgroup
2707 hierarchy. This option is likely to be come the built-in
2708 default in the kernel anyway, and the non-hierarchical mode
2709 never made much sense in the intrinsically hierarchical
2712 * A new field _SYSTEMD_SLICE= is logged along with all journal
2713 messages containing the slice a message was generated
2714 from. This is useful to allow easy per-customer filtering of
2715 logs among other things.
2717 * systemd-journald will no longer adjust the group of journal
2718 files it creates to the "systemd-journal" group. Instead we
2719 rely on the journal directory to be owned by the
2720 "systemd-journal" group, and its setgid bit set, so that the
2721 kernel file system layer will automatically enforce that
2722 journal files inherit this group assignment. The reason for
2723 this change is that we cannot allow NSS look-ups from
2724 journald which would be necessary to resolve
2725 "systemd-journal" to a numeric GID, because this might
2726 create deadlocks if NSS involves synchronous queries to
2727 other daemons (such as nscd, or sssd) which in turn are
2728 logging clients of journald and might block on it, which
2729 would then dead lock. A tmpfiles.d(5) snippet included in
2730 systemd will make sure the setgid bit and group are
2731 properly set on the journal directory if it exists on every
2732 boot. However, we recommend adjusting it manually after
2733 upgrades too (or from RPM scriptlets), so that the change is
2734 not delayed until next reboot.
2736 * Backlight and random seed files in /var/lib/ have moved into
2737 the /var/lib/systemd/ directory, in order to centralize all
2738 systemd generated files in one directory.
2740 * Boot time performance measurements (as displayed by
2741 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
2742 performance information if that's available to determine how
2743 much time BIOS and boot loader initialization required. With
2744 a sufficiently new BIOS you hence no longer need to boot
2745 with Gummiboot to get access to such information.
2747 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
2748 Cristian Rodríguez, Dave Reisner, David Herrmann, David
2749 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
2750 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
2751 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
2752 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
2753 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
2755 -- Berlin, 2013-10-02
2759 * The Restart= option for services now understands a new
2760 on-watchdog setting, which will restart the service
2761 automatically if the service stops sending out watchdog keep
2762 alive messages (as configured with WatchdogSec=).
2764 * The getty generator (which is responsible for bringing up a
2765 getty on configured serial consoles) will no longer only
2766 start a getty on the primary kernel console but on all
2767 others, too. This makes the order in which console= is
2768 specified on the kernel command line less important.
2770 * libsystemd-logind gained a new sd_session_get_vt() call to
2771 retrieve the VT number of a session.
2773 * If the option "tries=0" is set for an entry of /etc/crypttab
2774 its passphrase is queried indefinitely instead of any
2775 maximum number of tries.
2777 * If a service with a configure PID file terminates its PID
2778 file will now be removed automatically if it still exists
2779 afterwards. This should put an end to stale PID files.
2781 * systemd-run will now also take relative binary path names
2782 for execution and no longer insists on absolute paths.
2784 * InaccessibleDirectories= and ReadOnlyDirectories= now take
2785 paths that are optionally prefixed with "-" to indicate that
2786 it should not be considered a failure if they do not exist.
2788 * journalctl -o (and similar commands) now understands a new
2789 output mode "short-precise", it is similar to "short" but
2790 shows timestamps with usec accuracy.
2792 * The option "discard" (as known from Debian) is now
2793 synonymous to "allow-discards" in /etc/crypttab. In fact,
2794 "discard" is preferred now (since it is easier to remember
2797 * Some licensing clean-ups were made, so that more code is now
2798 LGPL-2.1 licensed than before.
2800 * A minimal tool to save/restore the display backlight
2801 brightness across reboots has been added. It will store the
2802 backlight setting as late as possible at shutdown, and
2803 restore it as early as possible during reboot.
2805 * A logic to automatically discover and enable home and swap
2806 partitions on GPT disks has been added. With this in place
2807 /etc/fstab becomes optional for many setups as systemd can
2808 discover certain partitions located on the root disk
2809 automatically. Home partitions are recognized under their
2810 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
2811 partitions are recognized under their GPT type ID
2812 0657fd6da4ab43c484e50933c84b4f4f.
2814 * systemd will no longer pass any environment from the kernel
2815 or initrd to system services. If you want to set an
2816 environment for all services, do so via the kernel command
2817 line systemd.setenv= assignment.
2819 * The systemd-sysctl tool no longer natively reads the file
2820 /etc/sysctl.conf. If desired, the file should be symlinked
2821 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
2822 legacy support by a symlink rather than built-in code, it
2823 also makes the otherwise hidden order of application of the
2824 different files visible. (Note that this partly reverts to a
2825 pre-198 application order of sysctl knobs!)
2827 * The "systemctl set-log-level" and "systemctl dump" commands
2828 have been moved to systemd-analyze.
2830 * systemd-run learned the new --remain-after-exit switch,
2831 which causes the scope unit not to be cleaned up
2832 automatically after the process terminated.
2834 * tmpfiles learned a new --exclude-prefix= switch to exclude
2835 certain paths from operation.
2837 * journald will now automatically flush all messages to disk
2838 as soon as a message at the log level CRIT, ALERT or EMERG
2841 Contributions from: Andrew Cook, Brandon Philips, Christian
2842 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
2843 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
2844 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
2845 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
2846 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
2847 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
2848 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
2849 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
2850 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
2851 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
2852 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
2853 William Giokas, Zbigniew Jędrzejewski-Szmek
2855 -- Berlin, 2013-09-13
2859 * The documentation has been updated to cover the various new
2860 concepts introduced with 205.
2862 * Unit files now understand the new %v specifier which
2863 resolves to the kernel version string as returned by "uname
2866 * systemctl now supports filtering the unit list output by
2867 load state, active state and sub state, using the new
2870 * "systemctl status" will now show the results of the
2871 condition checks (like ConditionPathExists= and similar) of
2872 the last start attempts of the unit. They are also logged to
2875 * "journalctl -b" may now be used to look for boot output of a
2876 specific boot. Try "journalctl -b -1" for the previous boot,
2877 but the syntax is substantially more powerful.
2879 * "journalctl --show-cursor" has been added which prints the
2880 cursor string the last shown log line. This may then be used
2881 with the new "journalctl --after-cursor=" switch to continue
2882 browsing logs from that point on.
2884 * "journalctl --force" may now be used to force regeneration
2887 * Creation of "dead" device nodes has been moved from udev
2888 into kmod and tmpfiles. Previously, udev would read the kmod
2889 databases to pre-generate dead device nodes based on meta
2890 information contained in kernel modules, so that these would
2891 be auto-loaded on access rather then at boot. As this
2892 does not really have much to do with the exposing actual
2893 kernel devices to userspace this has always been slightly
2894 alien in the udev codebase. Following the new scheme kmod
2895 will now generate a runtime snippet for tmpfiles from the
2896 module meta information and it now is tmpfiles' job to the
2897 create the nodes. This also allows overriding access and
2898 other parameters for the nodes using the usual tmpfiles
2899 facilities. As side effect this allows us to remove the
2900 CAP_SYS_MKNOD capability bit from udevd entirely.
2902 * logind's device ACLs may now be applied to these "dead"
2903 devices nodes too, thus finally allowing managed access to
2904 devices such as /dev/snd/sequencer whithout loading the
2905 backing module right-away.
2907 * A new RPM macro has been added that may be used to apply
2908 tmpfiles configuration during package installation.
2910 * systemd-detect-virt and ConditionVirtualization= now can
2911 detect User-Mode-Linux machines (UML).
2913 * journald will now implicitly log the effective capabilities
2914 set of processes in the message metadata.
2916 * systemd-cryptsetup has gained support for TrueCrypt volumes.
2918 * The initrd interface has been simplified (more specifically,
2919 support for passing performance data via environment
2920 variables and fsck results via files in /run has been
2921 removed). These features were non-essential, and are
2922 nowadays available in a much nicer way by having systemd in
2923 the initrd serialize its state and have the hosts systemd
2924 deserialize it again.
2926 * The udev "keymap" data files and tools to apply keyboard
2927 specific mappings of scan to key codes, and force-release
2928 scan code lists have been entirely replaced by a udev
2929 "keyboard" builtin and a hwdb data file.
2931 * systemd will now honour the kernel's "quiet" command line
2932 argument also during late shutdown, resulting in a
2933 completely silent shutdown when used.
2935 * There's now an option to control the SO_REUSEPORT socket
2936 option in .socket units.
2938 * Instance units will now automatically get a per-template
2939 subslice of system.slice unless something else is explicitly
2940 configured. For example, instances of sshd@.service will now
2941 implicitly be placed in system-sshd.slice rather than
2942 system.slice as before.
2944 * Test coverage support may now be enabled at build time.
2946 Contributions from: Dave Reisner, Frederic Crozat, Harald
2947 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
2948 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
2949 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
2950 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
2951 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
2952 Giokas, Zbigniew Jędrzejewski-Szmek
2954 -- Berlin, 2013-07-23
2958 * Two new unit types have been introduced:
2960 Scope units are very similar to service units, however, are
2961 created out of pre-existing processes -- instead of PID 1
2962 forking off the processes. By using scope units it is
2963 possible for system services and applications to group their
2964 own child processes (worker processes) in a powerful way
2965 which then maybe used to organize them, or kill them
2966 together, or apply resource limits on them.
2968 Slice units may be used to partition system resources in an
2969 hierarchical fashion and then assign other units to them. By
2970 default there are now three slices: system.slice (for all
2971 system services), user.slice (for all user sessions),
2972 machine.slice (for VMs and containers).
2974 Slices and scopes have been introduced primarily in
2975 context of the work to move cgroup handling to a
2976 single-writer scheme, where only PID 1
2977 creates/removes/manages cgroups.
2979 * There's a new concept of "transient" units. In contrast to
2980 normal units these units are created via an API at runtime,
2981 not from configuration from disk. More specifically this
2982 means it is now possible to run arbitrary programs as
2983 independent services, with all execution parameters passed
2984 in via bus APIs rather than read from disk. Transient units
2985 make systemd substantially more dynamic then it ever was,
2986 and useful as a general batch manager.
2988 * logind has been updated to make use of scope and slice units
2989 for managing user sessions. As a user logs in he will get
2990 his own private slice unit, to which all sessions are added
2991 as scope units. We also added support for automatically
2992 adding an instance of user@.service for the user into the
2993 slice. Effectively logind will no longer create cgroup
2994 hierarchies on its own now, it will defer entirely to PID 1
2995 for this by means of scope, service and slice units. Since
2996 user sessions this way become entities managed by PID 1
2997 the output of "systemctl" is now a lot more comprehensive.
2999 * A new mini-daemon "systemd-machined" has been added which
3000 may be used by virtualization managers to register local
3001 VMs/containers. nspawn has been updated accordingly, and
3002 libvirt will be updated shortly. machined will collect a bit
3003 of meta information about the VMs/containers, and assign
3004 them their own scope unit (see above). The collected
3005 meta-data is then made available via the "machinectl" tool,
3006 and exposed in "ps" and similar tools. machined/machinectl
3007 is compile-time optional.
3009 * As discussed earlier, the low-level cgroup configuration
3010 options ControlGroup=, ControlGroupModify=,
3011 ControlGroupPersistent=, ControlGroupAttribute= have been
3012 removed. Please use high-level attribute settings instead as
3013 well as slice units.
3015 * A new bus call SetUnitProperties() has been added to alter
3016 various runtime parameters of a unit. This is primarily
3017 useful to alter cgroup parameters dynamically in a nice way,
3018 but will be extended later on to make more properties
3019 modifiable at runtime. systemctl gained a new set-properties
3020 command that wraps this call.
3022 * A new tool "systemd-run" has been added which can be used to
3023 run arbitrary command lines as transient services or scopes,
3024 while configuring a number of settings via the command
3025 line. This tool is currently very basic, however already
3026 very useful. We plan to extend this tool to even allow
3027 queuing of execution jobs with time triggers from the
3028 command line, similar in fashion to "at".
3030 * nspawn will now inform the user explicitly that kernels with
3031 audit enabled break containers, and suggest the user to turn
3034 * Support for detecting the IMA and AppArmor security
3035 frameworks with ConditionSecurity= has been added.
3037 * journalctl gained a new "-k" switch for showing only kernel
3038 messages, mimicking dmesg output; in addition to "--user"
3039 and "--system" switches for showing only user's own logs
3042 * systemd-delta can now show information about drop-in
3043 snippets extending unit files.
3045 * libsystemd-bus has been substantially updated but is still
3046 not available as public API.
3048 * systemd will now look for the "debug" argument on the kernel
3049 command line and enable debug logging, similar to what
3050 "systemd.log_level=debug" already did before.
3052 * "systemctl set-default", "systemctl get-default" has been
3053 added to configure the default.target symlink, which
3054 controls what to boot into by default.
3056 * "systemctl set-log-level" has been added as a convenient
3057 way to raise and lower systemd logging threshold.
3059 * "systemd-analyze plot" will now show the time the various
3060 generators needed for execution, as well as information
3061 about the unit file loading.
3063 * libsystemd-journal gained a new sd_journal_open_files() call
3064 for opening specific journal files. journactl also gained a
3065 new switch to expose this new functionality. Previously we
3066 only supported opening all files from a directory, or all
3067 files from the system, as opening individual files only is
3068 racy due to journal file rotation.
3070 * systemd gained the new DefaultEnvironment= setting in
3071 /etc/systemd/system.conf to set environment variables for
3074 * If a privileged process logs a journal message with the
3075 OBJECT_PID= field set, then journald will automatically
3076 augment this with additional OBJECT_UID=, OBJECT_GID=,
3077 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
3078 system services want to log events about specific client
3079 processes. journactl/systemctl has been updated to make use
3080 of this information if all log messages regarding a specific
3083 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
3084 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
3085 Reisner, David Coppa, David King, David Strauss, Eelco
3086 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
3087 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
3088 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
3089 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
3090 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
3091 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
3092 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
3093 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
3094 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
3095 Łukasz Stelmach, 장동준
3099 * The Python bindings gained some minimal support for the APIs
3100 exposed by libsystemd-logind.
3102 * ConditionSecurity= gained support for detecting SMACK. Since
3103 this condition already supports SELinux and AppArmor we only
3104 miss IMA for this. Patches welcome!
3106 Contributions from: Karol Lewandowski, Lennart Poettering,
3107 Zbigniew Jędrzejewski-Szmek
3111 * systemd-nspawn will now create /etc/resolv.conf if
3112 necessary, before bind-mounting the host's file onto it.
3114 * systemd-nspawn will now store meta information about a
3115 container on the container's cgroup as extended attribute
3116 fields, including the root directory.
3118 * The cgroup hierarchy has been reworked in many ways. All
3119 objects any of the components systemd creates in the cgroup
3120 tree are now suffixed. More specifically, user sessions are
3121 now placed in cgroups suffixed with ".session", users in
3122 cgroups suffixed with ".user", and nspawn containers in
3123 cgroups suffixed with ".nspawn". Furthermore, all cgroup
3124 names are now escaped in a simple scheme to avoid collision
3125 of userspace object names with kernel filenames. This work
3126 is preparation for making these objects relocatable in the
3127 cgroup tree, in order to allow easy resource partitioning of
3128 these objects without causing naming conflicts.
3130 * systemctl list-dependencies gained the new switches
3131 --plain, --reverse, --after and --before.
3133 * systemd-inhibit now shows the process name of processes that
3134 have taken an inhibitor lock.
3136 * nss-myhostname will now also resolve "localhost"
3137 implicitly. This makes /etc/hosts an optional file and
3138 nicely handles that on IPv6 ::1 maps to both "localhost" and
3141 * libsystemd-logind.so gained a new call
3142 sd_get_machine_names() to enumerate running containers and
3143 VMs (currently only supported by very new libvirt and
3144 nspawn). sd_login_monitor can now be used to watch
3145 VMs/containers coming and going.
3147 * .include is not allowed recursively anymore, and only in
3148 unit files. Usually it is better to use drop-in snippets in
3149 .d/*.conf anyway, as introduced with systemd 198.
3151 * systemd-analyze gained a new "critical-chain" command that
3152 determines the slowest chain of units run during system
3153 boot-up. It is very useful for tracking down where
3154 optimizing boot time is the most beneficial.
3156 * systemd will no longer allow manipulating service paths in
3157 the name=systemd:/system cgroup tree using ControlGroup= in
3158 units. (But is still fine with it in all other dirs.)
3160 * There's a new systemd-nspawn@.service service file that may
3161 be used to easily run nspawn containers as system
3162 services. With the container's root directory in
3163 /var/lib/container/foobar it is now sufficient to run
3164 "systemctl start systemd-nspawn@foobar.service" to boot it.
3166 * systemd-cgls gained a new parameter "--machine" to list only
3167 the processes within a certain container.
3169 * ConditionSecurity= now can check for "apparmor". We still
3170 are lacking checks for SMACK and IMA for this condition
3171 check though. Patches welcome!
3173 * A new configuration file /etc/systemd/sleep.conf has been
3174 added that may be used to configure which kernel operation
3175 systemd is supposed to execute when "suspend", "hibernate"
3176 or "hybrid-sleep" is requested. This makes the new kernel
3177 "freeze" state accessible to the user.
3179 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
3180 the passed argument if applicable.
3182 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
3183 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
3184 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
3185 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
3186 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
3187 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
3188 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
3193 * The output of 'systemctl list-jobs' got some polishing. The
3194 '--type=' argument may now be passed more than once. A new
3195 command 'systemctl list-sockets' has been added which shows
3196 a list of kernel sockets systemd is listening on with the
3197 socket units they belong to, plus the units these socket
3200 * The experimental libsystemd-bus library got substantial
3201 updates to work in conjunction with the (also experimental)
3202 kdbus kernel project. It works well enough to exchange
3203 messages with some sophistication. Note that kdbus is not
3204 ready yet, and the library is mostly an elaborate test case
3205 for now, and not installable.
3207 * systemd gained a new unit 'systemd-static-nodes.service'
3208 that generates static device nodes earlier during boot, and
3209 can run in conjunction with udev.
3211 * libsystemd-login gained a new call sd_pid_get_user_unit()
3212 to retrieve the user systemd unit a process is running
3213 in. This is useful for systems where systemd is used as
3216 * systemd-nspawn now places all containers in the new /machine
3217 top-level cgroup directory in the name=systemd
3218 hierarchy. libvirt will soon do the same, so that we get a
3219 uniform separation of /system, /user and /machine for system
3220 services, user processes and containers/virtual
3221 machines. This new cgroup hierarchy is also useful to stick
3222 stable names to specific container instances, which can be
3223 recognized later this way (this name may be controlled
3224 via systemd-nspawn's new -M switch). libsystemd-login also
3225 gained a new call sd_pid_get_machine_name() to retrieve the
3226 name of the container/VM a specific process belongs to.
3228 * bootchart can now store its data in the journal.
3230 * libsystemd-journal gained a new call
3231 sd_journal_add_conjunction() for AND expressions to the
3232 matching logic. This can be used to express more complex
3233 logical expressions.
3235 * journactl can now take multiple --unit= and --user-unit=
3238 * The cryptsetup logic now understands the "luks.key=" kernel
3239 command line switch for specifying a file to read the
3240 decryption key from. Also, if a configured key file is not
3241 found the tool will now automatically fall back to prompting
3244 * Python systemd.journal module was updated to wrap recently
3245 added functions from libsystemd-journal. The interface was
3246 changed to bring the low level interface in s.j._Reader
3247 closer to the C API, and the high level interface in
3248 s.j.Reader was updated to wrap and convert all data about
3251 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
3252 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
3253 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
3254 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
3255 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
3256 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3260 * journalctl --update-catalog now understands a new --root=
3261 option to operate on catalogs found in a different root
3264 * During shutdown after systemd has terminated all running
3265 services a final killing loop kills all remaining left-over
3266 processes. We will now print the name of these processes
3267 when we send SIGKILL to them, since this usually indicates a
3270 * If /etc/crypttab refers to password files stored on
3271 configured mount points automatic dependencies will now be
3272 generated to ensure the specific mount is established first
3273 before the key file is attempted to be read.
3275 * 'systemctl status' will now show information about the
3276 network sockets a socket unit is listening on.
3278 * 'systemctl status' will also shown information about any
3279 drop-in configuration file for units. (Drop-In configuration
3280 files in this context are files such as
3281 /etc/systemd/systemd/foobar.service.d/*.conf)
3283 * systemd-cgtop now optionally shows summed up CPU times of
3284 cgroups. Press '%' while running cgtop to switch between
3285 percentage and absolute mode. This is useful to determine
3286 which cgroups use up the most CPU time over the entire
3287 runtime of the system. systemd-cgtop has also been updated
3288 to be 'pipeable' for processing with further shell tools.
3290 * 'hostnamectl set-hostname' will now allow setting of FQDN
3293 * The formatting and parsing of time span values has been
3294 changed. The parser now understands fractional expressions
3295 such as "5.5h". The formatter will now output fractional
3296 expressions for all time spans under 1min, i.e. "5.123456s"
3297 rather than "5s 123ms 456us". For time spans under 1s
3298 millisecond values are shown, for those under 1ms
3299 microsecond values are shown. This should greatly improve
3300 all time-related output of systemd.
3302 * libsystemd-login and libsystemd-journal gained new
3303 functions for querying the poll() events mask and poll()
3304 timeout value for integration into arbitrary event
3307 * localectl gained the ability to list available X11 keymaps
3308 (models, layouts, variants, options).
3310 * 'systemd-analyze dot' gained the ability to filter for
3311 specific units via shell-style globs, to create smaller,
3312 more useful graphs. I.e. it is now possible to create simple
3313 graphs of all the dependencies between only target units, or
3314 of all units that Avahi has dependencies with.
3316 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
3317 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
3318 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
3319 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
3320 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
3321 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
3322 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
3326 * The boot-time readahead implementation for rotating media
3327 will now read the read-ahead data in multiple passes which
3328 consist of all read requests made in equidistant time
3329 intervals. This means instead of strictly reading read-ahead
3330 data in its physical order on disk we now try to find a
3331 middle ground between physical and access time order.
3333 * /etc/os-release files gained a new BUILD_ID= field for usage
3334 on operating systems that provide continuous builds of OS
3337 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
3338 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
3339 William Douglas, Zbigniew Jędrzejewski-Szmek
3343 * systemd-python gained an API exposing libsystemd-daemon.
3345 * The SMACK setup logic gained support for uploading CIPSO
3348 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
3349 ReadOnlyDirectories= and InaccessibleDirectories= has
3350 changed. The private /tmp and /var/tmp directories are now
3351 shared by all processes of a service (which means
3352 ExecStartPre= may now leave data in /tmp that ExecStart= of
3353 the same service can still access). When a service is
3354 stopped its temporary directories are immediately deleted
3355 (normal clean-up with tmpfiles is still done in addition to
3358 * By default, systemd will now set a couple of sysctl
3359 variables in the kernel: the safe sysrq options are turned
3360 on, IP route verification is turned on, and source routing
3361 disabled. The recently added hardlink and softlink
3362 protection of the kernel is turned on. These settings should
3363 be reasonably safe, and good defaults for all new systems.
3365 * The predictable network naming logic may now be turned off
3366 with a new kernel command line switch: net.ifnames=0.
3368 * A new libsystemd-bus module has been added that implements a
3369 pretty complete D-Bus client library. For details see:
3371 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
3373 * journald will now explicitly flush the journal files to disk
3374 at the latest 5min after each write. The file will then also
3375 be marked offline until the next write. This should increase
3376 reliability in case of a crash. The synchronization delay
3377 can be configured via SyncIntervalSec= in journald.conf.
3379 * There's a new remote-fs-setup.target unit that can be used
3380 to pull in specific services when at least one remote file
3381 system is to be mounted.
3383 * There are new targets timers.target and paths.target as
3384 canonical targets to pull user timer and path units in
3385 from. This complements sockets.target with a similar
3386 purpose for socket units.
3388 * libudev gained a new call udev_device_set_attribute_value()
3389 to set sysfs attributes of a device.
3391 * The udev daemon now sets the default number of worker
3392 processes executed in parallel based on the number of available
3393 CPUs instead of the amount of available RAM. This is supposed
3394 to provide a more reliable default and limit a too aggressive
3395 paralellism for setups with 1000s of devices connected.
3397 Contributions from: Auke Kok, Colin Walters, Cristian
3398 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
3399 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
3400 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
3401 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
3402 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
3403 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
3404 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
3405 Zbigniew Jędrzejewski-Szmek
3409 * Configuration of unit files may now be extended via drop-in
3410 files without having to edit/override the unit files
3411 themselves. More specifically, if the administrator wants to
3412 change one value for a service file foobar.service he can
3413 now do so by dropping in a configuration snippet into
3414 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
3415 will load all these snippets and apply them on top of the
3416 main unit configuration file, possibly extending or
3417 overriding its settings. Using these drop-in snippets is
3418 generally nicer than the two earlier options for changing
3419 unit files locally: copying the files from
3420 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
3421 them there; or creating a new file in /etc/systemd/system/
3422 that incorporates the original one via ".include". Drop-in
3423 snippets into these .d/ directories can be placed in any
3424 directory systemd looks for units in, and the usual
3425 overriding semantics between /usr/lib, /etc and /run apply
3428 * Most unit file settings which take lists of items can now be
3429 reset by assigning the empty string to them. For example,
3430 normally, settings such as Environment=FOO=BAR append a new
3431 environment variable assignment to the environment block,
3432 each time they are used. By assigning Environment= the empty
3433 string the environment block can be reset to empty. This is
3434 particularly useful with the .d/*.conf drop-in snippets
3435 mentioned above, since this adds the ability to reset list
3436 settings from vendor unit files via these drop-ins.
3438 * systemctl gained a new "list-dependencies" command for
3439 listing the dependencies of a unit recursively.
3441 * Inhibitors are now honored and listed by "systemctl
3442 suspend", "systemctl poweroff" (and similar) too, not only
3443 GNOME. These commands will also list active sessions by
3446 * Resource limits (as exposed by the various control group
3447 controllers) can now be controlled dynamically at runtime
3448 for all units. More specifically, you can now use a command
3449 like "systemctl set-cgroup-attr foobar.service cpu.shares
3450 2000" to alter the CPU shares a specific service gets. These
3451 settings are stored persistently on disk, and thus allow the
3452 administrator to easily adjust the resource usage of
3453 services with a few simple commands. This dynamic resource
3454 management logic is also available to other programs via the
3455 bus. Almost any kernel cgroup attribute and controller is
3458 * systemd-vconsole-setup will now copy all font settings to
3459 all allocated VTs, where it previously applied them only to
3462 * libsystemd-login gained the new sd_session_get_tty() API
3465 * This release drops support for a few legacy or
3466 distribution-specific LSB facility names when parsing init
3467 scripts: $x-display-manager, $mail-transfer-agent,
3468 $mail-transport-agent, $mail-transfer-agent, $smtp,
3469 $null. Also, the mail-transfer-agent.target unit backing
3470 this has been removed. Distributions which want to retain
3471 compatibility with this should carry the burden for
3472 supporting this themselves and patch support for these back
3473 in, if they really need to. Also, the facilities $syslog and
3474 $local_fs are now ignored, since systemd does not support
3475 early-boot LSB init scripts anymore, and these facilities
3476 are implied anyway for normal services. syslog.target has
3479 * There are new bus calls on PID1's Manager object for
3480 cancelling jobs, and removing snapshot units. Previously,
3481 both calls were only available on the Job and Snapshot
3484 * systemd-journal-gatewayd gained SSL support.
3486 * The various "environment" files, such as /etc/locale.conf
3487 now support continuation lines with a backslash ("\") as
3488 last character in the line, similarly in style (but different)
3489 to how this is supported in shells.
3491 * For normal user processes the _SYSTEMD_USER_UNIT= field is
3492 now implicitly appended to every log entry logged. systemctl
3493 has been updated to filter by this field when operating on a
3494 user systemd instance.
3496 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
3497 CAP_AUDIT_CONTROL capabilities to the capabilities set for
3498 the container. This makes it easier to boot unmodified
3499 Fedora systems in a container, which however still requires
3500 audit=0 to be passed on the kernel command line. Auditing in
3501 kernel and userspace is unfortunately still too broken in
3502 context of containers, hence we recommend compiling it out
3503 of the kernel or using audit=0. Hopefully this will be fixed
3504 one day for good in the kernel.
3506 * nspawn gained the new --bind= and --bind-ro= parameters to
3507 bind mount specific directories from the host into the
3510 * nspawn will now mount its own devpts file system instance
3511 into the container, in order not to leak pty devices from
3512 the host into the container.
3514 * systemd will now read the firmware boot time performance
3515 information from the EFI variables, if the used boot loader
3516 supports this, and takes it into account for boot performance
3517 analysis via "systemd-analyze". This is currently supported
3518 only in conjunction with Gummiboot, but could be supported
3519 by other boot loaders too. For details see:
3521 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
3523 * A new generator has been added that automatically mounts the
3524 EFI System Partition (ESP) to /boot, if that directory
3525 exists, is empty, and no other file system has been
3526 configured to be mounted there.
3528 * logind will now send out PrepareForSleep(false) out
3529 unconditionally, after coming back from suspend. This may be
3530 used by applications as asynchronous notification for
3531 system resume events.
3533 * "systemctl unlock-sessions" has been added, that allows
3534 unlocking the screens of all user sessions at once, similar
3535 to how "systemctl lock-sessions" already locked all users
3536 sessions. This is backed by a new D-Bus call UnlockSessions().
3538 * "loginctl seat-status" will now show the master device of a
3539 seat. (i.e. the device of a seat that needs to be around for
3540 the seat to be considered available, usually the graphics
3543 * tmpfiles gained a new "X" line type, that allows
3544 configuration of files and directories (with wildcards) that
3545 shall be excluded from automatic cleanup ("aging").
3547 * udev default rules set the device node permissions now only
3548 at "add" events, and do not change them any longer with a
3549 later "change" event.
3551 * The log messages for lid events and power/sleep keypresses
3552 now carry a message ID.
3554 * We now have a substantially larger unit test suite, but this
3555 continues to be work in progress.
3557 * udevadm hwdb gained a new --root= parameter to change the
3558 root directory to operate relative to.
3560 * logind will now issue a background sync() request to the kernel
3561 early at shutdown, so that dirty buffers are flushed to disk early
3562 instead of at the last moment, in order to optimize shutdown
3565 * A new bootctl tool has been added that is an interface for
3566 certain boot loader operations. This is currently a preview
3567 and is likely to be extended into a small mechanism daemon
3568 like timedated, localed, hostnamed, and can be used by
3569 graphical UIs to enumerate available boot options, and
3570 request boot into firmware operations.
3572 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
3573 the rest of the package. It also has been updated to work
3574 correctly in initrds.
3576 * Policykit previously has been runtime optional, and is now
3577 also compile time optional via a configure switch.
3579 * systemd-analyze has been reimplemented in C. Also "systemctl
3580 dot" has moved into systemd-analyze.
3582 * "systemctl status" with no further parameters will now print
3583 the status of all active or failed units.
3585 * Operations such as "systemctl start" can now be executed
3586 with a new mode "--irreversible" which may be used to queue
3587 operations that cannot accidentally be reversed by a later
3588 job queuing. This is by default used to make shutdown
3589 requests more robust.
3591 * The Python API of systemd now gained a new module for
3592 reading journal files.
3594 * A new tool kernel-install has been added that can install
3595 kernel images according to the Boot Loader Specification:
3597 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
3599 * Boot time console output has been improved to provide
3600 animated boot time output for hanging jobs.
3602 * A new tool systemd-activate has been added which can be used
3603 to test socket activation with, directly from the command
3604 line. This should make it much easier to test and debug
3605 socket activation in daemons.
3607 * journalctl gained a new "--reverse" (or -r) option to show
3608 journal output in reverse order (i.e. newest line first).
3610 * journalctl gained a new "--pager-end" (or -e) option to jump
3611 to immediately jump to the end of the journal in the
3612 pager. This is only supported in conjunction with "less".
3614 * journalctl gained a new "--user-unit=" option, that works
3615 similarly to "--unit=" but filters for user units rather than
3618 * A number of unit files to ease adoption of systemd in
3619 initrds has been added. This moves some minimal logic from
3620 the various initrd implementations into systemd proper.
3622 * The journal files are now owned by a new group
3623 "systemd-journal", which exists specifically to allow access
3624 to the journal, and nothing else. Previously, we used the
3625 "adm" group for that, which however possibly covers more
3626 than just journal/log file access. This new group is now
3627 already used by systemd-journal-gatewayd to ensure this
3628 daemon gets access to the journal files and as little else
3629 as possible. Note that "make install" will also set FS ACLs
3630 up for /var/log/journal to give "adm" and "wheel" read
3631 access to it, in addition to "systemd-journal" which owns
3632 the journal files. We recommend that packaging scripts also
3633 add read access to "adm" + "wheel" to /var/log/journal, and
3634 all existing/future journal files. To normal users and
3635 administrators little changes, however packagers need to
3636 ensure to create the "systemd-journal" system group at
3637 package installation time.
3639 * The systemd-journal-gatewayd now runs as unprivileged user
3640 systemd-journal-gateway:systemd-journal-gateway. Packaging
3641 scripts need to create these system user/group at
3644 * timedated now exposes a new boolean property CanNTP that
3645 indicates whether a local NTP service is available or not.
3647 * systemd-detect-virt will now also detect xen PVs
3649 * The pstore file system is now mounted by default, if it is
3652 * In addition to the SELinux and IMA policies we will now also
3653 load SMACK policies at early boot.
3655 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
3656 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
3657 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
3658 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
3659 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
3660 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
3661 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
3662 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
3663 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
3664 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
3665 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
3666 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
3667 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
3668 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
3672 * Timer units now support calendar time events in addition to
3673 monotonic time events. That means you can now trigger a unit
3674 based on a calendar time specification such as "Thu,Fri
3675 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
3676 or fifth day of any month of the year 2013, given that it is
3677 a thursday or friday. This brings timer event support
3678 considerably closer to cron's capabilities. For details on
3679 the supported calendar time specification language see
3682 * udev now supports a number of different naming policies for
3683 network interfaces for predictable names, and a combination
3684 of these policies is now the default. Please see this wiki
3685 document for details:
3687 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
3689 * Auke Kok's bootchart implementation has been added to the
3690 systemd tree. It is an optional component that can graph the
3691 boot in quite some detail. It is one of the best bootchart
3692 implementations around and minimal in its code and
3695 * nss-myhostname has been integrated into the systemd source
3696 tree. nss-myhostname guarantees that the local hostname
3697 always stays resolvable via NSS. It has been a weak
3698 requirement of systemd-hostnamed since a long time, and
3699 since its code is actually trivial we decided to just
3700 include it in systemd's source tree. It can be turned off
3701 with a configure switch.
3703 * The read-ahead logic is now capable of properly detecting
3704 whether a btrfs file system is on SSD or rotating media, in
3705 order to optimize the read-ahead scheme. Previously, it was
3706 only capable of detecting this on traditional file systems
3709 * In udev, additional device properties are now read from the
3710 IAB in addition to the OUI database. Also, Bluetooth company
3711 identities are attached to the devices as well.
3713 * In service files %U may be used as specifier that is
3714 replaced by the configured user name of the service.
3716 * nspawn may now be invoked without a controlling TTY. This
3717 makes it suitable for invocation as its own service. This
3718 may be used to set up a simple containerized server system
3719 using only core OS tools.
3721 * systemd and nspawn can now accept socket file descriptors
3722 when they are started for socket activation. This enables
3723 implementation of socket activated nspawn
3724 containers. i.e. think about autospawning an entire OS image
3725 when the first SSH or HTTP connection is received. We expect
3726 that similar functionality will also be added to libvirt-lxc
3729 * journalctl will now suppress ANSI color codes when
3730 presenting log data.
3732 * systemctl will no longer show control group information for
3733 a unit if a the control group is empty anyway.
3735 * logind can now automatically suspend/hibernate/shutdown the
3738 * /etc/machine-info and hostnamed now also expose the chassis
3739 type of the system. This can be used to determine whether
3740 the local system is a laptop, desktop, handset or
3741 tablet. This information may either be configured by the
3742 user/vendor or is automatically determined from ACPI and DMI
3743 information if possible.
3745 * A number of PolicyKit actions are now bound together with
3746 "imply" rules. This should simplify creating UIs because
3747 many actions will now authenticate similar ones as well.
3749 * Unit files learnt a new condition ConditionACPower= which
3750 may be used to conditionalize a unit depending on whether an
3751 AC power source is connected or not, of whether the system
3752 is running on battery power.
3754 * systemctl gained a new "is-failed" verb that may be used in
3755 shell scripts and suchlike to check whether a specific unit
3756 is in the "failed" state.
3758 * The EnvironmentFile= setting in unit files now supports file
3759 globbing, and can hence be used to easily read a number of
3760 environment files at once.
3762 * systemd will no longer detect and recognize specific
3763 distributions. All distribution-specific #ifdeffery has been
3764 removed, systemd is now fully generic and
3765 distribution-agnostic. Effectively, not too much is lost as
3766 a lot of the code is still accessible via explicit configure
3767 switches. However, support for some distribution specific
3768 legacy configuration file formats has been dropped. We
3769 recommend distributions to simply adopt the configuration
3770 files everybody else uses now and convert the old
3771 configuration from packaging scripts. Most distributions
3772 already did that. If that's not possible or desirable,
3773 distributions are welcome to forward port the specific
3774 pieces of code locally from the git history.
3776 * When logging a message about a unit systemd will now always
3777 log the unit name in the message meta data.
3779 * localectl will now also discover system locale data that is
3780 not stored in locale archives, but directly unpacked.
3782 * logind will no longer unconditionally use framebuffer
3783 devices as seat masters, i.e. as devices that are required
3784 to be existing before a seat is considered preset. Instead,
3785 it will now look for all devices that are tagged as
3786 "seat-master" in udev. By default framebuffer devices will
3787 be marked as such, but depending on local systems other
3788 devices might be marked as well. This may be used to
3789 integrate graphics cards using closed source drivers (such
3790 as NVidia ones) more nicely into logind. Note however, that
3791 we recommend using the open source NVidia drivers instead,
3792 and no udev rules for the closed-source drivers will be
3793 shipped from us upstream.
3795 Contributions from: Adam Williamson, Alessandro Crismani, Auke
3796 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
3797 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
3798 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
3799 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
3800 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
3801 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
3802 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
3803 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
3804 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
3805 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
3806 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
3811 * udev gained support for loading additional device properties
3812 from an indexed database that is keyed by vendor/product IDs
3813 and similar device identifiers. For the beginning this
3814 "hwdb" is populated with data from the well-known PCI and
3815 USB database, but also includes PNP, ACPI and OID data. In
3816 the longer run this indexed database shall grow into
3817 becoming the one central database for non-essential
3818 userspace device metadata. Previously, data from the PCI/USB
3819 database was only attached to select devices, since the
3820 lookup was a relatively expensive operation due to O(n) time
3821 complexity (with n being the number of entries in the
3822 database). Since this is now O(1), we decided to add in this
3823 data for all devices where this is available, by
3824 default. Note that the indexed database needs to be rebuilt
3825 when new data files are installed. To achieve this you need
3826 to update your packaging scripts to invoke "udevadm hwdb
3827 --update" after installation of hwdb data files. For
3828 RPM-based distributions we introduced the new
3829 %udev_hwdb_update macro for this purpose.
3831 * The Journal gained support for the "Message Catalog", an
3832 indexed database to link up additional information with
3833 journal entries. For further details please check:
3835 http://www.freedesktop.org/wiki/Software/systemd/catalog
3837 The indexed message catalog database also needs to be
3838 rebuilt after installation of message catalog files. Use
3839 "journalctl --update-catalog" for this. For RPM-based
3840 distributions we introduced the %journal_catalog_update
3841 macro for this purpose.
3843 * The Python Journal bindings gained support for the standard
3844 Python logging framework.
3846 * The Journal API gained new functions for checking whether
3847 the underlying file system of a journal file is capable of
3848 properly reporting file change notifications, or whether
3849 applications that want to reflect journal changes "live"
3850 need to recheck journal files continuously in appropriate
3853 * It is now possible to set the "age" field for tmpfiles
3854 entries to 0, indicating that files matching this entry
3855 shall always be removed when the directories are cleaned up.
3857 * coredumpctl gained a new "gdb" verb which invokes gdb
3858 right-away on the selected coredump.
3860 * There's now support for "hybrid sleep" on kernels that
3861 support this, in addition to "suspend" and "hibernate". Use
3862 "systemctl hybrid-sleep" to make use of this.
3864 * logind's HandleSuspendKey= setting (and related settings)
3865 now gained support for a new "lock" setting to simply
3866 request the screen lock on all local sessions, instead of
3867 actually executing a suspend or hibernation.
3869 * systemd will now mount the EFI variables file system by
3872 * Socket units now gained support for configuration of the
3873 SMACK security label.
3875 * timedatectl will now output the time of the last and next
3876 daylight saving change.
3878 * We dropped support for various legacy and distro-specific
3879 concepts, such as insserv, early-boot SysV services
3880 (i.e. those for non-standard runlevels such as 'b' or 'S')
3881 or ArchLinux /etc/rc.conf support. We recommend the
3882 distributions who still need support this to either continue
3883 to maintain the necessary patches downstream, or find a
3884 different solution. (Talk to us if you have questions!)
3886 * Various systemd components will now bypass PolicyKit checks
3887 for root and otherwise handle properly if PolicyKit is not
3888 found to be around. This should fix most issues for
3889 PolicyKit-less systems. Quite frankly this should have been
3890 this way since day one. It is absolutely our intention to
3891 make systemd work fine on PolicyKit-less systems, and we
3892 consider it a bug if something does not work as it should if
3893 PolicyKit is not around.
3895 * For embedded systems it is now possible to build udev and
3896 systemd without blkid and/or kmod support.
3898 * "systemctl switch-root" is now capable of switching root
3899 more than once. I.e. in addition to transitions from the
3900 initrd to the host OS it is now possible to transition to
3901 further OS images from the host. This is useful to implement
3902 offline updating tools.
3904 * Various other additions have been made to the RPM macros
3905 shipped with systemd. Use %udev_rules_update() after
3906 installing new udev rules files. %_udevhwdbdir,
3907 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
3908 %_sysctldir are now available which resolve to the right
3909 directories for packages to place various data files in.
3911 * journalctl gained the new --full switch (in addition to
3912 --all, to disable ellipsation for long messages.
3914 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
3915 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
3916 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
3917 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
3918 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
3919 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
3920 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
3921 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
3922 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
3926 * journalctl gained new --since= and --until= switches to
3927 filter by time. It also now supports nice filtering for
3928 units via --unit=/-u.
3930 * Type=oneshot services may use ExecReload= and do the
3933 * The journal daemon now supports time-based rotation and
3934 vacuuming, in addition to the usual disk-space based
3937 * The journal will now index the available field values for
3938 each field name. This enables clients to show pretty drop
3939 downs of available match values when filtering. The bash
3940 completion of journalctl has been updated
3941 accordingly. journalctl gained a new switch -F to list all
3942 values a certain field takes in the journal database.
3944 * More service events are now written as structured messages
3945 to the journal, and made recognizable via message IDs.
3947 * The timedated, localed and hostnamed mini-services which
3948 previously only provided support for changing time, locale
3949 and hostname settings from graphical DEs such as GNOME now
3950 also have a minimal (but very useful) text-based client
3951 utility each. This is probably the nicest way to changing
3952 these settings from the command line now, especially since
3953 it lists available options and is fully integrated with bash
3956 * There's now a new tool "systemd-coredumpctl" to list and
3957 extract coredumps from the journal.
3959 * We now install a README each in /var/log/ and
3960 /etc/rc.d/init.d explaining where the system logs and init
3961 scripts went. This hopefully should help folks who go to
3962 that dirs and look into the otherwise now empty void and
3963 scratch their heads.
3965 * When user-services are invoked (by systemd --user) the
3966 $MANAGERPID env var is set to the PID of systemd.
3968 * SIGRTMIN+24 when sent to a --user instance will now result
3969 in immediate termination of systemd.
3971 * gatewayd received numerous feature additions such as a
3972 "follow" mode, for live syncing and filtering.
3974 * browse.html now allows filtering and showing detailed
3975 information on specific entries. Keyboard navigation and
3976 mouse screen support has been added.
3978 * gatewayd/journalctl now supports HTML5/JSON
3979 Server-Sent-Events as output.
3981 * The SysV init script compatibility logic will now
3982 heuristically determine whether a script supports the
3983 "reload" verb, and only then make this available as
3986 * "systemctl status --follow" has been removed, use "journalctl
3989 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
3990 have been removed since they are hardly useful to be
3993 * And I'd like to take the opportunity to specifically mention
3994 Zbigniew for his great contributions. Zbigniew, you rock!
3996 Contributions from: Andrew Eikum, Christian Hesse, Colin
3997 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
3998 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
3999 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
4000 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
4001 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
4002 Jędrzejewski-Szmek, Сковорода Никита Андреевич
4006 * If /etc/vconsole.conf is non-existent or empty we will no
4007 longer load any console font or key map at boot by
4008 default. Instead the kernel defaults will be left
4009 intact. This is definitely the right thing to do, as no
4010 configuration should mean no configuration, and hard-coding
4011 font names that are different on all archs is probably a bad
4012 idea. Also, the kernel default key map and font should be
4013 good enough for most cases anyway, and mostly identical to
4014 the userspace fonts/key maps we previously overloaded them
4015 with. If distributions want to continue to default to a
4016 non-kernel font or key map they should ship a default
4017 /etc/vconsole.conf with the appropriate contents.
4019 Contributions from: Colin Walters, Daniel J Walsh, Dave
4020 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
4021 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
4025 * journalctl gained a new --cursor= switch to show entries
4026 starting from the specified location in the journal.
4028 * We now enforce a size limit on journal entry fields exported
4029 with "-o json" in journalctl. Fields larger than 4K will be
4030 assigned null. This can be turned off with --all.
4032 * An (optional) journal gateway daemon is now available as
4033 "systemd-journal-gatewayd.service". This service provides
4034 access to the journal via HTTP and JSON. This functionality
4035 will be used to implement live log synchronization in both
4036 pull and push modes, but has various other users too, such
4037 as easy log access for debugging of embedded devices. Right
4038 now it is already useful to retrieve the journal via HTTP:
4040 # systemctl start systemd-journal-gatewayd.service
4041 # wget http://localhost:19531/entries
4043 This will download the journal contents in a
4044 /var/log/messages compatible format. The same as JSON:
4046 # curl -H"Accept: application/json" http://localhost:19531/entries
4048 This service is also accessible via a web browser where a
4049 single static HTML5 app is served that uses the JSON logic
4050 to enable the user to do some basic browsing of the
4051 journal. This will be extended later on. Here's an example
4052 screenshot of this app in its current state:
4054 http://0pointer.de/public/journal-gatewayd
4056 Contributions from: Kay Sievers, Lennart Poettering, Robert
4057 Milasan, Tom Gundersen
4061 * The bash completion logic is now available for journalctl
4064 * We do not mount the "cpuset" controller anymore together with
4065 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
4066 started if no parameters are assigned to it. "cpuset" hence
4067 broke code that assumed it it could create "cpu" groups and
4070 * journalctl -f will now subscribe to terminal size changes,
4071 and line break accordingly.
4073 Contributions from: Dave Reisner, Kay Sievers, Lennart
4074 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
4078 * nspawn will now create a symlink /etc/localtime in the
4079 container environment, copying the host's timezone
4080 setting. Previously this has been done via a bind mount, but
4081 since symlinks cannot be bind mounted this has now been
4082 changed to create/update the appropriate symlink.
4084 * journalctl -n's line number argument is now optional, and
4085 will default to 10 if omitted.
4087 * journald will now log the maximum size the journal files may
4088 take up on disk. This is particularly useful if the default
4089 built-in logic of determining this parameter from the file
4090 system size is used. Use "systemctl status
4091 systemd-journald.service" to see this information.
4093 * The multi-seat X wrapper tool has been stripped down. As X
4094 is now capable of enumerating graphics devices via udev in a
4095 seat-aware way the wrapper is not strictly necessary
4096 anymore. A stripped down temporary stop-gap is still shipped
4097 until the upstream display managers have been updated to
4098 fully support the new X logic. Expect this wrapper to be
4099 removed entirely in one of the next releases.
4101 * HandleSleepKey= in logind.conf has been split up into
4102 HandleSuspendKey= and HandleHibernateKey=. The old setting
4103 is not available anymore. X11 and the kernel are
4104 distinguishing between these keys and we should too. This
4105 also means the inhibition lock for these keys has been split
4108 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
4109 Poettering, Lukas Nykryn, Václav Pavlín
4113 * Whenever a unit changes state we will now log this to the
4114 journal and show along the unit's own log output in
4117 * ConditionPathIsMountPoint= can now properly detect bind
4118 mount points too. (Previously, a bind mount of one file
4119 system to another place in the same file system could not be
4120 detected as mount, since they shared struct stat's st_dev
4123 * We will now mount the cgroup controllers cpu, cpuacct,
4124 cpuset and the controllers net_cls, net_prio together by
4127 * nspawn containers will now have a virtualized boot
4128 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
4129 over with a randomized ID at container initialization). This
4130 has the effect of making "journalctl -b" do the right thing
4133 * The JSON output journal serialization has been updated not
4134 to generate "endless" list objects anymore, but rather one
4135 JSON object per line. This is more in line how most JSON
4136 parsers expect JSON objects. The new output mode
4137 "json-pretty" has been added to provide similar output, but
4138 neatly aligned for readability by humans.
4140 * We dropped all explicit sync() invocations in the shutdown
4141 code. The kernel does this implicitly anyway in the kernel
4142 reboot() syscall. halt(8)'s -n option is now a compatibility
4145 * We now support virtualized reboot() in containers, as
4146 supported by newer kernels. We will fall back to exit() if
4147 CAP_SYS_REBOOT is not available to the container. Also,
4148 nspawn makes use of this now and will actually reboot the
4149 container if the containerized OS asks for that.
4151 * journalctl will only show local log output by default
4152 now. Use --merge (-m) to show remote log output, too.
4154 * libsystemd-journal gained the new sd_journal_get_usage()
4155 call to determine the current disk usage of all journal
4156 files. This is exposed in the new "journalctl --disk-usage"
4159 * journald gained a new configuration setting SplitMode= in
4160 journald.conf which may be used to control how user journals
4161 are split off. See journald.conf(5) for details.
4163 * A new condition type ConditionFileNotEmpty= has been added.
4165 * tmpfiles' "w" lines now support file globbing, to write
4166 multiple files at once.
4168 * We added Python bindings for the journal submission
4169 APIs. More Python APIs for a number of selected APIs will
4170 likely follow. Note that we intend to add native bindings
4171 only for the Python language, as we consider it common
4172 enough to deserve bindings shipped within systemd. There are
4173 various projects outside of systemd that provide bindings
4174 for languages such as PHP or Lua.
4176 * Many conditions will now resolve specifiers such as %i. In
4177 addition, PathChanged= and related directives of .path units
4178 now support specifiers as well.
4180 * There's now a new RPM macro definition for the system preset
4183 * journald will now warn if it ca not forward a message to the
4184 syslog daemon because its socket is full.
4186 * timedated will no longer write or process /etc/timezone,
4187 except on Debian. As we do not support late mounted /usr
4188 anymore /etc/localtime always being a symlink is now safe,
4189 and hence the information in /etc/timezone is not necessary
4192 * logind will now always reserve one VT for a text getty (VT6
4193 by default). Previously if more than 6 X sessions where
4194 started they took up all the VTs with auto-spawned gettys,
4195 so that no text gettys were available anymore.
4197 * udev will now automatically inform the btrfs kernel logic
4198 about btrfs RAID components showing up. This should make
4199 simple hotplug based btrfs RAID assembly work.
4201 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
4202 (but not for its children which will stay at the kernel
4203 default). This should allow setups with a lot more listening
4206 * systemd will now always pass the configured timezone to the
4207 kernel at boot. timedated will do the same when the timezone
4210 * logind's inhibition logic has been updated. By default,
4211 logind will now handle the lid switch, the power and sleep
4212 keys all the time, even in graphical sessions. If DEs want
4213 to handle these events on their own they should take the new
4214 handle-power-key, handle-sleep-key and handle-lid-switch
4215 inhibitors during their runtime. A simple way to achieve
4216 that is to invoke the DE wrapped in an invocation of:
4218 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
4220 * Access to unit operations is now checked via SELinux taking
4221 the unit file label and client process label into account.
4223 * systemd will now notify the administrator in the journal
4224 when he over-mounts a non-empty directory.
4226 * There are new specifiers that are resolved in unit files,
4227 for the host name (%H), the machine ID (%m) and the boot ID
4230 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
4231 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
4232 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
4233 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
4234 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
4235 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
4236 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
4240 * Support for reading structured kernel messages from
4241 /dev/kmsg has now been added and is enabled by default.
4243 * Support for reading kernel messages from /proc/kmsg has now
4244 been removed. If you want kernel messages in the journal
4245 make sure to run a recent kernel (>= 3.5) that supports
4246 reading structured messages from /dev/kmsg (see
4247 above). /proc/kmsg is now exclusive property of classic
4248 syslog daemons again.
4250 * The libudev API gained the new
4251 udev_device_new_from_device_id() call.
4253 * The logic for file system namespace (ReadOnlyDirectory=,
4254 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
4255 require pivot_root() anymore. This means fewer temporary
4256 directories are created below /tmp for this feature.
4258 * nspawn containers will now see and receive all submounts
4259 made on the host OS below the root file system of the
4262 * Forward Secure Sealing is now supported for Journal files,
4263 which provide cryptographical sealing of journal files so
4264 that attackers cannot alter log history anymore without this
4265 being detectable. Lennart will soon post a blog story about
4266 this explaining it in more detail.
4268 * There are two new service settings RestartPreventExitStatus=
4269 and SuccessExitStatus= which allow configuration of exit
4270 status (exit code or signal) which will be excepted from the
4271 restart logic, resp. consider successful.
4273 * journalctl gained the new --verify switch that can be used
4274 to check the integrity of the structure of journal files and
4275 (if Forward Secure Sealing is enabled) the contents of
4278 * nspawn containers will now be run with /dev/stdin, /dev/fd/
4279 and similar symlinks pre-created. This makes running shells
4280 as container init process a lot more fun.
4282 * The fstab support can now handle PARTUUID= and PARTLABEL=
4285 * A new ConditionHost= condition has been added to match
4286 against the hostname (with globs) and machine ID. This is
4287 useful for clusters where a single OS image is used to
4288 provision a large number of hosts which shall run slightly
4289 different sets of services.
4291 * Services which hit the restart limit will now be placed in a
4294 Contributions from: Bertram Poettering, Dave Reisner, Huang
4295 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
4296 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
4300 * When running in --user mode systemd will now become a
4301 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
4302 tree a lot more organized.
4304 * A new PartOf= unit dependency type has been introduced that
4305 may be used to group services in a natural way.
4307 * "systemctl enable" may now be used to enable instances of
4310 * journalctl now prints error log levels in red, and
4311 warning/notice log levels in bright white. It also supports
4312 filtering by log level now.
4314 * cgtop gained a new -n switch (similar to top), to configure
4315 the maximum number of iterations to run for. It also gained
4316 -b, to run in batch mode (accepting no input).
4318 * The suffix ".service" may now be omitted on most systemctl
4319 command lines involving service unit names.
4321 * There's a new bus call in logind to lock all sessions, as
4322 well as a loginctl verb for it "lock-sessions".
4324 * libsystemd-logind.so gained a new call sd_journal_perror()
4325 that works similar to libc perror() but logs to the journal
4326 and encodes structured information about the error number.
4328 * /etc/crypttab entries now understand the new keyfile-size=
4331 * shutdown(8) now can send a (configurable) wall message when
4332 a shutdown is cancelled.
4334 * The mount propagation mode for the root file system will now
4335 default to "shared", which is useful to make containers work
4336 nicely out-of-the-box so that they receive new mounts from
4337 the host. This can be undone locally by running "mount
4338 --make-rprivate /" if needed.
4340 * The prefdm.service file has been removed. Distributions
4341 should maintain this unit downstream if they intend to keep
4342 it around. However, we recommend writing normal unit files
4343 for display managers instead.
4345 * Since systemd is a crucial part of the OS we will now
4346 default to a number of compiler switches that improve
4347 security (hardening) such as read-only relocations, stack
4348 protection, and suchlike.
4350 * The TimeoutSec= setting for services is now split into
4351 TimeoutStartSec= and TimeoutStopSec= to allow configuration
4352 of individual time outs for the start and the stop phase of
4355 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
4356 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
4357 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
4358 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
4359 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
4360 Gundersen, Zbigniew Jędrzejewski-Szmek
4364 * The journal and id128 C APIs are now fully documented as man
4367 * Extra safety checks have been added when transitioning from
4368 the initial RAM disk to the main system to avoid accidental
4371 * /etc/crypttab entries now understand the new keyfile-offset=
4374 * systemctl -t can now be used to filter by unit load state.
4376 * The journal C API gained the new sd_journal_wait() call to
4377 make writing synchronous journal clients easier.
4379 * journalctl gained the new -D switch to show journals from a
4382 * journalctl now displays a special marker between log
4383 messages of two different boots.
4385 * The journal is now explicitly flushed to /var via a service
4386 systemd-journal-flush.service, rather than implicitly simply
4387 by seeing /var/log/journal to be writable.
4389 * journalctl (and the journal C APIs) can now match for much
4390 more complex expressions, with alternatives and
4393 * When transitioning from the initial RAM disk to the main
4394 system we will now kill all processes in a killing spree to
4395 ensure no processes stay around by accident.
4397 * Three new specifiers may be used in unit files: %u, %h, %s
4398 resolve to the user name, user home directory resp. user
4399 shell. This is useful for running systemd user instances.
4401 * We now automatically rotate journal files if their data
4402 object hash table gets a fill level > 75%. We also size the
4403 hash table based on the configured maximum file size. This
4404 together should lower hash collisions drastically and thus
4405 speed things up a bit.
4407 * journalctl gained the new "--header" switch to introspect
4408 header data of journal files.
4410 * A new setting SystemCallFilters= has been added to services
4411 which may be used to apply blacklists or whitelists to
4412 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
4414 * nspawn gained a new --link-journal= switch (and quicker: -j)
4415 to link the container journal with the host. This makes it
4416 very easy to centralize log viewing on the host for all
4417 guests while still keeping the journal files separated.
4419 * Many bugfixes and optimizations
4421 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
4422 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
4423 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
4428 * Several tools now understand kernel command line arguments,
4429 which are only read when run in an initial RAM disk. They
4430 usually follow closely their normal counterparts, but are
4433 * There's a new tool to analyze the readahead files that are
4434 automatically generated at boot. Use:
4436 /usr/lib/systemd/systemd-readahead analyze /.readahead
4438 * We now provide an early debug shell on tty9 if this enabled. Use:
4440 systemctl enable debug-shell.service
4442 * All plymouth related units have been moved into the Plymouth
4443 package. Please make sure to upgrade your Plymouth version
4446 * systemd-tmpfiles now supports getting passed the basename of
4447 a configuration file only, in which case it will look for it
4448 in all appropriate directories automatically.
4450 * udevadm info now takes a /dev or /sys path as argument, and
4451 does the right thing. Example:
4453 udevadm info /dev/sda
4454 udevadm info /sys/class/block/sda
4456 * systemctl now prints a warning if a unit is stopped but a
4457 unit that might trigger it continues to run. Example: a
4458 service is stopped but the socket that activates it is left
4461 * "systemctl status" will now mention if the log output was
4462 shortened due to rotation since a service has been started.
4464 * The journal API now exposes functions to determine the
4465 "cutoff" times due to rotation.
4467 * journald now understands SIGUSR1 and SIGUSR2 for triggering
4468 immediately flushing of runtime logs to /var if possible,
4469 resp. for triggering immediate rotation of the journal
4472 * It is now considered an error if a service is attempted to
4473 be stopped that is not loaded.
4475 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
4477 * systemd-analyze now supports Python 3
4479 * tmpfiles now supports cleaning up directories via aging
4480 where the first level dirs are always kept around but
4481 directories beneath it automatically aged. This is enabled
4482 by prefixing the age field with '~'.
4484 * Seat objects now expose CanGraphical, CanTTY properties
4485 which is required to deal with very fast bootups where the
4486 display manager might be running before the graphics drivers
4487 completed initialization.
4489 * Seat objects now expose a State property.
4491 * We now include RPM macros for service enabling/disabling
4492 based on the preset logic. We recommend RPM based
4493 distributions to make use of these macros if possible. This
4494 makes it simpler to reuse RPM spec files across
4497 * We now make sure that the collected systemd unit name is
4498 always valid when services log to the journal via
4501 * There's a new man page kernel-command-line(7) detailing all
4502 command line options we understand.
4504 * The fstab generator may now be disabled at boot by passing
4505 fstab=0 on the kernel command line.
4507 * A new kernel command line option modules-load= is now understood
4508 to load a specific kernel module statically, early at boot.
4510 * Unit names specified on the systemctl command line are now
4511 automatically escaped as needed. Also, if file system or
4512 device paths are specified they are automatically turned
4513 into the appropriate mount or device unit names. Example:
4515 systemctl status /home
4516 systemctl status /dev/sda
4518 * The SysVConsole= configuration option has been removed from
4519 system.conf parsing.
4521 * The SysV search path is no longer exported on the D-Bus
4524 * The Names= option is been removed from unit file parsing.
4526 * There's a new man page bootup(7) detailing the boot process.
4528 * Every unit and every generator we ship with systemd now
4529 comes with full documentation. The self-explanatory boot is
4532 * A couple of services gained "systemd-" prefixes in their
4533 name if they wrap systemd code, rather than only external
4534 code. Among them fsck@.service which is now
4535 systemd-fsck@.service.
4537 * The HaveWatchdog property has been removed from the D-Bus
4540 * systemd.confirm_spawn= on the kernel command line should now
4543 * There's a new man page crypttab(5) which details all options
4544 we actually understand.
4546 * systemd-nspawn gained a new --capability= switch to pass
4547 additional capabilities to the container.
4549 * timedated will now read known NTP implementation unit names
4550 from /usr/lib/systemd/ntp-units.d/*.list,
4551 systemd-timedated-ntp.target has been removed.
4553 * journalctl gained a new switch "-b" that lists log data of
4554 the current boot only.
4556 * The notify socket is in the abstract namespace again, in
4557 order to support daemons which chroot() at start-up.
4559 * There is a new Storage= configuration option for journald
4560 which allows configuration of where log data should go. This
4561 also provides a way to disable journal logging entirely, so
4562 that data collected is only forwarded to the console, the
4563 kernel log buffer or another syslog implementation.
4565 * Many bugfixes and optimizations
4567 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
4568 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
4569 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
4570 Shawn Landden, Tom Gundersen
4574 * "systemctl help <unit>" now shows the man page if one is
4577 * Several new man pages have been added.
4579 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
4580 MaxLevelConsole= can now be specified in
4581 journald.conf. These options allow reducing the amount of
4582 data stored on disk or forwarded by the log level.
4584 * TimerSlackNSec= can now be specified in system.conf for
4585 PID1. This allows system-wide power savings.
4587 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
4588 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
4593 * logind is now capable of (optionally) handling power and
4594 sleep keys as well as the lid switch.
4596 * journalctl now understands the syntax "journalctl
4597 /usr/bin/avahi-daemon" to get all log output of a specific
4600 * CapabilityBoundingSet= in system.conf now also influences
4601 the capability bound set of usermode helpers of the kernel.
4603 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
4604 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
4605 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
4606 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
4610 * Note that we skipped 139 releases here in order to set the
4611 new version to something that is greater than both udev's
4612 and systemd's most recent version number.
4614 * udev: all udev sources are merged into the systemd source tree now.
4615 All future udev development will happen in the systemd tree. It
4616 is still fully supported to use the udev daemon and tools without
4617 systemd running, like in initramfs or other init systems. Building
4618 udev though, will require the *build* of the systemd tree, but
4619 udev can be properly *run* without systemd.
4621 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
4622 should be used to create dead device nodes as workarounds for broken
4625 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
4626 no longer supported. udev_monitor_new_from_netlink() needs to be
4627 used to subscribe to events.
4629 * udev: when udevd is started by systemd, processes which are left
4630 behind by forking them off of udev rules, are unconditionally cleaned
4631 up and killed now after the event handling has finished. Services or
4632 daemons must be started as systemd services. Services can be
4633 pulled-in by udev to get started, but they can no longer be directly
4634 forked by udev rules.
4636 * udev: the daemon binary is called systemd-udevd now and installed
4637 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
4638 to adapt to that, create symlink, or rename the binary after building
4641 * libudev no longer provides these symbols:
4642 udev_monitor_from_socket()
4643 udev_queue_get_failed_list_entry()
4644 udev_get_{dev,sys,run}_path()
4645 The versions number was bumped and symbol versioning introduced.
4647 * systemd-loginctl and systemd-journalctl have been renamed
4648 to loginctl and journalctl to match systemctl.
4650 * The config files: /etc/systemd/systemd-logind.conf and
4651 /etc/systemd/systemd-journald.conf have been renamed to
4652 logind.conf and journald.conf. Package updates should rename
4653 the files to the new names on upgrade.
4655 * For almost all files the license is now LGPL2.1+, changed
4656 from the previous GPL2.0+. Exceptions are some minor stuff
4657 of udev (which will be changed to LGPL2.1 eventually, too),
4658 and the MIT licensed sd-daemon.[ch] library that is suitable
4659 to be used as drop-in files.
4661 * systemd and logind now handle system sleep states, in
4662 particular suspending and hibernating.
4664 * logind now implements a sleep/shutdown/idle inhibiting logic
4665 suitable for a variety of uses. Soonishly Lennart will blog
4666 about this in more detail.
4668 * var-run.mount and var-lock.mount are no longer provided
4669 (which prevously bind mounted these directories to their new
4670 places). Distributions which have not converted these
4671 directories to symlinks should consider stealing these files
4672 from git history and add them downstream.
4674 * We introduced the Documentation= field for units and added
4675 this to all our shipped units. This is useful to make it
4676 easier to explore the boot and the purpose of the various
4679 * All smaller setup units (such as
4680 systemd-vconsole-setup.service) now detect properly if they
4681 are run in a container and are skipped when
4682 appropriate. This guarantees an entirely noise-free boot in
4683 Linux container environments such as systemd-nspawn.
4685 * A framework for implementing offline system updates is now
4686 integrated, for details see:
4687 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
4689 * A new service type Type=idle is available now which helps us
4690 avoiding ugly interleaving of getty output and boot status
4693 * There's now a system-wide CapabilityBoundingSet= option to
4694 globally reduce the set of capabilities for the
4695 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
4696 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
4697 even CAP_NET_ADMIN system-wide for secure systems.
4699 * There are now system-wide DefaultLimitXXX= options to
4700 globally change the defaults of the various resource limits
4701 for all units started by PID 1.
4703 * Harald Hoyer's systemd test suite has been integrated into
4704 systemd which allows easy testing of systemd builds in qemu
4705 and nspawn. (This is really awesome! Ask us for details!)
4707 * The fstab parser is now implemented as generator, not inside
4710 * systemctl will now warn you if .mount units generated from
4711 /etc/fstab are out of date due to changes in fstab that
4712 have not been read by systemd yet.
4714 * systemd is now suitable for usage in initrds. Dracut has
4715 already been updated to make use of this. With this in place
4716 initrds get a slight bit faster but primarily are much
4717 easier to introspect and debug since "systemctl status" in
4718 the host system can be used to introspect initrd services,
4719 and the journal from the initrd is kept around too.
4721 * systemd-delta has been added, a tool to explore differences
4722 between user/admin configuration and vendor defaults.
4724 * PrivateTmp= now affects both /tmp and /var/tmp.
4726 * Boot time status messages are now much prettier and feature
4727 proper english language. Booting up systemd has never been
4730 * Read-ahead pack files now include the inode number of all
4731 files to pre-cache. When the inode changes the pre-caching
4732 is not attempted. This should be nicer to deal with updated
4733 packages which might result in changes of read-ahead
4736 * We now temporaritly lower the kernel's read_ahead_kb variable
4737 when collecting read-ahead data to ensure the kernel's
4738 built-in read-ahead does not add noise to our measurements
4739 of necessary blocks to pre-cache.
4741 * There's now RequiresMountsFor= to add automatic dependencies
4742 for all mounts necessary for a specific file system path.
4744 * MountAuto= and SwapAuto= have been removed from
4745 system.conf. Mounting file systems at boot has to take place
4748 * nspawn now learned a new switch --uuid= to set the machine
4749 ID on the command line.
4751 * nspawn now learned the -b switch to automatically search
4754 * vt102 is now the default TERM for serial TTYs, upgraded from
4757 * systemd-logind now works on VT-less systems.
4759 * The build tree has been reorganized. The individual
4760 components now have directories of their own.
4762 * A new condition type ConditionPathIsReadWrite= is now available.
4764 * nspawn learned the new -C switch to create cgroups for the
4765 container in other hierarchies.
4767 * We now have support for hardware watchdogs, configurable in
4770 * The scheduled shutdown logic now has a public API.
4772 * We now mount /tmp as tmpfs by default, but this can be
4773 masked and /etc/fstab can override it.
4775 * Since udisks does not make use of /media anymore we are not
4776 mounting a tmpfs on it anymore.
4778 * journalctl gained a new --local switch to only interleave
4779 locally generated journal files.
4781 * We can now load the IMA policy at boot automatically.
4783 * The GTK tools have been split off into a systemd-ui.
4785 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
4786 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
4787 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
4788 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
4789 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
4790 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
4791 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
4792 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
4793 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
4798 * This is mostly a bugfix release
4800 * Support optional initialization of the machine ID from the
4801 KVM or container configured UUID.
4803 * Support immediate reboots with "systemctl reboot -ff"
4805 * Show /etc/os-release data in systemd-analyze output
4807 * Many bugfixes for the journal, including endianness fixes and
4808 ensuring that disk space enforcement works
4810 * sd-login.h is C++ comptaible again
4812 * Extend the /etc/os-release format on request of the Debian
4815 * We now refuse non-UTF8 strings used in various configuration
4816 and unit files. This is done to ensure we do not pass invalid
4817 data over D-Bus or expose it elsewhere.
4819 * Register Mimo USB Screens as suitable for automatic seat
4822 * Read SELinux client context from journal clients in a race
4825 * Reorder configuration file lookup order. /etc now always
4826 overrides /run in order to allow the administrator to always
4827 and unconditionally override vendor supplied or
4828 automatically generated data.
4830 * The various user visible bits of the journal now have man
4831 pages. We still lack man pages for the journal API calls
4834 * We now ship all man pages in HTML format again in the
4837 Contributions from: Dave Reisner, Dirk Eibach, Frederic
4838 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
4839 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
4844 * This is mostly a bugfix release
4846 * systems lacking /etc/os-release are no longer supported.
4848 * Various functionality updates to libsystemd-login.so
4850 * Track class of PAM logins to distinguish greeters from
4853 Contributions from: Kay Sievers, Lennart Poettering, Michael
4858 * This is an important bugfix release for v41.
4860 * Building man pages is now optional which should be useful
4861 for those building systemd from git but unwilling to install
4864 * Watchdog support for supervising services is now usable. In
4865 a future release support for hardware watchdogs
4866 (i.e. /dev/watchdog) will be added building on this.
4868 * Service start rate limiting is now configurable and can be
4869 turned off per service. When a start rate limit is hit a
4870 reboot can automatically be triggered.
4872 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
4874 Contributions from: Benjamin Franzke, Bill Nottingham,
4875 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
4876 Schmidt, Michał Górny, Piotr Drąg
4880 * The systemd binary is installed /usr/lib/systemd/systemd now;
4881 An existing /sbin/init symlink needs to be adapted with the
4884 * The code that loads kernel modules has been ported to invoke
4885 libkmod directly, instead of modprobe. This means we do not
4886 support systems with module-init-tools anymore.
4888 * Watchdog support is now already useful, but still not
4891 * A new kernel command line option systemd.setenv= is
4892 understood to set system wide environment variables
4893 dynamically at boot.
4895 * We now limit the set of capabilities of systemd-journald.
4897 * We now set SIGPIPE to ignore by default, since it only is
4898 useful in shell pipelines, and has little use in general
4899 code. This can be disabled with IgnoreSIPIPE=no in unit
4902 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
4903 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
4908 * This is mostly a bugfix release
4910 * We now expose the reason why a service failed in the
4911 "Result" D-Bus property.
4913 * Rudimentary service watchdog support (will be completed over
4914 the next few releases.)
4916 * When systemd forks off in order execute some service we will
4917 now immediately changes its argv[0] to reflect which process
4918 it will execute. This is useful to minimize the time window
4919 with a generic argv[0], which makes bootcharts more useful
4921 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
4922 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
4923 Mike Kazantsev, Ray Strode
4927 * This is mostly a test release, but incorporates many
4930 * New systemd-cgtop tool to show control groups by their
4933 * Linking against libacl for ACLs is optional again. If
4934 disabled, support tracking device access for active logins
4935 goes becomes unavailable, and so does access to the user
4936 journals by the respective users.
4938 * If a group "adm" exists, journal files are automatically
4939 owned by them, thus allow members of this group full access
4940 to the system journal as well as all user journals.
4942 * The journal now stores the SELinux context of the logging
4943 client for all entries.
4945 * Add C++ inclusion guards to all public headers
4947 * New output mode "cat" in the journal to print only text
4948 messages, without any meta data like date or time.
4950 * Include tiny X server wrapper as a temporary stop-gap to
4951 teach XOrg udev display enumeration. This is used by display
4952 managers such as gdm, and will go away as soon as XOrg
4953 learned native udev hotplugging for display devices.
4955 * Add new systemd-cat tool for executing arbitrary programs
4956 with STDERR/STDOUT connected to the journal. Can also act as
4957 BSD logger replacement, and does so by default.
4959 * Optionally store all locally generated coredumps in the
4960 journal along with meta data.
4962 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
4963 writing short strings to files (for usage for /sys), and for
4964 creating symlinks, character and block device nodes.
4966 * New unit file option ControlGroupPersistent= to make cgroups
4967 persistent, following the mechanisms outlined in
4968 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
4970 * Support multiple local RTCs in a sane way
4972 * No longer monopolize IO when replaying readahead data on
4973 rotating disks, since we might starve non-file-system IO to
4974 death, since fanotify() will not see accesses done by blkid,
4977 * Do not show kernel threads in systemd-cgls anymore, unless
4978 requested with new -k switch.
4980 Contributions from: Dan Horák, Kay Sievers, Lennart
4981 Poettering, Michal Schmidt
4985 * This is mostly a test release, but incorporates many
4988 * The git repository moved to:
4989 git://anongit.freedesktop.org/systemd/systemd
4990 ssh://git.freedesktop.org/git/systemd/systemd
4992 * First release with the journal
4993 http://0pointer.de/blog/projects/the-journal.html
4995 * The journal replaces both systemd-kmsg-syslogd and
4996 systemd-stdout-bridge.
4998 * New sd_pid_get_unit() API call in libsystemd-logind
5000 * Many systemadm clean-ups
5002 * Introduce remote-fs-pre.target which is ordered before all
5003 remote mounts and may be used to start services before all
5006 * Added Mageia support
5008 * Add bash completion for systemd-loginctl
5010 * Actively monitor PID file creation for daemons which exit in
5011 the parent process before having finished writing the PID
5012 file in the daemon process. Daemons which do this need to be
5013 fixed (i.e. PID file creation must have finished before the
5014 parent exits), but we now react a bit more gracefully to them.
5016 * Add colourful boot output, mimicking the well-known output
5017 of existing distributions.
5019 * New option PassCredentials= for socket units, for
5020 compatibility with a recent kernel ABI breakage.
5022 * /etc/rc.local is now hooked in via a generator binary, and
5023 thus will no longer act as synchronization point during
5026 * systemctl list-unit-files now supports --root=.
5028 * systemd-tmpfiles now understands two new commands: z, Z for
5029 relabelling files according to the SELinux database. This is
5030 useful to apply SELinux labels to specific files in /sys,
5033 * Output of SysV services is now forwarded to both the console
5034 and the journal by default, not only just the console.
5036 * New man pages for all APIs from libsystemd-login.
5038 * The build tree got reorganized and a the build system is a
5039 lot more modular allowing embedded setups to specifically
5040 select the components of systemd they are interested in.
5042 * Support for Linux systems lacking the kernel VT subsystem is
5045 * configure's --with-rootdir= got renamed to
5046 --with-rootprefix= to follow the naming used by udev and
5049 * Unless specified otherwise we will now install to /usr instead
5050 of /usr/local by default.
5052 * Processes with '@' in argv[0][0] are now excluded from the
5053 final shut-down killing spree, following the logic explained
5055 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
5057 * All processes remaining in a service cgroup when we enter
5058 the START or START_PRE states are now killed with
5059 SIGKILL. That means it is no longer possible to spawn
5060 background processes from ExecStart= lines (which was never
5061 supported anyway, and bad style).
5063 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
5064 reloading of units together.
5066 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
5067 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
5068 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
5069 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
5070 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek