1 systemd System and Service Manager
5 * The sd-bus.h and sd-event.h APIs have now been declared
6 stable and have been added to the official interface of
7 libsystemd.so. sd-bus implements an alternative D-Bus client
8 library, that is relatively easy to use, very efficient and
9 supports both classic D-Bus as well as kdbus as transport
10 backend. sd-event is a generic event loop abstraction that
11 is built around Linux epoll, but adds features such as event
12 prioritization or efficient timer handling. Both APIs are good
13 choices for C programs looking for a bus and/or event loop
14 implementation that is minimal and does not have to be
15 portable to other kernels.
17 * kdbus support is no longer compile-time optional. It is now
18 always built-in. However, it can still be disabled at
19 runtime using the kdbus=0 kernel command line setting, and
20 that setting may be changed to default to off, by specifying
21 --disable-kdbus at build-time. Note though that the kernel
22 command line setting has no effect if the kdbus.ko kernel
23 module is not installed, in which case kdbus is (obviously)
24 also disabled. We encourage all downstream distributions to
25 begin testing kdbus by adding it to the kernel images in the
26 development distributions, and leaving kdbus support in
29 * The minimal required util-linux version has been bumped to
32 * Support for chkconfig (--enable-chkconfig) was removed in
33 favor of calling an abstraction tool
34 /lib/systemd/systemd-sysv-install. This needs to be
35 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
36 in README for details.
38 * If there's a systemd unit and a SysV init script for the
39 same service name, and the user executes "systemctl enable"
40 for it (or a related call), then this will now enable both
41 (or execute the related operation on both), not just the
44 * The libudev API documentation has been converted from gtkdoc
47 * gudev has been removed from the systemd tree, it is now an
50 * The systemd-cgtop tool learnt a new --raw switch to generate
51 "raw" (machine parsable) output.
53 * networkd's IPForwarding= .network file setting learnt the
54 new setting "kernel", which ensures that networkd does not
55 change the IP forwarding sysctl from the default kernel
58 * The systemd-logind bus API now exposes a new boolean
59 property "Docked" that reports whether logind considers the
60 system "docked", i.e. connected to a docking station or not.
62 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
63 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
64 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
65 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
66 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
67 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
68 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
69 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
70 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
71 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
72 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
73 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
74 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
75 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
76 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
77 Fink, Zbigniew Jędrzejewski-Szmek
83 * The gudev library has been extracted into a separate repository
84 available at: https://git.gnome.org/browse/libgudev/
85 It is now managed as part of the Gnome project. Distributions
86 are recommended to pass --disable-gudev to systemd and use
87 gudev from the Gnome project instead. gudev is still included
88 in systemd, for now. It will be removed soon, though. Please
89 also see the announcement-thread on systemd-devel:
90 http://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
92 * systemd now exposes a CPUUsageNSec= property for each
93 service unit on the bus, that contains the overall consumed
94 CPU time of a service (the sum of what each process of the
95 service consumed). This value is only available if
96 CPUAccounting= is turned on for a service, and is then shown
97 in the "systemctl status" output.
99 * Support for configuring alternative mappings of the old SysV
100 runlevels to systemd targets has been removed. They are now
101 hardcoded in a way that runlevels 2, 3, 4 all map to
102 multi-user.target and 5 to graphical.target (which
103 previously was already the default behaviour).
105 * The auto-mounter logic gained support for mount point
106 expiry, using a new TimeoutIdleSec= setting in .automount
107 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
109 * The EFI System Partition (ESP) as mounted to /boot by
110 systemd-efi-boot-generator will now be unmounted
111 automatically after 2 minutes of not being used. This should
112 minimize the risk of ESP corruptions.
114 * New /etc/fstab options x-systemd.requires= and
115 x-systemd.requires-mounts-for= are now supported to express
116 additional dependencies for mounts. This is useful for
117 journalling file systems that support external journal
118 devices or overlay file systems that require underlying file
119 systems to be mounted.
121 * systemd does not support direct live-upgrades (via systemctl
122 daemon-reexec) from versions older than v44 anymore. As no
123 distribution we are aware of shipped such old versions in a
124 stable release this should not be problematic.
126 * When systemd forks off a new per-connection service instance
127 it will now set the $REMOTE_ADDR environment variable to the
128 remote IP address, and $REMOTE_PORT environment variable to
129 the remote IP port. This behaviour is similar to the
130 corresponding environment variables defined by CGI.
132 * systemd-networkd gained support for uplink failure
133 detection. The BindCarrier= option allows binding interface
134 configuration dynamically to the link sense of other
135 interfaces. This is useful to achieve behaviour like in
138 * systemd-networkd gained support for configuring the DHCP
139 client identifier to use when requesting leases.
141 * systemd-networkd now has a per-network UseNTP= option to
142 configure whether NTP server information acquired via DHCP
143 is passed on to services like systemd-timesyncd.
145 * systemd-networkd gained support for vti6 tunnels.
147 * Note that systemd-networkd manages the sysctl variable
148 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
149 it is configured for since v219. The variable controls IP
150 forwarding, and is a per-interface alternative to the global
151 /proc/sys/net/ipv[46]/ip_forward. This setting is
152 configurable in the IPForward= option, which defaults to
153 "no". This means if networkd is used for an interface it is
154 no longer sufficient to set the global sysctl option to turn
155 on IP forwarding! Instead, the .network file option
156 IPForward= needs to be turned on! Note that the
157 implementation of this behaviour was broken in v219 and has
160 * Many bonding and vxlan options are now configurable in
163 * systemd-nspawn gained a new --property= setting to set unit
164 properties for the container scope. This is useful for
165 setting resource parameters (e.g "CPUShares=500") on
166 containers started from the command line.
168 * systemd-nspawn gained a new --private-users= switch to make
169 use of user namespacing available on recent Linux kernels.
171 * systemd-nspawn may now be called as part of a shell pipeline
172 in which case the pipes used for stdin and stdout are passed
173 directly to the process invoked in the container, without
174 indirection via a pseudo tty.
176 * systemd-nspawn gained a new switch to control the UNIX
177 signal to use when killing the init process of the container
180 * systemd-nspawn gained a new --overlay= switch for mounting
181 overlay file systems into the container using the new kernel
184 * When a container image is imported via systemd-importd and
185 the host file system is not btrfs, a loopback block device
186 file is created in /var/lib/machines.raw with a btrfs file
187 system inside. It is then mounted to /var/lib/machines to
188 enable btrfs features for container management. The loopback
189 file and btrfs file system is grown as needed when container
190 images are imported via systemd-importd.
192 * systemd-machined/systemd-importd gained support for btrfs
193 quota, to enforce container disk space limits on disk. This
194 is exposed in "machinectl set-limit".
196 * systemd-importd now can import containers from local .tar,
197 .raw and .qcow2 images, and export them to .tar and .raw. It
198 can also import dkr v2 images now from the network (on top
201 * systemd-importd gained support for verifying downloaded
202 images with gpg2 (previously only gpg1 was supported).
204 * systemd-machined, systemd-logind, systemd: most bus calls
205 are now accessible to unprivileged processes via
206 PolicyKit. Also, systemd-logind will now allow users to kill
207 their own sessions without further privileges or
210 * systemd-shutdownd has been removed. This service was
211 previously responsible for implementing scheduled shutdowns
212 as exposed in /usr/bin/shutdown's time parameter. This
213 functionality has now been moved into systemd-logind and is
214 accessible via a bus interface.
216 * "systemctl reboot" gained a new switch --firmware-setup that
217 can be used to reboot into the EFI firmware setup, if that
218 is available. systemd-logind now exposes an API on the bus
219 to trigger such reboots, in case graphical desktop UIs want
220 to cover this functionality.
222 * "systemctl enable", "systemctl disable" and "systemctl mask"
223 now support a new "--now" switch. If specified the units
224 that are enabled will also be started, and the ones
225 disabled/masked also stopped.
227 * The Gummiboot EFI boot loader tool has been merged into
228 systemd, and renamed to "systemd-boot". The bootctl tool has been
229 updated to support systemd-boot.
231 * An EFI kernel stub has been added that may be used to create
232 kernel EFI binaries that contain not only the actual kernel,
233 but also an initrd, boot splash, command line and OS release
234 information. This combined binary can then be signed as a
235 single image, so that the firmware can verify it all in one
236 step. systemd-boot has special support for EFI binaries created
237 like this and can extract OS release information from them
238 and show them in the boot menu. This functionality is useful
239 to implement cryptographically verified boot schemes.
241 * Optional support has been added to systemd-fsck to pass
242 fsck's progress report to an AF_UNIX socket in the file
245 * udev will no longer create device symlinks for all block
246 devices by default. A blacklist for excluding special block
247 devices from this logic has been turned into a whitelist
248 that requires picking block devices explicitly that require
251 * A new (currently still internal) API sd-device.h has been
252 added to libsystemd. This modernized API is supposed to
253 replace libudev eventually. In fact, already much of libudev
254 is now just a wrapper around sd-device.h.
256 * A new hwdb database for storing metadata about pointing
257 stick devices has been added.
259 * systemd-tmpfiles gained support for setting file attributes
260 similar to the "chattr" tool with new 'h' and 'H' lines.
262 * systemd-journald will no longer unconditionally set the
263 btrfs NOCOW flag on new journal files. This is instead done
264 with tmpfiles snippet using the new 'h' line type. This
265 allows easy disabling of this logic, by masking the
266 journal-nocow.conf tmpfiles file.
268 * systemd-journald will now translate audit message types to
269 human readable identifiers when writing them to the
270 journal. This should improve readability of audit messages.
272 * The LUKS logic gained support for the offset= and skip=
273 options in /etc/crypttab, as previously implemented by
276 * /usr/lib/os-release gained a new optional field VARIANT= for
277 distributions that support multiple variants (such as a
278 desktop edition, a server edition, ...)
280 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
281 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
282 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
283 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
284 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
285 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
286 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
287 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
288 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
289 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
290 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
291 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
292 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
293 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
294 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
295 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
296 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
297 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
298 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
299 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
300 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
301 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
302 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
303 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
304 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
305 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
306 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
308 -- Berlin, 2015-05-22
312 * Introduce a new API "sd-hwdb.h" for querying the hardware
313 metadata database. With this minimal interface one can query
314 and enumerate the udev hwdb, decoupled from the old libudev
315 library. libudev's interface for this is now only a wrapper
316 around sd-hwdb. A new tool systemd-hwdb has been added to
317 interface with and update the database.
319 * When any of systemd's tools copies files (for example due to
320 tmpfiles' C lines) a btrfs reflink will attempted first,
321 before bytewise copying is done.
323 * systemd-nspawn gained a new --ephemeral switch. When
324 specified a btrfs snapshot is taken of the container's root
325 directory, and immediately removed when the container
326 terminates again. Thus, a container can be started whose
327 changes never alter the container's root directory, and are
328 lost on container termination. This switch can also be used
329 for starting a container off the root file system of the
330 host without affecting the host OS. This switch is only
331 available on btrfs file systems.
333 * systemd-nspawn gained a new --template= switch. It takes the
334 path to a container tree to use as template for the tree
335 specified via --directory=, should that directory be
336 missing. This allows instantiating containers dynamically,
337 on first run. This switch is only available on btrfs file
340 * When a .mount unit refers to a mount point on which multiple
341 mounts are stacked, and the .mount unit is stopped all of
342 the stacked mount points will now be unmounted until no
345 * systemd now has an explicit notion of supported and
346 unsupported unit types. Jobs enqueued for unsupported unit
347 types will now fail with an "unsupported" error code. More
348 specifically .swap, .automount and .device units are not
349 supported in containers, .busname units are not supported on
350 non-kdbus systems. .swap and .automount are also not
351 supported if their respective kernel compile time options
354 * machinectl gained support for two new "copy-from" and
355 "copy-to" commands for copying files from a running
356 container to the host or vice versa.
358 * machinectl gained support for a new "bind" command to bind
359 mount host directories into local containers. This is
360 currently only supported for nspawn containers.
362 * networkd gained support for configuring bridge forwarding
363 database entries (fdb) from .network files.
365 * A new tiny daemon "systemd-importd" has been added that can
366 download container images in tar, raw, qcow2 or dkr formats,
367 and make them available locally in /var/lib/machines, so
368 that they can run as nspawn containers. The daemon can GPG
369 verify the downloads (not supported for dkr, since it has no
370 provisions for verifying downloads). It will transparently
371 decompress bz2, xz, gzip compressed downloads if necessary,
372 and restore sparse files on disk. The daemon uses privilege
373 separation to ensure the actual download logic runs with
374 fewer privileges than the deamon itself. machinectl has
375 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
376 make the functionality of importd available to the
377 user. With this in place the Fedora and Ubuntu "Cloud"
378 images can be downloaded and booted as containers unmodified
379 (the Fedora images lack the appropriate GPG signature files
380 currently, so they cannot be verified, but this will change
381 soon, hopefully). Note that downloading images is currently
382 only fully supported on btrfs.
384 * machinectl is now able to list container images found in
385 /var/lib/machines, along with some metadata about sizes of
386 disk and similar. If the directory is located on btrfs and
387 quota is enabled, this includes quota display. A new command
388 "image-status" has been added that shows additional
389 information about images.
391 * machinectl is now able to clone container images
392 efficiently, if the underlying file system (btrfs) supports
393 it, with the new "machinectl list-images" command. It also
394 gained commands for renaming and removing images, as well as
395 marking them read-only or read-write (supported also on
396 legacy file systems).
398 * networkd gained support for collecting LLDP network
399 announcements, from hardware that supports this. This is
400 shown in networkctl output.
402 * systemd-run gained support for a new -t (--pty) switch for
403 invoking a binary on a pty whose input and output is
404 connected to the invoking terminal. This allows executing
405 processes as system services while interactively
406 communicating with them via the terminal. Most interestingly
407 this is supported across container boundaries. Invoking
408 "systemd-run -t /bin/bash" is an alternative to running a
409 full login session, the difference being that the former
410 will not register a session, nor go through the PAM session
413 * tmpfiles gained support for a new "v" line type for creating
414 btrfs subvolumes. If the underlying file system is a legacy
415 file system, this automatically degrades to creating a
416 normal directory. Among others /var/lib/machines is now
417 created like this at boot, should it be missing.
419 * The directory /var/lib/containers/ has been deprecated and
420 been replaced by /var/lib/machines. The term "machines" has
421 been used in the systemd context as generic term for both
422 VMs and containers, and hence appears more appropriate for
423 this, as the directory can also contain raw images bootable
426 * systemd-nspawn when invoked with -M but without --directory=
427 or --image= is now capable of searching for the container
428 root directory, subvolume or disk image automatically, in
429 /var/lib/machines. systemd-nspawn@.service has been updated
430 to make use of this, thus allowing it to be used for raw
433 * A new machines.target unit has been introduced that is
434 supposed to group all containers/VMs invoked as services on
435 the system. systemd-nspawn@.service has been updated to
438 * machinectl gained a new "start" command, for invoking a
439 container as a service. "machinectl start foo" is mostly
440 equivalent to "systemctl start systemd-nspawn@foo.service",
441 but handles escaping in a nicer way.
443 * systemd-nspawn will now mount most of the cgroupfs tree
444 read-only into each container, with the exception of the
445 container's own subtree in the name=systemd hierarchy.
447 * journald now sets the special FS_NOCOW file flag for its
448 journal files. This should improve performance on btrfs, by
449 avoiding heavy fragmentation when journald's write-pattern
450 is used on COW file systems. It degrades btrfs' data
451 integrity guarantees for the files to the same levels as for
452 ext3/ext4 however. This should be OK though as journald does
453 its own data integrity checks and all its objects are
454 checksummed on disk. Also, journald should handle btrfs disk
455 full events a lot more gracefully now, by processing SIGBUS
456 errors, and not relying on fallocate() anymore.
458 * When journald detects that journal files it is writing to
459 have been deleted it will immediately start new journal
462 * systemd now provides a way to store file descriptors
463 per-service in PID 1.This is useful for daemons to ensure
464 that fds they require are not lost during a daemon
465 restart. The fds are passed to the deamon on the next
466 invocation in the same way socket activation fds are
467 passed. This is now used by journald to ensure that the
468 various sockets connected to all the system's stdout/stderr
469 are not lost when journald is restarted. File descriptors
470 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
471 an extension to sd_notify(). Note that a limit is enforced
472 on the number of fds a service can store in PID 1, and it
473 defaults to 0, so that no fds may be stored, unless this is
474 explicitly turned on.
476 * The default TERM variable to use for units connected to a
477 terminal, when no other value is explicitly is set is now
478 vt220 rather than vt102. This should be fairly safe still,
479 but allows PgUp/PgDn work.
481 * The /etc/crypttab option header= as known from Debian is now
484 * "loginctl user-status" and "loginctl session-status" will
485 now show the last 10 lines of log messages of the
486 user/session following the status output. Similar,
487 "machinectl status" will show the last 10 log lines
488 associated with a virtual machine or container
489 service. (Note that this is usually not the log messages
490 done in the VM/container itself, but simply what the
491 container manager logs. For nspawn this includes all console
494 * "loginctl session-status" without further argument will now
495 show the status of the session of the caller. Similar,
496 "lock-session", "unlock-session", "activate",
497 "enable-linger", "disable-linger" may now be called without
498 session/user parameter in which case they apply to the
499 caller's session/user.
501 * An X11 session scriptlet is now shipped that uploads
502 $DISPLAY and $XAUTHORITY into the environment of the systemd
503 --user daemon if a session begins. This should improve
504 compatibility with X11 enabled applications run as systemd
507 * Generators are now subject to masking via /etc and /run, the
508 same way as unit files.
510 * networkd .network files gained support for configuring
511 per-link IPv4/IPv6 packet forwarding as well as IPv4
512 masquerading. This is by default turned on for veth links to
513 containers, as registered by systemd-nspawn. This means that
514 nspawn containers run with --network-veth will now get
515 automatic routed access to the host's networks without any
516 further configuration or setup, as long as networkd runs on
519 * systemd-nspawn gained the --port= (-p) switch to expose TCP
520 or UDP posts of a container on the host. With this in place
521 it is possible to run containers with private veth links
522 (--network-veth), and have their functionality exposed on
523 the host as if their services were running directly on the
526 * systemd-nspawn's --network-veth switch now gained a short
527 version "-n", since with the changes above it is now truly
528 useful out-of-the-box. The systemd-nspawn@.service has been
529 updated to make use of it too by default.
531 * systemd-nspawn will now maintain a per-image R/W lock, to
532 ensure that the same image is not started more than once
533 writable. (It's OK to run an image multiple times
534 simultaneously in read-only mode.)
536 * systemd-nspawn's --image= option is now capable of
537 dissecting and booting MBR and GPT disk images that contain
538 only a single active Linux partition. Previously it
539 supported only GPT disk images with proper GPT type
540 IDs. This allows running cloud images from major
541 distributions directly with systemd-nspawn, without
544 * In addition to collecting mouse dpi data in the udev
545 hardware database, there's now support for collecting angle
546 information for mouse scroll wheels. The database is
547 supposed to guarantee similar scrolling behavior on mice
548 that it knows about. There's also support for collecting
549 information about Touchpad types.
551 * udev's input_id built-in will now also collect touch screen
552 dimension data and attach it to probed devices.
554 * /etc/os-release gained support for a Distribution Privacy
557 * networkd gained support for creating "ipvlan", "gretap",
558 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
560 * systemd-tmpfiles gained support for "a" lines for setting
563 * systemd-nspawn will now mount /tmp in the container to
564 tmpfs, automatically.
566 * systemd now exposes the memory.usage_in_bytes cgroup
567 attribute and shows it for each service in the "systemctl
568 status" output, if available.
570 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
571 immediate reboot is triggered. This useful if shutdown is
572 hung and is unable to complete, to expedite the
573 operation. Note that this kind of reboot will still unmount
574 all file systems, and hence should not result in fsck being
577 * A .device unit for an optical block device will now be
578 considered active only when a medium is in the drive. Also,
579 mount units are now bound to their backing devices thus
580 triggering automatic unmounting when devices become
581 unavailable. With this in place systemd will now
582 automatically unmount left-over mounts when a CD-ROM is
583 ejected or an USB stick is yanked from the system.
585 * networkd-wait-online now has support for waiting for
586 specific interfaces only (with globbing), and for giving up
587 after a configurable timeout.
589 * networkd now exits when idle. It will be automatically
590 restarted as soon as interfaces show up, are removed or
591 change state. networkd will stay around as long as there is
592 at least one DHCP state machine or similar around, that keep
595 * networkd may now configure IPv6 link-local addressing in
596 addition to IPv4 link-local addressing.
598 * The IPv6 "token" for use in SLAAC may now be configured for
599 each .network interface in networkd.
601 * Routes configured with networkd may now be assigned a scope
604 * networkd's [Match] sections now support globbing and lists
605 of multiple space-separated matches per item.
607 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
608 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
609 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
610 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
611 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
612 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
613 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
614 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
615 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
616 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
617 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
618 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
619 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
620 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
621 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
622 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
623 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
624 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
625 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
626 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
627 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
628 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
629 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
630 Hoffmann, Zbigniew Jędrzejewski-Szmek
632 -- Berlin, 2015-02-16
636 * When querying unit file enablement status (for example via
637 "systemctl is-enabled"), a new state "indirect" is now known
638 which indicates that a unit might not be enabled itself, but
639 another unit listed in its Alias= setting might be.
641 * Similar to the various existing ConditionXYZ= settings for
642 units there are now matching AssertXYZ= settings. While
643 failing conditions cause a unit to be skipped, but its job
644 to succeed, failing assertions declared like this will cause
645 a unit start operation and its job to fail.
647 * hostnamed now knows a new chassis type "embedded".
649 * systemctl gained a new "edit" command. When used on a unit
650 file this allows extending unit files with .d/ drop-in
651 configuration snippets or editing the full file (after
652 copying it from /usr/lib to /etc). This will invoke the
653 user's editor (as configured with $EDITOR), and reload the
654 modified configuration after editing.
656 * "systemctl status" now shows the suggested enablement state
657 for a unit, as declared in the (usually vendor-supplied)
660 * nss-myhostname will now resolve the single-label host name
661 "gateway" to the locally configured default IP routing
662 gateways, ordered by their metrics. This assigns a stable
663 name to the used gateways, regardless which ones are
664 currently configured. Note that the name will only be
665 resolved after all other name sources (if nss-myhostname is
666 configured properly) and should hence not negatively impact
667 systems that use the single-label host name "gateway" in
670 * systemd-inhibit now allows filtering by mode when listing
673 * Scope and service units gained a new "Delegate" boolean
674 property, which when set allows processes running inside the
675 unit to further partition resources. This is primarily
676 useful for systemd user instances as well as container
679 * journald will now pick up audit messages directly from
680 the kernel, and log them like any other log message. The
681 audit fields are split up and fully indexed. This means that
682 journalctl in many ways is now a (nicer!) alternative to
683 ausearch, the traditional audit client. Note that this
684 implements only a minimal audit client, if you want the
685 special audit modes like reboot-on-log-overflow, please use
686 the traditional auditd instead, which can be used in
687 parallel to journald.
689 * The ConditionSecurity= unit file option now understands the
690 special string "audit" to check whether auditing is
693 * journalctl gained two new commands --vacuum-size= and
694 --vacuum-time= to delete old journal files until the
695 remaining ones take up no more the specified size on disk,
696 or are not older than the specified time.
698 * A new, native PPPoE library has been added to sd-network,
699 systemd's library of light-weight networking protocols. This
700 library will be used in a future version of networkd to
701 enable PPPoE communication without an external pppd daemon.
703 * The busctl tool now understands a new "capture" verb that
704 works similar to "monitor", but writes a packet capture
705 trace to STDOUT that can be redirected to a file which is
706 compatible with libcap's capture file format. This can then
707 be loaded in Wireshark and similar tools to inspect bus
710 * The busctl tool now understands a new "tree" verb that shows
711 the object trees of a specific service on the bus, or of all
714 * The busctl tool now understands a new "introspect" verb that
715 shows all interfaces and members of objects on the bus,
716 including their signature and values. This is particularly
717 useful to get more information about bus objects shown by
718 the new "busctl tree" command.
720 * The busctl tool now understands new verbs "call",
721 "set-property" and "get-property" for invoking bus method
722 calls, setting and getting bus object properties in a
725 * busctl gained a new --augment-creds= argument that controls
726 whether the tool shall augment credential information it
727 gets from the bus with data from /proc, in a possibly
730 * nspawn's --link-journal= switch gained two new values
731 "try-guest" and "try-host" that work like "guest" and
732 "host", but do not fail if the host has no persistent
733 journalling enabled. -j is now equivalent to
734 --link-journal=try-guest.
736 * macvlan network devices created by nspawn will now have
737 stable MAC addresses.
739 * A new SmackProcessLabel= unit setting has been added, which
740 controls the SMACK security label processes forked off by
741 the respective unit shall use.
743 * If compiled with --enable-xkbcommon, systemd-localed will
744 verify x11 keymap settings by compiling the given keymap. It
745 will spew out warnings if the compilation fails. This
746 requires libxkbcommon to be installed.
748 * When a coredump is collected a larger number of metadata
749 fields is now collected and included in the journal records
750 created for it. More specifically control group membership,
751 environment variables, memory maps, working directory,
752 chroot directory, /proc/$PID/status, and a list of open file
753 descriptors is now stored in the log entry.
755 * The udev hwdb now contains DPI information for mice. For
758 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
760 * All systemd programs that read standalone configuration
761 files in /etc now also support a corresponding series of
762 .conf.d configuration directories in /etc/, /run/,
763 /usr/local/lib/, /usr/lib/, and (if configured with
764 --enable-split-usr) /lib/. In particular, the following
765 configuration files now have corresponding configuration
766 directories: system.conf user.conf, logind.conf,
767 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
768 resolved.conf, timesyncd.conf, journal-remote.conf, and
769 journal-upload.conf. Note that distributions should use the
770 configuration directories in /usr/lib/; the directories in
771 /etc/ are reserved for the system administrator.
773 * systemd-rfkill will no longer take the rfkill device name
774 into account when storing rfkill state on disk, as the name
775 might be dynamically assigned and not stable. Instead, the
776 ID_PATH udev variable combined with the rfkill type (wlan,
777 bluetooth, ...) is used.
779 * A new service systemd-machine-id-commit.service has been
780 added. When used on systems where /etc is read-only during
781 boot, and /etc/machine-id is not initialized (but an empty
782 file), this service will copy the temporary machine ID
783 created as replacement into /etc after the system is fully
784 booted up. This is useful for systems that are freshly
785 installed with a non-initialized machine ID, but should get
786 a fixed machine ID for subsequent boots.
788 * networkd's .netdev files now provide a large set of
789 configuration parameters for VXLAN devices. Similar, the
790 bridge port cost parameter is now configurable in .network
791 files. There's also new support for configuring IP source
792 routing. networkd .link files gained support for a new
793 OriginalName= match that is useful to match against the
794 original interface name the kernel assigned. .network files
795 may include MTU= and MACAddress= fields for altering the MTU
796 and MAC address while being connected to a specific network
799 * The LUKS logic gained supported for configuring
800 UUID-specific key files. There's also new support for naming
801 LUKS device from the kernel command line, using the new
804 * Timer units may now be transiently created via the bus API
805 (this was previously already available for scope and service
806 units). In addition it is now possible to create multiple
807 transient units at the same time with a single bus call. The
808 "systemd-run" tool has been updated to make use of this for
809 running commands on a specified time, in at(1)-style.
811 * tmpfiles gained support for "t" lines, for assigning
812 extended attributes to files. Among other uses this may be
813 used to assign SMACK labels to files.
815 Contributions from: Alin Rauta, Alison Chaiken, Andrej
816 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
817 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
818 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
819 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
820 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
821 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
822 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
823 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
824 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
825 Michael Biebl, Michael Chapman, Michael Marineau, Michal
826 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
827 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
828 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
829 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
830 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
831 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
832 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
834 -- Berlin, 2014-12-10
838 * journalctl gained the new options -t/--identifier= to match
839 on the syslog identifier (aka "tag"), as well as --utc to
840 show log timestamps in the UTC timezone. journalctl now also
841 accepts -n/--lines=all to disable line capping in a pager.
843 * journalctl gained a new switch, --flush, that synchronously
844 flushes logs from /run/log/journal to /var/log/journal if
845 persistent storage is enabled. systemd-journal-flush.service
846 now waits until the operation is complete.
848 * Services can notify the manager before they start a reload
849 (by sending RELOADING=1) or shutdown (by sending
850 STOPPING=1). This allows the manager to track and show the
851 internal state of daemons and closes a race condition when
852 the process is still running but has closed its D-Bus
855 * Services with Type=oneshot do not have to have any ExecStart
858 * User units are now loaded also from
859 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
860 /run/systemd/user directory that was already previously
861 supported, but is under the control of the user.
863 * Job timeouts (i.e. time-outs on the time a job that is
864 queued stays in the run queue) can now optionally result in
865 immediate reboot or power-off actions (JobTimeoutAction= and
866 JobTimeoutRebootArgument=). This is useful on ".target"
867 units, to limit the maximum time a target remains
868 undispatched in the run queue, and to trigger an emergency
869 operation in such a case. This is now used by default to
870 turn off the system if boot-up (as defined by everything in
871 basic.target) hangs and does not complete for at least
872 15min. Also, if power-off or reboot hang for at least 30min
873 an immediate power-off/reboot operation is triggered. This
874 functionality is particularly useful to increase reliability
875 on embedded devices, but also on laptops which might
876 accidentally get powered on when carried in a backpack and
877 whose boot stays stuck in a hard disk encryption passphrase
880 * systemd-logind can be configured to also handle lid switch
881 events even when the machine is docked or multiple displays
882 are attached (HandleLidSwitchDocked= option).
884 * A helper binary and a service have been added which can be
885 used to resume from hibernation in the initramfs. A
886 generator will parse the resume= option on the kernel
887 command line to trigger resume.
889 * A user console daemon systemd-consoled has been
890 added. Currently, it is a preview, and will so far open a
891 single terminal on each session of the user marked as
892 Desktop=systemd-console.
894 * Route metrics can be specified for DHCP routes added by
897 * The SELinux context of socket-activated services can be set
898 from the information provided by the networking stack
899 (SELinuxContextFromNet= option).
901 * Userspace firmware loading support has been removed and
902 the minimum supported kernel version is thus bumped to 3.7.
904 * Timeout for udev workers has been increased from 1 to 3
905 minutes, but a warning will be printed after 1 minute to
906 help diagnose kernel modules that take a long time to load.
908 * Udev rules can now remove tags on devices with TAG-="foobar".
910 * systemd's readahead implementation has been removed. In many
911 circumstances it didn't give expected benefits even for
912 rotational disk drives and was becoming less relevant in the
913 age of SSDs. As none of the developers has been using
914 rotating media anymore, and nobody stepped up to actively
915 maintain this component of systemd it has now been removed.
917 * Swap units can use Options= to specify discard options.
918 Discard options specified for swaps in /etc/fstab are now
921 * Docker containers are now detected as a separate type of
924 * The Password Agent protocol gained support for queries where
925 the user input is shown, useful e.g. for user names.
926 systemd-ask-password gained a new --echo option to turn that
929 * The default sysctl.d/ snippets will now set:
931 net.core.default_qdisc = fq_codel
933 This selects Fair Queuing Controlled Delay as the default
934 queuing discipline for network interfaces. fq_codel helps
935 fight the network bufferbloat problem. It is believed to be
936 a good default with no tuning required for most workloads.
937 Downstream distributions may override this choice. On 10Gbit
938 servers that do not do forwarding, "fq" may perform better.
939 Systems without a good clocksource should use "pfifo_fast".
941 * If kdbus is enabled during build a new option BusPolicy= is
942 available for service units, that allows locking all service
943 processes into a stricter bus policy, in order to limit
944 access to various bus services, or even hide most of them
945 from the service's view entirely.
947 * networkctl will now show the .network and .link file
948 networkd has applied to a specific interface.
950 * sd-login gained a new API call sd_session_get_desktop() to
951 query which desktop environment has been selected for a
954 * UNIX utmp support is now compile-time optional to support
957 * systemctl gained two new commands "add-wants" and
958 "add-requires" for pulling in units from specific targets
961 * If the word "rescue" is specified on the kernel command line
962 the system will now boot into rescue mode (aka
963 rescue.target), which was previously available only by
964 specifying "1" or "systemd.unit=rescue.target" on the kernel
965 command line. This new kernel command line option nicely
966 mirrors the already existing "emergency" kernel command line
969 * New kernel command line options mount.usr=, mount.usrflags=,
970 mount.usrfstype= have been added that match root=, rootflags=,
971 rootfstype= but allow mounting a specific file system to
974 * The $NOTIFY_SOCKET is now also passed to control processes of
975 services, not only the main process.
977 * This version reenables support for fsck's -l switch. This
978 means at least version v2.25 of util-linux is required for
979 operation, otherwise dead-locks on device nodes may
980 occur. Again: you need to update util-linux to at least
981 v2.25 when updating systemd to v217.
983 * The "multi-seat-x" tool has been removed from systemd, as
984 its functionality has been integrated into X servers 1.16,
985 and the tool is hence redundant. It is recommended to update
986 display managers invoking this tool to simply invoke X
987 directly from now on, again.
989 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
990 message flag has been added for all of systemd's PolicyKit
991 authenticated method calls has been added. In particular
992 this now allows optional interactive authorization via
993 PolicyKit for many of PID1's privileged operations such as
994 unit file enabling and disabling.
996 * "udevadm hwdb --update" learnt a new switch "--usr" for
997 placing the rebuilt hardware database in /usr instead of
998 /etc. When used only hardware database entries stored in
999 /usr will be used, and any user database entries in /etc are
1000 ignored. This functionality is useful for vendors to ship a
1001 pre-built database on systems where local configuration is
1002 unnecessary or unlikely.
1004 * Calendar time specifications in .timer units now also
1005 understand the strings "semi-annually", "quarterly" and
1006 "minutely" as shortcuts (in addition to the preexisting
1007 "anually", "hourly", ...).
1009 * systemd-tmpfiles will now correctly create files in /dev
1010 at boot which are marked for creation only at boot. It is
1011 recommended to always create static device nodes with 'c!'
1012 and 'b!', so that they are created only at boot and not
1013 overwritten at runtime.
1015 * When the watchdog logic is used for a service (WatchdogSec=)
1016 and the watchdog timeout is hit the service will now be
1017 terminated with SIGABRT (instead of just SIGTERM), in order
1018 to make sure a proper coredump and backtrace is
1019 generated. This ensures that hanging services will result in
1020 similar coredump/backtrace behaviour as services that hit a
1023 Contributions from: Andreas Henriksson, Andrei Borzenkov,
1024 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
1025 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
1026 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
1027 Herrmann, David Sommerseth, David Strauss, Emil Renner
1028 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
1029 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
1030 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
1031 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
1032 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
1033 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
1034 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
1035 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
1036 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
1037 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
1038 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
1039 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
1040 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
1041 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
1042 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
1043 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
1046 -- Berlin, 2014-10-28
1050 * timedated no longer reads NTP implementation unit names from
1051 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
1052 implementations should add a
1054 Conflicts=systemd-timesyncd.service
1056 to their unit files to take over and replace systemd's NTP
1057 default functionality.
1059 * systemd-sysusers gained a new line type "r" for configuring
1060 which UID/GID ranges to allocate system users/groups
1061 from. Lines of type "u" may now add an additional column
1062 that specifies the home directory for the system user to be
1063 created. Also, systemd-sysusers may now optionally read user
1064 information from STDIN instead of a file. This is useful for
1065 invoking it from RPM preinst scriptlets that need to create
1066 users before the first RPM file is installed since these
1067 files might need to be owned by them. A new
1068 %sysusers_create_inline RPM macro has been introduced to do
1069 just that. systemd-sysusers now updates the shadow files as
1070 well as the user/group databases, which should enhance
1071 compatibility with certain tools like grpck.
1073 * A number of bus APIs of PID 1 now optionally consult
1074 PolicyKit to permit access for otherwise unprivileged
1075 clients under certain conditions. Note that this currently
1076 doesn't support interactive authentication yet, but this is
1077 expected to be added eventually, too.
1079 * /etc/machine-info now has new fields for configuring the
1080 deployment environment of the machine, as well as the
1081 location of the machine. hostnamectl has been updated with
1082 new command to update these fields.
1084 * systemd-timesyncd has been updated to automatically acquire
1085 NTP server information from systemd-networkd, which might
1086 have been discovered via DHCP.
1088 * systemd-resolved now includes a caching DNS stub resolver
1089 and a complete LLMNR name resolution implementation. A new
1090 NSS module "nss-resolve" has been added which can be used
1091 instead of glibc's own "nss-dns" to resolve hostnames via
1092 systemd-resolved. Hostnames, addresses and arbitrary RRs may
1093 be resolved via systemd-resolved D-Bus APIs. In contrast to
1094 the glibc internal resolver systemd-resolved is aware of
1095 multi-homed system, and keeps DNS server and caches separate
1096 and per-interface. Queries are sent simultaneously on all
1097 interfaces that have DNS servers configured, in order to
1098 properly handle VPNs and local LANs which might resolve
1099 separate sets of domain names. systemd-resolved may acquire
1100 DNS server information from systemd-networkd automatically,
1101 which in turn might have discovered them via DHCP. A tool
1102 "systemd-resolve-host" has been added that may be used to
1103 query the DNS logic in resolved. systemd-resolved implements
1104 IDNA and automatically uses IDNA or UTF-8 encoding depending
1105 on whether classic DNS or LLMNR is used as transport. In the
1106 next releases we intend to add a DNSSEC and mDNS/DNS-SD
1107 implementation to systemd-resolved.
1109 * A new NSS module nss-mymachines has been added, that
1110 automatically resolves the names of all local registered
1111 containers to their respective IP addresses.
1113 * A new client tool "networkctl" for systemd-networkd has been
1114 added. It currently is entirely passive and will query
1115 networking configuration from udev, rtnetlink and networkd,
1116 and present it to the user in a very friendly
1117 way. Eventually, we hope to extend it to become a full
1118 control utility for networkd.
1120 * .socket units gained a new DeferAcceptSec= setting that
1121 controls the kernels' TCP_DEFER_ACCEPT sockopt for
1122 TCP. Similar, support for controlling TCP keep-alive
1123 settings has been added (KeepAliveTimeSec=,
1124 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
1125 turning off Nagle's algorithm on TCP has been added
1128 * logind learned a new session type "web", for use in projects
1129 like Cockpit which register web clients as PAM sessions.
1131 * timer units with at least one OnCalendar= setting will now
1132 be started only after timer-sync.target has been
1133 reached. This way they will not elapse before the system
1134 clock has been corrected by a local NTP client or
1135 similar. This is particular useful on RTC-less embedded
1136 machines, that come up with an invalid system clock.
1138 * systemd-nspawn's --network-veth= switch should now result in
1139 stable MAC addresses for both the outer and the inner side
1142 * systemd-nspawn gained a new --volatile= switch for running
1143 container instances with /etc or /var unpopulated.
1145 * The kdbus client code has been updated to use the new Linux
1146 3.17 memfd subsystem instead of the old kdbus-specific one.
1148 * systemd-networkd's DHCP client and server now support
1149 FORCERENEW. There are also new configuration options to
1150 configure the vendor client identifier and broadcast mode
1153 * systemd will no longer inform the kernel about the current
1154 timezone, as this is necessarily incorrect and racy as the
1155 kernel has no understanding of DST and similar
1156 concepts. This hence means FAT timestamps will be always
1157 considered UTC, similar to what Android is already
1158 doing. Also, when the RTC is configured to the local time
1159 (rather than UTC) systemd will never synchronize back to it,
1160 as this might confuse Windows at a later boot.
1162 * systemd-analyze gained a new command "verify" for offline
1163 validation of unit files.
1165 * systemd-networkd gained support for a couple of additional
1166 settings for bonding networking setups. Also, the metric for
1167 statically configured routes may now be configured. For
1168 network interfaces where this is appropriate the peer IP
1169 address may now be configured.
1171 * systemd-networkd's DHCP client will no longer request
1172 broadcasting by default, as this tripped up some networks.
1173 For hardware where broadcast is required the feature should
1174 be switched back on using RequestBroadcast=yes.
1176 * systemd-networkd will now set up IPv4LL addresses (when
1177 enabled) even if DHCP is configured successfully.
1179 * udev will now default to respect network device names given
1180 by the kernel when the kernel indicates that these are
1181 predictable. This behavior can be tweaked by changing
1182 NamePolicy= in the relevant .link file.
1184 * A new library systemd-terminal has been added that
1185 implements full TTY stream parsing and rendering. This
1186 library is supposed to be used later on for implementing a
1187 full userspace VT subsystem, replacing the current kernel
1190 * A new tool systemd-journal-upload has been added to push
1191 journal data to a remote system running
1192 systemd-journal-remote.
1194 * journald will no longer forward all local data to another
1195 running syslog daemon. This change has been made because
1196 rsyslog (which appears to be the most commonly used syslog
1197 implementation these days) no longer makes use of this, and
1198 instead pulls the data out of the journal on its own. Since
1199 forwarding the messages to a non-existent syslog server is
1200 more expensive than we assumed we have now turned this
1201 off. If you run a syslog server that is not a recent rsyslog
1202 version, you have to turn this option on again
1203 (ForwardToSyslog= in journald.conf).
1205 * journald now optionally supports the LZ4 compressor for
1206 larger journal fields. This compressor should perform much
1207 better than XZ which was the previous default.
1209 * machinectl now shows the IP addresses of local containers,
1210 if it knows them, plus the interface name of the container.
1212 * A new tool "systemd-escape" has been added that makes it
1213 easy to escape strings to build unit names and similar.
1215 * sd_notify() messages may now include a new ERRNO= field
1216 which is parsed and collected by systemd and shown among the
1217 "systemctl status" output for a service.
1219 * A new component "systemd-firstboot" has been added that
1220 queries the most basic systemd information (timezone,
1221 hostname, root password) interactively on first
1222 boot. Alternatively it may also be used to provision these
1223 things offline on OS images installed into directories.
1225 * The default sysctl.d/ snippets will now set
1227 net.ipv4.conf.default.promote_secondaries=1
1229 This has the benefit of no flushing secondary IP addresses
1230 when primary addresses are removed.
1232 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
1233 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
1234 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
1235 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
1236 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
1237 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
1238 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
1239 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
1240 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
1241 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
1242 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
1243 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
1244 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
1245 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
1246 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
1248 -- Berlin, 2014-08-19
1252 * A new tool systemd-sysusers has been added. This tool
1253 creates system users and groups in /etc/passwd and
1254 /etc/group, based on static declarative system user/group
1255 definitions in /usr/lib/sysusers.d/. This is useful to
1256 enable factory resets and volatile systems that boot up with
1257 an empty /etc directory, and thus need system users and
1258 groups created during early boot. systemd now also ships
1259 with two default sysusers.d/ files for the most basic
1260 users and groups systemd and the core operating system
1263 * A new tmpfiles snippet has been added that rebuilds the
1264 essential files in /etc on boot, should they be missing.
1266 * A directive for ensuring automatic clean-up of
1267 /var/cache/man/ has been removed from the default
1268 configuration. This line should now be shipped by the man
1269 implementation. The necessary change has been made to the
1270 man-db implementation. Note that you need to update your man
1271 implementation to one that ships this line, otherwise no
1272 automatic clean-up of /var/cache/man will take place.
1274 * A new condition ConditionNeedsUpdate= has been added that
1275 may conditionalize services to only run when /etc or /var
1276 are "older" than the vendor operating system resources in
1277 /usr. This is useful for reconstructing or updating /etc
1278 after an offline update of /usr or a factory reset, on the
1279 next reboot. Services that want to run once after such an
1280 update or reset should use this condition and order
1281 themselves before the new systemd-update-done.service, which
1282 will mark the two directories as fully updated. A number of
1283 service files have been added making use of this, to rebuild
1284 the udev hardware database, the journald message catalog and
1285 dynamic loader cache (ldconfig). The systemd-sysusers tool
1286 described above also makes use of this now. With this in
1287 place it is now possible to start up a minimal operating
1288 system with /etc empty cleanly. For more information on the
1289 concepts involved see this recent blog story:
1291 http://0pointer.de/blog/projects/stateless.html
1293 * A new system group "input" has been introduced, and all
1294 input device nodes get this group assigned. This is useful
1295 for system-level software to get access to input devices. It
1296 complements what is already done for "audio" and "video".
1298 * systemd-networkd learnt minimal DHCPv4 server support in
1299 addition to the existing DHCPv4 client support. It also
1300 learnt DHCPv6 client and IPv6 Router Solicitation client
1301 support. The DHCPv4 client gained support for static routes
1302 passed in from the server. Note that the [DHCPv4] section
1303 known in older systemd-networkd versions has been renamed to
1304 [DHCP] and is now also used by the DHCPv6 client. Existing
1305 .network files using settings of this section should be
1306 updated, though compatibility is maintained. Optionally, the
1307 client hostname may now be sent to the DHCP server.
1309 * networkd gained support for vxlan virtual networks as well
1310 as tun/tap and dummy devices.
1312 * networkd gained support for automatic allocation of address
1313 ranges for interfaces from a system-wide pool of
1314 addresses. This is useful for dynamically managing a large
1315 number of interfaces with a single network configuration
1316 file. In particular this is useful to easily assign
1317 appropriate IP addresses to the veth links of a large number
1318 of nspawn instances.
1320 * RPM macros for processing sysusers, sysctl and binfmt
1321 drop-in snippets at package installation time have been
1324 * The /etc/os-release file should now be placed in
1325 /usr/lib/os-release. The old location is automatically
1326 created as symlink. /usr/lib is the more appropriate
1327 location of this file, since it shall actually describe the
1328 vendor operating system shipped in /usr, and not the
1329 configuration stored in /etc.
1331 * .mount units gained a new boolean SloppyOptions= setting
1332 that maps to mount(8)'s -s option which enables permissive
1333 parsing of unknown mount options.
1335 * tmpfiles learnt a new "L+" directive which creates a symlink
1336 but (unlike "L") deletes a pre-existing file first, should
1337 it already exist and not already be the correct
1338 symlink. Similar, "b+", "c+" and "p+" directives have been
1339 added as well, which create block and character devices, as
1340 well as fifos in the filesystem, possibly removing any
1341 pre-existing files of different types.
1343 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
1344 'argument' field (which so far specified the source to
1345 symlink/copy the files from) is now optional. If omitted the
1346 same file os copied from /usr/share/factory/ suffixed by the
1347 full destination path. This is useful for populating /etc
1348 with essential files, by copying them from vendor defaults
1349 shipped in /usr/share/factory/etc.
1351 * A new command "systemctl preset-all" has been added that
1352 applies the service preset settings to all installed unit
1353 files. A new switch --preset-mode= has been added that
1354 controls whether only enable or only disable operations
1357 * A new command "systemctl is-system-running" has been added
1358 that allows checking the overall state of the system, for
1359 example whether it is fully up and running.
1361 * When the system boots up with an empty /etc, the equivalent
1362 to "systemctl preset-all" is executed during early boot, to
1363 make sure all default services are enabled after a factory
1366 * systemd now contains a minimal preset file that enables the
1367 most basic services systemd ships by default.
1369 * Unit files' [Install] section gained a new DefaultInstance=
1370 field for defining the default instance to create if a
1371 template unit is enabled with no instance specified.
1373 * A new passive target cryptsetup-pre.target has been added
1374 that may be used by services that need to make they run and
1375 finish before the first LUKS cryptographic device is set up.
1377 * The /dev/loop-control and /dev/btrfs-control device nodes
1378 are now owned by the "disk" group by default, opening up
1379 access to this group.
1381 * systemd-coredump will now automatically generate a
1382 stack trace of all core dumps taking place on the system,
1383 based on elfutils' libdw library. This stack trace is logged
1386 * systemd-coredump may now optionally store coredumps directly
1387 on disk (in /var/lib/systemd/coredump, possibly compressed),
1388 instead of storing them unconditionally in the journal. This
1389 mode is the new default. A new configuration file
1390 /etc/systemd/coredump.conf has been added to configure this
1391 and other parameters of systemd-coredump.
1393 * coredumpctl gained a new "info" verb to show details about a
1394 specific coredump. A new switch "-1" has also been added
1395 that makes sure to only show information about the most
1396 recent entry instead of all entries. Also, as the tool is
1397 generally useful now the "systemd-" prefix of the binary
1398 name has been removed. Distributions that want to maintain
1399 compatibility with the old name should add a symlink from
1400 the old name to the new name.
1402 * journald's SplitMode= now defaults to "uid". This makes sure
1403 that unprivileged users can access their own coredumps with
1404 coredumpctl without restrictions.
1406 * New kernel command line options "systemd.wants=" (for
1407 pulling an additional unit during boot), "systemd.mask="
1408 (for masking a specific unit for the boot), and
1409 "systemd.debug-shell" (for enabling the debug shell on tty9)
1410 have been added. This is implemented in the new generator
1411 "systemd-debug-generator".
1413 * systemd-nspawn will now by default filter a couple of
1414 syscalls for containers, among them those required for
1415 kernel module loading, direct x86 IO port access, swap
1416 management, and kexec. Most importantly though
1417 open_by_handle_at() is now prohibited for containers,
1418 closing a hole similar to a recently discussed vulnerability
1419 in docker regarding access to files on file hierarchies the
1420 container should normally not have access to. Note that for
1421 nspawn we generally make no security claims anyway (and
1422 this is explicitly documented in the man page), so this is
1423 just a fix for one of the most obvious problems.
1425 * A new man page file-hierarchy(7) has been added that
1426 contains a minimized, modernized version of the file system
1427 layout systemd expects, similar in style to the FHS
1428 specification or hier(5). A new tool systemd-path(1) has
1429 been added to query many of these paths for the local
1432 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
1433 longer done. Since the directory now has a per-user size
1434 limit, and is cleaned on logout this appears unnecessary,
1435 in particular since this now brings the lifecycle of this
1436 directory closer in line with how IPC objects are handled.
1438 * systemd.pc now exports a number of additional directories,
1439 including $libdir (which is useful to identify the library
1440 path for the primary architecture of the system), and a
1441 couple of drop-in directories.
1443 * udev's predictable network interface names now use the dev_port
1444 sysfs attribute, introduced in linux 3.15 instead of dev_id to
1445 distinguish between ports of the same PCI function. dev_id should
1446 only be used for ports using the same HW address, hence the need
1449 * machined has been updated to export the OS version of a
1450 container (read from /etc/os-release and
1451 /usr/lib/os-release) on the bus. This is now shown in
1452 "machinectl status" for a machine.
1454 * A new service setting RestartForceExitStatus= has been
1455 added. If configured to a set of exit signals or process
1456 return values, the service will be restarted when the main
1457 daemon process exits with any of them, regardless of the
1460 * systemctl's -H switch for connecting to remote systemd
1461 machines has been extended so that it may be used to
1462 directly connect to a specific container on the
1463 host. "systemctl -H root@foobar:waldi" will now connect as
1464 user "root" to host "foobar", and then proceed directly to
1465 the container named "waldi". Note that currently you have to
1466 authenticate as user "root" for this to work, as entering
1467 containers is a privileged operation.
1469 Contributions from: Andreas Henriksson, Benjamin Steinwender,
1470 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
1471 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
1472 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
1473 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
1474 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
1475 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
1476 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
1477 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
1478 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
1479 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
1480 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
1482 -- Berlin, 2014-07-03
1486 * As an experimental feature, udev now tries to lock the
1487 disk device node (flock(LOCK_SH|LOCK_NB)) while it
1488 executes events for the disk or any of its partitions.
1489 Applications like partitioning programs can lock the
1490 disk device node (flock(LOCK_EX)) and claim temporary
1491 device ownership that way; udev will entirely skip all event
1492 handling for this disk and its partitions. If the disk
1493 was opened for writing, the close will trigger a partition
1494 table rescan in udev's "watch" facility, and if needed
1495 synthesize "change" events for the disk and all its partitions.
1496 This is now unconditionally enabled, and if it turns out to
1497 cause major problems, we might turn it on only for specific
1498 devices, or might need to disable it entirely. Device Mapper
1499 devices are excluded from this logic.
1501 * We temporarily dropped the "-l" switch for fsck invocations,
1502 since they collide with the flock() logic above. util-linux
1503 upstream has been changed already to avoid this conflict,
1504 and we will readd "-l" as soon as util-linux with this
1505 change has been released.
1507 * The dependency on libattr has been removed. Since a long
1508 time, the extended attribute calls have moved to glibc, and
1509 libattr is thus unnecessary.
1511 * Virtualization detection works without priviliges now. This
1512 means the systemd-detect-virt binary no longer requires
1513 CAP_SYS_PTRACE file capabilities, and our daemons can run
1514 with fewer privileges.
1516 * systemd-networkd now runs under its own "systemd-network"
1517 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
1518 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
1519 loses the ability to write to files owned by root this way.
1521 * Similar, systemd-resolved now runs under its own
1522 "systemd-resolve" user with no capabilities remaining.
1524 * Similar, systemd-bus-proxyd now runs under its own
1525 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
1527 * systemd-networkd gained support for setting up "veth"
1528 virtual ethernet devices for container connectivity, as well
1529 as GRE and VTI tunnels.
1531 * systemd-networkd will no longer automatically attempt to
1532 manually load kernel modules necessary for certain tunnel
1533 transports. Instead, it is assumed the kernel loads them
1534 automatically when required. This only works correctly on
1535 very new kernels. On older kernels, please consider adding
1536 the kernel modules to /etc/modules-load.d/ as a work-around.
1538 * The resolv.conf file systemd-resolved generates has been
1539 moved to /run/systemd/resolve/. If you have a symlink from
1540 /etc/resolv.conf, it might be necessary to correct it.
1542 * Two new service settings, ProtectHome= and ProtectSystem=,
1543 have been added. When enabled, they will make the user data
1544 (such as /home) inaccessible or read-only and the system
1545 (such as /usr) read-only, for specific services. This allows
1546 very light-weight per-service sandboxing to avoid
1547 modifications of user data or system files from
1548 services. These two new switches have been enabled for all
1549 of systemd's long-running services, where appropriate.
1551 * Socket units gained new SocketUser= and SocketGroup=
1552 settings to set the owner user and group of AF_UNIX sockets
1553 and FIFOs in the file system.
1555 * Socket units gained a new RemoveOnStop= setting. If enabled,
1556 all FIFOS and sockets in the file system will be removed
1557 when the specific socket unit is stopped.
1559 * Socket units gained a new Symlinks= setting. It takes a list
1560 of symlinks to create to file system sockets or FIFOs
1561 created by the specific Unix sockets. This is useful to
1562 manage symlinks to socket nodes with the same life-cycle as
1565 * The /dev/log socket and /dev/initctl FIFO have been moved to
1566 /run, and have been replaced by symlinks. This allows
1567 connecting to these facilities even if PrivateDevices=yes is
1568 used for a service (which makes /dev/log itself unavailable,
1569 but /run is left). This also has the benefit of ensuring
1570 that /dev only contains device nodes, directories and
1571 symlinks, and nothing else.
1573 * sd-daemon gained two new calls sd_pid_notify() and
1574 sd_pid_notifyf(). They are similar to sd_notify() and
1575 sd_notifyf(), but allow overriding of the source PID of
1576 notification messages if permissions permit this. This is
1577 useful to send notify messages on behalf of a different
1578 process (for example, the parent process). The
1579 systemd-notify tool has been updated to make use of this
1580 when sending messages (so that notification messages now
1581 originate from the shell script invoking systemd-notify and
1582 not the systemd-notify process itself. This should minimize
1583 a race where systemd fails to associate notification
1584 messages to services when the originating process already
1587 * A new "on-abnormal" setting for Restart= has been added. If
1588 set, it will result in automatic restarts on all "abnormal"
1589 reasons for a process to exit, which includes unclean
1590 signals, core dumps, timeouts and watchdog timeouts, but
1591 does not include clean and unclean exit codes or clean
1592 signals. Restart=on-abnormal is an alternative for
1593 Restart=on-failure for services that shall be able to
1594 terminate and avoid restarts on certain errors, by
1595 indicating so with an unclean exit code. Restart=on-failure
1596 or Restart=on-abnormal is now the recommended setting for
1597 all long-running services.
1599 * If the InaccessibleDirectories= service setting points to a
1600 mount point (or if there are any submounts contained within
1601 it), it is now attempted to completely unmount it, to make
1602 the file systems truly unavailable for the respective
1605 * The ReadOnlyDirectories= service setting and
1606 systemd-nspawn's --read-only parameter are now recursively
1607 applied to all submounts, too.
1609 * Mount units may now be created transiently via the bus APIs.
1611 * The support for SysV and LSB init scripts has been removed
1612 from the systemd daemon itself. Instead, it is now
1613 implemented as a generator that creates native systemd units
1614 from these scripts when needed. This enables us to remove a
1615 substantial amount of legacy code from PID 1, following the
1616 fact that many distributions only ship a very small number
1617 of LSB/SysV init scripts nowadays.
1619 * Privileged Xen (dom0) domains are not considered
1620 virtualization anymore by the virtualization detection
1621 logic. After all, they generally have unrestricted access to
1622 the hardware and usually are used to manage the unprivileged
1625 * systemd-tmpfiles gained a new "C" line type, for copying
1626 files or entire directories.
1628 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
1629 lines. So far, they have been non-globbing versions of the
1630 latter, and have thus been redundant. In future, it is
1631 recommended to only use "z". "m" has hence been removed
1632 from the documentation, even though it stays supported.
1634 * A tmpfiles snippet to recreate the most basic structure in
1635 /var has been added. This is enough to create the /var/run →
1636 /run symlink and create a couple of structural
1637 directories. This allows systems to boot up with an empty or
1638 volatile /var. Of course, while with this change, the core OS
1639 now is capable with dealing with a volatile /var, not all
1640 user services are ready for it. However, we hope that sooner
1641 or later, many service daemons will be changed upstream so
1642 that they are able to automatically create their necessary
1643 directories in /var at boot, should they be missing. This is
1644 the first step to allow state-less systems that only require
1645 the vendor image for /usr to boot.
1647 * systemd-nspawn has gained a new --tmpfs= switch to mount an
1648 empty tmpfs instance to a specific directory. This is
1649 particularly useful for making use of the automatic
1650 reconstruction of /var (see above), by passing --tmpfs=/var.
1652 * Access modes specified in tmpfiles snippets may now be
1653 prefixed with "~", which indicates that they shall be masked
1654 by whether the existing file or directory is currently
1655 writable, readable or executable at all. Also, if specified,
1656 the sgid/suid/sticky bits will be masked for all
1659 * A new passive target unit "network-pre.target" has been
1660 added which is useful for services that shall run before any
1661 network is configured, for example firewall scripts.
1663 * The "floppy" group that previously owned the /dev/fd*
1664 devices is no longer used. The "disk" group is now used
1665 instead. Distributions should probably deprecate usage of
1668 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
1669 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
1670 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
1671 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
1672 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
1673 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
1676 -- Berlin, 2014-06-11
1680 * A new "systemd-timesyncd" daemon has been added for
1681 synchronizing the system clock across the network. It
1682 implements an SNTP client. In contrast to NTP
1683 implementations such as chrony or the NTP reference server,
1684 this only implements a client side, and does not bother with
1685 the full NTP complexity, focusing only on querying time from
1686 one remote server and synchronizing the local clock to
1687 it. Unless you intend to serve NTP to networked clients or
1688 want to connect to local hardware clocks, this simple NTP
1689 client should be more than appropriate for most
1690 installations. The daemon runs with minimal privileges, and
1691 has been hooked up with networkd to only operate when
1692 network connectivity is available. The daemon saves the
1693 current clock to disk every time a new NTP sync has been
1694 acquired, and uses this to possibly correct the system clock
1695 early at bootup, in order to accommodate for systems that
1696 lack an RTC such as the Raspberry Pi and embedded devices,
1697 and to make sure that time monotonically progresses on these
1698 systems, even if it is not always correct. To make use of
1699 this daemon, a new system user and group "systemd-timesync"
1700 needs to be created on installation of systemd.
1702 * The queue "seqnum" interface of libudev has been disabled, as
1703 it was generally incompatible with device namespacing as
1704 sequence numbers of devices go "missing" if the devices are
1705 part of a different namespace.
1707 * "systemctl list-timers" and "systemctl list-sockets" gained
1708 a --recursive switch for showing units of these types also
1709 for all local containers, similar in style to the already
1710 supported --recursive switch for "systemctl list-units".
1712 * A new RebootArgument= setting has been added for service
1713 units, which may be used to specify a kernel reboot argument
1714 to use when triggering reboots with StartLimitAction=.
1716 * A new FailureAction= setting has been added for service
1717 units which may be used to specify an operation to trigger
1718 when a service fails. This works similarly to
1719 StartLimitAction=, but unlike it, controls what is done
1720 immediately rather than only after several attempts to
1721 restart the service in question.
1723 * hostnamed got updated to also expose the kernel name,
1724 release, and version on the bus. This is useful for
1725 executing commands like hostnamectl with the -H switch.
1726 systemd-analyze makes use of this to properly display
1727 details when running non-locally.
1729 * The bootchart tool can now show cgroup information in the
1730 graphs it generates.
1732 * The CFS CPU quota cgroup attribute is now exposed for
1733 services. The new CPUQuota= switch has been added for this
1734 which takes a percentage value. Setting this will have the
1735 result that a service may never get more CPU time than the
1736 specified percentage, even if the machine is otherwise idle.
1738 * systemd-networkd learned IPIP and SIT tunnel support.
1740 * LSB init scripts exposing a dependency on $network will now
1741 get a dependency on network-online.target rather than simply
1742 network.target. This should bring LSB handling closer to
1743 what it was on SysV systems.
1745 * A new fsck.repair= kernel option has been added to control
1746 how fsck shall deal with unclean file systems at boot.
1748 * The (.ini) configuration file parser will now silently
1749 ignore sections whose name begins with "X-". This may be
1750 used to maintain application-specific extension sections in unit
1753 * machined gained a new API to query the IP addresses of
1754 registered containers. "machinectl status" has been updated
1755 to show these addresses in its output.
1757 * A new call sd_uid_get_display() has been added to the
1758 sd-login APIs for querying the "primary" session of a
1759 user. The "primary" session of the user is elected from the
1760 user's sessions and generally a graphical session is
1761 preferred over a text one.
1763 * A minimal systemd-resolved daemon has been added. It
1764 currently simply acts as a companion to systemd-networkd and
1765 manages resolv.conf based on per-interface DNS
1766 configuration, possibly supplied via DHCP. In the long run
1767 we hope to extend this into a local DNSSEC enabled DNS and
1770 * The systemd-networkd-wait-online tool is now enabled by
1771 default. It will delay network-online.target until a network
1772 connection has been configured. The tool primarily integrates
1773 with networkd, but will also make a best effort to make sense
1774 of network configuration performed in some other way.
1776 * Two new service options StartupCPUShares= and
1777 StartupBlockIOWeight= have been added that work similarly to
1778 CPUShares= and BlockIOWeight= however only apply during
1779 system startup. This is useful to prioritize certain services
1780 differently during bootup than during normal runtime.
1782 * hostnamed has been changed to prefer the statically
1783 configured hostname in /etc/hostname (unless set to
1784 'localhost' or empty) over any dynamic one supplied by
1785 dhcp. With this change, the rules for picking the hostname
1786 match more closely the rules of other configuration settings
1787 where the local administrator's configuration in /etc always
1788 overrides any other settings.
1790 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
1791 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
1792 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
1793 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
1794 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
1795 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
1796 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
1797 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
1798 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
1799 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
1800 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
1801 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
1802 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
1803 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
1804 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
1805 Lindskog, WaLyong Cho, Will Woods, Zbigniew
1808 -- Beijing, 2014-05-28
1812 * When restoring the screen brightness at boot, stay away from
1813 the darkest setting or from the lowest 5% of the available
1814 range, depending on which is the larger value of both. This
1815 should effectively protect the user from rebooting into a
1816 black screen, should the brightness have been set to minimum
1819 * sd-login gained a new sd_machine_get_class() call to
1820 determine the class ("vm" or "container") of a machine
1821 registered with machined.
1823 * sd-login gained new calls
1824 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
1825 to query the identity of the peer of a local AF_UNIX
1826 connection. They operate similarly to their sd_pid_get_xyz()
1829 * PID 1 will now maintain a system-wide system state engine
1830 with the states "starting", "running", "degraded",
1831 "maintenance", "stopping". These states are bound to system
1832 startup, normal runtime, runtime with at least one failed
1833 service, rescue/emergency mode and system shutdown. This
1834 state is shown in the "systemctl status" output when no unit
1835 name is passed. It is useful to determine system state, in
1836 particularly when doing so for many systems or containers at
1839 * A new command "list-machines" has been added to "systemctl"
1840 that lists all local OS containers and shows their system
1841 state (see above), if systemd runs inside of them.
1843 * systemctl gained a new "-r" switch to recursively enumerate
1844 units on all local containers, when used with the
1845 "list-unit" command (which is the default one that is
1846 executed when no parameters are specified).
1848 * The GPT automatic partition discovery logic will now honour
1849 two GPT partition flags: one may be set on a partition to
1850 cause it to be mounted read-only, and the other may be set
1851 on a partition to ignore it during automatic discovery.
1853 * Two new GPT type UUIDs have been added for automatic root
1854 partition discovery, for 32-bit and 64-bit ARM. This is not
1855 particularly useful for discovering the root directory on
1856 these architectures during bare-metal boots (since UEFI is
1857 not common there), but still very useful to allow booting of
1858 ARM disk images in nspawn with the -i option.
1860 * MAC addresses of interfaces created with nspawn's
1861 --network-interface= switch will now be generated from the
1862 machine name, and thus be stable between multiple invocations
1865 * logind will now automatically remove all IPC objects owned
1866 by a user if she or he fully logs out. This makes sure that
1867 users who are logged out cannot continue to consume IPC
1868 resources. This covers SysV memory, semaphores and message
1869 queues as well as POSIX shared memory and message
1870 queues. Traditionally, SysV and POSIX IPC had no life-cycle
1871 limits. With this functionality, that is corrected. This may
1872 be turned off by using the RemoveIPC= switch of logind.conf.
1874 * The systemd-machine-id-setup and tmpfiles tools gained a
1875 --root= switch to operate on a specific root directory,
1878 * journald can now forward logged messages to the TTYs of all
1879 logged in users ("wall"). This is the default for all
1880 emergency messages now.
1882 * A new tool systemd-journal-remote has been added to stream
1883 journal log messages across the network.
1885 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
1886 controller trees are mounted into it. Note that the
1887 directories mounted beneath it are not read-only. This is a
1888 security measure and is particularly useful because glibc
1889 actually includes a search logic to pick any tmpfs it can
1890 find to implement shm_open() if /dev/shm is not available
1891 (which it might very well be in namespaced setups).
1893 * machinectl gained a new "poweroff" command to cleanly power
1894 down a local OS container.
1896 * The PrivateDevices= unit file setting will now also drop the
1897 CAP_MKNOD capability from the capability bound set, and
1898 imply DevicePolicy=closed.
1900 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
1901 comprehensively on all long-running systemd services where
1902 this is appropriate.
1904 * systemd-udevd will now run in a disassociated mount
1905 namespace. To mount directories from udev rules, make sure to
1906 pull in mount units via SYSTEMD_WANTS properties.
1908 * The kdbus support gained support for uploading policy into
1909 the kernel. sd-bus gained support for creating "monitoring"
1910 connections that can eavesdrop into all bus communication
1911 for debugging purposes.
1913 * Timestamps may now be specified in seconds since the UNIX
1914 epoch Jan 1st, 1970 by specifying "@" followed by the value
1917 * Native tcpwrap support in systemd has been removed. tcpwrap
1918 is old code, not really maintained anymore and has serious
1919 shortcomings, and better options such as firewalls
1920 exist. For setups that require tcpwrap usage, please
1921 consider invoking your socket-activated service via tcpd,
1922 like on traditional inetd.
1924 * A new system.conf configuration option
1925 DefaultTimerAccuracySec= has been added that controls the
1926 default AccuracySec= setting of .timer units.
1928 * Timer units gained a new WakeSystem= switch. If enabled,
1929 timers configured this way will cause the system to resume
1930 from system suspend (if the system supports that, which most
1933 * Timer units gained a new Persistent= switch. If enabled,
1934 timers configured this way will save to disk when they have
1935 been last triggered. This information is then used on next
1936 reboot to possible execute overdue timer events, that
1937 could not take place because the system was powered off.
1938 This enables simple anacron-like behaviour for timer units.
1940 * systemctl's "list-timers" will now also list the time a
1941 timer unit was last triggered in addition to the next time
1942 it will be triggered.
1944 * systemd-networkd will now assign predictable IPv4LL
1945 addresses to its local interfaces.
1947 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
1948 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
1949 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
1950 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
1951 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
1952 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
1953 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
1954 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
1957 -- Berlin, 2014-03-25
1961 * A new unit file setting RestrictAddressFamilies= has been
1962 added to restrict which socket address families unit
1963 processes gain access to. This takes address family names
1964 like "AF_INET" or "AF_UNIX", and is useful to minimize the
1965 attack surface of services via exotic protocol stacks. This
1966 is built on seccomp system call filters.
1968 * Two new unit file settings RuntimeDirectory= and
1969 RuntimeDirectoryMode= have been added that may be used to
1970 manage a per-daemon runtime directories below /run. This is
1971 an alternative for setting up directory permissions with
1972 tmpfiles snippets, and has the advantage that the runtime
1973 directory's lifetime is bound to the daemon runtime and that
1974 the daemon starts up with an empty directory each time. This
1975 is particularly useful when writing services that drop
1976 privileges using the User= or Group= setting.
1978 * The DeviceAllow= unit setting now supports globbing for
1979 matching against device group names.
1981 * The systemd configuration file system.conf gained new
1982 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
1983 DefaultMemoryAccounting= to globally turn on/off accounting
1984 for specific resources (cgroups) for all units. These
1985 settings may still be overridden individually in each unit
1988 * systemd-gpt-auto-generator is now able to discover /srv and
1989 root partitions in addition to /home and swap partitions. It
1990 also supports LUKS-encrypted partitions now. With this in
1991 place, automatic discovery of partitions to mount following
1992 the Discoverable Partitions Specification
1993 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
1994 is now a lot more complete. This allows booting without
1995 /etc/fstab and without root= on the kernel command line on
1996 systems prepared appropriately.
1998 * systemd-nspawn gained a new --image= switch which allows
1999 booting up disk images and Linux installations on any block
2000 device that follow the Discoverable Partitions Specification
2001 (see above). This means that installations made with
2002 appropriately updated installers may now be started and
2003 deployed using container managers, completely
2004 unmodified. (We hope that libvirt-lxc will add support for
2005 this feature soon, too.)
2007 * systemd-nspawn gained a new --network-macvlan= setting to
2008 set up a private macvlan interface for the
2009 container. Similarly, systemd-networkd gained a new
2010 Kind=macvlan setting in .netdev files.
2012 * systemd-networkd now supports configuring local addresses
2015 * A new tool systemd-network-wait-online has been added to
2016 synchronously wait for network connectivity using
2019 * The sd-bus.h bus API gained a new sd_bus_track object for
2020 tracking the life-cycle of bus peers. Note that sd-bus.h is
2021 still not a public API though (unless you specify
2022 --enable-kdbus on the configure command line, which however
2023 voids your warranty and you get no API stability guarantee).
2025 * The $XDG_RUNTIME_DIR runtime directories for each user are
2026 now individual tmpfs instances, which has the benefit of
2027 introducing separate pools for each user, with individual
2028 size limits, and thus making sure that unprivileged clients
2029 can no longer negatively impact the system or other users by
2030 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
2031 RuntimeDirectorySize= has been introduced that allows
2032 controlling the default size limit for all users. It
2033 defaults to 10% of the available physical memory. This is no
2034 replacement for quotas on tmpfs though (which the kernel
2035 still does not support), as /dev/shm and /tmp are still
2036 shared resources used by both the system and unprivileged
2039 * logind will now automatically turn off automatic suspending
2040 on laptop lid close when more than one display is
2041 connected. This was previously expected to be implemented
2042 individually in desktop environments (such as GNOME),
2043 however has been added to logind now, in order to fix a
2044 boot-time race where a desktop environment might not have
2045 been started yet and thus not been able to take an inhibitor
2046 lock at the time where logind already suspends the system
2047 due to a closed lid.
2049 * logind will now wait at least 30s after each system
2050 suspend/resume cycle, and 3min after system boot before
2051 suspending the system due to a closed laptop lid. This
2052 should give USB docking stations and similar enough time to
2053 be probed and configured after system resume and boot in
2054 order to then act as suspend blocker.
2056 * systemd-run gained a new --property= setting which allows
2057 initialization of resource control properties (and others)
2058 for the created scope or service unit. Example: "systemd-run
2059 --property=BlockIOWeight=10 updatedb" may be used to run
2060 updatedb at a low block IO scheduling weight.
2062 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
2063 now also work in --scope mode.
2065 * When systemd is compiled with kdbus support, basic support
2066 for enforced policies is now in place. (Note that enabling
2067 kdbus still voids your warranty and no API compatibility
2070 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
2071 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
2072 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
2073 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
2074 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
2075 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
2076 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
2077 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
2078 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
2079 Zbigniew Jędrzejewski-Szmek
2081 -- Berlin, 2014-03-12
2085 * systemd will now relabel /dev after loading the SMACK policy
2086 according to SMACK rules.
2088 * A new unit file option AppArmorProfile= has been added to
2089 set the AppArmor profile for the processes of a unit.
2091 * A new condition check ConditionArchitecture= has been added
2092 to conditionalize units based on the system architecture, as
2093 reported by uname()'s "machine" field.
2095 * systemd-networkd now supports matching on the system
2096 virtualization, architecture, kernel command line, host name
2099 * logind is now a lot more aggressive when suspending the
2100 machine due to a closed laptop lid. Instead of acting only
2101 on the lid close action, it will continuously watch the lid
2102 status and act on it. This is useful for laptops where the
2103 power button is on the outside of the chassis so that it can
2104 be reached without opening the lid (such as the Lenovo
2105 Yoga). On those machines, logind will now immediately
2106 re-suspend the machine if the power button has been
2107 accidentally pressed while the laptop was suspended and in a
2108 backpack or similar.
2110 * logind will now watch SW_DOCK switches and inhibit reaction
2111 to the lid switch if it is pressed. This means that logind
2112 will not suspend the machine anymore if the lid is closed
2113 and the system is docked, if the laptop supports SW_DOCK
2114 notifications via the input layer. Note that ACPI docking
2115 stations do not generate this currently. Also note that this
2116 logic is usually not fully sufficient and Desktop
2117 Environments should take a lid switch inhibitor lock when an
2118 external display is connected, as systemd will not watch
2121 * nspawn will now make use of the devices cgroup controller by
2122 default, and only permit creation of and access to the usual
2123 API device nodes like /dev/null or /dev/random, as well as
2124 access to (but not creation of) the pty devices.
2126 * We will now ship a default .network file for
2127 systemd-networkd that automatically configures DHCP for
2128 network interfaces created by nspawn's --network-veth or
2129 --network-bridge= switches.
2131 * systemd will now understand the usual M, K, G, T suffixes
2132 according to SI conventions (i.e. to the base 1000) when
2133 referring to throughput and hardware metrics. It will stay
2134 with IEC conventions (i.e. to the base 1024) for software
2135 metrics, according to what is customary according to
2136 Wikipedia. We explicitly document which base applies for
2137 each configuration option.
2139 * The DeviceAllow= setting in unit files now supports a syntax
2140 to whitelist an entire group of devices node majors at once,
2141 based on the /proc/devices listing. For example, with the
2142 string "char-pts", it is now possible to whitelist all
2143 current and future pseudo-TTYs at once.
2145 * sd-event learned a new "post" event source. Event sources of
2146 this type are triggered by the dispatching of any event
2147 source of a type that is not "post". This is useful for
2148 implementing clean-up and check event sources that are
2149 triggered by other work being done in the program.
2151 * systemd-networkd is no longer statically enabled, but uses
2152 the usual [Install] sections so that it can be
2153 enabled/disabled using systemctl. It still is enabled by
2156 * When creating a veth interface pair with systemd-nspawn, the
2157 host side will now be prefixed with "vb-" if
2158 --network-bridge= is used, and with "ve-" if --network-veth
2159 is used. This way, it is easy to distinguish these cases on
2160 the host, for example to apply different configuration to
2161 them with systemd-networkd.
2163 * The compatibility libraries for libsystemd-journal.so,
2164 libsystem-id128.so, libsystemd-login.so and
2165 libsystemd-daemon.so do not make use of IFUNC
2166 anymore. Instead, we now build libsystemd.so multiple times
2167 under these alternative names. This means that the footprint
2168 is drastically increased, but given that these are
2169 transitional compatibility libraries, this should not matter
2170 much. This change has been made necessary to support the ARM
2171 platform for these compatibility libraries, as the ARM
2172 toolchain is not really at the same level as the toolchain
2173 for other architectures like x86 and does not support
2174 IFUNC. Please make sure to use --enable-compat-libs only
2175 during a transitional period!
2177 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
2178 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
2179 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
2180 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
2181 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
2182 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
2183 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
2184 Zbigniew Jędrzejewski-Szmek
2186 -- Berlin, 2014-02-24
2190 * A new component "systemd-networkd" has been added that can
2191 be used to configure local network interfaces statically or
2192 via DHCP. It is capable of bringing up bridges, VLANs, and
2193 bonding. Currently, no hook-ups for interactive network
2194 configuration are provided. Use this for your initrd,
2195 container, embedded, or server setup if you need a simple,
2196 yet powerful, network configuration solution. This
2197 configuration subsystem is quite nifty, as it allows wildcard
2198 hotplug matching in interfaces. For example, with a single
2199 configuration snippet, you can configure that all Ethernet
2200 interfaces showing up are automatically added to a bridge,
2201 or similar. It supports link-sensing and more.
2203 * A new tool "systemd-socket-proxyd" has been added which can
2204 act as a bidirectional proxy for TCP sockets. This is
2205 useful for adding socket activation support to services that
2206 do not actually support socket activation, including virtual
2207 machines and the like.
2209 * Add a new tool to save/restore rfkill state on
2212 * Save/restore state of keyboard backlights in addition to
2213 display backlights on shutdown/boot.
2215 * udev learned a new SECLABEL{} construct to label device
2216 nodes with a specific security label when they appear. For
2217 now, only SECLABEL{selinux} is supported, but the syntax is
2218 prepared for additional security frameworks.
2220 * udev gained a new scheme to configure link-level attributes
2221 from files in /etc/systemd/network/*.link. These files can
2222 match against MAC address, device path, driver name and type,
2223 and will apply attributes like the naming policy, link speed,
2224 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
2225 address assignment policy (randomized, ...).
2227 * The configuration of network interface naming rules for
2228 "permanent interface names" has changed: a new NamePolicy=
2229 setting in the [Link] section of .link files determines the
2230 priority of possible naming schemes (onboard, slot, mac,
2231 path). The default value of this setting is determined by
2232 /usr/lib/net/links/99-default.link. Old
2233 80-net-name-slot.rules udev configuration file has been
2234 removed, so local configuration overriding this file should
2235 be adapated to override 99-default.link instead.
2237 * When the User= switch is used in a unit file, also
2238 initialize $SHELL= based on the user database entry.
2240 * systemd no longer depends on libdbus. All communication is
2241 now done with sd-bus, systemd's low-level bus library
2244 * kdbus support has been added to PID 1 itself. When kdbus is
2245 enabled, this causes PID 1 to set up the system bus and
2246 enable support for a new ".busname" unit type that
2247 encapsulates bus name activation on kdbus. It works a little
2248 bit like ".socket" units, except for bus names. A new
2249 generator has been added that converts classic dbus1 service
2250 activation files automatically into native systemd .busname
2253 * sd-bus: add a light-weight vtable implementation that allows
2254 defining objects on the bus with a simple static const
2255 vtable array of its methods, signals and properties.
2257 * systemd will not generate or install static dbus
2258 introspection data anymore to /usr/share/dbus-1/interfaces,
2259 as the precise format of these files is unclear, and
2260 nothing makes use of it.
2262 * A proxy daemon is now provided to proxy clients connecting
2263 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
2264 compatibility with classic D-Bus.
2266 * A bus driver implementation has been added that supports the
2267 classic D-Bus bus driver calls on kdbus, also for
2268 compatibility purposes.
2270 * A new API "sd-event.h" has been added that implements a
2271 minimal event loop API built around epoll. It provides a
2272 couple of features that direct epoll usage is lacking:
2273 prioritization of events, scales to large numbers of timer
2274 events, per-event timer slack (accuracy), system-wide
2275 coalescing of timer events, exit handlers, watchdog
2276 supervision support using systemd's sd_notify() API, child
2279 * A new API "sd-rntl.h" has been added that provides an API
2280 around the route netlink interface of the kernel, similar in
2281 style to "sd-bus.h".
2283 * A new API "sd-dhcp-client.h" has been added that provides a
2284 small DHCPv4 client-side implementation. This is used by
2287 * There is a new kernel command line option
2288 "systemd.restore_state=0|1". When set to "0", none of the
2289 systemd tools will restore saved runtime state to hardware
2290 devices. More specifically, the rfkill and backlight states
2293 * The FsckPassNo= compatibility option in mount/service units
2294 has been removed. The fstab generator will now add the
2295 necessary dependencies automatically, and does not require
2296 PID1's support for that anymore.
2298 * journalctl gained a new switch, --list-boots, that lists
2299 recent boots with their times and boot IDs.
2301 * The various tools like systemctl, loginctl, timedatectl,
2302 busctl, systemd-run, ... have gained a new switch "-M" to
2303 connect to a specific, local OS container (as direct
2304 connection, without requiring SSH). This works on any
2305 container that is registered with machined, such as those
2306 created by libvirt-lxc or nspawn.
2308 * systemd-run and systemd-analyze also gained support for "-H"
2309 to connect to remote hosts via SSH. This is particularly
2310 useful for systemd-run because it enables queuing of jobs
2311 onto remote systems.
2313 * machinectl gained a new command "login" to open a getty
2314 login in any local container. This works with any container
2315 that is registered with machined (such as those created by
2316 libvirt-lxc or nspawn), and which runs systemd inside.
2318 * machinectl gained a new "reboot" command that may be used to
2319 trigger a reboot on a specific container that is registered
2320 with machined. This works on any container that runs an init
2321 system of some kind.
2323 * systemctl gained a new "list-timers" command to print a nice
2324 listing of installed timer units with the times they elapse
2327 * Alternative reboot() parameters may now be specified on the
2328 "systemctl reboot" command line and are passed to the
2329 reboot() system call.
2331 * systemctl gained a new --job-mode= switch to configure the
2332 mode to queue a job with. This is a more generic version of
2333 --fail, --irreversible, and --ignore-dependencies, which are
2334 still available but not advertised anymore.
2336 * /etc/systemd/system.conf gained new settings to configure
2337 various default timeouts of units, as well as the default
2338 start limit interval and burst. These may still be overridden
2341 * PID1 will now export on the bus profile data of the security
2342 policy upload process (such as the SELinux policy upload to
2345 * journald: when forwarding logs to the console, include
2346 timestamps (following the setting in
2347 /sys/module/printk/parameters/time).
2349 * OnCalendar= in timer units now understands the special
2350 strings "yearly" and "annually". (Both are equivalent)
2352 * The accuracy of timer units is now configurable with the new
2353 AccuracySec= setting. It defaults to 1min.
2355 * A new dependency type JoinsNamespaceOf= has been added that
2356 allows running two services within the same /tmp and network
2357 namespace, if PrivateNetwork= or PrivateTmp= are used.
2359 * A new command "cat" has been added to systemctl. It outputs
2360 the original unit file of a unit, and concatenates the
2361 contents of additional "drop-in" unit file snippets, so that
2362 the full configuration is shown.
2364 * systemctl now supports globbing on the various "list-xyz"
2365 commands, like "list-units" or "list-sockets", as well as on
2366 those commands which take multiple unit names.
2368 * journalctl's --unit= switch gained support for globbing.
2370 * All systemd daemons now make use of the watchdog logic so
2371 that systemd automatically notices when they hang.
2373 * If the $container_ttys environment variable is set,
2374 getty-generator will automatically spawn a getty for each
2375 listed tty. This is useful for container managers to request
2376 login gettys to be spawned on as many ttys as needed.
2378 * %h, %s, %U specifier support is not available anymore when
2379 used in unit files for PID 1. This is because NSS calls are
2380 not safe from PID 1. They stay available for --user
2381 instances of systemd, and as special case for the root user.
2383 * loginctl gained a new "--no-legend" switch to turn off output
2386 * The "sd-login.h" API gained three new calls:
2387 sd_session_is_remote(), sd_session_get_remote_user(),
2388 sd_session_get_remote_host() to query information about
2391 * The udev hardware database now also carries vendor/product
2392 information of SDIO devices.
2394 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
2395 determine whether watchdog notifications are requested by
2398 * Socket-activated per-connection services now include a
2399 short description of the connection parameters in the
2402 * tmpfiles gained a new "--boot" option. When this is not used,
2403 only lines where the command character is not suffixed with
2404 "!" are executed. When this option is specified, those
2405 options are executed too. This partitions tmpfiles
2406 directives into those that can be safely executed at any
2407 time, and those which should be run only at boot (for
2408 example, a line that creates /run/nologin).
2410 * A new API "sd-resolve.h" has been added which provides a simple
2411 asynchronous wrapper around glibc NSS host name resolution
2412 calls, such as getaddrinfo(). In contrast to glibc's
2413 getaddrinfo_a(), it does not use signals. In contrast to most
2414 other asynchronous name resolution libraries, this one does
2415 not reimplement DNS, but reuses NSS, so that alternate
2416 host name resolution systems continue to work, such as mDNS,
2417 LDAP, etc. This API is based on libasyncns, but it has been
2418 cleaned up for inclusion in systemd.
2420 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
2421 "sd-daemon.h" are no longer found in individual libraries
2422 libsystemd-journal.so, libsystemd-login.so,
2423 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
2424 merged them into a single library, libsystemd.so, which
2425 provides all symbols. The reason for this is cyclic
2426 dependencies, as these libraries tend to use each other's
2427 symbols. So far, we have managed to workaround that by linking
2428 a copy of a good part of our code into each of these
2429 libraries again and again, which, however, makes certain
2430 things hard to do, like sharing static variables. Also, it
2431 substantially increases footprint. With this change, there
2432 is only one library for the basic APIs systemd
2433 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
2434 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
2435 library as well, however are subject to the --enable-kdbus
2436 switch (see below). Note that "sd-dhcp-client.h" is not part
2437 of this library (this is because it only consumes, never
2438 provides, services of/to other APIs). To make the transition
2439 easy from the separate libraries to the unified one, we
2440 provide the --enable-compat-libs compile-time switch which
2441 will generate stub libraries that are compatible with the
2442 old ones but redirect all calls to the new one.
2444 * All of the kdbus logic and the new APIs "sd-bus.h",
2445 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
2446 and "sd-utf8.h" are compile-time optional via the
2447 "--enable-kdbus" switch, and they are not compiled in by
2448 default. To make use of kdbus, you have to explicitly enable
2449 the switch. Note however, that neither the kernel nor the
2450 userspace API for all of this is considered stable yet. We
2451 want to maintain the freedom to still change the APIs for
2452 now. By specifying this build-time switch, you acknowledge
2453 that you are aware of the instability of the current
2456 * Also, note that while kdbus is pretty much complete,
2457 it lacks one thing: proper policy support. This means you
2458 can build a fully working system with all features; however,
2459 it will be highly insecure. Policy support will be added in
2460 one of the next releases, at the same time that we will
2461 declare the APIs stable.
2463 * When the kernel command line argument "kdbus" is specified,
2464 systemd will automatically load the kdbus.ko kernel module. At
2465 this stage of development, it is only useful for testing kdbus
2466 and should not be used in production. Note: if "--enable-kdbus"
2467 is specified, and the kdbus.ko kernel module is available, and
2468 "kdbus" is added to the kernel command line, the entire system
2469 runs with kdbus instead of dbus-daemon, with the above mentioned
2470 problem of missing the system policy enforcement. Also a future
2471 version of kdbus.ko or a newer systemd will not be compatible with
2472 each other, and will unlikely be able to boot the machine if only
2473 one of them is updated.
2475 * systemctl gained a new "import-environment" command which
2476 uploads the caller's environment (or parts thereof) into the
2477 service manager so that it is inherited by services started
2478 by the manager. This is useful to upload variables like
2479 $DISPLAY into the user service manager.
2481 * A new PrivateDevices= switch has been added to service units
2482 which allows running a service with a namespaced /dev
2483 directory that does not contain any device nodes for
2484 physical devices. More specifically, it only includes devices
2485 such as /dev/null, /dev/urandom, and /dev/zero which are API
2488 * logind has been extended to support behaviour like VT
2489 switching on seats that do not support a VT. This makes
2490 multi-session available on seats that are not the first seat
2491 (seat0), and on systems where kernel support for VTs has
2492 been disabled at compile-time.
2494 * If a process holds a delay lock for system sleep or shutdown
2495 and fails to release it in time, we will now log its
2496 identity. This makes it easier to identify processes that
2497 cause slow suspends or power-offs.
2499 * When parsing /etc/crypttab, support for a new key-slot=
2500 option as supported by Debian is added. It allows indicating
2501 which LUKS slot to use on disk, speeding up key loading.
2503 * The sd_journald_sendv() API call has been checked and
2504 officially declared to be async-signal-safe so that it may
2505 be invoked from signal handlers for logging purposes.
2507 * Boot-time status output is now enabled automatically after a
2508 short timeout if boot does not progress, in order to give
2509 the user an indication what she or he is waiting for.
2511 * The boot-time output has been improved to show how much time
2512 remains until jobs expire.
2514 * The KillMode= switch in service units gained a new possible
2515 value "mixed". If set, and the unit is shut down, then the
2516 initial SIGTERM signal is sent only to the main daemon
2517 process, while the following SIGKILL signal is sent to
2518 all remaining processes of the service.
2520 * When a scope unit is registered, a new property "Controller"
2521 may be set. If set to a valid bus name, systemd will send a
2522 RequestStop() signal to this name when it would like to shut
2523 down the scope. This may be used to hook manager logic into
2524 the shutdown logic of scope units. Also, scope units may now
2525 be put in a special "abandoned" state, in which case the
2526 manager process which created them takes no further
2527 responsibilities for it.
2529 * When reading unit files, systemd will now verify
2530 the access mode of these files, and warn about certain
2531 suspicious combinations. This has been added to make it
2532 easier to track down packaging bugs where unit files are
2533 marked executable or world-writable.
2535 * systemd-nspawn gained a new "--setenv=" switch to set
2536 container-wide environment variables. The similar option in
2537 systemd-activate was renamed from "--environment=" to
2538 "--setenv=" for consistency.
2540 * systemd-nspawn has been updated to create a new kdbus domain
2541 for each container that is invoked, thus allowing each
2542 container to have its own set of system and user buses,
2543 independent of the host.
2545 * systemd-nspawn gained a new --drop-capability= switch to run
2546 the container with less capabilities than the default. Both
2547 --drop-capability= and --capability= now take the special
2548 string "all" for dropping or keeping all capabilities.
2550 * systemd-nspawn gained new switches for executing containers
2551 with specific SELinux labels set.
2553 * systemd-nspawn gained a new --quiet switch to not generate
2554 any additional output but the container's own console
2557 * systemd-nspawn gained a new --share-system switch to run a
2558 container without PID namespacing enabled.
2560 * systemd-nspawn gained a new --register= switch to control
2561 whether the container is registered with systemd-machined or
2562 not. This is useful for containers that do not run full
2563 OS images, but only specific apps.
2565 * systemd-nspawn gained a new --keep-unit which may be used
2566 when invoked as the only program from a service unit, and
2567 results in registration of the unit service itself in
2568 systemd-machined, instead of a newly opened scope unit.
2570 * systemd-nspawn gained a new --network-interface= switch for
2571 moving arbitrary interfaces to the container. The new
2572 --network-veth switch creates a virtual Ethernet connection
2573 between host and container. The new --network-bridge=
2574 switch then allows assigning the host side of this virtual
2575 Ethernet connection to a bridge device.
2577 * systemd-nspawn gained a new --personality= switch for
2578 setting the kernel personality for the container. This is
2579 useful when running a 32-bit container on a 64-bit host. A
2580 similar option Personality= is now also available for service
2583 * logind will now also track a "Desktop" identifier for each
2584 session which encodes the desktop environment of it. This is
2585 useful for desktop environments that want to identify
2586 multiple running sessions of itself easily.
2588 * A new SELinuxContext= setting for service units has been
2589 added that allows setting a specific SELinux execution
2590 context for a service.
2592 * Most systemd client tools will now honour $SYSTEMD_LESS for
2593 settings of the "less" pager. By default, these tools will
2594 override $LESS to allow certain operations to work, such as
2595 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
2596 influence this logic.
2598 * systemd's "seccomp" hook-up has been changed to make use of
2599 the libseccomp library instead of using its own
2600 implementation. This has benefits for portability among
2603 * For usage together with SystemCallFilter=, a new
2604 SystemCallErrorNumber= setting has been introduced that
2605 allows configuration of a system error number to be returned
2606 on filtered system calls, instead of immediately killing the
2607 process. Also, SystemCallArchitectures= has been added to
2608 limit access to system calls of a particular architecture
2609 (in order to turn off support for unused secondary
2610 architectures). There is also a global
2611 SystemCallArchitectures= setting in system.conf now to turn
2612 off support for non-native system calls system-wide.
2614 * systemd requires a kernel with a working name_to_handle_at(),
2615 please see the kernel config requirements in the README file.
2617 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
2618 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
2619 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
2620 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
2621 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
2622 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
2623 Elia Pinto, Florian Weimer, George McCollister, Goffredo
2624 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
2625 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
2626 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
2627 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
2628 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
2629 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
2630 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
2631 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
2632 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
2633 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
2634 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
2635 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
2636 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
2637 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
2638 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
2639 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
2640 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
2642 -- Berlin, 2014-02-20
2646 * logind has gained support for facilitating privileged input
2647 and drm device access for unprivileged clients. This work is
2648 useful to allow Wayland display servers (and similar
2649 programs, such as kmscon) to run under the user's ID and
2650 access input and drm devices which are normally
2651 protected. When this is used (and the kernel is new enough)
2652 logind will "mute" IO on the file descriptors passed to
2653 Wayland as long as it is in the background and "unmute" it
2654 if it returns into the foreground. This allows secure
2655 session switching without allowing background sessions to
2656 eavesdrop on input and display data. This also introduces
2657 session switching support if VT support is turned off in the
2658 kernel, and on seats that are not seat0.
2660 * A new kernel command line option luks.options= is understood
2661 now which allows specifying LUKS options for usage for LUKS
2662 encrypted partitions specified with luks.uuid=.
2664 * tmpfiles.d(5) snippets may now use specifier expansion in
2665 path names. More specifically %m, %b, %H, %v, are now
2666 replaced by the local machine id, boot id, hostname, and
2667 kernel version number.
2669 * A new tmpfiles.d(5) command "m" has been introduced which
2670 may be used to change the owner/group/access mode of a file
2671 or directory if it exists, but do nothing if it does not.
2673 * This release removes high-level support for the
2674 MemorySoftLimit= cgroup setting. The underlying kernel
2675 cgroup attribute memory.soft_limit= is currently badly
2676 designed and likely to be removed from the kernel API in its
2677 current form, hence we should not expose it for now.
2679 * The memory.use_hierarchy cgroup attribute is now enabled for
2680 all cgroups systemd creates in the memory cgroup
2681 hierarchy. This option is likely to be come the built-in
2682 default in the kernel anyway, and the non-hierarchical mode
2683 never made much sense in the intrinsically hierarchical
2686 * A new field _SYSTEMD_SLICE= is logged along with all journal
2687 messages containing the slice a message was generated
2688 from. This is useful to allow easy per-customer filtering of
2689 logs among other things.
2691 * systemd-journald will no longer adjust the group of journal
2692 files it creates to the "systemd-journal" group. Instead we
2693 rely on the journal directory to be owned by the
2694 "systemd-journal" group, and its setgid bit set, so that the
2695 kernel file system layer will automatically enforce that
2696 journal files inherit this group assignment. The reason for
2697 this change is that we cannot allow NSS look-ups from
2698 journald which would be necessary to resolve
2699 "systemd-journal" to a numeric GID, because this might
2700 create deadlocks if NSS involves synchronous queries to
2701 other daemons (such as nscd, or sssd) which in turn are
2702 logging clients of journald and might block on it, which
2703 would then dead lock. A tmpfiles.d(5) snippet included in
2704 systemd will make sure the setgid bit and group are
2705 properly set on the journal directory if it exists on every
2706 boot. However, we recommend adjusting it manually after
2707 upgrades too (or from RPM scriptlets), so that the change is
2708 not delayed until next reboot.
2710 * Backlight and random seed files in /var/lib/ have moved into
2711 the /var/lib/systemd/ directory, in order to centralize all
2712 systemd generated files in one directory.
2714 * Boot time performance measurements (as displayed by
2715 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
2716 performance information if that's available to determine how
2717 much time BIOS and boot loader initialization required. With
2718 a sufficiently new BIOS you hence no longer need to boot
2719 with Gummiboot to get access to such information.
2721 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
2722 Cristian Rodríguez, Dave Reisner, David Herrmann, David
2723 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
2724 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
2725 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
2726 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
2727 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
2729 -- Berlin, 2013-10-02
2733 * The Restart= option for services now understands a new
2734 on-watchdog setting, which will restart the service
2735 automatically if the service stops sending out watchdog keep
2736 alive messages (as configured with WatchdogSec=).
2738 * The getty generator (which is responsible for bringing up a
2739 getty on configured serial consoles) will no longer only
2740 start a getty on the primary kernel console but on all
2741 others, too. This makes the order in which console= is
2742 specified on the kernel command line less important.
2744 * libsystemd-logind gained a new sd_session_get_vt() call to
2745 retrieve the VT number of a session.
2747 * If the option "tries=0" is set for an entry of /etc/crypttab
2748 its passphrase is queried indefinitely instead of any
2749 maximum number of tries.
2751 * If a service with a configure PID file terminates its PID
2752 file will now be removed automatically if it still exists
2753 afterwards. This should put an end to stale PID files.
2755 * systemd-run will now also take relative binary path names
2756 for execution and no longer insists on absolute paths.
2758 * InaccessibleDirectories= and ReadOnlyDirectories= now take
2759 paths that are optionally prefixed with "-" to indicate that
2760 it should not be considered a failure if they do not exist.
2762 * journalctl -o (and similar commands) now understands a new
2763 output mode "short-precise", it is similar to "short" but
2764 shows timestamps with usec accuracy.
2766 * The option "discard" (as known from Debian) is now
2767 synonymous to "allow-discards" in /etc/crypttab. In fact,
2768 "discard" is preferred now (since it is easier to remember
2771 * Some licensing clean-ups were made, so that more code is now
2772 LGPL-2.1 licensed than before.
2774 * A minimal tool to save/restore the display backlight
2775 brightness across reboots has been added. It will store the
2776 backlight setting as late as possible at shutdown, and
2777 restore it as early as possible during reboot.
2779 * A logic to automatically discover and enable home and swap
2780 partitions on GPT disks has been added. With this in place
2781 /etc/fstab becomes optional for many setups as systemd can
2782 discover certain partitions located on the root disk
2783 automatically. Home partitions are recognized under their
2784 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
2785 partitions are recognized under their GPT type ID
2786 0657fd6da4ab43c484e50933c84b4f4f.
2788 * systemd will no longer pass any environment from the kernel
2789 or initrd to system services. If you want to set an
2790 environment for all services, do so via the kernel command
2791 line systemd.setenv= assignment.
2793 * The systemd-sysctl tool no longer natively reads the file
2794 /etc/sysctl.conf. If desired, the file should be symlinked
2795 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
2796 legacy support by a symlink rather than built-in code, it
2797 also makes the otherwise hidden order of application of the
2798 different files visible. (Note that this partly reverts to a
2799 pre-198 application order of sysctl knobs!)
2801 * The "systemctl set-log-level" and "systemctl dump" commands
2802 have been moved to systemd-analyze.
2804 * systemd-run learned the new --remain-after-exit switch,
2805 which causes the scope unit not to be cleaned up
2806 automatically after the process terminated.
2808 * tmpfiles learned a new --exclude-prefix= switch to exclude
2809 certain paths from operation.
2811 * journald will now automatically flush all messages to disk
2812 as soon as a message at the log level CRIT, ALERT or EMERG
2815 Contributions from: Andrew Cook, Brandon Philips, Christian
2816 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
2817 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
2818 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
2819 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
2820 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
2821 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
2822 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
2823 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
2824 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
2825 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
2826 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
2827 William Giokas, Zbigniew Jędrzejewski-Szmek
2829 -- Berlin, 2013-09-13
2833 * The documentation has been updated to cover the various new
2834 concepts introduced with 205.
2836 * Unit files now understand the new %v specifier which
2837 resolves to the kernel version string as returned by "uname
2840 * systemctl now supports filtering the unit list output by
2841 load state, active state and sub state, using the new
2844 * "systemctl status" will now show the results of the
2845 condition checks (like ConditionPathExists= and similar) of
2846 the last start attempts of the unit. They are also logged to
2849 * "journalctl -b" may now be used to look for boot output of a
2850 specific boot. Try "journalctl -b -1" for the previous boot,
2851 but the syntax is substantially more powerful.
2853 * "journalctl --show-cursor" has been added which prints the
2854 cursor string the last shown log line. This may then be used
2855 with the new "journalctl --after-cursor=" switch to continue
2856 browsing logs from that point on.
2858 * "journalctl --force" may now be used to force regeneration
2861 * Creation of "dead" device nodes has been moved from udev
2862 into kmod and tmpfiles. Previously, udev would read the kmod
2863 databases to pre-generate dead device nodes based on meta
2864 information contained in kernel modules, so that these would
2865 be auto-loaded on access rather then at boot. As this
2866 does not really have much to do with the exposing actual
2867 kernel devices to userspace this has always been slightly
2868 alien in the udev codebase. Following the new scheme kmod
2869 will now generate a runtime snippet for tmpfiles from the
2870 module meta information and it now is tmpfiles' job to the
2871 create the nodes. This also allows overriding access and
2872 other parameters for the nodes using the usual tmpfiles
2873 facilities. As side effect this allows us to remove the
2874 CAP_SYS_MKNOD capability bit from udevd entirely.
2876 * logind's device ACLs may now be applied to these "dead"
2877 devices nodes too, thus finally allowing managed access to
2878 devices such as /dev/snd/sequencer whithout loading the
2879 backing module right-away.
2881 * A new RPM macro has been added that may be used to apply
2882 tmpfiles configuration during package installation.
2884 * systemd-detect-virt and ConditionVirtualization= now can
2885 detect User-Mode-Linux machines (UML).
2887 * journald will now implicitly log the effective capabilities
2888 set of processes in the message metadata.
2890 * systemd-cryptsetup has gained support for TrueCrypt volumes.
2892 * The initrd interface has been simplified (more specifically,
2893 support for passing performance data via environment
2894 variables and fsck results via files in /run has been
2895 removed). These features were non-essential, and are
2896 nowadays available in a much nicer way by having systemd in
2897 the initrd serialize its state and have the hosts systemd
2898 deserialize it again.
2900 * The udev "keymap" data files and tools to apply keyboard
2901 specific mappings of scan to key codes, and force-release
2902 scan code lists have been entirely replaced by a udev
2903 "keyboard" builtin and a hwdb data file.
2905 * systemd will now honour the kernel's "quiet" command line
2906 argument also during late shutdown, resulting in a
2907 completely silent shutdown when used.
2909 * There's now an option to control the SO_REUSEPORT socket
2910 option in .socket units.
2912 * Instance units will now automatically get a per-template
2913 subslice of system.slice unless something else is explicitly
2914 configured. For example, instances of sshd@.service will now
2915 implicitly be placed in system-sshd.slice rather than
2916 system.slice as before.
2918 * Test coverage support may now be enabled at build time.
2920 Contributions from: Dave Reisner, Frederic Crozat, Harald
2921 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
2922 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
2923 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
2924 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
2925 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
2926 Giokas, Zbigniew Jędrzejewski-Szmek
2928 -- Berlin, 2013-07-23
2932 * Two new unit types have been introduced:
2934 Scope units are very similar to service units, however, are
2935 created out of pre-existing processes -- instead of PID 1
2936 forking off the processes. By using scope units it is
2937 possible for system services and applications to group their
2938 own child processes (worker processes) in a powerful way
2939 which then maybe used to organize them, or kill them
2940 together, or apply resource limits on them.
2942 Slice units may be used to partition system resources in an
2943 hierarchical fashion and then assign other units to them. By
2944 default there are now three slices: system.slice (for all
2945 system services), user.slice (for all user sessions),
2946 machine.slice (for VMs and containers).
2948 Slices and scopes have been introduced primarily in
2949 context of the work to move cgroup handling to a
2950 single-writer scheme, where only PID 1
2951 creates/removes/manages cgroups.
2953 * There's a new concept of "transient" units. In contrast to
2954 normal units these units are created via an API at runtime,
2955 not from configuration from disk. More specifically this
2956 means it is now possible to run arbitrary programs as
2957 independent services, with all execution parameters passed
2958 in via bus APIs rather than read from disk. Transient units
2959 make systemd substantially more dynamic then it ever was,
2960 and useful as a general batch manager.
2962 * logind has been updated to make use of scope and slice units
2963 for managing user sessions. As a user logs in he will get
2964 his own private slice unit, to which all sessions are added
2965 as scope units. We also added support for automatically
2966 adding an instance of user@.service for the user into the
2967 slice. Effectively logind will no longer create cgroup
2968 hierarchies on its own now, it will defer entirely to PID 1
2969 for this by means of scope, service and slice units. Since
2970 user sessions this way become entities managed by PID 1
2971 the output of "systemctl" is now a lot more comprehensive.
2973 * A new mini-daemon "systemd-machined" has been added which
2974 may be used by virtualization managers to register local
2975 VMs/containers. nspawn has been updated accordingly, and
2976 libvirt will be updated shortly. machined will collect a bit
2977 of meta information about the VMs/containers, and assign
2978 them their own scope unit (see above). The collected
2979 meta-data is then made available via the "machinectl" tool,
2980 and exposed in "ps" and similar tools. machined/machinectl
2981 is compile-time optional.
2983 * As discussed earlier, the low-level cgroup configuration
2984 options ControlGroup=, ControlGroupModify=,
2985 ControlGroupPersistent=, ControlGroupAttribute= have been
2986 removed. Please use high-level attribute settings instead as
2987 well as slice units.
2989 * A new bus call SetUnitProperties() has been added to alter
2990 various runtime parameters of a unit. This is primarily
2991 useful to alter cgroup parameters dynamically in a nice way,
2992 but will be extended later on to make more properties
2993 modifiable at runtime. systemctl gained a new set-properties
2994 command that wraps this call.
2996 * A new tool "systemd-run" has been added which can be used to
2997 run arbitrary command lines as transient services or scopes,
2998 while configuring a number of settings via the command
2999 line. This tool is currently very basic, however already
3000 very useful. We plan to extend this tool to even allow
3001 queuing of execution jobs with time triggers from the
3002 command line, similar in fashion to "at".
3004 * nspawn will now inform the user explicitly that kernels with
3005 audit enabled break containers, and suggest the user to turn
3008 * Support for detecting the IMA and AppArmor security
3009 frameworks with ConditionSecurity= has been added.
3011 * journalctl gained a new "-k" switch for showing only kernel
3012 messages, mimicking dmesg output; in addition to "--user"
3013 and "--system" switches for showing only user's own logs
3016 * systemd-delta can now show information about drop-in
3017 snippets extending unit files.
3019 * libsystemd-bus has been substantially updated but is still
3020 not available as public API.
3022 * systemd will now look for the "debug" argument on the kernel
3023 command line and enable debug logging, similar to what
3024 "systemd.log_level=debug" already did before.
3026 * "systemctl set-default", "systemctl get-default" has been
3027 added to configure the default.target symlink, which
3028 controls what to boot into by default.
3030 * "systemctl set-log-level" has been added as a convenient
3031 way to raise and lower systemd logging threshold.
3033 * "systemd-analyze plot" will now show the time the various
3034 generators needed for execution, as well as information
3035 about the unit file loading.
3037 * libsystemd-journal gained a new sd_journal_open_files() call
3038 for opening specific journal files. journactl also gained a
3039 new switch to expose this new functionality. Previously we
3040 only supported opening all files from a directory, or all
3041 files from the system, as opening individual files only is
3042 racy due to journal file rotation.
3044 * systemd gained the new DefaultEnvironment= setting in
3045 /etc/systemd/system.conf to set environment variables for
3048 * If a privileged process logs a journal message with the
3049 OBJECT_PID= field set, then journald will automatically
3050 augment this with additional OBJECT_UID=, OBJECT_GID=,
3051 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
3052 system services want to log events about specific client
3053 processes. journactl/systemctl has been updated to make use
3054 of this information if all log messages regarding a specific
3057 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
3058 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
3059 Reisner, David Coppa, David King, David Strauss, Eelco
3060 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
3061 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
3062 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
3063 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
3064 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
3065 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
3066 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
3067 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
3068 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
3069 Łukasz Stelmach, 장동준
3073 * The Python bindings gained some minimal support for the APIs
3074 exposed by libsystemd-logind.
3076 * ConditionSecurity= gained support for detecting SMACK. Since
3077 this condition already supports SELinux and AppArmor we only
3078 miss IMA for this. Patches welcome!
3080 Contributions from: Karol Lewandowski, Lennart Poettering,
3081 Zbigniew Jędrzejewski-Szmek
3085 * systemd-nspawn will now create /etc/resolv.conf if
3086 necessary, before bind-mounting the host's file onto it.
3088 * systemd-nspawn will now store meta information about a
3089 container on the container's cgroup as extended attribute
3090 fields, including the root directory.
3092 * The cgroup hierarchy has been reworked in many ways. All
3093 objects any of the components systemd creates in the cgroup
3094 tree are now suffixed. More specifically, user sessions are
3095 now placed in cgroups suffixed with ".session", users in
3096 cgroups suffixed with ".user", and nspawn containers in
3097 cgroups suffixed with ".nspawn". Furthermore, all cgroup
3098 names are now escaped in a simple scheme to avoid collision
3099 of userspace object names with kernel filenames. This work
3100 is preparation for making these objects relocatable in the
3101 cgroup tree, in order to allow easy resource partitioning of
3102 these objects without causing naming conflicts.
3104 * systemctl list-dependencies gained the new switches
3105 --plain, --reverse, --after and --before.
3107 * systemd-inhibit now shows the process name of processes that
3108 have taken an inhibitor lock.
3110 * nss-myhostname will now also resolve "localhost"
3111 implicitly. This makes /etc/hosts an optional file and
3112 nicely handles that on IPv6 ::1 maps to both "localhost" and
3115 * libsystemd-logind.so gained a new call
3116 sd_get_machine_names() to enumerate running containers and
3117 VMs (currently only supported by very new libvirt and
3118 nspawn). sd_login_monitor can now be used to watch
3119 VMs/containers coming and going.
3121 * .include is not allowed recursively anymore, and only in
3122 unit files. Usually it is better to use drop-in snippets in
3123 .d/*.conf anyway, as introduced with systemd 198.
3125 * systemd-analyze gained a new "critical-chain" command that
3126 determines the slowest chain of units run during system
3127 boot-up. It is very useful for tracking down where
3128 optimizing boot time is the most beneficial.
3130 * systemd will no longer allow manipulating service paths in
3131 the name=systemd:/system cgroup tree using ControlGroup= in
3132 units. (But is still fine with it in all other dirs.)
3134 * There's a new systemd-nspawn@.service service file that may
3135 be used to easily run nspawn containers as system
3136 services. With the container's root directory in
3137 /var/lib/container/foobar it is now sufficient to run
3138 "systemctl start systemd-nspawn@foobar.service" to boot it.
3140 * systemd-cgls gained a new parameter "--machine" to list only
3141 the processes within a certain container.
3143 * ConditionSecurity= now can check for "apparmor". We still
3144 are lacking checks for SMACK and IMA for this condition
3145 check though. Patches welcome!
3147 * A new configuration file /etc/systemd/sleep.conf has been
3148 added that may be used to configure which kernel operation
3149 systemd is supposed to execute when "suspend", "hibernate"
3150 or "hybrid-sleep" is requested. This makes the new kernel
3151 "freeze" state accessible to the user.
3153 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
3154 the passed argument if applicable.
3156 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
3157 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
3158 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
3159 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
3160 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
3161 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
3162 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
3167 * The output of 'systemctl list-jobs' got some polishing. The
3168 '--type=' argument may now be passed more than once. A new
3169 command 'systemctl list-sockets' has been added which shows
3170 a list of kernel sockets systemd is listening on with the
3171 socket units they belong to, plus the units these socket
3174 * The experimental libsystemd-bus library got substantial
3175 updates to work in conjunction with the (also experimental)
3176 kdbus kernel project. It works well enough to exchange
3177 messages with some sophistication. Note that kdbus is not
3178 ready yet, and the library is mostly an elaborate test case
3179 for now, and not installable.
3181 * systemd gained a new unit 'systemd-static-nodes.service'
3182 that generates static device nodes earlier during boot, and
3183 can run in conjunction with udev.
3185 * libsystemd-login gained a new call sd_pid_get_user_unit()
3186 to retrieve the user systemd unit a process is running
3187 in. This is useful for systems where systemd is used as
3190 * systemd-nspawn now places all containers in the new /machine
3191 top-level cgroup directory in the name=systemd
3192 hierarchy. libvirt will soon do the same, so that we get a
3193 uniform separation of /system, /user and /machine for system
3194 services, user processes and containers/virtual
3195 machines. This new cgroup hierarchy is also useful to stick
3196 stable names to specific container instances, which can be
3197 recognized later this way (this name may be controlled
3198 via systemd-nspawn's new -M switch). libsystemd-login also
3199 gained a new call sd_pid_get_machine_name() to retrieve the
3200 name of the container/VM a specific process belongs to.
3202 * bootchart can now store its data in the journal.
3204 * libsystemd-journal gained a new call
3205 sd_journal_add_conjunction() for AND expressions to the
3206 matching logic. This can be used to express more complex
3207 logical expressions.
3209 * journactl can now take multiple --unit= and --user-unit=
3212 * The cryptsetup logic now understands the "luks.key=" kernel
3213 command line switch for specifying a file to read the
3214 decryption key from. Also, if a configured key file is not
3215 found the tool will now automatically fall back to prompting
3218 * Python systemd.journal module was updated to wrap recently
3219 added functions from libsystemd-journal. The interface was
3220 changed to bring the low level interface in s.j._Reader
3221 closer to the C API, and the high level interface in
3222 s.j.Reader was updated to wrap and convert all data about
3225 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
3226 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
3227 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
3228 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
3229 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
3230 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3234 * journalctl --update-catalog now understands a new --root=
3235 option to operate on catalogs found in a different root
3238 * During shutdown after systemd has terminated all running
3239 services a final killing loop kills all remaining left-over
3240 processes. We will now print the name of these processes
3241 when we send SIGKILL to them, since this usually indicates a
3244 * If /etc/crypttab refers to password files stored on
3245 configured mount points automatic dependencies will now be
3246 generated to ensure the specific mount is established first
3247 before the key file is attempted to be read.
3249 * 'systemctl status' will now show information about the
3250 network sockets a socket unit is listening on.
3252 * 'systemctl status' will also shown information about any
3253 drop-in configuration file for units. (Drop-In configuration
3254 files in this context are files such as
3255 /etc/systemd/systemd/foobar.service.d/*.conf)
3257 * systemd-cgtop now optionally shows summed up CPU times of
3258 cgroups. Press '%' while running cgtop to switch between
3259 percentage and absolute mode. This is useful to determine
3260 which cgroups use up the most CPU time over the entire
3261 runtime of the system. systemd-cgtop has also been updated
3262 to be 'pipeable' for processing with further shell tools.
3264 * 'hostnamectl set-hostname' will now allow setting of FQDN
3267 * The formatting and parsing of time span values has been
3268 changed. The parser now understands fractional expressions
3269 such as "5.5h". The formatter will now output fractional
3270 expressions for all time spans under 1min, i.e. "5.123456s"
3271 rather than "5s 123ms 456us". For time spans under 1s
3272 millisecond values are shown, for those under 1ms
3273 microsecond values are shown. This should greatly improve
3274 all time-related output of systemd.
3276 * libsystemd-login and libsystemd-journal gained new
3277 functions for querying the poll() events mask and poll()
3278 timeout value for integration into arbitrary event
3281 * localectl gained the ability to list available X11 keymaps
3282 (models, layouts, variants, options).
3284 * 'systemd-analyze dot' gained the ability to filter for
3285 specific units via shell-style globs, to create smaller,
3286 more useful graphs. I.e. it is now possible to create simple
3287 graphs of all the dependencies between only target units, or
3288 of all units that Avahi has dependencies with.
3290 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
3291 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
3292 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
3293 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
3294 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
3295 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
3296 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
3300 * The boot-time readahead implementation for rotating media
3301 will now read the read-ahead data in multiple passes which
3302 consist of all read requests made in equidistant time
3303 intervals. This means instead of strictly reading read-ahead
3304 data in its physical order on disk we now try to find a
3305 middle ground between physical and access time order.
3307 * /etc/os-release files gained a new BUILD_ID= field for usage
3308 on operating systems that provide continuous builds of OS
3311 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
3312 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
3313 William Douglas, Zbigniew Jędrzejewski-Szmek
3317 * systemd-python gained an API exposing libsystemd-daemon.
3319 * The SMACK setup logic gained support for uploading CIPSO
3322 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
3323 ReadOnlyDirectories= and InaccessibleDirectories= has
3324 changed. The private /tmp and /var/tmp directories are now
3325 shared by all processes of a service (which means
3326 ExecStartPre= may now leave data in /tmp that ExecStart= of
3327 the same service can still access). When a service is
3328 stopped its temporary directories are immediately deleted
3329 (normal clean-up with tmpfiles is still done in addition to
3332 * By default, systemd will now set a couple of sysctl
3333 variables in the kernel: the safe sysrq options are turned
3334 on, IP route verification is turned on, and source routing
3335 disabled. The recently added hardlink and softlink
3336 protection of the kernel is turned on. These settings should
3337 be reasonably safe, and good defaults for all new systems.
3339 * The predictable network naming logic may now be turned off
3340 with a new kernel command line switch: net.ifnames=0.
3342 * A new libsystemd-bus module has been added that implements a
3343 pretty complete D-Bus client library. For details see:
3345 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
3347 * journald will now explicitly flush the journal files to disk
3348 at the latest 5min after each write. The file will then also
3349 be marked offline until the next write. This should increase
3350 reliability in case of a crash. The synchronization delay
3351 can be configured via SyncIntervalSec= in journald.conf.
3353 * There's a new remote-fs-setup.target unit that can be used
3354 to pull in specific services when at least one remote file
3355 system is to be mounted.
3357 * There are new targets timers.target and paths.target as
3358 canonical targets to pull user timer and path units in
3359 from. This complements sockets.target with a similar
3360 purpose for socket units.
3362 * libudev gained a new call udev_device_set_attribute_value()
3363 to set sysfs attributes of a device.
3365 * The udev daemon now sets the default number of worker
3366 processes executed in parallel based on the number of available
3367 CPUs instead of the amount of available RAM. This is supposed
3368 to provide a more reliable default and limit a too aggressive
3369 paralellism for setups with 1000s of devices connected.
3371 Contributions from: Auke Kok, Colin Walters, Cristian
3372 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
3373 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
3374 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
3375 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
3376 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
3377 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
3378 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
3379 Zbigniew Jędrzejewski-Szmek
3383 * Configuration of unit files may now be extended via drop-in
3384 files without having to edit/override the unit files
3385 themselves. More specifically, if the administrator wants to
3386 change one value for a service file foobar.service he can
3387 now do so by dropping in a configuration snippet into
3388 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
3389 will load all these snippets and apply them on top of the
3390 main unit configuration file, possibly extending or
3391 overriding its settings. Using these drop-in snippets is
3392 generally nicer than the two earlier options for changing
3393 unit files locally: copying the files from
3394 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
3395 them there; or creating a new file in /etc/systemd/system/
3396 that incorporates the original one via ".include". Drop-in
3397 snippets into these .d/ directories can be placed in any
3398 directory systemd looks for units in, and the usual
3399 overriding semantics between /usr/lib, /etc and /run apply
3402 * Most unit file settings which take lists of items can now be
3403 reset by assigning the empty string to them. For example,
3404 normally, settings such as Environment=FOO=BAR append a new
3405 environment variable assignment to the environment block,
3406 each time they are used. By assigning Environment= the empty
3407 string the environment block can be reset to empty. This is
3408 particularly useful with the .d/*.conf drop-in snippets
3409 mentioned above, since this adds the ability to reset list
3410 settings from vendor unit files via these drop-ins.
3412 * systemctl gained a new "list-dependencies" command for
3413 listing the dependencies of a unit recursively.
3415 * Inhibitors are now honored and listed by "systemctl
3416 suspend", "systemctl poweroff" (and similar) too, not only
3417 GNOME. These commands will also list active sessions by
3420 * Resource limits (as exposed by the various control group
3421 controllers) can now be controlled dynamically at runtime
3422 for all units. More specifically, you can now use a command
3423 like "systemctl set-cgroup-attr foobar.service cpu.shares
3424 2000" to alter the CPU shares a specific service gets. These
3425 settings are stored persistently on disk, and thus allow the
3426 administrator to easily adjust the resource usage of
3427 services with a few simple commands. This dynamic resource
3428 management logic is also available to other programs via the
3429 bus. Almost any kernel cgroup attribute and controller is
3432 * systemd-vconsole-setup will now copy all font settings to
3433 all allocated VTs, where it previously applied them only to
3436 * libsystemd-login gained the new sd_session_get_tty() API
3439 * This release drops support for a few legacy or
3440 distribution-specific LSB facility names when parsing init
3441 scripts: $x-display-manager, $mail-transfer-agent,
3442 $mail-transport-agent, $mail-transfer-agent, $smtp,
3443 $null. Also, the mail-transfer-agent.target unit backing
3444 this has been removed. Distributions which want to retain
3445 compatibility with this should carry the burden for
3446 supporting this themselves and patch support for these back
3447 in, if they really need to. Also, the facilities $syslog and
3448 $local_fs are now ignored, since systemd does not support
3449 early-boot LSB init scripts anymore, and these facilities
3450 are implied anyway for normal services. syslog.target has
3453 * There are new bus calls on PID1's Manager object for
3454 cancelling jobs, and removing snapshot units. Previously,
3455 both calls were only available on the Job and Snapshot
3458 * systemd-journal-gatewayd gained SSL support.
3460 * The various "environment" files, such as /etc/locale.conf
3461 now support continuation lines with a backslash ("\") as
3462 last character in the line, similarly in style (but different)
3463 to how this is supported in shells.
3465 * For normal user processes the _SYSTEMD_USER_UNIT= field is
3466 now implicitly appended to every log entry logged. systemctl
3467 has been updated to filter by this field when operating on a
3468 user systemd instance.
3470 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
3471 CAP_AUDIT_CONTROL capabilities to the capabilities set for
3472 the container. This makes it easier to boot unmodified
3473 Fedora systems in a container, which however still requires
3474 audit=0 to be passed on the kernel command line. Auditing in
3475 kernel and userspace is unfortunately still too broken in
3476 context of containers, hence we recommend compiling it out
3477 of the kernel or using audit=0. Hopefully this will be fixed
3478 one day for good in the kernel.
3480 * nspawn gained the new --bind= and --bind-ro= parameters to
3481 bind mount specific directories from the host into the
3484 * nspawn will now mount its own devpts file system instance
3485 into the container, in order not to leak pty devices from
3486 the host into the container.
3488 * systemd will now read the firmware boot time performance
3489 information from the EFI variables, if the used boot loader
3490 supports this, and takes it into account for boot performance
3491 analysis via "systemd-analyze". This is currently supported
3492 only in conjunction with Gummiboot, but could be supported
3493 by other boot loaders too. For details see:
3495 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
3497 * A new generator has been added that automatically mounts the
3498 EFI System Partition (ESP) to /boot, if that directory
3499 exists, is empty, and no other file system has been
3500 configured to be mounted there.
3502 * logind will now send out PrepareForSleep(false) out
3503 unconditionally, after coming back from suspend. This may be
3504 used by applications as asynchronous notification for
3505 system resume events.
3507 * "systemctl unlock-sessions" has been added, that allows
3508 unlocking the screens of all user sessions at once, similar
3509 to how "systemctl lock-sessions" already locked all users
3510 sessions. This is backed by a new D-Bus call UnlockSessions().
3512 * "loginctl seat-status" will now show the master device of a
3513 seat. (i.e. the device of a seat that needs to be around for
3514 the seat to be considered available, usually the graphics
3517 * tmpfiles gained a new "X" line type, that allows
3518 configuration of files and directories (with wildcards) that
3519 shall be excluded from automatic cleanup ("aging").
3521 * udev default rules set the device node permissions now only
3522 at "add" events, and do not change them any longer with a
3523 later "change" event.
3525 * The log messages for lid events and power/sleep keypresses
3526 now carry a message ID.
3528 * We now have a substantially larger unit test suite, but this
3529 continues to be work in progress.
3531 * udevadm hwdb gained a new --root= parameter to change the
3532 root directory to operate relative to.
3534 * logind will now issue a background sync() request to the kernel
3535 early at shutdown, so that dirty buffers are flushed to disk early
3536 instead of at the last moment, in order to optimize shutdown
3539 * A new bootctl tool has been added that is an interface for
3540 certain boot loader operations. This is currently a preview
3541 and is likely to be extended into a small mechanism daemon
3542 like timedated, localed, hostnamed, and can be used by
3543 graphical UIs to enumerate available boot options, and
3544 request boot into firmware operations.
3546 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
3547 the rest of the package. It also has been updated to work
3548 correctly in initrds.
3550 * Policykit previously has been runtime optional, and is now
3551 also compile time optional via a configure switch.
3553 * systemd-analyze has been reimplemented in C. Also "systemctl
3554 dot" has moved into systemd-analyze.
3556 * "systemctl status" with no further parameters will now print
3557 the status of all active or failed units.
3559 * Operations such as "systemctl start" can now be executed
3560 with a new mode "--irreversible" which may be used to queue
3561 operations that cannot accidentally be reversed by a later
3562 job queuing. This is by default used to make shutdown
3563 requests more robust.
3565 * The Python API of systemd now gained a new module for
3566 reading journal files.
3568 * A new tool kernel-install has been added that can install
3569 kernel images according to the Boot Loader Specification:
3571 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
3573 * Boot time console output has been improved to provide
3574 animated boot time output for hanging jobs.
3576 * A new tool systemd-activate has been added which can be used
3577 to test socket activation with, directly from the command
3578 line. This should make it much easier to test and debug
3579 socket activation in daemons.
3581 * journalctl gained a new "--reverse" (or -r) option to show
3582 journal output in reverse order (i.e. newest line first).
3584 * journalctl gained a new "--pager-end" (or -e) option to jump
3585 to immediately jump to the end of the journal in the
3586 pager. This is only supported in conjunction with "less".
3588 * journalctl gained a new "--user-unit=" option, that works
3589 similarly to "--unit=" but filters for user units rather than
3592 * A number of unit files to ease adoption of systemd in
3593 initrds has been added. This moves some minimal logic from
3594 the various initrd implementations into systemd proper.
3596 * The journal files are now owned by a new group
3597 "systemd-journal", which exists specifically to allow access
3598 to the journal, and nothing else. Previously, we used the
3599 "adm" group for that, which however possibly covers more
3600 than just journal/log file access. This new group is now
3601 already used by systemd-journal-gatewayd to ensure this
3602 daemon gets access to the journal files and as little else
3603 as possible. Note that "make install" will also set FS ACLs
3604 up for /var/log/journal to give "adm" and "wheel" read
3605 access to it, in addition to "systemd-journal" which owns
3606 the journal files. We recommend that packaging scripts also
3607 add read access to "adm" + "wheel" to /var/log/journal, and
3608 all existing/future journal files. To normal users and
3609 administrators little changes, however packagers need to
3610 ensure to create the "systemd-journal" system group at
3611 package installation time.
3613 * The systemd-journal-gatewayd now runs as unprivileged user
3614 systemd-journal-gateway:systemd-journal-gateway. Packaging
3615 scripts need to create these system user/group at
3618 * timedated now exposes a new boolean property CanNTP that
3619 indicates whether a local NTP service is available or not.
3621 * systemd-detect-virt will now also detect xen PVs
3623 * The pstore file system is now mounted by default, if it is
3626 * In addition to the SELinux and IMA policies we will now also
3627 load SMACK policies at early boot.
3629 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
3630 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
3631 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
3632 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
3633 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
3634 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
3635 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
3636 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
3637 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
3638 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
3639 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
3640 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
3641 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
3642 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
3646 * Timer units now support calendar time events in addition to
3647 monotonic time events. That means you can now trigger a unit
3648 based on a calendar time specification such as "Thu,Fri
3649 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
3650 or fifth day of any month of the year 2013, given that it is
3651 a thursday or friday. This brings timer event support
3652 considerably closer to cron's capabilities. For details on
3653 the supported calendar time specification language see
3656 * udev now supports a number of different naming policies for
3657 network interfaces for predictable names, and a combination
3658 of these policies is now the default. Please see this wiki
3659 document for details:
3661 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
3663 * Auke Kok's bootchart implementation has been added to the
3664 systemd tree. It is an optional component that can graph the
3665 boot in quite some detail. It is one of the best bootchart
3666 implementations around and minimal in its code and
3669 * nss-myhostname has been integrated into the systemd source
3670 tree. nss-myhostname guarantees that the local hostname
3671 always stays resolvable via NSS. It has been a weak
3672 requirement of systemd-hostnamed since a long time, and
3673 since its code is actually trivial we decided to just
3674 include it in systemd's source tree. It can be turned off
3675 with a configure switch.
3677 * The read-ahead logic is now capable of properly detecting
3678 whether a btrfs file system is on SSD or rotating media, in
3679 order to optimize the read-ahead scheme. Previously, it was
3680 only capable of detecting this on traditional file systems
3683 * In udev, additional device properties are now read from the
3684 IAB in addition to the OUI database. Also, Bluetooth company
3685 identities are attached to the devices as well.
3687 * In service files %U may be used as specifier that is
3688 replaced by the configured user name of the service.
3690 * nspawn may now be invoked without a controlling TTY. This
3691 makes it suitable for invocation as its own service. This
3692 may be used to set up a simple containerized server system
3693 using only core OS tools.
3695 * systemd and nspawn can now accept socket file descriptors
3696 when they are started for socket activation. This enables
3697 implementation of socket activated nspawn
3698 containers. i.e. think about autospawning an entire OS image
3699 when the first SSH or HTTP connection is received. We expect
3700 that similar functionality will also be added to libvirt-lxc
3703 * journalctl will now suppress ANSI color codes when
3704 presenting log data.
3706 * systemctl will no longer show control group information for
3707 a unit if a the control group is empty anyway.
3709 * logind can now automatically suspend/hibernate/shutdown the
3712 * /etc/machine-info and hostnamed now also expose the chassis
3713 type of the system. This can be used to determine whether
3714 the local system is a laptop, desktop, handset or
3715 tablet. This information may either be configured by the
3716 user/vendor or is automatically determined from ACPI and DMI
3717 information if possible.
3719 * A number of PolicyKit actions are now bound together with
3720 "imply" rules. This should simplify creating UIs because
3721 many actions will now authenticate similar ones as well.
3723 * Unit files learnt a new condition ConditionACPower= which
3724 may be used to conditionalize a unit depending on whether an
3725 AC power source is connected or not, of whether the system
3726 is running on battery power.
3728 * systemctl gained a new "is-failed" verb that may be used in
3729 shell scripts and suchlike to check whether a specific unit
3730 is in the "failed" state.
3732 * The EnvironmentFile= setting in unit files now supports file
3733 globbing, and can hence be used to easily read a number of
3734 environment files at once.
3736 * systemd will no longer detect and recognize specific
3737 distributions. All distribution-specific #ifdeffery has been
3738 removed, systemd is now fully generic and
3739 distribution-agnostic. Effectively, not too much is lost as
3740 a lot of the code is still accessible via explicit configure
3741 switches. However, support for some distribution specific
3742 legacy configuration file formats has been dropped. We
3743 recommend distributions to simply adopt the configuration
3744 files everybody else uses now and convert the old
3745 configuration from packaging scripts. Most distributions
3746 already did that. If that's not possible or desirable,
3747 distributions are welcome to forward port the specific
3748 pieces of code locally from the git history.
3750 * When logging a message about a unit systemd will now always
3751 log the unit name in the message meta data.
3753 * localectl will now also discover system locale data that is
3754 not stored in locale archives, but directly unpacked.
3756 * logind will no longer unconditionally use framebuffer
3757 devices as seat masters, i.e. as devices that are required
3758 to be existing before a seat is considered preset. Instead,
3759 it will now look for all devices that are tagged as
3760 "seat-master" in udev. By default framebuffer devices will
3761 be marked as such, but depending on local systems other
3762 devices might be marked as well. This may be used to
3763 integrate graphics cards using closed source drivers (such
3764 as NVidia ones) more nicely into logind. Note however, that
3765 we recommend using the open source NVidia drivers instead,
3766 and no udev rules for the closed-source drivers will be
3767 shipped from us upstream.
3769 Contributions from: Adam Williamson, Alessandro Crismani, Auke
3770 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
3771 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
3772 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
3773 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
3774 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
3775 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
3776 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
3777 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
3778 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
3779 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
3780 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
3785 * udev gained support for loading additional device properties
3786 from an indexed database that is keyed by vendor/product IDs
3787 and similar device identifiers. For the beginning this
3788 "hwdb" is populated with data from the well-known PCI and
3789 USB database, but also includes PNP, ACPI and OID data. In
3790 the longer run this indexed database shall grow into
3791 becoming the one central database for non-essential
3792 userspace device metadata. Previously, data from the PCI/USB
3793 database was only attached to select devices, since the
3794 lookup was a relatively expensive operation due to O(n) time
3795 complexity (with n being the number of entries in the
3796 database). Since this is now O(1), we decided to add in this
3797 data for all devices where this is available, by
3798 default. Note that the indexed database needs to be rebuilt
3799 when new data files are installed. To achieve this you need
3800 to update your packaging scripts to invoke "udevadm hwdb
3801 --update" after installation of hwdb data files. For
3802 RPM-based distributions we introduced the new
3803 %udev_hwdb_update macro for this purpose.
3805 * The Journal gained support for the "Message Catalog", an
3806 indexed database to link up additional information with
3807 journal entries. For further details please check:
3809 http://www.freedesktop.org/wiki/Software/systemd/catalog
3811 The indexed message catalog database also needs to be
3812 rebuilt after installation of message catalog files. Use
3813 "journalctl --update-catalog" for this. For RPM-based
3814 distributions we introduced the %journal_catalog_update
3815 macro for this purpose.
3817 * The Python Journal bindings gained support for the standard
3818 Python logging framework.
3820 * The Journal API gained new functions for checking whether
3821 the underlying file system of a journal file is capable of
3822 properly reporting file change notifications, or whether
3823 applications that want to reflect journal changes "live"
3824 need to recheck journal files continuously in appropriate
3827 * It is now possible to set the "age" field for tmpfiles
3828 entries to 0, indicating that files matching this entry
3829 shall always be removed when the directories are cleaned up.
3831 * coredumpctl gained a new "gdb" verb which invokes gdb
3832 right-away on the selected coredump.
3834 * There's now support for "hybrid sleep" on kernels that
3835 support this, in addition to "suspend" and "hibernate". Use
3836 "systemctl hybrid-sleep" to make use of this.
3838 * logind's HandleSuspendKey= setting (and related settings)
3839 now gained support for a new "lock" setting to simply
3840 request the screen lock on all local sessions, instead of
3841 actually executing a suspend or hibernation.
3843 * systemd will now mount the EFI variables file system by
3846 * Socket units now gained support for configuration of the
3847 SMACK security label.
3849 * timedatectl will now output the time of the last and next
3850 daylight saving change.
3852 * We dropped support for various legacy and distro-specific
3853 concepts, such as insserv, early-boot SysV services
3854 (i.e. those for non-standard runlevels such as 'b' or 'S')
3855 or ArchLinux /etc/rc.conf support. We recommend the
3856 distributions who still need support this to either continue
3857 to maintain the necessary patches downstream, or find a
3858 different solution. (Talk to us if you have questions!)
3860 * Various systemd components will now bypass PolicyKit checks
3861 for root and otherwise handle properly if PolicyKit is not
3862 found to be around. This should fix most issues for
3863 PolicyKit-less systems. Quite frankly this should have been
3864 this way since day one. It is absolutely our intention to
3865 make systemd work fine on PolicyKit-less systems, and we
3866 consider it a bug if something does not work as it should if
3867 PolicyKit is not around.
3869 * For embedded systems it is now possible to build udev and
3870 systemd without blkid and/or kmod support.
3872 * "systemctl switch-root" is now capable of switching root
3873 more than once. I.e. in addition to transitions from the
3874 initrd to the host OS it is now possible to transition to
3875 further OS images from the host. This is useful to implement
3876 offline updating tools.
3878 * Various other additions have been made to the RPM macros
3879 shipped with systemd. Use %udev_rules_update() after
3880 installing new udev rules files. %_udevhwdbdir,
3881 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
3882 %_sysctldir are now available which resolve to the right
3883 directories for packages to place various data files in.
3885 * journalctl gained the new --full switch (in addition to
3886 --all, to disable ellipsation for long messages.
3888 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
3889 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
3890 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
3891 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
3892 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
3893 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
3894 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
3895 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
3896 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
3900 * journalctl gained new --since= and --until= switches to
3901 filter by time. It also now supports nice filtering for
3902 units via --unit=/-u.
3904 * Type=oneshot services may use ExecReload= and do the
3907 * The journal daemon now supports time-based rotation and
3908 vacuuming, in addition to the usual disk-space based
3911 * The journal will now index the available field values for
3912 each field name. This enables clients to show pretty drop
3913 downs of available match values when filtering. The bash
3914 completion of journalctl has been updated
3915 accordingly. journalctl gained a new switch -F to list all
3916 values a certain field takes in the journal database.
3918 * More service events are now written as structured messages
3919 to the journal, and made recognizable via message IDs.
3921 * The timedated, localed and hostnamed mini-services which
3922 previously only provided support for changing time, locale
3923 and hostname settings from graphical DEs such as GNOME now
3924 also have a minimal (but very useful) text-based client
3925 utility each. This is probably the nicest way to changing
3926 these settings from the command line now, especially since
3927 it lists available options and is fully integrated with bash
3930 * There's now a new tool "systemd-coredumpctl" to list and
3931 extract coredumps from the journal.
3933 * We now install a README each in /var/log/ and
3934 /etc/rc.d/init.d explaining where the system logs and init
3935 scripts went. This hopefully should help folks who go to
3936 that dirs and look into the otherwise now empty void and
3937 scratch their heads.
3939 * When user-services are invoked (by systemd --user) the
3940 $MANAGERPID env var is set to the PID of systemd.
3942 * SIGRTMIN+24 when sent to a --user instance will now result
3943 in immediate termination of systemd.
3945 * gatewayd received numerous feature additions such as a
3946 "follow" mode, for live syncing and filtering.
3948 * browse.html now allows filtering and showing detailed
3949 information on specific entries. Keyboard navigation and
3950 mouse screen support has been added.
3952 * gatewayd/journalctl now supports HTML5/JSON
3953 Server-Sent-Events as output.
3955 * The SysV init script compatibility logic will now
3956 heuristically determine whether a script supports the
3957 "reload" verb, and only then make this available as
3960 * "systemctl status --follow" has been removed, use "journalctl
3963 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
3964 have been removed since they are hardly useful to be
3967 * And I'd like to take the opportunity to specifically mention
3968 Zbigniew for his great contributions. Zbigniew, you rock!
3970 Contributions from: Andrew Eikum, Christian Hesse, Colin
3971 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
3972 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
3973 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
3974 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
3975 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
3976 Jędrzejewski-Szmek, Сковорода Никита Андреевич
3980 * If /etc/vconsole.conf is non-existent or empty we will no
3981 longer load any console font or key map at boot by
3982 default. Instead the kernel defaults will be left
3983 intact. This is definitely the right thing to do, as no
3984 configuration should mean no configuration, and hard-coding
3985 font names that are different on all archs is probably a bad
3986 idea. Also, the kernel default key map and font should be
3987 good enough for most cases anyway, and mostly identical to
3988 the userspace fonts/key maps we previously overloaded them
3989 with. If distributions want to continue to default to a
3990 non-kernel font or key map they should ship a default
3991 /etc/vconsole.conf with the appropriate contents.
3993 Contributions from: Colin Walters, Daniel J Walsh, Dave
3994 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
3995 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3999 * journalctl gained a new --cursor= switch to show entries
4000 starting from the specified location in the journal.
4002 * We now enforce a size limit on journal entry fields exported
4003 with "-o json" in journalctl. Fields larger than 4K will be
4004 assigned null. This can be turned off with --all.
4006 * An (optional) journal gateway daemon is now available as
4007 "systemd-journal-gatewayd.service". This service provides
4008 access to the journal via HTTP and JSON. This functionality
4009 will be used to implement live log synchronization in both
4010 pull and push modes, but has various other users too, such
4011 as easy log access for debugging of embedded devices. Right
4012 now it is already useful to retrieve the journal via HTTP:
4014 # systemctl start systemd-journal-gatewayd.service
4015 # wget http://localhost:19531/entries
4017 This will download the journal contents in a
4018 /var/log/messages compatible format. The same as JSON:
4020 # curl -H"Accept: application/json" http://localhost:19531/entries
4022 This service is also accessible via a web browser where a
4023 single static HTML5 app is served that uses the JSON logic
4024 to enable the user to do some basic browsing of the
4025 journal. This will be extended later on. Here's an example
4026 screenshot of this app in its current state:
4028 http://0pointer.de/public/journal-gatewayd
4030 Contributions from: Kay Sievers, Lennart Poettering, Robert
4031 Milasan, Tom Gundersen
4035 * The bash completion logic is now available for journalctl
4038 * We do not mount the "cpuset" controller anymore together with
4039 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
4040 started if no parameters are assigned to it. "cpuset" hence
4041 broke code that assumed it it could create "cpu" groups and
4044 * journalctl -f will now subscribe to terminal size changes,
4045 and line break accordingly.
4047 Contributions from: Dave Reisner, Kay Sievers, Lennart
4048 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
4052 * nspawn will now create a symlink /etc/localtime in the
4053 container environment, copying the host's timezone
4054 setting. Previously this has been done via a bind mount, but
4055 since symlinks cannot be bind mounted this has now been
4056 changed to create/update the appropriate symlink.
4058 * journalctl -n's line number argument is now optional, and
4059 will default to 10 if omitted.
4061 * journald will now log the maximum size the journal files may
4062 take up on disk. This is particularly useful if the default
4063 built-in logic of determining this parameter from the file
4064 system size is used. Use "systemctl status
4065 systemd-journald.service" to see this information.
4067 * The multi-seat X wrapper tool has been stripped down. As X
4068 is now capable of enumerating graphics devices via udev in a
4069 seat-aware way the wrapper is not strictly necessary
4070 anymore. A stripped down temporary stop-gap is still shipped
4071 until the upstream display managers have been updated to
4072 fully support the new X logic. Expect this wrapper to be
4073 removed entirely in one of the next releases.
4075 * HandleSleepKey= in logind.conf has been split up into
4076 HandleSuspendKey= and HandleHibernateKey=. The old setting
4077 is not available anymore. X11 and the kernel are
4078 distinguishing between these keys and we should too. This
4079 also means the inhibition lock for these keys has been split
4082 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
4083 Poettering, Lukas Nykryn, Václav Pavlín
4087 * Whenever a unit changes state we will now log this to the
4088 journal and show along the unit's own log output in
4091 * ConditionPathIsMountPoint= can now properly detect bind
4092 mount points too. (Previously, a bind mount of one file
4093 system to another place in the same file system could not be
4094 detected as mount, since they shared struct stat's st_dev
4097 * We will now mount the cgroup controllers cpu, cpuacct,
4098 cpuset and the controllers net_cls, net_prio together by
4101 * nspawn containers will now have a virtualized boot
4102 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
4103 over with a randomized ID at container initialization). This
4104 has the effect of making "journalctl -b" do the right thing
4107 * The JSON output journal serialization has been updated not
4108 to generate "endless" list objects anymore, but rather one
4109 JSON object per line. This is more in line how most JSON
4110 parsers expect JSON objects. The new output mode
4111 "json-pretty" has been added to provide similar output, but
4112 neatly aligned for readability by humans.
4114 * We dropped all explicit sync() invocations in the shutdown
4115 code. The kernel does this implicitly anyway in the kernel
4116 reboot() syscall. halt(8)'s -n option is now a compatibility
4119 * We now support virtualized reboot() in containers, as
4120 supported by newer kernels. We will fall back to exit() if
4121 CAP_SYS_REBOOT is not available to the container. Also,
4122 nspawn makes use of this now and will actually reboot the
4123 container if the containerized OS asks for that.
4125 * journalctl will only show local log output by default
4126 now. Use --merge (-m) to show remote log output, too.
4128 * libsystemd-journal gained the new sd_journal_get_usage()
4129 call to determine the current disk usage of all journal
4130 files. This is exposed in the new "journalctl --disk-usage"
4133 * journald gained a new configuration setting SplitMode= in
4134 journald.conf which may be used to control how user journals
4135 are split off. See journald.conf(5) for details.
4137 * A new condition type ConditionFileNotEmpty= has been added.
4139 * tmpfiles' "w" lines now support file globbing, to write
4140 multiple files at once.
4142 * We added Python bindings for the journal submission
4143 APIs. More Python APIs for a number of selected APIs will
4144 likely follow. Note that we intend to add native bindings
4145 only for the Python language, as we consider it common
4146 enough to deserve bindings shipped within systemd. There are
4147 various projects outside of systemd that provide bindings
4148 for languages such as PHP or Lua.
4150 * Many conditions will now resolve specifiers such as %i. In
4151 addition, PathChanged= and related directives of .path units
4152 now support specifiers as well.
4154 * There's now a new RPM macro definition for the system preset
4157 * journald will now warn if it ca not forward a message to the
4158 syslog daemon because its socket is full.
4160 * timedated will no longer write or process /etc/timezone,
4161 except on Debian. As we do not support late mounted /usr
4162 anymore /etc/localtime always being a symlink is now safe,
4163 and hence the information in /etc/timezone is not necessary
4166 * logind will now always reserve one VT for a text getty (VT6
4167 by default). Previously if more than 6 X sessions where
4168 started they took up all the VTs with auto-spawned gettys,
4169 so that no text gettys were available anymore.
4171 * udev will now automatically inform the btrfs kernel logic
4172 about btrfs RAID components showing up. This should make
4173 simple hotplug based btrfs RAID assembly work.
4175 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
4176 (but not for its children which will stay at the kernel
4177 default). This should allow setups with a lot more listening
4180 * systemd will now always pass the configured timezone to the
4181 kernel at boot. timedated will do the same when the timezone
4184 * logind's inhibition logic has been updated. By default,
4185 logind will now handle the lid switch, the power and sleep
4186 keys all the time, even in graphical sessions. If DEs want
4187 to handle these events on their own they should take the new
4188 handle-power-key, handle-sleep-key and handle-lid-switch
4189 inhibitors during their runtime. A simple way to achieve
4190 that is to invoke the DE wrapped in an invocation of:
4192 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
4194 * Access to unit operations is now checked via SELinux taking
4195 the unit file label and client process label into account.
4197 * systemd will now notify the administrator in the journal
4198 when he over-mounts a non-empty directory.
4200 * There are new specifiers that are resolved in unit files,
4201 for the host name (%H), the machine ID (%m) and the boot ID
4204 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
4205 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
4206 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
4207 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
4208 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
4209 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
4210 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
4214 * Support for reading structured kernel messages from
4215 /dev/kmsg has now been added and is enabled by default.
4217 * Support for reading kernel messages from /proc/kmsg has now
4218 been removed. If you want kernel messages in the journal
4219 make sure to run a recent kernel (>= 3.5) that supports
4220 reading structured messages from /dev/kmsg (see
4221 above). /proc/kmsg is now exclusive property of classic
4222 syslog daemons again.
4224 * The libudev API gained the new
4225 udev_device_new_from_device_id() call.
4227 * The logic for file system namespace (ReadOnlyDirectory=,
4228 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
4229 require pivot_root() anymore. This means fewer temporary
4230 directories are created below /tmp for this feature.
4232 * nspawn containers will now see and receive all submounts
4233 made on the host OS below the root file system of the
4236 * Forward Secure Sealing is now supported for Journal files,
4237 which provide cryptographical sealing of journal files so
4238 that attackers cannot alter log history anymore without this
4239 being detectable. Lennart will soon post a blog story about
4240 this explaining it in more detail.
4242 * There are two new service settings RestartPreventExitStatus=
4243 and SuccessExitStatus= which allow configuration of exit
4244 status (exit code or signal) which will be excepted from the
4245 restart logic, resp. consider successful.
4247 * journalctl gained the new --verify switch that can be used
4248 to check the integrity of the structure of journal files and
4249 (if Forward Secure Sealing is enabled) the contents of
4252 * nspawn containers will now be run with /dev/stdin, /dev/fd/
4253 and similar symlinks pre-created. This makes running shells
4254 as container init process a lot more fun.
4256 * The fstab support can now handle PARTUUID= and PARTLABEL=
4259 * A new ConditionHost= condition has been added to match
4260 against the hostname (with globs) and machine ID. This is
4261 useful for clusters where a single OS image is used to
4262 provision a large number of hosts which shall run slightly
4263 different sets of services.
4265 * Services which hit the restart limit will now be placed in a
4268 Contributions from: Bertram Poettering, Dave Reisner, Huang
4269 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
4270 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
4274 * When running in --user mode systemd will now become a
4275 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
4276 tree a lot more organized.
4278 * A new PartOf= unit dependency type has been introduced that
4279 may be used to group services in a natural way.
4281 * "systemctl enable" may now be used to enable instances of
4284 * journalctl now prints error log levels in red, and
4285 warning/notice log levels in bright white. It also supports
4286 filtering by log level now.
4288 * cgtop gained a new -n switch (similar to top), to configure
4289 the maximum number of iterations to run for. It also gained
4290 -b, to run in batch mode (accepting no input).
4292 * The suffix ".service" may now be omitted on most systemctl
4293 command lines involving service unit names.
4295 * There's a new bus call in logind to lock all sessions, as
4296 well as a loginctl verb for it "lock-sessions".
4298 * libsystemd-logind.so gained a new call sd_journal_perror()
4299 that works similar to libc perror() but logs to the journal
4300 and encodes structured information about the error number.
4302 * /etc/crypttab entries now understand the new keyfile-size=
4305 * shutdown(8) now can send a (configurable) wall message when
4306 a shutdown is cancelled.
4308 * The mount propagation mode for the root file system will now
4309 default to "shared", which is useful to make containers work
4310 nicely out-of-the-box so that they receive new mounts from
4311 the host. This can be undone locally by running "mount
4312 --make-rprivate /" if needed.
4314 * The prefdm.service file has been removed. Distributions
4315 should maintain this unit downstream if they intend to keep
4316 it around. However, we recommend writing normal unit files
4317 for display managers instead.
4319 * Since systemd is a crucial part of the OS we will now
4320 default to a number of compiler switches that improve
4321 security (hardening) such as read-only relocations, stack
4322 protection, and suchlike.
4324 * The TimeoutSec= setting for services is now split into
4325 TimeoutStartSec= and TimeoutStopSec= to allow configuration
4326 of individual time outs for the start and the stop phase of
4329 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
4330 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
4331 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
4332 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
4333 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
4334 Gundersen, Zbigniew Jędrzejewski-Szmek
4338 * The journal and id128 C APIs are now fully documented as man
4341 * Extra safety checks have been added when transitioning from
4342 the initial RAM disk to the main system to avoid accidental
4345 * /etc/crypttab entries now understand the new keyfile-offset=
4348 * systemctl -t can now be used to filter by unit load state.
4350 * The journal C API gained the new sd_journal_wait() call to
4351 make writing synchronous journal clients easier.
4353 * journalctl gained the new -D switch to show journals from a
4356 * journalctl now displays a special marker between log
4357 messages of two different boots.
4359 * The journal is now explicitly flushed to /var via a service
4360 systemd-journal-flush.service, rather than implicitly simply
4361 by seeing /var/log/journal to be writable.
4363 * journalctl (and the journal C APIs) can now match for much
4364 more complex expressions, with alternatives and
4367 * When transitioning from the initial RAM disk to the main
4368 system we will now kill all processes in a killing spree to
4369 ensure no processes stay around by accident.
4371 * Three new specifiers may be used in unit files: %u, %h, %s
4372 resolve to the user name, user home directory resp. user
4373 shell. This is useful for running systemd user instances.
4375 * We now automatically rotate journal files if their data
4376 object hash table gets a fill level > 75%. We also size the
4377 hash table based on the configured maximum file size. This
4378 together should lower hash collisions drastically and thus
4379 speed things up a bit.
4381 * journalctl gained the new "--header" switch to introspect
4382 header data of journal files.
4384 * A new setting SystemCallFilters= has been added to services
4385 which may be used to apply blacklists or whitelists to
4386 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
4388 * nspawn gained a new --link-journal= switch (and quicker: -j)
4389 to link the container journal with the host. This makes it
4390 very easy to centralize log viewing on the host for all
4391 guests while still keeping the journal files separated.
4393 * Many bugfixes and optimizations
4395 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
4396 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
4397 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
4402 * Several tools now understand kernel command line arguments,
4403 which are only read when run in an initial RAM disk. They
4404 usually follow closely their normal counterparts, but are
4407 * There's a new tool to analyze the readahead files that are
4408 automatically generated at boot. Use:
4410 /usr/lib/systemd/systemd-readahead analyze /.readahead
4412 * We now provide an early debug shell on tty9 if this enabled. Use:
4414 systemctl enable debug-shell.service
4416 * All plymouth related units have been moved into the Plymouth
4417 package. Please make sure to upgrade your Plymouth version
4420 * systemd-tmpfiles now supports getting passed the basename of
4421 a configuration file only, in which case it will look for it
4422 in all appropriate directories automatically.
4424 * udevadm info now takes a /dev or /sys path as argument, and
4425 does the right thing. Example:
4427 udevadm info /dev/sda
4428 udevadm info /sys/class/block/sda
4430 * systemctl now prints a warning if a unit is stopped but a
4431 unit that might trigger it continues to run. Example: a
4432 service is stopped but the socket that activates it is left
4435 * "systemctl status" will now mention if the log output was
4436 shortened due to rotation since a service has been started.
4438 * The journal API now exposes functions to determine the
4439 "cutoff" times due to rotation.
4441 * journald now understands SIGUSR1 and SIGUSR2 for triggering
4442 immediately flushing of runtime logs to /var if possible,
4443 resp. for triggering immediate rotation of the journal
4446 * It is now considered an error if a service is attempted to
4447 be stopped that is not loaded.
4449 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
4451 * systemd-analyze now supports Python 3
4453 * tmpfiles now supports cleaning up directories via aging
4454 where the first level dirs are always kept around but
4455 directories beneath it automatically aged. This is enabled
4456 by prefixing the age field with '~'.
4458 * Seat objects now expose CanGraphical, CanTTY properties
4459 which is required to deal with very fast bootups where the
4460 display manager might be running before the graphics drivers
4461 completed initialization.
4463 * Seat objects now expose a State property.
4465 * We now include RPM macros for service enabling/disabling
4466 based on the preset logic. We recommend RPM based
4467 distributions to make use of these macros if possible. This
4468 makes it simpler to reuse RPM spec files across
4471 * We now make sure that the collected systemd unit name is
4472 always valid when services log to the journal via
4475 * There's a new man page kernel-command-line(7) detailing all
4476 command line options we understand.
4478 * The fstab generator may now be disabled at boot by passing
4479 fstab=0 on the kernel command line.
4481 * A new kernel command line option modules-load= is now understood
4482 to load a specific kernel module statically, early at boot.
4484 * Unit names specified on the systemctl command line are now
4485 automatically escaped as needed. Also, if file system or
4486 device paths are specified they are automatically turned
4487 into the appropriate mount or device unit names. Example:
4489 systemctl status /home
4490 systemctl status /dev/sda
4492 * The SysVConsole= configuration option has been removed from
4493 system.conf parsing.
4495 * The SysV search path is no longer exported on the D-Bus
4498 * The Names= option is been removed from unit file parsing.
4500 * There's a new man page bootup(7) detailing the boot process.
4502 * Every unit and every generator we ship with systemd now
4503 comes with full documentation. The self-explanatory boot is
4506 * A couple of services gained "systemd-" prefixes in their
4507 name if they wrap systemd code, rather than only external
4508 code. Among them fsck@.service which is now
4509 systemd-fsck@.service.
4511 * The HaveWatchdog property has been removed from the D-Bus
4514 * systemd.confirm_spawn= on the kernel command line should now
4517 * There's a new man page crypttab(5) which details all options
4518 we actually understand.
4520 * systemd-nspawn gained a new --capability= switch to pass
4521 additional capabilities to the container.
4523 * timedated will now read known NTP implementation unit names
4524 from /usr/lib/systemd/ntp-units.d/*.list,
4525 systemd-timedated-ntp.target has been removed.
4527 * journalctl gained a new switch "-b" that lists log data of
4528 the current boot only.
4530 * The notify socket is in the abstract namespace again, in
4531 order to support daemons which chroot() at start-up.
4533 * There is a new Storage= configuration option for journald
4534 which allows configuration of where log data should go. This
4535 also provides a way to disable journal logging entirely, so
4536 that data collected is only forwarded to the console, the
4537 kernel log buffer or another syslog implementation.
4539 * Many bugfixes and optimizations
4541 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
4542 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
4543 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
4544 Shawn Landden, Tom Gundersen
4548 * "systemctl help <unit>" now shows the man page if one is
4551 * Several new man pages have been added.
4553 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
4554 MaxLevelConsole= can now be specified in
4555 journald.conf. These options allow reducing the amount of
4556 data stored on disk or forwarded by the log level.
4558 * TimerSlackNSec= can now be specified in system.conf for
4559 PID1. This allows system-wide power savings.
4561 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
4562 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
4567 * logind is now capable of (optionally) handling power and
4568 sleep keys as well as the lid switch.
4570 * journalctl now understands the syntax "journalctl
4571 /usr/bin/avahi-daemon" to get all log output of a specific
4574 * CapabilityBoundingSet= in system.conf now also influences
4575 the capability bound set of usermode helpers of the kernel.
4577 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
4578 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
4579 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
4580 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
4584 * Note that we skipped 139 releases here in order to set the
4585 new version to something that is greater than both udev's
4586 and systemd's most recent version number.
4588 * udev: all udev sources are merged into the systemd source tree now.
4589 All future udev development will happen in the systemd tree. It
4590 is still fully supported to use the udev daemon and tools without
4591 systemd running, like in initramfs or other init systems. Building
4592 udev though, will require the *build* of the systemd tree, but
4593 udev can be properly *run* without systemd.
4595 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
4596 should be used to create dead device nodes as workarounds for broken
4599 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
4600 no longer supported. udev_monitor_new_from_netlink() needs to be
4601 used to subscribe to events.
4603 * udev: when udevd is started by systemd, processes which are left
4604 behind by forking them off of udev rules, are unconditionally cleaned
4605 up and killed now after the event handling has finished. Services or
4606 daemons must be started as systemd services. Services can be
4607 pulled-in by udev to get started, but they can no longer be directly
4608 forked by udev rules.
4610 * udev: the daemon binary is called systemd-udevd now and installed
4611 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
4612 to adapt to that, create symlink, or rename the binary after building
4615 * libudev no longer provides these symbols:
4616 udev_monitor_from_socket()
4617 udev_queue_get_failed_list_entry()
4618 udev_get_{dev,sys,run}_path()
4619 The versions number was bumped and symbol versioning introduced.
4621 * systemd-loginctl and systemd-journalctl have been renamed
4622 to loginctl and journalctl to match systemctl.
4624 * The config files: /etc/systemd/systemd-logind.conf and
4625 /etc/systemd/systemd-journald.conf have been renamed to
4626 logind.conf and journald.conf. Package updates should rename
4627 the files to the new names on upgrade.
4629 * For almost all files the license is now LGPL2.1+, changed
4630 from the previous GPL2.0+. Exceptions are some minor stuff
4631 of udev (which will be changed to LGPL2.1 eventually, too),
4632 and the MIT licensed sd-daemon.[ch] library that is suitable
4633 to be used as drop-in files.
4635 * systemd and logind now handle system sleep states, in
4636 particular suspending and hibernating.
4638 * logind now implements a sleep/shutdown/idle inhibiting logic
4639 suitable for a variety of uses. Soonishly Lennart will blog
4640 about this in more detail.
4642 * var-run.mount and var-lock.mount are no longer provided
4643 (which prevously bind mounted these directories to their new
4644 places). Distributions which have not converted these
4645 directories to symlinks should consider stealing these files
4646 from git history and add them downstream.
4648 * We introduced the Documentation= field for units and added
4649 this to all our shipped units. This is useful to make it
4650 easier to explore the boot and the purpose of the various
4653 * All smaller setup units (such as
4654 systemd-vconsole-setup.service) now detect properly if they
4655 are run in a container and are skipped when
4656 appropriate. This guarantees an entirely noise-free boot in
4657 Linux container environments such as systemd-nspawn.
4659 * A framework for implementing offline system updates is now
4660 integrated, for details see:
4661 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
4663 * A new service type Type=idle is available now which helps us
4664 avoiding ugly interleaving of getty output and boot status
4667 * There's now a system-wide CapabilityBoundingSet= option to
4668 globally reduce the set of capabilities for the
4669 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
4670 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
4671 even CAP_NET_ADMIN system-wide for secure systems.
4673 * There are now system-wide DefaultLimitXXX= options to
4674 globally change the defaults of the various resource limits
4675 for all units started by PID 1.
4677 * Harald Hoyer's systemd test suite has been integrated into
4678 systemd which allows easy testing of systemd builds in qemu
4679 and nspawn. (This is really awesome! Ask us for details!)
4681 * The fstab parser is now implemented as generator, not inside
4684 * systemctl will now warn you if .mount units generated from
4685 /etc/fstab are out of date due to changes in fstab that
4686 have not been read by systemd yet.
4688 * systemd is now suitable for usage in initrds. Dracut has
4689 already been updated to make use of this. With this in place
4690 initrds get a slight bit faster but primarily are much
4691 easier to introspect and debug since "systemctl status" in
4692 the host system can be used to introspect initrd services,
4693 and the journal from the initrd is kept around too.
4695 * systemd-delta has been added, a tool to explore differences
4696 between user/admin configuration and vendor defaults.
4698 * PrivateTmp= now affects both /tmp and /var/tmp.
4700 * Boot time status messages are now much prettier and feature
4701 proper english language. Booting up systemd has never been
4704 * Read-ahead pack files now include the inode number of all
4705 files to pre-cache. When the inode changes the pre-caching
4706 is not attempted. This should be nicer to deal with updated
4707 packages which might result in changes of read-ahead
4710 * We now temporaritly lower the kernel's read_ahead_kb variable
4711 when collecting read-ahead data to ensure the kernel's
4712 built-in read-ahead does not add noise to our measurements
4713 of necessary blocks to pre-cache.
4715 * There's now RequiresMountsFor= to add automatic dependencies
4716 for all mounts necessary for a specific file system path.
4718 * MountAuto= and SwapAuto= have been removed from
4719 system.conf. Mounting file systems at boot has to take place
4722 * nspawn now learned a new switch --uuid= to set the machine
4723 ID on the command line.
4725 * nspawn now learned the -b switch to automatically search
4728 * vt102 is now the default TERM for serial TTYs, upgraded from
4731 * systemd-logind now works on VT-less systems.
4733 * The build tree has been reorganized. The individual
4734 components now have directories of their own.
4736 * A new condition type ConditionPathIsReadWrite= is now available.
4738 * nspawn learned the new -C switch to create cgroups for the
4739 container in other hierarchies.
4741 * We now have support for hardware watchdogs, configurable in
4744 * The scheduled shutdown logic now has a public API.
4746 * We now mount /tmp as tmpfs by default, but this can be
4747 masked and /etc/fstab can override it.
4749 * Since udisks does not make use of /media anymore we are not
4750 mounting a tmpfs on it anymore.
4752 * journalctl gained a new --local switch to only interleave
4753 locally generated journal files.
4755 * We can now load the IMA policy at boot automatically.
4757 * The GTK tools have been split off into a systemd-ui.
4759 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
4760 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
4761 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
4762 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
4763 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
4764 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
4765 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
4766 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
4767 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
4772 * This is mostly a bugfix release
4774 * Support optional initialization of the machine ID from the
4775 KVM or container configured UUID.
4777 * Support immediate reboots with "systemctl reboot -ff"
4779 * Show /etc/os-release data in systemd-analyze output
4781 * Many bugfixes for the journal, including endianness fixes and
4782 ensuring that disk space enforcement works
4784 * sd-login.h is C++ comptaible again
4786 * Extend the /etc/os-release format on request of the Debian
4789 * We now refuse non-UTF8 strings used in various configuration
4790 and unit files. This is done to ensure we do not pass invalid
4791 data over D-Bus or expose it elsewhere.
4793 * Register Mimo USB Screens as suitable for automatic seat
4796 * Read SELinux client context from journal clients in a race
4799 * Reorder configuration file lookup order. /etc now always
4800 overrides /run in order to allow the administrator to always
4801 and unconditionally override vendor supplied or
4802 automatically generated data.
4804 * The various user visible bits of the journal now have man
4805 pages. We still lack man pages for the journal API calls
4808 * We now ship all man pages in HTML format again in the
4811 Contributions from: Dave Reisner, Dirk Eibach, Frederic
4812 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
4813 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
4818 * This is mostly a bugfix release
4820 * systems lacking /etc/os-release are no longer supported.
4822 * Various functionality updates to libsystemd-login.so
4824 * Track class of PAM logins to distinguish greeters from
4827 Contributions from: Kay Sievers, Lennart Poettering, Michael
4832 * This is an important bugfix release for v41.
4834 * Building man pages is now optional which should be useful
4835 for those building systemd from git but unwilling to install
4838 * Watchdog support for supervising services is now usable. In
4839 a future release support for hardware watchdogs
4840 (i.e. /dev/watchdog) will be added building on this.
4842 * Service start rate limiting is now configurable and can be
4843 turned off per service. When a start rate limit is hit a
4844 reboot can automatically be triggered.
4846 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
4848 Contributions from: Benjamin Franzke, Bill Nottingham,
4849 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
4850 Schmidt, Michał Górny, Piotr Drąg
4854 * The systemd binary is installed /usr/lib/systemd/systemd now;
4855 An existing /sbin/init symlink needs to be adapted with the
4858 * The code that loads kernel modules has been ported to invoke
4859 libkmod directly, instead of modprobe. This means we do not
4860 support systems with module-init-tools anymore.
4862 * Watchdog support is now already useful, but still not
4865 * A new kernel command line option systemd.setenv= is
4866 understood to set system wide environment variables
4867 dynamically at boot.
4869 * We now limit the set of capabilities of systemd-journald.
4871 * We now set SIGPIPE to ignore by default, since it only is
4872 useful in shell pipelines, and has little use in general
4873 code. This can be disabled with IgnoreSIPIPE=no in unit
4876 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
4877 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
4882 * This is mostly a bugfix release
4884 * We now expose the reason why a service failed in the
4885 "Result" D-Bus property.
4887 * Rudimentary service watchdog support (will be completed over
4888 the next few releases.)
4890 * When systemd forks off in order execute some service we will
4891 now immediately changes its argv[0] to reflect which process
4892 it will execute. This is useful to minimize the time window
4893 with a generic argv[0], which makes bootcharts more useful
4895 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
4896 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
4897 Mike Kazantsev, Ray Strode
4901 * This is mostly a test release, but incorporates many
4904 * New systemd-cgtop tool to show control groups by their
4907 * Linking against libacl for ACLs is optional again. If
4908 disabled, support tracking device access for active logins
4909 goes becomes unavailable, and so does access to the user
4910 journals by the respective users.
4912 * If a group "adm" exists, journal files are automatically
4913 owned by them, thus allow members of this group full access
4914 to the system journal as well as all user journals.
4916 * The journal now stores the SELinux context of the logging
4917 client for all entries.
4919 * Add C++ inclusion guards to all public headers
4921 * New output mode "cat" in the journal to print only text
4922 messages, without any meta data like date or time.
4924 * Include tiny X server wrapper as a temporary stop-gap to
4925 teach XOrg udev display enumeration. This is used by display
4926 managers such as gdm, and will go away as soon as XOrg
4927 learned native udev hotplugging for display devices.
4929 * Add new systemd-cat tool for executing arbitrary programs
4930 with STDERR/STDOUT connected to the journal. Can also act as
4931 BSD logger replacement, and does so by default.
4933 * Optionally store all locally generated coredumps in the
4934 journal along with meta data.
4936 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
4937 writing short strings to files (for usage for /sys), and for
4938 creating symlinks, character and block device nodes.
4940 * New unit file option ControlGroupPersistent= to make cgroups
4941 persistent, following the mechanisms outlined in
4942 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
4944 * Support multiple local RTCs in a sane way
4946 * No longer monopolize IO when replaying readahead data on
4947 rotating disks, since we might starve non-file-system IO to
4948 death, since fanotify() will not see accesses done by blkid,
4951 * Do not show kernel threads in systemd-cgls anymore, unless
4952 requested with new -k switch.
4954 Contributions from: Dan Horák, Kay Sievers, Lennart
4955 Poettering, Michal Schmidt
4959 * This is mostly a test release, but incorporates many
4962 * The git repository moved to:
4963 git://anongit.freedesktop.org/systemd/systemd
4964 ssh://git.freedesktop.org/git/systemd/systemd
4966 * First release with the journal
4967 http://0pointer.de/blog/projects/the-journal.html
4969 * The journal replaces both systemd-kmsg-syslogd and
4970 systemd-stdout-bridge.
4972 * New sd_pid_get_unit() API call in libsystemd-logind
4974 * Many systemadm clean-ups
4976 * Introduce remote-fs-pre.target which is ordered before all
4977 remote mounts and may be used to start services before all
4980 * Added Mageia support
4982 * Add bash completion for systemd-loginctl
4984 * Actively monitor PID file creation for daemons which exit in
4985 the parent process before having finished writing the PID
4986 file in the daemon process. Daemons which do this need to be
4987 fixed (i.e. PID file creation must have finished before the
4988 parent exits), but we now react a bit more gracefully to them.
4990 * Add colourful boot output, mimicking the well-known output
4991 of existing distributions.
4993 * New option PassCredentials= for socket units, for
4994 compatibility with a recent kernel ABI breakage.
4996 * /etc/rc.local is now hooked in via a generator binary, and
4997 thus will no longer act as synchronization point during
5000 * systemctl list-unit-files now supports --root=.
5002 * systemd-tmpfiles now understands two new commands: z, Z for
5003 relabelling files according to the SELinux database. This is
5004 useful to apply SELinux labels to specific files in /sys,
5007 * Output of SysV services is now forwarded to both the console
5008 and the journal by default, not only just the console.
5010 * New man pages for all APIs from libsystemd-login.
5012 * The build tree got reorganized and a the build system is a
5013 lot more modular allowing embedded setups to specifically
5014 select the components of systemd they are interested in.
5016 * Support for Linux systems lacking the kernel VT subsystem is
5019 * configure's --with-rootdir= got renamed to
5020 --with-rootprefix= to follow the naming used by udev and
5023 * Unless specified otherwise we will now install to /usr instead
5024 of /usr/local by default.
5026 * Processes with '@' in argv[0][0] are now excluded from the
5027 final shut-down killing spree, following the logic explained
5029 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
5031 * All processes remaining in a service cgroup when we enter
5032 the START or START_PRE states are now killed with
5033 SIGKILL. That means it is no longer possible to spawn
5034 background processes from ExecStart= lines (which was never
5035 supported anyway, and bad style).
5037 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
5038 reloading of units together.
5040 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
5041 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
5042 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
5043 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
5044 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek