1 systemd System and Service Manager
5 * Support for chkconfig (--enable-chkconfig) was removed in favour of
6 calling an abstraction /lib/systemd/systemd-sysv-install. This needs
7 to be implemented for your distribution. See "SYSV INIT.D SCRIPTS" in
10 Contributions from: ...
16 * The gudev library has been extracted into a separate repository
17 available at: https://git.gnome.org/browse/libgudev/
18 It is now managed as part of the Gnome project. Distributions
19 are recommended to pass --disable-gudev to systemd and use
20 gudev from the Gnome project instead. gudev is still included
21 in systemd, for now. It will be removed soon, though. Please
22 also see the announcement-thread on systemd-devel:
23 http://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
25 * systemd now exposes a CPUUsageNSec= property for each
26 service unit on the bus, that contains the overall consumed
27 CPU time of a service (the sum of what each process of the
28 service consumed). This value is only available if
29 CPUAccounting= is turned on for a service, and is then shown
30 in the "systemctl status" output.
32 * Support for configuring alternative mappings of the old SysV
33 runlevels to systemd targets has been removed. They are now
34 hardcoded in a way that runlevels 2, 3, 4 all map to
35 multi-user.target and 5 to graphical.target (which
36 previously was already the default behaviour).
38 * The auto-mounter logic gained support for mount point
39 expiry, using a new TimeoutIdleSec= setting in .automount
40 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
42 * The EFI System Partition (ESP) as mounted to /boot by
43 systemd-efi-boot-generator will now be unmounted
44 automatically after 2 minutes of not being used. This should
45 minimize the risk of ESP corruptions.
47 * New /etc/fstab options x-systemd.requires= and
48 x-systemd.requires-mounts-for= are now supported to express
49 additional dependencies for mounts. This is useful for
50 journalling file systems that support external journal
51 devices or overlay file systems that require underlying file
52 systems to be mounted.
54 * systemd does not support direct live-upgrades (via systemctl
55 daemon-reexec) from versions older than v44 anymore. As no
56 distribution we are aware of shipped such old versions in a
57 stable release this should not be problematic.
59 * When systemd forks off a new per-connection service instance
60 it will now set the $REMOTE_ADDR environment variable to the
61 remote IP address, and $REMOTE_PORT environment variable to
62 the remote IP port. This behaviour is similar to the
63 corresponding environment variables defined by CGI.
65 * systemd-networkd gained support for uplink failure
66 detection. The BindCarrier= option allows binding interface
67 configuration dynamically to the link sense of other
68 interfaces. This is useful to achieve behaviour like in
71 * systemd-networkd gained support for configuring the DHCP
72 client identifier to use when requesting leases.
74 * systemd-networkd now has a per-network UseNTP= option to
75 configure whether NTP server information acquired via DHCP
76 is passed on to services like systemd-timesyncd.
78 * systemd-networkd gained support for vti6 tunnels.
80 * Note that systemd-networkd manages the sysctl variable
81 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
82 it is configured for since v219. The variable controls IP
83 forwarding, and is a per-interface alternative to the global
84 /proc/sys/net/ipv[46]/ip_forward. This setting is
85 configurable in the IPForward= option, which defaults to
86 "no". This means if networkd is used for an interface it is
87 no longer sufficient to set the global sysctl option to turn
88 on IP forwarding! Instead, the .network file option
89 IPForward= needs to be turned on! Note that the
90 implementation of this behaviour was broken in v219 and has
93 * Many bonding and vxlan options are now configurable in
96 * systemd-nspawn gained a new --property= setting to set unit
97 properties for the container scope. This is useful for
98 setting resource parameters (e.g "CPUShares=500") on
99 containers started from the command line.
101 * systemd-nspawn gained a new --private-users= switch to make
102 use of user namespacing available on recent Linux kernels.
104 * systemd-nspawn may now be called as part of a shell pipeline
105 in which case the pipes used for stdin and stdout are passed
106 directly to the process invoked in the container, without
107 indirection via a pseudo tty.
109 * systemd-nspawn gained a new switch to control the UNIX
110 signal to use when killing the init process of the container
113 * systemd-nspawn gained a new --overlay= switch for mounting
114 overlay file systems into the container using the new kernel
117 * When a container image is imported via systemd-importd and
118 the host file system is not btrfs, a loopback block device
119 file is created in /var/lib/machines.raw with a btrfs file
120 system inside. It is then mounted to /var/lib/machines to
121 enable btrfs features for container management. The loopback
122 file and btrfs file system is grown as needed when container
123 images are imported via systemd-importd.
125 * systemd-machined/systemd-importd gained support for btrfs
126 quota, to enforce container disk space limits on disk. This
127 is exposed in "machinectl set-limit".
129 * systemd-importd now can import containers from local .tar,
130 .raw and .qcow2 images, and export them to .tar and .raw. It
131 can also import dkr v2 images now from the network (on top
134 * systemd-importd gained support for verifying downloaded
135 images with gpg2 (previously only gpg1 was supported).
137 * systemd-machined, systemd-logind, systemd: most bus calls
138 are now accessible to unprivileged processes via
139 PolicyKit. Also, systemd-logind will now allow users to kill
140 their own sessions without further privileges or
143 * systemd-shutdownd has been removed. This service was
144 previously responsible for implementing scheduled shutdowns
145 as exposed in /usr/bin/shutdown's time parameter. This
146 functionality has now been moved into systemd-logind and is
147 accessible via a bus interface.
149 * "systemctl reboot" gained a new switch --firmware-setup that
150 can be used to reboot into the EFI firmware setup, if that
151 is available. systemd-logind now exposes an API on the bus
152 to trigger such reboots, in case graphical desktop UIs want
153 to cover this functionality.
155 * "systemctl enable", "systemctl disable" and "systemctl mask"
156 now support a new "--now" switch. If specified the units
157 that are enabled will also be started, and the ones
158 disabled/masked also stopped.
160 * The Gummiboot EFI boot loader tool has been merged into
161 systemd, and renamed to "systemd-boot". The bootctl tool has been
162 updated to support systemd-boot.
164 * An EFI kernel stub has been added that may be used to create
165 kernel EFI binaries that contain not only the actual kernel,
166 but also an initrd, boot splash, command line and OS release
167 information. This combined binary can then be signed as a
168 single image, so that the firmware can verify it all in one
169 step. systemd-boot has special support for EFI binaries created
170 like this and can extract OS release information from them
171 and show them in the boot menu. This functionality is useful
172 to implement cryptographically verified boot schemes.
174 * Optional support has been added to systemd-fsck to pass
175 fsck's progress report to an AF_UNIX socket in the file
178 * udev will no longer create device symlinks for all block
179 devices by default. A blacklist for excluding special block
180 devices from this logic has been turned into a whitelist
181 that requires picking block devices explicitly that require
184 * A new (currently still internal) API sd-device.h has been
185 added to libsystemd. This modernized API is supposed to
186 replace libudev eventually. In fact, already much of libudev
187 is now just a wrapper around sd-device.h.
189 * A new hwdb database for storing metadata about pointing
190 stick devices has been added.
192 * systemd-tmpfiles gained support for setting file attributes
193 similar to the "chattr" tool with new 'h' and 'H' lines.
195 * systemd-journald will no longer unconditionally set the
196 btrfs NOCOW flag on new journal files. This is instead done
197 with tmpfiles snippet using the new 'h' line type. This
198 allows easy disabling of this logic, by masking the
199 journal-nocow.conf tmpfiles file.
201 * systemd-journald will now translate audit message types to
202 human readable identifiers when writing them to the
203 journal. This should improve readability of audit messages.
205 * The LUKS logic gained support for the offset= and skip=
206 options in /etc/crypttab, as previously implemented by
209 * /usr/lib/os-release gained a new optional field VARIANT= for
210 distributions that support multiple variants (such as a
211 desktop edition, a server edition, ...)
213 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
214 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
215 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
216 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
217 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
218 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
219 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
220 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
221 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
222 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
223 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
224 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
225 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
226 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
227 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
228 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
229 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
230 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
231 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
232 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
233 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
234 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
235 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
236 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
237 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
238 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
239 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
241 -- Berlin, 2015-05-22
245 * Introduce a new API "sd-hwdb.h" for querying the hardware
246 metadata database. With this minimal interface one can query
247 and enumerate the udev hwdb, decoupled from the old libudev
248 library. libudev's interface for this is now only a wrapper
249 around sd-hwdb. A new tool systemd-hwdb has been added to
250 interface with and update the database.
252 * When any of systemd's tools copies files (for example due to
253 tmpfiles' C lines) a btrfs reflink will attempted first,
254 before bytewise copying is done.
256 * systemd-nspawn gained a new --ephemeral switch. When
257 specified a btrfs snapshot is taken of the container's root
258 directory, and immediately removed when the container
259 terminates again. Thus, a container can be started whose
260 changes never alter the container's root directory, and are
261 lost on container termination. This switch can also be used
262 for starting a container off the root file system of the
263 host without affecting the host OS. This switch is only
264 available on btrfs file systems.
266 * systemd-nspawn gained a new --template= switch. It takes the
267 path to a container tree to use as template for the tree
268 specified via --directory=, should that directory be
269 missing. This allows instantiating containers dynamically,
270 on first run. This switch is only available on btrfs file
273 * When a .mount unit refers to a mount point on which multiple
274 mounts are stacked, and the .mount unit is stopped all of
275 the stacked mount points will now be unmounted until no
278 * systemd now has an explicit notion of supported and
279 unsupported unit types. Jobs enqueued for unsupported unit
280 types will now fail with an "unsupported" error code. More
281 specifically .swap, .automount and .device units are not
282 supported in containers, .busname units are not supported on
283 non-kdbus systems. .swap and .automount are also not
284 supported if their respective kernel compile time options
287 * machinectl gained support for two new "copy-from" and
288 "copy-to" commands for copying files from a running
289 container to the host or vice versa.
291 * machinectl gained support for a new "bind" command to bind
292 mount host directories into local containers. This is
293 currently only supported for nspawn containers.
295 * networkd gained support for configuring bridge forwarding
296 database entries (fdb) from .network files.
298 * A new tiny daemon "systemd-importd" has been added that can
299 download container images in tar, raw, qcow2 or dkr formats,
300 and make them available locally in /var/lib/machines, so
301 that they can run as nspawn containers. The daemon can GPG
302 verify the downloads (not supported for dkr, since it has no
303 provisions for verifying downloads). It will transparently
304 decompress bz2, xz, gzip compressed downloads if necessary,
305 and restore sparse files on disk. The daemon uses privilege
306 separation to ensure the actual download logic runs with
307 fewer privileges than the deamon itself. machinectl has
308 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
309 make the functionality of importd available to the
310 user. With this in place the Fedora and Ubuntu "Cloud"
311 images can be downloaded and booted as containers unmodified
312 (the Fedora images lack the appropriate GPG signature files
313 currently, so they cannot be verified, but this will change
314 soon, hopefully). Note that downloading images is currently
315 only fully supported on btrfs.
317 * machinectl is now able to list container images found in
318 /var/lib/machines, along with some metadata about sizes of
319 disk and similar. If the directory is located on btrfs and
320 quota is enabled, this includes quota display. A new command
321 "image-status" has been added that shows additional
322 information about images.
324 * machinectl is now able to clone container images
325 efficiently, if the underlying file system (btrfs) supports
326 it, with the new "machinectl list-images" command. It also
327 gained commands for renaming and removing images, as well as
328 marking them read-only or read-write (supported also on
329 legacy file systems).
331 * networkd gained support for collecting LLDP network
332 announcements, from hardware that supports this. This is
333 shown in networkctl output.
335 * systemd-run gained support for a new -t (--pty) switch for
336 invoking a binary on a pty whose input and output is
337 connected to the invoking terminal. This allows executing
338 processes as system services while interactively
339 communicating with them via the terminal. Most interestingly
340 this is supported across container boundaries. Invoking
341 "systemd-run -t /bin/bash" is an alternative to running a
342 full login session, the difference being that the former
343 will not register a session, nor go through the PAM session
346 * tmpfiles gained support for a new "v" line type for creating
347 btrfs subvolumes. If the underlying file system is a legacy
348 file system, this automatically degrades to creating a
349 normal directory. Among others /var/lib/machines is now
350 created like this at boot, should it be missing.
352 * The directory /var/lib/containers/ has been deprecated and
353 been replaced by /var/lib/machines. The term "machines" has
354 been used in the systemd context as generic term for both
355 VMs and containers, and hence appears more appropriate for
356 this, as the directory can also contain raw images bootable
359 * systemd-nspawn when invoked with -M but without --directory=
360 or --image= is now capable of searching for the container
361 root directory, subvolume or disk image automatically, in
362 /var/lib/machines. systemd-nspawn@.service has been updated
363 to make use of this, thus allowing it to be used for raw
366 * A new machines.target unit has been introduced that is
367 supposed to group all containers/VMs invoked as services on
368 the system. systemd-nspawn@.service has been updated to
371 * machinectl gained a new "start" command, for invoking a
372 container as a service. "machinectl start foo" is mostly
373 equivalent to "systemctl start systemd-nspawn@foo.service",
374 but handles escaping in a nicer way.
376 * systemd-nspawn will now mount most of the cgroupfs tree
377 read-only into each container, with the exception of the
378 container's own subtree in the name=systemd hierarchy.
380 * journald now sets the special FS_NOCOW file flag for its
381 journal files. This should improve performance on btrfs, by
382 avoiding heavy fragmentation when journald's write-pattern
383 is used on COW file systems. It degrades btrfs' data
384 integrity guarantees for the files to the same levels as for
385 ext3/ext4 however. This should be OK though as journald does
386 its own data integrity checks and all its objects are
387 checksummed on disk. Also, journald should handle btrfs disk
388 full events a lot more gracefully now, by processing SIGBUS
389 errors, and not relying on fallocate() anymore.
391 * When journald detects that journal files it is writing to
392 have been deleted it will immediately start new journal
395 * systemd now provides a way to store file descriptors
396 per-service in PID 1.This is useful for daemons to ensure
397 that fds they require are not lost during a daemon
398 restart. The fds are passed to the deamon on the next
399 invocation in the same way socket activation fds are
400 passed. This is now used by journald to ensure that the
401 various sockets connected to all the system's stdout/stderr
402 are not lost when journald is restarted. File descriptors
403 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
404 an extension to sd_notify(). Note that a limit is enforced
405 on the number of fds a service can store in PID 1, and it
406 defaults to 0, so that no fds may be stored, unless this is
407 explicitly turned on.
409 * The default TERM variable to use for units connected to a
410 terminal, when no other value is explicitly is set is now
411 vt220 rather than vt102. This should be fairly safe still,
412 but allows PgUp/PgDn work.
414 * The /etc/crypttab option header= as known from Debian is now
417 * "loginctl user-status" and "loginctl session-status" will
418 now show the last 10 lines of log messages of the
419 user/session following the status output. Similar,
420 "machinectl status" will show the last 10 log lines
421 associated with a virtual machine or container
422 service. (Note that this is usually not the log messages
423 done in the VM/container itself, but simply what the
424 container manager logs. For nspawn this includes all console
427 * "loginctl session-status" without further argument will now
428 show the status of the session of the caller. Similar,
429 "lock-session", "unlock-session", "activate",
430 "enable-linger", "disable-linger" may now be called without
431 session/user parameter in which case they apply to the
432 caller's session/user.
434 * An X11 session scriptlet is now shipped that uploads
435 $DISPLAY and $XAUTHORITY into the environment of the systemd
436 --user daemon if a session begins. This should improve
437 compatibility with X11 enabled applications run as systemd
440 * Generators are now subject to masking via /etc and /run, the
441 same way as unit files.
443 * networkd .network files gained support for configuring
444 per-link IPv4/IPv6 packet forwarding as well as IPv4
445 masquerading. This is by default turned on for veth links to
446 containers, as registered by systemd-nspawn. This means that
447 nspawn containers run with --network-veth will now get
448 automatic routed access to the host's networks without any
449 further configuration or setup, as long as networkd runs on
452 * systemd-nspawn gained the --port= (-p) switch to expose TCP
453 or UDP posts of a container on the host. With this in place
454 it is possible to run containers with private veth links
455 (--network-veth), and have their functionality exposed on
456 the host as if their services were running directly on the
459 * systemd-nspawn's --network-veth switch now gained a short
460 version "-n", since with the changes above it is now truly
461 useful out-of-the-box. The systemd-nspawn@.service has been
462 updated to make use of it too by default.
464 * systemd-nspawn will now maintain a per-image R/W lock, to
465 ensure that the same image is not started more than once
466 writable. (It's OK to run an image multiple times
467 simultaneously in read-only mode.)
469 * systemd-nspawn's --image= option is now capable of
470 dissecting and booting MBR and GPT disk images that contain
471 only a single active Linux partition. Previously it
472 supported only GPT disk images with proper GPT type
473 IDs. This allows running cloud images from major
474 distributions directly with systemd-nspawn, without
477 * In addition to collecting mouse dpi data in the udev
478 hardware database, there's now support for collecting angle
479 information for mouse scroll wheels. The database is
480 supposed to guarantee similar scrolling behavior on mice
481 that it knows about. There's also support for collecting
482 information about Touchpad types.
484 * udev's input_id built-in will now also collect touch screen
485 dimension data and attach it to probed devices.
487 * /etc/os-release gained support for a Distribution Privacy
490 * networkd gained support for creating "ipvlan", "gretap",
491 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
493 * systemd-tmpfiles gained support for "a" lines for setting
496 * systemd-nspawn will now mount /tmp in the container to
497 tmpfs, automatically.
499 * systemd now exposes the memory.usage_in_bytes cgroup
500 attribute and shows it for each service in the "systemctl
501 status" output, if available.
503 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
504 immediate reboot is triggered. This useful if shutdown is
505 hung and is unable to complete, to expedite the
506 operation. Note that this kind of reboot will still unmount
507 all file systems, and hence should not result in fsck being
510 * A .device unit for an optical block device will now be
511 considered active only when a medium is in the drive. Also,
512 mount units are now bound to their backing devices thus
513 triggering automatic unmounting when devices become
514 unavailable. With this in place systemd will now
515 automatically unmount left-over mounts when a CD-ROM is
516 ejected or an USB stick is yanked from the system.
518 * networkd-wait-online now has support for waiting for
519 specific interfaces only (with globbing), and for giving up
520 after a configurable timeout.
522 * networkd now exits when idle. It will be automatically
523 restarted as soon as interfaces show up, are removed or
524 change state. networkd will stay around as long as there is
525 at least one DHCP state machine or similar around, that keep
528 * networkd may now configure IPv6 link-local addressing in
529 addition to IPv4 link-local addressing.
531 * The IPv6 "token" for use in SLAAC may now be configured for
532 each .network interface in networkd.
534 * Routes configured with networkd may now be assigned a scope
537 * networkd's [Match] sections now support globbing and lists
538 of multiple space-separated matches per item.
540 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
541 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
542 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
543 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
544 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
545 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
546 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
547 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
548 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
549 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
550 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
551 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
552 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
553 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
554 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
555 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
556 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
557 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
558 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
559 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
560 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
561 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
562 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
563 Hoffmann, Zbigniew Jędrzejewski-Szmek
565 -- Berlin, 2015-02-16
569 * When querying unit file enablement status (for example via
570 "systemctl is-enabled"), a new state "indirect" is now known
571 which indicates that a unit might not be enabled itself, but
572 another unit listed in its Alias= setting might be.
574 * Similar to the various existing ConditionXYZ= settings for
575 units there are now matching AssertXYZ= settings. While
576 failing conditions cause a unit to be skipped, but its job
577 to succeed, failing assertions declared like this will cause
578 a unit start operation and its job to fail.
580 * hostnamed now knows a new chassis type "embedded".
582 * systemctl gained a new "edit" command. When used on a unit
583 file this allows extending unit files with .d/ drop-in
584 configuration snippets or editing the full file (after
585 copying it from /usr/lib to /etc). This will invoke the
586 user's editor (as configured with $EDITOR), and reload the
587 modified configuration after editing.
589 * "systemctl status" now shows the suggested enablement state
590 for a unit, as declared in the (usually vendor-supplied)
593 * nss-myhostname will now resolve the single-label host name
594 "gateway" to the locally configured default IP routing
595 gateways, ordered by their metrics. This assigns a stable
596 name to the used gateways, regardless which ones are
597 currently configured. Note that the name will only be
598 resolved after all other name sources (if nss-myhostname is
599 configured properly) and should hence not negatively impact
600 systems that use the single-label host name "gateway" in
603 * systemd-inhibit now allows filtering by mode when listing
606 * Scope and service units gained a new "Delegate" boolean
607 property, which when set allows processes running inside the
608 unit to further partition resources. This is primarily
609 useful for systemd user instances as well as container
612 * journald will now pick up audit messages directly from
613 the kernel, and log them like any other log message. The
614 audit fields are split up and fully indexed. This means that
615 journalctl in many ways is now a (nicer!) alternative to
616 ausearch, the traditional audit client. Note that this
617 implements only a minimal audit client, if you want the
618 special audit modes like reboot-on-log-overflow, please use
619 the traditional auditd instead, which can be used in
620 parallel to journald.
622 * The ConditionSecurity= unit file option now understands the
623 special string "audit" to check whether auditing is
626 * journalctl gained two new commands --vacuum-size= and
627 --vacuum-time= to delete old journal files until the
628 remaining ones take up no more the specified size on disk,
629 or are not older than the specified time.
631 * A new, native PPPoE library has been added to sd-network,
632 systemd's library of light-weight networking protocols. This
633 library will be used in a future version of networkd to
634 enable PPPoE communication without an external pppd daemon.
636 * The busctl tool now understands a new "capture" verb that
637 works similar to "monitor", but writes a packet capture
638 trace to STDOUT that can be redirected to a file which is
639 compatible with libcap's capture file format. This can then
640 be loaded in Wireshark and similar tools to inspect bus
643 * The busctl tool now understands a new "tree" verb that shows
644 the object trees of a specific service on the bus, or of all
647 * The busctl tool now understands a new "introspect" verb that
648 shows all interfaces and members of objects on the bus,
649 including their signature and values. This is particularly
650 useful to get more information about bus objects shown by
651 the new "busctl tree" command.
653 * The busctl tool now understands new verbs "call",
654 "set-property" and "get-property" for invoking bus method
655 calls, setting and getting bus object properties in a
658 * busctl gained a new --augment-creds= argument that controls
659 whether the tool shall augment credential information it
660 gets from the bus with data from /proc, in a possibly
663 * nspawn's --link-journal= switch gained two new values
664 "try-guest" and "try-host" that work like "guest" and
665 "host", but do not fail if the host has no persistent
666 journalling enabled. -j is now equivalent to
667 --link-journal=try-guest.
669 * macvlan network devices created by nspawn will now have
670 stable MAC addresses.
672 * A new SmackProcessLabel= unit setting has been added, which
673 controls the SMACK security label processes forked off by
674 the respective unit shall use.
676 * If compiled with --enable-xkbcommon, systemd-localed will
677 verify x11 keymap settings by compiling the given keymap. It
678 will spew out warnings if the compilation fails. This
679 requires libxkbcommon to be installed.
681 * When a coredump is collected a larger number of metadata
682 fields is now collected and included in the journal records
683 created for it. More specifically control group membership,
684 environment variables, memory maps, working directory,
685 chroot directory, /proc/$PID/status, and a list of open file
686 descriptors is now stored in the log entry.
688 * The udev hwdb now contains DPI information for mice. For
691 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
693 * All systemd programs that read standalone configuration
694 files in /etc now also support a corresponding series of
695 .conf.d configuration directories in /etc/, /run/,
696 /usr/local/lib/, /usr/lib/, and (if configured with
697 --enable-split-usr) /lib/. In particular, the following
698 configuration files now have corresponding configuration
699 directories: system.conf user.conf, logind.conf,
700 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
701 resolved.conf, timesyncd.conf, journal-remote.conf, and
702 journal-upload.conf. Note that distributions should use the
703 configuration directories in /usr/lib/; the directories in
704 /etc/ are reserved for the system administrator.
706 * systemd-rfkill will no longer take the rfkill device name
707 into account when storing rfkill state on disk, as the name
708 might be dynamically assigned and not stable. Instead, the
709 ID_PATH udev variable combined with the rfkill type (wlan,
710 bluetooth, ...) is used.
712 * A new service systemd-machine-id-commit.service has been
713 added. When used on systems where /etc is read-only during
714 boot, and /etc/machine-id is not initialized (but an empty
715 file), this service will copy the temporary machine ID
716 created as replacement into /etc after the system is fully
717 booted up. This is useful for systems that are freshly
718 installed with a non-initialized machine ID, but should get
719 a fixed machine ID for subsequent boots.
721 * networkd's .netdev files now provide a large set of
722 configuration parameters for VXLAN devices. Similar, the
723 bridge port cost parameter is now configurable in .network
724 files. There's also new support for configuring IP source
725 routing. networkd .link files gained support for a new
726 OriginalName= match that is useful to match against the
727 original interface name the kernel assigned. .network files
728 may include MTU= and MACAddress= fields for altering the MTU
729 and MAC address while being connected to a specific network
732 * The LUKS logic gained supported for configuring
733 UUID-specific key files. There's also new support for naming
734 LUKS device from the kernel command line, using the new
737 * Timer units may now be transiently created via the bus API
738 (this was previously already available for scope and service
739 units). In addition it is now possible to create multiple
740 transient units at the same time with a single bus call. The
741 "systemd-run" tool has been updated to make use of this for
742 running commands on a specified time, in at(1)-style.
744 * tmpfiles gained support for "t" lines, for assigning
745 extended attributes to files. Among other uses this may be
746 used to assign SMACK labels to files.
748 Contributions from: Alin Rauta, Alison Chaiken, Andrej
749 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
750 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
751 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
752 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
753 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
754 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
755 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
756 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
757 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
758 Michael Biebl, Michael Chapman, Michael Marineau, Michal
759 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
760 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
761 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
762 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
763 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
764 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
765 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
767 -- Berlin, 2014-12-10
771 * journalctl gained the new options -t/--identifier= to match
772 on the syslog identifier (aka "tag"), as well as --utc to
773 show log timestamps in the UTC timezone. journalctl now also
774 accepts -n/--lines=all to disable line capping in a pager.
776 * journalctl gained a new switch, --flush, that synchronously
777 flushes logs from /run/log/journal to /var/log/journal if
778 persistent storage is enabled. systemd-journal-flush.service
779 now waits until the operation is complete.
781 * Services can notify the manager before they start a reload
782 (by sending RELOADING=1) or shutdown (by sending
783 STOPPING=1). This allows the manager to track and show the
784 internal state of daemons and closes a race condition when
785 the process is still running but has closed its D-Bus
788 * Services with Type=oneshot do not have to have any ExecStart
791 * User units are now loaded also from
792 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
793 /run/systemd/user directory that was already previously
794 supported, but is under the control of the user.
796 * Job timeouts (i.e. time-outs on the time a job that is
797 queued stays in the run queue) can now optionally result in
798 immediate reboot or power-off actions (JobTimeoutAction= and
799 JobTimeoutRebootArgument=). This is useful on ".target"
800 units, to limit the maximum time a target remains
801 undispatched in the run queue, and to trigger an emergency
802 operation in such a case. This is now used by default to
803 turn off the system if boot-up (as defined by everything in
804 basic.target) hangs and does not complete for at least
805 15min. Also, if power-off or reboot hang for at least 30min
806 an immediate power-off/reboot operation is triggered. This
807 functionality is particularly useful to increase reliability
808 on embedded devices, but also on laptops which might
809 accidentally get powered on when carried in a backpack and
810 whose boot stays stuck in a hard disk encryption passphrase
813 * systemd-logind can be configured to also handle lid switch
814 events even when the machine is docked or multiple displays
815 are attached (HandleLidSwitchDocked= option).
817 * A helper binary and a service have been added which can be
818 used to resume from hibernation in the initramfs. A
819 generator will parse the resume= option on the kernel
820 command line to trigger resume.
822 * A user console daemon systemd-consoled has been
823 added. Currently, it is a preview, and will so far open a
824 single terminal on each session of the user marked as
825 Desktop=systemd-console.
827 * Route metrics can be specified for DHCP routes added by
830 * The SELinux context of socket-activated services can be set
831 from the information provided by the networking stack
832 (SELinuxContextFromNet= option).
834 * Userspace firmware loading support has been removed and
835 the minimum supported kernel version is thus bumped to 3.7.
837 * Timeout for udev workers has been increased from 1 to 3
838 minutes, but a warning will be printed after 1 minute to
839 help diagnose kernel modules that take a long time to load.
841 * Udev rules can now remove tags on devices with TAG-="foobar".
843 * systemd's readahead implementation has been removed. In many
844 circumstances it didn't give expected benefits even for
845 rotational disk drives and was becoming less relevant in the
846 age of SSDs. As none of the developers has been using
847 rotating media anymore, and nobody stepped up to actively
848 maintain this component of systemd it has now been removed.
850 * Swap units can use Options= to specify discard options.
851 Discard options specified for swaps in /etc/fstab are now
854 * Docker containers are now detected as a separate type of
857 * The Password Agent protocol gained support for queries where
858 the user input is shown, useful e.g. for user names.
859 systemd-ask-password gained a new --echo option to turn that
862 * The default sysctl.d/ snippets will now set:
864 net.core.default_qdisc = fq_codel
866 This selects Fair Queuing Controlled Delay as the default
867 queuing discipline for network interfaces. fq_codel helps
868 fight the network bufferbloat problem. It is believed to be
869 a good default with no tuning required for most workloads.
870 Downstream distributions may override this choice. On 10Gbit
871 servers that do not do forwarding, "fq" may perform better.
872 Systems without a good clocksource should use "pfifo_fast".
874 * If kdbus is enabled during build a new option BusPolicy= is
875 available for service units, that allows locking all service
876 processes into a stricter bus policy, in order to limit
877 access to various bus services, or even hide most of them
878 from the service's view entirely.
880 * networkctl will now show the .network and .link file
881 networkd has applied to a specific interface.
883 * sd-login gained a new API call sd_session_get_desktop() to
884 query which desktop environment has been selected for a
887 * UNIX utmp support is now compile-time optional to support
890 * systemctl gained two new commands "add-wants" and
891 "add-requires" for pulling in units from specific targets
894 * If the word "rescue" is specified on the kernel command line
895 the system will now boot into rescue mode (aka
896 rescue.target), which was previously available only by
897 specifying "1" or "systemd.unit=rescue.target" on the kernel
898 command line. This new kernel command line option nicely
899 mirrors the already existing "emergency" kernel command line
902 * New kernel command line options mount.usr=, mount.usrflags=,
903 mount.usrfstype= have been added that match root=, rootflags=,
904 rootfstype= but allow mounting a specific file system to
907 * The $NOTIFY_SOCKET is now also passed to control processes of
908 services, not only the main process.
910 * This version reenables support for fsck's -l switch. This
911 means at least version v2.25 of util-linux is required for
912 operation, otherwise dead-locks on device nodes may
913 occur. Again: you need to update util-linux to at least
914 v2.25 when updating systemd to v217.
916 * The "multi-seat-x" tool has been removed from systemd, as
917 its functionality has been integrated into X servers 1.16,
918 and the tool is hence redundant. It is recommended to update
919 display managers invoking this tool to simply invoke X
920 directly from now on, again.
922 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
923 message flag has been added for all of systemd's PolicyKit
924 authenticated method calls has been added. In particular
925 this now allows optional interactive authorization via
926 PolicyKit for many of PID1's privileged operations such as
927 unit file enabling and disabling.
929 * "udevadm hwdb --update" learnt a new switch "--usr" for
930 placing the rebuilt hardware database in /usr instead of
931 /etc. When used only hardware database entries stored in
932 /usr will be used, and any user database entries in /etc are
933 ignored. This functionality is useful for vendors to ship a
934 pre-built database on systems where local configuration is
935 unnecessary or unlikely.
937 * Calendar time specifications in .timer units now also
938 understand the strings "semi-annually", "quarterly" and
939 "minutely" as shortcuts (in addition to the preexisting
940 "anually", "hourly", ...).
942 * systemd-tmpfiles will now correctly create files in /dev
943 at boot which are marked for creation only at boot. It is
944 recommended to always create static device nodes with 'c!'
945 and 'b!', so that they are created only at boot and not
946 overwritten at runtime.
948 * When the watchdog logic is used for a service (WatchdogSec=)
949 and the watchdog timeout is hit the service will now be
950 terminated with SIGABRT (instead of just SIGTERM), in order
951 to make sure a proper coredump and backtrace is
952 generated. This ensures that hanging services will result in
953 similar coredump/backtrace behaviour as services that hit a
956 Contributions from: Andreas Henriksson, Andrei Borzenkov,
957 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
958 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
959 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
960 Herrmann, David Sommerseth, David Strauss, Emil Renner
961 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
962 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
963 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
964 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
965 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
966 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
967 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
968 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
969 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
970 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
971 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
972 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
973 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
974 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
975 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
976 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
979 -- Berlin, 2014-10-28
983 * timedated no longer reads NTP implementation unit names from
984 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
985 implementations should add a
987 Conflicts=systemd-timesyncd.service
989 to their unit files to take over and replace systemd's NTP
990 default functionality.
992 * systemd-sysusers gained a new line type "r" for configuring
993 which UID/GID ranges to allocate system users/groups
994 from. Lines of type "u" may now add an additional column
995 that specifies the home directory for the system user to be
996 created. Also, systemd-sysusers may now optionally read user
997 information from STDIN instead of a file. This is useful for
998 invoking it from RPM preinst scriptlets that need to create
999 users before the first RPM file is installed since these
1000 files might need to be owned by them. A new
1001 %sysusers_create_inline RPM macro has been introduced to do
1002 just that. systemd-sysusers now updates the shadow files as
1003 well as the user/group databases, which should enhance
1004 compatibility with certain tools like grpck.
1006 * A number of bus APIs of PID 1 now optionally consult
1007 PolicyKit to permit access for otherwise unprivileged
1008 clients under certain conditions. Note that this currently
1009 doesn't support interactive authentication yet, but this is
1010 expected to be added eventually, too.
1012 * /etc/machine-info now has new fields for configuring the
1013 deployment environment of the machine, as well as the
1014 location of the machine. hostnamectl has been updated with
1015 new command to update these fields.
1017 * systemd-timesyncd has been updated to automatically acquire
1018 NTP server information from systemd-networkd, which might
1019 have been discovered via DHCP.
1021 * systemd-resolved now includes a caching DNS stub resolver
1022 and a complete LLMNR name resolution implementation. A new
1023 NSS module "nss-resolve" has been added which can be used
1024 instead of glibc's own "nss-dns" to resolve hostnames via
1025 systemd-resolved. Hostnames, addresses and arbitrary RRs may
1026 be resolved via systemd-resolved D-Bus APIs. In contrast to
1027 the glibc internal resolver systemd-resolved is aware of
1028 multi-homed system, and keeps DNS server and caches separate
1029 and per-interface. Queries are sent simultaneously on all
1030 interfaces that have DNS servers configured, in order to
1031 properly handle VPNs and local LANs which might resolve
1032 separate sets of domain names. systemd-resolved may acquire
1033 DNS server information from systemd-networkd automatically,
1034 which in turn might have discovered them via DHCP. A tool
1035 "systemd-resolve-host" has been added that may be used to
1036 query the DNS logic in resolved. systemd-resolved implements
1037 IDNA and automatically uses IDNA or UTF-8 encoding depending
1038 on whether classic DNS or LLMNR is used as transport. In the
1039 next releases we intend to add a DNSSEC and mDNS/DNS-SD
1040 implementation to systemd-resolved.
1042 * A new NSS module nss-mymachines has been added, that
1043 automatically resolves the names of all local registered
1044 containers to their respective IP addresses.
1046 * A new client tool "networkctl" for systemd-networkd has been
1047 added. It currently is entirely passive and will query
1048 networking configuration from udev, rtnetlink and networkd,
1049 and present it to the user in a very friendly
1050 way. Eventually, we hope to extend it to become a full
1051 control utility for networkd.
1053 * .socket units gained a new DeferAcceptSec= setting that
1054 controls the kernels' TCP_DEFER_ACCEPT sockopt for
1055 TCP. Similar, support for controlling TCP keep-alive
1056 settings has been added (KeepAliveTimeSec=,
1057 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
1058 turning off Nagle's algorithm on TCP has been added
1061 * logind learned a new session type "web", for use in projects
1062 like Cockpit which register web clients as PAM sessions.
1064 * timer units with at least one OnCalendar= setting will now
1065 be started only after timer-sync.target has been
1066 reached. This way they will not elapse before the system
1067 clock has been corrected by a local NTP client or
1068 similar. This is particular useful on RTC-less embedded
1069 machines, that come up with an invalid system clock.
1071 * systemd-nspawn's --network-veth= switch should now result in
1072 stable MAC addresses for both the outer and the inner side
1075 * systemd-nspawn gained a new --volatile= switch for running
1076 container instances with /etc or /var unpopulated.
1078 * The kdbus client code has been updated to use the new Linux
1079 3.17 memfd subsystem instead of the old kdbus-specific one.
1081 * systemd-networkd's DHCP client and server now support
1082 FORCERENEW. There are also new configuration options to
1083 configure the vendor client identifier and broadcast mode
1086 * systemd will no longer inform the kernel about the current
1087 timezone, as this is necessarily incorrect and racy as the
1088 kernel has no understanding of DST and similar
1089 concepts. This hence means FAT timestamps will be always
1090 considered UTC, similar to what Android is already
1091 doing. Also, when the RTC is configured to the local time
1092 (rather than UTC) systemd will never synchronize back to it,
1093 as this might confuse Windows at a later boot.
1095 * systemd-analyze gained a new command "verify" for offline
1096 validation of unit files.
1098 * systemd-networkd gained support for a couple of additional
1099 settings for bonding networking setups. Also, the metric for
1100 statically configured routes may now be configured. For
1101 network interfaces where this is appropriate the peer IP
1102 address may now be configured.
1104 * systemd-networkd's DHCP client will no longer request
1105 broadcasting by default, as this tripped up some networks.
1106 For hardware where broadcast is required the feature should
1107 be switched back on using RequestBroadcast=yes.
1109 * systemd-networkd will now set up IPv4LL addresses (when
1110 enabled) even if DHCP is configured successfully.
1112 * udev will now default to respect network device names given
1113 by the kernel when the kernel indicates that these are
1114 predictable. This behavior can be tweaked by changing
1115 NamePolicy= in the relevant .link file.
1117 * A new library systemd-terminal has been added that
1118 implements full TTY stream parsing and rendering. This
1119 library is supposed to be used later on for implementing a
1120 full userspace VT subsystem, replacing the current kernel
1123 * A new tool systemd-journal-upload has been added to push
1124 journal data to a remote system running
1125 systemd-journal-remote.
1127 * journald will no longer forward all local data to another
1128 running syslog daemon. This change has been made because
1129 rsyslog (which appears to be the most commonly used syslog
1130 implementation these days) no longer makes use of this, and
1131 instead pulls the data out of the journal on its own. Since
1132 forwarding the messages to a non-existent syslog server is
1133 more expensive than we assumed we have now turned this
1134 off. If you run a syslog server that is not a recent rsyslog
1135 version, you have to turn this option on again
1136 (ForwardToSyslog= in journald.conf).
1138 * journald now optionally supports the LZ4 compressor for
1139 larger journal fields. This compressor should perform much
1140 better than XZ which was the previous default.
1142 * machinectl now shows the IP addresses of local containers,
1143 if it knows them, plus the interface name of the container.
1145 * A new tool "systemd-escape" has been added that makes it
1146 easy to escape strings to build unit names and similar.
1148 * sd_notify() messages may now include a new ERRNO= field
1149 which is parsed and collected by systemd and shown among the
1150 "systemctl status" output for a service.
1152 * A new component "systemd-firstboot" has been added that
1153 queries the most basic systemd information (timezone,
1154 hostname, root password) interactively on first
1155 boot. Alternatively it may also be used to provision these
1156 things offline on OS images installed into directories.
1158 * The default sysctl.d/ snippets will now set
1160 net.ipv4.conf.default.promote_secondaries=1
1162 This has the benefit of no flushing secondary IP addresses
1163 when primary addresses are removed.
1165 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
1166 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
1167 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
1168 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
1169 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
1170 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
1171 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
1172 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
1173 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
1174 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
1175 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
1176 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
1177 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
1178 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
1179 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
1181 -- Berlin, 2014-08-19
1185 * A new tool systemd-sysusers has been added. This tool
1186 creates system users and groups in /etc/passwd and
1187 /etc/group, based on static declarative system user/group
1188 definitions in /usr/lib/sysusers.d/. This is useful to
1189 enable factory resets and volatile systems that boot up with
1190 an empty /etc directory, and thus need system users and
1191 groups created during early boot. systemd now also ships
1192 with two default sysusers.d/ files for the most basic
1193 users and groups systemd and the core operating system
1196 * A new tmpfiles snippet has been added that rebuilds the
1197 essential files in /etc on boot, should they be missing.
1199 * A directive for ensuring automatic clean-up of
1200 /var/cache/man/ has been removed from the default
1201 configuration. This line should now be shipped by the man
1202 implementation. The necessary change has been made to the
1203 man-db implementation. Note that you need to update your man
1204 implementation to one that ships this line, otherwise no
1205 automatic clean-up of /var/cache/man will take place.
1207 * A new condition ConditionNeedsUpdate= has been added that
1208 may conditionalize services to only run when /etc or /var
1209 are "older" than the vendor operating system resources in
1210 /usr. This is useful for reconstructing or updating /etc
1211 after an offline update of /usr or a factory reset, on the
1212 next reboot. Services that want to run once after such an
1213 update or reset should use this condition and order
1214 themselves before the new systemd-update-done.service, which
1215 will mark the two directories as fully updated. A number of
1216 service files have been added making use of this, to rebuild
1217 the udev hardware database, the journald message catalog and
1218 dynamic loader cache (ldconfig). The systemd-sysusers tool
1219 described above also makes use of this now. With this in
1220 place it is now possible to start up a minimal operating
1221 system with /etc empty cleanly. For more information on the
1222 concepts involved see this recent blog story:
1224 http://0pointer.de/blog/projects/stateless.html
1226 * A new system group "input" has been introduced, and all
1227 input device nodes get this group assigned. This is useful
1228 for system-level software to get access to input devices. It
1229 complements what is already done for "audio" and "video".
1231 * systemd-networkd learnt minimal DHCPv4 server support in
1232 addition to the existing DHCPv4 client support. It also
1233 learnt DHCPv6 client and IPv6 Router Solicitation client
1234 support. The DHCPv4 client gained support for static routes
1235 passed in from the server. Note that the [DHCPv4] section
1236 known in older systemd-networkd versions has been renamed to
1237 [DHCP] and is now also used by the DHCPv6 client. Existing
1238 .network files using settings of this section should be
1239 updated, though compatibility is maintained. Optionally, the
1240 client hostname may now be sent to the DHCP server.
1242 * networkd gained support for vxlan virtual networks as well
1243 as tun/tap and dummy devices.
1245 * networkd gained support for automatic allocation of address
1246 ranges for interfaces from a system-wide pool of
1247 addresses. This is useful for dynamically managing a large
1248 number of interfaces with a single network configuration
1249 file. In particular this is useful to easily assign
1250 appropriate IP addresses to the veth links of a large number
1251 of nspawn instances.
1253 * RPM macros for processing sysusers, sysctl and binfmt
1254 drop-in snippets at package installation time have been
1257 * The /etc/os-release file should now be placed in
1258 /usr/lib/os-release. The old location is automatically
1259 created as symlink. /usr/lib is the more appropriate
1260 location of this file, since it shall actually describe the
1261 vendor operating system shipped in /usr, and not the
1262 configuration stored in /etc.
1264 * .mount units gained a new boolean SloppyOptions= setting
1265 that maps to mount(8)'s -s option which enables permissive
1266 parsing of unknown mount options.
1268 * tmpfiles learnt a new "L+" directive which creates a symlink
1269 but (unlike "L") deletes a pre-existing file first, should
1270 it already exist and not already be the correct
1271 symlink. Similar, "b+", "c+" and "p+" directives have been
1272 added as well, which create block and character devices, as
1273 well as fifos in the filesystem, possibly removing any
1274 pre-existing files of different types.
1276 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
1277 'argument' field (which so far specified the source to
1278 symlink/copy the files from) is now optional. If omitted the
1279 same file os copied from /usr/share/factory/ suffixed by the
1280 full destination path. This is useful for populating /etc
1281 with essential files, by copying them from vendor defaults
1282 shipped in /usr/share/factory/etc.
1284 * A new command "systemctl preset-all" has been added that
1285 applies the service preset settings to all installed unit
1286 files. A new switch --preset-mode= has been added that
1287 controls whether only enable or only disable operations
1290 * A new command "systemctl is-system-running" has been added
1291 that allows checking the overall state of the system, for
1292 example whether it is fully up and running.
1294 * When the system boots up with an empty /etc, the equivalent
1295 to "systemctl preset-all" is executed during early boot, to
1296 make sure all default services are enabled after a factory
1299 * systemd now contains a minimal preset file that enables the
1300 most basic services systemd ships by default.
1302 * Unit files' [Install] section gained a new DefaultInstance=
1303 field for defining the default instance to create if a
1304 template unit is enabled with no instance specified.
1306 * A new passive target cryptsetup-pre.target has been added
1307 that may be used by services that need to make they run and
1308 finish before the first LUKS cryptographic device is set up.
1310 * The /dev/loop-control and /dev/btrfs-control device nodes
1311 are now owned by the "disk" group by default, opening up
1312 access to this group.
1314 * systemd-coredump will now automatically generate a
1315 stack trace of all core dumps taking place on the system,
1316 based on elfutils' libdw library. This stack trace is logged
1319 * systemd-coredump may now optionally store coredumps directly
1320 on disk (in /var/lib/systemd/coredump, possibly compressed),
1321 instead of storing them unconditionally in the journal. This
1322 mode is the new default. A new configuration file
1323 /etc/systemd/coredump.conf has been added to configure this
1324 and other parameters of systemd-coredump.
1326 * coredumpctl gained a new "info" verb to show details about a
1327 specific coredump. A new switch "-1" has also been added
1328 that makes sure to only show information about the most
1329 recent entry instead of all entries. Also, as the tool is
1330 generally useful now the "systemd-" prefix of the binary
1331 name has been removed. Distributions that want to maintain
1332 compatibility with the old name should add a symlink from
1333 the old name to the new name.
1335 * journald's SplitMode= now defaults to "uid". This makes sure
1336 that unprivileged users can access their own coredumps with
1337 coredumpctl without restrictions.
1339 * New kernel command line options "systemd.wants=" (for
1340 pulling an additional unit during boot), "systemd.mask="
1341 (for masking a specific unit for the boot), and
1342 "systemd.debug-shell" (for enabling the debug shell on tty9)
1343 have been added. This is implemented in the new generator
1344 "systemd-debug-generator".
1346 * systemd-nspawn will now by default filter a couple of
1347 syscalls for containers, among them those required for
1348 kernel module loading, direct x86 IO port access, swap
1349 management, and kexec. Most importantly though
1350 open_by_handle_at() is now prohibited for containers,
1351 closing a hole similar to a recently discussed vulnerability
1352 in docker regarding access to files on file hierarchies the
1353 container should normally not have access to. Note that for
1354 nspawn we generally make no security claims anyway (and
1355 this is explicitly documented in the man page), so this is
1356 just a fix for one of the most obvious problems.
1358 * A new man page file-hierarchy(7) has been added that
1359 contains a minimized, modernized version of the file system
1360 layout systemd expects, similar in style to the FHS
1361 specification or hier(5). A new tool systemd-path(1) has
1362 been added to query many of these paths for the local
1365 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
1366 longer done. Since the directory now has a per-user size
1367 limit, and is cleaned on logout this appears unnecessary,
1368 in particular since this now brings the lifecycle of this
1369 directory closer in line with how IPC objects are handled.
1371 * systemd.pc now exports a number of additional directories,
1372 including $libdir (which is useful to identify the library
1373 path for the primary architecture of the system), and a
1374 couple of drop-in directories.
1376 * udev's predictable network interface names now use the dev_port
1377 sysfs attribute, introduced in linux 3.15 instead of dev_id to
1378 distinguish between ports of the same PCI function. dev_id should
1379 only be used for ports using the same HW address, hence the need
1382 * machined has been updated to export the OS version of a
1383 container (read from /etc/os-release and
1384 /usr/lib/os-release) on the bus. This is now shown in
1385 "machinectl status" for a machine.
1387 * A new service setting RestartForceExitStatus= has been
1388 added. If configured to a set of exit signals or process
1389 return values, the service will be restarted when the main
1390 daemon process exits with any of them, regardless of the
1393 * systemctl's -H switch for connecting to remote systemd
1394 machines has been extended so that it may be used to
1395 directly connect to a specific container on the
1396 host. "systemctl -H root@foobar:waldi" will now connect as
1397 user "root" to host "foobar", and then proceed directly to
1398 the container named "waldi". Note that currently you have to
1399 authenticate as user "root" for this to work, as entering
1400 containers is a privileged operation.
1402 Contributions from: Andreas Henriksson, Benjamin Steinwender,
1403 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
1404 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
1405 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
1406 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
1407 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
1408 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
1409 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
1410 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
1411 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
1412 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
1413 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
1415 -- Berlin, 2014-07-03
1419 * As an experimental feature, udev now tries to lock the
1420 disk device node (flock(LOCK_SH|LOCK_NB)) while it
1421 executes events for the disk or any of its partitions.
1422 Applications like partitioning programs can lock the
1423 disk device node (flock(LOCK_EX)) and claim temporary
1424 device ownership that way; udev will entirely skip all event
1425 handling for this disk and its partitions. If the disk
1426 was opened for writing, the close will trigger a partition
1427 table rescan in udev's "watch" facility, and if needed
1428 synthesize "change" events for the disk and all its partitions.
1429 This is now unconditionally enabled, and if it turns out to
1430 cause major problems, we might turn it on only for specific
1431 devices, or might need to disable it entirely. Device Mapper
1432 devices are excluded from this logic.
1434 * We temporarily dropped the "-l" switch for fsck invocations,
1435 since they collide with the flock() logic above. util-linux
1436 upstream has been changed already to avoid this conflict,
1437 and we will readd "-l" as soon as util-linux with this
1438 change has been released.
1440 * The dependency on libattr has been removed. Since a long
1441 time, the extended attribute calls have moved to glibc, and
1442 libattr is thus unnecessary.
1444 * Virtualization detection works without priviliges now. This
1445 means the systemd-detect-virt binary no longer requires
1446 CAP_SYS_PTRACE file capabilities, and our daemons can run
1447 with fewer privileges.
1449 * systemd-networkd now runs under its own "systemd-network"
1450 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
1451 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
1452 loses the ability to write to files owned by root this way.
1454 * Similar, systemd-resolved now runs under its own
1455 "systemd-resolve" user with no capabilities remaining.
1457 * Similar, systemd-bus-proxyd now runs under its own
1458 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
1460 * systemd-networkd gained support for setting up "veth"
1461 virtual ethernet devices for container connectivity, as well
1462 as GRE and VTI tunnels.
1464 * systemd-networkd will no longer automatically attempt to
1465 manually load kernel modules necessary for certain tunnel
1466 transports. Instead, it is assumed the kernel loads them
1467 automatically when required. This only works correctly on
1468 very new kernels. On older kernels, please consider adding
1469 the kernel modules to /etc/modules-load.d/ as a work-around.
1471 * The resolv.conf file systemd-resolved generates has been
1472 moved to /run/systemd/resolve/. If you have a symlink from
1473 /etc/resolv.conf, it might be necessary to correct it.
1475 * Two new service settings, ProtectHome= and ProtectSystem=,
1476 have been added. When enabled, they will make the user data
1477 (such as /home) inaccessible or read-only and the system
1478 (such as /usr) read-only, for specific services. This allows
1479 very light-weight per-service sandboxing to avoid
1480 modifications of user data or system files from
1481 services. These two new switches have been enabled for all
1482 of systemd's long-running services, where appropriate.
1484 * Socket units gained new SocketUser= and SocketGroup=
1485 settings to set the owner user and group of AF_UNIX sockets
1486 and FIFOs in the file system.
1488 * Socket units gained a new RemoveOnStop= setting. If enabled,
1489 all FIFOS and sockets in the file system will be removed
1490 when the specific socket unit is stopped.
1492 * Socket units gained a new Symlinks= setting. It takes a list
1493 of symlinks to create to file system sockets or FIFOs
1494 created by the specific Unix sockets. This is useful to
1495 manage symlinks to socket nodes with the same life-cycle as
1498 * The /dev/log socket and /dev/initctl FIFO have been moved to
1499 /run, and have been replaced by symlinks. This allows
1500 connecting to these facilities even if PrivateDevices=yes is
1501 used for a service (which makes /dev/log itself unavailable,
1502 but /run is left). This also has the benefit of ensuring
1503 that /dev only contains device nodes, directories and
1504 symlinks, and nothing else.
1506 * sd-daemon gained two new calls sd_pid_notify() and
1507 sd_pid_notifyf(). They are similar to sd_notify() and
1508 sd_notifyf(), but allow overriding of the source PID of
1509 notification messages if permissions permit this. This is
1510 useful to send notify messages on behalf of a different
1511 process (for example, the parent process). The
1512 systemd-notify tool has been updated to make use of this
1513 when sending messages (so that notification messages now
1514 originate from the shell script invoking systemd-notify and
1515 not the systemd-notify process itself. This should minimize
1516 a race where systemd fails to associate notification
1517 messages to services when the originating process already
1520 * A new "on-abnormal" setting for Restart= has been added. If
1521 set, it will result in automatic restarts on all "abnormal"
1522 reasons for a process to exit, which includes unclean
1523 signals, core dumps, timeouts and watchdog timeouts, but
1524 does not include clean and unclean exit codes or clean
1525 signals. Restart=on-abnormal is an alternative for
1526 Restart=on-failure for services that shall be able to
1527 terminate and avoid restarts on certain errors, by
1528 indicating so with an unclean exit code. Restart=on-failure
1529 or Restart=on-abnormal is now the recommended setting for
1530 all long-running services.
1532 * If the InaccessibleDirectories= service setting points to a
1533 mount point (or if there are any submounts contained within
1534 it), it is now attempted to completely unmount it, to make
1535 the file systems truly unavailable for the respective
1538 * The ReadOnlyDirectories= service setting and
1539 systemd-nspawn's --read-only parameter are now recursively
1540 applied to all submounts, too.
1542 * Mount units may now be created transiently via the bus APIs.
1544 * The support for SysV and LSB init scripts has been removed
1545 from the systemd daemon itself. Instead, it is now
1546 implemented as a generator that creates native systemd units
1547 from these scripts when needed. This enables us to remove a
1548 substantial amount of legacy code from PID 1, following the
1549 fact that many distributions only ship a very small number
1550 of LSB/SysV init scripts nowadays.
1552 * Privileged Xen (dom0) domains are not considered
1553 virtualization anymore by the virtualization detection
1554 logic. After all, they generally have unrestricted access to
1555 the hardware and usually are used to manage the unprivileged
1558 * systemd-tmpfiles gained a new "C" line type, for copying
1559 files or entire directories.
1561 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
1562 lines. So far, they have been non-globbing versions of the
1563 latter, and have thus been redundant. In future, it is
1564 recommended to only use "z". "m" has hence been removed
1565 from the documentation, even though it stays supported.
1567 * A tmpfiles snippet to recreate the most basic structure in
1568 /var has been added. This is enough to create the /var/run →
1569 /run symlink and create a couple of structural
1570 directories. This allows systems to boot up with an empty or
1571 volatile /var. Of course, while with this change, the core OS
1572 now is capable with dealing with a volatile /var, not all
1573 user services are ready for it. However, we hope that sooner
1574 or later, many service daemons will be changed upstream so
1575 that they are able to automatically create their necessary
1576 directories in /var at boot, should they be missing. This is
1577 the first step to allow state-less systems that only require
1578 the vendor image for /usr to boot.
1580 * systemd-nspawn has gained a new --tmpfs= switch to mount an
1581 empty tmpfs instance to a specific directory. This is
1582 particularly useful for making use of the automatic
1583 reconstruction of /var (see above), by passing --tmpfs=/var.
1585 * Access modes specified in tmpfiles snippets may now be
1586 prefixed with "~", which indicates that they shall be masked
1587 by whether the existing file or directory is currently
1588 writable, readable or executable at all. Also, if specified,
1589 the sgid/suid/sticky bits will be masked for all
1592 * A new passive target unit "network-pre.target" has been
1593 added which is useful for services that shall run before any
1594 network is configured, for example firewall scripts.
1596 * The "floppy" group that previously owned the /dev/fd*
1597 devices is no longer used. The "disk" group is now used
1598 instead. Distributions should probably deprecate usage of
1601 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
1602 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
1603 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
1604 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
1605 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
1606 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
1609 -- Berlin, 2014-06-11
1613 * A new "systemd-timesyncd" daemon has been added for
1614 synchronizing the system clock across the network. It
1615 implements an SNTP client. In contrast to NTP
1616 implementations such as chrony or the NTP reference server,
1617 this only implements a client side, and does not bother with
1618 the full NTP complexity, focusing only on querying time from
1619 one remote server and synchronizing the local clock to
1620 it. Unless you intend to serve NTP to networked clients or
1621 want to connect to local hardware clocks, this simple NTP
1622 client should be more than appropriate for most
1623 installations. The daemon runs with minimal privileges, and
1624 has been hooked up with networkd to only operate when
1625 network connectivity is available. The daemon saves the
1626 current clock to disk every time a new NTP sync has been
1627 acquired, and uses this to possibly correct the system clock
1628 early at bootup, in order to accommodate for systems that
1629 lack an RTC such as the Raspberry Pi and embedded devices,
1630 and to make sure that time monotonically progresses on these
1631 systems, even if it is not always correct. To make use of
1632 this daemon, a new system user and group "systemd-timesync"
1633 needs to be created on installation of systemd.
1635 * The queue "seqnum" interface of libudev has been disabled, as
1636 it was generally incompatible with device namespacing as
1637 sequence numbers of devices go "missing" if the devices are
1638 part of a different namespace.
1640 * "systemctl list-timers" and "systemctl list-sockets" gained
1641 a --recursive switch for showing units of these types also
1642 for all local containers, similar in style to the already
1643 supported --recursive switch for "systemctl list-units".
1645 * A new RebootArgument= setting has been added for service
1646 units, which may be used to specify a kernel reboot argument
1647 to use when triggering reboots with StartLimitAction=.
1649 * A new FailureAction= setting has been added for service
1650 units which may be used to specify an operation to trigger
1651 when a service fails. This works similarly to
1652 StartLimitAction=, but unlike it, controls what is done
1653 immediately rather than only after several attempts to
1654 restart the service in question.
1656 * hostnamed got updated to also expose the kernel name,
1657 release, and version on the bus. This is useful for
1658 executing commands like hostnamectl with the -H switch.
1659 systemd-analyze makes use of this to properly display
1660 details when running non-locally.
1662 * The bootchart tool can now show cgroup information in the
1663 graphs it generates.
1665 * The CFS CPU quota cgroup attribute is now exposed for
1666 services. The new CPUQuota= switch has been added for this
1667 which takes a percentage value. Setting this will have the
1668 result that a service may never get more CPU time than the
1669 specified percentage, even if the machine is otherwise idle.
1671 * systemd-networkd learned IPIP and SIT tunnel support.
1673 * LSB init scripts exposing a dependency on $network will now
1674 get a dependency on network-online.target rather than simply
1675 network.target. This should bring LSB handling closer to
1676 what it was on SysV systems.
1678 * A new fsck.repair= kernel option has been added to control
1679 how fsck shall deal with unclean file systems at boot.
1681 * The (.ini) configuration file parser will now silently
1682 ignore sections whose name begins with "X-". This may be
1683 used to maintain application-specific extension sections in unit
1686 * machined gained a new API to query the IP addresses of
1687 registered containers. "machinectl status" has been updated
1688 to show these addresses in its output.
1690 * A new call sd_uid_get_display() has been added to the
1691 sd-login APIs for querying the "primary" session of a
1692 user. The "primary" session of the user is elected from the
1693 user's sessions and generally a graphical session is
1694 preferred over a text one.
1696 * A minimal systemd-resolved daemon has been added. It
1697 currently simply acts as a companion to systemd-networkd and
1698 manages resolv.conf based on per-interface DNS
1699 configuration, possibly supplied via DHCP. In the long run
1700 we hope to extend this into a local DNSSEC enabled DNS and
1703 * The systemd-networkd-wait-online tool is now enabled by
1704 default. It will delay network-online.target until a network
1705 connection has been configured. The tool primarily integrates
1706 with networkd, but will also make a best effort to make sense
1707 of network configuration performed in some other way.
1709 * Two new service options StartupCPUShares= and
1710 StartupBlockIOWeight= have been added that work similarly to
1711 CPUShares= and BlockIOWeight= however only apply during
1712 system startup. This is useful to prioritize certain services
1713 differently during bootup than during normal runtime.
1715 * hostnamed has been changed to prefer the statically
1716 configured hostname in /etc/hostname (unless set to
1717 'localhost' or empty) over any dynamic one supplied by
1718 dhcp. With this change, the rules for picking the hostname
1719 match more closely the rules of other configuration settings
1720 where the local administrator's configuration in /etc always
1721 overrides any other settings.
1723 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
1724 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
1725 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
1726 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
1727 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
1728 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
1729 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
1730 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
1731 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
1732 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
1733 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
1734 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
1735 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
1736 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
1737 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
1738 Lindskog, WaLyong Cho, Will Woods, Zbigniew
1741 -- Beijing, 2014-05-28
1745 * When restoring the screen brightness at boot, stay away from
1746 the darkest setting or from the lowest 5% of the available
1747 range, depending on which is the larger value of both. This
1748 should effectively protect the user from rebooting into a
1749 black screen, should the brightness have been set to minimum
1752 * sd-login gained a new sd_machine_get_class() call to
1753 determine the class ("vm" or "container") of a machine
1754 registered with machined.
1756 * sd-login gained new calls
1757 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
1758 to query the identity of the peer of a local AF_UNIX
1759 connection. They operate similarly to their sd_pid_get_xyz()
1762 * PID 1 will now maintain a system-wide system state engine
1763 with the states "starting", "running", "degraded",
1764 "maintenance", "stopping". These states are bound to system
1765 startup, normal runtime, runtime with at least one failed
1766 service, rescue/emergency mode and system shutdown. This
1767 state is shown in the "systemctl status" output when no unit
1768 name is passed. It is useful to determine system state, in
1769 particularly when doing so for many systems or containers at
1772 * A new command "list-machines" has been added to "systemctl"
1773 that lists all local OS containers and shows their system
1774 state (see above), if systemd runs inside of them.
1776 * systemctl gained a new "-r" switch to recursively enumerate
1777 units on all local containers, when used with the
1778 "list-unit" command (which is the default one that is
1779 executed when no parameters are specified).
1781 * The GPT automatic partition discovery logic will now honour
1782 two GPT partition flags: one may be set on a partition to
1783 cause it to be mounted read-only, and the other may be set
1784 on a partition to ignore it during automatic discovery.
1786 * Two new GPT type UUIDs have been added for automatic root
1787 partition discovery, for 32-bit and 64-bit ARM. This is not
1788 particularly useful for discovering the root directory on
1789 these architectures during bare-metal boots (since UEFI is
1790 not common there), but still very useful to allow booting of
1791 ARM disk images in nspawn with the -i option.
1793 * MAC addresses of interfaces created with nspawn's
1794 --network-interface= switch will now be generated from the
1795 machine name, and thus be stable between multiple invocations
1798 * logind will now automatically remove all IPC objects owned
1799 by a user if she or he fully logs out. This makes sure that
1800 users who are logged out cannot continue to consume IPC
1801 resources. This covers SysV memory, semaphores and message
1802 queues as well as POSIX shared memory and message
1803 queues. Traditionally, SysV and POSIX IPC had no life-cycle
1804 limits. With this functionality, that is corrected. This may
1805 be turned off by using the RemoveIPC= switch of logind.conf.
1807 * The systemd-machine-id-setup and tmpfiles tools gained a
1808 --root= switch to operate on a specific root directory,
1811 * journald can now forward logged messages to the TTYs of all
1812 logged in users ("wall"). This is the default for all
1813 emergency messages now.
1815 * A new tool systemd-journal-remote has been added to stream
1816 journal log messages across the network.
1818 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
1819 controller trees are mounted into it. Note that the
1820 directories mounted beneath it are not read-only. This is a
1821 security measure and is particularly useful because glibc
1822 actually includes a search logic to pick any tmpfs it can
1823 find to implement shm_open() if /dev/shm is not available
1824 (which it might very well be in namespaced setups).
1826 * machinectl gained a new "poweroff" command to cleanly power
1827 down a local OS container.
1829 * The PrivateDevices= unit file setting will now also drop the
1830 CAP_MKNOD capability from the capability bound set, and
1831 imply DevicePolicy=closed.
1833 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
1834 comprehensively on all long-running systemd services where
1835 this is appropriate.
1837 * systemd-udevd will now run in a disassociated mount
1838 namespace. To mount directories from udev rules, make sure to
1839 pull in mount units via SYSTEMD_WANTS properties.
1841 * The kdbus support gained support for uploading policy into
1842 the kernel. sd-bus gained support for creating "monitoring"
1843 connections that can eavesdrop into all bus communication
1844 for debugging purposes.
1846 * Timestamps may now be specified in seconds since the UNIX
1847 epoch Jan 1st, 1970 by specifying "@" followed by the value
1850 * Native tcpwrap support in systemd has been removed. tcpwrap
1851 is old code, not really maintained anymore and has serious
1852 shortcomings, and better options such as firewalls
1853 exist. For setups that require tcpwrap usage, please
1854 consider invoking your socket-activated service via tcpd,
1855 like on traditional inetd.
1857 * A new system.conf configuration option
1858 DefaultTimerAccuracySec= has been added that controls the
1859 default AccuracySec= setting of .timer units.
1861 * Timer units gained a new WakeSystem= switch. If enabled,
1862 timers configured this way will cause the system to resume
1863 from system suspend (if the system supports that, which most
1866 * Timer units gained a new Persistent= switch. If enabled,
1867 timers configured this way will save to disk when they have
1868 been last triggered. This information is then used on next
1869 reboot to possible execute overdue timer events, that
1870 could not take place because the system was powered off.
1871 This enables simple anacron-like behaviour for timer units.
1873 * systemctl's "list-timers" will now also list the time a
1874 timer unit was last triggered in addition to the next time
1875 it will be triggered.
1877 * systemd-networkd will now assign predictable IPv4LL
1878 addresses to its local interfaces.
1880 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
1881 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
1882 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
1883 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
1884 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
1885 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
1886 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
1887 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
1890 -- Berlin, 2014-03-25
1894 * A new unit file setting RestrictAddressFamilies= has been
1895 added to restrict which socket address families unit
1896 processes gain access to. This takes address family names
1897 like "AF_INET" or "AF_UNIX", and is useful to minimize the
1898 attack surface of services via exotic protocol stacks. This
1899 is built on seccomp system call filters.
1901 * Two new unit file settings RuntimeDirectory= and
1902 RuntimeDirectoryMode= have been added that may be used to
1903 manage a per-daemon runtime directories below /run. This is
1904 an alternative for setting up directory permissions with
1905 tmpfiles snippets, and has the advantage that the runtime
1906 directory's lifetime is bound to the daemon runtime and that
1907 the daemon starts up with an empty directory each time. This
1908 is particularly useful when writing services that drop
1909 privileges using the User= or Group= setting.
1911 * The DeviceAllow= unit setting now supports globbing for
1912 matching against device group names.
1914 * The systemd configuration file system.conf gained new
1915 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
1916 DefaultMemoryAccounting= to globally turn on/off accounting
1917 for specific resources (cgroups) for all units. These
1918 settings may still be overridden individually in each unit
1921 * systemd-gpt-auto-generator is now able to discover /srv and
1922 root partitions in addition to /home and swap partitions. It
1923 also supports LUKS-encrypted partitions now. With this in
1924 place, automatic discovery of partitions to mount following
1925 the Discoverable Partitions Specification
1926 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
1927 is now a lot more complete. This allows booting without
1928 /etc/fstab and without root= on the kernel command line on
1929 systems prepared appropriately.
1931 * systemd-nspawn gained a new --image= switch which allows
1932 booting up disk images and Linux installations on any block
1933 device that follow the Discoverable Partitions Specification
1934 (see above). This means that installations made with
1935 appropriately updated installers may now be started and
1936 deployed using container managers, completely
1937 unmodified. (We hope that libvirt-lxc will add support for
1938 this feature soon, too.)
1940 * systemd-nspawn gained a new --network-macvlan= setting to
1941 set up a private macvlan interface for the
1942 container. Similarly, systemd-networkd gained a new
1943 Kind=macvlan setting in .netdev files.
1945 * systemd-networkd now supports configuring local addresses
1948 * A new tool systemd-network-wait-online has been added to
1949 synchronously wait for network connectivity using
1952 * The sd-bus.h bus API gained a new sd_bus_track object for
1953 tracking the life-cycle of bus peers. Note that sd-bus.h is
1954 still not a public API though (unless you specify
1955 --enable-kdbus on the configure command line, which however
1956 voids your warranty and you get no API stability guarantee).
1958 * The $XDG_RUNTIME_DIR runtime directories for each user are
1959 now individual tmpfs instances, which has the benefit of
1960 introducing separate pools for each user, with individual
1961 size limits, and thus making sure that unprivileged clients
1962 can no longer negatively impact the system or other users by
1963 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
1964 RuntimeDirectorySize= has been introduced that allows
1965 controlling the default size limit for all users. It
1966 defaults to 10% of the available physical memory. This is no
1967 replacement for quotas on tmpfs though (which the kernel
1968 still does not support), as /dev/shm and /tmp are still
1969 shared resources used by both the system and unprivileged
1972 * logind will now automatically turn off automatic suspending
1973 on laptop lid close when more than one display is
1974 connected. This was previously expected to be implemented
1975 individually in desktop environments (such as GNOME),
1976 however has been added to logind now, in order to fix a
1977 boot-time race where a desktop environment might not have
1978 been started yet and thus not been able to take an inhibitor
1979 lock at the time where logind already suspends the system
1980 due to a closed lid.
1982 * logind will now wait at least 30s after each system
1983 suspend/resume cycle, and 3min after system boot before
1984 suspending the system due to a closed laptop lid. This
1985 should give USB docking stations and similar enough time to
1986 be probed and configured after system resume and boot in
1987 order to then act as suspend blocker.
1989 * systemd-run gained a new --property= setting which allows
1990 initialization of resource control properties (and others)
1991 for the created scope or service unit. Example: "systemd-run
1992 --property=BlockIOWeight=10 updatedb" may be used to run
1993 updatedb at a low block IO scheduling weight.
1995 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
1996 now also work in --scope mode.
1998 * When systemd is compiled with kdbus support, basic support
1999 for enforced policies is now in place. (Note that enabling
2000 kdbus still voids your warranty and no API compatibility
2003 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
2004 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
2005 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
2006 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
2007 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
2008 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
2009 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
2010 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
2011 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
2012 Zbigniew Jędrzejewski-Szmek
2014 -- Berlin, 2014-03-12
2018 * systemd will now relabel /dev after loading the SMACK policy
2019 according to SMACK rules.
2021 * A new unit file option AppArmorProfile= has been added to
2022 set the AppArmor profile for the processes of a unit.
2024 * A new condition check ConditionArchitecture= has been added
2025 to conditionalize units based on the system architecture, as
2026 reported by uname()'s "machine" field.
2028 * systemd-networkd now supports matching on the system
2029 virtualization, architecture, kernel command line, host name
2032 * logind is now a lot more aggressive when suspending the
2033 machine due to a closed laptop lid. Instead of acting only
2034 on the lid close action, it will continuously watch the lid
2035 status and act on it. This is useful for laptops where the
2036 power button is on the outside of the chassis so that it can
2037 be reached without opening the lid (such as the Lenovo
2038 Yoga). On those machines, logind will now immediately
2039 re-suspend the machine if the power button has been
2040 accidentally pressed while the laptop was suspended and in a
2041 backpack or similar.
2043 * logind will now watch SW_DOCK switches and inhibit reaction
2044 to the lid switch if it is pressed. This means that logind
2045 will not suspend the machine anymore if the lid is closed
2046 and the system is docked, if the laptop supports SW_DOCK
2047 notifications via the input layer. Note that ACPI docking
2048 stations do not generate this currently. Also note that this
2049 logic is usually not fully sufficient and Desktop
2050 Environments should take a lid switch inhibitor lock when an
2051 external display is connected, as systemd will not watch
2054 * nspawn will now make use of the devices cgroup controller by
2055 default, and only permit creation of and access to the usual
2056 API device nodes like /dev/null or /dev/random, as well as
2057 access to (but not creation of) the pty devices.
2059 * We will now ship a default .network file for
2060 systemd-networkd that automatically configures DHCP for
2061 network interfaces created by nspawn's --network-veth or
2062 --network-bridge= switches.
2064 * systemd will now understand the usual M, K, G, T suffixes
2065 according to SI conventions (i.e. to the base 1000) when
2066 referring to throughput and hardware metrics. It will stay
2067 with IEC conventions (i.e. to the base 1024) for software
2068 metrics, according to what is customary according to
2069 Wikipedia. We explicitly document which base applies for
2070 each configuration option.
2072 * The DeviceAllow= setting in unit files now supports a syntax
2073 to whitelist an entire group of devices node majors at once,
2074 based on the /proc/devices listing. For example, with the
2075 string "char-pts", it is now possible to whitelist all
2076 current and future pseudo-TTYs at once.
2078 * sd-event learned a new "post" event source. Event sources of
2079 this type are triggered by the dispatching of any event
2080 source of a type that is not "post". This is useful for
2081 implementing clean-up and check event sources that are
2082 triggered by other work being done in the program.
2084 * systemd-networkd is no longer statically enabled, but uses
2085 the usual [Install] sections so that it can be
2086 enabled/disabled using systemctl. It still is enabled by
2089 * When creating a veth interface pair with systemd-nspawn, the
2090 host side will now be prefixed with "vb-" if
2091 --network-bridge= is used, and with "ve-" if --network-veth
2092 is used. This way, it is easy to distinguish these cases on
2093 the host, for example to apply different configuration to
2094 them with systemd-networkd.
2096 * The compatibility libraries for libsystemd-journal.so,
2097 libsystem-id128.so, libsystemd-login.so and
2098 libsystemd-daemon.so do not make use of IFUNC
2099 anymore. Instead, we now build libsystemd.so multiple times
2100 under these alternative names. This means that the footprint
2101 is drastically increased, but given that these are
2102 transitional compatibility libraries, this should not matter
2103 much. This change has been made necessary to support the ARM
2104 platform for these compatibility libraries, as the ARM
2105 toolchain is not really at the same level as the toolchain
2106 for other architectures like x86 and does not support
2107 IFUNC. Please make sure to use --enable-compat-libs only
2108 during a transitional period!
2110 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
2111 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
2112 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
2113 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
2114 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
2115 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
2116 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
2117 Zbigniew Jędrzejewski-Szmek
2119 -- Berlin, 2014-02-24
2123 * A new component "systemd-networkd" has been added that can
2124 be used to configure local network interfaces statically or
2125 via DHCP. It is capable of bringing up bridges, VLANs, and
2126 bonding. Currently, no hook-ups for interactive network
2127 configuration are provided. Use this for your initrd,
2128 container, embedded, or server setup if you need a simple,
2129 yet powerful, network configuration solution. This
2130 configuration subsystem is quite nifty, as it allows wildcard
2131 hotplug matching in interfaces. For example, with a single
2132 configuration snippet, you can configure that all Ethernet
2133 interfaces showing up are automatically added to a bridge,
2134 or similar. It supports link-sensing and more.
2136 * A new tool "systemd-socket-proxyd" has been added which can
2137 act as a bidirectional proxy for TCP sockets. This is
2138 useful for adding socket activation support to services that
2139 do not actually support socket activation, including virtual
2140 machines and the like.
2142 * Add a new tool to save/restore rfkill state on
2145 * Save/restore state of keyboard backlights in addition to
2146 display backlights on shutdown/boot.
2148 * udev learned a new SECLABEL{} construct to label device
2149 nodes with a specific security label when they appear. For
2150 now, only SECLABEL{selinux} is supported, but the syntax is
2151 prepared for additional security frameworks.
2153 * udev gained a new scheme to configure link-level attributes
2154 from files in /etc/systemd/network/*.link. These files can
2155 match against MAC address, device path, driver name and type,
2156 and will apply attributes like the naming policy, link speed,
2157 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
2158 address assignment policy (randomized, ...).
2160 * The configuration of network interface naming rules for
2161 "permanent interface names" has changed: a new NamePolicy=
2162 setting in the [Link] section of .link files determines the
2163 priority of possible naming schemes (onboard, slot, mac,
2164 path). The default value of this setting is determined by
2165 /usr/lib/net/links/99-default.link. Old
2166 80-net-name-slot.rules udev configuration file has been
2167 removed, so local configuration overriding this file should
2168 be adapated to override 99-default.link instead.
2170 * When the User= switch is used in a unit file, also
2171 initialize $SHELL= based on the user database entry.
2173 * systemd no longer depends on libdbus. All communication is
2174 now done with sd-bus, systemd's low-level bus library
2177 * kdbus support has been added to PID 1 itself. When kdbus is
2178 enabled, this causes PID 1 to set up the system bus and
2179 enable support for a new ".busname" unit type that
2180 encapsulates bus name activation on kdbus. It works a little
2181 bit like ".socket" units, except for bus names. A new
2182 generator has been added that converts classic dbus1 service
2183 activation files automatically into native systemd .busname
2186 * sd-bus: add a light-weight vtable implementation that allows
2187 defining objects on the bus with a simple static const
2188 vtable array of its methods, signals and properties.
2190 * systemd will not generate or install static dbus
2191 introspection data anymore to /usr/share/dbus-1/interfaces,
2192 as the precise format of these files is unclear, and
2193 nothing makes use of it.
2195 * A proxy daemon is now provided to proxy clients connecting
2196 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
2197 compatibility with classic D-Bus.
2199 * A bus driver implementation has been added that supports the
2200 classic D-Bus bus driver calls on kdbus, also for
2201 compatibility purposes.
2203 * A new API "sd-event.h" has been added that implements a
2204 minimal event loop API built around epoll. It provides a
2205 couple of features that direct epoll usage is lacking:
2206 prioritization of events, scales to large numbers of timer
2207 events, per-event timer slack (accuracy), system-wide
2208 coalescing of timer events, exit handlers, watchdog
2209 supervision support using systemd's sd_notify() API, child
2212 * A new API "sd-rntl.h" has been added that provides an API
2213 around the route netlink interface of the kernel, similar in
2214 style to "sd-bus.h".
2216 * A new API "sd-dhcp-client.h" has been added that provides a
2217 small DHCPv4 client-side implementation. This is used by
2220 * There is a new kernel command line option
2221 "systemd.restore_state=0|1". When set to "0", none of the
2222 systemd tools will restore saved runtime state to hardware
2223 devices. More specifically, the rfkill and backlight states
2226 * The FsckPassNo= compatibility option in mount/service units
2227 has been removed. The fstab generator will now add the
2228 necessary dependencies automatically, and does not require
2229 PID1's support for that anymore.
2231 * journalctl gained a new switch, --list-boots, that lists
2232 recent boots with their times and boot IDs.
2234 * The various tools like systemctl, loginctl, timedatectl,
2235 busctl, systemd-run, ... have gained a new switch "-M" to
2236 connect to a specific, local OS container (as direct
2237 connection, without requiring SSH). This works on any
2238 container that is registered with machined, such as those
2239 created by libvirt-lxc or nspawn.
2241 * systemd-run and systemd-analyze also gained support for "-H"
2242 to connect to remote hosts via SSH. This is particularly
2243 useful for systemd-run because it enables queuing of jobs
2244 onto remote systems.
2246 * machinectl gained a new command "login" to open a getty
2247 login in any local container. This works with any container
2248 that is registered with machined (such as those created by
2249 libvirt-lxc or nspawn), and which runs systemd inside.
2251 * machinectl gained a new "reboot" command that may be used to
2252 trigger a reboot on a specific container that is registered
2253 with machined. This works on any container that runs an init
2254 system of some kind.
2256 * systemctl gained a new "list-timers" command to print a nice
2257 listing of installed timer units with the times they elapse
2260 * Alternative reboot() parameters may now be specified on the
2261 "systemctl reboot" command line and are passed to the
2262 reboot() system call.
2264 * systemctl gained a new --job-mode= switch to configure the
2265 mode to queue a job with. This is a more generic version of
2266 --fail, --irreversible, and --ignore-dependencies, which are
2267 still available but not advertised anymore.
2269 * /etc/systemd/system.conf gained new settings to configure
2270 various default timeouts of units, as well as the default
2271 start limit interval and burst. These may still be overridden
2274 * PID1 will now export on the bus profile data of the security
2275 policy upload process (such as the SELinux policy upload to
2278 * journald: when forwarding logs to the console, include
2279 timestamps (following the setting in
2280 /sys/module/printk/parameters/time).
2282 * OnCalendar= in timer units now understands the special
2283 strings "yearly" and "annually". (Both are equivalent)
2285 * The accuracy of timer units is now configurable with the new
2286 AccuracySec= setting. It defaults to 1min.
2288 * A new dependency type JoinsNamespaceOf= has been added that
2289 allows running two services within the same /tmp and network
2290 namespace, if PrivateNetwork= or PrivateTmp= are used.
2292 * A new command "cat" has been added to systemctl. It outputs
2293 the original unit file of a unit, and concatenates the
2294 contents of additional "drop-in" unit file snippets, so that
2295 the full configuration is shown.
2297 * systemctl now supports globbing on the various "list-xyz"
2298 commands, like "list-units" or "list-sockets", as well as on
2299 those commands which take multiple unit names.
2301 * journalctl's --unit= switch gained support for globbing.
2303 * All systemd daemons now make use of the watchdog logic so
2304 that systemd automatically notices when they hang.
2306 * If the $container_ttys environment variable is set,
2307 getty-generator will automatically spawn a getty for each
2308 listed tty. This is useful for container managers to request
2309 login gettys to be spawned on as many ttys as needed.
2311 * %h, %s, %U specifier support is not available anymore when
2312 used in unit files for PID 1. This is because NSS calls are
2313 not safe from PID 1. They stay available for --user
2314 instances of systemd, and as special case for the root user.
2316 * loginctl gained a new "--no-legend" switch to turn off output
2319 * The "sd-login.h" API gained three new calls:
2320 sd_session_is_remote(), sd_session_get_remote_user(),
2321 sd_session_get_remote_host() to query information about
2324 * The udev hardware database now also carries vendor/product
2325 information of SDIO devices.
2327 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
2328 determine whether watchdog notifications are requested by
2331 * Socket-activated per-connection services now include a
2332 short description of the connection parameters in the
2335 * tmpfiles gained a new "--boot" option. When this is not used,
2336 only lines where the command character is not suffixed with
2337 "!" are executed. When this option is specified, those
2338 options are executed too. This partitions tmpfiles
2339 directives into those that can be safely executed at any
2340 time, and those which should be run only at boot (for
2341 example, a line that creates /run/nologin).
2343 * A new API "sd-resolve.h" has been added which provides a simple
2344 asynchronous wrapper around glibc NSS host name resolution
2345 calls, such as getaddrinfo(). In contrast to glibc's
2346 getaddrinfo_a(), it does not use signals. In contrast to most
2347 other asynchronous name resolution libraries, this one does
2348 not reimplement DNS, but reuses NSS, so that alternate
2349 host name resolution systems continue to work, such as mDNS,
2350 LDAP, etc. This API is based on libasyncns, but it has been
2351 cleaned up for inclusion in systemd.
2353 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
2354 "sd-daemon.h" are no longer found in individual libraries
2355 libsystemd-journal.so, libsystemd-login.so,
2356 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
2357 merged them into a single library, libsystemd.so, which
2358 provides all symbols. The reason for this is cyclic
2359 dependencies, as these libraries tend to use each other's
2360 symbols. So far, we have managed to workaround that by linking
2361 a copy of a good part of our code into each of these
2362 libraries again and again, which, however, makes certain
2363 things hard to do, like sharing static variables. Also, it
2364 substantially increases footprint. With this change, there
2365 is only one library for the basic APIs systemd
2366 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
2367 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
2368 library as well, however are subject to the --enable-kdbus
2369 switch (see below). Note that "sd-dhcp-client.h" is not part
2370 of this library (this is because it only consumes, never
2371 provides, services of/to other APIs). To make the transition
2372 easy from the separate libraries to the unified one, we
2373 provide the --enable-compat-libs compile-time switch which
2374 will generate stub libraries that are compatible with the
2375 old ones but redirect all calls to the new one.
2377 * All of the kdbus logic and the new APIs "sd-bus.h",
2378 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
2379 and "sd-utf8.h" are compile-time optional via the
2380 "--enable-kdbus" switch, and they are not compiled in by
2381 default. To make use of kdbus, you have to explicitly enable
2382 the switch. Note however, that neither the kernel nor the
2383 userspace API for all of this is considered stable yet. We
2384 want to maintain the freedom to still change the APIs for
2385 now. By specifying this build-time switch, you acknowledge
2386 that you are aware of the instability of the current
2389 * Also, note that while kdbus is pretty much complete,
2390 it lacks one thing: proper policy support. This means you
2391 can build a fully working system with all features; however,
2392 it will be highly insecure. Policy support will be added in
2393 one of the next releases, at the same time that we will
2394 declare the APIs stable.
2396 * When the kernel command line argument "kdbus" is specified,
2397 systemd will automatically load the kdbus.ko kernel module. At
2398 this stage of development, it is only useful for testing kdbus
2399 and should not be used in production. Note: if "--enable-kdbus"
2400 is specified, and the kdbus.ko kernel module is available, and
2401 "kdbus" is added to the kernel command line, the entire system
2402 runs with kdbus instead of dbus-daemon, with the above mentioned
2403 problem of missing the system policy enforcement. Also a future
2404 version of kdbus.ko or a newer systemd will not be compatible with
2405 each other, and will unlikely be able to boot the machine if only
2406 one of them is updated.
2408 * systemctl gained a new "import-environment" command which
2409 uploads the caller's environment (or parts thereof) into the
2410 service manager so that it is inherited by services started
2411 by the manager. This is useful to upload variables like
2412 $DISPLAY into the user service manager.
2414 * A new PrivateDevices= switch has been added to service units
2415 which allows running a service with a namespaced /dev
2416 directory that does not contain any device nodes for
2417 physical devices. More specifically, it only includes devices
2418 such as /dev/null, /dev/urandom, and /dev/zero which are API
2421 * logind has been extended to support behaviour like VT
2422 switching on seats that do not support a VT. This makes
2423 multi-session available on seats that are not the first seat
2424 (seat0), and on systems where kernel support for VTs has
2425 been disabled at compile-time.
2427 * If a process holds a delay lock for system sleep or shutdown
2428 and fails to release it in time, we will now log its
2429 identity. This makes it easier to identify processes that
2430 cause slow suspends or power-offs.
2432 * When parsing /etc/crypttab, support for a new key-slot=
2433 option as supported by Debian is added. It allows indicating
2434 which LUKS slot to use on disk, speeding up key loading.
2436 * The sd_journald_sendv() API call has been checked and
2437 officially declared to be async-signal-safe so that it may
2438 be invoked from signal handlers for logging purposes.
2440 * Boot-time status output is now enabled automatically after a
2441 short timeout if boot does not progress, in order to give
2442 the user an indication what she or he is waiting for.
2444 * The boot-time output has been improved to show how much time
2445 remains until jobs expire.
2447 * The KillMode= switch in service units gained a new possible
2448 value "mixed". If set, and the unit is shut down, then the
2449 initial SIGTERM signal is sent only to the main daemon
2450 process, while the following SIGKILL signal is sent to
2451 all remaining processes of the service.
2453 * When a scope unit is registered, a new property "Controller"
2454 may be set. If set to a valid bus name, systemd will send a
2455 RequestStop() signal to this name when it would like to shut
2456 down the scope. This may be used to hook manager logic into
2457 the shutdown logic of scope units. Also, scope units may now
2458 be put in a special "abandoned" state, in which case the
2459 manager process which created them takes no further
2460 responsibilities for it.
2462 * When reading unit files, systemd will now verify
2463 the access mode of these files, and warn about certain
2464 suspicious combinations. This has been added to make it
2465 easier to track down packaging bugs where unit files are
2466 marked executable or world-writable.
2468 * systemd-nspawn gained a new "--setenv=" switch to set
2469 container-wide environment variables. The similar option in
2470 systemd-activate was renamed from "--environment=" to
2471 "--setenv=" for consistency.
2473 * systemd-nspawn has been updated to create a new kdbus domain
2474 for each container that is invoked, thus allowing each
2475 container to have its own set of system and user buses,
2476 independent of the host.
2478 * systemd-nspawn gained a new --drop-capability= switch to run
2479 the container with less capabilities than the default. Both
2480 --drop-capability= and --capability= now take the special
2481 string "all" for dropping or keeping all capabilities.
2483 * systemd-nspawn gained new switches for executing containers
2484 with specific SELinux labels set.
2486 * systemd-nspawn gained a new --quiet switch to not generate
2487 any additional output but the container's own console
2490 * systemd-nspawn gained a new --share-system switch to run a
2491 container without PID namespacing enabled.
2493 * systemd-nspawn gained a new --register= switch to control
2494 whether the container is registered with systemd-machined or
2495 not. This is useful for containers that do not run full
2496 OS images, but only specific apps.
2498 * systemd-nspawn gained a new --keep-unit which may be used
2499 when invoked as the only program from a service unit, and
2500 results in registration of the unit service itself in
2501 systemd-machined, instead of a newly opened scope unit.
2503 * systemd-nspawn gained a new --network-interface= switch for
2504 moving arbitrary interfaces to the container. The new
2505 --network-veth switch creates a virtual Ethernet connection
2506 between host and container. The new --network-bridge=
2507 switch then allows assigning the host side of this virtual
2508 Ethernet connection to a bridge device.
2510 * systemd-nspawn gained a new --personality= switch for
2511 setting the kernel personality for the container. This is
2512 useful when running a 32-bit container on a 64-bit host. A
2513 similar option Personality= is now also available for service
2516 * logind will now also track a "Desktop" identifier for each
2517 session which encodes the desktop environment of it. This is
2518 useful for desktop environments that want to identify
2519 multiple running sessions of itself easily.
2521 * A new SELinuxContext= setting for service units has been
2522 added that allows setting a specific SELinux execution
2523 context for a service.
2525 * Most systemd client tools will now honour $SYSTEMD_LESS for
2526 settings of the "less" pager. By default, these tools will
2527 override $LESS to allow certain operations to work, such as
2528 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
2529 influence this logic.
2531 * systemd's "seccomp" hook-up has been changed to make use of
2532 the libseccomp library instead of using its own
2533 implementation. This has benefits for portability among
2536 * For usage together with SystemCallFilter=, a new
2537 SystemCallErrorNumber= setting has been introduced that
2538 allows configuration of a system error number to be returned
2539 on filtered system calls, instead of immediately killing the
2540 process. Also, SystemCallArchitectures= has been added to
2541 limit access to system calls of a particular architecture
2542 (in order to turn off support for unused secondary
2543 architectures). There is also a global
2544 SystemCallArchitectures= setting in system.conf now to turn
2545 off support for non-native system calls system-wide.
2547 * systemd requires a kernel with a working name_to_handle_at(),
2548 please see the kernel config requirements in the README file.
2550 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
2551 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
2552 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
2553 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
2554 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
2555 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
2556 Elia Pinto, Florian Weimer, George McCollister, Goffredo
2557 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
2558 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
2559 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
2560 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
2561 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
2562 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
2563 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
2564 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
2565 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
2566 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
2567 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
2568 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
2569 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
2570 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
2571 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
2572 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
2573 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
2575 -- Berlin, 2014-02-20
2579 * logind has gained support for facilitating privileged input
2580 and drm device access for unprivileged clients. This work is
2581 useful to allow Wayland display servers (and similar
2582 programs, such as kmscon) to run under the user's ID and
2583 access input and drm devices which are normally
2584 protected. When this is used (and the kernel is new enough)
2585 logind will "mute" IO on the file descriptors passed to
2586 Wayland as long as it is in the background and "unmute" it
2587 if it returns into the foreground. This allows secure
2588 session switching without allowing background sessions to
2589 eavesdrop on input and display data. This also introduces
2590 session switching support if VT support is turned off in the
2591 kernel, and on seats that are not seat0.
2593 * A new kernel command line option luks.options= is understood
2594 now which allows specifying LUKS options for usage for LUKS
2595 encrypted partitions specified with luks.uuid=.
2597 * tmpfiles.d(5) snippets may now use specifier expansion in
2598 path names. More specifically %m, %b, %H, %v, are now
2599 replaced by the local machine id, boot id, hostname, and
2600 kernel version number.
2602 * A new tmpfiles.d(5) command "m" has been introduced which
2603 may be used to change the owner/group/access mode of a file
2604 or directory if it exists, but do nothing if it does not.
2606 * This release removes high-level support for the
2607 MemorySoftLimit= cgroup setting. The underlying kernel
2608 cgroup attribute memory.soft_limit= is currently badly
2609 designed and likely to be removed from the kernel API in its
2610 current form, hence we should not expose it for now.
2612 * The memory.use_hierarchy cgroup attribute is now enabled for
2613 all cgroups systemd creates in the memory cgroup
2614 hierarchy. This option is likely to be come the built-in
2615 default in the kernel anyway, and the non-hierarchical mode
2616 never made much sense in the intrinsically hierarchical
2619 * A new field _SYSTEMD_SLICE= is logged along with all journal
2620 messages containing the slice a message was generated
2621 from. This is useful to allow easy per-customer filtering of
2622 logs among other things.
2624 * systemd-journald will no longer adjust the group of journal
2625 files it creates to the "systemd-journal" group. Instead we
2626 rely on the journal directory to be owned by the
2627 "systemd-journal" group, and its setgid bit set, so that the
2628 kernel file system layer will automatically enforce that
2629 journal files inherit this group assignment. The reason for
2630 this change is that we cannot allow NSS look-ups from
2631 journald which would be necessary to resolve
2632 "systemd-journal" to a numeric GID, because this might
2633 create deadlocks if NSS involves synchronous queries to
2634 other daemons (such as nscd, or sssd) which in turn are
2635 logging clients of journald and might block on it, which
2636 would then dead lock. A tmpfiles.d(5) snippet included in
2637 systemd will make sure the setgid bit and group are
2638 properly set on the journal directory if it exists on every
2639 boot. However, we recommend adjusting it manually after
2640 upgrades too (or from RPM scriptlets), so that the change is
2641 not delayed until next reboot.
2643 * Backlight and random seed files in /var/lib/ have moved into
2644 the /var/lib/systemd/ directory, in order to centralize all
2645 systemd generated files in one directory.
2647 * Boot time performance measurements (as displayed by
2648 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
2649 performance information if that's available to determine how
2650 much time BIOS and boot loader initialization required. With
2651 a sufficiently new BIOS you hence no longer need to boot
2652 with Gummiboot to get access to such information.
2654 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
2655 Cristian Rodríguez, Dave Reisner, David Herrmann, David
2656 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
2657 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
2658 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
2659 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
2660 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
2662 -- Berlin, 2013-10-02
2666 * The Restart= option for services now understands a new
2667 on-watchdog setting, which will restart the service
2668 automatically if the service stops sending out watchdog keep
2669 alive messages (as configured with WatchdogSec=).
2671 * The getty generator (which is responsible for bringing up a
2672 getty on configured serial consoles) will no longer only
2673 start a getty on the primary kernel console but on all
2674 others, too. This makes the order in which console= is
2675 specified on the kernel command line less important.
2677 * libsystemd-logind gained a new sd_session_get_vt() call to
2678 retrieve the VT number of a session.
2680 * If the option "tries=0" is set for an entry of /etc/crypttab
2681 its passphrase is queried indefinitely instead of any
2682 maximum number of tries.
2684 * If a service with a configure PID file terminates its PID
2685 file will now be removed automatically if it still exists
2686 afterwards. This should put an end to stale PID files.
2688 * systemd-run will now also take relative binary path names
2689 for execution and no longer insists on absolute paths.
2691 * InaccessibleDirectories= and ReadOnlyDirectories= now take
2692 paths that are optionally prefixed with "-" to indicate that
2693 it should not be considered a failure if they do not exist.
2695 * journalctl -o (and similar commands) now understands a new
2696 output mode "short-precise", it is similar to "short" but
2697 shows timestamps with usec accuracy.
2699 * The option "discard" (as known from Debian) is now
2700 synonymous to "allow-discards" in /etc/crypttab. In fact,
2701 "discard" is preferred now (since it is easier to remember
2704 * Some licensing clean-ups were made, so that more code is now
2705 LGPL-2.1 licensed than before.
2707 * A minimal tool to save/restore the display backlight
2708 brightness across reboots has been added. It will store the
2709 backlight setting as late as possible at shutdown, and
2710 restore it as early as possible during reboot.
2712 * A logic to automatically discover and enable home and swap
2713 partitions on GPT disks has been added. With this in place
2714 /etc/fstab becomes optional for many setups as systemd can
2715 discover certain partitions located on the root disk
2716 automatically. Home partitions are recognized under their
2717 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
2718 partitions are recognized under their GPT type ID
2719 0657fd6da4ab43c484e50933c84b4f4f.
2721 * systemd will no longer pass any environment from the kernel
2722 or initrd to system services. If you want to set an
2723 environment for all services, do so via the kernel command
2724 line systemd.setenv= assignment.
2726 * The systemd-sysctl tool no longer natively reads the file
2727 /etc/sysctl.conf. If desired, the file should be symlinked
2728 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
2729 legacy support by a symlink rather than built-in code, it
2730 also makes the otherwise hidden order of application of the
2731 different files visible. (Note that this partly reverts to a
2732 pre-198 application order of sysctl knobs!)
2734 * The "systemctl set-log-level" and "systemctl dump" commands
2735 have been moved to systemd-analyze.
2737 * systemd-run learned the new --remain-after-exit switch,
2738 which causes the scope unit not to be cleaned up
2739 automatically after the process terminated.
2741 * tmpfiles learned a new --exclude-prefix= switch to exclude
2742 certain paths from operation.
2744 * journald will now automatically flush all messages to disk
2745 as soon as a message at the log level CRIT, ALERT or EMERG
2748 Contributions from: Andrew Cook, Brandon Philips, Christian
2749 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
2750 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
2751 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
2752 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
2753 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
2754 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
2755 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
2756 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
2757 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
2758 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
2759 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
2760 William Giokas, Zbigniew Jędrzejewski-Szmek
2762 -- Berlin, 2013-09-13
2766 * The documentation has been updated to cover the various new
2767 concepts introduced with 205.
2769 * Unit files now understand the new %v specifier which
2770 resolves to the kernel version string as returned by "uname
2773 * systemctl now supports filtering the unit list output by
2774 load state, active state and sub state, using the new
2777 * "systemctl status" will now show the results of the
2778 condition checks (like ConditionPathExists= and similar) of
2779 the last start attempts of the unit. They are also logged to
2782 * "journalctl -b" may now be used to look for boot output of a
2783 specific boot. Try "journalctl -b -1" for the previous boot,
2784 but the syntax is substantially more powerful.
2786 * "journalctl --show-cursor" has been added which prints the
2787 cursor string the last shown log line. This may then be used
2788 with the new "journalctl --after-cursor=" switch to continue
2789 browsing logs from that point on.
2791 * "journalctl --force" may now be used to force regeneration
2794 * Creation of "dead" device nodes has been moved from udev
2795 into kmod and tmpfiles. Previously, udev would read the kmod
2796 databases to pre-generate dead device nodes based on meta
2797 information contained in kernel modules, so that these would
2798 be auto-loaded on access rather then at boot. As this
2799 does not really have much to do with the exposing actual
2800 kernel devices to userspace this has always been slightly
2801 alien in the udev codebase. Following the new scheme kmod
2802 will now generate a runtime snippet for tmpfiles from the
2803 module meta information and it now is tmpfiles' job to the
2804 create the nodes. This also allows overriding access and
2805 other parameters for the nodes using the usual tmpfiles
2806 facilities. As side effect this allows us to remove the
2807 CAP_SYS_MKNOD capability bit from udevd entirely.
2809 * logind's device ACLs may now be applied to these "dead"
2810 devices nodes too, thus finally allowing managed access to
2811 devices such as /dev/snd/sequencer whithout loading the
2812 backing module right-away.
2814 * A new RPM macro has been added that may be used to apply
2815 tmpfiles configuration during package installation.
2817 * systemd-detect-virt and ConditionVirtualization= now can
2818 detect User-Mode-Linux machines (UML).
2820 * journald will now implicitly log the effective capabilities
2821 set of processes in the message metadata.
2823 * systemd-cryptsetup has gained support for TrueCrypt volumes.
2825 * The initrd interface has been simplified (more specifically,
2826 support for passing performance data via environment
2827 variables and fsck results via files in /run has been
2828 removed). These features were non-essential, and are
2829 nowadays available in a much nicer way by having systemd in
2830 the initrd serialize its state and have the hosts systemd
2831 deserialize it again.
2833 * The udev "keymap" data files and tools to apply keyboard
2834 specific mappings of scan to key codes, and force-release
2835 scan code lists have been entirely replaced by a udev
2836 "keyboard" builtin and a hwdb data file.
2838 * systemd will now honour the kernel's "quiet" command line
2839 argument also during late shutdown, resulting in a
2840 completely silent shutdown when used.
2842 * There's now an option to control the SO_REUSEPORT socket
2843 option in .socket units.
2845 * Instance units will now automatically get a per-template
2846 subslice of system.slice unless something else is explicitly
2847 configured. For example, instances of sshd@.service will now
2848 implicitly be placed in system-sshd.slice rather than
2849 system.slice as before.
2851 * Test coverage support may now be enabled at build time.
2853 Contributions from: Dave Reisner, Frederic Crozat, Harald
2854 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
2855 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
2856 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
2857 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
2858 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
2859 Giokas, Zbigniew Jędrzejewski-Szmek
2861 -- Berlin, 2013-07-23
2865 * Two new unit types have been introduced:
2867 Scope units are very similar to service units, however, are
2868 created out of pre-existing processes -- instead of PID 1
2869 forking off the processes. By using scope units it is
2870 possible for system services and applications to group their
2871 own child processes (worker processes) in a powerful way
2872 which then maybe used to organize them, or kill them
2873 together, or apply resource limits on them.
2875 Slice units may be used to partition system resources in an
2876 hierarchical fashion and then assign other units to them. By
2877 default there are now three slices: system.slice (for all
2878 system services), user.slice (for all user sessions),
2879 machine.slice (for VMs and containers).
2881 Slices and scopes have been introduced primarily in
2882 context of the work to move cgroup handling to a
2883 single-writer scheme, where only PID 1
2884 creates/removes/manages cgroups.
2886 * There's a new concept of "transient" units. In contrast to
2887 normal units these units are created via an API at runtime,
2888 not from configuration from disk. More specifically this
2889 means it is now possible to run arbitrary programs as
2890 independent services, with all execution parameters passed
2891 in via bus APIs rather than read from disk. Transient units
2892 make systemd substantially more dynamic then it ever was,
2893 and useful as a general batch manager.
2895 * logind has been updated to make use of scope and slice units
2896 for managing user sessions. As a user logs in he will get
2897 his own private slice unit, to which all sessions are added
2898 as scope units. We also added support for automatically
2899 adding an instance of user@.service for the user into the
2900 slice. Effectively logind will no longer create cgroup
2901 hierarchies on its own now, it will defer entirely to PID 1
2902 for this by means of scope, service and slice units. Since
2903 user sessions this way become entities managed by PID 1
2904 the output of "systemctl" is now a lot more comprehensive.
2906 * A new mini-daemon "systemd-machined" has been added which
2907 may be used by virtualization managers to register local
2908 VMs/containers. nspawn has been updated accordingly, and
2909 libvirt will be updated shortly. machined will collect a bit
2910 of meta information about the VMs/containers, and assign
2911 them their own scope unit (see above). The collected
2912 meta-data is then made available via the "machinectl" tool,
2913 and exposed in "ps" and similar tools. machined/machinectl
2914 is compile-time optional.
2916 * As discussed earlier, the low-level cgroup configuration
2917 options ControlGroup=, ControlGroupModify=,
2918 ControlGroupPersistent=, ControlGroupAttribute= have been
2919 removed. Please use high-level attribute settings instead as
2920 well as slice units.
2922 * A new bus call SetUnitProperties() has been added to alter
2923 various runtime parameters of a unit. This is primarily
2924 useful to alter cgroup parameters dynamically in a nice way,
2925 but will be extended later on to make more properties
2926 modifiable at runtime. systemctl gained a new set-properties
2927 command that wraps this call.
2929 * A new tool "systemd-run" has been added which can be used to
2930 run arbitrary command lines as transient services or scopes,
2931 while configuring a number of settings via the command
2932 line. This tool is currently very basic, however already
2933 very useful. We plan to extend this tool to even allow
2934 queuing of execution jobs with time triggers from the
2935 command line, similar in fashion to "at".
2937 * nspawn will now inform the user explicitly that kernels with
2938 audit enabled break containers, and suggest the user to turn
2941 * Support for detecting the IMA and AppArmor security
2942 frameworks with ConditionSecurity= has been added.
2944 * journalctl gained a new "-k" switch for showing only kernel
2945 messages, mimicking dmesg output; in addition to "--user"
2946 and "--system" switches for showing only user's own logs
2949 * systemd-delta can now show information about drop-in
2950 snippets extending unit files.
2952 * libsystemd-bus has been substantially updated but is still
2953 not available as public API.
2955 * systemd will now look for the "debug" argument on the kernel
2956 command line and enable debug logging, similar to what
2957 "systemd.log_level=debug" already did before.
2959 * "systemctl set-default", "systemctl get-default" has been
2960 added to configure the default.target symlink, which
2961 controls what to boot into by default.
2963 * "systemctl set-log-level" has been added as a convenient
2964 way to raise and lower systemd logging threshold.
2966 * "systemd-analyze plot" will now show the time the various
2967 generators needed for execution, as well as information
2968 about the unit file loading.
2970 * libsystemd-journal gained a new sd_journal_open_files() call
2971 for opening specific journal files. journactl also gained a
2972 new switch to expose this new functionality. Previously we
2973 only supported opening all files from a directory, or all
2974 files from the system, as opening individual files only is
2975 racy due to journal file rotation.
2977 * systemd gained the new DefaultEnvironment= setting in
2978 /etc/systemd/system.conf to set environment variables for
2981 * If a privileged process logs a journal message with the
2982 OBJECT_PID= field set, then journald will automatically
2983 augment this with additional OBJECT_UID=, OBJECT_GID=,
2984 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
2985 system services want to log events about specific client
2986 processes. journactl/systemctl has been updated to make use
2987 of this information if all log messages regarding a specific
2990 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
2991 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
2992 Reisner, David Coppa, David King, David Strauss, Eelco
2993 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
2994 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
2995 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
2996 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
2997 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
2998 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
2999 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
3000 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
3001 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
3002 Łukasz Stelmach, 장동준
3006 * The Python bindings gained some minimal support for the APIs
3007 exposed by libsystemd-logind.
3009 * ConditionSecurity= gained support for detecting SMACK. Since
3010 this condition already supports SELinux and AppArmor we only
3011 miss IMA for this. Patches welcome!
3013 Contributions from: Karol Lewandowski, Lennart Poettering,
3014 Zbigniew Jędrzejewski-Szmek
3018 * systemd-nspawn will now create /etc/resolv.conf if
3019 necessary, before bind-mounting the host's file onto it.
3021 * systemd-nspawn will now store meta information about a
3022 container on the container's cgroup as extended attribute
3023 fields, including the root directory.
3025 * The cgroup hierarchy has been reworked in many ways. All
3026 objects any of the components systemd creates in the cgroup
3027 tree are now suffixed. More specifically, user sessions are
3028 now placed in cgroups suffixed with ".session", users in
3029 cgroups suffixed with ".user", and nspawn containers in
3030 cgroups suffixed with ".nspawn". Furthermore, all cgroup
3031 names are now escaped in a simple scheme to avoid collision
3032 of userspace object names with kernel filenames. This work
3033 is preparation for making these objects relocatable in the
3034 cgroup tree, in order to allow easy resource partitioning of
3035 these objects without causing naming conflicts.
3037 * systemctl list-dependencies gained the new switches
3038 --plain, --reverse, --after and --before.
3040 * systemd-inhibit now shows the process name of processes that
3041 have taken an inhibitor lock.
3043 * nss-myhostname will now also resolve "localhost"
3044 implicitly. This makes /etc/hosts an optional file and
3045 nicely handles that on IPv6 ::1 maps to both "localhost" and
3048 * libsystemd-logind.so gained a new call
3049 sd_get_machine_names() to enumerate running containers and
3050 VMs (currently only supported by very new libvirt and
3051 nspawn). sd_login_monitor can now be used to watch
3052 VMs/containers coming and going.
3054 * .include is not allowed recursively anymore, and only in
3055 unit files. Usually it is better to use drop-in snippets in
3056 .d/*.conf anyway, as introduced with systemd 198.
3058 * systemd-analyze gained a new "critical-chain" command that
3059 determines the slowest chain of units run during system
3060 boot-up. It is very useful for tracking down where
3061 optimizing boot time is the most beneficial.
3063 * systemd will no longer allow manipulating service paths in
3064 the name=systemd:/system cgroup tree using ControlGroup= in
3065 units. (But is still fine with it in all other dirs.)
3067 * There's a new systemd-nspawn@.service service file that may
3068 be used to easily run nspawn containers as system
3069 services. With the container's root directory in
3070 /var/lib/container/foobar it is now sufficient to run
3071 "systemctl start systemd-nspawn@foobar.service" to boot it.
3073 * systemd-cgls gained a new parameter "--machine" to list only
3074 the processes within a certain container.
3076 * ConditionSecurity= now can check for "apparmor". We still
3077 are lacking checks for SMACK and IMA for this condition
3078 check though. Patches welcome!
3080 * A new configuration file /etc/systemd/sleep.conf has been
3081 added that may be used to configure which kernel operation
3082 systemd is supposed to execute when "suspend", "hibernate"
3083 or "hybrid-sleep" is requested. This makes the new kernel
3084 "freeze" state accessible to the user.
3086 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
3087 the passed argument if applicable.
3089 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
3090 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
3091 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
3092 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
3093 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
3094 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
3095 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
3100 * The output of 'systemctl list-jobs' got some polishing. The
3101 '--type=' argument may now be passed more than once. A new
3102 command 'systemctl list-sockets' has been added which shows
3103 a list of kernel sockets systemd is listening on with the
3104 socket units they belong to, plus the units these socket
3107 * The experimental libsystemd-bus library got substantial
3108 updates to work in conjunction with the (also experimental)
3109 kdbus kernel project. It works well enough to exchange
3110 messages with some sophistication. Note that kdbus is not
3111 ready yet, and the library is mostly an elaborate test case
3112 for now, and not installable.
3114 * systemd gained a new unit 'systemd-static-nodes.service'
3115 that generates static device nodes earlier during boot, and
3116 can run in conjunction with udev.
3118 * libsystemd-login gained a new call sd_pid_get_user_unit()
3119 to retrieve the user systemd unit a process is running
3120 in. This is useful for systems where systemd is used as
3123 * systemd-nspawn now places all containers in the new /machine
3124 top-level cgroup directory in the name=systemd
3125 hierarchy. libvirt will soon do the same, so that we get a
3126 uniform separation of /system, /user and /machine for system
3127 services, user processes and containers/virtual
3128 machines. This new cgroup hierarchy is also useful to stick
3129 stable names to specific container instances, which can be
3130 recognized later this way (this name may be controlled
3131 via systemd-nspawn's new -M switch). libsystemd-login also
3132 gained a new call sd_pid_get_machine_name() to retrieve the
3133 name of the container/VM a specific process belongs to.
3135 * bootchart can now store its data in the journal.
3137 * libsystemd-journal gained a new call
3138 sd_journal_add_conjunction() for AND expressions to the
3139 matching logic. This can be used to express more complex
3140 logical expressions.
3142 * journactl can now take multiple --unit= and --user-unit=
3145 * The cryptsetup logic now understands the "luks.key=" kernel
3146 command line switch for specifying a file to read the
3147 decryption key from. Also, if a configured key file is not
3148 found the tool will now automatically fall back to prompting
3151 * Python systemd.journal module was updated to wrap recently
3152 added functions from libsystemd-journal. The interface was
3153 changed to bring the low level interface in s.j._Reader
3154 closer to the C API, and the high level interface in
3155 s.j.Reader was updated to wrap and convert all data about
3158 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
3159 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
3160 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
3161 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
3162 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
3163 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3167 * journalctl --update-catalog now understands a new --root=
3168 option to operate on catalogs found in a different root
3171 * During shutdown after systemd has terminated all running
3172 services a final killing loop kills all remaining left-over
3173 processes. We will now print the name of these processes
3174 when we send SIGKILL to them, since this usually indicates a
3177 * If /etc/crypttab refers to password files stored on
3178 configured mount points automatic dependencies will now be
3179 generated to ensure the specific mount is established first
3180 before the key file is attempted to be read.
3182 * 'systemctl status' will now show information about the
3183 network sockets a socket unit is listening on.
3185 * 'systemctl status' will also shown information about any
3186 drop-in configuration file for units. (Drop-In configuration
3187 files in this context are files such as
3188 /etc/systemd/systemd/foobar.service.d/*.conf)
3190 * systemd-cgtop now optionally shows summed up CPU times of
3191 cgroups. Press '%' while running cgtop to switch between
3192 percentage and absolute mode. This is useful to determine
3193 which cgroups use up the most CPU time over the entire
3194 runtime of the system. systemd-cgtop has also been updated
3195 to be 'pipeable' for processing with further shell tools.
3197 * 'hostnamectl set-hostname' will now allow setting of FQDN
3200 * The formatting and parsing of time span values has been
3201 changed. The parser now understands fractional expressions
3202 such as "5.5h". The formatter will now output fractional
3203 expressions for all time spans under 1min, i.e. "5.123456s"
3204 rather than "5s 123ms 456us". For time spans under 1s
3205 millisecond values are shown, for those under 1ms
3206 microsecond values are shown. This should greatly improve
3207 all time-related output of systemd.
3209 * libsystemd-login and libsystemd-journal gained new
3210 functions for querying the poll() events mask and poll()
3211 timeout value for integration into arbitrary event
3214 * localectl gained the ability to list available X11 keymaps
3215 (models, layouts, variants, options).
3217 * 'systemd-analyze dot' gained the ability to filter for
3218 specific units via shell-style globs, to create smaller,
3219 more useful graphs. I.e. it is now possible to create simple
3220 graphs of all the dependencies between only target units, or
3221 of all units that Avahi has dependencies with.
3223 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
3224 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
3225 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
3226 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
3227 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
3228 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
3229 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
3233 * The boot-time readahead implementation for rotating media
3234 will now read the read-ahead data in multiple passes which
3235 consist of all read requests made in equidistant time
3236 intervals. This means instead of strictly reading read-ahead
3237 data in its physical order on disk we now try to find a
3238 middle ground between physical and access time order.
3240 * /etc/os-release files gained a new BUILD_ID= field for usage
3241 on operating systems that provide continuous builds of OS
3244 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
3245 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
3246 William Douglas, Zbigniew Jędrzejewski-Szmek
3250 * systemd-python gained an API exposing libsystemd-daemon.
3252 * The SMACK setup logic gained support for uploading CIPSO
3255 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
3256 ReadOnlyDirectories= and InaccessibleDirectories= has
3257 changed. The private /tmp and /var/tmp directories are now
3258 shared by all processes of a service (which means
3259 ExecStartPre= may now leave data in /tmp that ExecStart= of
3260 the same service can still access). When a service is
3261 stopped its temporary directories are immediately deleted
3262 (normal clean-up with tmpfiles is still done in addition to
3265 * By default, systemd will now set a couple of sysctl
3266 variables in the kernel: the safe sysrq options are turned
3267 on, IP route verification is turned on, and source routing
3268 disabled. The recently added hardlink and softlink
3269 protection of the kernel is turned on. These settings should
3270 be reasonably safe, and good defaults for all new systems.
3272 * The predictable network naming logic may now be turned off
3273 with a new kernel command line switch: net.ifnames=0.
3275 * A new libsystemd-bus module has been added that implements a
3276 pretty complete D-Bus client library. For details see:
3278 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
3280 * journald will now explicitly flush the journal files to disk
3281 at the latest 5min after each write. The file will then also
3282 be marked offline until the next write. This should increase
3283 reliability in case of a crash. The synchronization delay
3284 can be configured via SyncIntervalSec= in journald.conf.
3286 * There's a new remote-fs-setup.target unit that can be used
3287 to pull in specific services when at least one remote file
3288 system is to be mounted.
3290 * There are new targets timers.target and paths.target as
3291 canonical targets to pull user timer and path units in
3292 from. This complements sockets.target with a similar
3293 purpose for socket units.
3295 * libudev gained a new call udev_device_set_attribute_value()
3296 to set sysfs attributes of a device.
3298 * The udev daemon now sets the default number of worker
3299 processes executed in parallel based on the number of available
3300 CPUs instead of the amount of available RAM. This is supposed
3301 to provide a more reliable default and limit a too aggressive
3302 paralellism for setups with 1000s of devices connected.
3304 Contributions from: Auke Kok, Colin Walters, Cristian
3305 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
3306 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
3307 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
3308 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
3309 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
3310 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
3311 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
3312 Zbigniew Jędrzejewski-Szmek
3316 * Configuration of unit files may now be extended via drop-in
3317 files without having to edit/override the unit files
3318 themselves. More specifically, if the administrator wants to
3319 change one value for a service file foobar.service he can
3320 now do so by dropping in a configuration snippet into
3321 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
3322 will load all these snippets and apply them on top of the
3323 main unit configuration file, possibly extending or
3324 overriding its settings. Using these drop-in snippets is
3325 generally nicer than the two earlier options for changing
3326 unit files locally: copying the files from
3327 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
3328 them there; or creating a new file in /etc/systemd/system/
3329 that incorporates the original one via ".include". Drop-in
3330 snippets into these .d/ directories can be placed in any
3331 directory systemd looks for units in, and the usual
3332 overriding semantics between /usr/lib, /etc and /run apply
3335 * Most unit file settings which take lists of items can now be
3336 reset by assigning the empty string to them. For example,
3337 normally, settings such as Environment=FOO=BAR append a new
3338 environment variable assignment to the environment block,
3339 each time they are used. By assigning Environment= the empty
3340 string the environment block can be reset to empty. This is
3341 particularly useful with the .d/*.conf drop-in snippets
3342 mentioned above, since this adds the ability to reset list
3343 settings from vendor unit files via these drop-ins.
3345 * systemctl gained a new "list-dependencies" command for
3346 listing the dependencies of a unit recursively.
3348 * Inhibitors are now honored and listed by "systemctl
3349 suspend", "systemctl poweroff" (and similar) too, not only
3350 GNOME. These commands will also list active sessions by
3353 * Resource limits (as exposed by the various control group
3354 controllers) can now be controlled dynamically at runtime
3355 for all units. More specifically, you can now use a command
3356 like "systemctl set-cgroup-attr foobar.service cpu.shares
3357 2000" to alter the CPU shares a specific service gets. These
3358 settings are stored persistently on disk, and thus allow the
3359 administrator to easily adjust the resource usage of
3360 services with a few simple commands. This dynamic resource
3361 management logic is also available to other programs via the
3362 bus. Almost any kernel cgroup attribute and controller is
3365 * systemd-vconsole-setup will now copy all font settings to
3366 all allocated VTs, where it previously applied them only to
3369 * libsystemd-login gained the new sd_session_get_tty() API
3372 * This release drops support for a few legacy or
3373 distribution-specific LSB facility names when parsing init
3374 scripts: $x-display-manager, $mail-transfer-agent,
3375 $mail-transport-agent, $mail-transfer-agent, $smtp,
3376 $null. Also, the mail-transfer-agent.target unit backing
3377 this has been removed. Distributions which want to retain
3378 compatibility with this should carry the burden for
3379 supporting this themselves and patch support for these back
3380 in, if they really need to. Also, the facilities $syslog and
3381 $local_fs are now ignored, since systemd does not support
3382 early-boot LSB init scripts anymore, and these facilities
3383 are implied anyway for normal services. syslog.target has
3386 * There are new bus calls on PID1's Manager object for
3387 cancelling jobs, and removing snapshot units. Previously,
3388 both calls were only available on the Job and Snapshot
3391 * systemd-journal-gatewayd gained SSL support.
3393 * The various "environment" files, such as /etc/locale.conf
3394 now support continuation lines with a backslash ("\") as
3395 last character in the line, similarly in style (but different)
3396 to how this is supported in shells.
3398 * For normal user processes the _SYSTEMD_USER_UNIT= field is
3399 now implicitly appended to every log entry logged. systemctl
3400 has been updated to filter by this field when operating on a
3401 user systemd instance.
3403 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
3404 CAP_AUDIT_CONTROL capabilities to the capabilities set for
3405 the container. This makes it easier to boot unmodified
3406 Fedora systems in a container, which however still requires
3407 audit=0 to be passed on the kernel command line. Auditing in
3408 kernel and userspace is unfortunately still too broken in
3409 context of containers, hence we recommend compiling it out
3410 of the kernel or using audit=0. Hopefully this will be fixed
3411 one day for good in the kernel.
3413 * nspawn gained the new --bind= and --bind-ro= parameters to
3414 bind mount specific directories from the host into the
3417 * nspawn will now mount its own devpts file system instance
3418 into the container, in order not to leak pty devices from
3419 the host into the container.
3421 * systemd will now read the firmware boot time performance
3422 information from the EFI variables, if the used boot loader
3423 supports this, and takes it into account for boot performance
3424 analysis via "systemd-analyze". This is currently supported
3425 only in conjunction with Gummiboot, but could be supported
3426 by other boot loaders too. For details see:
3428 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
3430 * A new generator has been added that automatically mounts the
3431 EFI System Partition (ESP) to /boot, if that directory
3432 exists, is empty, and no other file system has been
3433 configured to be mounted there.
3435 * logind will now send out PrepareForSleep(false) out
3436 unconditionally, after coming back from suspend. This may be
3437 used by applications as asynchronous notification for
3438 system resume events.
3440 * "systemctl unlock-sessions" has been added, that allows
3441 unlocking the screens of all user sessions at once, similar
3442 to how "systemctl lock-sessions" already locked all users
3443 sessions. This is backed by a new D-Bus call UnlockSessions().
3445 * "loginctl seat-status" will now show the master device of a
3446 seat. (i.e. the device of a seat that needs to be around for
3447 the seat to be considered available, usually the graphics
3450 * tmpfiles gained a new "X" line type, that allows
3451 configuration of files and directories (with wildcards) that
3452 shall be excluded from automatic cleanup ("aging").
3454 * udev default rules set the device node permissions now only
3455 at "add" events, and do not change them any longer with a
3456 later "change" event.
3458 * The log messages for lid events and power/sleep keypresses
3459 now carry a message ID.
3461 * We now have a substantially larger unit test suite, but this
3462 continues to be work in progress.
3464 * udevadm hwdb gained a new --root= parameter to change the
3465 root directory to operate relative to.
3467 * logind will now issue a background sync() request to the kernel
3468 early at shutdown, so that dirty buffers are flushed to disk early
3469 instead of at the last moment, in order to optimize shutdown
3472 * A new bootctl tool has been added that is an interface for
3473 certain boot loader operations. This is currently a preview
3474 and is likely to be extended into a small mechanism daemon
3475 like timedated, localed, hostnamed, and can be used by
3476 graphical UIs to enumerate available boot options, and
3477 request boot into firmware operations.
3479 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
3480 the rest of the package. It also has been updated to work
3481 correctly in initrds.
3483 * Policykit previously has been runtime optional, and is now
3484 also compile time optional via a configure switch.
3486 * systemd-analyze has been reimplemented in C. Also "systemctl
3487 dot" has moved into systemd-analyze.
3489 * "systemctl status" with no further parameters will now print
3490 the status of all active or failed units.
3492 * Operations such as "systemctl start" can now be executed
3493 with a new mode "--irreversible" which may be used to queue
3494 operations that cannot accidentally be reversed by a later
3495 job queuing. This is by default used to make shutdown
3496 requests more robust.
3498 * The Python API of systemd now gained a new module for
3499 reading journal files.
3501 * A new tool kernel-install has been added that can install
3502 kernel images according to the Boot Loader Specification:
3504 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
3506 * Boot time console output has been improved to provide
3507 animated boot time output for hanging jobs.
3509 * A new tool systemd-activate has been added which can be used
3510 to test socket activation with, directly from the command
3511 line. This should make it much easier to test and debug
3512 socket activation in daemons.
3514 * journalctl gained a new "--reverse" (or -r) option to show
3515 journal output in reverse order (i.e. newest line first).
3517 * journalctl gained a new "--pager-end" (or -e) option to jump
3518 to immediately jump to the end of the journal in the
3519 pager. This is only supported in conjunction with "less".
3521 * journalctl gained a new "--user-unit=" option, that works
3522 similarly to "--unit=" but filters for user units rather than
3525 * A number of unit files to ease adoption of systemd in
3526 initrds has been added. This moves some minimal logic from
3527 the various initrd implementations into systemd proper.
3529 * The journal files are now owned by a new group
3530 "systemd-journal", which exists specifically to allow access
3531 to the journal, and nothing else. Previously, we used the
3532 "adm" group for that, which however possibly covers more
3533 than just journal/log file access. This new group is now
3534 already used by systemd-journal-gatewayd to ensure this
3535 daemon gets access to the journal files and as little else
3536 as possible. Note that "make install" will also set FS ACLs
3537 up for /var/log/journal to give "adm" and "wheel" read
3538 access to it, in addition to "systemd-journal" which owns
3539 the journal files. We recommend that packaging scripts also
3540 add read access to "adm" + "wheel" to /var/log/journal, and
3541 all existing/future journal files. To normal users and
3542 administrators little changes, however packagers need to
3543 ensure to create the "systemd-journal" system group at
3544 package installation time.
3546 * The systemd-journal-gatewayd now runs as unprivileged user
3547 systemd-journal-gateway:systemd-journal-gateway. Packaging
3548 scripts need to create these system user/group at
3551 * timedated now exposes a new boolean property CanNTP that
3552 indicates whether a local NTP service is available or not.
3554 * systemd-detect-virt will now also detect xen PVs
3556 * The pstore file system is now mounted by default, if it is
3559 * In addition to the SELinux and IMA policies we will now also
3560 load SMACK policies at early boot.
3562 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
3563 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
3564 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
3565 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
3566 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
3567 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
3568 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
3569 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
3570 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
3571 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
3572 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
3573 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
3574 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
3575 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
3579 * Timer units now support calendar time events in addition to
3580 monotonic time events. That means you can now trigger a unit
3581 based on a calendar time specification such as "Thu,Fri
3582 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
3583 or fifth day of any month of the year 2013, given that it is
3584 a thursday or friday. This brings timer event support
3585 considerably closer to cron's capabilities. For details on
3586 the supported calendar time specification language see
3589 * udev now supports a number of different naming policies for
3590 network interfaces for predictable names, and a combination
3591 of these policies is now the default. Please see this wiki
3592 document for details:
3594 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
3596 * Auke Kok's bootchart implementation has been added to the
3597 systemd tree. It is an optional component that can graph the
3598 boot in quite some detail. It is one of the best bootchart
3599 implementations around and minimal in its code and
3602 * nss-myhostname has been integrated into the systemd source
3603 tree. nss-myhostname guarantees that the local hostname
3604 always stays resolvable via NSS. It has been a weak
3605 requirement of systemd-hostnamed since a long time, and
3606 since its code is actually trivial we decided to just
3607 include it in systemd's source tree. It can be turned off
3608 with a configure switch.
3610 * The read-ahead logic is now capable of properly detecting
3611 whether a btrfs file system is on SSD or rotating media, in
3612 order to optimize the read-ahead scheme. Previously, it was
3613 only capable of detecting this on traditional file systems
3616 * In udev, additional device properties are now read from the
3617 IAB in addition to the OUI database. Also, Bluetooth company
3618 identities are attached to the devices as well.
3620 * In service files %U may be used as specifier that is
3621 replaced by the configured user name of the service.
3623 * nspawn may now be invoked without a controlling TTY. This
3624 makes it suitable for invocation as its own service. This
3625 may be used to set up a simple containerized server system
3626 using only core OS tools.
3628 * systemd and nspawn can now accept socket file descriptors
3629 when they are started for socket activation. This enables
3630 implementation of socket activated nspawn
3631 containers. i.e. think about autospawning an entire OS image
3632 when the first SSH or HTTP connection is received. We expect
3633 that similar functionality will also be added to libvirt-lxc
3636 * journalctl will now suppress ANSI color codes when
3637 presenting log data.
3639 * systemctl will no longer show control group information for
3640 a unit if a the control group is empty anyway.
3642 * logind can now automatically suspend/hibernate/shutdown the
3645 * /etc/machine-info and hostnamed now also expose the chassis
3646 type of the system. This can be used to determine whether
3647 the local system is a laptop, desktop, handset or
3648 tablet. This information may either be configured by the
3649 user/vendor or is automatically determined from ACPI and DMI
3650 information if possible.
3652 * A number of PolicyKit actions are now bound together with
3653 "imply" rules. This should simplify creating UIs because
3654 many actions will now authenticate similar ones as well.
3656 * Unit files learnt a new condition ConditionACPower= which
3657 may be used to conditionalize a unit depending on whether an
3658 AC power source is connected or not, of whether the system
3659 is running on battery power.
3661 * systemctl gained a new "is-failed" verb that may be used in
3662 shell scripts and suchlike to check whether a specific unit
3663 is in the "failed" state.
3665 * The EnvironmentFile= setting in unit files now supports file
3666 globbing, and can hence be used to easily read a number of
3667 environment files at once.
3669 * systemd will no longer detect and recognize specific
3670 distributions. All distribution-specific #ifdeffery has been
3671 removed, systemd is now fully generic and
3672 distribution-agnostic. Effectively, not too much is lost as
3673 a lot of the code is still accessible via explicit configure
3674 switches. However, support for some distribution specific
3675 legacy configuration file formats has been dropped. We
3676 recommend distributions to simply adopt the configuration
3677 files everybody else uses now and convert the old
3678 configuration from packaging scripts. Most distributions
3679 already did that. If that's not possible or desirable,
3680 distributions are welcome to forward port the specific
3681 pieces of code locally from the git history.
3683 * When logging a message about a unit systemd will now always
3684 log the unit name in the message meta data.
3686 * localectl will now also discover system locale data that is
3687 not stored in locale archives, but directly unpacked.
3689 * logind will no longer unconditionally use framebuffer
3690 devices as seat masters, i.e. as devices that are required
3691 to be existing before a seat is considered preset. Instead,
3692 it will now look for all devices that are tagged as
3693 "seat-master" in udev. By default framebuffer devices will
3694 be marked as such, but depending on local systems other
3695 devices might be marked as well. This may be used to
3696 integrate graphics cards using closed source drivers (such
3697 as NVidia ones) more nicely into logind. Note however, that
3698 we recommend using the open source NVidia drivers instead,
3699 and no udev rules for the closed-source drivers will be
3700 shipped from us upstream.
3702 Contributions from: Adam Williamson, Alessandro Crismani, Auke
3703 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
3704 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
3705 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
3706 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
3707 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
3708 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
3709 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
3710 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
3711 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
3712 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
3713 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
3718 * udev gained support for loading additional device properties
3719 from an indexed database that is keyed by vendor/product IDs
3720 and similar device identifiers. For the beginning this
3721 "hwdb" is populated with data from the well-known PCI and
3722 USB database, but also includes PNP, ACPI and OID data. In
3723 the longer run this indexed database shall grow into
3724 becoming the one central database for non-essential
3725 userspace device metadata. Previously, data from the PCI/USB
3726 database was only attached to select devices, since the
3727 lookup was a relatively expensive operation due to O(n) time
3728 complexity (with n being the number of entries in the
3729 database). Since this is now O(1), we decided to add in this
3730 data for all devices where this is available, by
3731 default. Note that the indexed database needs to be rebuilt
3732 when new data files are installed. To achieve this you need
3733 to update your packaging scripts to invoke "udevadm hwdb
3734 --update" after installation of hwdb data files. For
3735 RPM-based distributions we introduced the new
3736 %udev_hwdb_update macro for this purpose.
3738 * The Journal gained support for the "Message Catalog", an
3739 indexed database to link up additional information with
3740 journal entries. For further details please check:
3742 http://www.freedesktop.org/wiki/Software/systemd/catalog
3744 The indexed message catalog database also needs to be
3745 rebuilt after installation of message catalog files. Use
3746 "journalctl --update-catalog" for this. For RPM-based
3747 distributions we introduced the %journal_catalog_update
3748 macro for this purpose.
3750 * The Python Journal bindings gained support for the standard
3751 Python logging framework.
3753 * The Journal API gained new functions for checking whether
3754 the underlying file system of a journal file is capable of
3755 properly reporting file change notifications, or whether
3756 applications that want to reflect journal changes "live"
3757 need to recheck journal files continuously in appropriate
3760 * It is now possible to set the "age" field for tmpfiles
3761 entries to 0, indicating that files matching this entry
3762 shall always be removed when the directories are cleaned up.
3764 * coredumpctl gained a new "gdb" verb which invokes gdb
3765 right-away on the selected coredump.
3767 * There's now support for "hybrid sleep" on kernels that
3768 support this, in addition to "suspend" and "hibernate". Use
3769 "systemctl hybrid-sleep" to make use of this.
3771 * logind's HandleSuspendKey= setting (and related settings)
3772 now gained support for a new "lock" setting to simply
3773 request the screen lock on all local sessions, instead of
3774 actually executing a suspend or hibernation.
3776 * systemd will now mount the EFI variables file system by
3779 * Socket units now gained support for configuration of the
3780 SMACK security label.
3782 * timedatectl will now output the time of the last and next
3783 daylight saving change.
3785 * We dropped support for various legacy and distro-specific
3786 concepts, such as insserv, early-boot SysV services
3787 (i.e. those for non-standard runlevels such as 'b' or 'S')
3788 or ArchLinux /etc/rc.conf support. We recommend the
3789 distributions who still need support this to either continue
3790 to maintain the necessary patches downstream, or find a
3791 different solution. (Talk to us if you have questions!)
3793 * Various systemd components will now bypass PolicyKit checks
3794 for root and otherwise handle properly if PolicyKit is not
3795 found to be around. This should fix most issues for
3796 PolicyKit-less systems. Quite frankly this should have been
3797 this way since day one. It is absolutely our intention to
3798 make systemd work fine on PolicyKit-less systems, and we
3799 consider it a bug if something does not work as it should if
3800 PolicyKit is not around.
3802 * For embedded systems it is now possible to build udev and
3803 systemd without blkid and/or kmod support.
3805 * "systemctl switch-root" is now capable of switching root
3806 more than once. I.e. in addition to transitions from the
3807 initrd to the host OS it is now possible to transition to
3808 further OS images from the host. This is useful to implement
3809 offline updating tools.
3811 * Various other additions have been made to the RPM macros
3812 shipped with systemd. Use %udev_rules_update() after
3813 installing new udev rules files. %_udevhwdbdir,
3814 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
3815 %_sysctldir are now available which resolve to the right
3816 directories for packages to place various data files in.
3818 * journalctl gained the new --full switch (in addition to
3819 --all, to disable ellipsation for long messages.
3821 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
3822 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
3823 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
3824 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
3825 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
3826 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
3827 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
3828 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
3829 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
3833 * journalctl gained new --since= and --until= switches to
3834 filter by time. It also now supports nice filtering for
3835 units via --unit=/-u.
3837 * Type=oneshot services may use ExecReload= and do the
3840 * The journal daemon now supports time-based rotation and
3841 vacuuming, in addition to the usual disk-space based
3844 * The journal will now index the available field values for
3845 each field name. This enables clients to show pretty drop
3846 downs of available match values when filtering. The bash
3847 completion of journalctl has been updated
3848 accordingly. journalctl gained a new switch -F to list all
3849 values a certain field takes in the journal database.
3851 * More service events are now written as structured messages
3852 to the journal, and made recognizable via message IDs.
3854 * The timedated, localed and hostnamed mini-services which
3855 previously only provided support for changing time, locale
3856 and hostname settings from graphical DEs such as GNOME now
3857 also have a minimal (but very useful) text-based client
3858 utility each. This is probably the nicest way to changing
3859 these settings from the command line now, especially since
3860 it lists available options and is fully integrated with bash
3863 * There's now a new tool "systemd-coredumpctl" to list and
3864 extract coredumps from the journal.
3866 * We now install a README each in /var/log/ and
3867 /etc/rc.d/init.d explaining where the system logs and init
3868 scripts went. This hopefully should help folks who go to
3869 that dirs and look into the otherwise now empty void and
3870 scratch their heads.
3872 * When user-services are invoked (by systemd --user) the
3873 $MANAGERPID env var is set to the PID of systemd.
3875 * SIGRTMIN+24 when sent to a --user instance will now result
3876 in immediate termination of systemd.
3878 * gatewayd received numerous feature additions such as a
3879 "follow" mode, for live syncing and filtering.
3881 * browse.html now allows filtering and showing detailed
3882 information on specific entries. Keyboard navigation and
3883 mouse screen support has been added.
3885 * gatewayd/journalctl now supports HTML5/JSON
3886 Server-Sent-Events as output.
3888 * The SysV init script compatibility logic will now
3889 heuristically determine whether a script supports the
3890 "reload" verb, and only then make this available as
3893 * "systemctl status --follow" has been removed, use "journalctl
3896 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
3897 have been removed since they are hardly useful to be
3900 * And I'd like to take the opportunity to specifically mention
3901 Zbigniew for his great contributions. Zbigniew, you rock!
3903 Contributions from: Andrew Eikum, Christian Hesse, Colin
3904 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
3905 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
3906 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
3907 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
3908 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
3909 Jędrzejewski-Szmek, Сковорода Никита Андреевич
3913 * If /etc/vconsole.conf is non-existent or empty we will no
3914 longer load any console font or key map at boot by
3915 default. Instead the kernel defaults will be left
3916 intact. This is definitely the right thing to do, as no
3917 configuration should mean no configuration, and hard-coding
3918 font names that are different on all archs is probably a bad
3919 idea. Also, the kernel default key map and font should be
3920 good enough for most cases anyway, and mostly identical to
3921 the userspace fonts/key maps we previously overloaded them
3922 with. If distributions want to continue to default to a
3923 non-kernel font or key map they should ship a default
3924 /etc/vconsole.conf with the appropriate contents.
3926 Contributions from: Colin Walters, Daniel J Walsh, Dave
3927 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
3928 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3932 * journalctl gained a new --cursor= switch to show entries
3933 starting from the specified location in the journal.
3935 * We now enforce a size limit on journal entry fields exported
3936 with "-o json" in journalctl. Fields larger than 4K will be
3937 assigned null. This can be turned off with --all.
3939 * An (optional) journal gateway daemon is now available as
3940 "systemd-journal-gatewayd.service". This service provides
3941 access to the journal via HTTP and JSON. This functionality
3942 will be used to implement live log synchronization in both
3943 pull and push modes, but has various other users too, such
3944 as easy log access for debugging of embedded devices. Right
3945 now it is already useful to retrieve the journal via HTTP:
3947 # systemctl start systemd-journal-gatewayd.service
3948 # wget http://localhost:19531/entries
3950 This will download the journal contents in a
3951 /var/log/messages compatible format. The same as JSON:
3953 # curl -H"Accept: application/json" http://localhost:19531/entries
3955 This service is also accessible via a web browser where a
3956 single static HTML5 app is served that uses the JSON logic
3957 to enable the user to do some basic browsing of the
3958 journal. This will be extended later on. Here's an example
3959 screenshot of this app in its current state:
3961 http://0pointer.de/public/journal-gatewayd
3963 Contributions from: Kay Sievers, Lennart Poettering, Robert
3964 Milasan, Tom Gundersen
3968 * The bash completion logic is now available for journalctl
3971 * We do not mount the "cpuset" controller anymore together with
3972 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
3973 started if no parameters are assigned to it. "cpuset" hence
3974 broke code that assumed it it could create "cpu" groups and
3977 * journalctl -f will now subscribe to terminal size changes,
3978 and line break accordingly.
3980 Contributions from: Dave Reisner, Kay Sievers, Lennart
3981 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
3985 * nspawn will now create a symlink /etc/localtime in the
3986 container environment, copying the host's timezone
3987 setting. Previously this has been done via a bind mount, but
3988 since symlinks cannot be bind mounted this has now been
3989 changed to create/update the appropriate symlink.
3991 * journalctl -n's line number argument is now optional, and
3992 will default to 10 if omitted.
3994 * journald will now log the maximum size the journal files may
3995 take up on disk. This is particularly useful if the default
3996 built-in logic of determining this parameter from the file
3997 system size is used. Use "systemctl status
3998 systemd-journald.service" to see this information.
4000 * The multi-seat X wrapper tool has been stripped down. As X
4001 is now capable of enumerating graphics devices via udev in a
4002 seat-aware way the wrapper is not strictly necessary
4003 anymore. A stripped down temporary stop-gap is still shipped
4004 until the upstream display managers have been updated to
4005 fully support the new X logic. Expect this wrapper to be
4006 removed entirely in one of the next releases.
4008 * HandleSleepKey= in logind.conf has been split up into
4009 HandleSuspendKey= and HandleHibernateKey=. The old setting
4010 is not available anymore. X11 and the kernel are
4011 distinguishing between these keys and we should too. This
4012 also means the inhibition lock for these keys has been split
4015 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
4016 Poettering, Lukas Nykryn, Václav Pavlín
4020 * Whenever a unit changes state we will now log this to the
4021 journal and show along the unit's own log output in
4024 * ConditionPathIsMountPoint= can now properly detect bind
4025 mount points too. (Previously, a bind mount of one file
4026 system to another place in the same file system could not be
4027 detected as mount, since they shared struct stat's st_dev
4030 * We will now mount the cgroup controllers cpu, cpuacct,
4031 cpuset and the controllers net_cls, net_prio together by
4034 * nspawn containers will now have a virtualized boot
4035 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
4036 over with a randomized ID at container initialization). This
4037 has the effect of making "journalctl -b" do the right thing
4040 * The JSON output journal serialization has been updated not
4041 to generate "endless" list objects anymore, but rather one
4042 JSON object per line. This is more in line how most JSON
4043 parsers expect JSON objects. The new output mode
4044 "json-pretty" has been added to provide similar output, but
4045 neatly aligned for readability by humans.
4047 * We dropped all explicit sync() invocations in the shutdown
4048 code. The kernel does this implicitly anyway in the kernel
4049 reboot() syscall. halt(8)'s -n option is now a compatibility
4052 * We now support virtualized reboot() in containers, as
4053 supported by newer kernels. We will fall back to exit() if
4054 CAP_SYS_REBOOT is not available to the container. Also,
4055 nspawn makes use of this now and will actually reboot the
4056 container if the containerized OS asks for that.
4058 * journalctl will only show local log output by default
4059 now. Use --merge (-m) to show remote log output, too.
4061 * libsystemd-journal gained the new sd_journal_get_usage()
4062 call to determine the current disk usage of all journal
4063 files. This is exposed in the new "journalctl --disk-usage"
4066 * journald gained a new configuration setting SplitMode= in
4067 journald.conf which may be used to control how user journals
4068 are split off. See journald.conf(5) for details.
4070 * A new condition type ConditionFileNotEmpty= has been added.
4072 * tmpfiles' "w" lines now support file globbing, to write
4073 multiple files at once.
4075 * We added Python bindings for the journal submission
4076 APIs. More Python APIs for a number of selected APIs will
4077 likely follow. Note that we intend to add native bindings
4078 only for the Python language, as we consider it common
4079 enough to deserve bindings shipped within systemd. There are
4080 various projects outside of systemd that provide bindings
4081 for languages such as PHP or Lua.
4083 * Many conditions will now resolve specifiers such as %i. In
4084 addition, PathChanged= and related directives of .path units
4085 now support specifiers as well.
4087 * There's now a new RPM macro definition for the system preset
4090 * journald will now warn if it ca not forward a message to the
4091 syslog daemon because its socket is full.
4093 * timedated will no longer write or process /etc/timezone,
4094 except on Debian. As we do not support late mounted /usr
4095 anymore /etc/localtime always being a symlink is now safe,
4096 and hence the information in /etc/timezone is not necessary
4099 * logind will now always reserve one VT for a text getty (VT6
4100 by default). Previously if more than 6 X sessions where
4101 started they took up all the VTs with auto-spawned gettys,
4102 so that no text gettys were available anymore.
4104 * udev will now automatically inform the btrfs kernel logic
4105 about btrfs RAID components showing up. This should make
4106 simple hotplug based btrfs RAID assembly work.
4108 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
4109 (but not for its children which will stay at the kernel
4110 default). This should allow setups with a lot more listening
4113 * systemd will now always pass the configured timezone to the
4114 kernel at boot. timedated will do the same when the timezone
4117 * logind's inhibition logic has been updated. By default,
4118 logind will now handle the lid switch, the power and sleep
4119 keys all the time, even in graphical sessions. If DEs want
4120 to handle these events on their own they should take the new
4121 handle-power-key, handle-sleep-key and handle-lid-switch
4122 inhibitors during their runtime. A simple way to achieve
4123 that is to invoke the DE wrapped in an invocation of:
4125 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
4127 * Access to unit operations is now checked via SELinux taking
4128 the unit file label and client process label into account.
4130 * systemd will now notify the administrator in the journal
4131 when he over-mounts a non-empty directory.
4133 * There are new specifiers that are resolved in unit files,
4134 for the host name (%H), the machine ID (%m) and the boot ID
4137 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
4138 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
4139 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
4140 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
4141 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
4142 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
4143 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
4147 * Support for reading structured kernel messages from
4148 /dev/kmsg has now been added and is enabled by default.
4150 * Support for reading kernel messages from /proc/kmsg has now
4151 been removed. If you want kernel messages in the journal
4152 make sure to run a recent kernel (>= 3.5) that supports
4153 reading structured messages from /dev/kmsg (see
4154 above). /proc/kmsg is now exclusive property of classic
4155 syslog daemons again.
4157 * The libudev API gained the new
4158 udev_device_new_from_device_id() call.
4160 * The logic for file system namespace (ReadOnlyDirectory=,
4161 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
4162 require pivot_root() anymore. This means fewer temporary
4163 directories are created below /tmp for this feature.
4165 * nspawn containers will now see and receive all submounts
4166 made on the host OS below the root file system of the
4169 * Forward Secure Sealing is now supported for Journal files,
4170 which provide cryptographical sealing of journal files so
4171 that attackers cannot alter log history anymore without this
4172 being detectable. Lennart will soon post a blog story about
4173 this explaining it in more detail.
4175 * There are two new service settings RestartPreventExitStatus=
4176 and SuccessExitStatus= which allow configuration of exit
4177 status (exit code or signal) which will be excepted from the
4178 restart logic, resp. consider successful.
4180 * journalctl gained the new --verify switch that can be used
4181 to check the integrity of the structure of journal files and
4182 (if Forward Secure Sealing is enabled) the contents of
4185 * nspawn containers will now be run with /dev/stdin, /dev/fd/
4186 and similar symlinks pre-created. This makes running shells
4187 as container init process a lot more fun.
4189 * The fstab support can now handle PARTUUID= and PARTLABEL=
4192 * A new ConditionHost= condition has been added to match
4193 against the hostname (with globs) and machine ID. This is
4194 useful for clusters where a single OS image is used to
4195 provision a large number of hosts which shall run slightly
4196 different sets of services.
4198 * Services which hit the restart limit will now be placed in a
4201 Contributions from: Bertram Poettering, Dave Reisner, Huang
4202 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
4203 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
4207 * When running in --user mode systemd will now become a
4208 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
4209 tree a lot more organized.
4211 * A new PartOf= unit dependency type has been introduced that
4212 may be used to group services in a natural way.
4214 * "systemctl enable" may now be used to enable instances of
4217 * journalctl now prints error log levels in red, and
4218 warning/notice log levels in bright white. It also supports
4219 filtering by log level now.
4221 * cgtop gained a new -n switch (similar to top), to configure
4222 the maximum number of iterations to run for. It also gained
4223 -b, to run in batch mode (accepting no input).
4225 * The suffix ".service" may now be omitted on most systemctl
4226 command lines involving service unit names.
4228 * There's a new bus call in logind to lock all sessions, as
4229 well as a loginctl verb for it "lock-sessions".
4231 * libsystemd-logind.so gained a new call sd_journal_perror()
4232 that works similar to libc perror() but logs to the journal
4233 and encodes structured information about the error number.
4235 * /etc/crypttab entries now understand the new keyfile-size=
4238 * shutdown(8) now can send a (configurable) wall message when
4239 a shutdown is cancelled.
4241 * The mount propagation mode for the root file system will now
4242 default to "shared", which is useful to make containers work
4243 nicely out-of-the-box so that they receive new mounts from
4244 the host. This can be undone locally by running "mount
4245 --make-rprivate /" if needed.
4247 * The prefdm.service file has been removed. Distributions
4248 should maintain this unit downstream if they intend to keep
4249 it around. However, we recommend writing normal unit files
4250 for display managers instead.
4252 * Since systemd is a crucial part of the OS we will now
4253 default to a number of compiler switches that improve
4254 security (hardening) such as read-only relocations, stack
4255 protection, and suchlike.
4257 * The TimeoutSec= setting for services is now split into
4258 TimeoutStartSec= and TimeoutStopSec= to allow configuration
4259 of individual time outs for the start and the stop phase of
4262 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
4263 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
4264 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
4265 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
4266 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
4267 Gundersen, Zbigniew Jędrzejewski-Szmek
4271 * The journal and id128 C APIs are now fully documented as man
4274 * Extra safety checks have been added when transitioning from
4275 the initial RAM disk to the main system to avoid accidental
4278 * /etc/crypttab entries now understand the new keyfile-offset=
4281 * systemctl -t can now be used to filter by unit load state.
4283 * The journal C API gained the new sd_journal_wait() call to
4284 make writing synchronous journal clients easier.
4286 * journalctl gained the new -D switch to show journals from a
4289 * journalctl now displays a special marker between log
4290 messages of two different boots.
4292 * The journal is now explicitly flushed to /var via a service
4293 systemd-journal-flush.service, rather than implicitly simply
4294 by seeing /var/log/journal to be writable.
4296 * journalctl (and the journal C APIs) can now match for much
4297 more complex expressions, with alternatives and
4300 * When transitioning from the initial RAM disk to the main
4301 system we will now kill all processes in a killing spree to
4302 ensure no processes stay around by accident.
4304 * Three new specifiers may be used in unit files: %u, %h, %s
4305 resolve to the user name, user home directory resp. user
4306 shell. This is useful for running systemd user instances.
4308 * We now automatically rotate journal files if their data
4309 object hash table gets a fill level > 75%. We also size the
4310 hash table based on the configured maximum file size. This
4311 together should lower hash collisions drastically and thus
4312 speed things up a bit.
4314 * journalctl gained the new "--header" switch to introspect
4315 header data of journal files.
4317 * A new setting SystemCallFilters= has been added to services
4318 which may be used to apply blacklists or whitelists to
4319 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
4321 * nspawn gained a new --link-journal= switch (and quicker: -j)
4322 to link the container journal with the host. This makes it
4323 very easy to centralize log viewing on the host for all
4324 guests while still keeping the journal files separated.
4326 * Many bugfixes and optimizations
4328 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
4329 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
4330 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
4335 * Several tools now understand kernel command line arguments,
4336 which are only read when run in an initial RAM disk. They
4337 usually follow closely their normal counterparts, but are
4340 * There's a new tool to analyze the readahead files that are
4341 automatically generated at boot. Use:
4343 /usr/lib/systemd/systemd-readahead analyze /.readahead
4345 * We now provide an early debug shell on tty9 if this enabled. Use:
4347 systemctl enable debug-shell.service
4349 * All plymouth related units have been moved into the Plymouth
4350 package. Please make sure to upgrade your Plymouth version
4353 * systemd-tmpfiles now supports getting passed the basename of
4354 a configuration file only, in which case it will look for it
4355 in all appropriate directories automatically.
4357 * udevadm info now takes a /dev or /sys path as argument, and
4358 does the right thing. Example:
4360 udevadm info /dev/sda
4361 udevadm info /sys/class/block/sda
4363 * systemctl now prints a warning if a unit is stopped but a
4364 unit that might trigger it continues to run. Example: a
4365 service is stopped but the socket that activates it is left
4368 * "systemctl status" will now mention if the log output was
4369 shortened due to rotation since a service has been started.
4371 * The journal API now exposes functions to determine the
4372 "cutoff" times due to rotation.
4374 * journald now understands SIGUSR1 and SIGUSR2 for triggering
4375 immediately flushing of runtime logs to /var if possible,
4376 resp. for triggering immediate rotation of the journal
4379 * It is now considered an error if a service is attempted to
4380 be stopped that is not loaded.
4382 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
4384 * systemd-analyze now supports Python 3
4386 * tmpfiles now supports cleaning up directories via aging
4387 where the first level dirs are always kept around but
4388 directories beneath it automatically aged. This is enabled
4389 by prefixing the age field with '~'.
4391 * Seat objects now expose CanGraphical, CanTTY properties
4392 which is required to deal with very fast bootups where the
4393 display manager might be running before the graphics drivers
4394 completed initialization.
4396 * Seat objects now expose a State property.
4398 * We now include RPM macros for service enabling/disabling
4399 based on the preset logic. We recommend RPM based
4400 distributions to make use of these macros if possible. This
4401 makes it simpler to reuse RPM spec files across
4404 * We now make sure that the collected systemd unit name is
4405 always valid when services log to the journal via
4408 * There's a new man page kernel-command-line(7) detailing all
4409 command line options we understand.
4411 * The fstab generator may now be disabled at boot by passing
4412 fstab=0 on the kernel command line.
4414 * A new kernel command line option modules-load= is now understood
4415 to load a specific kernel module statically, early at boot.
4417 * Unit names specified on the systemctl command line are now
4418 automatically escaped as needed. Also, if file system or
4419 device paths are specified they are automatically turned
4420 into the appropriate mount or device unit names. Example:
4422 systemctl status /home
4423 systemctl status /dev/sda
4425 * The SysVConsole= configuration option has been removed from
4426 system.conf parsing.
4428 * The SysV search path is no longer exported on the D-Bus
4431 * The Names= option is been removed from unit file parsing.
4433 * There's a new man page bootup(7) detailing the boot process.
4435 * Every unit and every generator we ship with systemd now
4436 comes with full documentation. The self-explanatory boot is
4439 * A couple of services gained "systemd-" prefixes in their
4440 name if they wrap systemd code, rather than only external
4441 code. Among them fsck@.service which is now
4442 systemd-fsck@.service.
4444 * The HaveWatchdog property has been removed from the D-Bus
4447 * systemd.confirm_spawn= on the kernel command line should now
4450 * There's a new man page crypttab(5) which details all options
4451 we actually understand.
4453 * systemd-nspawn gained a new --capability= switch to pass
4454 additional capabilities to the container.
4456 * timedated will now read known NTP implementation unit names
4457 from /usr/lib/systemd/ntp-units.d/*.list,
4458 systemd-timedated-ntp.target has been removed.
4460 * journalctl gained a new switch "-b" that lists log data of
4461 the current boot only.
4463 * The notify socket is in the abstract namespace again, in
4464 order to support daemons which chroot() at start-up.
4466 * There is a new Storage= configuration option for journald
4467 which allows configuration of where log data should go. This
4468 also provides a way to disable journal logging entirely, so
4469 that data collected is only forwarded to the console, the
4470 kernel log buffer or another syslog implementation.
4472 * Many bugfixes and optimizations
4474 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
4475 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
4476 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
4477 Shawn Landden, Tom Gundersen
4481 * "systemctl help <unit>" now shows the man page if one is
4484 * Several new man pages have been added.
4486 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
4487 MaxLevelConsole= can now be specified in
4488 journald.conf. These options allow reducing the amount of
4489 data stored on disk or forwarded by the log level.
4491 * TimerSlackNSec= can now be specified in system.conf for
4492 PID1. This allows system-wide power savings.
4494 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
4495 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
4500 * logind is now capable of (optionally) handling power and
4501 sleep keys as well as the lid switch.
4503 * journalctl now understands the syntax "journalctl
4504 /usr/bin/avahi-daemon" to get all log output of a specific
4507 * CapabilityBoundingSet= in system.conf now also influences
4508 the capability bound set of usermode helpers of the kernel.
4510 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
4511 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
4512 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
4513 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
4517 * Note that we skipped 139 releases here in order to set the
4518 new version to something that is greater than both udev's
4519 and systemd's most recent version number.
4521 * udev: all udev sources are merged into the systemd source tree now.
4522 All future udev development will happen in the systemd tree. It
4523 is still fully supported to use the udev daemon and tools without
4524 systemd running, like in initramfs or other init systems. Building
4525 udev though, will require the *build* of the systemd tree, but
4526 udev can be properly *run* without systemd.
4528 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
4529 should be used to create dead device nodes as workarounds for broken
4532 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
4533 no longer supported. udev_monitor_new_from_netlink() needs to be
4534 used to subscribe to events.
4536 * udev: when udevd is started by systemd, processes which are left
4537 behind by forking them off of udev rules, are unconditionally cleaned
4538 up and killed now after the event handling has finished. Services or
4539 daemons must be started as systemd services. Services can be
4540 pulled-in by udev to get started, but they can no longer be directly
4541 forked by udev rules.
4543 * udev: the daemon binary is called systemd-udevd now and installed
4544 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
4545 to adapt to that, create symlink, or rename the binary after building
4548 * libudev no longer provides these symbols:
4549 udev_monitor_from_socket()
4550 udev_queue_get_failed_list_entry()
4551 udev_get_{dev,sys,run}_path()
4552 The versions number was bumped and symbol versioning introduced.
4554 * systemd-loginctl and systemd-journalctl have been renamed
4555 to loginctl and journalctl to match systemctl.
4557 * The config files: /etc/systemd/systemd-logind.conf and
4558 /etc/systemd/systemd-journald.conf have been renamed to
4559 logind.conf and journald.conf. Package updates should rename
4560 the files to the new names on upgrade.
4562 * For almost all files the license is now LGPL2.1+, changed
4563 from the previous GPL2.0+. Exceptions are some minor stuff
4564 of udev (which will be changed to LGPL2.1 eventually, too),
4565 and the MIT licensed sd-daemon.[ch] library that is suitable
4566 to be used as drop-in files.
4568 * systemd and logind now handle system sleep states, in
4569 particular suspending and hibernating.
4571 * logind now implements a sleep/shutdown/idle inhibiting logic
4572 suitable for a variety of uses. Soonishly Lennart will blog
4573 about this in more detail.
4575 * var-run.mount and var-lock.mount are no longer provided
4576 (which prevously bind mounted these directories to their new
4577 places). Distributions which have not converted these
4578 directories to symlinks should consider stealing these files
4579 from git history and add them downstream.
4581 * We introduced the Documentation= field for units and added
4582 this to all our shipped units. This is useful to make it
4583 easier to explore the boot and the purpose of the various
4586 * All smaller setup units (such as
4587 systemd-vconsole-setup.service) now detect properly if they
4588 are run in a container and are skipped when
4589 appropriate. This guarantees an entirely noise-free boot in
4590 Linux container environments such as systemd-nspawn.
4592 * A framework for implementing offline system updates is now
4593 integrated, for details see:
4594 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
4596 * A new service type Type=idle is available now which helps us
4597 avoiding ugly interleaving of getty output and boot status
4600 * There's now a system-wide CapabilityBoundingSet= option to
4601 globally reduce the set of capabilities for the
4602 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
4603 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
4604 even CAP_NET_ADMIN system-wide for secure systems.
4606 * There are now system-wide DefaultLimitXXX= options to
4607 globally change the defaults of the various resource limits
4608 for all units started by PID 1.
4610 * Harald Hoyer's systemd test suite has been integrated into
4611 systemd which allows easy testing of systemd builds in qemu
4612 and nspawn. (This is really awesome! Ask us for details!)
4614 * The fstab parser is now implemented as generator, not inside
4617 * systemctl will now warn you if .mount units generated from
4618 /etc/fstab are out of date due to changes in fstab that
4619 have not been read by systemd yet.
4621 * systemd is now suitable for usage in initrds. Dracut has
4622 already been updated to make use of this. With this in place
4623 initrds get a slight bit faster but primarily are much
4624 easier to introspect and debug since "systemctl status" in
4625 the host system can be used to introspect initrd services,
4626 and the journal from the initrd is kept around too.
4628 * systemd-delta has been added, a tool to explore differences
4629 between user/admin configuration and vendor defaults.
4631 * PrivateTmp= now affects both /tmp and /var/tmp.
4633 * Boot time status messages are now much prettier and feature
4634 proper english language. Booting up systemd has never been
4637 * Read-ahead pack files now include the inode number of all
4638 files to pre-cache. When the inode changes the pre-caching
4639 is not attempted. This should be nicer to deal with updated
4640 packages which might result in changes of read-ahead
4643 * We now temporaritly lower the kernel's read_ahead_kb variable
4644 when collecting read-ahead data to ensure the kernel's
4645 built-in read-ahead does not add noise to our measurements
4646 of necessary blocks to pre-cache.
4648 * There's now RequiresMountsFor= to add automatic dependencies
4649 for all mounts necessary for a specific file system path.
4651 * MountAuto= and SwapAuto= have been removed from
4652 system.conf. Mounting file systems at boot has to take place
4655 * nspawn now learned a new switch --uuid= to set the machine
4656 ID on the command line.
4658 * nspawn now learned the -b switch to automatically search
4661 * vt102 is now the default TERM for serial TTYs, upgraded from
4664 * systemd-logind now works on VT-less systems.
4666 * The build tree has been reorganized. The individual
4667 components now have directories of their own.
4669 * A new condition type ConditionPathIsReadWrite= is now available.
4671 * nspawn learned the new -C switch to create cgroups for the
4672 container in other hierarchies.
4674 * We now have support for hardware watchdogs, configurable in
4677 * The scheduled shutdown logic now has a public API.
4679 * We now mount /tmp as tmpfs by default, but this can be
4680 masked and /etc/fstab can override it.
4682 * Since udisks does not make use of /media anymore we are not
4683 mounting a tmpfs on it anymore.
4685 * journalctl gained a new --local switch to only interleave
4686 locally generated journal files.
4688 * We can now load the IMA policy at boot automatically.
4690 * The GTK tools have been split off into a systemd-ui.
4692 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
4693 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
4694 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
4695 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
4696 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
4697 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
4698 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
4699 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
4700 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
4705 * This is mostly a bugfix release
4707 * Support optional initialization of the machine ID from the
4708 KVM or container configured UUID.
4710 * Support immediate reboots with "systemctl reboot -ff"
4712 * Show /etc/os-release data in systemd-analyze output
4714 * Many bugfixes for the journal, including endianness fixes and
4715 ensuring that disk space enforcement works
4717 * sd-login.h is C++ comptaible again
4719 * Extend the /etc/os-release format on request of the Debian
4722 * We now refuse non-UTF8 strings used in various configuration
4723 and unit files. This is done to ensure we do not pass invalid
4724 data over D-Bus or expose it elsewhere.
4726 * Register Mimo USB Screens as suitable for automatic seat
4729 * Read SELinux client context from journal clients in a race
4732 * Reorder configuration file lookup order. /etc now always
4733 overrides /run in order to allow the administrator to always
4734 and unconditionally override vendor supplied or
4735 automatically generated data.
4737 * The various user visible bits of the journal now have man
4738 pages. We still lack man pages for the journal API calls
4741 * We now ship all man pages in HTML format again in the
4744 Contributions from: Dave Reisner, Dirk Eibach, Frederic
4745 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
4746 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
4751 * This is mostly a bugfix release
4753 * systems lacking /etc/os-release are no longer supported.
4755 * Various functionality updates to libsystemd-login.so
4757 * Track class of PAM logins to distinguish greeters from
4760 Contributions from: Kay Sievers, Lennart Poettering, Michael
4765 * This is an important bugfix release for v41.
4767 * Building man pages is now optional which should be useful
4768 for those building systemd from git but unwilling to install
4771 * Watchdog support for supervising services is now usable. In
4772 a future release support for hardware watchdogs
4773 (i.e. /dev/watchdog) will be added building on this.
4775 * Service start rate limiting is now configurable and can be
4776 turned off per service. When a start rate limit is hit a
4777 reboot can automatically be triggered.
4779 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
4781 Contributions from: Benjamin Franzke, Bill Nottingham,
4782 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
4783 Schmidt, Michał Górny, Piotr Drąg
4787 * The systemd binary is installed /usr/lib/systemd/systemd now;
4788 An existing /sbin/init symlink needs to be adapted with the
4791 * The code that loads kernel modules has been ported to invoke
4792 libkmod directly, instead of modprobe. This means we do not
4793 support systems with module-init-tools anymore.
4795 * Watchdog support is now already useful, but still not
4798 * A new kernel command line option systemd.setenv= is
4799 understood to set system wide environment variables
4800 dynamically at boot.
4802 * We now limit the set of capabilities of systemd-journald.
4804 * We now set SIGPIPE to ignore by default, since it only is
4805 useful in shell pipelines, and has little use in general
4806 code. This can be disabled with IgnoreSIPIPE=no in unit
4809 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
4810 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
4815 * This is mostly a bugfix release
4817 * We now expose the reason why a service failed in the
4818 "Result" D-Bus property.
4820 * Rudimentary service watchdog support (will be completed over
4821 the next few releases.)
4823 * When systemd forks off in order execute some service we will
4824 now immediately changes its argv[0] to reflect which process
4825 it will execute. This is useful to minimize the time window
4826 with a generic argv[0], which makes bootcharts more useful
4828 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
4829 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
4830 Mike Kazantsev, Ray Strode
4834 * This is mostly a test release, but incorporates many
4837 * New systemd-cgtop tool to show control groups by their
4840 * Linking against libacl for ACLs is optional again. If
4841 disabled, support tracking device access for active logins
4842 goes becomes unavailable, and so does access to the user
4843 journals by the respective users.
4845 * If a group "adm" exists, journal files are automatically
4846 owned by them, thus allow members of this group full access
4847 to the system journal as well as all user journals.
4849 * The journal now stores the SELinux context of the logging
4850 client for all entries.
4852 * Add C++ inclusion guards to all public headers
4854 * New output mode "cat" in the journal to print only text
4855 messages, without any meta data like date or time.
4857 * Include tiny X server wrapper as a temporary stop-gap to
4858 teach XOrg udev display enumeration. This is used by display
4859 managers such as gdm, and will go away as soon as XOrg
4860 learned native udev hotplugging for display devices.
4862 * Add new systemd-cat tool for executing arbitrary programs
4863 with STDERR/STDOUT connected to the journal. Can also act as
4864 BSD logger replacement, and does so by default.
4866 * Optionally store all locally generated coredumps in the
4867 journal along with meta data.
4869 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
4870 writing short strings to files (for usage for /sys), and for
4871 creating symlinks, character and block device nodes.
4873 * New unit file option ControlGroupPersistent= to make cgroups
4874 persistent, following the mechanisms outlined in
4875 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
4877 * Support multiple local RTCs in a sane way
4879 * No longer monopolize IO when replaying readahead data on
4880 rotating disks, since we might starve non-file-system IO to
4881 death, since fanotify() will not see accesses done by blkid,
4884 * Do not show kernel threads in systemd-cgls anymore, unless
4885 requested with new -k switch.
4887 Contributions from: Dan Horák, Kay Sievers, Lennart
4888 Poettering, Michal Schmidt
4892 * This is mostly a test release, but incorporates many
4895 * The git repository moved to:
4896 git://anongit.freedesktop.org/systemd/systemd
4897 ssh://git.freedesktop.org/git/systemd/systemd
4899 * First release with the journal
4900 http://0pointer.de/blog/projects/the-journal.html
4902 * The journal replaces both systemd-kmsg-syslogd and
4903 systemd-stdout-bridge.
4905 * New sd_pid_get_unit() API call in libsystemd-logind
4907 * Many systemadm clean-ups
4909 * Introduce remote-fs-pre.target which is ordered before all
4910 remote mounts and may be used to start services before all
4913 * Added Mageia support
4915 * Add bash completion for systemd-loginctl
4917 * Actively monitor PID file creation for daemons which exit in
4918 the parent process before having finished writing the PID
4919 file in the daemon process. Daemons which do this need to be
4920 fixed (i.e. PID file creation must have finished before the
4921 parent exits), but we now react a bit more gracefully to them.
4923 * Add colourful boot output, mimicking the well-known output
4924 of existing distributions.
4926 * New option PassCredentials= for socket units, for
4927 compatibility with a recent kernel ABI breakage.
4929 * /etc/rc.local is now hooked in via a generator binary, and
4930 thus will no longer act as synchronization point during
4933 * systemctl list-unit-files now supports --root=.
4935 * systemd-tmpfiles now understands two new commands: z, Z for
4936 relabelling files according to the SELinux database. This is
4937 useful to apply SELinux labels to specific files in /sys,
4940 * Output of SysV services is now forwarded to both the console
4941 and the journal by default, not only just the console.
4943 * New man pages for all APIs from libsystemd-login.
4945 * The build tree got reorganized and a the build system is a
4946 lot more modular allowing embedded setups to specifically
4947 select the components of systemd they are interested in.
4949 * Support for Linux systems lacking the kernel VT subsystem is
4952 * configure's --with-rootdir= got renamed to
4953 --with-rootprefix= to follow the naming used by udev and
4956 * Unless specified otherwise we will now install to /usr instead
4957 of /usr/local by default.
4959 * Processes with '@' in argv[0][0] are now excluded from the
4960 final shut-down killing spree, following the logic explained
4962 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
4964 * All processes remaining in a service cgroup when we enter
4965 the START or START_PRE states are now killed with
4966 SIGKILL. That means it is no longer possible to spawn
4967 background processes from ExecStart= lines (which was never
4968 supported anyway, and bad style).
4970 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
4971 reloading of units together.
4973 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
4974 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
4975 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
4976 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
4977 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek