1 systemd System and Service Manager
5 * The gudev library has been extracted into a separate repository
6 available at: https://git.gnome.org/browse/libgudev/
7 It is now managed as part of the Gnome project. Distributions
8 are recommended to pass --disable-gudev to systemd and use
9 gudev from the Gnome project instead. gudev is still included
10 in systemd, for now. It will be removed soon, though. Please
11 also see the announcement-thread on systemd-devel:
12 http://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
14 * systemd now exposes a CPUUsageNSec= property for each
15 service unit on the bus, that contains the overall consumed
16 CPU time of a service (the sum of what each process of the
17 service consumed). This value is only available if
18 CPUAccounting= is turned on for a service, and is then shown
19 in the "systemctl status" output.
21 * Support for configuring alternative mappings of the old SysV
22 runlevels to systemd targets has been removed. They are now
23 hardcoded in a way that runlevels 2, 3, 4 all map to
24 multi-user.target and 5 to graphical.target (which
25 previously was already the default behaviour).
27 * The auto-mounter logic gained support for mount point
28 expiry, using a new TimeoutIdleSec= setting in .automount
29 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
31 * The EFI System Partition (ESP) as mounted to /boot by
32 systemd-efi-boot-generator will now be unmounted
33 automatically after 2 minutes of not being used. This should
34 minimize the risk of ESP corruptions.
36 * New /etc/fstab options x-systemd.requires= and
37 x-systemd.requires-mounts-for= are now supported to express
38 additional dependencies for mounts. This is useful for
39 journalling file systems that support external journal
40 devices or overlay file systems that require underlying file
41 systems to be mounted.
43 * systemd does not support direct live-upgrades (via systemctl
44 daemon-reexec) from versions older than v44 anymore. As no
45 distribution we are aware of shipped such old versions in a
46 stable release this should not be problematic.
48 * When systemd forks off a new per-connection service instance
49 it will now set the $REMOTE_ADDR environment variable to the
50 remote IP address, and $REMOTE_PORT environment variable to
51 the remote IP port. This behaviour is similar to the
52 corresponding environment variables defined by CGI.
54 * systemd-networkd gained support for uplink failure
55 detection. The BindCarrier= option allows binding interface
56 configuration dynamically to the link sense of other
57 interfaces. This is useful to achieve behaviour like in
60 * systemd-networkd gained support for configuring the DHCP
61 client identifier to use when requesting leases.
63 * systemd-networkd now has a per-network UseNTP= option to
64 configure whether NTP server information acquired via DHCP
65 is passed on to services like systemd-timesyncd.
67 * systemd-networkd gained support for vti6 tunnels.
69 * Note that systemd-networkd manages the sysctl variable
70 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
71 it is configured for since v219. The variable controls IP
72 forwarding, and is a per-interface alternative to the global
73 /proc/sys/net/ipv[46]/ip_forward. This setting is
74 configurable in the IPForward= option, which defaults to
75 "no". This means if networkd is used for an interface it is
76 no longer sufficient to set the global sysctl option to turn
77 on IP forwarding! Instead, the .network file option
78 IPForward= needs to be turned on! Note that the
79 implementation of this behaviour was broken in v219 and has
82 * Many bonding and vxlan options are now configurable in
85 * systemd-nspawn gained a new --property= setting to set unit
86 properties for the container scope. This is useful for
87 setting resource parameters (e.g "CPUShares=500") on
88 containers started from the command line.
90 * systemd-nspawn gained a new --private-users= switch to make
91 use of user namespacing available on recent Linux kernels.
93 * systemd-nspawn may now be called as part of a shell pipeline
94 in which case the pipes used for stdin and stdout are passed
95 directly to the process invoked in the container, without
96 indirection via a pseudo tty.
98 * systemd-nspawn gained a new switch to control the UNIX
99 signal to use when killing the init process of the container
102 * systemd-nspawn gained a new --overlay= switch for mounting
103 overlay file systems into the container using the new kernel
106 * When a container image is imported via systemd-importd and
107 the host file system is not btrfs, a loopback block device
108 file is created in /var/lib/machines.raw with a btrfs file
109 system inside. It is then mounted to /var/lib/machines to
110 enable btrfs features for container management. The loopback
111 file and btrfs file system is grown as needed when container
112 images are imported via systemd-importd.
114 * systemd-machined/systemd-importd gained support for btrfs
115 quota, to enforce container disk space limits on disk. This
116 is exposed in "machinectl set-limit".
118 * systemd-importd now can import containers from local .tar,
119 .raw and .qcow2 images, and export them to .tar and .raw. It
120 can also import dkr v2 images now from the network (on top
123 * systemd-importd gained support for verifying downloaded
124 images with gpg2 (previously only gpg1 was supported).
126 * systemd-machined, systemd-logind, systemd: most bus calls
127 are now accessible to unprivileged processes via
128 PolicyKit. Also, systemd-logind will now allow users to kill
129 their own sessions without further privileges or
132 * systemd-shutdownd has been removed. This service was
133 previously responsible for implementing scheduled shutdowns
134 as exposed in /usr/bin/shutdown's time parameter. This
135 functionality has now been moved into systemd-logind and is
136 accessible via a bus interface.
138 * "systemctl reboot" gained a new switch --firmware-setup that
139 can be used to reboot into the EFI firmware setup, if that
140 is available. systemd-logind now exposes an API on the bus
141 to trigger such reboots, in case graphical desktop UIs want
142 to cover this functionality.
144 * "systemctl enable", "systemctl disable" and "systemctl mask"
145 now support a new "--now" switch. If specified the units
146 that are enabled will also be started, and the ones
147 disabled/masked also stopped.
149 * The Gummiboot EFI boot loader tool has been merged into
150 systemd, and renamed to "systemd-boot". The bootctl tool has been
151 updated to support systemd-boot.
153 * An EFI kernel stub has been added that may be used to create
154 kernel EFI binaries that contain not only the actual kernel,
155 but also an initrd, boot splash, command line and OS release
156 information. This combined binary can then be signed as a
157 single image, so that the firmware can verify it all in one
158 step. systemd-boot has special support for EFI binaries created
159 like this and can extract OS release information from them
160 and show them in the boot menu. This functionality is useful
161 to implement cryptographically verified boot schemes.
163 * Optional support has been added to systemd-fsck to pass
164 fsck's progress report to an AF_UNIX socket in the file
167 * udev will no longer create device symlinks for all block
168 devices by default. A blacklist for excluding special block
169 devices from this logic has been turned into a whitelist
170 that requires picking block devices explicitly that require
173 * A new (currently still internal) API sd-device.h has been
174 added to libsystemd. This modernized API is supposed to
175 replace libudev eventually. In fact, already much of libudev
176 is now just a wrapper around sd-device.h.
178 * A new hwdb database for storing metadata about pointing
179 stick devices has been added.
181 * systemd-tmpfiles gained support for setting file attributes
182 similar to the "chattr" tool with new 'h' and 'H' lines.
184 * systemd-journald will no longer unconditionally set the
185 btrfs NOCOW flag on new journal files. This is instead done
186 with tmpfiles snippet using the new 'h' line type. This
187 allows easy disabling of this logic, by masking the
188 journal-nocow.conf tmpfiles file.
190 * systemd-journald will now translate audit message types to
191 human readable identifiers when writing them to the
192 journal. This should improve readability of audit messages.
194 * The LUKS logic gained support for the offset= and skip=
195 options in /etc/crypttab, as previously implemented by
198 * /usr/lib/os-release gained a new optional field VARIANT= for
199 distributions that support multiple variants (such as a
200 desktop edition, a server edition, ...)
202 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
203 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
204 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
205 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
206 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
207 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
208 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
209 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
210 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
211 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
212 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
213 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
214 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
215 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
216 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
217 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
218 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
219 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
220 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
221 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
222 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
223 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
224 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
225 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
226 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
227 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
228 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
230 -- Berlin, 2015-05-??
234 * Introduce a new API "sd-hwdb.h" for querying the hardware
235 metadata database. With this minimal interface one can query
236 and enumerate the udev hwdb, decoupled from the old libudev
237 library. libudev's interface for this is now only a wrapper
238 around sd-hwdb. A new tool systemd-hwdb has been added to
239 interface with and update the database.
241 * When any of systemd's tools copies files (for example due to
242 tmpfiles' C lines) a btrfs reflink will attempted first,
243 before bytewise copying is done.
245 * systemd-nspawn gained a new --ephemeral switch. When
246 specified a btrfs snapshot is taken of the container's root
247 directory, and immediately removed when the container
248 terminates again. Thus, a container can be started whose
249 changes never alter the container's root directory, and are
250 lost on container termination. This switch can also be used
251 for starting a container off the root file system of the
252 host without affecting the host OS. This switch is only
253 available on btrfs file systems.
255 * systemd-nspawn gained a new --template= switch. It takes the
256 path to a container tree to use as template for the tree
257 specified via --directory=, should that directory be
258 missing. This allows instantiating containers dynamically,
259 on first run. This switch is only available on btrfs file
262 * When a .mount unit refers to a mount point on which multiple
263 mounts are stacked, and the .mount unit is stopped all of
264 the stacked mount points will now be unmounted until no
267 * systemd now has an explicit notion of supported and
268 unsupported unit types. Jobs enqueued for unsupported unit
269 types will now fail with an "unsupported" error code. More
270 specifically .swap, .automount and .device units are not
271 supported in containers, .busname units are not supported on
272 non-kdbus systems. .swap and .automount are also not
273 supported if their respective kernel compile time options
276 * machinectl gained support for two new "copy-from" and
277 "copy-to" commands for copying files from a running
278 container to the host or vice versa.
280 * machinectl gained support for a new "bind" command to bind
281 mount host directories into local containers. This is
282 currently only supported for nspawn containers.
284 * networkd gained support for configuring bridge forwarding
285 database entries (fdb) from .network files.
287 * A new tiny daemon "systemd-importd" has been added that can
288 download container images in tar, raw, qcow2 or dkr formats,
289 and make them available locally in /var/lib/machines, so
290 that they can run as nspawn containers. The daemon can GPG
291 verify the downloads (not supported for dkr, since it has no
292 provisions for verifying downloads). It will transparently
293 decompress bz2, xz, gzip compressed downloads if necessary,
294 and restore sparse files on disk. The daemon uses privilege
295 separation to ensure the actual download logic runs with
296 fewer privileges than the deamon itself. machinectl has
297 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
298 make the functionality of importd available to the
299 user. With this in place the Fedora and Ubuntu "Cloud"
300 images can be downloaded and booted as containers unmodified
301 (the Fedora images lack the appropriate GPG signature files
302 currently, so they cannot be verified, but this will change
303 soon, hopefully). Note that downloading images is currently
304 only fully supported on btrfs.
306 * machinectl is now able to list container images found in
307 /var/lib/machines, along with some metadata about sizes of
308 disk and similar. If the directory is located on btrfs and
309 quota is enabled, this includes quota display. A new command
310 "image-status" has been added that shows additional
311 information about images.
313 * machinectl is now able to clone container images
314 efficiently, if the underlying file system (btrfs) supports
315 it, with the new "machinectl list-images" command. It also
316 gained commands for renaming and removing images, as well as
317 marking them read-only or read-write (supported also on
318 legacy file systems).
320 * networkd gained support for collecting LLDP network
321 announcements, from hardware that supports this. This is
322 shown in networkctl output.
324 * systemd-run gained support for a new -t (--pty) switch for
325 invoking a binary on a pty whose input and output is
326 connected to the invoking terminal. This allows executing
327 processes as system services while interactively
328 communicating with them via the terminal. Most interestingly
329 this is supported across container boundaries. Invoking
330 "systemd-run -t /bin/bash" is an alternative to running a
331 full login session, the difference being that the former
332 will not register a session, nor go through the PAM session
335 * tmpfiles gained support for a new "v" line type for creating
336 btrfs subvolumes. If the underlying file system is a legacy
337 file system, this automatically degrades to creating a
338 normal directory. Among others /var/lib/machines is now
339 created like this at boot, should it be missing.
341 * The directory /var/lib/containers/ has been deprecated and
342 been replaced by /var/lib/machines. The term "machines" has
343 been used in the systemd context as generic term for both
344 VMs and containers, and hence appears more appropriate for
345 this, as the directory can also contain raw images bootable
348 * systemd-nspawn when invoked with -M but without --directory=
349 or --image= is now capable of searching for the container
350 root directory, subvolume or disk image automatically, in
351 /var/lib/machines. systemd-nspawn@.service has been updated
352 to make use of this, thus allowing it to be used for raw
355 * A new machines.target unit has been introduced that is
356 supposed to group all containers/VMs invoked as services on
357 the system. systemd-nspawn@.service has been updated to
360 * machinectl gained a new "start" command, for invoking a
361 container as a service. "machinectl start foo" is mostly
362 equivalent to "systemctl start systemd-nspawn@foo.service",
363 but handles escaping in a nicer way.
365 * systemd-nspawn will now mount most of the cgroupfs tree
366 read-only into each container, with the exception of the
367 container's own subtree in the name=systemd hierarchy.
369 * journald now sets the special FS_NOCOW file flag for its
370 journal files. This should improve performance on btrfs, by
371 avoiding heavy fragmentation when journald's write-pattern
372 is used on COW file systems. It degrades btrfs' data
373 integrity guarantees for the files to the same levels as for
374 ext3/ext4 however. This should be OK though as journald does
375 its own data integrity checks and all its objects are
376 checksummed on disk. Also, journald should handle btrfs disk
377 full events a lot more gracefully now, by processing SIGBUS
378 errors, and not relying on fallocate() anymore.
380 * When journald detects that journal files it is writing to
381 have been deleted it will immediately start new journal
384 * systemd now provides a way to store file descriptors
385 per-service in PID 1.This is useful for daemons to ensure
386 that fds they require are not lost during a daemon
387 restart. The fds are passed to the deamon on the next
388 invocation in the same way socket activation fds are
389 passed. This is now used by journald to ensure that the
390 various sockets connected to all the system's stdout/stderr
391 are not lost when journald is restarted. File descriptors
392 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
393 an extension to sd_notify(). Note that a limit is enforced
394 on the number of fds a service can store in PID 1, and it
395 defaults to 0, so that no fds may be stored, unless this is
396 explicitly turned on.
398 * The default TERM variable to use for units connected to a
399 terminal, when no other value is explicitly is set is now
400 vt220 rather than vt102. This should be fairly safe still,
401 but allows PgUp/PgDn work.
403 * The /etc/crypttab option header= as known from Debian is now
406 * "loginctl user-status" and "loginctl session-status" will
407 now show the last 10 lines of log messages of the
408 user/session following the status output. Similar,
409 "machinectl status" will show the last 10 log lines
410 associated with a virtual machine or container
411 service. (Note that this is usually not the log messages
412 done in the VM/container itself, but simply what the
413 container manager logs. For nspawn this includes all console
416 * "loginctl session-status" without further argument will now
417 show the status of the session of the caller. Similar,
418 "lock-session", "unlock-session", "activate",
419 "enable-linger", "disable-linger" may now be called without
420 session/user parameter in which case they apply to the
421 caller's session/user.
423 * An X11 session scriptlet is now shipped that uploads
424 $DISPLAY and $XAUTHORITY into the environment of the systemd
425 --user daemon if a session begins. This should improve
426 compatibility with X11 enabled applications run as systemd
429 * Generators are now subject to masking via /etc and /run, the
430 same way as unit files.
432 * networkd .network files gained support for configuring
433 per-link IPv4/IPv6 packet forwarding as well as IPv4
434 masquerading. This is by default turned on for veth links to
435 containers, as registered by systemd-nspawn. This means that
436 nspawn containers run with --network-veth will now get
437 automatic routed access to the host's networks without any
438 further configuration or setup, as long as networkd runs on
441 * systemd-nspawn gained the --port= (-p) switch to expose TCP
442 or UDP posts of a container on the host. With this in place
443 it is possible to run containers with private veth links
444 (--network-veth), and have their functionality exposed on
445 the host as if their services were running directly on the
448 * systemd-nspawn's --network-veth switch now gained a short
449 version "-n", since with the changes above it is now truly
450 useful out-of-the-box. The systemd-nspawn@.service has been
451 updated to make use of it too by default.
453 * systemd-nspawn will now maintain a per-image R/W lock, to
454 ensure that the same image is not started more than once
455 writable. (It's OK to run an image multiple times
456 simultaneously in read-only mode.)
458 * systemd-nspawn's --image= option is now capable of
459 dissecting and booting MBR and GPT disk images that contain
460 only a single active Linux partition. Previously it
461 supported only GPT disk images with proper GPT type
462 IDs. This allows running cloud images from major
463 distributions directly with systemd-nspawn, without
466 * In addition to collecting mouse dpi data in the udev
467 hardware database, there's now support for collecting angle
468 information for mouse scroll wheels. The database is
469 supposed to guarantee similar scrolling behavior on mice
470 that it knows about. There's also support for collecting
471 information about Touchpad types.
473 * udev's input_id built-in will now also collect touch screen
474 dimension data and attach it to probed devices.
476 * /etc/os-release gained support for a Distribution Privacy
479 * networkd gained support for creating "ipvlan", "gretap",
480 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
482 * systemd-tmpfiles gained support for "a" lines for setting
485 * systemd-nspawn will now mount /tmp in the container to
486 tmpfs, automatically.
488 * systemd now exposes the memory.usage_in_bytes cgroup
489 attribute and shows it for each service in the "systemctl
490 status" output, if available.
492 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
493 immediate reboot is triggered. This useful if shutdown is
494 hung and is unable to complete, to expedite the
495 operation. Note that this kind of reboot will still unmount
496 all file systems, and hence should not result in fsck being
499 * A .device unit for an optical block device will now be
500 considered active only when a medium is in the drive. Also,
501 mount units are now bound to their backing devices thus
502 triggering automatic unmounting when devices become
503 unavailable. With this in place systemd will now
504 automatically unmount left-over mounts when a CD-ROM is
505 ejected or an USB stick is yanked from the system.
507 * networkd-wait-online now has support for waiting for
508 specific interfaces only (with globbing), and for giving up
509 after a configurable timeout.
511 * networkd now exits when idle. It will be automatically
512 restarted as soon as interfaces show up, are removed or
513 change state. networkd will stay around as long as there is
514 at least one DHCP state machine or similar around, that keep
517 * networkd may now configure IPv6 link-local addressing in
518 addition to IPv4 link-local addressing.
520 * The IPv6 "token" for use in SLAAC may now be configured for
521 each .network interface in networkd.
523 * Routes configured with networkd may now be assigned a scope
526 * networkd's [Match] sections now support globbing and lists
527 of multiple space-separated matches per item.
529 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
530 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
531 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
532 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
533 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
534 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
535 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
536 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
537 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
538 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
539 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
540 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
541 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
542 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
543 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
544 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
545 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
546 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
547 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
548 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
549 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
550 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
551 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
552 Hoffmann, Zbigniew Jędrzejewski-Szmek
554 -- Berlin, 2015-02-16
558 * When querying unit file enablement status (for example via
559 "systemctl is-enabled"), a new state "indirect" is now known
560 which indicates that a unit might not be enabled itself, but
561 another unit listed in its Alias= setting might be.
563 * Similar to the various existing ConditionXYZ= settings for
564 units there are now matching AssertXYZ= settings. While
565 failing conditions cause a unit to be skipped, but its job
566 to succeed, failing assertions declared like this will cause
567 a unit start operation and its job to fail.
569 * hostnamed now knows a new chassis type "embedded".
571 * systemctl gained a new "edit" command. When used on a unit
572 file this allows extending unit files with .d/ drop-in
573 configuration snippets or editing the full file (after
574 copying it from /usr/lib to /etc). This will invoke the
575 user's editor (as configured with $EDITOR), and reload the
576 modified configuration after editing.
578 * "systemctl status" now shows the suggested enablement state
579 for a unit, as declared in the (usually vendor-supplied)
582 * nss-myhostname will now resolve the single-label host name
583 "gateway" to the locally configured default IP routing
584 gateways, ordered by their metrics. This assigns a stable
585 name to the used gateways, regardless which ones are
586 currently configured. Note that the name will only be
587 resolved after all other name sources (if nss-myhostname is
588 configured properly) and should hence not negatively impact
589 systems that use the single-label host name "gateway" in
592 * systemd-inhibit now allows filtering by mode when listing
595 * Scope and service units gained a new "Delegate" boolean
596 property, which when set allows processes running inside the
597 unit to further partition resources. This is primarily
598 useful for systemd user instances as well as container
601 * journald will now pick up audit messages directly from
602 the kernel, and log them like any other log message. The
603 audit fields are split up and fully indexed. This means that
604 journalctl in many ways is now a (nicer!) alternative to
605 ausearch, the traditional audit client. Note that this
606 implements only a minimal audit client, if you want the
607 special audit modes like reboot-on-log-overflow, please use
608 the traditional auditd instead, which can be used in
609 parallel to journald.
611 * The ConditionSecurity= unit file option now understands the
612 special string "audit" to check whether auditing is
615 * journalctl gained two new commands --vacuum-size= and
616 --vacuum-time= to delete old journal files until the
617 remaining ones take up no more the specified size on disk,
618 or are not older than the specified time.
620 * A new, native PPPoE library has been added to sd-network,
621 systemd's library of light-weight networking protocols. This
622 library will be used in a future version of networkd to
623 enable PPPoE communication without an external pppd daemon.
625 * The busctl tool now understands a new "capture" verb that
626 works similar to "monitor", but writes a packet capture
627 trace to STDOUT that can be redirected to a file which is
628 compatible with libcap's capture file format. This can then
629 be loaded in Wireshark and similar tools to inspect bus
632 * The busctl tool now understands a new "tree" verb that shows
633 the object trees of a specific service on the bus, or of all
636 * The busctl tool now understands a new "introspect" verb that
637 shows all interfaces and members of objects on the bus,
638 including their signature and values. This is particularly
639 useful to get more information about bus objects shown by
640 the new "busctl tree" command.
642 * The busctl tool now understands new verbs "call",
643 "set-property" and "get-property" for invoking bus method
644 calls, setting and getting bus object properties in a
647 * busctl gained a new --augment-creds= argument that controls
648 whether the tool shall augment credential information it
649 gets from the bus with data from /proc, in a possibly
652 * nspawn's --link-journal= switch gained two new values
653 "try-guest" and "try-host" that work like "guest" and
654 "host", but do not fail if the host has no persistent
655 journalling enabled. -j is now equivalent to
656 --link-journal=try-guest.
658 * macvlan network devices created by nspawn will now have
659 stable MAC addresses.
661 * A new SmackProcessLabel= unit setting has been added, which
662 controls the SMACK security label processes forked off by
663 the respective unit shall use.
665 * If compiled with --enable-xkbcommon, systemd-localed will
666 verify x11 keymap settings by compiling the given keymap. It
667 will spew out warnings if the compilation fails. This
668 requires libxkbcommon to be installed.
670 * When a coredump is collected a larger number of metadata
671 fields is now collected and included in the journal records
672 created for it. More specifically control group membership,
673 environment variables, memory maps, working directory,
674 chroot directory, /proc/$PID/status, and a list of open file
675 descriptors is now stored in the log entry.
677 * The udev hwdb now contains DPI information for mice. For
680 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
682 * All systemd programs that read standalone configuration
683 files in /etc now also support a corresponding series of
684 .conf.d configuration directories in /etc/, /run/,
685 /usr/local/lib/, /usr/lib/, and (if configured with
686 --enable-split-usr) /lib/. In particular, the following
687 configuration files now have corresponding configuration
688 directories: system.conf user.conf, logind.conf,
689 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
690 resolved.conf, timesyncd.conf, journal-remote.conf, and
691 journal-upload.conf. Note that distributions should use the
692 configuration directories in /usr/lib/; the directories in
693 /etc/ are reserved for the system administrator.
695 * systemd-rfkill will no longer take the rfkill device name
696 into account when storing rfkill state on disk, as the name
697 might be dynamically assigned and not stable. Instead, the
698 ID_PATH udev variable combined with the rfkill type (wlan,
699 bluetooth, ...) is used.
701 * A new service systemd-machine-id-commit.service has been
702 added. When used on systems where /etc is read-only during
703 boot, and /etc/machine-id is not initialized (but an empty
704 file), this service will copy the temporary machine ID
705 created as replacement into /etc after the system is fully
706 booted up. This is useful for systems that are freshly
707 installed with a non-initialized machine ID, but should get
708 a fixed machine ID for subsequent boots.
710 * networkd's .netdev files now provide a large set of
711 configuration parameters for VXLAN devices. Similar, the
712 bridge port cost parameter is now configurable in .network
713 files. There's also new support for configuring IP source
714 routing. networkd .link files gained support for a new
715 OriginalName= match that is useful to match against the
716 original interface name the kernel assigned. .network files
717 may include MTU= and MACAddress= fields for altering the MTU
718 and MAC address while being connected to a specific network
721 * The LUKS logic gained supported for configuring
722 UUID-specific key files. There's also new support for naming
723 LUKS device from the kernel command line, using the new
726 * Timer units may now be transiently created via the bus API
727 (this was previously already available for scope and service
728 units). In addition it is now possible to create multiple
729 transient units at the same time with a single bus call. The
730 "systemd-run" tool has been updated to make use of this for
731 running commands on a specified time, in at(1)-style.
733 * tmpfiles gained support for "t" lines, for assigning
734 extended attributes to files. Among other uses this may be
735 used to assign SMACK labels to files.
737 Contributions from: Alin Rauta, Alison Chaiken, Andrej
738 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
739 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
740 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
741 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
742 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
743 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
744 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
745 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
746 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
747 Michael Biebl, Michael Chapman, Michael Marineau, Michal
748 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
749 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
750 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
751 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
752 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
753 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
754 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
756 -- Berlin, 2014-12-10
760 * journalctl gained the new options -t/--identifier= to match
761 on the syslog identifier (aka "tag"), as well as --utc to
762 show log timestamps in the UTC timezone. journalctl now also
763 accepts -n/--lines=all to disable line capping in a pager.
765 * journalctl gained a new switch, --flush, that synchronously
766 flushes logs from /run/log/journal to /var/log/journal if
767 persistent storage is enabled. systemd-journal-flush.service
768 now waits until the operation is complete.
770 * Services can notify the manager before they start a reload
771 (by sending RELOADING=1) or shutdown (by sending
772 STOPPING=1). This allows the manager to track and show the
773 internal state of daemons and closes a race condition when
774 the process is still running but has closed its D-Bus
777 * Services with Type=oneshot do not have to have any ExecStart
780 * User units are now loaded also from
781 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
782 /run/systemd/user directory that was already previously
783 supported, but is under the control of the user.
785 * Job timeouts (i.e. time-outs on the time a job that is
786 queued stays in the run queue) can now optionally result in
787 immediate reboot or power-off actions (JobTimeoutAction= and
788 JobTimeoutRebootArgument=). This is useful on ".target"
789 units, to limit the maximum time a target remains
790 undispatched in the run queue, and to trigger an emergency
791 operation in such a case. This is now used by default to
792 turn off the system if boot-up (as defined by everything in
793 basic.target) hangs and does not complete for at least
794 15min. Also, if power-off or reboot hang for at least 30min
795 an immediate power-off/reboot operation is triggered. This
796 functionality is particularly useful to increase reliability
797 on embedded devices, but also on laptops which might
798 accidentally get powered on when carried in a backpack and
799 whose boot stays stuck in a hard disk encryption passphrase
802 * systemd-logind can be configured to also handle lid switch
803 events even when the machine is docked or multiple displays
804 are attached (HandleLidSwitchDocked= option).
806 * A helper binary and a service have been added which can be
807 used to resume from hibernation in the initramfs. A
808 generator will parse the resume= option on the kernel
809 command line to trigger resume.
811 * A user console daemon systemd-consoled has been
812 added. Currently, it is a preview, and will so far open a
813 single terminal on each session of the user marked as
814 Desktop=systemd-console.
816 * Route metrics can be specified for DHCP routes added by
819 * The SELinux context of socket-activated services can be set
820 from the information provided by the networking stack
821 (SELinuxContextFromNet= option).
823 * Userspace firmware loading support has been removed and
824 the minimum supported kernel version is thus bumped to 3.7.
826 * Timeout for udev workers has been increased from 1 to 3
827 minutes, but a warning will be printed after 1 minute to
828 help diagnose kernel modules that take a long time to load.
830 * Udev rules can now remove tags on devices with TAG-="foobar".
832 * systemd's readahead implementation has been removed. In many
833 circumstances it didn't give expected benefits even for
834 rotational disk drives and was becoming less relevant in the
835 age of SSDs. As none of the developers has been using
836 rotating media anymore, and nobody stepped up to actively
837 maintain this component of systemd it has now been removed.
839 * Swap units can use Options= to specify discard options.
840 Discard options specified for swaps in /etc/fstab are now
843 * Docker containers are now detected as a separate type of
846 * The Password Agent protocol gained support for queries where
847 the user input is shown, useful e.g. for user names.
848 systemd-ask-password gained a new --echo option to turn that
851 * The default sysctl.d/ snippets will now set:
853 net.core.default_qdisc = fq_codel
855 This selects Fair Queuing Controlled Delay as the default
856 queuing discipline for network interfaces. fq_codel helps
857 fight the network bufferbloat problem. It is believed to be
858 a good default with no tuning required for most workloads.
859 Downstream distributions may override this choice. On 10Gbit
860 servers that do not do forwarding, "fq" may perform better.
861 Systems without a good clocksource should use "pfifo_fast".
863 * If kdbus is enabled during build a new option BusPolicy= is
864 available for service units, that allows locking all service
865 processes into a stricter bus policy, in order to limit
866 access to various bus services, or even hide most of them
867 from the service's view entirely.
869 * networkctl will now show the .network and .link file
870 networkd has applied to a specific interface.
872 * sd-login gained a new API call sd_session_get_desktop() to
873 query which desktop environment has been selected for a
876 * UNIX utmp support is now compile-time optional to support
879 * systemctl gained two new commands "add-wants" and
880 "add-requires" for pulling in units from specific targets
883 * If the word "rescue" is specified on the kernel command line
884 the system will now boot into rescue mode (aka
885 rescue.target), which was previously available only by
886 specifying "1" or "systemd.unit=rescue.target" on the kernel
887 command line. This new kernel command line option nicely
888 mirrors the already existing "emergency" kernel command line
891 * New kernel command line options mount.usr=, mount.usrflags=,
892 mount.usrfstype= have been added that match root=, rootflags=,
893 rootfstype= but allow mounting a specific file system to
896 * The $NOTIFY_SOCKET is now also passed to control processes of
897 services, not only the main process.
899 * This version reenables support for fsck's -l switch. This
900 means at least version v2.25 of util-linux is required for
901 operation, otherwise dead-locks on device nodes may
902 occur. Again: you need to update util-linux to at least
903 v2.25 when updating systemd to v217.
905 * The "multi-seat-x" tool has been removed from systemd, as
906 its functionality has been integrated into X servers 1.16,
907 and the tool is hence redundant. It is recommended to update
908 display managers invoking this tool to simply invoke X
909 directly from now on, again.
911 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
912 message flag has been added for all of systemd's PolicyKit
913 authenticated method calls has been added. In particular
914 this now allows optional interactive authorization via
915 PolicyKit for many of PID1's privileged operations such as
916 unit file enabling and disabling.
918 * "udevadm hwdb --update" learnt a new switch "--usr" for
919 placing the rebuilt hardware database in /usr instead of
920 /etc. When used only hardware database entries stored in
921 /usr will be used, and any user database entries in /etc are
922 ignored. This functionality is useful for vendors to ship a
923 pre-built database on systems where local configuration is
924 unnecessary or unlikely.
926 * Calendar time specifications in .timer units now also
927 understand the strings "semi-annually", "quarterly" and
928 "minutely" as shortcuts (in addition to the preexisting
929 "anually", "hourly", ...).
931 * systemd-tmpfiles will now correctly create files in /dev
932 at boot which are marked for creation only at boot. It is
933 recommended to always create static device nodes with 'c!'
934 and 'b!', so that they are created only at boot and not
935 overwritten at runtime.
937 * When the watchdog logic is used for a service (WatchdogSec=)
938 and the watchdog timeout is hit the service will now be
939 terminated with SIGABRT (instead of just SIGTERM), in order
940 to make sure a proper coredump and backtrace is
941 generated. This ensures that hanging services will result in
942 similar coredump/backtrace behaviour as services that hit a
945 Contributions from: Andreas Henriksson, Andrei Borzenkov,
946 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
947 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
948 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
949 Herrmann, David Sommerseth, David Strauss, Emil Renner
950 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
951 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
952 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
953 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
954 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
955 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
956 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
957 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
958 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
959 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
960 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
961 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
962 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
963 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
964 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
965 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
968 -- Berlin, 2014-10-28
972 * timedated no longer reads NTP implementation unit names from
973 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
974 implementations should add a
976 Conflicts=systemd-timesyncd.service
978 to their unit files to take over and replace systemd's NTP
979 default functionality.
981 * systemd-sysusers gained a new line type "r" for configuring
982 which UID/GID ranges to allocate system users/groups
983 from. Lines of type "u" may now add an additional column
984 that specifies the home directory for the system user to be
985 created. Also, systemd-sysusers may now optionally read user
986 information from STDIN instead of a file. This is useful for
987 invoking it from RPM preinst scriptlets that need to create
988 users before the first RPM file is installed since these
989 files might need to be owned by them. A new
990 %sysusers_create_inline RPM macro has been introduced to do
991 just that. systemd-sysusers now updates the shadow files as
992 well as the user/group databases, which should enhance
993 compatibility with certain tools like grpck.
995 * A number of bus APIs of PID 1 now optionally consult
996 PolicyKit to permit access for otherwise unprivileged
997 clients under certain conditions. Note that this currently
998 doesn't support interactive authentication yet, but this is
999 expected to be added eventually, too.
1001 * /etc/machine-info now has new fields for configuring the
1002 deployment environment of the machine, as well as the
1003 location of the machine. hostnamectl has been updated with
1004 new command to update these fields.
1006 * systemd-timesyncd has been updated to automatically acquire
1007 NTP server information from systemd-networkd, which might
1008 have been discovered via DHCP.
1010 * systemd-resolved now includes a caching DNS stub resolver
1011 and a complete LLMNR name resolution implementation. A new
1012 NSS module "nss-resolve" has been added which can be used
1013 instead of glibc's own "nss-dns" to resolve hostnames via
1014 systemd-resolved. Hostnames, addresses and arbitrary RRs may
1015 be resolved via systemd-resolved D-Bus APIs. In contrast to
1016 the glibc internal resolver systemd-resolved is aware of
1017 multi-homed system, and keeps DNS server and caches separate
1018 and per-interface. Queries are sent simultaneously on all
1019 interfaces that have DNS servers configured, in order to
1020 properly handle VPNs and local LANs which might resolve
1021 separate sets of domain names. systemd-resolved may acquire
1022 DNS server information from systemd-networkd automatically,
1023 which in turn might have discovered them via DHCP. A tool
1024 "systemd-resolve-host" has been added that may be used to
1025 query the DNS logic in resolved. systemd-resolved implements
1026 IDNA and automatically uses IDNA or UTF-8 encoding depending
1027 on whether classic DNS or LLMNR is used as transport. In the
1028 next releases we intend to add a DNSSEC and mDNS/DNS-SD
1029 implementation to systemd-resolved.
1031 * A new NSS module nss-mymachines has been added, that
1032 automatically resolves the names of all local registered
1033 containers to their respective IP addresses.
1035 * A new client tool "networkctl" for systemd-networkd has been
1036 added. It currently is entirely passive and will query
1037 networking configuration from udev, rtnetlink and networkd,
1038 and present it to the user in a very friendly
1039 way. Eventually, we hope to extend it to become a full
1040 control utility for networkd.
1042 * .socket units gained a new DeferAcceptSec= setting that
1043 controls the kernels' TCP_DEFER_ACCEPT sockopt for
1044 TCP. Similar, support for controlling TCP keep-alive
1045 settings has been added (KeepAliveTimeSec=,
1046 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
1047 turning off Nagle's algorithm on TCP has been added
1050 * logind learned a new session type "web", for use in projects
1051 like Cockpit which register web clients as PAM sessions.
1053 * timer units with at least one OnCalendar= setting will now
1054 be started only after timer-sync.target has been
1055 reached. This way they will not elapse before the system
1056 clock has been corrected by a local NTP client or
1057 similar. This is particular useful on RTC-less embedded
1058 machines, that come up with an invalid system clock.
1060 * systemd-nspawn's --network-veth= switch should now result in
1061 stable MAC addresses for both the outer and the inner side
1064 * systemd-nspawn gained a new --volatile= switch for running
1065 container instances with /etc or /var unpopulated.
1067 * The kdbus client code has been updated to use the new Linux
1068 3.17 memfd subsystem instead of the old kdbus-specific one.
1070 * systemd-networkd's DHCP client and server now support
1071 FORCERENEW. There are also new configuration options to
1072 configure the vendor client identifier and broadcast mode
1075 * systemd will no longer inform the kernel about the current
1076 timezone, as this is necessarily incorrect and racy as the
1077 kernel has no understanding of DST and similar
1078 concepts. This hence means FAT timestamps will be always
1079 considered UTC, similar to what Android is already
1080 doing. Also, when the RTC is configured to the local time
1081 (rather than UTC) systemd will never synchronize back to it,
1082 as this might confuse Windows at a later boot.
1084 * systemd-analyze gained a new command "verify" for offline
1085 validation of unit files.
1087 * systemd-networkd gained support for a couple of additional
1088 settings for bonding networking setups. Also, the metric for
1089 statically configured routes may now be configured. For
1090 network interfaces where this is appropriate the peer IP
1091 address may now be configured.
1093 * systemd-networkd's DHCP client will no longer request
1094 broadcasting by default, as this tripped up some networks.
1095 For hardware where broadcast is required the feature should
1096 be switched back on using RequestBroadcast=yes.
1098 * systemd-networkd will now set up IPv4LL addresses (when
1099 enabled) even if DHCP is configured successfully.
1101 * udev will now default to respect network device names given
1102 by the kernel when the kernel indicates that these are
1103 predictable. This behavior can be tweaked by changing
1104 NamePolicy= in the relevant .link file.
1106 * A new library systemd-terminal has been added that
1107 implements full TTY stream parsing and rendering. This
1108 library is supposed to be used later on for implementing a
1109 full userspace VT subsystem, replacing the current kernel
1112 * A new tool systemd-journal-upload has been added to push
1113 journal data to a remote system running
1114 systemd-journal-remote.
1116 * journald will no longer forward all local data to another
1117 running syslog daemon. This change has been made because
1118 rsyslog (which appears to be the most commonly used syslog
1119 implementation these days) no longer makes use of this, and
1120 instead pulls the data out of the journal on its own. Since
1121 forwarding the messages to a non-existent syslog server is
1122 more expensive than we assumed we have now turned this
1123 off. If you run a syslog server that is not a recent rsyslog
1124 version, you have to turn this option on again
1125 (ForwardToSyslog= in journald.conf).
1127 * journald now optionally supports the LZ4 compressor for
1128 larger journal fields. This compressor should perform much
1129 better than XZ which was the previous default.
1131 * machinectl now shows the IP addresses of local containers,
1132 if it knows them, plus the interface name of the container.
1134 * A new tool "systemd-escape" has been added that makes it
1135 easy to escape strings to build unit names and similar.
1137 * sd_notify() messages may now include a new ERRNO= field
1138 which is parsed and collected by systemd and shown among the
1139 "systemctl status" output for a service.
1141 * A new component "systemd-firstboot" has been added that
1142 queries the most basic systemd information (timezone,
1143 hostname, root password) interactively on first
1144 boot. Alternatively it may also be used to provision these
1145 things offline on OS images installed into directories.
1147 * The default sysctl.d/ snippets will now set
1149 net.ipv4.conf.default.promote_secondaries=1
1151 This has the benefit of no flushing secondary IP addresses
1152 when primary addresses are removed.
1154 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
1155 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
1156 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
1157 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
1158 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
1159 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
1160 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
1161 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
1162 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
1163 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
1164 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
1165 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
1166 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
1167 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
1168 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
1170 -- Berlin, 2014-08-19
1174 * A new tool systemd-sysusers has been added. This tool
1175 creates system users and groups in /etc/passwd and
1176 /etc/group, based on static declarative system user/group
1177 definitions in /usr/lib/sysusers.d/. This is useful to
1178 enable factory resets and volatile systems that boot up with
1179 an empty /etc directory, and thus need system users and
1180 groups created during early boot. systemd now also ships
1181 with two default sysusers.d/ files for the most basic
1182 users and groups systemd and the core operating system
1185 * A new tmpfiles snippet has been added that rebuilds the
1186 essential files in /etc on boot, should they be missing.
1188 * A directive for ensuring automatic clean-up of
1189 /var/cache/man/ has been removed from the default
1190 configuration. This line should now be shipped by the man
1191 implementation. The necessary change has been made to the
1192 man-db implementation. Note that you need to update your man
1193 implementation to one that ships this line, otherwise no
1194 automatic clean-up of /var/cache/man will take place.
1196 * A new condition ConditionNeedsUpdate= has been added that
1197 may conditionalize services to only run when /etc or /var
1198 are "older" than the vendor operating system resources in
1199 /usr. This is useful for reconstructing or updating /etc
1200 after an offline update of /usr or a factory reset, on the
1201 next reboot. Services that want to run once after such an
1202 update or reset should use this condition and order
1203 themselves before the new systemd-update-done.service, which
1204 will mark the two directories as fully updated. A number of
1205 service files have been added making use of this, to rebuild
1206 the udev hardware database, the journald message catalog and
1207 dynamic loader cache (ldconfig). The systemd-sysusers tool
1208 described above also makes use of this now. With this in
1209 place it is now possible to start up a minimal operating
1210 system with /etc empty cleanly. For more information on the
1211 concepts involved see this recent blog story:
1213 http://0pointer.de/blog/projects/stateless.html
1215 * A new system group "input" has been introduced, and all
1216 input device nodes get this group assigned. This is useful
1217 for system-level software to get access to input devices. It
1218 complements what is already done for "audio" and "video".
1220 * systemd-networkd learnt minimal DHCPv4 server support in
1221 addition to the existing DHCPv4 client support. It also
1222 learnt DHCPv6 client and IPv6 Router Solicitation client
1223 support. The DHCPv4 client gained support for static routes
1224 passed in from the server. Note that the [DHCPv4] section
1225 known in older systemd-networkd versions has been renamed to
1226 [DHCP] and is now also used by the DHCPv6 client. Existing
1227 .network files using settings of this section should be
1228 updated, though compatibility is maintained. Optionally, the
1229 client hostname may now be sent to the DHCP server.
1231 * networkd gained support for vxlan virtual networks as well
1232 as tun/tap and dummy devices.
1234 * networkd gained support for automatic allocation of address
1235 ranges for interfaces from a system-wide pool of
1236 addresses. This is useful for dynamically managing a large
1237 number of interfaces with a single network configuration
1238 file. In particular this is useful to easily assign
1239 appropriate IP addresses to the veth links of a large number
1240 of nspawn instances.
1242 * RPM macros for processing sysusers, sysctl and binfmt
1243 drop-in snippets at package installation time have been
1246 * The /etc/os-release file should now be placed in
1247 /usr/lib/os-release. The old location is automatically
1248 created as symlink. /usr/lib is the more appropriate
1249 location of this file, since it shall actually describe the
1250 vendor operating system shipped in /usr, and not the
1251 configuration stored in /etc.
1253 * .mount units gained a new boolean SloppyOptions= setting
1254 that maps to mount(8)'s -s option which enables permissive
1255 parsing of unknown mount options.
1257 * tmpfiles learnt a new "L+" directive which creates a symlink
1258 but (unlike "L") deletes a pre-existing file first, should
1259 it already exist and not already be the correct
1260 symlink. Similar, "b+", "c+" and "p+" directives have been
1261 added as well, which create block and character devices, as
1262 well as fifos in the filesystem, possibly removing any
1263 pre-existing files of different types.
1265 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
1266 'argument' field (which so far specified the source to
1267 symlink/copy the files from) is now optional. If omitted the
1268 same file os copied from /usr/share/factory/ suffixed by the
1269 full destination path. This is useful for populating /etc
1270 with essential files, by copying them from vendor defaults
1271 shipped in /usr/share/factory/etc.
1273 * A new command "systemctl preset-all" has been added that
1274 applies the service preset settings to all installed unit
1275 files. A new switch --preset-mode= has been added that
1276 controls whether only enable or only disable operations
1279 * A new command "systemctl is-system-running" has been added
1280 that allows checking the overall state of the system, for
1281 example whether it is fully up and running.
1283 * When the system boots up with an empty /etc, the equivalent
1284 to "systemctl preset-all" is executed during early boot, to
1285 make sure all default services are enabled after a factory
1288 * systemd now contains a minimal preset file that enables the
1289 most basic services systemd ships by default.
1291 * Unit files' [Install] section gained a new DefaultInstance=
1292 field for defining the default instance to create if a
1293 template unit is enabled with no instance specified.
1295 * A new passive target cryptsetup-pre.target has been added
1296 that may be used by services that need to make they run and
1297 finish before the first LUKS cryptographic device is set up.
1299 * The /dev/loop-control and /dev/btrfs-control device nodes
1300 are now owned by the "disk" group by default, opening up
1301 access to this group.
1303 * systemd-coredump will now automatically generate a
1304 stack trace of all core dumps taking place on the system,
1305 based on elfutils' libdw library. This stack trace is logged
1308 * systemd-coredump may now optionally store coredumps directly
1309 on disk (in /var/lib/systemd/coredump, possibly compressed),
1310 instead of storing them unconditionally in the journal. This
1311 mode is the new default. A new configuration file
1312 /etc/systemd/coredump.conf has been added to configure this
1313 and other parameters of systemd-coredump.
1315 * coredumpctl gained a new "info" verb to show details about a
1316 specific coredump. A new switch "-1" has also been added
1317 that makes sure to only show information about the most
1318 recent entry instead of all entries. Also, as the tool is
1319 generally useful now the "systemd-" prefix of the binary
1320 name has been removed. Distributions that want to maintain
1321 compatibility with the old name should add a symlink from
1322 the old name to the new name.
1324 * journald's SplitMode= now defaults to "uid". This makes sure
1325 that unprivileged users can access their own coredumps with
1326 coredumpctl without restrictions.
1328 * New kernel command line options "systemd.wants=" (for
1329 pulling an additional unit during boot), "systemd.mask="
1330 (for masking a specific unit for the boot), and
1331 "systemd.debug-shell" (for enabling the debug shell on tty9)
1332 have been added. This is implemented in the new generator
1333 "systemd-debug-generator".
1335 * systemd-nspawn will now by default filter a couple of
1336 syscalls for containers, among them those required for
1337 kernel module loading, direct x86 IO port access, swap
1338 management, and kexec. Most importantly though
1339 open_by_handle_at() is now prohibited for containers,
1340 closing a hole similar to a recently discussed vulnerability
1341 in docker regarding access to files on file hierarchies the
1342 container should normally not have access to. Note that for
1343 nspawn we generally make no security claims anyway (and
1344 this is explicitly documented in the man page), so this is
1345 just a fix for one of the most obvious problems.
1347 * A new man page file-hierarchy(7) has been added that
1348 contains a minimized, modernized version of the file system
1349 layout systemd expects, similar in style to the FHS
1350 specification or hier(5). A new tool systemd-path(1) has
1351 been added to query many of these paths for the local
1354 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
1355 longer done. Since the directory now has a per-user size
1356 limit, and is cleaned on logout this appears unnecessary,
1357 in particular since this now brings the lifecycle of this
1358 directory closer in line with how IPC objects are handled.
1360 * systemd.pc now exports a number of additional directories,
1361 including $libdir (which is useful to identify the library
1362 path for the primary architecture of the system), and a
1363 couple of drop-in directories.
1365 * udev's predictable network interface names now use the dev_port
1366 sysfs attribute, introduced in linux 3.15 instead of dev_id to
1367 distinguish between ports of the same PCI function. dev_id should
1368 only be used for ports using the same HW address, hence the need
1371 * machined has been updated to export the OS version of a
1372 container (read from /etc/os-release and
1373 /usr/lib/os-release) on the bus. This is now shown in
1374 "machinectl status" for a machine.
1376 * A new service setting RestartForceExitStatus= has been
1377 added. If configured to a set of exit signals or process
1378 return values, the service will be restarted when the main
1379 daemon process exits with any of them, regardless of the
1382 * systemctl's -H switch for connecting to remote systemd
1383 machines has been extended so that it may be used to
1384 directly connect to a specific container on the
1385 host. "systemctl -H root@foobar:waldi" will now connect as
1386 user "root" to host "foobar", and then proceed directly to
1387 the container named "waldi". Note that currently you have to
1388 authenticate as user "root" for this to work, as entering
1389 containers is a privileged operation.
1391 Contributions from: Andreas Henriksson, Benjamin Steinwender,
1392 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
1393 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
1394 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
1395 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
1396 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
1397 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
1398 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
1399 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
1400 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
1401 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
1402 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
1404 -- Berlin, 2014-07-03
1408 * As an experimental feature, udev now tries to lock the
1409 disk device node (flock(LOCK_SH|LOCK_NB)) while it
1410 executes events for the disk or any of its partitions.
1411 Applications like partitioning programs can lock the
1412 disk device node (flock(LOCK_EX)) and claim temporary
1413 device ownership that way; udev will entirely skip all event
1414 handling for this disk and its partitions. If the disk
1415 was opened for writing, the close will trigger a partition
1416 table rescan in udev's "watch" facility, and if needed
1417 synthesize "change" events for the disk and all its partitions.
1418 This is now unconditionally enabled, and if it turns out to
1419 cause major problems, we might turn it on only for specific
1420 devices, or might need to disable it entirely. Device Mapper
1421 devices are excluded from this logic.
1423 * We temporarily dropped the "-l" switch for fsck invocations,
1424 since they collide with the flock() logic above. util-linux
1425 upstream has been changed already to avoid this conflict,
1426 and we will readd "-l" as soon as util-linux with this
1427 change has been released.
1429 * The dependency on libattr has been removed. Since a long
1430 time, the extended attribute calls have moved to glibc, and
1431 libattr is thus unnecessary.
1433 * Virtualization detection works without priviliges now. This
1434 means the systemd-detect-virt binary no longer requires
1435 CAP_SYS_PTRACE file capabilities, and our daemons can run
1436 with fewer privileges.
1438 * systemd-networkd now runs under its own "systemd-network"
1439 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
1440 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
1441 loses the ability to write to files owned by root this way.
1443 * Similar, systemd-resolved now runs under its own
1444 "systemd-resolve" user with no capabilities remaining.
1446 * Similar, systemd-bus-proxyd now runs under its own
1447 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
1449 * systemd-networkd gained support for setting up "veth"
1450 virtual ethernet devices for container connectivity, as well
1451 as GRE and VTI tunnels.
1453 * systemd-networkd will no longer automatically attempt to
1454 manually load kernel modules necessary for certain tunnel
1455 transports. Instead, it is assumed the kernel loads them
1456 automatically when required. This only works correctly on
1457 very new kernels. On older kernels, please consider adding
1458 the kernel modules to /etc/modules-load.d/ as a work-around.
1460 * The resolv.conf file systemd-resolved generates has been
1461 moved to /run/systemd/resolve/. If you have a symlink from
1462 /etc/resolv.conf, it might be necessary to correct it.
1464 * Two new service settings, ProtectHome= and ProtectSystem=,
1465 have been added. When enabled, they will make the user data
1466 (such as /home) inaccessible or read-only and the system
1467 (such as /usr) read-only, for specific services. This allows
1468 very light-weight per-service sandboxing to avoid
1469 modifications of user data or system files from
1470 services. These two new switches have been enabled for all
1471 of systemd's long-running services, where appropriate.
1473 * Socket units gained new SocketUser= and SocketGroup=
1474 settings to set the owner user and group of AF_UNIX sockets
1475 and FIFOs in the file system.
1477 * Socket units gained a new RemoveOnStop= setting. If enabled,
1478 all FIFOS and sockets in the file system will be removed
1479 when the specific socket unit is stopped.
1481 * Socket units gained a new Symlinks= setting. It takes a list
1482 of symlinks to create to file system sockets or FIFOs
1483 created by the specific Unix sockets. This is useful to
1484 manage symlinks to socket nodes with the same life-cycle as
1487 * The /dev/log socket and /dev/initctl FIFO have been moved to
1488 /run, and have been replaced by symlinks. This allows
1489 connecting to these facilities even if PrivateDevices=yes is
1490 used for a service (which makes /dev/log itself unavailable,
1491 but /run is left). This also has the benefit of ensuring
1492 that /dev only contains device nodes, directories and
1493 symlinks, and nothing else.
1495 * sd-daemon gained two new calls sd_pid_notify() and
1496 sd_pid_notifyf(). They are similar to sd_notify() and
1497 sd_notifyf(), but allow overriding of the source PID of
1498 notification messages if permissions permit this. This is
1499 useful to send notify messages on behalf of a different
1500 process (for example, the parent process). The
1501 systemd-notify tool has been updated to make use of this
1502 when sending messages (so that notification messages now
1503 originate from the shell script invoking systemd-notify and
1504 not the systemd-notify process itself. This should minimize
1505 a race where systemd fails to associate notification
1506 messages to services when the originating process already
1509 * A new "on-abnormal" setting for Restart= has been added. If
1510 set, it will result in automatic restarts on all "abnormal"
1511 reasons for a process to exit, which includes unclean
1512 signals, core dumps, timeouts and watchdog timeouts, but
1513 does not include clean and unclean exit codes or clean
1514 signals. Restart=on-abnormal is an alternative for
1515 Restart=on-failure for services that shall be able to
1516 terminate and avoid restarts on certain errors, by
1517 indicating so with an unclean exit code. Restart=on-failure
1518 or Restart=on-abnormal is now the recommended setting for
1519 all long-running services.
1521 * If the InaccessibleDirectories= service setting points to a
1522 mount point (or if there are any submounts contained within
1523 it), it is now attempted to completely unmount it, to make
1524 the file systems truly unavailable for the respective
1527 * The ReadOnlyDirectories= service setting and
1528 systemd-nspawn's --read-only parameter are now recursively
1529 applied to all submounts, too.
1531 * Mount units may now be created transiently via the bus APIs.
1533 * The support for SysV and LSB init scripts has been removed
1534 from the systemd daemon itself. Instead, it is now
1535 implemented as a generator that creates native systemd units
1536 from these scripts when needed. This enables us to remove a
1537 substantial amount of legacy code from PID 1, following the
1538 fact that many distributions only ship a very small number
1539 of LSB/SysV init scripts nowadays.
1541 * Privileged Xen (dom0) domains are not considered
1542 virtualization anymore by the virtualization detection
1543 logic. After all, they generally have unrestricted access to
1544 the hardware and usually are used to manage the unprivileged
1547 * systemd-tmpfiles gained a new "C" line type, for copying
1548 files or entire directories.
1550 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
1551 lines. So far, they have been non-globbing versions of the
1552 latter, and have thus been redundant. In future, it is
1553 recommended to only use "z". "m" has hence been removed
1554 from the documentation, even though it stays supported.
1556 * A tmpfiles snippet to recreate the most basic structure in
1557 /var has been added. This is enough to create the /var/run →
1558 /run symlink and create a couple of structural
1559 directories. This allows systems to boot up with an empty or
1560 volatile /var. Of course, while with this change, the core OS
1561 now is capable with dealing with a volatile /var, not all
1562 user services are ready for it. However, we hope that sooner
1563 or later, many service daemons will be changed upstream so
1564 that they are able to automatically create their necessary
1565 directories in /var at boot, should they be missing. This is
1566 the first step to allow state-less systems that only require
1567 the vendor image for /usr to boot.
1569 * systemd-nspawn has gained a new --tmpfs= switch to mount an
1570 empty tmpfs instance to a specific directory. This is
1571 particularly useful for making use of the automatic
1572 reconstruction of /var (see above), by passing --tmpfs=/var.
1574 * Access modes specified in tmpfiles snippets may now be
1575 prefixed with "~", which indicates that they shall be masked
1576 by whether the existing file or directory is currently
1577 writable, readable or executable at all. Also, if specified,
1578 the sgid/suid/sticky bits will be masked for all
1581 * A new passive target unit "network-pre.target" has been
1582 added which is useful for services that shall run before any
1583 network is configured, for example firewall scripts.
1585 * The "floppy" group that previously owned the /dev/fd*
1586 devices is no longer used. The "disk" group is now used
1587 instead. Distributions should probably deprecate usage of
1590 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
1591 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
1592 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
1593 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
1594 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
1595 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
1598 -- Berlin, 2014-06-11
1602 * A new "systemd-timesyncd" daemon has been added for
1603 synchronizing the system clock across the network. It
1604 implements an SNTP client. In contrast to NTP
1605 implementations such as chrony or the NTP reference server,
1606 this only implements a client side, and does not bother with
1607 the full NTP complexity, focusing only on querying time from
1608 one remote server and synchronizing the local clock to
1609 it. Unless you intend to serve NTP to networked clients or
1610 want to connect to local hardware clocks, this simple NTP
1611 client should be more than appropriate for most
1612 installations. The daemon runs with minimal privileges, and
1613 has been hooked up with networkd to only operate when
1614 network connectivity is available. The daemon saves the
1615 current clock to disk every time a new NTP sync has been
1616 acquired, and uses this to possibly correct the system clock
1617 early at bootup, in order to accommodate for systems that
1618 lack an RTC such as the Raspberry Pi and embedded devices,
1619 and to make sure that time monotonically progresses on these
1620 systems, even if it is not always correct. To make use of
1621 this daemon, a new system user and group "systemd-timesync"
1622 needs to be created on installation of systemd.
1624 * The queue "seqnum" interface of libudev has been disabled, as
1625 it was generally incompatible with device namespacing as
1626 sequence numbers of devices go "missing" if the devices are
1627 part of a different namespace.
1629 * "systemctl list-timers" and "systemctl list-sockets" gained
1630 a --recursive switch for showing units of these types also
1631 for all local containers, similar in style to the already
1632 supported --recursive switch for "systemctl list-units".
1634 * A new RebootArgument= setting has been added for service
1635 units, which may be used to specify a kernel reboot argument
1636 to use when triggering reboots with StartLimitAction=.
1638 * A new FailureAction= setting has been added for service
1639 units which may be used to specify an operation to trigger
1640 when a service fails. This works similarly to
1641 StartLimitAction=, but unlike it, controls what is done
1642 immediately rather than only after several attempts to
1643 restart the service in question.
1645 * hostnamed got updated to also expose the kernel name,
1646 release, and version on the bus. This is useful for
1647 executing commands like hostnamectl with the -H switch.
1648 systemd-analyze makes use of this to properly display
1649 details when running non-locally.
1651 * The bootchart tool can now show cgroup information in the
1652 graphs it generates.
1654 * The CFS CPU quota cgroup attribute is now exposed for
1655 services. The new CPUQuota= switch has been added for this
1656 which takes a percentage value. Setting this will have the
1657 result that a service may never get more CPU time than the
1658 specified percentage, even if the machine is otherwise idle.
1660 * systemd-networkd learned IPIP and SIT tunnel support.
1662 * LSB init scripts exposing a dependency on $network will now
1663 get a dependency on network-online.target rather than simply
1664 network.target. This should bring LSB handling closer to
1665 what it was on SysV systems.
1667 * A new fsck.repair= kernel option has been added to control
1668 how fsck shall deal with unclean file systems at boot.
1670 * The (.ini) configuration file parser will now silently
1671 ignore sections whose name begins with "X-". This may be
1672 used to maintain application-specific extension sections in unit
1675 * machined gained a new API to query the IP addresses of
1676 registered containers. "machinectl status" has been updated
1677 to show these addresses in its output.
1679 * A new call sd_uid_get_display() has been added to the
1680 sd-login APIs for querying the "primary" session of a
1681 user. The "primary" session of the user is elected from the
1682 user's sessions and generally a graphical session is
1683 preferred over a text one.
1685 * A minimal systemd-resolved daemon has been added. It
1686 currently simply acts as a companion to systemd-networkd and
1687 manages resolv.conf based on per-interface DNS
1688 configuration, possibly supplied via DHCP. In the long run
1689 we hope to extend this into a local DNSSEC enabled DNS and
1692 * The systemd-networkd-wait-online tool is now enabled by
1693 default. It will delay network-online.target until a network
1694 connection has been configured. The tool primarily integrates
1695 with networkd, but will also make a best effort to make sense
1696 of network configuration performed in some other way.
1698 * Two new service options StartupCPUShares= and
1699 StartupBlockIOWeight= have been added that work similarly to
1700 CPUShares= and BlockIOWeight= however only apply during
1701 system startup. This is useful to prioritize certain services
1702 differently during bootup than during normal runtime.
1704 * hostnamed has been changed to prefer the statically
1705 configured hostname in /etc/hostname (unless set to
1706 'localhost' or empty) over any dynamic one supplied by
1707 dhcp. With this change, the rules for picking the hostname
1708 match more closely the rules of other configuration settings
1709 where the local administrator's configuration in /etc always
1710 overrides any other settings.
1712 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
1713 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
1714 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
1715 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
1716 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
1717 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
1718 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
1719 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
1720 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
1721 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
1722 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
1723 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
1724 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
1725 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
1726 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
1727 Lindskog, WaLyong Cho, Will Woods, Zbigniew
1730 -- Beijing, 2014-05-28
1734 * When restoring the screen brightness at boot, stay away from
1735 the darkest setting or from the lowest 5% of the available
1736 range, depending on which is the larger value of both. This
1737 should effectively protect the user from rebooting into a
1738 black screen, should the brightness have been set to minimum
1741 * sd-login gained a new sd_machine_get_class() call to
1742 determine the class ("vm" or "container") of a machine
1743 registered with machined.
1745 * sd-login gained new calls
1746 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
1747 to query the identity of the peer of a local AF_UNIX
1748 connection. They operate similarly to their sd_pid_get_xyz()
1751 * PID 1 will now maintain a system-wide system state engine
1752 with the states "starting", "running", "degraded",
1753 "maintenance", "stopping". These states are bound to system
1754 startup, normal runtime, runtime with at least one failed
1755 service, rescue/emergency mode and system shutdown. This
1756 state is shown in the "systemctl status" output when no unit
1757 name is passed. It is useful to determine system state, in
1758 particularly when doing so for many systems or containers at
1761 * A new command "list-machines" has been added to "systemctl"
1762 that lists all local OS containers and shows their system
1763 state (see above), if systemd runs inside of them.
1765 * systemctl gained a new "-r" switch to recursively enumerate
1766 units on all local containers, when used with the
1767 "list-unit" command (which is the default one that is
1768 executed when no parameters are specified).
1770 * The GPT automatic partition discovery logic will now honour
1771 two GPT partition flags: one may be set on a partition to
1772 cause it to be mounted read-only, and the other may be set
1773 on a partition to ignore it during automatic discovery.
1775 * Two new GPT type UUIDs have been added for automatic root
1776 partition discovery, for 32-bit and 64-bit ARM. This is not
1777 particularly useful for discovering the root directory on
1778 these architectures during bare-metal boots (since UEFI is
1779 not common there), but still very useful to allow booting of
1780 ARM disk images in nspawn with the -i option.
1782 * MAC addresses of interfaces created with nspawn's
1783 --network-interface= switch will now be generated from the
1784 machine name, and thus be stable between multiple invocations
1787 * logind will now automatically remove all IPC objects owned
1788 by a user if she or he fully logs out. This makes sure that
1789 users who are logged out cannot continue to consume IPC
1790 resources. This covers SysV memory, semaphores and message
1791 queues as well as POSIX shared memory and message
1792 queues. Traditionally, SysV and POSIX IPC had no life-cycle
1793 limits. With this functionality, that is corrected. This may
1794 be turned off by using the RemoveIPC= switch of logind.conf.
1796 * The systemd-machine-id-setup and tmpfiles tools gained a
1797 --root= switch to operate on a specific root directory,
1800 * journald can now forward logged messages to the TTYs of all
1801 logged in users ("wall"). This is the default for all
1802 emergency messages now.
1804 * A new tool systemd-journal-remote has been added to stream
1805 journal log messages across the network.
1807 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
1808 controller trees are mounted into it. Note that the
1809 directories mounted beneath it are not read-only. This is a
1810 security measure and is particularly useful because glibc
1811 actually includes a search logic to pick any tmpfs it can
1812 find to implement shm_open() if /dev/shm is not available
1813 (which it might very well be in namespaced setups).
1815 * machinectl gained a new "poweroff" command to cleanly power
1816 down a local OS container.
1818 * The PrivateDevices= unit file setting will now also drop the
1819 CAP_MKNOD capability from the capability bound set, and
1820 imply DevicePolicy=closed.
1822 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
1823 comprehensively on all long-running systemd services where
1824 this is appropriate.
1826 * systemd-udevd will now run in a disassociated mount
1827 namespace. To mount directories from udev rules, make sure to
1828 pull in mount units via SYSTEMD_WANTS properties.
1830 * The kdbus support gained support for uploading policy into
1831 the kernel. sd-bus gained support for creating "monitoring"
1832 connections that can eavesdrop into all bus communication
1833 for debugging purposes.
1835 * Timestamps may now be specified in seconds since the UNIX
1836 epoch Jan 1st, 1970 by specifying "@" followed by the value
1839 * Native tcpwrap support in systemd has been removed. tcpwrap
1840 is old code, not really maintained anymore and has serious
1841 shortcomings, and better options such as firewalls
1842 exist. For setups that require tcpwrap usage, please
1843 consider invoking your socket-activated service via tcpd,
1844 like on traditional inetd.
1846 * A new system.conf configuration option
1847 DefaultTimerAccuracySec= has been added that controls the
1848 default AccuracySec= setting of .timer units.
1850 * Timer units gained a new WakeSystem= switch. If enabled,
1851 timers configured this way will cause the system to resume
1852 from system suspend (if the system supports that, which most
1855 * Timer units gained a new Persistent= switch. If enabled,
1856 timers configured this way will save to disk when they have
1857 been last triggered. This information is then used on next
1858 reboot to possible execute overdue timer events, that
1859 could not take place because the system was powered off.
1860 This enables simple anacron-like behaviour for timer units.
1862 * systemctl's "list-timers" will now also list the time a
1863 timer unit was last triggered in addition to the next time
1864 it will be triggered.
1866 * systemd-networkd will now assign predictable IPv4LL
1867 addresses to its local interfaces.
1869 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
1870 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
1871 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
1872 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
1873 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
1874 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
1875 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
1876 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
1879 -- Berlin, 2014-03-25
1883 * A new unit file setting RestrictAddressFamilies= has been
1884 added to restrict which socket address families unit
1885 processes gain access to. This takes address family names
1886 like "AF_INET" or "AF_UNIX", and is useful to minimize the
1887 attack surface of services via exotic protocol stacks. This
1888 is built on seccomp system call filters.
1890 * Two new unit file settings RuntimeDirectory= and
1891 RuntimeDirectoryMode= have been added that may be used to
1892 manage a per-daemon runtime directories below /run. This is
1893 an alternative for setting up directory permissions with
1894 tmpfiles snippets, and has the advantage that the runtime
1895 directory's lifetime is bound to the daemon runtime and that
1896 the daemon starts up with an empty directory each time. This
1897 is particularly useful when writing services that drop
1898 privileges using the User= or Group= setting.
1900 * The DeviceAllow= unit setting now supports globbing for
1901 matching against device group names.
1903 * The systemd configuration file system.conf gained new
1904 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
1905 DefaultMemoryAccounting= to globally turn on/off accounting
1906 for specific resources (cgroups) for all units. These
1907 settings may still be overridden individually in each unit
1910 * systemd-gpt-auto-generator is now able to discover /srv and
1911 root partitions in addition to /home and swap partitions. It
1912 also supports LUKS-encrypted partitions now. With this in
1913 place, automatic discovery of partitions to mount following
1914 the Discoverable Partitions Specification
1915 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
1916 is now a lot more complete. This allows booting without
1917 /etc/fstab and without root= on the kernel command line on
1918 systems prepared appropriately.
1920 * systemd-nspawn gained a new --image= switch which allows
1921 booting up disk images and Linux installations on any block
1922 device that follow the Discoverable Partitions Specification
1923 (see above). This means that installations made with
1924 appropriately updated installers may now be started and
1925 deployed using container managers, completely
1926 unmodified. (We hope that libvirt-lxc will add support for
1927 this feature soon, too.)
1929 * systemd-nspawn gained a new --network-macvlan= setting to
1930 set up a private macvlan interface for the
1931 container. Similarly, systemd-networkd gained a new
1932 Kind=macvlan setting in .netdev files.
1934 * systemd-networkd now supports configuring local addresses
1937 * A new tool systemd-network-wait-online has been added to
1938 synchronously wait for network connectivity using
1941 * The sd-bus.h bus API gained a new sd_bus_track object for
1942 tracking the life-cycle of bus peers. Note that sd-bus.h is
1943 still not a public API though (unless you specify
1944 --enable-kdbus on the configure command line, which however
1945 voids your warranty and you get no API stability guarantee).
1947 * The $XDG_RUNTIME_DIR runtime directories for each user are
1948 now individual tmpfs instances, which has the benefit of
1949 introducing separate pools for each user, with individual
1950 size limits, and thus making sure that unprivileged clients
1951 can no longer negatively impact the system or other users by
1952 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
1953 RuntimeDirectorySize= has been introduced that allows
1954 controlling the default size limit for all users. It
1955 defaults to 10% of the available physical memory. This is no
1956 replacement for quotas on tmpfs though (which the kernel
1957 still does not support), as /dev/shm and /tmp are still
1958 shared resources used by both the system and unprivileged
1961 * logind will now automatically turn off automatic suspending
1962 on laptop lid close when more than one display is
1963 connected. This was previously expected to be implemented
1964 individually in desktop environments (such as GNOME),
1965 however has been added to logind now, in order to fix a
1966 boot-time race where a desktop environment might not have
1967 been started yet and thus not been able to take an inhibitor
1968 lock at the time where logind already suspends the system
1969 due to a closed lid.
1971 * logind will now wait at least 30s after each system
1972 suspend/resume cycle, and 3min after system boot before
1973 suspending the system due to a closed laptop lid. This
1974 should give USB docking stations and similar enough time to
1975 be probed and configured after system resume and boot in
1976 order to then act as suspend blocker.
1978 * systemd-run gained a new --property= setting which allows
1979 initialization of resource control properties (and others)
1980 for the created scope or service unit. Example: "systemd-run
1981 --property=BlockIOWeight=10 updatedb" may be used to run
1982 updatedb at a low block IO scheduling weight.
1984 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
1985 now also work in --scope mode.
1987 * When systemd is compiled with kdbus support, basic support
1988 for enforced policies is now in place. (Note that enabling
1989 kdbus still voids your warranty and no API compatibility
1992 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
1993 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
1994 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
1995 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
1996 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
1997 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
1998 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
1999 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
2000 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
2001 Zbigniew Jędrzejewski-Szmek
2003 -- Berlin, 2014-03-12
2007 * systemd will now relabel /dev after loading the SMACK policy
2008 according to SMACK rules.
2010 * A new unit file option AppArmorProfile= has been added to
2011 set the AppArmor profile for the processes of a unit.
2013 * A new condition check ConditionArchitecture= has been added
2014 to conditionalize units based on the system architecture, as
2015 reported by uname()'s "machine" field.
2017 * systemd-networkd now supports matching on the system
2018 virtualization, architecture, kernel command line, host name
2021 * logind is now a lot more aggressive when suspending the
2022 machine due to a closed laptop lid. Instead of acting only
2023 on the lid close action, it will continuously watch the lid
2024 status and act on it. This is useful for laptops where the
2025 power button is on the outside of the chassis so that it can
2026 be reached without opening the lid (such as the Lenovo
2027 Yoga). On those machines, logind will now immediately
2028 re-suspend the machine if the power button has been
2029 accidentally pressed while the laptop was suspended and in a
2030 backpack or similar.
2032 * logind will now watch SW_DOCK switches and inhibit reaction
2033 to the lid switch if it is pressed. This means that logind
2034 will not suspend the machine anymore if the lid is closed
2035 and the system is docked, if the laptop supports SW_DOCK
2036 notifications via the input layer. Note that ACPI docking
2037 stations do not generate this currently. Also note that this
2038 logic is usually not fully sufficient and Desktop
2039 Environments should take a lid switch inhibitor lock when an
2040 external display is connected, as systemd will not watch
2043 * nspawn will now make use of the devices cgroup controller by
2044 default, and only permit creation of and access to the usual
2045 API device nodes like /dev/null or /dev/random, as well as
2046 access to (but not creation of) the pty devices.
2048 * We will now ship a default .network file for
2049 systemd-networkd that automatically configures DHCP for
2050 network interfaces created by nspawn's --network-veth or
2051 --network-bridge= switches.
2053 * systemd will now understand the usual M, K, G, T suffixes
2054 according to SI conventions (i.e. to the base 1000) when
2055 referring to throughput and hardware metrics. It will stay
2056 with IEC conventions (i.e. to the base 1024) for software
2057 metrics, according to what is customary according to
2058 Wikipedia. We explicitly document which base applies for
2059 each configuration option.
2061 * The DeviceAllow= setting in unit files now supports a syntax
2062 to whitelist an entire group of devices node majors at once,
2063 based on the /proc/devices listing. For example, with the
2064 string "char-pts", it is now possible to whitelist all
2065 current and future pseudo-TTYs at once.
2067 * sd-event learned a new "post" event source. Event sources of
2068 this type are triggered by the dispatching of any event
2069 source of a type that is not "post". This is useful for
2070 implementing clean-up and check event sources that are
2071 triggered by other work being done in the program.
2073 * systemd-networkd is no longer statically enabled, but uses
2074 the usual [Install] sections so that it can be
2075 enabled/disabled using systemctl. It still is enabled by
2078 * When creating a veth interface pair with systemd-nspawn, the
2079 host side will now be prefixed with "vb-" if
2080 --network-bridge= is used, and with "ve-" if --network-veth
2081 is used. This way, it is easy to distinguish these cases on
2082 the host, for example to apply different configuration to
2083 them with systemd-networkd.
2085 * The compatibility libraries for libsystemd-journal.so,
2086 libsystem-id128.so, libsystemd-login.so and
2087 libsystemd-daemon.so do not make use of IFUNC
2088 anymore. Instead, we now build libsystemd.so multiple times
2089 under these alternative names. This means that the footprint
2090 is drastically increased, but given that these are
2091 transitional compatibility libraries, this should not matter
2092 much. This change has been made necessary to support the ARM
2093 platform for these compatibility libraries, as the ARM
2094 toolchain is not really at the same level as the toolchain
2095 for other architectures like x86 and does not support
2096 IFUNC. Please make sure to use --enable-compat-libs only
2097 during a transitional period!
2099 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
2100 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
2101 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
2102 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
2103 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
2104 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
2105 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
2106 Zbigniew Jędrzejewski-Szmek
2108 -- Berlin, 2014-02-24
2112 * A new component "systemd-networkd" has been added that can
2113 be used to configure local network interfaces statically or
2114 via DHCP. It is capable of bringing up bridges, VLANs, and
2115 bonding. Currently, no hook-ups for interactive network
2116 configuration are provided. Use this for your initrd,
2117 container, embedded, or server setup if you need a simple,
2118 yet powerful, network configuration solution. This
2119 configuration subsystem is quite nifty, as it allows wildcard
2120 hotplug matching in interfaces. For example, with a single
2121 configuration snippet, you can configure that all Ethernet
2122 interfaces showing up are automatically added to a bridge,
2123 or similar. It supports link-sensing and more.
2125 * A new tool "systemd-socket-proxyd" has been added which can
2126 act as a bidirectional proxy for TCP sockets. This is
2127 useful for adding socket activation support to services that
2128 do not actually support socket activation, including virtual
2129 machines and the like.
2131 * Add a new tool to save/restore rfkill state on
2134 * Save/restore state of keyboard backlights in addition to
2135 display backlights on shutdown/boot.
2137 * udev learned a new SECLABEL{} construct to label device
2138 nodes with a specific security label when they appear. For
2139 now, only SECLABEL{selinux} is supported, but the syntax is
2140 prepared for additional security frameworks.
2142 * udev gained a new scheme to configure link-level attributes
2143 from files in /etc/systemd/network/*.link. These files can
2144 match against MAC address, device path, driver name and type,
2145 and will apply attributes like the naming policy, link speed,
2146 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
2147 address assignment policy (randomized, ...).
2149 * The configuration of network interface naming rules for
2150 "permanent interface names" has changed: a new NamePolicy=
2151 setting in the [Link] section of .link files determines the
2152 priority of possible naming schemes (onboard, slot, mac,
2153 path). The default value of this setting is determined by
2154 /usr/lib/net/links/99-default.link. Old
2155 80-net-name-slot.rules udev configuration file has been
2156 removed, so local configuration overriding this file should
2157 be adapated to override 99-default.link instead.
2159 * When the User= switch is used in a unit file, also
2160 initialize $SHELL= based on the user database entry.
2162 * systemd no longer depends on libdbus. All communication is
2163 now done with sd-bus, systemd's low-level bus library
2166 * kdbus support has been added to PID 1 itself. When kdbus is
2167 enabled, this causes PID 1 to set up the system bus and
2168 enable support for a new ".busname" unit type that
2169 encapsulates bus name activation on kdbus. It works a little
2170 bit like ".socket" units, except for bus names. A new
2171 generator has been added that converts classic dbus1 service
2172 activation files automatically into native systemd .busname
2175 * sd-bus: add a light-weight vtable implementation that allows
2176 defining objects on the bus with a simple static const
2177 vtable array of its methods, signals and properties.
2179 * systemd will not generate or install static dbus
2180 introspection data anymore to /usr/share/dbus-1/interfaces,
2181 as the precise format of these files is unclear, and
2182 nothing makes use of it.
2184 * A proxy daemon is now provided to proxy clients connecting
2185 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
2186 compatibility with classic D-Bus.
2188 * A bus driver implementation has been added that supports the
2189 classic D-Bus bus driver calls on kdbus, also for
2190 compatibility purposes.
2192 * A new API "sd-event.h" has been added that implements a
2193 minimal event loop API built around epoll. It provides a
2194 couple of features that direct epoll usage is lacking:
2195 prioritization of events, scales to large numbers of timer
2196 events, per-event timer slack (accuracy), system-wide
2197 coalescing of timer events, exit handlers, watchdog
2198 supervision support using systemd's sd_notify() API, child
2201 * A new API "sd-rntl.h" has been added that provides an API
2202 around the route netlink interface of the kernel, similar in
2203 style to "sd-bus.h".
2205 * A new API "sd-dhcp-client.h" has been added that provides a
2206 small DHCPv4 client-side implementation. This is used by
2209 * There is a new kernel command line option
2210 "systemd.restore_state=0|1". When set to "0", none of the
2211 systemd tools will restore saved runtime state to hardware
2212 devices. More specifically, the rfkill and backlight states
2215 * The FsckPassNo= compatibility option in mount/service units
2216 has been removed. The fstab generator will now add the
2217 necessary dependencies automatically, and does not require
2218 PID1's support for that anymore.
2220 * journalctl gained a new switch, --list-boots, that lists
2221 recent boots with their times and boot IDs.
2223 * The various tools like systemctl, loginctl, timedatectl,
2224 busctl, systemd-run, ... have gained a new switch "-M" to
2225 connect to a specific, local OS container (as direct
2226 connection, without requiring SSH). This works on any
2227 container that is registered with machined, such as those
2228 created by libvirt-lxc or nspawn.
2230 * systemd-run and systemd-analyze also gained support for "-H"
2231 to connect to remote hosts via SSH. This is particularly
2232 useful for systemd-run because it enables queuing of jobs
2233 onto remote systems.
2235 * machinectl gained a new command "login" to open a getty
2236 login in any local container. This works with any container
2237 that is registered with machined (such as those created by
2238 libvirt-lxc or nspawn), and which runs systemd inside.
2240 * machinectl gained a new "reboot" command that may be used to
2241 trigger a reboot on a specific container that is registered
2242 with machined. This works on any container that runs an init
2243 system of some kind.
2245 * systemctl gained a new "list-timers" command to print a nice
2246 listing of installed timer units with the times they elapse
2249 * Alternative reboot() parameters may now be specified on the
2250 "systemctl reboot" command line and are passed to the
2251 reboot() system call.
2253 * systemctl gained a new --job-mode= switch to configure the
2254 mode to queue a job with. This is a more generic version of
2255 --fail, --irreversible, and --ignore-dependencies, which are
2256 still available but not advertised anymore.
2258 * /etc/systemd/system.conf gained new settings to configure
2259 various default timeouts of units, as well as the default
2260 start limit interval and burst. These may still be overridden
2263 * PID1 will now export on the bus profile data of the security
2264 policy upload process (such as the SELinux policy upload to
2267 * journald: when forwarding logs to the console, include
2268 timestamps (following the setting in
2269 /sys/module/printk/parameters/time).
2271 * OnCalendar= in timer units now understands the special
2272 strings "yearly" and "annually". (Both are equivalent)
2274 * The accuracy of timer units is now configurable with the new
2275 AccuracySec= setting. It defaults to 1min.
2277 * A new dependency type JoinsNamespaceOf= has been added that
2278 allows running two services within the same /tmp and network
2279 namespace, if PrivateNetwork= or PrivateTmp= are used.
2281 * A new command "cat" has been added to systemctl. It outputs
2282 the original unit file of a unit, and concatenates the
2283 contents of additional "drop-in" unit file snippets, so that
2284 the full configuration is shown.
2286 * systemctl now supports globbing on the various "list-xyz"
2287 commands, like "list-units" or "list-sockets", as well as on
2288 those commands which take multiple unit names.
2290 * journalctl's --unit= switch gained support for globbing.
2292 * All systemd daemons now make use of the watchdog logic so
2293 that systemd automatically notices when they hang.
2295 * If the $container_ttys environment variable is set,
2296 getty-generator will automatically spawn a getty for each
2297 listed tty. This is useful for container managers to request
2298 login gettys to be spawned on as many ttys as needed.
2300 * %h, %s, %U specifier support is not available anymore when
2301 used in unit files for PID 1. This is because NSS calls are
2302 not safe from PID 1. They stay available for --user
2303 instances of systemd, and as special case for the root user.
2305 * loginctl gained a new "--no-legend" switch to turn off output
2308 * The "sd-login.h" API gained three new calls:
2309 sd_session_is_remote(), sd_session_get_remote_user(),
2310 sd_session_get_remote_host() to query information about
2313 * The udev hardware database now also carries vendor/product
2314 information of SDIO devices.
2316 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
2317 determine whether watchdog notifications are requested by
2320 * Socket-activated per-connection services now include a
2321 short description of the connection parameters in the
2324 * tmpfiles gained a new "--boot" option. When this is not used,
2325 only lines where the command character is not suffixed with
2326 "!" are executed. When this option is specified, those
2327 options are executed too. This partitions tmpfiles
2328 directives into those that can be safely executed at any
2329 time, and those which should be run only at boot (for
2330 example, a line that creates /run/nologin).
2332 * A new API "sd-resolve.h" has been added which provides a simple
2333 asynchronous wrapper around glibc NSS host name resolution
2334 calls, such as getaddrinfo(). In contrast to glibc's
2335 getaddrinfo_a(), it does not use signals. In contrast to most
2336 other asynchronous name resolution libraries, this one does
2337 not reimplement DNS, but reuses NSS, so that alternate
2338 host name resolution systems continue to work, such as mDNS,
2339 LDAP, etc. This API is based on libasyncns, but it has been
2340 cleaned up for inclusion in systemd.
2342 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
2343 "sd-daemon.h" are no longer found in individual libraries
2344 libsystemd-journal.so, libsystemd-login.so,
2345 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
2346 merged them into a single library, libsystemd.so, which
2347 provides all symbols. The reason for this is cyclic
2348 dependencies, as these libraries tend to use each other's
2349 symbols. So far, we have managed to workaround that by linking
2350 a copy of a good part of our code into each of these
2351 libraries again and again, which, however, makes certain
2352 things hard to do, like sharing static variables. Also, it
2353 substantially increases footprint. With this change, there
2354 is only one library for the basic APIs systemd
2355 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
2356 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
2357 library as well, however are subject to the --enable-kdbus
2358 switch (see below). Note that "sd-dhcp-client.h" is not part
2359 of this library (this is because it only consumes, never
2360 provides, services of/to other APIs). To make the transition
2361 easy from the separate libraries to the unified one, we
2362 provide the --enable-compat-libs compile-time switch which
2363 will generate stub libraries that are compatible with the
2364 old ones but redirect all calls to the new one.
2366 * All of the kdbus logic and the new APIs "sd-bus.h",
2367 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
2368 and "sd-utf8.h" are compile-time optional via the
2369 "--enable-kdbus" switch, and they are not compiled in by
2370 default. To make use of kdbus, you have to explicitly enable
2371 the switch. Note however, that neither the kernel nor the
2372 userspace API for all of this is considered stable yet. We
2373 want to maintain the freedom to still change the APIs for
2374 now. By specifying this build-time switch, you acknowledge
2375 that you are aware of the instability of the current
2378 * Also, note that while kdbus is pretty much complete,
2379 it lacks one thing: proper policy support. This means you
2380 can build a fully working system with all features; however,
2381 it will be highly insecure. Policy support will be added in
2382 one of the next releases, at the same time that we will
2383 declare the APIs stable.
2385 * When the kernel command line argument "kdbus" is specified,
2386 systemd will automatically load the kdbus.ko kernel module. At
2387 this stage of development, it is only useful for testing kdbus
2388 and should not be used in production. Note: if "--enable-kdbus"
2389 is specified, and the kdbus.ko kernel module is available, and
2390 "kdbus" is added to the kernel command line, the entire system
2391 runs with kdbus instead of dbus-daemon, with the above mentioned
2392 problem of missing the system policy enforcement. Also a future
2393 version of kdbus.ko or a newer systemd will not be compatible with
2394 each other, and will unlikely be able to boot the machine if only
2395 one of them is updated.
2397 * systemctl gained a new "import-environment" command which
2398 uploads the caller's environment (or parts thereof) into the
2399 service manager so that it is inherited by services started
2400 by the manager. This is useful to upload variables like
2401 $DISPLAY into the user service manager.
2403 * A new PrivateDevices= switch has been added to service units
2404 which allows running a service with a namespaced /dev
2405 directory that does not contain any device nodes for
2406 physical devices. More specifically, it only includes devices
2407 such as /dev/null, /dev/urandom, and /dev/zero which are API
2410 * logind has been extended to support behaviour like VT
2411 switching on seats that do not support a VT. This makes
2412 multi-session available on seats that are not the first seat
2413 (seat0), and on systems where kernel support for VTs has
2414 been disabled at compile-time.
2416 * If a process holds a delay lock for system sleep or shutdown
2417 and fails to release it in time, we will now log its
2418 identity. This makes it easier to identify processes that
2419 cause slow suspends or power-offs.
2421 * When parsing /etc/crypttab, support for a new key-slot=
2422 option as supported by Debian is added. It allows indicating
2423 which LUKS slot to use on disk, speeding up key loading.
2425 * The sd_journald_sendv() API call has been checked and
2426 officially declared to be async-signal-safe so that it may
2427 be invoked from signal handlers for logging purposes.
2429 * Boot-time status output is now enabled automatically after a
2430 short timeout if boot does not progress, in order to give
2431 the user an indication what she or he is waiting for.
2433 * The boot-time output has been improved to show how much time
2434 remains until jobs expire.
2436 * The KillMode= switch in service units gained a new possible
2437 value "mixed". If set, and the unit is shut down, then the
2438 initial SIGTERM signal is sent only to the main daemon
2439 process, while the following SIGKILL signal is sent to
2440 all remaining processes of the service.
2442 * When a scope unit is registered, a new property "Controller"
2443 may be set. If set to a valid bus name, systemd will send a
2444 RequestStop() signal to this name when it would like to shut
2445 down the scope. This may be used to hook manager logic into
2446 the shutdown logic of scope units. Also, scope units may now
2447 be put in a special "abandoned" state, in which case the
2448 manager process which created them takes no further
2449 responsibilities for it.
2451 * When reading unit files, systemd will now verify
2452 the access mode of these files, and warn about certain
2453 suspicious combinations. This has been added to make it
2454 easier to track down packaging bugs where unit files are
2455 marked executable or world-writable.
2457 * systemd-nspawn gained a new "--setenv=" switch to set
2458 container-wide environment variables. The similar option in
2459 systemd-activate was renamed from "--environment=" to
2460 "--setenv=" for consistency.
2462 * systemd-nspawn has been updated to create a new kdbus domain
2463 for each container that is invoked, thus allowing each
2464 container to have its own set of system and user buses,
2465 independent of the host.
2467 * systemd-nspawn gained a new --drop-capability= switch to run
2468 the container with less capabilities than the default. Both
2469 --drop-capability= and --capability= now take the special
2470 string "all" for dropping or keeping all capabilities.
2472 * systemd-nspawn gained new switches for executing containers
2473 with specific SELinux labels set.
2475 * systemd-nspawn gained a new --quiet switch to not generate
2476 any additional output but the container's own console
2479 * systemd-nspawn gained a new --share-system switch to run a
2480 container without PID namespacing enabled.
2482 * systemd-nspawn gained a new --register= switch to control
2483 whether the container is registered with systemd-machined or
2484 not. This is useful for containers that do not run full
2485 OS images, but only specific apps.
2487 * systemd-nspawn gained a new --keep-unit which may be used
2488 when invoked as the only program from a service unit, and
2489 results in registration of the unit service itself in
2490 systemd-machined, instead of a newly opened scope unit.
2492 * systemd-nspawn gained a new --network-interface= switch for
2493 moving arbitrary interfaces to the container. The new
2494 --network-veth switch creates a virtual Ethernet connection
2495 between host and container. The new --network-bridge=
2496 switch then allows assigning the host side of this virtual
2497 Ethernet connection to a bridge device.
2499 * systemd-nspawn gained a new --personality= switch for
2500 setting the kernel personality for the container. This is
2501 useful when running a 32-bit container on a 64-bit host. A
2502 similar option Personality= is now also available for service
2505 * logind will now also track a "Desktop" identifier for each
2506 session which encodes the desktop environment of it. This is
2507 useful for desktop environments that want to identify
2508 multiple running sessions of itself easily.
2510 * A new SELinuxContext= setting for service units has been
2511 added that allows setting a specific SELinux execution
2512 context for a service.
2514 * Most systemd client tools will now honour $SYSTEMD_LESS for
2515 settings of the "less" pager. By default, these tools will
2516 override $LESS to allow certain operations to work, such as
2517 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
2518 influence this logic.
2520 * systemd's "seccomp" hook-up has been changed to make use of
2521 the libseccomp library instead of using its own
2522 implementation. This has benefits for portability among
2525 * For usage together with SystemCallFilter=, a new
2526 SystemCallErrorNumber= setting has been introduced that
2527 allows configuration of a system error number to be returned
2528 on filtered system calls, instead of immediately killing the
2529 process. Also, SystemCallArchitectures= has been added to
2530 limit access to system calls of a particular architecture
2531 (in order to turn off support for unused secondary
2532 architectures). There is also a global
2533 SystemCallArchitectures= setting in system.conf now to turn
2534 off support for non-native system calls system-wide.
2536 * systemd requires a kernel with a working name_to_handle_at(),
2537 please see the kernel config requirements in the README file.
2539 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
2540 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
2541 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
2542 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
2543 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
2544 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
2545 Elia Pinto, Florian Weimer, George McCollister, Goffredo
2546 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
2547 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
2548 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
2549 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
2550 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
2551 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
2552 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
2553 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
2554 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
2555 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
2556 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
2557 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
2558 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
2559 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
2560 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
2561 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
2562 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
2564 -- Berlin, 2014-02-20
2568 * logind has gained support for facilitating privileged input
2569 and drm device access for unprivileged clients. This work is
2570 useful to allow Wayland display servers (and similar
2571 programs, such as kmscon) to run under the user's ID and
2572 access input and drm devices which are normally
2573 protected. When this is used (and the kernel is new enough)
2574 logind will "mute" IO on the file descriptors passed to
2575 Wayland as long as it is in the background and "unmute" it
2576 if it returns into the foreground. This allows secure
2577 session switching without allowing background sessions to
2578 eavesdrop on input and display data. This also introduces
2579 session switching support if VT support is turned off in the
2580 kernel, and on seats that are not seat0.
2582 * A new kernel command line option luks.options= is understood
2583 now which allows specifying LUKS options for usage for LUKS
2584 encrypted partitions specified with luks.uuid=.
2586 * tmpfiles.d(5) snippets may now use specifier expansion in
2587 path names. More specifically %m, %b, %H, %v, are now
2588 replaced by the local machine id, boot id, hostname, and
2589 kernel version number.
2591 * A new tmpfiles.d(5) command "m" has been introduced which
2592 may be used to change the owner/group/access mode of a file
2593 or directory if it exists, but do nothing if it does not.
2595 * This release removes high-level support for the
2596 MemorySoftLimit= cgroup setting. The underlying kernel
2597 cgroup attribute memory.soft_limit= is currently badly
2598 designed and likely to be removed from the kernel API in its
2599 current form, hence we should not expose it for now.
2601 * The memory.use_hierarchy cgroup attribute is now enabled for
2602 all cgroups systemd creates in the memory cgroup
2603 hierarchy. This option is likely to be come the built-in
2604 default in the kernel anyway, and the non-hierarchical mode
2605 never made much sense in the intrinsically hierarchical
2608 * A new field _SYSTEMD_SLICE= is logged along with all journal
2609 messages containing the slice a message was generated
2610 from. This is useful to allow easy per-customer filtering of
2611 logs among other things.
2613 * systemd-journald will no longer adjust the group of journal
2614 files it creates to the "systemd-journal" group. Instead we
2615 rely on the journal directory to be owned by the
2616 "systemd-journal" group, and its setgid bit set, so that the
2617 kernel file system layer will automatically enforce that
2618 journal files inherit this group assignment. The reason for
2619 this change is that we cannot allow NSS look-ups from
2620 journald which would be necessary to resolve
2621 "systemd-journal" to a numeric GID, because this might
2622 create deadlocks if NSS involves synchronous queries to
2623 other daemons (such as nscd, or sssd) which in turn are
2624 logging clients of journald and might block on it, which
2625 would then dead lock. A tmpfiles.d(5) snippet included in
2626 systemd will make sure the setgid bit and group are
2627 properly set on the journal directory if it exists on every
2628 boot. However, we recommend adjusting it manually after
2629 upgrades too (or from RPM scriptlets), so that the change is
2630 not delayed until next reboot.
2632 * Backlight and random seed files in /var/lib/ have moved into
2633 the /var/lib/systemd/ directory, in order to centralize all
2634 systemd generated files in one directory.
2636 * Boot time performance measurements (as displayed by
2637 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
2638 performance information if that's available to determine how
2639 much time BIOS and boot loader initialization required. With
2640 a sufficiently new BIOS you hence no longer need to boot
2641 with Gummiboot to get access to such information.
2643 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
2644 Cristian Rodríguez, Dave Reisner, David Herrmann, David
2645 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
2646 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
2647 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
2648 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
2649 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
2651 -- Berlin, 2013-10-02
2655 * The Restart= option for services now understands a new
2656 on-watchdog setting, which will restart the service
2657 automatically if the service stops sending out watchdog keep
2658 alive messages (as configured with WatchdogSec=).
2660 * The getty generator (which is responsible for bringing up a
2661 getty on configured serial consoles) will no longer only
2662 start a getty on the primary kernel console but on all
2663 others, too. This makes the order in which console= is
2664 specified on the kernel command line less important.
2666 * libsystemd-logind gained a new sd_session_get_vt() call to
2667 retrieve the VT number of a session.
2669 * If the option "tries=0" is set for an entry of /etc/crypttab
2670 its passphrase is queried indefinitely instead of any
2671 maximum number of tries.
2673 * If a service with a configure PID file terminates its PID
2674 file will now be removed automatically if it still exists
2675 afterwards. This should put an end to stale PID files.
2677 * systemd-run will now also take relative binary path names
2678 for execution and no longer insists on absolute paths.
2680 * InaccessibleDirectories= and ReadOnlyDirectories= now take
2681 paths that are optionally prefixed with "-" to indicate that
2682 it should not be considered a failure if they do not exist.
2684 * journalctl -o (and similar commands) now understands a new
2685 output mode "short-precise", it is similar to "short" but
2686 shows timestamps with usec accuracy.
2688 * The option "discard" (as known from Debian) is now
2689 synonymous to "allow-discards" in /etc/crypttab. In fact,
2690 "discard" is preferred now (since it is easier to remember
2693 * Some licensing clean-ups were made, so that more code is now
2694 LGPL-2.1 licensed than before.
2696 * A minimal tool to save/restore the display backlight
2697 brightness across reboots has been added. It will store the
2698 backlight setting as late as possible at shutdown, and
2699 restore it as early as possible during reboot.
2701 * A logic to automatically discover and enable home and swap
2702 partitions on GPT disks has been added. With this in place
2703 /etc/fstab becomes optional for many setups as systemd can
2704 discover certain partitions located on the root disk
2705 automatically. Home partitions are recognized under their
2706 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
2707 partitions are recognized under their GPT type ID
2708 0657fd6da4ab43c484e50933c84b4f4f.
2710 * systemd will no longer pass any environment from the kernel
2711 or initrd to system services. If you want to set an
2712 environment for all services, do so via the kernel command
2713 line systemd.setenv= assignment.
2715 * The systemd-sysctl tool no longer natively reads the file
2716 /etc/sysctl.conf. If desired, the file should be symlinked
2717 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
2718 legacy support by a symlink rather than built-in code, it
2719 also makes the otherwise hidden order of application of the
2720 different files visible. (Note that this partly reverts to a
2721 pre-198 application order of sysctl knobs!)
2723 * The "systemctl set-log-level" and "systemctl dump" commands
2724 have been moved to systemd-analyze.
2726 * systemd-run learned the new --remain-after-exit switch,
2727 which causes the scope unit not to be cleaned up
2728 automatically after the process terminated.
2730 * tmpfiles learned a new --exclude-prefix= switch to exclude
2731 certain paths from operation.
2733 * journald will now automatically flush all messages to disk
2734 as soon as a message at the log level CRIT, ALERT or EMERG
2737 Contributions from: Andrew Cook, Brandon Philips, Christian
2738 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
2739 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
2740 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
2741 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
2742 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
2743 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
2744 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
2745 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
2746 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
2747 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
2748 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
2749 William Giokas, Zbigniew Jędrzejewski-Szmek
2751 -- Berlin, 2013-09-13
2755 * The documentation has been updated to cover the various new
2756 concepts introduced with 205.
2758 * Unit files now understand the new %v specifier which
2759 resolves to the kernel version string as returned by "uname
2762 * systemctl now supports filtering the unit list output by
2763 load state, active state and sub state, using the new
2766 * "systemctl status" will now show the results of the
2767 condition checks (like ConditionPathExists= and similar) of
2768 the last start attempts of the unit. They are also logged to
2771 * "journalctl -b" may now be used to look for boot output of a
2772 specific boot. Try "journalctl -b -1" for the previous boot,
2773 but the syntax is substantially more powerful.
2775 * "journalctl --show-cursor" has been added which prints the
2776 cursor string the last shown log line. This may then be used
2777 with the new "journalctl --after-cursor=" switch to continue
2778 browsing logs from that point on.
2780 * "journalctl --force" may now be used to force regeneration
2783 * Creation of "dead" device nodes has been moved from udev
2784 into kmod and tmpfiles. Previously, udev would read the kmod
2785 databases to pre-generate dead device nodes based on meta
2786 information contained in kernel modules, so that these would
2787 be auto-loaded on access rather then at boot. As this
2788 does not really have much to do with the exposing actual
2789 kernel devices to userspace this has always been slightly
2790 alien in the udev codebase. Following the new scheme kmod
2791 will now generate a runtime snippet for tmpfiles from the
2792 module meta information and it now is tmpfiles' job to the
2793 create the nodes. This also allows overriding access and
2794 other parameters for the nodes using the usual tmpfiles
2795 facilities. As side effect this allows us to remove the
2796 CAP_SYS_MKNOD capability bit from udevd entirely.
2798 * logind's device ACLs may now be applied to these "dead"
2799 devices nodes too, thus finally allowing managed access to
2800 devices such as /dev/snd/sequencer whithout loading the
2801 backing module right-away.
2803 * A new RPM macro has been added that may be used to apply
2804 tmpfiles configuration during package installation.
2806 * systemd-detect-virt and ConditionVirtualization= now can
2807 detect User-Mode-Linux machines (UML).
2809 * journald will now implicitly log the effective capabilities
2810 set of processes in the message metadata.
2812 * systemd-cryptsetup has gained support for TrueCrypt volumes.
2814 * The initrd interface has been simplified (more specifically,
2815 support for passing performance data via environment
2816 variables and fsck results via files in /run has been
2817 removed). These features were non-essential, and are
2818 nowadays available in a much nicer way by having systemd in
2819 the initrd serialize its state and have the hosts systemd
2820 deserialize it again.
2822 * The udev "keymap" data files and tools to apply keyboard
2823 specific mappings of scan to key codes, and force-release
2824 scan code lists have been entirely replaced by a udev
2825 "keyboard" builtin and a hwdb data file.
2827 * systemd will now honour the kernel's "quiet" command line
2828 argument also during late shutdown, resulting in a
2829 completely silent shutdown when used.
2831 * There's now an option to control the SO_REUSEPORT socket
2832 option in .socket units.
2834 * Instance units will now automatically get a per-template
2835 subslice of system.slice unless something else is explicitly
2836 configured. For example, instances of sshd@.service will now
2837 implicitly be placed in system-sshd.slice rather than
2838 system.slice as before.
2840 * Test coverage support may now be enabled at build time.
2842 Contributions from: Dave Reisner, Frederic Crozat, Harald
2843 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
2844 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
2845 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
2846 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
2847 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
2848 Giokas, Zbigniew Jędrzejewski-Szmek
2850 -- Berlin, 2013-07-23
2854 * Two new unit types have been introduced:
2856 Scope units are very similar to service units, however, are
2857 created out of pre-existing processes -- instead of PID 1
2858 forking off the processes. By using scope units it is
2859 possible for system services and applications to group their
2860 own child processes (worker processes) in a powerful way
2861 which then maybe used to organize them, or kill them
2862 together, or apply resource limits on them.
2864 Slice units may be used to partition system resources in an
2865 hierarchical fashion and then assign other units to them. By
2866 default there are now three slices: system.slice (for all
2867 system services), user.slice (for all user sessions),
2868 machine.slice (for VMs and containers).
2870 Slices and scopes have been introduced primarily in
2871 context of the work to move cgroup handling to a
2872 single-writer scheme, where only PID 1
2873 creates/removes/manages cgroups.
2875 * There's a new concept of "transient" units. In contrast to
2876 normal units these units are created via an API at runtime,
2877 not from configuration from disk. More specifically this
2878 means it is now possible to run arbitrary programs as
2879 independent services, with all execution parameters passed
2880 in via bus APIs rather than read from disk. Transient units
2881 make systemd substantially more dynamic then it ever was,
2882 and useful as a general batch manager.
2884 * logind has been updated to make use of scope and slice units
2885 for managing user sessions. As a user logs in he will get
2886 his own private slice unit, to which all sessions are added
2887 as scope units. We also added support for automatically
2888 adding an instance of user@.service for the user into the
2889 slice. Effectively logind will no longer create cgroup
2890 hierarchies on its own now, it will defer entirely to PID 1
2891 for this by means of scope, service and slice units. Since
2892 user sessions this way become entities managed by PID 1
2893 the output of "systemctl" is now a lot more comprehensive.
2895 * A new mini-daemon "systemd-machined" has been added which
2896 may be used by virtualization managers to register local
2897 VMs/containers. nspawn has been updated accordingly, and
2898 libvirt will be updated shortly. machined will collect a bit
2899 of meta information about the VMs/containers, and assign
2900 them their own scope unit (see above). The collected
2901 meta-data is then made available via the "machinectl" tool,
2902 and exposed in "ps" and similar tools. machined/machinectl
2903 is compile-time optional.
2905 * As discussed earlier, the low-level cgroup configuration
2906 options ControlGroup=, ControlGroupModify=,
2907 ControlGroupPersistent=, ControlGroupAttribute= have been
2908 removed. Please use high-level attribute settings instead as
2909 well as slice units.
2911 * A new bus call SetUnitProperties() has been added to alter
2912 various runtime parameters of a unit. This is primarily
2913 useful to alter cgroup parameters dynamically in a nice way,
2914 but will be extended later on to make more properties
2915 modifiable at runtime. systemctl gained a new set-properties
2916 command that wraps this call.
2918 * A new tool "systemd-run" has been added which can be used to
2919 run arbitrary command lines as transient services or scopes,
2920 while configuring a number of settings via the command
2921 line. This tool is currently very basic, however already
2922 very useful. We plan to extend this tool to even allow
2923 queuing of execution jobs with time triggers from the
2924 command line, similar in fashion to "at".
2926 * nspawn will now inform the user explicitly that kernels with
2927 audit enabled break containers, and suggest the user to turn
2930 * Support for detecting the IMA and AppArmor security
2931 frameworks with ConditionSecurity= has been added.
2933 * journalctl gained a new "-k" switch for showing only kernel
2934 messages, mimicking dmesg output; in addition to "--user"
2935 and "--system" switches for showing only user's own logs
2938 * systemd-delta can now show information about drop-in
2939 snippets extending unit files.
2941 * libsystemd-bus has been substantially updated but is still
2942 not available as public API.
2944 * systemd will now look for the "debug" argument on the kernel
2945 command line and enable debug logging, similar to what
2946 "systemd.log_level=debug" already did before.
2948 * "systemctl set-default", "systemctl get-default" has been
2949 added to configure the default.target symlink, which
2950 controls what to boot into by default.
2952 * "systemctl set-log-level" has been added as a convenient
2953 way to raise and lower systemd logging threshold.
2955 * "systemd-analyze plot" will now show the time the various
2956 generators needed for execution, as well as information
2957 about the unit file loading.
2959 * libsystemd-journal gained a new sd_journal_open_files() call
2960 for opening specific journal files. journactl also gained a
2961 new switch to expose this new functionality. Previously we
2962 only supported opening all files from a directory, or all
2963 files from the system, as opening individual files only is
2964 racy due to journal file rotation.
2966 * systemd gained the new DefaultEnvironment= setting in
2967 /etc/systemd/system.conf to set environment variables for
2970 * If a privileged process logs a journal message with the
2971 OBJECT_PID= field set, then journald will automatically
2972 augment this with additional OBJECT_UID=, OBJECT_GID=,
2973 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
2974 system services want to log events about specific client
2975 processes. journactl/systemctl has been updated to make use
2976 of this information if all log messages regarding a specific
2979 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
2980 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
2981 Reisner, David Coppa, David King, David Strauss, Eelco
2982 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
2983 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
2984 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
2985 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
2986 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
2987 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
2988 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
2989 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
2990 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
2991 Łukasz Stelmach, 장동준
2995 * The Python bindings gained some minimal support for the APIs
2996 exposed by libsystemd-logind.
2998 * ConditionSecurity= gained support for detecting SMACK. Since
2999 this condition already supports SELinux and AppArmor we only
3000 miss IMA for this. Patches welcome!
3002 Contributions from: Karol Lewandowski, Lennart Poettering,
3003 Zbigniew Jędrzejewski-Szmek
3007 * systemd-nspawn will now create /etc/resolv.conf if
3008 necessary, before bind-mounting the host's file onto it.
3010 * systemd-nspawn will now store meta information about a
3011 container on the container's cgroup as extended attribute
3012 fields, including the root directory.
3014 * The cgroup hierarchy has been reworked in many ways. All
3015 objects any of the components systemd creates in the cgroup
3016 tree are now suffixed. More specifically, user sessions are
3017 now placed in cgroups suffixed with ".session", users in
3018 cgroups suffixed with ".user", and nspawn containers in
3019 cgroups suffixed with ".nspawn". Furthermore, all cgroup
3020 names are now escaped in a simple scheme to avoid collision
3021 of userspace object names with kernel filenames. This work
3022 is preparation for making these objects relocatable in the
3023 cgroup tree, in order to allow easy resource partitioning of
3024 these objects without causing naming conflicts.
3026 * systemctl list-dependencies gained the new switches
3027 --plain, --reverse, --after and --before.
3029 * systemd-inhibit now shows the process name of processes that
3030 have taken an inhibitor lock.
3032 * nss-myhostname will now also resolve "localhost"
3033 implicitly. This makes /etc/hosts an optional file and
3034 nicely handles that on IPv6 ::1 maps to both "localhost" and
3037 * libsystemd-logind.so gained a new call
3038 sd_get_machine_names() to enumerate running containers and
3039 VMs (currently only supported by very new libvirt and
3040 nspawn). sd_login_monitor can now be used to watch
3041 VMs/containers coming and going.
3043 * .include is not allowed recursively anymore, and only in
3044 unit files. Usually it is better to use drop-in snippets in
3045 .d/*.conf anyway, as introduced with systemd 198.
3047 * systemd-analyze gained a new "critical-chain" command that
3048 determines the slowest chain of units run during system
3049 boot-up. It is very useful for tracking down where
3050 optimizing boot time is the most beneficial.
3052 * systemd will no longer allow manipulating service paths in
3053 the name=systemd:/system cgroup tree using ControlGroup= in
3054 units. (But is still fine with it in all other dirs.)
3056 * There's a new systemd-nspawn@.service service file that may
3057 be used to easily run nspawn containers as system
3058 services. With the container's root directory in
3059 /var/lib/container/foobar it is now sufficient to run
3060 "systemctl start systemd-nspawn@foobar.service" to boot it.
3062 * systemd-cgls gained a new parameter "--machine" to list only
3063 the processes within a certain container.
3065 * ConditionSecurity= now can check for "apparmor". We still
3066 are lacking checks for SMACK and IMA for this condition
3067 check though. Patches welcome!
3069 * A new configuration file /etc/systemd/sleep.conf has been
3070 added that may be used to configure which kernel operation
3071 systemd is supposed to execute when "suspend", "hibernate"
3072 or "hybrid-sleep" is requested. This makes the new kernel
3073 "freeze" state accessible to the user.
3075 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
3076 the passed argument if applicable.
3078 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
3079 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
3080 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
3081 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
3082 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
3083 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
3084 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
3089 * The output of 'systemctl list-jobs' got some polishing. The
3090 '--type=' argument may now be passed more than once. A new
3091 command 'systemctl list-sockets' has been added which shows
3092 a list of kernel sockets systemd is listening on with the
3093 socket units they belong to, plus the units these socket
3096 * The experimental libsystemd-bus library got substantial
3097 updates to work in conjunction with the (also experimental)
3098 kdbus kernel project. It works well enough to exchange
3099 messages with some sophistication. Note that kdbus is not
3100 ready yet, and the library is mostly an elaborate test case
3101 for now, and not installable.
3103 * systemd gained a new unit 'systemd-static-nodes.service'
3104 that generates static device nodes earlier during boot, and
3105 can run in conjunction with udev.
3107 * libsystemd-login gained a new call sd_pid_get_user_unit()
3108 to retrieve the user systemd unit a process is running
3109 in. This is useful for systems where systemd is used as
3112 * systemd-nspawn now places all containers in the new /machine
3113 top-level cgroup directory in the name=systemd
3114 hierarchy. libvirt will soon do the same, so that we get a
3115 uniform separation of /system, /user and /machine for system
3116 services, user processes and containers/virtual
3117 machines. This new cgroup hierarchy is also useful to stick
3118 stable names to specific container instances, which can be
3119 recognized later this way (this name may be controlled
3120 via systemd-nspawn's new -M switch). libsystemd-login also
3121 gained a new call sd_pid_get_machine_name() to retrieve the
3122 name of the container/VM a specific process belongs to.
3124 * bootchart can now store its data in the journal.
3126 * libsystemd-journal gained a new call
3127 sd_journal_add_conjunction() for AND expressions to the
3128 matching logic. This can be used to express more complex
3129 logical expressions.
3131 * journactl can now take multiple --unit= and --user-unit=
3134 * The cryptsetup logic now understands the "luks.key=" kernel
3135 command line switch for specifying a file to read the
3136 decryption key from. Also, if a configured key file is not
3137 found the tool will now automatically fall back to prompting
3140 * Python systemd.journal module was updated to wrap recently
3141 added functions from libsystemd-journal. The interface was
3142 changed to bring the low level interface in s.j._Reader
3143 closer to the C API, and the high level interface in
3144 s.j.Reader was updated to wrap and convert all data about
3147 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
3148 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
3149 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
3150 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
3151 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
3152 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3156 * journalctl --update-catalog now understands a new --root=
3157 option to operate on catalogs found in a different root
3160 * During shutdown after systemd has terminated all running
3161 services a final killing loop kills all remaining left-over
3162 processes. We will now print the name of these processes
3163 when we send SIGKILL to them, since this usually indicates a
3166 * If /etc/crypttab refers to password files stored on
3167 configured mount points automatic dependencies will now be
3168 generated to ensure the specific mount is established first
3169 before the key file is attempted to be read.
3171 * 'systemctl status' will now show information about the
3172 network sockets a socket unit is listening on.
3174 * 'systemctl status' will also shown information about any
3175 drop-in configuration file for units. (Drop-In configuration
3176 files in this context are files such as
3177 /etc/systemd/systemd/foobar.service.d/*.conf)
3179 * systemd-cgtop now optionally shows summed up CPU times of
3180 cgroups. Press '%' while running cgtop to switch between
3181 percentage and absolute mode. This is useful to determine
3182 which cgroups use up the most CPU time over the entire
3183 runtime of the system. systemd-cgtop has also been updated
3184 to be 'pipeable' for processing with further shell tools.
3186 * 'hostnamectl set-hostname' will now allow setting of FQDN
3189 * The formatting and parsing of time span values has been
3190 changed. The parser now understands fractional expressions
3191 such as "5.5h". The formatter will now output fractional
3192 expressions for all time spans under 1min, i.e. "5.123456s"
3193 rather than "5s 123ms 456us". For time spans under 1s
3194 millisecond values are shown, for those under 1ms
3195 microsecond values are shown. This should greatly improve
3196 all time-related output of systemd.
3198 * libsystemd-login and libsystemd-journal gained new
3199 functions for querying the poll() events mask and poll()
3200 timeout value for integration into arbitrary event
3203 * localectl gained the ability to list available X11 keymaps
3204 (models, layouts, variants, options).
3206 * 'systemd-analyze dot' gained the ability to filter for
3207 specific units via shell-style globs, to create smaller,
3208 more useful graphs. I.e. it is now possible to create simple
3209 graphs of all the dependencies between only target units, or
3210 of all units that Avahi has dependencies with.
3212 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
3213 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
3214 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
3215 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
3216 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
3217 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
3218 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
3222 * The boot-time readahead implementation for rotating media
3223 will now read the read-ahead data in multiple passes which
3224 consist of all read requests made in equidistant time
3225 intervals. This means instead of strictly reading read-ahead
3226 data in its physical order on disk we now try to find a
3227 middle ground between physical and access time order.
3229 * /etc/os-release files gained a new BUILD_ID= field for usage
3230 on operating systems that provide continuous builds of OS
3233 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
3234 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
3235 William Douglas, Zbigniew Jędrzejewski-Szmek
3239 * systemd-python gained an API exposing libsystemd-daemon.
3241 * The SMACK setup logic gained support for uploading CIPSO
3244 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
3245 ReadOnlyDirectories= and InaccessibleDirectories= has
3246 changed. The private /tmp and /var/tmp directories are now
3247 shared by all processes of a service (which means
3248 ExecStartPre= may now leave data in /tmp that ExecStart= of
3249 the same service can still access). When a service is
3250 stopped its temporary directories are immediately deleted
3251 (normal clean-up with tmpfiles is still done in addition to
3254 * By default, systemd will now set a couple of sysctl
3255 variables in the kernel: the safe sysrq options are turned
3256 on, IP route verification is turned on, and source routing
3257 disabled. The recently added hardlink and softlink
3258 protection of the kernel is turned on. These settings should
3259 be reasonably safe, and good defaults for all new systems.
3261 * The predictable network naming logic may now be turned off
3262 with a new kernel command line switch: net.ifnames=0.
3264 * A new libsystemd-bus module has been added that implements a
3265 pretty complete D-Bus client library. For details see:
3267 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
3269 * journald will now explicitly flush the journal files to disk
3270 at the latest 5min after each write. The file will then also
3271 be marked offline until the next write. This should increase
3272 reliability in case of a crash. The synchronization delay
3273 can be configured via SyncIntervalSec= in journald.conf.
3275 * There's a new remote-fs-setup.target unit that can be used
3276 to pull in specific services when at least one remote file
3277 system is to be mounted.
3279 * There are new targets timers.target and paths.target as
3280 canonical targets to pull user timer and path units in
3281 from. This complements sockets.target with a similar
3282 purpose for socket units.
3284 * libudev gained a new call udev_device_set_attribute_value()
3285 to set sysfs attributes of a device.
3287 * The udev daemon now sets the default number of worker
3288 processes executed in parallel based on the number of available
3289 CPUs instead of the amount of available RAM. This is supposed
3290 to provide a more reliable default and limit a too aggressive
3291 paralellism for setups with 1000s of devices connected.
3293 Contributions from: Auke Kok, Colin Walters, Cristian
3294 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
3295 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
3296 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
3297 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
3298 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
3299 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
3300 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
3301 Zbigniew Jędrzejewski-Szmek
3305 * Configuration of unit files may now be extended via drop-in
3306 files without having to edit/override the unit files
3307 themselves. More specifically, if the administrator wants to
3308 change one value for a service file foobar.service he can
3309 now do so by dropping in a configuration snippet into
3310 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
3311 will load all these snippets and apply them on top of the
3312 main unit configuration file, possibly extending or
3313 overriding its settings. Using these drop-in snippets is
3314 generally nicer than the two earlier options for changing
3315 unit files locally: copying the files from
3316 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
3317 them there; or creating a new file in /etc/systemd/system/
3318 that incorporates the original one via ".include". Drop-in
3319 snippets into these .d/ directories can be placed in any
3320 directory systemd looks for units in, and the usual
3321 overriding semantics between /usr/lib, /etc and /run apply
3324 * Most unit file settings which take lists of items can now be
3325 reset by assigning the empty string to them. For example,
3326 normally, settings such as Environment=FOO=BAR append a new
3327 environment variable assignment to the environment block,
3328 each time they are used. By assigning Environment= the empty
3329 string the environment block can be reset to empty. This is
3330 particularly useful with the .d/*.conf drop-in snippets
3331 mentioned above, since this adds the ability to reset list
3332 settings from vendor unit files via these drop-ins.
3334 * systemctl gained a new "list-dependencies" command for
3335 listing the dependencies of a unit recursively.
3337 * Inhibitors are now honored and listed by "systemctl
3338 suspend", "systemctl poweroff" (and similar) too, not only
3339 GNOME. These commands will also list active sessions by
3342 * Resource limits (as exposed by the various control group
3343 controllers) can now be controlled dynamically at runtime
3344 for all units. More specifically, you can now use a command
3345 like "systemctl set-cgroup-attr foobar.service cpu.shares
3346 2000" to alter the CPU shares a specific service gets. These
3347 settings are stored persistently on disk, and thus allow the
3348 administrator to easily adjust the resource usage of
3349 services with a few simple commands. This dynamic resource
3350 management logic is also available to other programs via the
3351 bus. Almost any kernel cgroup attribute and controller is
3354 * systemd-vconsole-setup will now copy all font settings to
3355 all allocated VTs, where it previously applied them only to
3358 * libsystemd-login gained the new sd_session_get_tty() API
3361 * This release drops support for a few legacy or
3362 distribution-specific LSB facility names when parsing init
3363 scripts: $x-display-manager, $mail-transfer-agent,
3364 $mail-transport-agent, $mail-transfer-agent, $smtp,
3365 $null. Also, the mail-transfer-agent.target unit backing
3366 this has been removed. Distributions which want to retain
3367 compatibility with this should carry the burden for
3368 supporting this themselves and patch support for these back
3369 in, if they really need to. Also, the facilities $syslog and
3370 $local_fs are now ignored, since systemd does not support
3371 early-boot LSB init scripts anymore, and these facilities
3372 are implied anyway for normal services. syslog.target has
3375 * There are new bus calls on PID1's Manager object for
3376 cancelling jobs, and removing snapshot units. Previously,
3377 both calls were only available on the Job and Snapshot
3380 * systemd-journal-gatewayd gained SSL support.
3382 * The various "environment" files, such as /etc/locale.conf
3383 now support continuation lines with a backslash ("\") as
3384 last character in the line, similarly in style (but different)
3385 to how this is supported in shells.
3387 * For normal user processes the _SYSTEMD_USER_UNIT= field is
3388 now implicitly appended to every log entry logged. systemctl
3389 has been updated to filter by this field when operating on a
3390 user systemd instance.
3392 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
3393 CAP_AUDIT_CONTROL capabilities to the capabilities set for
3394 the container. This makes it easier to boot unmodified
3395 Fedora systems in a container, which however still requires
3396 audit=0 to be passed on the kernel command line. Auditing in
3397 kernel and userspace is unfortunately still too broken in
3398 context of containers, hence we recommend compiling it out
3399 of the kernel or using audit=0. Hopefully this will be fixed
3400 one day for good in the kernel.
3402 * nspawn gained the new --bind= and --bind-ro= parameters to
3403 bind mount specific directories from the host into the
3406 * nspawn will now mount its own devpts file system instance
3407 into the container, in order not to leak pty devices from
3408 the host into the container.
3410 * systemd will now read the firmware boot time performance
3411 information from the EFI variables, if the used boot loader
3412 supports this, and takes it into account for boot performance
3413 analysis via "systemd-analyze". This is currently supported
3414 only in conjunction with Gummiboot, but could be supported
3415 by other boot loaders too. For details see:
3417 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
3419 * A new generator has been added that automatically mounts the
3420 EFI System Partition (ESP) to /boot, if that directory
3421 exists, is empty, and no other file system has been
3422 configured to be mounted there.
3424 * logind will now send out PrepareForSleep(false) out
3425 unconditionally, after coming back from suspend. This may be
3426 used by applications as asynchronous notification for
3427 system resume events.
3429 * "systemctl unlock-sessions" has been added, that allows
3430 unlocking the screens of all user sessions at once, similar
3431 to how "systemctl lock-sessions" already locked all users
3432 sessions. This is backed by a new D-Bus call UnlockSessions().
3434 * "loginctl seat-status" will now show the master device of a
3435 seat. (i.e. the device of a seat that needs to be around for
3436 the seat to be considered available, usually the graphics
3439 * tmpfiles gained a new "X" line type, that allows
3440 configuration of files and directories (with wildcards) that
3441 shall be excluded from automatic cleanup ("aging").
3443 * udev default rules set the device node permissions now only
3444 at "add" events, and do not change them any longer with a
3445 later "change" event.
3447 * The log messages for lid events and power/sleep keypresses
3448 now carry a message ID.
3450 * We now have a substantially larger unit test suite, but this
3451 continues to be work in progress.
3453 * udevadm hwdb gained a new --root= parameter to change the
3454 root directory to operate relative to.
3456 * logind will now issue a background sync() request to the kernel
3457 early at shutdown, so that dirty buffers are flushed to disk early
3458 instead of at the last moment, in order to optimize shutdown
3461 * A new bootctl tool has been added that is an interface for
3462 certain boot loader operations. This is currently a preview
3463 and is likely to be extended into a small mechanism daemon
3464 like timedated, localed, hostnamed, and can be used by
3465 graphical UIs to enumerate available boot options, and
3466 request boot into firmware operations.
3468 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
3469 the rest of the package. It also has been updated to work
3470 correctly in initrds.
3472 * Policykit previously has been runtime optional, and is now
3473 also compile time optional via a configure switch.
3475 * systemd-analyze has been reimplemented in C. Also "systemctl
3476 dot" has moved into systemd-analyze.
3478 * "systemctl status" with no further parameters will now print
3479 the status of all active or failed units.
3481 * Operations such as "systemctl start" can now be executed
3482 with a new mode "--irreversible" which may be used to queue
3483 operations that cannot accidentally be reversed by a later
3484 job queuing. This is by default used to make shutdown
3485 requests more robust.
3487 * The Python API of systemd now gained a new module for
3488 reading journal files.
3490 * A new tool kernel-install has been added that can install
3491 kernel images according to the Boot Loader Specification:
3493 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
3495 * Boot time console output has been improved to provide
3496 animated boot time output for hanging jobs.
3498 * A new tool systemd-activate has been added which can be used
3499 to test socket activation with, directly from the command
3500 line. This should make it much easier to test and debug
3501 socket activation in daemons.
3503 * journalctl gained a new "--reverse" (or -r) option to show
3504 journal output in reverse order (i.e. newest line first).
3506 * journalctl gained a new "--pager-end" (or -e) option to jump
3507 to immediately jump to the end of the journal in the
3508 pager. This is only supported in conjunction with "less".
3510 * journalctl gained a new "--user-unit=" option, that works
3511 similarly to "--unit=" but filters for user units rather than
3514 * A number of unit files to ease adoption of systemd in
3515 initrds has been added. This moves some minimal logic from
3516 the various initrd implementations into systemd proper.
3518 * The journal files are now owned by a new group
3519 "systemd-journal", which exists specifically to allow access
3520 to the journal, and nothing else. Previously, we used the
3521 "adm" group for that, which however possibly covers more
3522 than just journal/log file access. This new group is now
3523 already used by systemd-journal-gatewayd to ensure this
3524 daemon gets access to the journal files and as little else
3525 as possible. Note that "make install" will also set FS ACLs
3526 up for /var/log/journal to give "adm" and "wheel" read
3527 access to it, in addition to "systemd-journal" which owns
3528 the journal files. We recommend that packaging scripts also
3529 add read access to "adm" + "wheel" to /var/log/journal, and
3530 all existing/future journal files. To normal users and
3531 administrators little changes, however packagers need to
3532 ensure to create the "systemd-journal" system group at
3533 package installation time.
3535 * The systemd-journal-gatewayd now runs as unprivileged user
3536 systemd-journal-gateway:systemd-journal-gateway. Packaging
3537 scripts need to create these system user/group at
3540 * timedated now exposes a new boolean property CanNTP that
3541 indicates whether a local NTP service is available or not.
3543 * systemd-detect-virt will now also detect xen PVs
3545 * The pstore file system is now mounted by default, if it is
3548 * In addition to the SELinux and IMA policies we will now also
3549 load SMACK policies at early boot.
3551 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
3552 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
3553 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
3554 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
3555 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
3556 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
3557 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
3558 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
3559 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
3560 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
3561 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
3562 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
3563 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
3564 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
3568 * Timer units now support calendar time events in addition to
3569 monotonic time events. That means you can now trigger a unit
3570 based on a calendar time specification such as "Thu,Fri
3571 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
3572 or fifth day of any month of the year 2013, given that it is
3573 a thursday or friday. This brings timer event support
3574 considerably closer to cron's capabilities. For details on
3575 the supported calendar time specification language see
3578 * udev now supports a number of different naming policies for
3579 network interfaces for predictable names, and a combination
3580 of these policies is now the default. Please see this wiki
3581 document for details:
3583 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
3585 * Auke Kok's bootchart implementation has been added to the
3586 systemd tree. It is an optional component that can graph the
3587 boot in quite some detail. It is one of the best bootchart
3588 implementations around and minimal in its code and
3591 * nss-myhostname has been integrated into the systemd source
3592 tree. nss-myhostname guarantees that the local hostname
3593 always stays resolvable via NSS. It has been a weak
3594 requirement of systemd-hostnamed since a long time, and
3595 since its code is actually trivial we decided to just
3596 include it in systemd's source tree. It can be turned off
3597 with a configure switch.
3599 * The read-ahead logic is now capable of properly detecting
3600 whether a btrfs file system is on SSD or rotating media, in
3601 order to optimize the read-ahead scheme. Previously, it was
3602 only capable of detecting this on traditional file systems
3605 * In udev, additional device properties are now read from the
3606 IAB in addition to the OUI database. Also, Bluetooth company
3607 identities are attached to the devices as well.
3609 * In service files %U may be used as specifier that is
3610 replaced by the configured user name of the service.
3612 * nspawn may now be invoked without a controlling TTY. This
3613 makes it suitable for invocation as its own service. This
3614 may be used to set up a simple containerized server system
3615 using only core OS tools.
3617 * systemd and nspawn can now accept socket file descriptors
3618 when they are started for socket activation. This enables
3619 implementation of socket activated nspawn
3620 containers. i.e. think about autospawning an entire OS image
3621 when the first SSH or HTTP connection is received. We expect
3622 that similar functionality will also be added to libvirt-lxc
3625 * journalctl will now suppress ANSI color codes when
3626 presenting log data.
3628 * systemctl will no longer show control group information for
3629 a unit if a the control group is empty anyway.
3631 * logind can now automatically suspend/hibernate/shutdown the
3634 * /etc/machine-info and hostnamed now also expose the chassis
3635 type of the system. This can be used to determine whether
3636 the local system is a laptop, desktop, handset or
3637 tablet. This information may either be configured by the
3638 user/vendor or is automatically determined from ACPI and DMI
3639 information if possible.
3641 * A number of PolicyKit actions are now bound together with
3642 "imply" rules. This should simplify creating UIs because
3643 many actions will now authenticate similar ones as well.
3645 * Unit files learnt a new condition ConditionACPower= which
3646 may be used to conditionalize a unit depending on whether an
3647 AC power source is connected or not, of whether the system
3648 is running on battery power.
3650 * systemctl gained a new "is-failed" verb that may be used in
3651 shell scripts and suchlike to check whether a specific unit
3652 is in the "failed" state.
3654 * The EnvironmentFile= setting in unit files now supports file
3655 globbing, and can hence be used to easily read a number of
3656 environment files at once.
3658 * systemd will no longer detect and recognize specific
3659 distributions. All distribution-specific #ifdeffery has been
3660 removed, systemd is now fully generic and
3661 distribution-agnostic. Effectively, not too much is lost as
3662 a lot of the code is still accessible via explicit configure
3663 switches. However, support for some distribution specific
3664 legacy configuration file formats has been dropped. We
3665 recommend distributions to simply adopt the configuration
3666 files everybody else uses now and convert the old
3667 configuration from packaging scripts. Most distributions
3668 already did that. If that's not possible or desirable,
3669 distributions are welcome to forward port the specific
3670 pieces of code locally from the git history.
3672 * When logging a message about a unit systemd will now always
3673 log the unit name in the message meta data.
3675 * localectl will now also discover system locale data that is
3676 not stored in locale archives, but directly unpacked.
3678 * logind will no longer unconditionally use framebuffer
3679 devices as seat masters, i.e. as devices that are required
3680 to be existing before a seat is considered preset. Instead,
3681 it will now look for all devices that are tagged as
3682 "seat-master" in udev. By default framebuffer devices will
3683 be marked as such, but depending on local systems other
3684 devices might be marked as well. This may be used to
3685 integrate graphics cards using closed source drivers (such
3686 as NVidia ones) more nicely into logind. Note however, that
3687 we recommend using the open source NVidia drivers instead,
3688 and no udev rules for the closed-source drivers will be
3689 shipped from us upstream.
3691 Contributions from: Adam Williamson, Alessandro Crismani, Auke
3692 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
3693 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
3694 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
3695 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
3696 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
3697 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
3698 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
3699 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
3700 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
3701 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
3702 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
3707 * udev gained support for loading additional device properties
3708 from an indexed database that is keyed by vendor/product IDs
3709 and similar device identifiers. For the beginning this
3710 "hwdb" is populated with data from the well-known PCI and
3711 USB database, but also includes PNP, ACPI and OID data. In
3712 the longer run this indexed database shall grow into
3713 becoming the one central database for non-essential
3714 userspace device metadata. Previously, data from the PCI/USB
3715 database was only attached to select devices, since the
3716 lookup was a relatively expensive operation due to O(n) time
3717 complexity (with n being the number of entries in the
3718 database). Since this is now O(1), we decided to add in this
3719 data for all devices where this is available, by
3720 default. Note that the indexed database needs to be rebuilt
3721 when new data files are installed. To achieve this you need
3722 to update your packaging scripts to invoke "udevadm hwdb
3723 --update" after installation of hwdb data files. For
3724 RPM-based distributions we introduced the new
3725 %udev_hwdb_update macro for this purpose.
3727 * The Journal gained support for the "Message Catalog", an
3728 indexed database to link up additional information with
3729 journal entries. For further details please check:
3731 http://www.freedesktop.org/wiki/Software/systemd/catalog
3733 The indexed message catalog database also needs to be
3734 rebuilt after installation of message catalog files. Use
3735 "journalctl --update-catalog" for this. For RPM-based
3736 distributions we introduced the %journal_catalog_update
3737 macro for this purpose.
3739 * The Python Journal bindings gained support for the standard
3740 Python logging framework.
3742 * The Journal API gained new functions for checking whether
3743 the underlying file system of a journal file is capable of
3744 properly reporting file change notifications, or whether
3745 applications that want to reflect journal changes "live"
3746 need to recheck journal files continuously in appropriate
3749 * It is now possible to set the "age" field for tmpfiles
3750 entries to 0, indicating that files matching this entry
3751 shall always be removed when the directories are cleaned up.
3753 * coredumpctl gained a new "gdb" verb which invokes gdb
3754 right-away on the selected coredump.
3756 * There's now support for "hybrid sleep" on kernels that
3757 support this, in addition to "suspend" and "hibernate". Use
3758 "systemctl hybrid-sleep" to make use of this.
3760 * logind's HandleSuspendKey= setting (and related settings)
3761 now gained support for a new "lock" setting to simply
3762 request the screen lock on all local sessions, instead of
3763 actually executing a suspend or hibernation.
3765 * systemd will now mount the EFI variables file system by
3768 * Socket units now gained support for configuration of the
3769 SMACK security label.
3771 * timedatectl will now output the time of the last and next
3772 daylight saving change.
3774 * We dropped support for various legacy and distro-specific
3775 concepts, such as insserv, early-boot SysV services
3776 (i.e. those for non-standard runlevels such as 'b' or 'S')
3777 or ArchLinux /etc/rc.conf support. We recommend the
3778 distributions who still need support this to either continue
3779 to maintain the necessary patches downstream, or find a
3780 different solution. (Talk to us if you have questions!)
3782 * Various systemd components will now bypass PolicyKit checks
3783 for root and otherwise handle properly if PolicyKit is not
3784 found to be around. This should fix most issues for
3785 PolicyKit-less systems. Quite frankly this should have been
3786 this way since day one. It is absolutely our intention to
3787 make systemd work fine on PolicyKit-less systems, and we
3788 consider it a bug if something does not work as it should if
3789 PolicyKit is not around.
3791 * For embedded systems it is now possible to build udev and
3792 systemd without blkid and/or kmod support.
3794 * "systemctl switch-root" is now capable of switching root
3795 more than once. I.e. in addition to transitions from the
3796 initrd to the host OS it is now possible to transition to
3797 further OS images from the host. This is useful to implement
3798 offline updating tools.
3800 * Various other additions have been made to the RPM macros
3801 shipped with systemd. Use %udev_rules_update() after
3802 installing new udev rules files. %_udevhwdbdir,
3803 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
3804 %_sysctldir are now available which resolve to the right
3805 directories for packages to place various data files in.
3807 * journalctl gained the new --full switch (in addition to
3808 --all, to disable ellipsation for long messages.
3810 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
3811 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
3812 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
3813 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
3814 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
3815 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
3816 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
3817 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
3818 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
3822 * journalctl gained new --since= and --until= switches to
3823 filter by time. It also now supports nice filtering for
3824 units via --unit=/-u.
3826 * Type=oneshot services may use ExecReload= and do the
3829 * The journal daemon now supports time-based rotation and
3830 vacuuming, in addition to the usual disk-space based
3833 * The journal will now index the available field values for
3834 each field name. This enables clients to show pretty drop
3835 downs of available match values when filtering. The bash
3836 completion of journalctl has been updated
3837 accordingly. journalctl gained a new switch -F to list all
3838 values a certain field takes in the journal database.
3840 * More service events are now written as structured messages
3841 to the journal, and made recognizable via message IDs.
3843 * The timedated, localed and hostnamed mini-services which
3844 previously only provided support for changing time, locale
3845 and hostname settings from graphical DEs such as GNOME now
3846 also have a minimal (but very useful) text-based client
3847 utility each. This is probably the nicest way to changing
3848 these settings from the command line now, especially since
3849 it lists available options and is fully integrated with bash
3852 * There's now a new tool "systemd-coredumpctl" to list and
3853 extract coredumps from the journal.
3855 * We now install a README each in /var/log/ and
3856 /etc/rc.d/init.d explaining where the system logs and init
3857 scripts went. This hopefully should help folks who go to
3858 that dirs and look into the otherwise now empty void and
3859 scratch their heads.
3861 * When user-services are invoked (by systemd --user) the
3862 $MANAGERPID env var is set to the PID of systemd.
3864 * SIGRTMIN+24 when sent to a --user instance will now result
3865 in immediate termination of systemd.
3867 * gatewayd received numerous feature additions such as a
3868 "follow" mode, for live syncing and filtering.
3870 * browse.html now allows filtering and showing detailed
3871 information on specific entries. Keyboard navigation and
3872 mouse screen support has been added.
3874 * gatewayd/journalctl now supports HTML5/JSON
3875 Server-Sent-Events as output.
3877 * The SysV init script compatibility logic will now
3878 heuristically determine whether a script supports the
3879 "reload" verb, and only then make this available as
3882 * "systemctl status --follow" has been removed, use "journalctl
3885 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
3886 have been removed since they are hardly useful to be
3889 * And I'd like to take the opportunity to specifically mention
3890 Zbigniew for his great contributions. Zbigniew, you rock!
3892 Contributions from: Andrew Eikum, Christian Hesse, Colin
3893 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
3894 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
3895 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
3896 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
3897 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
3898 Jędrzejewski-Szmek, Сковорода Никита Андреевич
3902 * If /etc/vconsole.conf is non-existent or empty we will no
3903 longer load any console font or key map at boot by
3904 default. Instead the kernel defaults will be left
3905 intact. This is definitely the right thing to do, as no
3906 configuration should mean no configuration, and hard-coding
3907 font names that are different on all archs is probably a bad
3908 idea. Also, the kernel default key map and font should be
3909 good enough for most cases anyway, and mostly identical to
3910 the userspace fonts/key maps we previously overloaded them
3911 with. If distributions want to continue to default to a
3912 non-kernel font or key map they should ship a default
3913 /etc/vconsole.conf with the appropriate contents.
3915 Contributions from: Colin Walters, Daniel J Walsh, Dave
3916 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
3917 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3921 * journalctl gained a new --cursor= switch to show entries
3922 starting from the specified location in the journal.
3924 * We now enforce a size limit on journal entry fields exported
3925 with "-o json" in journalctl. Fields larger than 4K will be
3926 assigned null. This can be turned off with --all.
3928 * An (optional) journal gateway daemon is now available as
3929 "systemd-journal-gatewayd.service". This service provides
3930 access to the journal via HTTP and JSON. This functionality
3931 will be used to implement live log synchronization in both
3932 pull and push modes, but has various other users too, such
3933 as easy log access for debugging of embedded devices. Right
3934 now it is already useful to retrieve the journal via HTTP:
3936 # systemctl start systemd-journal-gatewayd.service
3937 # wget http://localhost:19531/entries
3939 This will download the journal contents in a
3940 /var/log/messages compatible format. The same as JSON:
3942 # curl -H"Accept: application/json" http://localhost:19531/entries
3944 This service is also accessible via a web browser where a
3945 single static HTML5 app is served that uses the JSON logic
3946 to enable the user to do some basic browsing of the
3947 journal. This will be extended later on. Here's an example
3948 screenshot of this app in its current state:
3950 http://0pointer.de/public/journal-gatewayd
3952 Contributions from: Kay Sievers, Lennart Poettering, Robert
3953 Milasan, Tom Gundersen
3957 * The bash completion logic is now available for journalctl
3960 * We do not mount the "cpuset" controller anymore together with
3961 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
3962 started if no parameters are assigned to it. "cpuset" hence
3963 broke code that assumed it it could create "cpu" groups and
3966 * journalctl -f will now subscribe to terminal size changes,
3967 and line break accordingly.
3969 Contributions from: Dave Reisner, Kay Sievers, Lennart
3970 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
3974 * nspawn will now create a symlink /etc/localtime in the
3975 container environment, copying the host's timezone
3976 setting. Previously this has been done via a bind mount, but
3977 since symlinks cannot be bind mounted this has now been
3978 changed to create/update the appropriate symlink.
3980 * journalctl -n's line number argument is now optional, and
3981 will default to 10 if omitted.
3983 * journald will now log the maximum size the journal files may
3984 take up on disk. This is particularly useful if the default
3985 built-in logic of determining this parameter from the file
3986 system size is used. Use "systemctl status
3987 systemd-journald.service" to see this information.
3989 * The multi-seat X wrapper tool has been stripped down. As X
3990 is now capable of enumerating graphics devices via udev in a
3991 seat-aware way the wrapper is not strictly necessary
3992 anymore. A stripped down temporary stop-gap is still shipped
3993 until the upstream display managers have been updated to
3994 fully support the new X logic. Expect this wrapper to be
3995 removed entirely in one of the next releases.
3997 * HandleSleepKey= in logind.conf has been split up into
3998 HandleSuspendKey= and HandleHibernateKey=. The old setting
3999 is not available anymore. X11 and the kernel are
4000 distuingishing between these keys and we should too. This
4001 also means the inhibition lock for these keys has been split
4004 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
4005 Poettering, Lukas Nykryn, Václav Pavlín
4009 * Whenever a unit changes state we will now log this to the
4010 journal and show along the unit's own log output in
4013 * ConditionPathIsMountPoint= can now properly detect bind
4014 mount points too. (Previously, a bind mount of one file
4015 system to another place in the same file system could not be
4016 detected as mount, since they shared struct stat's st_dev
4019 * We will now mount the cgroup controllers cpu, cpuacct,
4020 cpuset and the controllers net_cls, net_prio together by
4023 * nspawn containers will now have a virtualized boot
4024 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
4025 over with a randomized ID at container initialization). This
4026 has the effect of making "journalctl -b" do the right thing
4029 * The JSON output journal serialization has been updated not
4030 to generate "endless" list objects anymore, but rather one
4031 JSON object per line. This is more in line how most JSON
4032 parsers expect JSON objects. The new output mode
4033 "json-pretty" has been added to provide similar output, but
4034 neatly aligned for readability by humans.
4036 * We dropped all explicit sync() invocations in the shutdown
4037 code. The kernel does this implicitly anyway in the kernel
4038 reboot() syscall. halt(8)'s -n option is now a compatibility
4041 * We now support virtualized reboot() in containers, as
4042 supported by newer kernels. We will fall back to exit() if
4043 CAP_SYS_REBOOT is not available to the container. Also,
4044 nspawn makes use of this now and will actually reboot the
4045 container if the containerized OS asks for that.
4047 * journalctl will only show local log output by default
4048 now. Use --merge (-m) to show remote log output, too.
4050 * libsystemd-journal gained the new sd_journal_get_usage()
4051 call to determine the current disk usage of all journal
4052 files. This is exposed in the new "journalctl --disk-usage"
4055 * journald gained a new configuration setting SplitMode= in
4056 journald.conf which may be used to control how user journals
4057 are split off. See journald.conf(5) for details.
4059 * A new condition type ConditionFileNotEmpty= has been added.
4061 * tmpfiles' "w" lines now support file globbing, to write
4062 multiple files at once.
4064 * We added Python bindings for the journal submission
4065 APIs. More Python APIs for a number of selected APIs will
4066 likely follow. Note that we intend to add native bindings
4067 only for the Python language, as we consider it common
4068 enough to deserve bindings shipped within systemd. There are
4069 various projects outside of systemd that provide bindings
4070 for languages such as PHP or Lua.
4072 * Many conditions will now resolve specifiers such as %i. In
4073 addition, PathChanged= and related directives of .path units
4074 now support specifiers as well.
4076 * There's now a new RPM macro definition for the system preset
4079 * journald will now warn if it ca not forward a message to the
4080 syslog daemon because its socket is full.
4082 * timedated will no longer write or process /etc/timezone,
4083 except on Debian. As we do not support late mounted /usr
4084 anymore /etc/localtime always being a symlink is now safe,
4085 and hence the information in /etc/timezone is not necessary
4088 * logind will now always reserve one VT for a text getty (VT6
4089 by default). Previously if more than 6 X sessions where
4090 started they took up all the VTs with auto-spawned gettys,
4091 so that no text gettys were available anymore.
4093 * udev will now automatically inform the btrfs kernel logic
4094 about btrfs RAID components showing up. This should make
4095 simple hotplug based btrfs RAID assembly work.
4097 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
4098 (but not for its children which will stay at the kernel
4099 default). This should allow setups with a lot more listening
4102 * systemd will now always pass the configured timezone to the
4103 kernel at boot. timedated will do the same when the timezone
4106 * logind's inhibition logic has been updated. By default,
4107 logind will now handle the lid switch, the power and sleep
4108 keys all the time, even in graphical sessions. If DEs want
4109 to handle these events on their own they should take the new
4110 handle-power-key, handle-sleep-key and handle-lid-switch
4111 inhibitors during their runtime. A simple way to achieve
4112 that is to invoke the DE wrapped in an invocation of:
4114 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
4116 * Access to unit operations is now checked via SELinux taking
4117 the unit file label and client process label into account.
4119 * systemd will now notify the administrator in the journal
4120 when he over-mounts a non-empty directory.
4122 * There are new specifiers that are resolved in unit files,
4123 for the host name (%H), the machine ID (%m) and the boot ID
4126 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
4127 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
4128 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
4129 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
4130 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
4131 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
4132 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
4136 * Support for reading structured kernel messages from
4137 /dev/kmsg has now been added and is enabled by default.
4139 * Support for reading kernel messages from /proc/kmsg has now
4140 been removed. If you want kernel messages in the journal
4141 make sure to run a recent kernel (>= 3.5) that supports
4142 reading structured messages from /dev/kmsg (see
4143 above). /proc/kmsg is now exclusive property of classic
4144 syslog daemons again.
4146 * The libudev API gained the new
4147 udev_device_new_from_device_id() call.
4149 * The logic for file system namespace (ReadOnlyDirectory=,
4150 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
4151 require pivot_root() anymore. This means fewer temporary
4152 directories are created below /tmp for this feature.
4154 * nspawn containers will now see and receive all submounts
4155 made on the host OS below the root file system of the
4158 * Forward Secure Sealing is now supported for Journal files,
4159 which provide cryptographical sealing of journal files so
4160 that attackers cannot alter log history anymore without this
4161 being detectable. Lennart will soon post a blog story about
4162 this explaining it in more detail.
4164 * There are two new service settings RestartPreventExitStatus=
4165 and SuccessExitStatus= which allow configuration of exit
4166 status (exit code or signal) which will be excepted from the
4167 restart logic, resp. consider successful.
4169 * journalctl gained the new --verify switch that can be used
4170 to check the integrity of the structure of journal files and
4171 (if Forward Secure Sealing is enabled) the contents of
4174 * nspawn containers will now be run with /dev/stdin, /dev/fd/
4175 and similar symlinks pre-created. This makes running shells
4176 as container init process a lot more fun.
4178 * The fstab support can now handle PARTUUID= and PARTLABEL=
4181 * A new ConditionHost= condition has been added to match
4182 against the hostname (with globs) and machine ID. This is
4183 useful for clusters where a single OS image is used to
4184 provision a large number of hosts which shall run slightly
4185 different sets of services.
4187 * Services which hit the restart limit will now be placed in a
4190 Contributions from: Bertram Poettering, Dave Reisner, Huang
4191 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
4192 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
4196 * When running in --user mode systemd will now become a
4197 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
4198 tree a lot more organized.
4200 * A new PartOf= unit dependency type has been introduced that
4201 may be used to group services in a natural way.
4203 * "systemctl enable" may now be used to enable instances of
4206 * journalctl now prints error log levels in red, and
4207 warning/notice log levels in bright white. It also supports
4208 filtering by log level now.
4210 * cgtop gained a new -n switch (similar to top), to configure
4211 the maximum number of iterations to run for. It also gained
4212 -b, to run in batch mode (accepting no input).
4214 * The suffix ".service" may now be omitted on most systemctl
4215 command lines involving service unit names.
4217 * There's a new bus call in logind to lock all sessions, as
4218 well as a loginctl verb for it "lock-sessions".
4220 * libsystemd-logind.so gained a new call sd_journal_perror()
4221 that works similar to libc perror() but logs to the journal
4222 and encodes structured information about the error number.
4224 * /etc/crypttab entries now understand the new keyfile-size=
4227 * shutdown(8) now can send a (configurable) wall message when
4228 a shutdown is cancelled.
4230 * The mount propagation mode for the root file system will now
4231 default to "shared", which is useful to make containers work
4232 nicely out-of-the-box so that they receive new mounts from
4233 the host. This can be undone locally by running "mount
4234 --make-rprivate /" if needed.
4236 * The prefdm.service file has been removed. Distributions
4237 should maintain this unit downstream if they intend to keep
4238 it around. However, we recommend writing normal unit files
4239 for display managers instead.
4241 * Since systemd is a crucial part of the OS we will now
4242 default to a number of compiler switches that improve
4243 security (hardening) such as read-only relocations, stack
4244 protection, and suchlike.
4246 * The TimeoutSec= setting for services is now split into
4247 TimeoutStartSec= and TimeoutStopSec= to allow configuration
4248 of individual time outs for the start and the stop phase of
4251 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
4252 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
4253 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
4254 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
4255 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
4256 Gundersen, Zbigniew Jędrzejewski-Szmek
4260 * The journal and id128 C APIs are now fully documented as man
4263 * Extra safety checks have been added when transitioning from
4264 the initial RAM disk to the main system to avoid accidental
4267 * /etc/crypttab entries now understand the new keyfile-offset=
4270 * systemctl -t can now be used to filter by unit load state.
4272 * The journal C API gained the new sd_journal_wait() call to
4273 make writing synchronous journal clients easier.
4275 * journalctl gained the new -D switch to show journals from a
4278 * journalctl now displays a special marker between log
4279 messages of two different boots.
4281 * The journal is now explicitly flushed to /var via a service
4282 systemd-journal-flush.service, rather than implicitly simply
4283 by seeing /var/log/journal to be writable.
4285 * journalctl (and the journal C APIs) can now match for much
4286 more complex expressions, with alternatives and
4289 * When transitioning from the initial RAM disk to the main
4290 system we will now kill all processes in a killing spree to
4291 ensure no processes stay around by accident.
4293 * Three new specifiers may be used in unit files: %u, %h, %s
4294 resolve to the user name, user home directory resp. user
4295 shell. This is useful for running systemd user instances.
4297 * We now automatically rotate journal files if their data
4298 object hash table gets a fill level > 75%. We also size the
4299 hash table based on the configured maximum file size. This
4300 together should lower hash collisions drastically and thus
4301 speed things up a bit.
4303 * journalctl gained the new "--header" switch to introspect
4304 header data of journal files.
4306 * A new setting SystemCallFilters= has been added to services
4307 which may be used to apply blacklists or whitelists to
4308 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
4310 * nspawn gained a new --link-journal= switch (and quicker: -j)
4311 to link the container journal with the host. This makes it
4312 very easy to centralize log viewing on the host for all
4313 guests while still keeping the journal files separated.
4315 * Many bugfixes and optimizations
4317 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
4318 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
4319 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
4324 * Several tools now understand kernel command line arguments,
4325 which are only read when run in an initial RAM disk. They
4326 usually follow closely their normal counterparts, but are
4329 * There's a new tool to analyze the readahead files that are
4330 automatically generated at boot. Use:
4332 /usr/lib/systemd/systemd-readahead analyze /.readahead
4334 * We now provide an early debug shell on tty9 if this enabled. Use:
4336 systemctl enable debug-shell.service
4338 * All plymouth related units have been moved into the Plymouth
4339 package. Please make sure to upgrade your Plymouth version
4342 * systemd-tmpfiles now supports getting passed the basename of
4343 a configuration file only, in which case it will look for it
4344 in all appropriate directories automatically.
4346 * udevadm info now takes a /dev or /sys path as argument, and
4347 does the right thing. Example:
4349 udevadm info /dev/sda
4350 udevadm info /sys/class/block/sda
4352 * systemctl now prints a warning if a unit is stopped but a
4353 unit that might trigger it continues to run. Example: a
4354 service is stopped but the socket that activates it is left
4357 * "systemctl status" will now mention if the log output was
4358 shortened due to rotation since a service has been started.
4360 * The journal API now exposes functions to determine the
4361 "cutoff" times due to rotation.
4363 * journald now understands SIGUSR1 and SIGUSR2 for triggering
4364 immediately flushing of runtime logs to /var if possible,
4365 resp. for triggering immediate rotation of the journal
4368 * It is now considered an error if a service is attempted to
4369 be stopped that is not loaded.
4371 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
4373 * systemd-analyze now supports Python 3
4375 * tmpfiles now supports cleaning up directories via aging
4376 where the first level dirs are always kept around but
4377 directories beneath it automatically aged. This is enabled
4378 by prefixing the age field with '~'.
4380 * Seat objects now expose CanGraphical, CanTTY properties
4381 which is required to deal with very fast bootups where the
4382 display manager might be running before the graphics drivers
4383 completed initialization.
4385 * Seat objects now expose a State property.
4387 * We now include RPM macros for service enabling/disabling
4388 based on the preset logic. We recommend RPM based
4389 distributions to make use of these macros if possible. This
4390 makes it simpler to reuse RPM spec files across
4393 * We now make sure that the collected systemd unit name is
4394 always valid when services log to the journal via
4397 * There's a new man page kernel-command-line(7) detailing all
4398 command line options we understand.
4400 * The fstab generator may now be disabled at boot by passing
4401 fstab=0 on the kernel command line.
4403 * A new kernel command line option modules-load= is now understood
4404 to load a specific kernel module statically, early at boot.
4406 * Unit names specified on the systemctl command line are now
4407 automatically escaped as needed. Also, if file system or
4408 device paths are specified they are automatically turned
4409 into the appropriate mount or device unit names. Example:
4411 systemctl status /home
4412 systemctl status /dev/sda
4414 * The SysVConsole= configuration option has been removed from
4415 system.conf parsing.
4417 * The SysV search path is no longer exported on the D-Bus
4420 * The Names= option is been removed from unit file parsing.
4422 * There's a new man page bootup(7) detailing the boot process.
4424 * Every unit and every generator we ship with systemd now
4425 comes with full documentation. The self-explanatory boot is
4428 * A couple of services gained "systemd-" prefixes in their
4429 name if they wrap systemd code, rather than only external
4430 code. Among them fsck@.service which is now
4431 systemd-fsck@.service.
4433 * The HaveWatchdog property has been removed from the D-Bus
4436 * systemd.confirm_spawn= on the kernel command line should now
4439 * There's a new man page crypttab(5) which details all options
4440 we actually understand.
4442 * systemd-nspawn gained a new --capability= switch to pass
4443 additional capabilities to the container.
4445 * timedated will now read known NTP implementation unit names
4446 from /usr/lib/systemd/ntp-units.d/*.list,
4447 systemd-timedated-ntp.target has been removed.
4449 * journalctl gained a new switch "-b" that lists log data of
4450 the current boot only.
4452 * The notify socket is in the abstract namespace again, in
4453 order to support daemons which chroot() at start-up.
4455 * There is a new Storage= configuration option for journald
4456 which allows configuration of where log data should go. This
4457 also provides a way to disable journal logging entirely, so
4458 that data collected is only forwarded to the console, the
4459 kernel log buffer or another syslog implementation.
4461 * Many bugfixes and optimizations
4463 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
4464 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
4465 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
4466 Shawn Landden, Tom Gundersen
4470 * "systemctl help <unit>" now shows the man page if one is
4473 * Several new man pages have been added.
4475 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
4476 MaxLevelConsole= can now be specified in
4477 journald.conf. These options allow reducing the amount of
4478 data stored on disk or forwarded by the log level.
4480 * TimerSlackNSec= can now be specified in system.conf for
4481 PID1. This allows system-wide power savings.
4483 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
4484 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
4489 * logind is now capable of (optionally) handling power and
4490 sleep keys as well as the lid switch.
4492 * journalctl now understands the syntax "journalctl
4493 /usr/bin/avahi-daemon" to get all log output of a specific
4496 * CapabilityBoundingSet= in system.conf now also influences
4497 the capability bound set of usermode helpers of the kernel.
4499 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
4500 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
4501 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
4502 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
4506 * Note that we skipped 139 releases here in order to set the
4507 new version to something that is greater than both udev's
4508 and systemd's most recent version number.
4510 * udev: all udev sources are merged into the systemd source tree now.
4511 All future udev development will happen in the systemd tree. It
4512 is still fully supported to use the udev daemon and tools without
4513 systemd running, like in initramfs or other init systems. Building
4514 udev though, will require the *build* of the systemd tree, but
4515 udev can be properly *run* without systemd.
4517 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
4518 should be used to create dead device nodes as workarounds for broken
4521 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
4522 no longer supported. udev_monitor_new_from_netlink() needs to be
4523 used to subscribe to events.
4525 * udev: when udevd is started by systemd, processes which are left
4526 behind by forking them off of udev rules, are unconditionally cleaned
4527 up and killed now after the event handling has finished. Services or
4528 daemons must be started as systemd services. Services can be
4529 pulled-in by udev to get started, but they can no longer be directly
4530 forked by udev rules.
4532 * udev: the daemon binary is called systemd-udevd now and installed
4533 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
4534 to adapt to that, create symlink, or rename the binary after building
4537 * libudev no longer provides these symbols:
4538 udev_monitor_from_socket()
4539 udev_queue_get_failed_list_entry()
4540 udev_get_{dev,sys,run}_path()
4541 The versions number was bumped and symbol versioning introduced.
4543 * systemd-loginctl and systemd-journalctl have been renamed
4544 to loginctl and journalctl to match systemctl.
4546 * The config files: /etc/systemd/systemd-logind.conf and
4547 /etc/systemd/systemd-journald.conf have been renamed to
4548 logind.conf and journald.conf. Package updates should rename
4549 the files to the new names on upgrade.
4551 * For almost all files the license is now LGPL2.1+, changed
4552 from the previous GPL2.0+. Exceptions are some minor stuff
4553 of udev (which will be changed to LGPL2.1 eventually, too),
4554 and the MIT licensed sd-daemon.[ch] library that is suitable
4555 to be used as drop-in files.
4557 * systemd and logind now handle system sleep states, in
4558 particular suspending and hibernating.
4560 * logind now implements a sleep/shutdown/idle inhibiting logic
4561 suitable for a variety of uses. Soonishly Lennart will blog
4562 about this in more detail.
4564 * var-run.mount and var-lock.mount are no longer provided
4565 (which prevously bind mounted these directories to their new
4566 places). Distributions which have not converted these
4567 directories to symlinks should consider stealing these files
4568 from git history and add them downstream.
4570 * We introduced the Documentation= field for units and added
4571 this to all our shipped units. This is useful to make it
4572 easier to explore the boot and the purpose of the various
4575 * All smaller setup units (such as
4576 systemd-vconsole-setup.service) now detect properly if they
4577 are run in a container and are skipped when
4578 appropriate. This guarantees an entirely noise-free boot in
4579 Linux container environments such as systemd-nspawn.
4581 * A framework for implementing offline system updates is now
4582 integrated, for details see:
4583 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
4585 * A new service type Type=idle is available now which helps us
4586 avoiding ugly interleaving of getty output and boot status
4589 * There's now a system-wide CapabilityBoundingSet= option to
4590 globally reduce the set of capabilities for the
4591 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
4592 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
4593 even CAP_NET_ADMIN system-wide for secure systems.
4595 * There are now system-wide DefaultLimitXXX= options to
4596 globally change the defaults of the various resource limits
4597 for all units started by PID 1.
4599 * Harald Hoyer's systemd test suite has been integrated into
4600 systemd which allows easy testing of systemd builds in qemu
4601 and nspawn. (This is really awesome! Ask us for details!)
4603 * The fstab parser is now implemented as generator, not inside
4606 * systemctl will now warn you if .mount units generated from
4607 /etc/fstab are out of date due to changes in fstab that
4608 have not been read by systemd yet.
4610 * systemd is now suitable for usage in initrds. Dracut has
4611 already been updated to make use of this. With this in place
4612 initrds get a slight bit faster but primarily are much
4613 easier to introspect and debug since "systemctl status" in
4614 the host system can be used to introspect initrd services,
4615 and the journal from the initrd is kept around too.
4617 * systemd-delta has been added, a tool to explore differences
4618 between user/admin configuration and vendor defaults.
4620 * PrivateTmp= now affects both /tmp and /var/tmp.
4622 * Boot time status messages are now much prettier and feature
4623 proper english language. Booting up systemd has never been
4626 * Read-ahead pack files now include the inode number of all
4627 files to pre-cache. When the inode changes the pre-caching
4628 is not attempted. This should be nicer to deal with updated
4629 packages which might result in changes of read-ahead
4632 * We now temporaritly lower the kernel's read_ahead_kb variable
4633 when collecting read-ahead data to ensure the kernel's
4634 built-in read-ahead does not add noise to our measurements
4635 of necessary blocks to pre-cache.
4637 * There's now RequiresMountsFor= to add automatic dependencies
4638 for all mounts necessary for a specific file system path.
4640 * MountAuto= and SwapAuto= have been removed from
4641 system.conf. Mounting file systems at boot has to take place
4644 * nspawn now learned a new switch --uuid= to set the machine
4645 ID on the command line.
4647 * nspawn now learned the -b switch to automatically search
4650 * vt102 is now the default TERM for serial TTYs, upgraded from
4653 * systemd-logind now works on VT-less systems.
4655 * The build tree has been reorganized. The individual
4656 components now have directories of their own.
4658 * A new condition type ConditionPathIsReadWrite= is now available.
4660 * nspawn learned the new -C switch to create cgroups for the
4661 container in other hierarchies.
4663 * We now have support for hardware watchdogs, configurable in
4666 * The scheduled shutdown logic now has a public API.
4668 * We now mount /tmp as tmpfs by default, but this can be
4669 masked and /etc/fstab can override it.
4671 * Since udisks does not make use of /media anymore we are not
4672 mounting a tmpfs on it anymore.
4674 * journalctl gained a new --local switch to only interleave
4675 locally generated journal files.
4677 * We can now load the IMA policy at boot automatically.
4679 * The GTK tools have been split off into a systemd-ui.
4681 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
4682 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
4683 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
4684 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
4685 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
4686 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
4687 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
4688 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
4689 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
4694 * This is mostly a bugfix release
4696 * Support optional initialization of the machine ID from the
4697 KVM or container configured UUID.
4699 * Support immediate reboots with "systemctl reboot -ff"
4701 * Show /etc/os-release data in systemd-analyze output
4703 * Many bugfixes for the journal, including endianness fixes and
4704 ensuring that disk space enforcement works
4706 * sd-login.h is C++ comptaible again
4708 * Extend the /etc/os-release format on request of the Debian
4711 * We now refuse non-UTF8 strings used in various configuration
4712 and unit files. This is done to ensure we do not pass invalid
4713 data over D-Bus or expose it elsewhere.
4715 * Register Mimo USB Screens as suitable for automatic seat
4718 * Read SELinux client context from journal clients in a race
4721 * Reorder configuration file lookup order. /etc now always
4722 overrides /run in order to allow the administrator to always
4723 and unconditionally override vendor supplied or
4724 automatically generated data.
4726 * The various user visible bits of the journal now have man
4727 pages. We still lack man pages for the journal API calls
4730 * We now ship all man pages in HTML format again in the
4733 Contributions from: Dave Reisner, Dirk Eibach, Frederic
4734 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
4735 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
4740 * This is mostly a bugfix release
4742 * systems lacking /etc/os-release are no longer supported.
4744 * Various functionality updates to libsystemd-login.so
4746 * Track class of PAM logins to distuingish greeters from
4749 Contributions from: Kay Sievers, Lennart Poettering, Michael
4754 * This is an important bugfix release for v41.
4756 * Building man pages is now optional which should be useful
4757 for those building systemd from git but unwilling to install
4760 * Watchdog support for supervising services is now usable. In
4761 a future release support for hardware watchdogs
4762 (i.e. /dev/watchdog) will be added building on this.
4764 * Service start rate limiting is now configurable and can be
4765 turned off per service. When a start rate limit is hit a
4766 reboot can automatically be triggered.
4768 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
4770 Contributions from: Benjamin Franzke, Bill Nottingham,
4771 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
4772 Schmidt, Michał Górny, Piotr Drąg
4776 * The systemd binary is installed /usr/lib/systemd/systemd now;
4777 An existing /sbin/init symlink needs to be adapted with the
4780 * The code that loads kernel modules has been ported to invoke
4781 libkmod directly, instead of modprobe. This means we do not
4782 support systems with module-init-tools anymore.
4784 * Watchdog support is now already useful, but still not
4787 * A new kernel command line option systemd.setenv= is
4788 understood to set system wide environment variables
4789 dynamically at boot.
4791 * We now limit the set of capabilities of systemd-journald.
4793 * We now set SIGPIPE to ignore by default, since it only is
4794 useful in shell pipelines, and has little use in general
4795 code. This can be disabled with IgnoreSIPIPE=no in unit
4798 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
4799 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
4804 * This is mostly a bugfix release
4806 * We now expose the reason why a service failed in the
4807 "Result" D-Bus property.
4809 * Rudimentary service watchdog support (will be completed over
4810 the next few releases.)
4812 * When systemd forks off in order execute some service we will
4813 now immediately changes its argv[0] to reflect which process
4814 it will execute. This is useful to minimize the time window
4815 with a generic argv[0], which makes bootcharts more useful
4817 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
4818 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
4819 Mike Kazantsev, Ray Strode
4823 * This is mostly a test release, but incorporates many
4826 * New systemd-cgtop tool to show control groups by their
4829 * Linking against libacl for ACLs is optional again. If
4830 disabled, support tracking device access for active logins
4831 goes becomes unavailable, and so does access to the user
4832 journals by the respective users.
4834 * If a group "adm" exists, journal files are automatically
4835 owned by them, thus allow members of this group full access
4836 to the system journal as well as all user journals.
4838 * The journal now stores the SELinux context of the logging
4839 client for all entries.
4841 * Add C++ inclusion guards to all public headers
4843 * New output mode "cat" in the journal to print only text
4844 messages, without any meta data like date or time.
4846 * Include tiny X server wrapper as a temporary stop-gap to
4847 teach XOrg udev display enumeration. This is used by display
4848 managers such as gdm, and will go away as soon as XOrg
4849 learned native udev hotplugging for display devices.
4851 * Add new systemd-cat tool for executing arbitrary programs
4852 with STDERR/STDOUT connected to the journal. Can also act as
4853 BSD logger replacement, and does so by default.
4855 * Optionally store all locally generated coredumps in the
4856 journal along with meta data.
4858 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
4859 writing short strings to files (for usage for /sys), and for
4860 creating symlinks, character and block device nodes.
4862 * New unit file option ControlGroupPersistent= to make cgroups
4863 persistent, following the mechanisms outlined in
4864 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
4866 * Support multiple local RTCs in a sane way
4868 * No longer monopolize IO when replaying readahead data on
4869 rotating disks, since we might starve non-file-system IO to
4870 death, since fanotify() will not see accesses done by blkid,
4873 * Do not show kernel threads in systemd-cgls anymore, unless
4874 requested with new -k switch.
4876 Contributions from: Dan Horák, Kay Sievers, Lennart
4877 Poettering, Michal Schmidt
4881 * This is mostly a test release, but incorporates many
4884 * The git repository moved to:
4885 git://anongit.freedesktop.org/systemd/systemd
4886 ssh://git.freedesktop.org/git/systemd/systemd
4888 * First release with the journal
4889 http://0pointer.de/blog/projects/the-journal.html
4891 * The journal replaces both systemd-kmsg-syslogd and
4892 systemd-stdout-bridge.
4894 * New sd_pid_get_unit() API call in libsystemd-logind
4896 * Many systemadm clean-ups
4898 * Introduce remote-fs-pre.target which is ordered before all
4899 remote mounts and may be used to start services before all
4902 * Added Mageia support
4904 * Add bash completion for systemd-loginctl
4906 * Actively monitor PID file creation for daemons which exit in
4907 the parent process before having finished writing the PID
4908 file in the daemon process. Daemons which do this need to be
4909 fixed (i.e. PID file creation must have finished before the
4910 parent exits), but we now react a bit more gracefully to them.
4912 * Add colourful boot output, mimicking the well-known output
4913 of existing distributions.
4915 * New option PassCredentials= for socket units, for
4916 compatibility with a recent kernel ABI breakage.
4918 * /etc/rc.local is now hooked in via a generator binary, and
4919 thus will no longer act as synchronization point during
4922 * systemctl list-unit-files now supports --root=.
4924 * systemd-tmpfiles now understands two new commands: z, Z for
4925 relabelling files according to the SELinux database. This is
4926 useful to apply SELinux labels to specific files in /sys,
4929 * Output of SysV services is now forwarded to both the console
4930 and the journal by default, not only just the console.
4932 * New man pages for all APIs from libsystemd-login.
4934 * The build tree got reorganized and a the build system is a
4935 lot more modular allowing embedded setups to specifically
4936 select the components of systemd they are interested in.
4938 * Support for Linux systems lacking the kernel VT subsystem is
4941 * configure's --with-rootdir= got renamed to
4942 --with-rootprefix= to follow the naming used by udev and
4945 * Unless specified otherwise we will now install to /usr instead
4946 of /usr/local by default.
4948 * Processes with '@' in argv[0][0] are now excluded from the
4949 final shut-down killing spree, following the logic explained
4951 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
4953 * All processes remaining in a service cgroup when we enter
4954 the START or START_PRE states are now killed with
4955 SIGKILL. That means it is no longer possible to spawn
4956 background processes from ExecStart= lines (which was never
4957 supported anyway, and bad style).
4959 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
4960 reloading of units together.
4962 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
4963 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
4964 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
4965 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
4966 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek