1 systemd System and Service Manager
5 * systemd now exposes a CPUUsageNSec= property for each
6 service unit on the bus, that contains the overall consumed
7 CPU time of a service (the sum of what each process of the
8 service consumed). This value is only available if
9 CPUAccounting= is turned on for a service, and is then shown
10 in the "systemctl status" output.
12 * Support for configuring alternative mappings of the old SysV
13 runlevels to systemd targets has been removed. They are now
14 hardcoded in a way that runlevels 2, 3, 4 all map to
15 multi-user.target and 5 to graphical.target (which
16 previously was already the default behaviour).
18 * The auto-mounter logic gained support for mount point
19 expiry, using a new TimeoutIdleSec= setting in .automount
20 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
22 * The EFI System Partition (ESP) as mounted to /boot by
23 systemd-efi-boot-generator will now be unmounted
24 automatically after 2 minutes of not being used. This should
25 minimize the risk of ESP corruptions.
27 * New /etc/fstab options x-systemd.requires= and
28 x-systemd.requires-mounts-for= are now supported to express
29 additional dependencies for mounts. This is useful for
30 journalling file systems that support external journal
31 devices or overlay file systems that require underlying file
32 systems to be mounted.
34 * systemd does not support direct live-upgrades (via systemctl
35 daemon-reexec) from versions older than v44 anymore. As no
36 distribution we are aware of shipped such old versions in a
37 stable release this should not be problematic.
39 * When systemd forks off a new per-connection service instance
40 it will now set the $REMOTE_ADDR environment variable to the
41 remote IP address, and $REMOTE_PORT environment variable to
42 the remote IP port. This behaviour is similar to the
43 corresponding environment variables defined by CGI.
45 * systemd-networkd gained support for uplink failure
46 detection. The BindCarrier= option allows binding interface
47 configuration dynamically to the link sense of other
48 interfaces. This is useful to achieve behaviour like in
51 * systemd-networkd gained support for configuring the DHCP
52 client identifier to use when requesting leases.
54 * systemd-networkd now has a per-network UseNTP= option to
55 configure whether NTP server information acquired via DHCP
56 is passed on to services like systemd-timesyncd.
58 * systemd-networkd gained support for vti6 tunnels.
60 * Note that systemd-networkd manages the sysctl variable
61 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
62 it is configured for since v219. The variable controls IP
63 forwarding, and is a per-interface alternative to the global
64 /proc/sys/net/ipv[46]/ip_forward. This setting is
65 configurable in the IPForward= option, which defaults to
66 "no". This means if networkd is used for an interface it is
67 no longer sufficient to set the global sysctl option to turn
68 on IP forwarding! Instead, the .network file option
69 IPForward= needs to be turned on! Note that the
70 implementation of this behaviour was broken in v219 and has
73 * Many bonding and vxlan options are now configurable in
76 * systemd-nspawn gained a new --property= setting to set unit
77 properties for the container scope. This is useful for
78 setting resource parameters (e.g "CPUShares=500") on
79 containers started from the command line.
81 * systemd-nspawn gained a new --private-users= switch to make
82 use of user namespacing available on recent Linux kernels.
84 * systemd-nspawn may now be called as part of a shell pipeline
85 in which case the pipes used for stdin and stdout are passed
86 directly to the process invoked in the container, without
87 indirection via a pseudo tty.
89 * systemd-nspawn gained a new switch to control the UNIX
90 signal to use when killing the init process of the container
93 * systemd-nspawn gained a new --overlay= switch for mounting
94 overlay file systems into the container using the new kernel
97 * When a container image is imported via systemd-importd and
98 the host file system is not btrfs, a loopback block device
99 file is created in /var/lib/machines.raw with a btrfs file
100 system inside. It is then mounted to /var/lib/machines to
101 enable btrfs features for container management. The loopback
102 file and btrfs file system is grown as needed when container
103 images are imported via systemd-importd.
105 * systemd-machined/systemd-importd gained support for btrfs
106 quota, to enforce container disk space limits on disk. This
107 is exposed in "machinectl set-limit".
109 * systemd-importd now can import containers from local .tar,
110 .raw and .qcow2 images, and export them to .tar and .raw. It
111 can also import dkr v2 images now from the network (on top
114 * systemd-importd gained support for verifying downloaded
115 images with gpg2 (previously only gpg1 was supported).
117 * systemd-machined, systemd-logind, systemd: most bus calls
118 are now accessible to unprivileged processes via
119 PolicyKit. Also, systemd-logind will now allow users to kill
120 their own sessions without further privileges or
123 * systemd-shutdownd has been removed. This service was
124 previously responsible for implementing scheduled shutdowns
125 as exposed in /usr/bin/shutdown's time parameter. This
126 functionality has now been moved into systemd-logind and is
127 accessible via a bus interface.
129 * "systemctl reboot" gained a new switch --firmware-setup that
130 can be used to reboot into the EFI firmware setup, if that
131 is available. systemd-logind now exposes an API on the bus
132 to trigger such reboots, in case graphical desktop UIs want
133 to cover this functionality.
135 * "systemctl enable", "systemctl disable" and "systemctl mask"
136 now support a new "--now" switch. If specified the units
137 that are enabled will also be started, and the ones
138 disabled/masked also stopped.
140 * The Gummiboot EFI boot loader tool has been merged into
141 systemd, and renamed to "systemd-boot". The bootctl tool has been
142 updated to support systemd-boot.
144 * An EFI kernel stub has been added that may be used to create
145 kernel EFI binaries that contain not only the actual kernel,
146 but also an initrd, boot splash, command line and OS release
147 information. This combined binary can then be signed as a
148 single image, so that the firmware can verify it all in one
149 step. systemd-boot has special support for EFI binaries created
150 like this and can extract OS release information from them
151 and show them in the boot menu. This functionality is useful
152 to implement cryptographically verified boot schemes.
154 * Optional support has been added to systemd-fsck to pass
155 fsck's progress report to an AF_UNIX socket in the file
158 * udev will no longer create device symlinks for all block
159 devices by default. A blacklist for excluding special block
160 devices from this logic has been turned into a whitelist
161 that requires picking block devices explicitly that require
164 * A new (currently still internal) API sd-device.h has been
165 added to libsystemd. This modernized API is supposed to
166 replace libudev eventually. In fact, already much of libudev
167 is now just a wrapper around sd-device.h.
169 * A new hwdb database for storing metadata about pointing
170 stick devices has been added.
172 * systemd-tmpfiles gained support for setting file attributes
173 similar to the "chattr" tool with new 'h' and 'H' lines.
175 * systemd-journald will no longer unconditionally set the
176 btrfs NOCOW flag on new journal files. This is instead done
177 with tmpfiles snippet using the new 'h' line type. This
178 allows easy disabling of this logic, by masking the
179 journal-nocow.conf tmpfiles file.
181 * systemd-journald will now translate audit message types to
182 human readable identifiers when writing them to the
183 journal. This should improve readability of audit messages.
185 * The LUKS logic gained support for the offset= and skip=
186 options in /etc/crypttab, as previously implemented by
189 * /usr/lib/os-release gained a new optional field VARIANT= for
190 distributions that support multiple variants (such as a
191 desktop edition, a server edition, ...)
193 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
194 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
195 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
196 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
197 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
198 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
199 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
200 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
201 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
202 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
203 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
204 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
205 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
206 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
207 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
208 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
209 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
210 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
211 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
212 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
213 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
214 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
215 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
216 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
217 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
218 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
219 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
221 -- Berlin, 2015-05-??
225 * Introduce a new API "sd-hwdb.h" for querying the hardware
226 metadata database. With this minimal interface one can query
227 and enumerate the udev hwdb, decoupled from the old libudev
228 library. libudev's interface for this is now only a wrapper
229 around sd-hwdb. A new tool systemd-hwdb has been added to
230 interface with and update the database.
232 * When any of systemd's tools copies files (for example due to
233 tmpfiles' C lines) a btrfs reflink will attempted first,
234 before bytewise copying is done.
236 * systemd-nspawn gained a new --ephemeral switch. When
237 specified a btrfs snapshot is taken of the container's root
238 directory, and immediately removed when the container
239 terminates again. Thus, a container can be started whose
240 changes never alter the container's root directory, and are
241 lost on container termination. This switch can also be used
242 for starting a container off the root file system of the
243 host without affecting the host OS. This switch is only
244 available on btrfs file systems.
246 * systemd-nspawn gained a new --template= switch. It takes the
247 path to a container tree to use as template for the tree
248 specified via --directory=, should that directory be
249 missing. This allows instantiating containers dynamically,
250 on first run. This switch is only available on btrfs file
253 * When a .mount unit refers to a mount point on which multiple
254 mounts are stacked, and the .mount unit is stopped all of
255 the stacked mount points will now be unmounted until no
258 * systemd now has an explicit notion of supported and
259 unsupported unit types. Jobs enqueued for unsupported unit
260 types will now fail with an "unsupported" error code. More
261 specifically .swap, .automount and .device units are not
262 supported in containers, .busname units are not supported on
263 non-kdbus systems. .swap and .automount are also not
264 supported if their respective kernel compile time options
267 * machinectl gained support for two new "copy-from" and
268 "copy-to" commands for copying files from a running
269 container to the host or vice versa.
271 * machinectl gained support for a new "bind" command to bind
272 mount host directories into local containers. This is
273 currently only supported for nspawn containers.
275 * networkd gained support for configuring bridge forwarding
276 database entries (fdb) from .network files.
278 * A new tiny daemon "systemd-importd" has been added that can
279 download container images in tar, raw, qcow2 or dkr formats,
280 and make them available locally in /var/lib/machines, so
281 that they can run as nspawn containers. The daemon can GPG
282 verify the downloads (not supported for dkr, since it has no
283 provisions for verifying downloads). It will transparently
284 decompress bz2, xz, gzip compressed downloads if necessary,
285 and restore sparse files on disk. The daemon uses privilege
286 separation to ensure the actual download logic runs with
287 fewer privileges than the deamon itself. machinectl has
288 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
289 make the functionality of importd available to the
290 user. With this in place the Fedora and Ubuntu "Cloud"
291 images can be downloaded and booted as containers unmodified
292 (the Fedora images lack the appropriate GPG signature files
293 currently, so they cannot be verified, but this will change
294 soon, hopefully). Note that downloading images is currently
295 only fully supported on btrfs.
297 * machinectl is now able to list container images found in
298 /var/lib/machines, along with some metadata about sizes of
299 disk and similar. If the directory is located on btrfs and
300 quota is enabled, this includes quota display. A new command
301 "image-status" has been added that shows additional
302 information about images.
304 * machinectl is now able to clone container images
305 efficiently, if the underlying file system (btrfs) supports
306 it, with the new "machinectl list-images" command. It also
307 gained commands for renaming and removing images, as well as
308 marking them read-only or read-write (supported also on
309 legacy file systems).
311 * networkd gained support for collecting LLDP network
312 announcements, from hardware that supports this. This is
313 shown in networkctl output.
315 * systemd-run gained support for a new -t (--pty) switch for
316 invoking a binary on a pty whose input and output is
317 connected to the invoking terminal. This allows executing
318 processes as system services while interactively
319 communicating with them via the terminal. Most interestingly
320 this is supported across container boundaries. Invoking
321 "systemd-run -t /bin/bash" is an alternative to running a
322 full login session, the difference being that the former
323 will not register a session, nor go through the PAM session
326 * tmpfiles gained support for a new "v" line type for creating
327 btrfs subvolumes. If the underlying file system is a legacy
328 file system, this automatically degrades to creating a
329 normal directory. Among others /var/lib/machines is now
330 created like this at boot, should it be missing.
332 * The directory /var/lib/containers/ has been deprecated and
333 been replaced by /var/lib/machines. The term "machines" has
334 been used in the systemd context as generic term for both
335 VMs and containers, and hence appears more appropriate for
336 this, as the directory can also contain raw images bootable
339 * systemd-nspawn when invoked with -M but without --directory=
340 or --image= is now capable of searching for the container
341 root directory, subvolume or disk image automatically, in
342 /var/lib/machines. systemd-nspawn@.service has been updated
343 to make use of this, thus allowing it to be used for raw
346 * A new machines.target unit has been introduced that is
347 supposed to group all containers/VMs invoked as services on
348 the system. systemd-nspawn@.service has been updated to
351 * machinectl gained a new "start" command, for invoking a
352 container as a service. "machinectl start foo" is mostly
353 equivalent to "systemctl start systemd-nspawn@foo.service",
354 but handles escaping in a nicer way.
356 * systemd-nspawn will now mount most of the cgroupfs tree
357 read-only into each container, with the exception of the
358 container's own subtree in the name=systemd hierarchy.
360 * journald now sets the special FS_NOCOW file flag for its
361 journal files. This should improve performance on btrfs, by
362 avoiding heavy fragmentation when journald's write-pattern
363 is used on COW file systems. It degrades btrfs' data
364 integrity guarantees for the files to the same levels as for
365 ext3/ext4 however. This should be OK though as journald does
366 its own data integrity checks and all its objects are
367 checksummed on disk. Also, journald should handle btrfs disk
368 full events a lot more gracefully now, by processing SIGBUS
369 errors, and not relying on fallocate() anymore.
371 * When journald detects that journal files it is writing to
372 have been deleted it will immediately start new journal
375 * systemd now provides a way to store file descriptors
376 per-service in PID 1.This is useful for daemons to ensure
377 that fds they require are not lost during a daemon
378 restart. The fds are passed to the deamon on the next
379 invocation in the same way socket activation fds are
380 passed. This is now used by journald to ensure that the
381 various sockets connected to all the system's stdout/stderr
382 are not lost when journald is restarted. File descriptors
383 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
384 an extension to sd_notify(). Note that a limit is enforced
385 on the number of fds a service can store in PID 1, and it
386 defaults to 0, so that no fds may be stored, unless this is
387 explicitly turned on.
389 * The default TERM variable to use for units connected to a
390 terminal, when no other value is explicitly is set is now
391 vt220 rather than vt102. This should be fairly safe still,
392 but allows PgUp/PgDn work.
394 * The /etc/crypttab option header= as known from Debian is now
397 * "loginctl user-status" and "loginctl session-status" will
398 now show the last 10 lines of log messages of the
399 user/session following the status output. Similar,
400 "machinectl status" will show the last 10 log lines
401 associated with a virtual machine or container
402 service. (Note that this is usually not the log messages
403 done in the VM/container itself, but simply what the
404 container manager logs. For nspawn this includes all console
407 * "loginctl session-status" without further argument will now
408 show the status of the session of the caller. Similar,
409 "lock-session", "unlock-session", "activate",
410 "enable-linger", "disable-linger" may now be called without
411 session/user parameter in which case they apply to the
412 caller's session/user.
414 * An X11 session scriptlet is now shipped that uploads
415 $DISPLAY and $XAUTHORITY into the environment of the systemd
416 --user daemon if a session begins. This should improve
417 compatibility with X11 enabled applications run as systemd
420 * Generators are now subject to masking via /etc and /run, the
421 same way as unit files.
423 * networkd .network files gained support for configuring
424 per-link IPv4/IPv6 packet forwarding as well as IPv4
425 masquerading. This is by default turned on for veth links to
426 containers, as registered by systemd-nspawn. This means that
427 nspawn containers run with --network-veth will now get
428 automatic routed access to the host's networks without any
429 further configuration or setup, as long as networkd runs on
432 * systemd-nspawn gained the --port= (-p) switch to expose TCP
433 or UDP posts of a container on the host. With this in place
434 it is possible to run containers with private veth links
435 (--network-veth), and have their functionality exposed on
436 the host as if their services were running directly on the
439 * systemd-nspawn's --network-veth switch now gained a short
440 version "-n", since with the changes above it is now truly
441 useful out-of-the-box. The systemd-nspawn@.service has been
442 updated to make use of it too by default.
444 * systemd-nspawn will now maintain a per-image R/W lock, to
445 ensure that the same image is not started more than once
446 writable. (It's OK to run an image multiple times
447 simultaneously in read-only mode.)
449 * systemd-nspawn's --image= option is now capable of
450 dissecting and booting MBR and GPT disk images that contain
451 only a single active Linux partition. Previously it
452 supported only GPT disk images with proper GPT type
453 IDs. This allows running cloud images from major
454 distributions directly with systemd-nspawn, without
457 * In addition to collecting mouse dpi data in the udev
458 hardware database, there's now support for collecting angle
459 information for mouse scroll wheels. The database is
460 supposed to guarantee similar scrolling behavior on mice
461 that it knows about. There's also support for collecting
462 information about Touchpad types.
464 * udev's input_id built-in will now also collect touch screen
465 dimension data and attach it to probed devices.
467 * /etc/os-release gained support for a Distribution Privacy
470 * networkd gained support for creating "ipvlan", "gretap",
471 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
473 * systemd-tmpfiles gained support for "a" lines for setting
476 * systemd-nspawn will now mount /tmp in the container to
477 tmpfs, automatically.
479 * systemd now exposes the memory.usage_in_bytes cgroup
480 attribute and shows it for each service in the "systemctl
481 status" output, if available.
483 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
484 immediate reboot is triggered. This useful if shutdown is
485 hung and is unable to complete, to expedite the
486 operation. Note that this kind of reboot will still unmount
487 all file systems, and hence should not result in fsck being
490 * A .device unit for an optical block device will now be
491 considered active only when a medium is in the drive. Also,
492 mount units are now bound to their backing devices thus
493 triggering automatic unmounting when devices become
494 unavailable. With this in place systemd will now
495 automatically unmount left-over mounts when a CD-ROM is
496 ejected or an USB stick is yanked from the system.
498 * networkd-wait-online now has support for waiting for
499 specific interfaces only (with globbing), and for giving up
500 after a configurable timeout.
502 * networkd now exits when idle. It will be automatically
503 restarted as soon as interfaces show up, are removed or
504 change state. networkd will stay around as long as there is
505 at least one DHCP state machine or similar around, that keep
508 * networkd may now configure IPv6 link-local addressing in
509 addition to IPv4 link-local addressing.
511 * The IPv6 "token" for use in SLAAC may now be configured for
512 each .network interface in networkd.
514 * Routes configured with networkd may now be assigned a scope
517 * networkd's [Match] sections now support globbing and lists
518 of multiple space-separated matches per item.
520 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
521 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
522 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
523 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
524 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
525 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
526 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
527 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
528 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
529 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
530 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
531 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
532 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
533 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
534 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
535 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
536 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
537 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
538 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
539 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
540 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
541 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
542 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
543 Hoffmann, Zbigniew Jędrzejewski-Szmek
545 -- Berlin, 2015-02-16
549 * When querying unit file enablement status (for example via
550 "systemctl is-enabled"), a new state "indirect" is now known
551 which indicates that a unit might not be enabled itself, but
552 another unit listed in its Alias= setting might be.
554 * Similar to the various existing ConditionXYZ= settings for
555 units there are now matching AssertXYZ= settings. While
556 failing conditions cause a unit to be skipped, but its job
557 to succeed, failing assertions declared like this will cause
558 a unit start operation and its job to fail.
560 * hostnamed now knows a new chassis type "embedded".
562 * systemctl gained a new "edit" command. When used on a unit
563 file this allows extending unit files with .d/ drop-in
564 configuration snippets or editing the full file (after
565 copying it from /usr/lib to /etc). This will invoke the
566 user's editor (as configured with $EDITOR), and reload the
567 modified configuration after editing.
569 * "systemctl status" now shows the suggested enablement state
570 for a unit, as declared in the (usually vendor-supplied)
573 * nss-myhostname will now resolve the single-label host name
574 "gateway" to the locally configured default IP routing
575 gateways, ordered by their metrics. This assigns a stable
576 name to the used gateways, regardless which ones are
577 currently configured. Note that the name will only be
578 resolved after all other name sources (if nss-myhostname is
579 configured properly) and should hence not negatively impact
580 systems that use the single-label host name "gateway" in
583 * systemd-inhibit now allows filtering by mode when listing
586 * Scope and service units gained a new "Delegate" boolean
587 property, which when set allows processes running inside the
588 unit to further partition resources. This is primarily
589 useful for systemd user instances as well as container
592 * journald will now pick up audit messages directly from
593 the kernel, and log them like any other log message. The
594 audit fields are split up and fully indexed. This means that
595 journalctl in many ways is now a (nicer!) alternative to
596 ausearch, the traditional audit client. Note that this
597 implements only a minimal audit client, if you want the
598 special audit modes like reboot-on-log-overflow, please use
599 the traditional auditd instead, which can be used in
600 parallel to journald.
602 * The ConditionSecurity= unit file option now understands the
603 special string "audit" to check whether auditing is
606 * journalctl gained two new commands --vacuum-size= and
607 --vacuum-time= to delete old journal files until the
608 remaining ones take up no more the specified size on disk,
609 or are not older than the specified time.
611 * A new, native PPPoE library has been added to sd-network,
612 systemd's library of light-weight networking protocols. This
613 library will be used in a future version of networkd to
614 enable PPPoE communication without an external pppd daemon.
616 * The busctl tool now understands a new "capture" verb that
617 works similar to "monitor", but writes a packet capture
618 trace to STDOUT that can be redirected to a file which is
619 compatible with libcap's capture file format. This can then
620 be loaded in Wireshark and similar tools to inspect bus
623 * The busctl tool now understands a new "tree" verb that shows
624 the object trees of a specific service on the bus, or of all
627 * The busctl tool now understands a new "introspect" verb that
628 shows all interfaces and members of objects on the bus,
629 including their signature and values. This is particularly
630 useful to get more information about bus objects shown by
631 the new "busctl tree" command.
633 * The busctl tool now understands new verbs "call",
634 "set-property" and "get-property" for invoking bus method
635 calls, setting and getting bus object properties in a
638 * busctl gained a new --augment-creds= argument that controls
639 whether the tool shall augment credential information it
640 gets from the bus with data from /proc, in a possibly
643 * nspawn's --link-journal= switch gained two new values
644 "try-guest" and "try-host" that work like "guest" and
645 "host", but do not fail if the host has no persistent
646 journalling enabled. -j is now equivalent to
647 --link-journal=try-guest.
649 * macvlan network devices created by nspawn will now have
650 stable MAC addresses.
652 * A new SmackProcessLabel= unit setting has been added, which
653 controls the SMACK security label processes forked off by
654 the respective unit shall use.
656 * If compiled with --enable-xkbcommon, systemd-localed will
657 verify x11 keymap settings by compiling the given keymap. It
658 will spew out warnings if the compilation fails. This
659 requires libxkbcommon to be installed.
661 * When a coredump is collected a larger number of metadata
662 fields is now collected and included in the journal records
663 created for it. More specifically control group membership,
664 environment variables, memory maps, working directory,
665 chroot directory, /proc/$PID/status, and a list of open file
666 descriptors is now stored in the log entry.
668 * The udev hwdb now contains DPI information for mice. For
671 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
673 * All systemd programs that read standalone configuration
674 files in /etc now also support a corresponding series of
675 .conf.d configuration directories in /etc/, /run/,
676 /usr/local/lib/, /usr/lib/, and (if configured with
677 --enable-split-usr) /lib/. In particular, the following
678 configuration files now have corresponding configuration
679 directories: system.conf user.conf, logind.conf,
680 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
681 resolved.conf, timesyncd.conf, journal-remote.conf, and
682 journal-upload.conf. Note that distributions should use the
683 configuration directories in /usr/lib/; the directories in
684 /etc/ are reserved for the system administrator.
686 * systemd-rfkill will no longer take the rfkill device name
687 into account when storing rfkill state on disk, as the name
688 might be dynamically assigned and not stable. Instead, the
689 ID_PATH udev variable combined with the rfkill type (wlan,
690 bluetooth, ...) is used.
692 * A new service systemd-machine-id-commit.service has been
693 added. When used on systems where /etc is read-only during
694 boot, and /etc/machine-id is not initialized (but an empty
695 file), this service will copy the temporary machine ID
696 created as replacement into /etc after the system is fully
697 booted up. This is useful for systems that are freshly
698 installed with a non-initialized machine ID, but should get
699 a fixed machine ID for subsequent boots.
701 * networkd's .netdev files now provide a large set of
702 configuration parameters for VXLAN devices. Similar, the
703 bridge port cost parameter is now configurable in .network
704 files. There's also new support for configuring IP source
705 routing. networkd .link files gained support for a new
706 OriginalName= match that is useful to match against the
707 original interface name the kernel assigned. .network files
708 may include MTU= and MACAddress= fields for altering the MTU
709 and MAC address while being connected to a specific network
712 * The LUKS logic gained supported for configuring
713 UUID-specific key files. There's also new support for naming
714 LUKS device from the kernel command line, using the new
717 * Timer units may now be transiently created via the bus API
718 (this was previously already available for scope and service
719 units). In addition it is now possible to create multiple
720 transient units at the same time with a single bus call. The
721 "systemd-run" tool has been updated to make use of this for
722 running commands on a specified time, in at(1)-style.
724 * tmpfiles gained support for "t" lines, for assigning
725 extended attributes to files. Among other uses this may be
726 used to assign SMACK labels to files.
728 Contributions from: Alin Rauta, Alison Chaiken, Andrej
729 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
730 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
731 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
732 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
733 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
734 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
735 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
736 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
737 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
738 Michael Biebl, Michael Chapman, Michael Marineau, Michal
739 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
740 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
741 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
742 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
743 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
744 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
745 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
747 -- Berlin, 2014-12-10
751 * journalctl gained the new options -t/--identifier= to match
752 on the syslog identifier (aka "tag"), as well as --utc to
753 show log timestamps in the UTC timezone. journalctl now also
754 accepts -n/--lines=all to disable line capping in a pager.
756 * journalctl gained a new switch, --flush, that synchronously
757 flushes logs from /run/log/journal to /var/log/journal if
758 persistent storage is enabled. systemd-journal-flush.service
759 now waits until the operation is complete.
761 * Services can notify the manager before they start a reload
762 (by sending RELOADING=1) or shutdown (by sending
763 STOPPING=1). This allows the manager to track and show the
764 internal state of daemons and closes a race condition when
765 the process is still running but has closed its D-Bus
768 * Services with Type=oneshot do not have to have any ExecStart
771 * User units are now loaded also from
772 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
773 /run/systemd/user directory that was already previously
774 supported, but is under the control of the user.
776 * Job timeouts (i.e. time-outs on the time a job that is
777 queued stays in the run queue) can now optionally result in
778 immediate reboot or power-off actions (JobTimeoutAction= and
779 JobTimeoutRebootArgument=). This is useful on ".target"
780 units, to limit the maximum time a target remains
781 undispatched in the run queue, and to trigger an emergency
782 operation in such a case. This is now used by default to
783 turn off the system if boot-up (as defined by everything in
784 basic.target) hangs and does not complete for at least
785 15min. Also, if power-off or reboot hang for at least 30min
786 an immediate power-off/reboot operation is triggered. This
787 functionality is particularly useful to increase reliability
788 on embedded devices, but also on laptops which might
789 accidentally get powered on when carried in a backpack and
790 whose boot stays stuck in a hard disk encryption passphrase
793 * systemd-logind can be configured to also handle lid switch
794 events even when the machine is docked or multiple displays
795 are attached (HandleLidSwitchDocked= option).
797 * A helper binary and a service have been added which can be
798 used to resume from hibernation in the initramfs. A
799 generator will parse the resume= option on the kernel
800 command line to trigger resume.
802 * A user console daemon systemd-consoled has been
803 added. Currently, it is a preview, and will so far open a
804 single terminal on each session of the user marked as
805 Desktop=systemd-console.
807 * Route metrics can be specified for DHCP routes added by
810 * The SELinux context of socket-activated services can be set
811 from the information provided by the networking stack
812 (SELinuxContextFromNet= option).
814 * Userspace firmware loading support has been removed and
815 the minimum supported kernel version is thus bumped to 3.7.
817 * Timeout for udev workers has been increased from 1 to 3
818 minutes, but a warning will be printed after 1 minute to
819 help diagnose kernel modules that take a long time to load.
821 * Udev rules can now remove tags on devices with TAG-="foobar".
823 * systemd's readahead implementation has been removed. In many
824 circumstances it didn't give expected benefits even for
825 rotational disk drives and was becoming less relevant in the
826 age of SSDs. As none of the developers has been using
827 rotating media anymore, and nobody stepped up to actively
828 maintain this component of systemd it has now been removed.
830 * Swap units can use Options= to specify discard options.
831 Discard options specified for swaps in /etc/fstab are now
834 * Docker containers are now detected as a separate type of
837 * The Password Agent protocol gained support for queries where
838 the user input is shown, useful e.g. for user names.
839 systemd-ask-password gained a new --echo option to turn that
842 * The default sysctl.d/ snippets will now set:
844 net.core.default_qdisc = fq_codel
846 This selects Fair Queuing Controlled Delay as the default
847 queuing discipline for network interfaces. fq_codel helps
848 fight the network bufferbloat problem. It is believed to be
849 a good default with no tuning required for most workloads.
850 Downstream distributions may override this choice. On 10Gbit
851 servers that do not do forwarding, "fq" may perform better.
852 Systems without a good clocksource should use "pfifo_fast".
854 * If kdbus is enabled during build a new option BusPolicy= is
855 available for service units, that allows locking all service
856 processes into a stricter bus policy, in order to limit
857 access to various bus services, or even hide most of them
858 from the service's view entirely.
860 * networkctl will now show the .network and .link file
861 networkd has applied to a specific interface.
863 * sd-login gained a new API call sd_session_get_desktop() to
864 query which desktop environment has been selected for a
867 * UNIX utmp support is now compile-time optional to support
870 * systemctl gained two new commands "add-wants" and
871 "add-requires" for pulling in units from specific targets
874 * If the word "rescue" is specified on the kernel command line
875 the system will now boot into rescue mode (aka
876 rescue.target), which was previously available only by
877 specifying "1" or "systemd.unit=rescue.target" on the kernel
878 command line. This new kernel command line option nicely
879 mirrors the already existing "emergency" kernel command line
882 * New kernel command line options mount.usr=, mount.usrflags=,
883 mount.usrfstype= have been added that match root=, rootflags=,
884 rootfstype= but allow mounting a specific file system to
887 * The $NOTIFY_SOCKET is now also passed to control processes of
888 services, not only the main process.
890 * This version reenables support for fsck's -l switch. This
891 means at least version v2.25 of util-linux is required for
892 operation, otherwise dead-locks on device nodes may
893 occur. Again: you need to update util-linux to at least
894 v2.25 when updating systemd to v217.
896 * The "multi-seat-x" tool has been removed from systemd, as
897 its functionality has been integrated into X servers 1.16,
898 and the tool is hence redundant. It is recommended to update
899 display managers invoking this tool to simply invoke X
900 directly from now on, again.
902 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
903 message flag has been added for all of systemd's PolicyKit
904 authenticated method calls has been added. In particular
905 this now allows optional interactive authorization via
906 PolicyKit for many of PID1's privileged operations such as
907 unit file enabling and disabling.
909 * "udevadm hwdb --update" learnt a new switch "--usr" for
910 placing the rebuilt hardware database in /usr instead of
911 /etc. When used only hardware database entries stored in
912 /usr will be used, and any user database entries in /etc are
913 ignored. This functionality is useful for vendors to ship a
914 pre-built database on systems where local configuration is
915 unnecessary or unlikely.
917 * Calendar time specifications in .timer units now also
918 understand the strings "semi-annually", "quarterly" and
919 "minutely" as shortcuts (in addition to the preexisting
920 "anually", "hourly", ...).
922 * systemd-tmpfiles will now correctly create files in /dev
923 at boot which are marked for creation only at boot. It is
924 recommended to always create static device nodes with 'c!'
925 and 'b!', so that they are created only at boot and not
926 overwritten at runtime.
928 * When the watchdog logic is used for a service (WatchdogSec=)
929 and the watchdog timeout is hit the service will now be
930 terminated with SIGABRT (instead of just SIGTERM), in order
931 to make sure a proper coredump and backtrace is
932 generated. This ensures that hanging services will result in
933 similar coredump/backtrace behaviour as services that hit a
936 Contributions from: Andreas Henriksson, Andrei Borzenkov,
937 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
938 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
939 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
940 Herrmann, David Sommerseth, David Strauss, Emil Renner
941 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
942 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
943 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
944 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
945 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
946 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
947 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
948 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
949 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
950 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
951 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
952 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
953 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
954 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
955 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
956 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
959 -- Berlin, 2014-10-28
963 * timedated no longer reads NTP implementation unit names from
964 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
965 implementations should add a
967 Conflicts=systemd-timesyncd.service
969 to their unit files to take over and replace systemd's NTP
970 default functionality.
972 * systemd-sysusers gained a new line type "r" for configuring
973 which UID/GID ranges to allocate system users/groups
974 from. Lines of type "u" may now add an additional column
975 that specifies the home directory for the system user to be
976 created. Also, systemd-sysusers may now optionally read user
977 information from STDIN instead of a file. This is useful for
978 invoking it from RPM preinst scriptlets that need to create
979 users before the first RPM file is installed since these
980 files might need to be owned by them. A new
981 %sysusers_create_inline RPM macro has been introduced to do
982 just that. systemd-sysusers now updates the shadow files as
983 well as the user/group databases, which should enhance
984 compatibility with certain tools like grpck.
986 * A number of bus APIs of PID 1 now optionally consult
987 PolicyKit to permit access for otherwise unprivileged
988 clients under certain conditions. Note that this currently
989 doesn't support interactive authentication yet, but this is
990 expected to be added eventually, too.
992 * /etc/machine-info now has new fields for configuring the
993 deployment environment of the machine, as well as the
994 location of the machine. hostnamectl has been updated with
995 new command to update these fields.
997 * systemd-timesyncd has been updated to automatically acquire
998 NTP server information from systemd-networkd, which might
999 have been discovered via DHCP.
1001 * systemd-resolved now includes a caching DNS stub resolver
1002 and a complete LLMNR name resolution implementation. A new
1003 NSS module "nss-resolve" has been added which can be used
1004 instead of glibc's own "nss-dns" to resolve hostnames via
1005 systemd-resolved. Hostnames, addresses and arbitrary RRs may
1006 be resolved via systemd-resolved D-Bus APIs. In contrast to
1007 the glibc internal resolver systemd-resolved is aware of
1008 multi-homed system, and keeps DNS server and caches separate
1009 and per-interface. Queries are sent simultaneously on all
1010 interfaces that have DNS servers configured, in order to
1011 properly handle VPNs and local LANs which might resolve
1012 separate sets of domain names. systemd-resolved may acquire
1013 DNS server information from systemd-networkd automatically,
1014 which in turn might have discovered them via DHCP. A tool
1015 "systemd-resolve-host" has been added that may be used to
1016 query the DNS logic in resolved. systemd-resolved implements
1017 IDNA and automatically uses IDNA or UTF-8 encoding depending
1018 on whether classic DNS or LLMNR is used as transport. In the
1019 next releases we intend to add a DNSSEC and mDNS/DNS-SD
1020 implementation to systemd-resolved.
1022 * A new NSS module nss-mymachines has been added, that
1023 automatically resolves the names of all local registered
1024 containers to their respective IP addresses.
1026 * A new client tool "networkctl" for systemd-networkd has been
1027 added. It currently is entirely passive and will query
1028 networking configuration from udev, rtnetlink and networkd,
1029 and present it to the user in a very friendly
1030 way. Eventually, we hope to extend it to become a full
1031 control utility for networkd.
1033 * .socket units gained a new DeferAcceptSec= setting that
1034 controls the kernels' TCP_DEFER_ACCEPT sockopt for
1035 TCP. Similar, support for controlling TCP keep-alive
1036 settings has been added (KeepAliveTimeSec=,
1037 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
1038 turning off Nagle's algorithm on TCP has been added
1041 * logind learned a new session type "web", for use in projects
1042 like Cockpit which register web clients as PAM sessions.
1044 * timer units with at least one OnCalendar= setting will now
1045 be started only after timer-sync.target has been
1046 reached. This way they will not elapse before the system
1047 clock has been corrected by a local NTP client or
1048 similar. This is particular useful on RTC-less embedded
1049 machines, that come up with an invalid system clock.
1051 * systemd-nspawn's --network-veth= switch should now result in
1052 stable MAC addresses for both the outer and the inner side
1055 * systemd-nspawn gained a new --volatile= switch for running
1056 container instances with /etc or /var unpopulated.
1058 * The kdbus client code has been updated to use the new Linux
1059 3.17 memfd subsystem instead of the old kdbus-specific one.
1061 * systemd-networkd's DHCP client and server now support
1062 FORCERENEW. There are also new configuration options to
1063 configure the vendor client identifier and broadcast mode
1066 * systemd will no longer inform the kernel about the current
1067 timezone, as this is necessarily incorrect and racy as the
1068 kernel has no understanding of DST and similar
1069 concepts. This hence means FAT timestamps will be always
1070 considered UTC, similar to what Android is already
1071 doing. Also, when the RTC is configured to the local time
1072 (rather than UTC) systemd will never synchronize back to it,
1073 as this might confuse Windows at a later boot.
1075 * systemd-analyze gained a new command "verify" for offline
1076 validation of unit files.
1078 * systemd-networkd gained support for a couple of additional
1079 settings for bonding networking setups. Also, the metric for
1080 statically configured routes may now be configured. For
1081 network interfaces where this is appropriate the peer IP
1082 address may now be configured.
1084 * systemd-networkd's DHCP client will no longer request
1085 broadcasting by default, as this tripped up some networks.
1086 For hardware where broadcast is required the feature should
1087 be switched back on using RequestBroadcast=yes.
1089 * systemd-networkd will now set up IPv4LL addresses (when
1090 enabled) even if DHCP is configured successfully.
1092 * udev will now default to respect network device names given
1093 by the kernel when the kernel indicates that these are
1094 predictable. This behavior can be tweaked by changing
1095 NamePolicy= in the relevant .link file.
1097 * A new library systemd-terminal has been added that
1098 implements full TTY stream parsing and rendering. This
1099 library is supposed to be used later on for implementing a
1100 full userspace VT subsystem, replacing the current kernel
1103 * A new tool systemd-journal-upload has been added to push
1104 journal data to a remote system running
1105 systemd-journal-remote.
1107 * journald will no longer forward all local data to another
1108 running syslog daemon. This change has been made because
1109 rsyslog (which appears to be the most commonly used syslog
1110 implementation these days) no longer makes use of this, and
1111 instead pulls the data out of the journal on its own. Since
1112 forwarding the messages to a non-existent syslog server is
1113 more expensive than we assumed we have now turned this
1114 off. If you run a syslog server that is not a recent rsyslog
1115 version, you have to turn this option on again
1116 (ForwardToSyslog= in journald.conf).
1118 * journald now optionally supports the LZ4 compressor for
1119 larger journal fields. This compressor should perform much
1120 better than XZ which was the previous default.
1122 * machinectl now shows the IP addresses of local containers,
1123 if it knows them, plus the interface name of the container.
1125 * A new tool "systemd-escape" has been added that makes it
1126 easy to escape strings to build unit names and similar.
1128 * sd_notify() messages may now include a new ERRNO= field
1129 which is parsed and collected by systemd and shown among the
1130 "systemctl status" output for a service.
1132 * A new component "systemd-firstboot" has been added that
1133 queries the most basic systemd information (timezone,
1134 hostname, root password) interactively on first
1135 boot. Alternatively it may also be used to provision these
1136 things offline on OS images installed into directories.
1138 * The default sysctl.d/ snippets will now set
1140 net.ipv4.conf.default.promote_secondaries=1
1142 This has the benefit of no flushing secondary IP addresses
1143 when primary addresses are removed.
1145 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
1146 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
1147 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
1148 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
1149 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
1150 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
1151 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
1152 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
1153 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
1154 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
1155 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
1156 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
1157 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
1158 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
1159 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
1161 -- Berlin, 2014-08-19
1165 * A new tool systemd-sysusers has been added. This tool
1166 creates system users and groups in /etc/passwd and
1167 /etc/group, based on static declarative system user/group
1168 definitions in /usr/lib/sysusers.d/. This is useful to
1169 enable factory resets and volatile systems that boot up with
1170 an empty /etc directory, and thus need system users and
1171 groups created during early boot. systemd now also ships
1172 with two default sysusers.d/ files for the most basic
1173 users and groups systemd and the core operating system
1176 * A new tmpfiles snippet has been added that rebuilds the
1177 essential files in /etc on boot, should they be missing.
1179 * A directive for ensuring automatic clean-up of
1180 /var/cache/man/ has been removed from the default
1181 configuration. This line should now be shipped by the man
1182 implementation. The necessary change has been made to the
1183 man-db implementation. Note that you need to update your man
1184 implementation to one that ships this line, otherwise no
1185 automatic clean-up of /var/cache/man will take place.
1187 * A new condition ConditionNeedsUpdate= has been added that
1188 may conditionalize services to only run when /etc or /var
1189 are "older" than the vendor operating system resources in
1190 /usr. This is useful for reconstructing or updating /etc
1191 after an offline update of /usr or a factory reset, on the
1192 next reboot. Services that want to run once after such an
1193 update or reset should use this condition and order
1194 themselves before the new systemd-update-done.service, which
1195 will mark the two directories as fully updated. A number of
1196 service files have been added making use of this, to rebuild
1197 the udev hardware database, the journald message catalog and
1198 dynamic loader cache (ldconfig). The systemd-sysusers tool
1199 described above also makes use of this now. With this in
1200 place it is now possible to start up a minimal operating
1201 system with /etc empty cleanly. For more information on the
1202 concepts involved see this recent blog story:
1204 http://0pointer.de/blog/projects/stateless.html
1206 * A new system group "input" has been introduced, and all
1207 input device nodes get this group assigned. This is useful
1208 for system-level software to get access to input devices. It
1209 complements what is already done for "audio" and "video".
1211 * systemd-networkd learnt minimal DHCPv4 server support in
1212 addition to the existing DHCPv4 client support. It also
1213 learnt DHCPv6 client and IPv6 Router Solicitation client
1214 support. The DHCPv4 client gained support for static routes
1215 passed in from the server. Note that the [DHCPv4] section
1216 known in older systemd-networkd versions has been renamed to
1217 [DHCP] and is now also used by the DHCPv6 client. Existing
1218 .network files using settings of this section should be
1219 updated, though compatibility is maintained. Optionally, the
1220 client hostname may now be sent to the DHCP server.
1222 * networkd gained support for vxlan virtual networks as well
1223 as tun/tap and dummy devices.
1225 * networkd gained support for automatic allocation of address
1226 ranges for interfaces from a system-wide pool of
1227 addresses. This is useful for dynamically managing a large
1228 number of interfaces with a single network configuration
1229 file. In particular this is useful to easily assign
1230 appropriate IP addresses to the veth links of a large number
1231 of nspawn instances.
1233 * RPM macros for processing sysusers, sysctl and binfmt
1234 drop-in snippets at package installation time have been
1237 * The /etc/os-release file should now be placed in
1238 /usr/lib/os-release. The old location is automatically
1239 created as symlink. /usr/lib is the more appropriate
1240 location of this file, since it shall actually describe the
1241 vendor operating system shipped in /usr, and not the
1242 configuration stored in /etc.
1244 * .mount units gained a new boolean SloppyOptions= setting
1245 that maps to mount(8)'s -s option which enables permissive
1246 parsing of unknown mount options.
1248 * tmpfiles learnt a new "L+" directive which creates a symlink
1249 but (unlike "L") deletes a pre-existing file first, should
1250 it already exist and not already be the correct
1251 symlink. Similar, "b+", "c+" and "p+" directives have been
1252 added as well, which create block and character devices, as
1253 well as fifos in the filesystem, possibly removing any
1254 pre-existing files of different types.
1256 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
1257 'argument' field (which so far specified the source to
1258 symlink/copy the files from) is now optional. If omitted the
1259 same file os copied from /usr/share/factory/ suffixed by the
1260 full destination path. This is useful for populating /etc
1261 with essential files, by copying them from vendor defaults
1262 shipped in /usr/share/factory/etc.
1264 * A new command "systemctl preset-all" has been added that
1265 applies the service preset settings to all installed unit
1266 files. A new switch --preset-mode= has been added that
1267 controls whether only enable or only disable operations
1270 * A new command "systemctl is-system-running" has been added
1271 that allows checking the overall state of the system, for
1272 example whether it is fully up and running.
1274 * When the system boots up with an empty /etc, the equivalent
1275 to "systemctl preset-all" is executed during early boot, to
1276 make sure all default services are enabled after a factory
1279 * systemd now contains a minimal preset file that enables the
1280 most basic services systemd ships by default.
1282 * Unit files' [Install] section gained a new DefaultInstance=
1283 field for defining the default instance to create if a
1284 template unit is enabled with no instance specified.
1286 * A new passive target cryptsetup-pre.target has been added
1287 that may be used by services that need to make they run and
1288 finish before the first LUKS cryptographic device is set up.
1290 * The /dev/loop-control and /dev/btrfs-control device nodes
1291 are now owned by the "disk" group by default, opening up
1292 access to this group.
1294 * systemd-coredump will now automatically generate a
1295 stack trace of all core dumps taking place on the system,
1296 based on elfutils' libdw library. This stack trace is logged
1299 * systemd-coredump may now optionally store coredumps directly
1300 on disk (in /var/lib/systemd/coredump, possibly compressed),
1301 instead of storing them unconditionally in the journal. This
1302 mode is the new default. A new configuration file
1303 /etc/systemd/coredump.conf has been added to configure this
1304 and other parameters of systemd-coredump.
1306 * coredumpctl gained a new "info" verb to show details about a
1307 specific coredump. A new switch "-1" has also been added
1308 that makes sure to only show information about the most
1309 recent entry instead of all entries. Also, as the tool is
1310 generally useful now the "systemd-" prefix of the binary
1311 name has been removed. Distributions that want to maintain
1312 compatibility with the old name should add a symlink from
1313 the old name to the new name.
1315 * journald's SplitMode= now defaults to "uid". This makes sure
1316 that unprivileged users can access their own coredumps with
1317 coredumpctl without restrictions.
1319 * New kernel command line options "systemd.wants=" (for
1320 pulling an additional unit during boot), "systemd.mask="
1321 (for masking a specific unit for the boot), and
1322 "systemd.debug-shell" (for enabling the debug shell on tty9)
1323 have been added. This is implemented in the new generator
1324 "systemd-debug-generator".
1326 * systemd-nspawn will now by default filter a couple of
1327 syscalls for containers, among them those required for
1328 kernel module loading, direct x86 IO port access, swap
1329 management, and kexec. Most importantly though
1330 open_by_handle_at() is now prohibited for containers,
1331 closing a hole similar to a recently discussed vulnerability
1332 in docker regarding access to files on file hierarchies the
1333 container should normally not have access to. Note that for
1334 nspawn we generally make no security claims anyway (and
1335 this is explicitly documented in the man page), so this is
1336 just a fix for one of the most obvious problems.
1338 * A new man page file-hierarchy(7) has been added that
1339 contains a minimized, modernized version of the file system
1340 layout systemd expects, similar in style to the FHS
1341 specification or hier(5). A new tool systemd-path(1) has
1342 been added to query many of these paths for the local
1345 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
1346 longer done. Since the directory now has a per-user size
1347 limit, and is cleaned on logout this appears unnecessary,
1348 in particular since this now brings the lifecycle of this
1349 directory closer in line with how IPC objects are handled.
1351 * systemd.pc now exports a number of additional directories,
1352 including $libdir (which is useful to identify the library
1353 path for the primary architecture of the system), and a
1354 couple of drop-in directories.
1356 * udev's predictable network interface names now use the dev_port
1357 sysfs attribute, introduced in linux 3.15 instead of dev_id to
1358 distinguish between ports of the same PCI function. dev_id should
1359 only be used for ports using the same HW address, hence the need
1362 * machined has been updated to export the OS version of a
1363 container (read from /etc/os-release and
1364 /usr/lib/os-release) on the bus. This is now shown in
1365 "machinectl status" for a machine.
1367 * A new service setting RestartForceExitStatus= has been
1368 added. If configured to a set of exit signals or process
1369 return values, the service will be restarted when the main
1370 daemon process exits with any of them, regardless of the
1373 * systemctl's -H switch for connecting to remote systemd
1374 machines has been extended so that it may be used to
1375 directly connect to a specific container on the
1376 host. "systemctl -H root@foobar:waldi" will now connect as
1377 user "root" to host "foobar", and then proceed directly to
1378 the container named "waldi". Note that currently you have to
1379 authenticate as user "root" for this to work, as entering
1380 containers is a privileged operation.
1382 Contributions from: Andreas Henriksson, Benjamin Steinwender,
1383 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
1384 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
1385 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
1386 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
1387 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
1388 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
1389 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
1390 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
1391 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
1392 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
1393 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
1395 -- Berlin, 2014-07-03
1399 * As an experimental feature, udev now tries to lock the
1400 disk device node (flock(LOCK_SH|LOCK_NB)) while it
1401 executes events for the disk or any of its partitions.
1402 Applications like partitioning programs can lock the
1403 disk device node (flock(LOCK_EX)) and claim temporary
1404 device ownership that way; udev will entirely skip all event
1405 handling for this disk and its partitions. If the disk
1406 was opened for writing, the close will trigger a partition
1407 table rescan in udev's "watch" facility, and if needed
1408 synthesize "change" events for the disk and all its partitions.
1409 This is now unconditionally enabled, and if it turns out to
1410 cause major problems, we might turn it on only for specific
1411 devices, or might need to disable it entirely. Device Mapper
1412 devices are excluded from this logic.
1414 * We temporarily dropped the "-l" switch for fsck invocations,
1415 since they collide with the flock() logic above. util-linux
1416 upstream has been changed already to avoid this conflict,
1417 and we will readd "-l" as soon as util-linux with this
1418 change has been released.
1420 * The dependency on libattr has been removed. Since a long
1421 time, the extended attribute calls have moved to glibc, and
1422 libattr is thus unnecessary.
1424 * Virtualization detection works without priviliges now. This
1425 means the systemd-detect-virt binary no longer requires
1426 CAP_SYS_PTRACE file capabilities, and our daemons can run
1427 with fewer privileges.
1429 * systemd-networkd now runs under its own "systemd-network"
1430 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
1431 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
1432 loses the ability to write to files owned by root this way.
1434 * Similar, systemd-resolved now runs under its own
1435 "systemd-resolve" user with no capabilities remaining.
1437 * Similar, systemd-bus-proxyd now runs under its own
1438 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
1440 * systemd-networkd gained support for setting up "veth"
1441 virtual ethernet devices for container connectivity, as well
1442 as GRE and VTI tunnels.
1444 * systemd-networkd will no longer automatically attempt to
1445 manually load kernel modules necessary for certain tunnel
1446 transports. Instead, it is assumed the kernel loads them
1447 automatically when required. This only works correctly on
1448 very new kernels. On older kernels, please consider adding
1449 the kernel modules to /etc/modules-load.d/ as a work-around.
1451 * The resolv.conf file systemd-resolved generates has been
1452 moved to /run/systemd/resolve/. If you have a symlink from
1453 /etc/resolv.conf, it might be necessary to correct it.
1455 * Two new service settings, ProtectHome= and ProtectSystem=,
1456 have been added. When enabled, they will make the user data
1457 (such as /home) inaccessible or read-only and the system
1458 (such as /usr) read-only, for specific services. This allows
1459 very light-weight per-service sandboxing to avoid
1460 modifications of user data or system files from
1461 services. These two new switches have been enabled for all
1462 of systemd's long-running services, where appropriate.
1464 * Socket units gained new SocketUser= and SocketGroup=
1465 settings to set the owner user and group of AF_UNIX sockets
1466 and FIFOs in the file system.
1468 * Socket units gained a new RemoveOnStop= setting. If enabled,
1469 all FIFOS and sockets in the file system will be removed
1470 when the specific socket unit is stopped.
1472 * Socket units gained a new Symlinks= setting. It takes a list
1473 of symlinks to create to file system sockets or FIFOs
1474 created by the specific Unix sockets. This is useful to
1475 manage symlinks to socket nodes with the same life-cycle as
1478 * The /dev/log socket and /dev/initctl FIFO have been moved to
1479 /run, and have been replaced by symlinks. This allows
1480 connecting to these facilities even if PrivateDevices=yes is
1481 used for a service (which makes /dev/log itself unavailable,
1482 but /run is left). This also has the benefit of ensuring
1483 that /dev only contains device nodes, directories and
1484 symlinks, and nothing else.
1486 * sd-daemon gained two new calls sd_pid_notify() and
1487 sd_pid_notifyf(). They are similar to sd_notify() and
1488 sd_notifyf(), but allow overriding of the source PID of
1489 notification messages if permissions permit this. This is
1490 useful to send notify messages on behalf of a different
1491 process (for example, the parent process). The
1492 systemd-notify tool has been updated to make use of this
1493 when sending messages (so that notification messages now
1494 originate from the shell script invoking systemd-notify and
1495 not the systemd-notify process itself. This should minimize
1496 a race where systemd fails to associate notification
1497 messages to services when the originating process already
1500 * A new "on-abnormal" setting for Restart= has been added. If
1501 set, it will result in automatic restarts on all "abnormal"
1502 reasons for a process to exit, which includes unclean
1503 signals, core dumps, timeouts and watchdog timeouts, but
1504 does not include clean and unclean exit codes or clean
1505 signals. Restart=on-abnormal is an alternative for
1506 Restart=on-failure for services that shall be able to
1507 terminate and avoid restarts on certain errors, by
1508 indicating so with an unclean exit code. Restart=on-failure
1509 or Restart=on-abnormal is now the recommended setting for
1510 all long-running services.
1512 * If the InaccessibleDirectories= service setting points to a
1513 mount point (or if there are any submounts contained within
1514 it), it is now attempted to completely unmount it, to make
1515 the file systems truly unavailable for the respective
1518 * The ReadOnlyDirectories= service setting and
1519 systemd-nspawn's --read-only parameter are now recursively
1520 applied to all submounts, too.
1522 * Mount units may now be created transiently via the bus APIs.
1524 * The support for SysV and LSB init scripts has been removed
1525 from the systemd daemon itself. Instead, it is now
1526 implemented as a generator that creates native systemd units
1527 from these scripts when needed. This enables us to remove a
1528 substantial amount of legacy code from PID 1, following the
1529 fact that many distributions only ship a very small number
1530 of LSB/SysV init scripts nowadays.
1532 * Privileged Xen (dom0) domains are not considered
1533 virtualization anymore by the virtualization detection
1534 logic. After all, they generally have unrestricted access to
1535 the hardware and usually are used to manage the unprivileged
1538 * systemd-tmpfiles gained a new "C" line type, for copying
1539 files or entire directories.
1541 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
1542 lines. So far, they have been non-globbing versions of the
1543 latter, and have thus been redundant. In future, it is
1544 recommended to only use "z". "m" has hence been removed
1545 from the documentation, even though it stays supported.
1547 * A tmpfiles snippet to recreate the most basic structure in
1548 /var has been added. This is enough to create the /var/run →
1549 /run symlink and create a couple of structural
1550 directories. This allows systems to boot up with an empty or
1551 volatile /var. Of course, while with this change, the core OS
1552 now is capable with dealing with a volatile /var, not all
1553 user services are ready for it. However, we hope that sooner
1554 or later, many service daemons will be changed upstream so
1555 that they are able to automatically create their necessary
1556 directories in /var at boot, should they be missing. This is
1557 the first step to allow state-less systems that only require
1558 the vendor image for /usr to boot.
1560 * systemd-nspawn has gained a new --tmpfs= switch to mount an
1561 empty tmpfs instance to a specific directory. This is
1562 particularly useful for making use of the automatic
1563 reconstruction of /var (see above), by passing --tmpfs=/var.
1565 * Access modes specified in tmpfiles snippets may now be
1566 prefixed with "~", which indicates that they shall be masked
1567 by whether the existing file or directory is currently
1568 writable, readable or executable at all. Also, if specified,
1569 the sgid/suid/sticky bits will be masked for all
1572 * A new passive target unit "network-pre.target" has been
1573 added which is useful for services that shall run before any
1574 network is configured, for example firewall scripts.
1576 * The "floppy" group that previously owned the /dev/fd*
1577 devices is no longer used. The "disk" group is now used
1578 instead. Distributions should probably deprecate usage of
1581 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
1582 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
1583 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
1584 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
1585 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
1586 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
1589 -- Berlin, 2014-06-11
1593 * A new "systemd-timesyncd" daemon has been added for
1594 synchronizing the system clock across the network. It
1595 implements an SNTP client. In contrast to NTP
1596 implementations such as chrony or the NTP reference server,
1597 this only implements a client side, and does not bother with
1598 the full NTP complexity, focusing only on querying time from
1599 one remote server and synchronizing the local clock to
1600 it. Unless you intend to serve NTP to networked clients or
1601 want to connect to local hardware clocks, this simple NTP
1602 client should be more than appropriate for most
1603 installations. The daemon runs with minimal privileges, and
1604 has been hooked up with networkd to only operate when
1605 network connectivity is available. The daemon saves the
1606 current clock to disk every time a new NTP sync has been
1607 acquired, and uses this to possibly correct the system clock
1608 early at bootup, in order to accommodate for systems that
1609 lack an RTC such as the Raspberry Pi and embedded devices,
1610 and to make sure that time monotonically progresses on these
1611 systems, even if it is not always correct. To make use of
1612 this daemon, a new system user and group "systemd-timesync"
1613 needs to be created on installation of systemd.
1615 * The queue "seqnum" interface of libudev has been disabled, as
1616 it was generally incompatible with device namespacing as
1617 sequence numbers of devices go "missing" if the devices are
1618 part of a different namespace.
1620 * "systemctl list-timers" and "systemctl list-sockets" gained
1621 a --recursive switch for showing units of these types also
1622 for all local containers, similar in style to the already
1623 supported --recursive switch for "systemctl list-units".
1625 * A new RebootArgument= setting has been added for service
1626 units, which may be used to specify a kernel reboot argument
1627 to use when triggering reboots with StartLimitAction=.
1629 * A new FailureAction= setting has been added for service
1630 units which may be used to specify an operation to trigger
1631 when a service fails. This works similarly to
1632 StartLimitAction=, but unlike it, controls what is done
1633 immediately rather than only after several attempts to
1634 restart the service in question.
1636 * hostnamed got updated to also expose the kernel name,
1637 release, and version on the bus. This is useful for
1638 executing commands like hostnamectl with the -H switch.
1639 systemd-analyze makes use of this to properly display
1640 details when running non-locally.
1642 * The bootchart tool can now show cgroup information in the
1643 graphs it generates.
1645 * The CFS CPU quota cgroup attribute is now exposed for
1646 services. The new CPUQuota= switch has been added for this
1647 which takes a percentage value. Setting this will have the
1648 result that a service may never get more CPU time than the
1649 specified percentage, even if the machine is otherwise idle.
1651 * systemd-networkd learned IPIP and SIT tunnel support.
1653 * LSB init scripts exposing a dependency on $network will now
1654 get a dependency on network-online.target rather than simply
1655 network.target. This should bring LSB handling closer to
1656 what it was on SysV systems.
1658 * A new fsck.repair= kernel option has been added to control
1659 how fsck shall deal with unclean file systems at boot.
1661 * The (.ini) configuration file parser will now silently
1662 ignore sections whose name begins with "X-". This may be
1663 used to maintain application-specific extension sections in unit
1666 * machined gained a new API to query the IP addresses of
1667 registered containers. "machinectl status" has been updated
1668 to show these addresses in its output.
1670 * A new call sd_uid_get_display() has been added to the
1671 sd-login APIs for querying the "primary" session of a
1672 user. The "primary" session of the user is elected from the
1673 user's sessions and generally a graphical session is
1674 preferred over a text one.
1676 * A minimal systemd-resolved daemon has been added. It
1677 currently simply acts as a companion to systemd-networkd and
1678 manages resolv.conf based on per-interface DNS
1679 configuration, possibly supplied via DHCP. In the long run
1680 we hope to extend this into a local DNSSEC enabled DNS and
1683 * The systemd-networkd-wait-online tool is now enabled by
1684 default. It will delay network-online.target until a network
1685 connection has been configured. The tool primarily integrates
1686 with networkd, but will also make a best effort to make sense
1687 of network configuration performed in some other way.
1689 * Two new service options StartupCPUShares= and
1690 StartupBlockIOWeight= have been added that work similarly to
1691 CPUShares= and BlockIOWeight= however only apply during
1692 system startup. This is useful to prioritize certain services
1693 differently during bootup than during normal runtime.
1695 * hostnamed has been changed to prefer the statically
1696 configured hostname in /etc/hostname (unless set to
1697 'localhost' or empty) over any dynamic one supplied by
1698 dhcp. With this change, the rules for picking the hostname
1699 match more closely the rules of other configuration settings
1700 where the local administrator's configuration in /etc always
1701 overrides any other settings.
1703 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
1704 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
1705 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
1706 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
1707 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
1708 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
1709 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
1710 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
1711 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
1712 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
1713 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
1714 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
1715 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
1716 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
1717 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
1718 Lindskog, WaLyong Cho, Will Woods, Zbigniew
1721 -- Beijing, 2014-05-28
1725 * When restoring the screen brightness at boot, stay away from
1726 the darkest setting or from the lowest 5% of the available
1727 range, depending on which is the larger value of both. This
1728 should effectively protect the user from rebooting into a
1729 black screen, should the brightness have been set to minimum
1732 * sd-login gained a new sd_machine_get_class() call to
1733 determine the class ("vm" or "container") of a machine
1734 registered with machined.
1736 * sd-login gained new calls
1737 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
1738 to query the identity of the peer of a local AF_UNIX
1739 connection. They operate similarly to their sd_pid_get_xyz()
1742 * PID 1 will now maintain a system-wide system state engine
1743 with the states "starting", "running", "degraded",
1744 "maintenance", "stopping". These states are bound to system
1745 startup, normal runtime, runtime with at least one failed
1746 service, rescue/emergency mode and system shutdown. This
1747 state is shown in the "systemctl status" output when no unit
1748 name is passed. It is useful to determine system state, in
1749 particularly when doing so for many systems or containers at
1752 * A new command "list-machines" has been added to "systemctl"
1753 that lists all local OS containers and shows their system
1754 state (see above), if systemd runs inside of them.
1756 * systemctl gained a new "-r" switch to recursively enumerate
1757 units on all local containers, when used with the
1758 "list-unit" command (which is the default one that is
1759 executed when no parameters are specified).
1761 * The GPT automatic partition discovery logic will now honour
1762 two GPT partition flags: one may be set on a partition to
1763 cause it to be mounted read-only, and the other may be set
1764 on a partition to ignore it during automatic discovery.
1766 * Two new GPT type UUIDs have been added for automatic root
1767 partition discovery, for 32-bit and 64-bit ARM. This is not
1768 particularly useful for discovering the root directory on
1769 these architectures during bare-metal boots (since UEFI is
1770 not common there), but still very useful to allow booting of
1771 ARM disk images in nspawn with the -i option.
1773 * MAC addresses of interfaces created with nspawn's
1774 --network-interface= switch will now be generated from the
1775 machine name, and thus be stable between multiple invocations
1778 * logind will now automatically remove all IPC objects owned
1779 by a user if she or he fully logs out. This makes sure that
1780 users who are logged out cannot continue to consume IPC
1781 resources. This covers SysV memory, semaphores and message
1782 queues as well as POSIX shared memory and message
1783 queues. Traditionally, SysV and POSIX IPC had no life-cycle
1784 limits. With this functionality, that is corrected. This may
1785 be turned off by using the RemoveIPC= switch of logind.conf.
1787 * The systemd-machine-id-setup and tmpfiles tools gained a
1788 --root= switch to operate on a specific root directory,
1791 * journald can now forward logged messages to the TTYs of all
1792 logged in users ("wall"). This is the default for all
1793 emergency messages now.
1795 * A new tool systemd-journal-remote has been added to stream
1796 journal log messages across the network.
1798 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
1799 controller trees are mounted into it. Note that the
1800 directories mounted beneath it are not read-only. This is a
1801 security measure and is particularly useful because glibc
1802 actually includes a search logic to pick any tmpfs it can
1803 find to implement shm_open() if /dev/shm is not available
1804 (which it might very well be in namespaced setups).
1806 * machinectl gained a new "poweroff" command to cleanly power
1807 down a local OS container.
1809 * The PrivateDevices= unit file setting will now also drop the
1810 CAP_MKNOD capability from the capability bound set, and
1811 imply DevicePolicy=closed.
1813 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
1814 comprehensively on all long-running systemd services where
1815 this is appropriate.
1817 * systemd-udevd will now run in a disassociated mount
1818 namespace. To mount directories from udev rules, make sure to
1819 pull in mount units via SYSTEMD_WANTS properties.
1821 * The kdbus support gained support for uploading policy into
1822 the kernel. sd-bus gained support for creating "monitoring"
1823 connections that can eavesdrop into all bus communication
1824 for debugging purposes.
1826 * Timestamps may now be specified in seconds since the UNIX
1827 epoch Jan 1st, 1970 by specifying "@" followed by the value
1830 * Native tcpwrap support in systemd has been removed. tcpwrap
1831 is old code, not really maintained anymore and has serious
1832 shortcomings, and better options such as firewalls
1833 exist. For setups that require tcpwrap usage, please
1834 consider invoking your socket-activated service via tcpd,
1835 like on traditional inetd.
1837 * A new system.conf configuration option
1838 DefaultTimerAccuracySec= has been added that controls the
1839 default AccuracySec= setting of .timer units.
1841 * Timer units gained a new WakeSystem= switch. If enabled,
1842 timers configured this way will cause the system to resume
1843 from system suspend (if the system supports that, which most
1846 * Timer units gained a new Persistent= switch. If enabled,
1847 timers configured this way will save to disk when they have
1848 been last triggered. This information is then used on next
1849 reboot to possible execute overdue timer events, that
1850 could not take place because the system was powered off.
1851 This enables simple anacron-like behaviour for timer units.
1853 * systemctl's "list-timers" will now also list the time a
1854 timer unit was last triggered in addition to the next time
1855 it will be triggered.
1857 * systemd-networkd will now assign predictable IPv4LL
1858 addresses to its local interfaces.
1860 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
1861 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
1862 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
1863 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
1864 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
1865 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
1866 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
1867 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
1870 -- Berlin, 2014-03-25
1874 * A new unit file setting RestrictAddressFamilies= has been
1875 added to restrict which socket address families unit
1876 processes gain access to. This takes address family names
1877 like "AF_INET" or "AF_UNIX", and is useful to minimize the
1878 attack surface of services via exotic protocol stacks. This
1879 is built on seccomp system call filters.
1881 * Two new unit file settings RuntimeDirectory= and
1882 RuntimeDirectoryMode= have been added that may be used to
1883 manage a per-daemon runtime directories below /run. This is
1884 an alternative for setting up directory permissions with
1885 tmpfiles snippets, and has the advantage that the runtime
1886 directory's lifetime is bound to the daemon runtime and that
1887 the daemon starts up with an empty directory each time. This
1888 is particularly useful when writing services that drop
1889 privileges using the User= or Group= setting.
1891 * The DeviceAllow= unit setting now supports globbing for
1892 matching against device group names.
1894 * The systemd configuration file system.conf gained new
1895 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
1896 DefaultMemoryAccounting= to globally turn on/off accounting
1897 for specific resources (cgroups) for all units. These
1898 settings may still be overridden individually in each unit
1901 * systemd-gpt-auto-generator is now able to discover /srv and
1902 root partitions in addition to /home and swap partitions. It
1903 also supports LUKS-encrypted partitions now. With this in
1904 place, automatic discovery of partitions to mount following
1905 the Discoverable Partitions Specification
1906 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
1907 is now a lot more complete. This allows booting without
1908 /etc/fstab and without root= on the kernel command line on
1909 systems prepared appropriately.
1911 * systemd-nspawn gained a new --image= switch which allows
1912 booting up disk images and Linux installations on any block
1913 device that follow the Discoverable Partitions Specification
1914 (see above). This means that installations made with
1915 appropriately updated installers may now be started and
1916 deployed using container managers, completely
1917 unmodified. (We hope that libvirt-lxc will add support for
1918 this feature soon, too.)
1920 * systemd-nspawn gained a new --network-macvlan= setting to
1921 set up a private macvlan interface for the
1922 container. Similarly, systemd-networkd gained a new
1923 Kind=macvlan setting in .netdev files.
1925 * systemd-networkd now supports configuring local addresses
1928 * A new tool systemd-network-wait-online has been added to
1929 synchronously wait for network connectivity using
1932 * The sd-bus.h bus API gained a new sd_bus_track object for
1933 tracking the life-cycle of bus peers. Note that sd-bus.h is
1934 still not a public API though (unless you specify
1935 --enable-kdbus on the configure command line, which however
1936 voids your warranty and you get no API stability guarantee).
1938 * The $XDG_RUNTIME_DIR runtime directories for each user are
1939 now individual tmpfs instances, which has the benefit of
1940 introducing separate pools for each user, with individual
1941 size limits, and thus making sure that unprivileged clients
1942 can no longer negatively impact the system or other users by
1943 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
1944 RuntimeDirectorySize= has been introduced that allows
1945 controlling the default size limit for all users. It
1946 defaults to 10% of the available physical memory. This is no
1947 replacement for quotas on tmpfs though (which the kernel
1948 still does not support), as /dev/shm and /tmp are still
1949 shared resources used by both the system and unprivileged
1952 * logind will now automatically turn off automatic suspending
1953 on laptop lid close when more than one display is
1954 connected. This was previously expected to be implemented
1955 individually in desktop environments (such as GNOME),
1956 however has been added to logind now, in order to fix a
1957 boot-time race where a desktop environment might not have
1958 been started yet and thus not been able to take an inhibitor
1959 lock at the time where logind already suspends the system
1960 due to a closed lid.
1962 * logind will now wait at least 30s after each system
1963 suspend/resume cycle, and 3min after system boot before
1964 suspending the system due to a closed laptop lid. This
1965 should give USB docking stations and similar enough time to
1966 be probed and configured after system resume and boot in
1967 order to then act as suspend blocker.
1969 * systemd-run gained a new --property= setting which allows
1970 initialization of resource control properties (and others)
1971 for the created scope or service unit. Example: "systemd-run
1972 --property=BlockIOWeight=10 updatedb" may be used to run
1973 updatedb at a low block IO scheduling weight.
1975 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
1976 now also work in --scope mode.
1978 * When systemd is compiled with kdbus support, basic support
1979 for enforced policies is now in place. (Note that enabling
1980 kdbus still voids your warranty and no API compatibility
1983 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
1984 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
1985 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
1986 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
1987 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
1988 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
1989 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
1990 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
1991 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
1992 Zbigniew Jędrzejewski-Szmek
1994 -- Berlin, 2014-03-12
1998 * systemd will now relabel /dev after loading the SMACK policy
1999 according to SMACK rules.
2001 * A new unit file option AppArmorProfile= has been added to
2002 set the AppArmor profile for the processes of a unit.
2004 * A new condition check ConditionArchitecture= has been added
2005 to conditionalize units based on the system architecture, as
2006 reported by uname()'s "machine" field.
2008 * systemd-networkd now supports matching on the system
2009 virtualization, architecture, kernel command line, host name
2012 * logind is now a lot more aggressive when suspending the
2013 machine due to a closed laptop lid. Instead of acting only
2014 on the lid close action, it will continuously watch the lid
2015 status and act on it. This is useful for laptops where the
2016 power button is on the outside of the chassis so that it can
2017 be reached without opening the lid (such as the Lenovo
2018 Yoga). On those machines, logind will now immediately
2019 re-suspend the machine if the power button has been
2020 accidentally pressed while the laptop was suspended and in a
2021 backpack or similar.
2023 * logind will now watch SW_DOCK switches and inhibit reaction
2024 to the lid switch if it is pressed. This means that logind
2025 will not suspend the machine anymore if the lid is closed
2026 and the system is docked, if the laptop supports SW_DOCK
2027 notifications via the input layer. Note that ACPI docking
2028 stations do not generate this currently. Also note that this
2029 logic is usually not fully sufficient and Desktop
2030 Environments should take a lid switch inhibitor lock when an
2031 external display is connected, as systemd will not watch
2034 * nspawn will now make use of the devices cgroup controller by
2035 default, and only permit creation of and access to the usual
2036 API device nodes like /dev/null or /dev/random, as well as
2037 access to (but not creation of) the pty devices.
2039 * We will now ship a default .network file for
2040 systemd-networkd that automatically configures DHCP for
2041 network interfaces created by nspawn's --network-veth or
2042 --network-bridge= switches.
2044 * systemd will now understand the usual M, K, G, T suffixes
2045 according to SI conventions (i.e. to the base 1000) when
2046 referring to throughput and hardware metrics. It will stay
2047 with IEC conventions (i.e. to the base 1024) for software
2048 metrics, according to what is customary according to
2049 Wikipedia. We explicitly document which base applies for
2050 each configuration option.
2052 * The DeviceAllow= setting in unit files now supports a syntax
2053 to whitelist an entire group of devices node majors at once,
2054 based on the /proc/devices listing. For example, with the
2055 string "char-pts", it is now possible to whitelist all
2056 current and future pseudo-TTYs at once.
2058 * sd-event learned a new "post" event source. Event sources of
2059 this type are triggered by the dispatching of any event
2060 source of a type that is not "post". This is useful for
2061 implementing clean-up and check event sources that are
2062 triggered by other work being done in the program.
2064 * systemd-networkd is no longer statically enabled, but uses
2065 the usual [Install] sections so that it can be
2066 enabled/disabled using systemctl. It still is enabled by
2069 * When creating a veth interface pair with systemd-nspawn, the
2070 host side will now be prefixed with "vb-" if
2071 --network-bridge= is used, and with "ve-" if --network-veth
2072 is used. This way, it is easy to distinguish these cases on
2073 the host, for example to apply different configuration to
2074 them with systemd-networkd.
2076 * The compatibility libraries for libsystemd-journal.so,
2077 libsystem-id128.so, libsystemd-login.so and
2078 libsystemd-daemon.so do not make use of IFUNC
2079 anymore. Instead, we now build libsystemd.so multiple times
2080 under these alternative names. This means that the footprint
2081 is drastically increased, but given that these are
2082 transitional compatibility libraries, this should not matter
2083 much. This change has been made necessary to support the ARM
2084 platform for these compatibility libraries, as the ARM
2085 toolchain is not really at the same level as the toolchain
2086 for other architectures like x86 and does not support
2087 IFUNC. Please make sure to use --enable-compat-libs only
2088 during a transitional period!
2090 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
2091 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
2092 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
2093 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
2094 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
2095 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
2096 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
2097 Zbigniew Jędrzejewski-Szmek
2099 -- Berlin, 2014-02-24
2103 * A new component "systemd-networkd" has been added that can
2104 be used to configure local network interfaces statically or
2105 via DHCP. It is capable of bringing up bridges, VLANs, and
2106 bonding. Currently, no hook-ups for interactive network
2107 configuration are provided. Use this for your initrd,
2108 container, embedded, or server setup if you need a simple,
2109 yet powerful, network configuration solution. This
2110 configuration subsystem is quite nifty, as it allows wildcard
2111 hotplug matching in interfaces. For example, with a single
2112 configuration snippet, you can configure that all Ethernet
2113 interfaces showing up are automatically added to a bridge,
2114 or similar. It supports link-sensing and more.
2116 * A new tool "systemd-socket-proxyd" has been added which can
2117 act as a bidirectional proxy for TCP sockets. This is
2118 useful for adding socket activation support to services that
2119 do not actually support socket activation, including virtual
2120 machines and the like.
2122 * Add a new tool to save/restore rfkill state on
2125 * Save/restore state of keyboard backlights in addition to
2126 display backlights on shutdown/boot.
2128 * udev learned a new SECLABEL{} construct to label device
2129 nodes with a specific security label when they appear. For
2130 now, only SECLABEL{selinux} is supported, but the syntax is
2131 prepared for additional security frameworks.
2133 * udev gained a new scheme to configure link-level attributes
2134 from files in /etc/systemd/network/*.link. These files can
2135 match against MAC address, device path, driver name and type,
2136 and will apply attributes like the naming policy, link speed,
2137 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
2138 address assignment policy (randomized, ...).
2140 * The configuration of network interface naming rules for
2141 "permanent interface names" has changed: a new NamePolicy=
2142 setting in the [Link] section of .link files determines the
2143 priority of possible naming schemes (onboard, slot, mac,
2144 path). The default value of this setting is determined by
2145 /usr/lib/net/links/99-default.link. Old
2146 80-net-name-slot.rules udev configuration file has been
2147 removed, so local configuration overriding this file should
2148 be adapated to override 99-default.link instead.
2150 * When the User= switch is used in a unit file, also
2151 initialize $SHELL= based on the user database entry.
2153 * systemd no longer depends on libdbus. All communication is
2154 now done with sd-bus, systemd's low-level bus library
2157 * kdbus support has been added to PID 1 itself. When kdbus is
2158 enabled, this causes PID 1 to set up the system bus and
2159 enable support for a new ".busname" unit type that
2160 encapsulates bus name activation on kdbus. It works a little
2161 bit like ".socket" units, except for bus names. A new
2162 generator has been added that converts classic dbus1 service
2163 activation files automatically into native systemd .busname
2166 * sd-bus: add a light-weight vtable implementation that allows
2167 defining objects on the bus with a simple static const
2168 vtable array of its methods, signals and properties.
2170 * systemd will not generate or install static dbus
2171 introspection data anymore to /usr/share/dbus-1/interfaces,
2172 as the precise format of these files is unclear, and
2173 nothing makes use of it.
2175 * A proxy daemon is now provided to proxy clients connecting
2176 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
2177 compatibility with classic D-Bus.
2179 * A bus driver implementation has been added that supports the
2180 classic D-Bus bus driver calls on kdbus, also for
2181 compatibility purposes.
2183 * A new API "sd-event.h" has been added that implements a
2184 minimal event loop API built around epoll. It provides a
2185 couple of features that direct epoll usage is lacking:
2186 prioritization of events, scales to large numbers of timer
2187 events, per-event timer slack (accuracy), system-wide
2188 coalescing of timer events, exit handlers, watchdog
2189 supervision support using systemd's sd_notify() API, child
2192 * A new API "sd-rntl.h" has been added that provides an API
2193 around the route netlink interface of the kernel, similar in
2194 style to "sd-bus.h".
2196 * A new API "sd-dhcp-client.h" has been added that provides a
2197 small DHCPv4 client-side implementation. This is used by
2200 * There is a new kernel command line option
2201 "systemd.restore_state=0|1". When set to "0", none of the
2202 systemd tools will restore saved runtime state to hardware
2203 devices. More specifically, the rfkill and backlight states
2206 * The FsckPassNo= compatibility option in mount/service units
2207 has been removed. The fstab generator will now add the
2208 necessary dependencies automatically, and does not require
2209 PID1's support for that anymore.
2211 * journalctl gained a new switch, --list-boots, that lists
2212 recent boots with their times and boot IDs.
2214 * The various tools like systemctl, loginctl, timedatectl,
2215 busctl, systemd-run, ... have gained a new switch "-M" to
2216 connect to a specific, local OS container (as direct
2217 connection, without requiring SSH). This works on any
2218 container that is registered with machined, such as those
2219 created by libvirt-lxc or nspawn.
2221 * systemd-run and systemd-analyze also gained support for "-H"
2222 to connect to remote hosts via SSH. This is particularly
2223 useful for systemd-run because it enables queuing of jobs
2224 onto remote systems.
2226 * machinectl gained a new command "login" to open a getty
2227 login in any local container. This works with any container
2228 that is registered with machined (such as those created by
2229 libvirt-lxc or nspawn), and which runs systemd inside.
2231 * machinectl gained a new "reboot" command that may be used to
2232 trigger a reboot on a specific container that is registered
2233 with machined. This works on any container that runs an init
2234 system of some kind.
2236 * systemctl gained a new "list-timers" command to print a nice
2237 listing of installed timer units with the times they elapse
2240 * Alternative reboot() parameters may now be specified on the
2241 "systemctl reboot" command line and are passed to the
2242 reboot() system call.
2244 * systemctl gained a new --job-mode= switch to configure the
2245 mode to queue a job with. This is a more generic version of
2246 --fail, --irreversible, and --ignore-dependencies, which are
2247 still available but not advertised anymore.
2249 * /etc/systemd/system.conf gained new settings to configure
2250 various default timeouts of units, as well as the default
2251 start limit interval and burst. These may still be overridden
2254 * PID1 will now export on the bus profile data of the security
2255 policy upload process (such as the SELinux policy upload to
2258 * journald: when forwarding logs to the console, include
2259 timestamps (following the setting in
2260 /sys/module/printk/parameters/time).
2262 * OnCalendar= in timer units now understands the special
2263 strings "yearly" and "annually". (Both are equivalent)
2265 * The accuracy of timer units is now configurable with the new
2266 AccuracySec= setting. It defaults to 1min.
2268 * A new dependency type JoinsNamespaceOf= has been added that
2269 allows running two services within the same /tmp and network
2270 namespace, if PrivateNetwork= or PrivateTmp= are used.
2272 * A new command "cat" has been added to systemctl. It outputs
2273 the original unit file of a unit, and concatenates the
2274 contents of additional "drop-in" unit file snippets, so that
2275 the full configuration is shown.
2277 * systemctl now supports globbing on the various "list-xyz"
2278 commands, like "list-units" or "list-sockets", as well as on
2279 those commands which take multiple unit names.
2281 * journalctl's --unit= switch gained support for globbing.
2283 * All systemd daemons now make use of the watchdog logic so
2284 that systemd automatically notices when they hang.
2286 * If the $container_ttys environment variable is set,
2287 getty-generator will automatically spawn a getty for each
2288 listed tty. This is useful for container managers to request
2289 login gettys to be spawned on as many ttys as needed.
2291 * %h, %s, %U specifier support is not available anymore when
2292 used in unit files for PID 1. This is because NSS calls are
2293 not safe from PID 1. They stay available for --user
2294 instances of systemd, and as special case for the root user.
2296 * loginctl gained a new "--no-legend" switch to turn off output
2299 * The "sd-login.h" API gained three new calls:
2300 sd_session_is_remote(), sd_session_get_remote_user(),
2301 sd_session_get_remote_host() to query information about
2304 * The udev hardware database now also carries vendor/product
2305 information of SDIO devices.
2307 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
2308 determine whether watchdog notifications are requested by
2311 * Socket-activated per-connection services now include a
2312 short description of the connection parameters in the
2315 * tmpfiles gained a new "--boot" option. When this is not used,
2316 only lines where the command character is not suffixed with
2317 "!" are executed. When this option is specified, those
2318 options are executed too. This partitions tmpfiles
2319 directives into those that can be safely executed at any
2320 time, and those which should be run only at boot (for
2321 example, a line that creates /run/nologin).
2323 * A new API "sd-resolve.h" has been added which provides a simple
2324 asynchronous wrapper around glibc NSS host name resolution
2325 calls, such as getaddrinfo(). In contrast to glibc's
2326 getaddrinfo_a(), it does not use signals. In contrast to most
2327 other asynchronous name resolution libraries, this one does
2328 not reimplement DNS, but reuses NSS, so that alternate
2329 host name resolution systems continue to work, such as mDNS,
2330 LDAP, etc. This API is based on libasyncns, but it has been
2331 cleaned up for inclusion in systemd.
2333 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
2334 "sd-daemon.h" are no longer found in individual libraries
2335 libsystemd-journal.so, libsystemd-login.so,
2336 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
2337 merged them into a single library, libsystemd.so, which
2338 provides all symbols. The reason for this is cyclic
2339 dependencies, as these libraries tend to use each other's
2340 symbols. So far, we have managed to workaround that by linking
2341 a copy of a good part of our code into each of these
2342 libraries again and again, which, however, makes certain
2343 things hard to do, like sharing static variables. Also, it
2344 substantially increases footprint. With this change, there
2345 is only one library for the basic APIs systemd
2346 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
2347 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
2348 library as well, however are subject to the --enable-kdbus
2349 switch (see below). Note that "sd-dhcp-client.h" is not part
2350 of this library (this is because it only consumes, never
2351 provides, services of/to other APIs). To make the transition
2352 easy from the separate libraries to the unified one, we
2353 provide the --enable-compat-libs compile-time switch which
2354 will generate stub libraries that are compatible with the
2355 old ones but redirect all calls to the new one.
2357 * All of the kdbus logic and the new APIs "sd-bus.h",
2358 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
2359 and "sd-utf8.h" are compile-time optional via the
2360 "--enable-kdbus" switch, and they are not compiled in by
2361 default. To make use of kdbus, you have to explicitly enable
2362 the switch. Note however, that neither the kernel nor the
2363 userspace API for all of this is considered stable yet. We
2364 want to maintain the freedom to still change the APIs for
2365 now. By specifying this build-time switch, you acknowledge
2366 that you are aware of the instability of the current
2369 * Also, note that while kdbus is pretty much complete,
2370 it lacks one thing: proper policy support. This means you
2371 can build a fully working system with all features; however,
2372 it will be highly insecure. Policy support will be added in
2373 one of the next releases, at the same time that we will
2374 declare the APIs stable.
2376 * When the kernel command line argument "kdbus" is specified,
2377 systemd will automatically load the kdbus.ko kernel module. At
2378 this stage of development, it is only useful for testing kdbus
2379 and should not be used in production. Note: if "--enable-kdbus"
2380 is specified, and the kdbus.ko kernel module is available, and
2381 "kdbus" is added to the kernel command line, the entire system
2382 runs with kdbus instead of dbus-daemon, with the above mentioned
2383 problem of missing the system policy enforcement. Also a future
2384 version of kdbus.ko or a newer systemd will not be compatible with
2385 each other, and will unlikely be able to boot the machine if only
2386 one of them is updated.
2388 * systemctl gained a new "import-environment" command which
2389 uploads the caller's environment (or parts thereof) into the
2390 service manager so that it is inherited by services started
2391 by the manager. This is useful to upload variables like
2392 $DISPLAY into the user service manager.
2394 * A new PrivateDevices= switch has been added to service units
2395 which allows running a service with a namespaced /dev
2396 directory that does not contain any device nodes for
2397 physical devices. More specifically, it only includes devices
2398 such as /dev/null, /dev/urandom, and /dev/zero which are API
2401 * logind has been extended to support behaviour like VT
2402 switching on seats that do not support a VT. This makes
2403 multi-session available on seats that are not the first seat
2404 (seat0), and on systems where kernel support for VTs has
2405 been disabled at compile-time.
2407 * If a process holds a delay lock for system sleep or shutdown
2408 and fails to release it in time, we will now log its
2409 identity. This makes it easier to identify processes that
2410 cause slow suspends or power-offs.
2412 * When parsing /etc/crypttab, support for a new key-slot=
2413 option as supported by Debian is added. It allows indicating
2414 which LUKS slot to use on disk, speeding up key loading.
2416 * The sd_journald_sendv() API call has been checked and
2417 officially declared to be async-signal-safe so that it may
2418 be invoked from signal handlers for logging purposes.
2420 * Boot-time status output is now enabled automatically after a
2421 short timeout if boot does not progress, in order to give
2422 the user an indication what she or he is waiting for.
2424 * The boot-time output has been improved to show how much time
2425 remains until jobs expire.
2427 * The KillMode= switch in service units gained a new possible
2428 value "mixed". If set, and the unit is shut down, then the
2429 initial SIGTERM signal is sent only to the main daemon
2430 process, while the following SIGKILL signal is sent to
2431 all remaining processes of the service.
2433 * When a scope unit is registered, a new property "Controller"
2434 may be set. If set to a valid bus name, systemd will send a
2435 RequestStop() signal to this name when it would like to shut
2436 down the scope. This may be used to hook manager logic into
2437 the shutdown logic of scope units. Also, scope units may now
2438 be put in a special "abandoned" state, in which case the
2439 manager process which created them takes no further
2440 responsibilities for it.
2442 * When reading unit files, systemd will now verify
2443 the access mode of these files, and warn about certain
2444 suspicious combinations. This has been added to make it
2445 easier to track down packaging bugs where unit files are
2446 marked executable or world-writable.
2448 * systemd-nspawn gained a new "--setenv=" switch to set
2449 container-wide environment variables. The similar option in
2450 systemd-activate was renamed from "--environment=" to
2451 "--setenv=" for consistency.
2453 * systemd-nspawn has been updated to create a new kdbus domain
2454 for each container that is invoked, thus allowing each
2455 container to have its own set of system and user buses,
2456 independent of the host.
2458 * systemd-nspawn gained a new --drop-capability= switch to run
2459 the container with less capabilities than the default. Both
2460 --drop-capability= and --capability= now take the special
2461 string "all" for dropping or keeping all capabilities.
2463 * systemd-nspawn gained new switches for executing containers
2464 with specific SELinux labels set.
2466 * systemd-nspawn gained a new --quiet switch to not generate
2467 any additional output but the container's own console
2470 * systemd-nspawn gained a new --share-system switch to run a
2471 container without PID namespacing enabled.
2473 * systemd-nspawn gained a new --register= switch to control
2474 whether the container is registered with systemd-machined or
2475 not. This is useful for containers that do not run full
2476 OS images, but only specific apps.
2478 * systemd-nspawn gained a new --keep-unit which may be used
2479 when invoked as the only program from a service unit, and
2480 results in registration of the unit service itself in
2481 systemd-machined, instead of a newly opened scope unit.
2483 * systemd-nspawn gained a new --network-interface= switch for
2484 moving arbitrary interfaces to the container. The new
2485 --network-veth switch creates a virtual Ethernet connection
2486 between host and container. The new --network-bridge=
2487 switch then allows assigning the host side of this virtual
2488 Ethernet connection to a bridge device.
2490 * systemd-nspawn gained a new --personality= switch for
2491 setting the kernel personality for the container. This is
2492 useful when running a 32-bit container on a 64-bit host. A
2493 similar option Personality= is now also available for service
2496 * logind will now also track a "Desktop" identifier for each
2497 session which encodes the desktop environment of it. This is
2498 useful for desktop environments that want to identify
2499 multiple running sessions of itself easily.
2501 * A new SELinuxContext= setting for service units has been
2502 added that allows setting a specific SELinux execution
2503 context for a service.
2505 * Most systemd client tools will now honour $SYSTEMD_LESS for
2506 settings of the "less" pager. By default, these tools will
2507 override $LESS to allow certain operations to work, such as
2508 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
2509 influence this logic.
2511 * systemd's "seccomp" hook-up has been changed to make use of
2512 the libseccomp library instead of using its own
2513 implementation. This has benefits for portability among
2516 * For usage together with SystemCallFilter=, a new
2517 SystemCallErrorNumber= setting has been introduced that
2518 allows configuration of a system error number to be returned
2519 on filtered system calls, instead of immediately killing the
2520 process. Also, SystemCallArchitectures= has been added to
2521 limit access to system calls of a particular architecture
2522 (in order to turn off support for unused secondary
2523 architectures). There is also a global
2524 SystemCallArchitectures= setting in system.conf now to turn
2525 off support for non-native system calls system-wide.
2527 * systemd requires a kernel with a working name_to_handle_at(),
2528 please see the kernel config requirements in the README file.
2530 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
2531 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
2532 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
2533 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
2534 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
2535 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
2536 Elia Pinto, Florian Weimer, George McCollister, Goffredo
2537 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
2538 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
2539 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
2540 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
2541 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
2542 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
2543 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
2544 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
2545 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
2546 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
2547 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
2548 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
2549 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
2550 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
2551 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
2552 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
2553 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
2555 -- Berlin, 2014-02-20
2559 * logind has gained support for facilitating privileged input
2560 and drm device access for unprivileged clients. This work is
2561 useful to allow Wayland display servers (and similar
2562 programs, such as kmscon) to run under the user's ID and
2563 access input and drm devices which are normally
2564 protected. When this is used (and the kernel is new enough)
2565 logind will "mute" IO on the file descriptors passed to
2566 Wayland as long as it is in the background and "unmute" it
2567 if it returns into the foreground. This allows secure
2568 session switching without allowing background sessions to
2569 eavesdrop on input and display data. This also introduces
2570 session switching support if VT support is turned off in the
2571 kernel, and on seats that are not seat0.
2573 * A new kernel command line option luks.options= is understood
2574 now which allows specifying LUKS options for usage for LUKS
2575 encrypted partitions specified with luks.uuid=.
2577 * tmpfiles.d(5) snippets may now use specifier expansion in
2578 path names. More specifically %m, %b, %H, %v, are now
2579 replaced by the local machine id, boot id, hostname, and
2580 kernel version number.
2582 * A new tmpfiles.d(5) command "m" has been introduced which
2583 may be used to change the owner/group/access mode of a file
2584 or directory if it exists, but do nothing if it does not.
2586 * This release removes high-level support for the
2587 MemorySoftLimit= cgroup setting. The underlying kernel
2588 cgroup attribute memory.soft_limit= is currently badly
2589 designed and likely to be removed from the kernel API in its
2590 current form, hence we should not expose it for now.
2592 * The memory.use_hierarchy cgroup attribute is now enabled for
2593 all cgroups systemd creates in the memory cgroup
2594 hierarchy. This option is likely to be come the built-in
2595 default in the kernel anyway, and the non-hierarchical mode
2596 never made much sense in the intrinsically hierarchical
2599 * A new field _SYSTEMD_SLICE= is logged along with all journal
2600 messages containing the slice a message was generated
2601 from. This is useful to allow easy per-customer filtering of
2602 logs among other things.
2604 * systemd-journald will no longer adjust the group of journal
2605 files it creates to the "systemd-journal" group. Instead we
2606 rely on the journal directory to be owned by the
2607 "systemd-journal" group, and its setgid bit set, so that the
2608 kernel file system layer will automatically enforce that
2609 journal files inherit this group assignment. The reason for
2610 this change is that we cannot allow NSS look-ups from
2611 journald which would be necessary to resolve
2612 "systemd-journal" to a numeric GID, because this might
2613 create deadlocks if NSS involves synchronous queries to
2614 other daemons (such as nscd, or sssd) which in turn are
2615 logging clients of journald and might block on it, which
2616 would then dead lock. A tmpfiles.d(5) snippet included in
2617 systemd will make sure the setgid bit and group are
2618 properly set on the journal directory if it exists on every
2619 boot. However, we recommend adjusting it manually after
2620 upgrades too (or from RPM scriptlets), so that the change is
2621 not delayed until next reboot.
2623 * Backlight and random seed files in /var/lib/ have moved into
2624 the /var/lib/systemd/ directory, in order to centralize all
2625 systemd generated files in one directory.
2627 * Boot time performance measurements (as displayed by
2628 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
2629 performance information if that's available to determine how
2630 much time BIOS and boot loader initialization required. With
2631 a sufficiently new BIOS you hence no longer need to boot
2632 with Gummiboot to get access to such information.
2634 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
2635 Cristian Rodríguez, Dave Reisner, David Herrmann, David
2636 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
2637 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
2638 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
2639 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
2640 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
2642 -- Berlin, 2013-10-02
2646 * The Restart= option for services now understands a new
2647 on-watchdog setting, which will restart the service
2648 automatically if the service stops sending out watchdog keep
2649 alive messages (as configured with WatchdogSec=).
2651 * The getty generator (which is responsible for bringing up a
2652 getty on configured serial consoles) will no longer only
2653 start a getty on the primary kernel console but on all
2654 others, too. This makes the order in which console= is
2655 specified on the kernel command line less important.
2657 * libsystemd-logind gained a new sd_session_get_vt() call to
2658 retrieve the VT number of a session.
2660 * If the option "tries=0" is set for an entry of /etc/crypttab
2661 its passphrase is queried indefinitely instead of any
2662 maximum number of tries.
2664 * If a service with a configure PID file terminates its PID
2665 file will now be removed automatically if it still exists
2666 afterwards. This should put an end to stale PID files.
2668 * systemd-run will now also take relative binary path names
2669 for execution and no longer insists on absolute paths.
2671 * InaccessibleDirectories= and ReadOnlyDirectories= now take
2672 paths that are optionally prefixed with "-" to indicate that
2673 it should not be considered a failure if they do not exist.
2675 * journalctl -o (and similar commands) now understands a new
2676 output mode "short-precise", it is similar to "short" but
2677 shows timestamps with usec accuracy.
2679 * The option "discard" (as known from Debian) is now
2680 synonymous to "allow-discards" in /etc/crypttab. In fact,
2681 "discard" is preferred now (since it is easier to remember
2684 * Some licensing clean-ups were made, so that more code is now
2685 LGPL-2.1 licensed than before.
2687 * A minimal tool to save/restore the display backlight
2688 brightness across reboots has been added. It will store the
2689 backlight setting as late as possible at shutdown, and
2690 restore it as early as possible during reboot.
2692 * A logic to automatically discover and enable home and swap
2693 partitions on GPT disks has been added. With this in place
2694 /etc/fstab becomes optional for many setups as systemd can
2695 discover certain partitions located on the root disk
2696 automatically. Home partitions are recognized under their
2697 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
2698 partitions are recognized under their GPT type ID
2699 0657fd6da4ab43c484e50933c84b4f4f.
2701 * systemd will no longer pass any environment from the kernel
2702 or initrd to system services. If you want to set an
2703 environment for all services, do so via the kernel command
2704 line systemd.setenv= assignment.
2706 * The systemd-sysctl tool no longer natively reads the file
2707 /etc/sysctl.conf. If desired, the file should be symlinked
2708 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
2709 legacy support by a symlink rather than built-in code, it
2710 also makes the otherwise hidden order of application of the
2711 different files visible. (Note that this partly reverts to a
2712 pre-198 application order of sysctl knobs!)
2714 * The "systemctl set-log-level" and "systemctl dump" commands
2715 have been moved to systemd-analyze.
2717 * systemd-run learned the new --remain-after-exit switch,
2718 which causes the scope unit not to be cleaned up
2719 automatically after the process terminated.
2721 * tmpfiles learned a new --exclude-prefix= switch to exclude
2722 certain paths from operation.
2724 * journald will now automatically flush all messages to disk
2725 as soon as a message at the log level CRIT, ALERT or EMERG
2728 Contributions from: Andrew Cook, Brandon Philips, Christian
2729 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
2730 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
2731 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
2732 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
2733 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
2734 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
2735 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
2736 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
2737 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
2738 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
2739 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
2740 William Giokas, Zbigniew Jędrzejewski-Szmek
2742 -- Berlin, 2013-09-13
2746 * The documentation has been updated to cover the various new
2747 concepts introduced with 205.
2749 * Unit files now understand the new %v specifier which
2750 resolves to the kernel version string as returned by "uname
2753 * systemctl now supports filtering the unit list output by
2754 load state, active state and sub state, using the new
2757 * "systemctl status" will now show the results of the
2758 condition checks (like ConditionPathExists= and similar) of
2759 the last start attempts of the unit. They are also logged to
2762 * "journalctl -b" may now be used to look for boot output of a
2763 specific boot. Try "journalctl -b -1" for the previous boot,
2764 but the syntax is substantially more powerful.
2766 * "journalctl --show-cursor" has been added which prints the
2767 cursor string the last shown log line. This may then be used
2768 with the new "journalctl --after-cursor=" switch to continue
2769 browsing logs from that point on.
2771 * "journalctl --force" may now be used to force regeneration
2774 * Creation of "dead" device nodes has been moved from udev
2775 into kmod and tmpfiles. Previously, udev would read the kmod
2776 databases to pre-generate dead device nodes based on meta
2777 information contained in kernel modules, so that these would
2778 be auto-loaded on access rather then at boot. As this
2779 does not really have much to do with the exposing actual
2780 kernel devices to userspace this has always been slightly
2781 alien in the udev codebase. Following the new scheme kmod
2782 will now generate a runtime snippet for tmpfiles from the
2783 module meta information and it now is tmpfiles' job to the
2784 create the nodes. This also allows overriding access and
2785 other parameters for the nodes using the usual tmpfiles
2786 facilities. As side effect this allows us to remove the
2787 CAP_SYS_MKNOD capability bit from udevd entirely.
2789 * logind's device ACLs may now be applied to these "dead"
2790 devices nodes too, thus finally allowing managed access to
2791 devices such as /dev/snd/sequencer whithout loading the
2792 backing module right-away.
2794 * A new RPM macro has been added that may be used to apply
2795 tmpfiles configuration during package installation.
2797 * systemd-detect-virt and ConditionVirtualization= now can
2798 detect User-Mode-Linux machines (UML).
2800 * journald will now implicitly log the effective capabilities
2801 set of processes in the message metadata.
2803 * systemd-cryptsetup has gained support for TrueCrypt volumes.
2805 * The initrd interface has been simplified (more specifically,
2806 support for passing performance data via environment
2807 variables and fsck results via files in /run has been
2808 removed). These features were non-essential, and are
2809 nowadays available in a much nicer way by having systemd in
2810 the initrd serialize its state and have the hosts systemd
2811 deserialize it again.
2813 * The udev "keymap" data files and tools to apply keyboard
2814 specific mappings of scan to key codes, and force-release
2815 scan code lists have been entirely replaced by a udev
2816 "keyboard" builtin and a hwdb data file.
2818 * systemd will now honour the kernel's "quiet" command line
2819 argument also during late shutdown, resulting in a
2820 completely silent shutdown when used.
2822 * There's now an option to control the SO_REUSEPORT socket
2823 option in .socket units.
2825 * Instance units will now automatically get a per-template
2826 subslice of system.slice unless something else is explicitly
2827 configured. For example, instances of sshd@.service will now
2828 implicitly be placed in system-sshd.slice rather than
2829 system.slice as before.
2831 * Test coverage support may now be enabled at build time.
2833 Contributions from: Dave Reisner, Frederic Crozat, Harald
2834 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
2835 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
2836 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
2837 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
2838 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
2839 Giokas, Zbigniew Jędrzejewski-Szmek
2841 -- Berlin, 2013-07-23
2845 * Two new unit types have been introduced:
2847 Scope units are very similar to service units, however, are
2848 created out of pre-existing processes -- instead of PID 1
2849 forking off the processes. By using scope units it is
2850 possible for system services and applications to group their
2851 own child processes (worker processes) in a powerful way
2852 which then maybe used to organize them, or kill them
2853 together, or apply resource limits on them.
2855 Slice units may be used to partition system resources in an
2856 hierarchical fashion and then assign other units to them. By
2857 default there are now three slices: system.slice (for all
2858 system services), user.slice (for all user sessions),
2859 machine.slice (for VMs and containers).
2861 Slices and scopes have been introduced primarily in
2862 context of the work to move cgroup handling to a
2863 single-writer scheme, where only PID 1
2864 creates/removes/manages cgroups.
2866 * There's a new concept of "transient" units. In contrast to
2867 normal units these units are created via an API at runtime,
2868 not from configuration from disk. More specifically this
2869 means it is now possible to run arbitrary programs as
2870 independent services, with all execution parameters passed
2871 in via bus APIs rather than read from disk. Transient units
2872 make systemd substantially more dynamic then it ever was,
2873 and useful as a general batch manager.
2875 * logind has been updated to make use of scope and slice units
2876 for managing user sessions. As a user logs in he will get
2877 his own private slice unit, to which all sessions are added
2878 as scope units. We also added support for automatically
2879 adding an instance of user@.service for the user into the
2880 slice. Effectively logind will no longer create cgroup
2881 hierarchies on its own now, it will defer entirely to PID 1
2882 for this by means of scope, service and slice units. Since
2883 user sessions this way become entities managed by PID 1
2884 the output of "systemctl" is now a lot more comprehensive.
2886 * A new mini-daemon "systemd-machined" has been added which
2887 may be used by virtualization managers to register local
2888 VMs/containers. nspawn has been updated accordingly, and
2889 libvirt will be updated shortly. machined will collect a bit
2890 of meta information about the VMs/containers, and assign
2891 them their own scope unit (see above). The collected
2892 meta-data is then made available via the "machinectl" tool,
2893 and exposed in "ps" and similar tools. machined/machinectl
2894 is compile-time optional.
2896 * As discussed earlier, the low-level cgroup configuration
2897 options ControlGroup=, ControlGroupModify=,
2898 ControlGroupPersistent=, ControlGroupAttribute= have been
2899 removed. Please use high-level attribute settings instead as
2900 well as slice units.
2902 * A new bus call SetUnitProperties() has been added to alter
2903 various runtime parameters of a unit. This is primarily
2904 useful to alter cgroup parameters dynamically in a nice way,
2905 but will be extended later on to make more properties
2906 modifiable at runtime. systemctl gained a new set-properties
2907 command that wraps this call.
2909 * A new tool "systemd-run" has been added which can be used to
2910 run arbitrary command lines as transient services or scopes,
2911 while configuring a number of settings via the command
2912 line. This tool is currently very basic, however already
2913 very useful. We plan to extend this tool to even allow
2914 queuing of execution jobs with time triggers from the
2915 command line, similar in fashion to "at".
2917 * nspawn will now inform the user explicitly that kernels with
2918 audit enabled break containers, and suggest the user to turn
2921 * Support for detecting the IMA and AppArmor security
2922 frameworks with ConditionSecurity= has been added.
2924 * journalctl gained a new "-k" switch for showing only kernel
2925 messages, mimicking dmesg output; in addition to "--user"
2926 and "--system" switches for showing only user's own logs
2929 * systemd-delta can now show information about drop-in
2930 snippets extending unit files.
2932 * libsystemd-bus has been substantially updated but is still
2933 not available as public API.
2935 * systemd will now look for the "debug" argument on the kernel
2936 command line and enable debug logging, similar to what
2937 "systemd.log_level=debug" already did before.
2939 * "systemctl set-default", "systemctl get-default" has been
2940 added to configure the default.target symlink, which
2941 controls what to boot into by default.
2943 * "systemctl set-log-level" has been added as a convenient
2944 way to raise and lower systemd logging threshold.
2946 * "systemd-analyze plot" will now show the time the various
2947 generators needed for execution, as well as information
2948 about the unit file loading.
2950 * libsystemd-journal gained a new sd_journal_open_files() call
2951 for opening specific journal files. journactl also gained a
2952 new switch to expose this new functionality. Previously we
2953 only supported opening all files from a directory, or all
2954 files from the system, as opening individual files only is
2955 racy due to journal file rotation.
2957 * systemd gained the new DefaultEnvironment= setting in
2958 /etc/systemd/system.conf to set environment variables for
2961 * If a privileged process logs a journal message with the
2962 OBJECT_PID= field set, then journald will automatically
2963 augment this with additional OBJECT_UID=, OBJECT_GID=,
2964 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
2965 system services want to log events about specific client
2966 processes. journactl/systemctl has been updated to make use
2967 of this information if all log messages regarding a specific
2970 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
2971 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
2972 Reisner, David Coppa, David King, David Strauss, Eelco
2973 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
2974 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
2975 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
2976 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
2977 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
2978 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
2979 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
2980 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
2981 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
2982 Łukasz Stelmach, 장동준
2986 * The Python bindings gained some minimal support for the APIs
2987 exposed by libsystemd-logind.
2989 * ConditionSecurity= gained support for detecting SMACK. Since
2990 this condition already supports SELinux and AppArmor we only
2991 miss IMA for this. Patches welcome!
2993 Contributions from: Karol Lewandowski, Lennart Poettering,
2994 Zbigniew Jędrzejewski-Szmek
2998 * systemd-nspawn will now create /etc/resolv.conf if
2999 necessary, before bind-mounting the host's file onto it.
3001 * systemd-nspawn will now store meta information about a
3002 container on the container's cgroup as extended attribute
3003 fields, including the root directory.
3005 * The cgroup hierarchy has been reworked in many ways. All
3006 objects any of the components systemd creates in the cgroup
3007 tree are now suffixed. More specifically, user sessions are
3008 now placed in cgroups suffixed with ".session", users in
3009 cgroups suffixed with ".user", and nspawn containers in
3010 cgroups suffixed with ".nspawn". Furthermore, all cgroup
3011 names are now escaped in a simple scheme to avoid collision
3012 of userspace object names with kernel filenames. This work
3013 is preparation for making these objects relocatable in the
3014 cgroup tree, in order to allow easy resource partitioning of
3015 these objects without causing naming conflicts.
3017 * systemctl list-dependencies gained the new switches
3018 --plain, --reverse, --after and --before.
3020 * systemd-inhibit now shows the process name of processes that
3021 have taken an inhibitor lock.
3023 * nss-myhostname will now also resolve "localhost"
3024 implicitly. This makes /etc/hosts an optional file and
3025 nicely handles that on IPv6 ::1 maps to both "localhost" and
3028 * libsystemd-logind.so gained a new call
3029 sd_get_machine_names() to enumerate running containers and
3030 VMs (currently only supported by very new libvirt and
3031 nspawn). sd_login_monitor can now be used to watch
3032 VMs/containers coming and going.
3034 * .include is not allowed recursively anymore, and only in
3035 unit files. Usually it is better to use drop-in snippets in
3036 .d/*.conf anyway, as introduced with systemd 198.
3038 * systemd-analyze gained a new "critical-chain" command that
3039 determines the slowest chain of units run during system
3040 boot-up. It is very useful for tracking down where
3041 optimizing boot time is the most beneficial.
3043 * systemd will no longer allow manipulating service paths in
3044 the name=systemd:/system cgroup tree using ControlGroup= in
3045 units. (But is still fine with it in all other dirs.)
3047 * There's a new systemd-nspawn@.service service file that may
3048 be used to easily run nspawn containers as system
3049 services. With the container's root directory in
3050 /var/lib/container/foobar it is now sufficient to run
3051 "systemctl start systemd-nspawn@foobar.service" to boot it.
3053 * systemd-cgls gained a new parameter "--machine" to list only
3054 the processes within a certain container.
3056 * ConditionSecurity= now can check for "apparmor". We still
3057 are lacking checks for SMACK and IMA for this condition
3058 check though. Patches welcome!
3060 * A new configuration file /etc/systemd/sleep.conf has been
3061 added that may be used to configure which kernel operation
3062 systemd is supposed to execute when "suspend", "hibernate"
3063 or "hybrid-sleep" is requested. This makes the new kernel
3064 "freeze" state accessible to the user.
3066 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
3067 the passed argument if applicable.
3069 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
3070 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
3071 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
3072 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
3073 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
3074 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
3075 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
3080 * The output of 'systemctl list-jobs' got some polishing. The
3081 '--type=' argument may now be passed more than once. A new
3082 command 'systemctl list-sockets' has been added which shows
3083 a list of kernel sockets systemd is listening on with the
3084 socket units they belong to, plus the units these socket
3087 * The experimental libsystemd-bus library got substantial
3088 updates to work in conjunction with the (also experimental)
3089 kdbus kernel project. It works well enough to exchange
3090 messages with some sophistication. Note that kdbus is not
3091 ready yet, and the library is mostly an elaborate test case
3092 for now, and not installable.
3094 * systemd gained a new unit 'systemd-static-nodes.service'
3095 that generates static device nodes earlier during boot, and
3096 can run in conjunction with udev.
3098 * libsystemd-login gained a new call sd_pid_get_user_unit()
3099 to retrieve the user systemd unit a process is running
3100 in. This is useful for systems where systemd is used as
3103 * systemd-nspawn now places all containers in the new /machine
3104 top-level cgroup directory in the name=systemd
3105 hierarchy. libvirt will soon do the same, so that we get a
3106 uniform separation of /system, /user and /machine for system
3107 services, user processes and containers/virtual
3108 machines. This new cgroup hierarchy is also useful to stick
3109 stable names to specific container instances, which can be
3110 recognized later this way (this name may be controlled
3111 via systemd-nspawn's new -M switch). libsystemd-login also
3112 gained a new call sd_pid_get_machine_name() to retrieve the
3113 name of the container/VM a specific process belongs to.
3115 * bootchart can now store its data in the journal.
3117 * libsystemd-journal gained a new call
3118 sd_journal_add_conjunction() for AND expressions to the
3119 matching logic. This can be used to express more complex
3120 logical expressions.
3122 * journactl can now take multiple --unit= and --user-unit=
3125 * The cryptsetup logic now understands the "luks.key=" kernel
3126 command line switch for specifying a file to read the
3127 decryption key from. Also, if a configured key file is not
3128 found the tool will now automatically fall back to prompting
3131 * Python systemd.journal module was updated to wrap recently
3132 added functions from libsystemd-journal. The interface was
3133 changed to bring the low level interface in s.j._Reader
3134 closer to the C API, and the high level interface in
3135 s.j.Reader was updated to wrap and convert all data about
3138 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
3139 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
3140 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
3141 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
3142 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
3143 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3147 * journalctl --update-catalog now understands a new --root=
3148 option to operate on catalogs found in a different root
3151 * During shutdown after systemd has terminated all running
3152 services a final killing loop kills all remaining left-over
3153 processes. We will now print the name of these processes
3154 when we send SIGKILL to them, since this usually indicates a
3157 * If /etc/crypttab refers to password files stored on
3158 configured mount points automatic dependencies will now be
3159 generated to ensure the specific mount is established first
3160 before the key file is attempted to be read.
3162 * 'systemctl status' will now show information about the
3163 network sockets a socket unit is listening on.
3165 * 'systemctl status' will also shown information about any
3166 drop-in configuration file for units. (Drop-In configuration
3167 files in this context are files such as
3168 /etc/systemd/systemd/foobar.service.d/*.conf)
3170 * systemd-cgtop now optionally shows summed up CPU times of
3171 cgroups. Press '%' while running cgtop to switch between
3172 percentage and absolute mode. This is useful to determine
3173 which cgroups use up the most CPU time over the entire
3174 runtime of the system. systemd-cgtop has also been updated
3175 to be 'pipeable' for processing with further shell tools.
3177 * 'hostnamectl set-hostname' will now allow setting of FQDN
3180 * The formatting and parsing of time span values has been
3181 changed. The parser now understands fractional expressions
3182 such as "5.5h". The formatter will now output fractional
3183 expressions for all time spans under 1min, i.e. "5.123456s"
3184 rather than "5s 123ms 456us". For time spans under 1s
3185 millisecond values are shown, for those under 1ms
3186 microsecond values are shown. This should greatly improve
3187 all time-related output of systemd.
3189 * libsystemd-login and libsystemd-journal gained new
3190 functions for querying the poll() events mask and poll()
3191 timeout value for integration into arbitrary event
3194 * localectl gained the ability to list available X11 keymaps
3195 (models, layouts, variants, options).
3197 * 'systemd-analyze dot' gained the ability to filter for
3198 specific units via shell-style globs, to create smaller,
3199 more useful graphs. I.e. it is now possible to create simple
3200 graphs of all the dependencies between only target units, or
3201 of all units that Avahi has dependencies with.
3203 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
3204 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
3205 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
3206 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
3207 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
3208 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
3209 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
3213 * The boot-time readahead implementation for rotating media
3214 will now read the read-ahead data in multiple passes which
3215 consist of all read requests made in equidistant time
3216 intervals. This means instead of strictly reading read-ahead
3217 data in its physical order on disk we now try to find a
3218 middle ground between physical and access time order.
3220 * /etc/os-release files gained a new BUILD_ID= field for usage
3221 on operating systems that provide continuous builds of OS
3224 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
3225 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
3226 William Douglas, Zbigniew Jędrzejewski-Szmek
3230 * systemd-python gained an API exposing libsystemd-daemon.
3232 * The SMACK setup logic gained support for uploading CIPSO
3235 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
3236 ReadOnlyDirectories= and InaccessibleDirectories= has
3237 changed. The private /tmp and /var/tmp directories are now
3238 shared by all processes of a service (which means
3239 ExecStartPre= may now leave data in /tmp that ExecStart= of
3240 the same service can still access). When a service is
3241 stopped its temporary directories are immediately deleted
3242 (normal clean-up with tmpfiles is still done in addition to
3245 * By default, systemd will now set a couple of sysctl
3246 variables in the kernel: the safe sysrq options are turned
3247 on, IP route verification is turned on, and source routing
3248 disabled. The recently added hardlink and softlink
3249 protection of the kernel is turned on. These settings should
3250 be reasonably safe, and good defaults for all new systems.
3252 * The predictable network naming logic may now be turned off
3253 with a new kernel command line switch: net.ifnames=0.
3255 * A new libsystemd-bus module has been added that implements a
3256 pretty complete D-Bus client library. For details see:
3258 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
3260 * journald will now explicitly flush the journal files to disk
3261 at the latest 5min after each write. The file will then also
3262 be marked offline until the next write. This should increase
3263 reliability in case of a crash. The synchronization delay
3264 can be configured via SyncIntervalSec= in journald.conf.
3266 * There's a new remote-fs-setup.target unit that can be used
3267 to pull in specific services when at least one remote file
3268 system is to be mounted.
3270 * There are new targets timers.target and paths.target as
3271 canonical targets to pull user timer and path units in
3272 from. This complements sockets.target with a similar
3273 purpose for socket units.
3275 * libudev gained a new call udev_device_set_attribute_value()
3276 to set sysfs attributes of a device.
3278 * The udev daemon now sets the default number of worker
3279 processes executed in parallel based on the number of available
3280 CPUs instead of the amount of available RAM. This is supposed
3281 to provide a more reliable default and limit a too aggressive
3282 paralellism for setups with 1000s of devices connected.
3284 Contributions from: Auke Kok, Colin Walters, Cristian
3285 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
3286 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
3287 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
3288 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
3289 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
3290 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
3291 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
3292 Zbigniew Jędrzejewski-Szmek
3296 * Configuration of unit files may now be extended via drop-in
3297 files without having to edit/override the unit files
3298 themselves. More specifically, if the administrator wants to
3299 change one value for a service file foobar.service he can
3300 now do so by dropping in a configuration snippet into
3301 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
3302 will load all these snippets and apply them on top of the
3303 main unit configuration file, possibly extending or
3304 overriding its settings. Using these drop-in snippets is
3305 generally nicer than the two earlier options for changing
3306 unit files locally: copying the files from
3307 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
3308 them there; or creating a new file in /etc/systemd/system/
3309 that incorporates the original one via ".include". Drop-in
3310 snippets into these .d/ directories can be placed in any
3311 directory systemd looks for units in, and the usual
3312 overriding semantics between /usr/lib, /etc and /run apply
3315 * Most unit file settings which take lists of items can now be
3316 reset by assigning the empty string to them. For example,
3317 normally, settings such as Environment=FOO=BAR append a new
3318 environment variable assignment to the environment block,
3319 each time they are used. By assigning Environment= the empty
3320 string the environment block can be reset to empty. This is
3321 particularly useful with the .d/*.conf drop-in snippets
3322 mentioned above, since this adds the ability to reset list
3323 settings from vendor unit files via these drop-ins.
3325 * systemctl gained a new "list-dependencies" command for
3326 listing the dependencies of a unit recursively.
3328 * Inhibitors are now honored and listed by "systemctl
3329 suspend", "systemctl poweroff" (and similar) too, not only
3330 GNOME. These commands will also list active sessions by
3333 * Resource limits (as exposed by the various control group
3334 controllers) can now be controlled dynamically at runtime
3335 for all units. More specifically, you can now use a command
3336 like "systemctl set-cgroup-attr foobar.service cpu.shares
3337 2000" to alter the CPU shares a specific service gets. These
3338 settings are stored persistently on disk, and thus allow the
3339 administrator to easily adjust the resource usage of
3340 services with a few simple commands. This dynamic resource
3341 management logic is also available to other programs via the
3342 bus. Almost any kernel cgroup attribute and controller is
3345 * systemd-vconsole-setup will now copy all font settings to
3346 all allocated VTs, where it previously applied them only to
3349 * libsystemd-login gained the new sd_session_get_tty() API
3352 * This release drops support for a few legacy or
3353 distribution-specific LSB facility names when parsing init
3354 scripts: $x-display-manager, $mail-transfer-agent,
3355 $mail-transport-agent, $mail-transfer-agent, $smtp,
3356 $null. Also, the mail-transfer-agent.target unit backing
3357 this has been removed. Distributions which want to retain
3358 compatibility with this should carry the burden for
3359 supporting this themselves and patch support for these back
3360 in, if they really need to. Also, the facilities $syslog and
3361 $local_fs are now ignored, since systemd does not support
3362 early-boot LSB init scripts anymore, and these facilities
3363 are implied anyway for normal services. syslog.target has
3366 * There are new bus calls on PID1's Manager object for
3367 cancelling jobs, and removing snapshot units. Previously,
3368 both calls were only available on the Job and Snapshot
3371 * systemd-journal-gatewayd gained SSL support.
3373 * The various "environment" files, such as /etc/locale.conf
3374 now support continuation lines with a backslash ("\") as
3375 last character in the line, similarly in style (but different)
3376 to how this is supported in shells.
3378 * For normal user processes the _SYSTEMD_USER_UNIT= field is
3379 now implicitly appended to every log entry logged. systemctl
3380 has been updated to filter by this field when operating on a
3381 user systemd instance.
3383 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
3384 CAP_AUDIT_CONTROL capabilities to the capabilities set for
3385 the container. This makes it easier to boot unmodified
3386 Fedora systems in a container, which however still requires
3387 audit=0 to be passed on the kernel command line. Auditing in
3388 kernel and userspace is unfortunately still too broken in
3389 context of containers, hence we recommend compiling it out
3390 of the kernel or using audit=0. Hopefully this will be fixed
3391 one day for good in the kernel.
3393 * nspawn gained the new --bind= and --bind-ro= parameters to
3394 bind mount specific directories from the host into the
3397 * nspawn will now mount its own devpts file system instance
3398 into the container, in order not to leak pty devices from
3399 the host into the container.
3401 * systemd will now read the firmware boot time performance
3402 information from the EFI variables, if the used boot loader
3403 supports this, and takes it into account for boot performance
3404 analysis via "systemd-analyze". This is currently supported
3405 only in conjunction with Gummiboot, but could be supported
3406 by other boot loaders too. For details see:
3408 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
3410 * A new generator has been added that automatically mounts the
3411 EFI System Partition (ESP) to /boot, if that directory
3412 exists, is empty, and no other file system has been
3413 configured to be mounted there.
3415 * logind will now send out PrepareForSleep(false) out
3416 unconditionally, after coming back from suspend. This may be
3417 used by applications as asynchronous notification for
3418 system resume events.
3420 * "systemctl unlock-sessions" has been added, that allows
3421 unlocking the screens of all user sessions at once, similar
3422 to how "systemctl lock-sessions" already locked all users
3423 sessions. This is backed by a new D-Bus call UnlockSessions().
3425 * "loginctl seat-status" will now show the master device of a
3426 seat. (i.e. the device of a seat that needs to be around for
3427 the seat to be considered available, usually the graphics
3430 * tmpfiles gained a new "X" line type, that allows
3431 configuration of files and directories (with wildcards) that
3432 shall be excluded from automatic cleanup ("aging").
3434 * udev default rules set the device node permissions now only
3435 at "add" events, and do not change them any longer with a
3436 later "change" event.
3438 * The log messages for lid events and power/sleep keypresses
3439 now carry a message ID.
3441 * We now have a substantially larger unit test suite, but this
3442 continues to be work in progress.
3444 * udevadm hwdb gained a new --root= parameter to change the
3445 root directory to operate relative to.
3447 * logind will now issue a background sync() request to the kernel
3448 early at shutdown, so that dirty buffers are flushed to disk early
3449 instead of at the last moment, in order to optimize shutdown
3452 * A new bootctl tool has been added that is an interface for
3453 certain boot loader operations. This is currently a preview
3454 and is likely to be extended into a small mechanism daemon
3455 like timedated, localed, hostnamed, and can be used by
3456 graphical UIs to enumerate available boot options, and
3457 request boot into firmware operations.
3459 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
3460 the rest of the package. It also has been updated to work
3461 correctly in initrds.
3463 * Policykit previously has been runtime optional, and is now
3464 also compile time optional via a configure switch.
3466 * systemd-analyze has been reimplemented in C. Also "systemctl
3467 dot" has moved into systemd-analyze.
3469 * "systemctl status" with no further parameters will now print
3470 the status of all active or failed units.
3472 * Operations such as "systemctl start" can now be executed
3473 with a new mode "--irreversible" which may be used to queue
3474 operations that cannot accidentally be reversed by a later
3475 job queuing. This is by default used to make shutdown
3476 requests more robust.
3478 * The Python API of systemd now gained a new module for
3479 reading journal files.
3481 * A new tool kernel-install has been added that can install
3482 kernel images according to the Boot Loader Specification:
3484 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
3486 * Boot time console output has been improved to provide
3487 animated boot time output for hanging jobs.
3489 * A new tool systemd-activate has been added which can be used
3490 to test socket activation with, directly from the command
3491 line. This should make it much easier to test and debug
3492 socket activation in daemons.
3494 * journalctl gained a new "--reverse" (or -r) option to show
3495 journal output in reverse order (i.e. newest line first).
3497 * journalctl gained a new "--pager-end" (or -e) option to jump
3498 to immediately jump to the end of the journal in the
3499 pager. This is only supported in conjunction with "less".
3501 * journalctl gained a new "--user-unit=" option, that works
3502 similarly to "--unit=" but filters for user units rather than
3505 * A number of unit files to ease adoption of systemd in
3506 initrds has been added. This moves some minimal logic from
3507 the various initrd implementations into systemd proper.
3509 * The journal files are now owned by a new group
3510 "systemd-journal", which exists specifically to allow access
3511 to the journal, and nothing else. Previously, we used the
3512 "adm" group for that, which however possibly covers more
3513 than just journal/log file access. This new group is now
3514 already used by systemd-journal-gatewayd to ensure this
3515 daemon gets access to the journal files and as little else
3516 as possible. Note that "make install" will also set FS ACLs
3517 up for /var/log/journal to give "adm" and "wheel" read
3518 access to it, in addition to "systemd-journal" which owns
3519 the journal files. We recommend that packaging scripts also
3520 add read access to "adm" + "wheel" to /var/log/journal, and
3521 all existing/future journal files. To normal users and
3522 administrators little changes, however packagers need to
3523 ensure to create the "systemd-journal" system group at
3524 package installation time.
3526 * The systemd-journal-gatewayd now runs as unprivileged user
3527 systemd-journal-gateway:systemd-journal-gateway. Packaging
3528 scripts need to create these system user/group at
3531 * timedated now exposes a new boolean property CanNTP that
3532 indicates whether a local NTP service is available or not.
3534 * systemd-detect-virt will now also detect xen PVs
3536 * The pstore file system is now mounted by default, if it is
3539 * In addition to the SELinux and IMA policies we will now also
3540 load SMACK policies at early boot.
3542 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
3543 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
3544 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
3545 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
3546 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
3547 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
3548 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
3549 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
3550 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
3551 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
3552 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
3553 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
3554 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
3555 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
3559 * Timer units now support calendar time events in addition to
3560 monotonic time events. That means you can now trigger a unit
3561 based on a calendar time specification such as "Thu,Fri
3562 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
3563 or fifth day of any month of the year 2013, given that it is
3564 a thursday or friday. This brings timer event support
3565 considerably closer to cron's capabilities. For details on
3566 the supported calendar time specification language see
3569 * udev now supports a number of different naming policies for
3570 network interfaces for predictable names, and a combination
3571 of these policies is now the default. Please see this wiki
3572 document for details:
3574 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
3576 * Auke Kok's bootchart implementation has been added to the
3577 systemd tree. It is an optional component that can graph the
3578 boot in quite some detail. It is one of the best bootchart
3579 implementations around and minimal in its code and
3582 * nss-myhostname has been integrated into the systemd source
3583 tree. nss-myhostname guarantees that the local hostname
3584 always stays resolvable via NSS. It has been a weak
3585 requirement of systemd-hostnamed since a long time, and
3586 since its code is actually trivial we decided to just
3587 include it in systemd's source tree. It can be turned off
3588 with a configure switch.
3590 * The read-ahead logic is now capable of properly detecting
3591 whether a btrfs file system is on SSD or rotating media, in
3592 order to optimize the read-ahead scheme. Previously, it was
3593 only capable of detecting this on traditional file systems
3596 * In udev, additional device properties are now read from the
3597 IAB in addition to the OUI database. Also, Bluetooth company
3598 identities are attached to the devices as well.
3600 * In service files %U may be used as specifier that is
3601 replaced by the configured user name of the service.
3603 * nspawn may now be invoked without a controlling TTY. This
3604 makes it suitable for invocation as its own service. This
3605 may be used to set up a simple containerized server system
3606 using only core OS tools.
3608 * systemd and nspawn can now accept socket file descriptors
3609 when they are started for socket activation. This enables
3610 implementation of socket activated nspawn
3611 containers. i.e. think about autospawning an entire OS image
3612 when the first SSH or HTTP connection is received. We expect
3613 that similar functionality will also be added to libvirt-lxc
3616 * journalctl will now suppress ANSI color codes when
3617 presenting log data.
3619 * systemctl will no longer show control group information for
3620 a unit if a the control group is empty anyway.
3622 * logind can now automatically suspend/hibernate/shutdown the
3625 * /etc/machine-info and hostnamed now also expose the chassis
3626 type of the system. This can be used to determine whether
3627 the local system is a laptop, desktop, handset or
3628 tablet. This information may either be configured by the
3629 user/vendor or is automatically determined from ACPI and DMI
3630 information if possible.
3632 * A number of PolicyKit actions are now bound together with
3633 "imply" rules. This should simplify creating UIs because
3634 many actions will now authenticate similar ones as well.
3636 * Unit files learnt a new condition ConditionACPower= which
3637 may be used to conditionalize a unit depending on whether an
3638 AC power source is connected or not, of whether the system
3639 is running on battery power.
3641 * systemctl gained a new "is-failed" verb that may be used in
3642 shell scripts and suchlike to check whether a specific unit
3643 is in the "failed" state.
3645 * The EnvironmentFile= setting in unit files now supports file
3646 globbing, and can hence be used to easily read a number of
3647 environment files at once.
3649 * systemd will no longer detect and recognize specific
3650 distributions. All distribution-specific #ifdeffery has been
3651 removed, systemd is now fully generic and
3652 distribution-agnostic. Effectively, not too much is lost as
3653 a lot of the code is still accessible via explicit configure
3654 switches. However, support for some distribution specific
3655 legacy configuration file formats has been dropped. We
3656 recommend distributions to simply adopt the configuration
3657 files everybody else uses now and convert the old
3658 configuration from packaging scripts. Most distributions
3659 already did that. If that's not possible or desirable,
3660 distributions are welcome to forward port the specific
3661 pieces of code locally from the git history.
3663 * When logging a message about a unit systemd will now always
3664 log the unit name in the message meta data.
3666 * localectl will now also discover system locale data that is
3667 not stored in locale archives, but directly unpacked.
3669 * logind will no longer unconditionally use framebuffer
3670 devices as seat masters, i.e. as devices that are required
3671 to be existing before a seat is considered preset. Instead,
3672 it will now look for all devices that are tagged as
3673 "seat-master" in udev. By default framebuffer devices will
3674 be marked as such, but depending on local systems other
3675 devices might be marked as well. This may be used to
3676 integrate graphics cards using closed source drivers (such
3677 as NVidia ones) more nicely into logind. Note however, that
3678 we recommend using the open source NVidia drivers instead,
3679 and no udev rules for the closed-source drivers will be
3680 shipped from us upstream.
3682 Contributions from: Adam Williamson, Alessandro Crismani, Auke
3683 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
3684 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
3685 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
3686 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
3687 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
3688 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
3689 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
3690 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
3691 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
3692 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
3693 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
3698 * udev gained support for loading additional device properties
3699 from an indexed database that is keyed by vendor/product IDs
3700 and similar device identifiers. For the beginning this
3701 "hwdb" is populated with data from the well-known PCI and
3702 USB database, but also includes PNP, ACPI and OID data. In
3703 the longer run this indexed database shall grow into
3704 becoming the one central database for non-essential
3705 userspace device metadata. Previously, data from the PCI/USB
3706 database was only attached to select devices, since the
3707 lookup was a relatively expensive operation due to O(n) time
3708 complexity (with n being the number of entries in the
3709 database). Since this is now O(1), we decided to add in this
3710 data for all devices where this is available, by
3711 default. Note that the indexed database needs to be rebuilt
3712 when new data files are installed. To achieve this you need
3713 to update your packaging scripts to invoke "udevadm hwdb
3714 --update" after installation of hwdb data files. For
3715 RPM-based distributions we introduced the new
3716 %udev_hwdb_update macro for this purpose.
3718 * The Journal gained support for the "Message Catalog", an
3719 indexed database to link up additional information with
3720 journal entries. For further details please check:
3722 http://www.freedesktop.org/wiki/Software/systemd/catalog
3724 The indexed message catalog database also needs to be
3725 rebuilt after installation of message catalog files. Use
3726 "journalctl --update-catalog" for this. For RPM-based
3727 distributions we introduced the %journal_catalog_update
3728 macro for this purpose.
3730 * The Python Journal bindings gained support for the standard
3731 Python logging framework.
3733 * The Journal API gained new functions for checking whether
3734 the underlying file system of a journal file is capable of
3735 properly reporting file change notifications, or whether
3736 applications that want to reflect journal changes "live"
3737 need to recheck journal files continuously in appropriate
3740 * It is now possible to set the "age" field for tmpfiles
3741 entries to 0, indicating that files matching this entry
3742 shall always be removed when the directories are cleaned up.
3744 * coredumpctl gained a new "gdb" verb which invokes gdb
3745 right-away on the selected coredump.
3747 * There's now support for "hybrid sleep" on kernels that
3748 support this, in addition to "suspend" and "hibernate". Use
3749 "systemctl hybrid-sleep" to make use of this.
3751 * logind's HandleSuspendKey= setting (and related settings)
3752 now gained support for a new "lock" setting to simply
3753 request the screen lock on all local sessions, instead of
3754 actually executing a suspend or hibernation.
3756 * systemd will now mount the EFI variables file system by
3759 * Socket units now gained support for configuration of the
3760 SMACK security label.
3762 * timedatectl will now output the time of the last and next
3763 daylight saving change.
3765 * We dropped support for various legacy and distro-specific
3766 concepts, such as insserv, early-boot SysV services
3767 (i.e. those for non-standard runlevels such as 'b' or 'S')
3768 or ArchLinux /etc/rc.conf support. We recommend the
3769 distributions who still need support this to either continue
3770 to maintain the necessary patches downstream, or find a
3771 different solution. (Talk to us if you have questions!)
3773 * Various systemd components will now bypass PolicyKit checks
3774 for root and otherwise handle properly if PolicyKit is not
3775 found to be around. This should fix most issues for
3776 PolicyKit-less systems. Quite frankly this should have been
3777 this way since day one. It is absolutely our intention to
3778 make systemd work fine on PolicyKit-less systems, and we
3779 consider it a bug if something does not work as it should if
3780 PolicyKit is not around.
3782 * For embedded systems it is now possible to build udev and
3783 systemd without blkid and/or kmod support.
3785 * "systemctl switch-root" is now capable of switching root
3786 more than once. I.e. in addition to transitions from the
3787 initrd to the host OS it is now possible to transition to
3788 further OS images from the host. This is useful to implement
3789 offline updating tools.
3791 * Various other additions have been made to the RPM macros
3792 shipped with systemd. Use %udev_rules_update() after
3793 installing new udev rules files. %_udevhwdbdir,
3794 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
3795 %_sysctldir are now available which resolve to the right
3796 directories for packages to place various data files in.
3798 * journalctl gained the new --full switch (in addition to
3799 --all, to disable ellipsation for long messages.
3801 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
3802 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
3803 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
3804 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
3805 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
3806 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
3807 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
3808 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
3809 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
3813 * journalctl gained new --since= and --until= switches to
3814 filter by time. It also now supports nice filtering for
3815 units via --unit=/-u.
3817 * Type=oneshot services may use ExecReload= and do the
3820 * The journal daemon now supports time-based rotation and
3821 vacuuming, in addition to the usual disk-space based
3824 * The journal will now index the available field values for
3825 each field name. This enables clients to show pretty drop
3826 downs of available match values when filtering. The bash
3827 completion of journalctl has been updated
3828 accordingly. journalctl gained a new switch -F to list all
3829 values a certain field takes in the journal database.
3831 * More service events are now written as structured messages
3832 to the journal, and made recognizable via message IDs.
3834 * The timedated, localed and hostnamed mini-services which
3835 previously only provided support for changing time, locale
3836 and hostname settings from graphical DEs such as GNOME now
3837 also have a minimal (but very useful) text-based client
3838 utility each. This is probably the nicest way to changing
3839 these settings from the command line now, especially since
3840 it lists available options and is fully integrated with bash
3843 * There's now a new tool "systemd-coredumpctl" to list and
3844 extract coredumps from the journal.
3846 * We now install a README each in /var/log/ and
3847 /etc/rc.d/init.d explaining where the system logs and init
3848 scripts went. This hopefully should help folks who go to
3849 that dirs and look into the otherwise now empty void and
3850 scratch their heads.
3852 * When user-services are invoked (by systemd --user) the
3853 $MANAGERPID env var is set to the PID of systemd.
3855 * SIGRTMIN+24 when sent to a --user instance will now result
3856 in immediate termination of systemd.
3858 * gatewayd received numerous feature additions such as a
3859 "follow" mode, for live syncing and filtering.
3861 * browse.html now allows filtering and showing detailed
3862 information on specific entries. Keyboard navigation and
3863 mouse screen support has been added.
3865 * gatewayd/journalctl now supports HTML5/JSON
3866 Server-Sent-Events as output.
3868 * The SysV init script compatibility logic will now
3869 heuristically determine whether a script supports the
3870 "reload" verb, and only then make this available as
3873 * "systemctl status --follow" has been removed, use "journalctl
3876 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
3877 have been removed since they are hardly useful to be
3880 * And I'd like to take the opportunity to specifically mention
3881 Zbigniew for his great contributions. Zbigniew, you rock!
3883 Contributions from: Andrew Eikum, Christian Hesse, Colin
3884 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
3885 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
3886 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
3887 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
3888 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
3889 Jędrzejewski-Szmek, Сковорода Никита Андреевич
3893 * If /etc/vconsole.conf is non-existent or empty we will no
3894 longer load any console font or key map at boot by
3895 default. Instead the kernel defaults will be left
3896 intact. This is definitely the right thing to do, as no
3897 configuration should mean no configuration, and hard-coding
3898 font names that are different on all archs is probably a bad
3899 idea. Also, the kernel default key map and font should be
3900 good enough for most cases anyway, and mostly identical to
3901 the userspace fonts/key maps we previously overloaded them
3902 with. If distributions want to continue to default to a
3903 non-kernel font or key map they should ship a default
3904 /etc/vconsole.conf with the appropriate contents.
3906 Contributions from: Colin Walters, Daniel J Walsh, Dave
3907 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
3908 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3912 * journalctl gained a new --cursor= switch to show entries
3913 starting from the specified location in the journal.
3915 * We now enforce a size limit on journal entry fields exported
3916 with "-o json" in journalctl. Fields larger than 4K will be
3917 assigned null. This can be turned off with --all.
3919 * An (optional) journal gateway daemon is now available as
3920 "systemd-journal-gatewayd.service". This service provides
3921 access to the journal via HTTP and JSON. This functionality
3922 will be used to implement live log synchronization in both
3923 pull and push modes, but has various other users too, such
3924 as easy log access for debugging of embedded devices. Right
3925 now it is already useful to retrieve the journal via HTTP:
3927 # systemctl start systemd-journal-gatewayd.service
3928 # wget http://localhost:19531/entries
3930 This will download the journal contents in a
3931 /var/log/messages compatible format. The same as JSON:
3933 # curl -H"Accept: application/json" http://localhost:19531/entries
3935 This service is also accessible via a web browser where a
3936 single static HTML5 app is served that uses the JSON logic
3937 to enable the user to do some basic browsing of the
3938 journal. This will be extended later on. Here's an example
3939 screenshot of this app in its current state:
3941 http://0pointer.de/public/journal-gatewayd
3943 Contributions from: Kay Sievers, Lennart Poettering, Robert
3944 Milasan, Tom Gundersen
3948 * The bash completion logic is now available for journalctl
3951 * We do not mount the "cpuset" controller anymore together with
3952 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
3953 started if no parameters are assigned to it. "cpuset" hence
3954 broke code that assumed it it could create "cpu" groups and
3957 * journalctl -f will now subscribe to terminal size changes,
3958 and line break accordingly.
3960 Contributions from: Dave Reisner, Kay Sievers, Lennart
3961 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
3965 * nspawn will now create a symlink /etc/localtime in the
3966 container environment, copying the host's timezone
3967 setting. Previously this has been done via a bind mount, but
3968 since symlinks cannot be bind mounted this has now been
3969 changed to create/update the appropriate symlink.
3971 * journalctl -n's line number argument is now optional, and
3972 will default to 10 if omitted.
3974 * journald will now log the maximum size the journal files may
3975 take up on disk. This is particularly useful if the default
3976 built-in logic of determining this parameter from the file
3977 system size is used. Use "systemctl status
3978 systemd-journald.service" to see this information.
3980 * The multi-seat X wrapper tool has been stripped down. As X
3981 is now capable of enumerating graphics devices via udev in a
3982 seat-aware way the wrapper is not strictly necessary
3983 anymore. A stripped down temporary stop-gap is still shipped
3984 until the upstream display managers have been updated to
3985 fully support the new X logic. Expect this wrapper to be
3986 removed entirely in one of the next releases.
3988 * HandleSleepKey= in logind.conf has been split up into
3989 HandleSuspendKey= and HandleHibernateKey=. The old setting
3990 is not available anymore. X11 and the kernel are
3991 distuingishing between these keys and we should too. This
3992 also means the inhibition lock for these keys has been split
3995 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
3996 Poettering, Lukas Nykryn, Václav Pavlín
4000 * Whenever a unit changes state we will now log this to the
4001 journal and show along the unit's own log output in
4004 * ConditionPathIsMountPoint= can now properly detect bind
4005 mount points too. (Previously, a bind mount of one file
4006 system to another place in the same file system could not be
4007 detected as mount, since they shared struct stat's st_dev
4010 * We will now mount the cgroup controllers cpu, cpuacct,
4011 cpuset and the controllers net_cls, net_prio together by
4014 * nspawn containers will now have a virtualized boot
4015 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
4016 over with a randomized ID at container initialization). This
4017 has the effect of making "journalctl -b" do the right thing
4020 * The JSON output journal serialization has been updated not
4021 to generate "endless" list objects anymore, but rather one
4022 JSON object per line. This is more in line how most JSON
4023 parsers expect JSON objects. The new output mode
4024 "json-pretty" has been added to provide similar output, but
4025 neatly aligned for readability by humans.
4027 * We dropped all explicit sync() invocations in the shutdown
4028 code. The kernel does this implicitly anyway in the kernel
4029 reboot() syscall. halt(8)'s -n option is now a compatibility
4032 * We now support virtualized reboot() in containers, as
4033 supported by newer kernels. We will fall back to exit() if
4034 CAP_SYS_REBOOT is not available to the container. Also,
4035 nspawn makes use of this now and will actually reboot the
4036 container if the containerized OS asks for that.
4038 * journalctl will only show local log output by default
4039 now. Use --merge (-m) to show remote log output, too.
4041 * libsystemd-journal gained the new sd_journal_get_usage()
4042 call to determine the current disk usage of all journal
4043 files. This is exposed in the new "journalctl --disk-usage"
4046 * journald gained a new configuration setting SplitMode= in
4047 journald.conf which may be used to control how user journals
4048 are split off. See journald.conf(5) for details.
4050 * A new condition type ConditionFileNotEmpty= has been added.
4052 * tmpfiles' "w" lines now support file globbing, to write
4053 multiple files at once.
4055 * We added Python bindings for the journal submission
4056 APIs. More Python APIs for a number of selected APIs will
4057 likely follow. Note that we intend to add native bindings
4058 only for the Python language, as we consider it common
4059 enough to deserve bindings shipped within systemd. There are
4060 various projects outside of systemd that provide bindings
4061 for languages such as PHP or Lua.
4063 * Many conditions will now resolve specifiers such as %i. In
4064 addition, PathChanged= and related directives of .path units
4065 now support specifiers as well.
4067 * There's now a new RPM macro definition for the system preset
4070 * journald will now warn if it ca not forward a message to the
4071 syslog daemon because its socket is full.
4073 * timedated will no longer write or process /etc/timezone,
4074 except on Debian. As we do not support late mounted /usr
4075 anymore /etc/localtime always being a symlink is now safe,
4076 and hence the information in /etc/timezone is not necessary
4079 * logind will now always reserve one VT for a text getty (VT6
4080 by default). Previously if more than 6 X sessions where
4081 started they took up all the VTs with auto-spawned gettys,
4082 so that no text gettys were available anymore.
4084 * udev will now automatically inform the btrfs kernel logic
4085 about btrfs RAID components showing up. This should make
4086 simple hotplug based btrfs RAID assembly work.
4088 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
4089 (but not for its children which will stay at the kernel
4090 default). This should allow setups with a lot more listening
4093 * systemd will now always pass the configured timezone to the
4094 kernel at boot. timedated will do the same when the timezone
4097 * logind's inhibition logic has been updated. By default,
4098 logind will now handle the lid switch, the power and sleep
4099 keys all the time, even in graphical sessions. If DEs want
4100 to handle these events on their own they should take the new
4101 handle-power-key, handle-sleep-key and handle-lid-switch
4102 inhibitors during their runtime. A simple way to achieve
4103 that is to invoke the DE wrapped in an invocation of:
4105 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
4107 * Access to unit operations is now checked via SELinux taking
4108 the unit file label and client process label into account.
4110 * systemd will now notify the administrator in the journal
4111 when he over-mounts a non-empty directory.
4113 * There are new specifiers that are resolved in unit files,
4114 for the host name (%H), the machine ID (%m) and the boot ID
4117 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
4118 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
4119 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
4120 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
4121 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
4122 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
4123 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
4127 * Support for reading structured kernel messages from
4128 /dev/kmsg has now been added and is enabled by default.
4130 * Support for reading kernel messages from /proc/kmsg has now
4131 been removed. If you want kernel messages in the journal
4132 make sure to run a recent kernel (>= 3.5) that supports
4133 reading structured messages from /dev/kmsg (see
4134 above). /proc/kmsg is now exclusive property of classic
4135 syslog daemons again.
4137 * The libudev API gained the new
4138 udev_device_new_from_device_id() call.
4140 * The logic for file system namespace (ReadOnlyDirectory=,
4141 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
4142 require pivot_root() anymore. This means fewer temporary
4143 directories are created below /tmp for this feature.
4145 * nspawn containers will now see and receive all submounts
4146 made on the host OS below the root file system of the
4149 * Forward Secure Sealing is now supported for Journal files,
4150 which provide cryptographical sealing of journal files so
4151 that attackers cannot alter log history anymore without this
4152 being detectable. Lennart will soon post a blog story about
4153 this explaining it in more detail.
4155 * There are two new service settings RestartPreventExitStatus=
4156 and SuccessExitStatus= which allow configuration of exit
4157 status (exit code or signal) which will be excepted from the
4158 restart logic, resp. consider successful.
4160 * journalctl gained the new --verify switch that can be used
4161 to check the integrity of the structure of journal files and
4162 (if Forward Secure Sealing is enabled) the contents of
4165 * nspawn containers will now be run with /dev/stdin, /dev/fd/
4166 and similar symlinks pre-created. This makes running shells
4167 as container init process a lot more fun.
4169 * The fstab support can now handle PARTUUID= and PARTLABEL=
4172 * A new ConditionHost= condition has been added to match
4173 against the hostname (with globs) and machine ID. This is
4174 useful for clusters where a single OS image is used to
4175 provision a large number of hosts which shall run slightly
4176 different sets of services.
4178 * Services which hit the restart limit will now be placed in a
4181 Contributions from: Bertram Poettering, Dave Reisner, Huang
4182 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
4183 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
4187 * When running in --user mode systemd will now become a
4188 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
4189 tree a lot more organized.
4191 * A new PartOf= unit dependency type has been introduced that
4192 may be used to group services in a natural way.
4194 * "systemctl enable" may now be used to enable instances of
4197 * journalctl now prints error log levels in red, and
4198 warning/notice log levels in bright white. It also supports
4199 filtering by log level now.
4201 * cgtop gained a new -n switch (similar to top), to configure
4202 the maximum number of iterations to run for. It also gained
4203 -b, to run in batch mode (accepting no input).
4205 * The suffix ".service" may now be omitted on most systemctl
4206 command lines involving service unit names.
4208 * There's a new bus call in logind to lock all sessions, as
4209 well as a loginctl verb for it "lock-sessions".
4211 * libsystemd-logind.so gained a new call sd_journal_perror()
4212 that works similar to libc perror() but logs to the journal
4213 and encodes structured information about the error number.
4215 * /etc/crypttab entries now understand the new keyfile-size=
4218 * shutdown(8) now can send a (configurable) wall message when
4219 a shutdown is cancelled.
4221 * The mount propagation mode for the root file system will now
4222 default to "shared", which is useful to make containers work
4223 nicely out-of-the-box so that they receive new mounts from
4224 the host. This can be undone locally by running "mount
4225 --make-rprivate /" if needed.
4227 * The prefdm.service file has been removed. Distributions
4228 should maintain this unit downstream if they intend to keep
4229 it around. However, we recommend writing normal unit files
4230 for display managers instead.
4232 * Since systemd is a crucial part of the OS we will now
4233 default to a number of compiler switches that improve
4234 security (hardening) such as read-only relocations, stack
4235 protection, and suchlike.
4237 * The TimeoutSec= setting for services is now split into
4238 TimeoutStartSec= and TimeoutStopSec= to allow configuration
4239 of individual time outs for the start and the stop phase of
4242 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
4243 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
4244 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
4245 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
4246 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
4247 Gundersen, Zbigniew Jędrzejewski-Szmek
4251 * The journal and id128 C APIs are now fully documented as man
4254 * Extra safety checks have been added when transitioning from
4255 the initial RAM disk to the main system to avoid accidental
4258 * /etc/crypttab entries now understand the new keyfile-offset=
4261 * systemctl -t can now be used to filter by unit load state.
4263 * The journal C API gained the new sd_journal_wait() call to
4264 make writing synchronous journal clients easier.
4266 * journalctl gained the new -D switch to show journals from a
4269 * journalctl now displays a special marker between log
4270 messages of two different boots.
4272 * The journal is now explicitly flushed to /var via a service
4273 systemd-journal-flush.service, rather than implicitly simply
4274 by seeing /var/log/journal to be writable.
4276 * journalctl (and the journal C APIs) can now match for much
4277 more complex expressions, with alternatives and
4280 * When transitioning from the initial RAM disk to the main
4281 system we will now kill all processes in a killing spree to
4282 ensure no processes stay around by accident.
4284 * Three new specifiers may be used in unit files: %u, %h, %s
4285 resolve to the user name, user home directory resp. user
4286 shell. This is useful for running systemd user instances.
4288 * We now automatically rotate journal files if their data
4289 object hash table gets a fill level > 75%. We also size the
4290 hash table based on the configured maximum file size. This
4291 together should lower hash collisions drastically and thus
4292 speed things up a bit.
4294 * journalctl gained the new "--header" switch to introspect
4295 header data of journal files.
4297 * A new setting SystemCallFilters= has been added to services
4298 which may be used to apply blacklists or whitelists to
4299 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
4301 * nspawn gained a new --link-journal= switch (and quicker: -j)
4302 to link the container journal with the host. This makes it
4303 very easy to centralize log viewing on the host for all
4304 guests while still keeping the journal files separated.
4306 * Many bugfixes and optimizations
4308 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
4309 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
4310 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
4315 * Several tools now understand kernel command line arguments,
4316 which are only read when run in an initial RAM disk. They
4317 usually follow closely their normal counterparts, but are
4320 * There's a new tool to analyze the readahead files that are
4321 automatically generated at boot. Use:
4323 /usr/lib/systemd/systemd-readahead analyze /.readahead
4325 * We now provide an early debug shell on tty9 if this enabled. Use:
4327 systemctl enable debug-shell.service
4329 * All plymouth related units have been moved into the Plymouth
4330 package. Please make sure to upgrade your Plymouth version
4333 * systemd-tmpfiles now supports getting passed the basename of
4334 a configuration file only, in which case it will look for it
4335 in all appropriate directories automatically.
4337 * udevadm info now takes a /dev or /sys path as argument, and
4338 does the right thing. Example:
4340 udevadm info /dev/sda
4341 udevadm info /sys/class/block/sda
4343 * systemctl now prints a warning if a unit is stopped but a
4344 unit that might trigger it continues to run. Example: a
4345 service is stopped but the socket that activates it is left
4348 * "systemctl status" will now mention if the log output was
4349 shortened due to rotation since a service has been started.
4351 * The journal API now exposes functions to determine the
4352 "cutoff" times due to rotation.
4354 * journald now understands SIGUSR1 and SIGUSR2 for triggering
4355 immediately flushing of runtime logs to /var if possible,
4356 resp. for triggering immediate rotation of the journal
4359 * It is now considered an error if a service is attempted to
4360 be stopped that is not loaded.
4362 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
4364 * systemd-analyze now supports Python 3
4366 * tmpfiles now supports cleaning up directories via aging
4367 where the first level dirs are always kept around but
4368 directories beneath it automatically aged. This is enabled
4369 by prefixing the age field with '~'.
4371 * Seat objects now expose CanGraphical, CanTTY properties
4372 which is required to deal with very fast bootups where the
4373 display manager might be running before the graphics drivers
4374 completed initialization.
4376 * Seat objects now expose a State property.
4378 * We now include RPM macros for service enabling/disabling
4379 based on the preset logic. We recommend RPM based
4380 distributions to make use of these macros if possible. This
4381 makes it simpler to reuse RPM spec files across
4384 * We now make sure that the collected systemd unit name is
4385 always valid when services log to the journal via
4388 * There's a new man page kernel-command-line(7) detailing all
4389 command line options we understand.
4391 * The fstab generator may now be disabled at boot by passing
4392 fstab=0 on the kernel command line.
4394 * A new kernel command line option modules-load= is now understood
4395 to load a specific kernel module statically, early at boot.
4397 * Unit names specified on the systemctl command line are now
4398 automatically escaped as needed. Also, if file system or
4399 device paths are specified they are automatically turned
4400 into the appropriate mount or device unit names. Example:
4402 systemctl status /home
4403 systemctl status /dev/sda
4405 * The SysVConsole= configuration option has been removed from
4406 system.conf parsing.
4408 * The SysV search path is no longer exported on the D-Bus
4411 * The Names= option is been removed from unit file parsing.
4413 * There's a new man page bootup(7) detailing the boot process.
4415 * Every unit and every generator we ship with systemd now
4416 comes with full documentation. The self-explanatory boot is
4419 * A couple of services gained "systemd-" prefixes in their
4420 name if they wrap systemd code, rather than only external
4421 code. Among them fsck@.service which is now
4422 systemd-fsck@.service.
4424 * The HaveWatchdog property has been removed from the D-Bus
4427 * systemd.confirm_spawn= on the kernel command line should now
4430 * There's a new man page crypttab(5) which details all options
4431 we actually understand.
4433 * systemd-nspawn gained a new --capability= switch to pass
4434 additional capabilities to the container.
4436 * timedated will now read known NTP implementation unit names
4437 from /usr/lib/systemd/ntp-units.d/*.list,
4438 systemd-timedated-ntp.target has been removed.
4440 * journalctl gained a new switch "-b" that lists log data of
4441 the current boot only.
4443 * The notify socket is in the abstract namespace again, in
4444 order to support daemons which chroot() at start-up.
4446 * There is a new Storage= configuration option for journald
4447 which allows configuration of where log data should go. This
4448 also provides a way to disable journal logging entirely, so
4449 that data collected is only forwarded to the console, the
4450 kernel log buffer or another syslog implementation.
4452 * Many bugfixes and optimizations
4454 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
4455 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
4456 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
4457 Shawn Landden, Tom Gundersen
4461 * "systemctl help <unit>" now shows the man page if one is
4464 * Several new man pages have been added.
4466 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
4467 MaxLevelConsole= can now be specified in
4468 journald.conf. These options allow reducing the amount of
4469 data stored on disk or forwarded by the log level.
4471 * TimerSlackNSec= can now be specified in system.conf for
4472 PID1. This allows system-wide power savings.
4474 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
4475 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
4480 * logind is now capable of (optionally) handling power and
4481 sleep keys as well as the lid switch.
4483 * journalctl now understands the syntax "journalctl
4484 /usr/bin/avahi-daemon" to get all log output of a specific
4487 * CapabilityBoundingSet= in system.conf now also influences
4488 the capability bound set of usermode helpers of the kernel.
4490 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
4491 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
4492 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
4493 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
4497 * Note that we skipped 139 releases here in order to set the
4498 new version to something that is greater than both udev's
4499 and systemd's most recent version number.
4501 * udev: all udev sources are merged into the systemd source tree now.
4502 All future udev development will happen in the systemd tree. It
4503 is still fully supported to use the udev daemon and tools without
4504 systemd running, like in initramfs or other init systems. Building
4505 udev though, will require the *build* of the systemd tree, but
4506 udev can be properly *run* without systemd.
4508 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
4509 should be used to create dead device nodes as workarounds for broken
4512 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
4513 no longer supported. udev_monitor_new_from_netlink() needs to be
4514 used to subscribe to events.
4516 * udev: when udevd is started by systemd, processes which are left
4517 behind by forking them off of udev rules, are unconditionally cleaned
4518 up and killed now after the event handling has finished. Services or
4519 daemons must be started as systemd services. Services can be
4520 pulled-in by udev to get started, but they can no longer be directly
4521 forked by udev rules.
4523 * udev: the daemon binary is called systemd-udevd now and installed
4524 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
4525 to adapt to that, create symlink, or rename the binary after building
4528 * libudev no longer provides these symbols:
4529 udev_monitor_from_socket()
4530 udev_queue_get_failed_list_entry()
4531 udev_get_{dev,sys,run}_path()
4532 The versions number was bumped and symbol versioning introduced.
4534 * systemd-loginctl and systemd-journalctl have been renamed
4535 to loginctl and journalctl to match systemctl.
4537 * The config files: /etc/systemd/systemd-logind.conf and
4538 /etc/systemd/systemd-journald.conf have been renamed to
4539 logind.conf and journald.conf. Package updates should rename
4540 the files to the new names on upgrade.
4542 * For almost all files the license is now LGPL2.1+, changed
4543 from the previous GPL2.0+. Exceptions are some minor stuff
4544 of udev (which will be changed to LGPL2.1 eventually, too),
4545 and the MIT licensed sd-daemon.[ch] library that is suitable
4546 to be used as drop-in files.
4548 * systemd and logind now handle system sleep states, in
4549 particular suspending and hibernating.
4551 * logind now implements a sleep/shutdown/idle inhibiting logic
4552 suitable for a variety of uses. Soonishly Lennart will blog
4553 about this in more detail.
4555 * var-run.mount and var-lock.mount are no longer provided
4556 (which prevously bind mounted these directories to their new
4557 places). Distributions which have not converted these
4558 directories to symlinks should consider stealing these files
4559 from git history and add them downstream.
4561 * We introduced the Documentation= field for units and added
4562 this to all our shipped units. This is useful to make it
4563 easier to explore the boot and the purpose of the various
4566 * All smaller setup units (such as
4567 systemd-vconsole-setup.service) now detect properly if they
4568 are run in a container and are skipped when
4569 appropriate. This guarantees an entirely noise-free boot in
4570 Linux container environments such as systemd-nspawn.
4572 * A framework for implementing offline system updates is now
4573 integrated, for details see:
4574 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
4576 * A new service type Type=idle is available now which helps us
4577 avoiding ugly interleaving of getty output and boot status
4580 * There's now a system-wide CapabilityBoundingSet= option to
4581 globally reduce the set of capabilities for the
4582 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
4583 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
4584 even CAP_NET_ADMIN system-wide for secure systems.
4586 * There are now system-wide DefaultLimitXXX= options to
4587 globally change the defaults of the various resource limits
4588 for all units started by PID 1.
4590 * Harald Hoyer's systemd test suite has been integrated into
4591 systemd which allows easy testing of systemd builds in qemu
4592 and nspawn. (This is really awesome! Ask us for details!)
4594 * The fstab parser is now implemented as generator, not inside
4597 * systemctl will now warn you if .mount units generated from
4598 /etc/fstab are out of date due to changes in fstab that
4599 have not been read by systemd yet.
4601 * systemd is now suitable for usage in initrds. Dracut has
4602 already been updated to make use of this. With this in place
4603 initrds get a slight bit faster but primarily are much
4604 easier to introspect and debug since "systemctl status" in
4605 the host system can be used to introspect initrd services,
4606 and the journal from the initrd is kept around too.
4608 * systemd-delta has been added, a tool to explore differences
4609 between user/admin configuration and vendor defaults.
4611 * PrivateTmp= now affects both /tmp and /var/tmp.
4613 * Boot time status messages are now much prettier and feature
4614 proper english language. Booting up systemd has never been
4617 * Read-ahead pack files now include the inode number of all
4618 files to pre-cache. When the inode changes the pre-caching
4619 is not attempted. This should be nicer to deal with updated
4620 packages which might result in changes of read-ahead
4623 * We now temporaritly lower the kernel's read_ahead_kb variable
4624 when collecting read-ahead data to ensure the kernel's
4625 built-in read-ahead does not add noise to our measurements
4626 of necessary blocks to pre-cache.
4628 * There's now RequiresMountsFor= to add automatic dependencies
4629 for all mounts necessary for a specific file system path.
4631 * MountAuto= and SwapAuto= have been removed from
4632 system.conf. Mounting file systems at boot has to take place
4635 * nspawn now learned a new switch --uuid= to set the machine
4636 ID on the command line.
4638 * nspawn now learned the -b switch to automatically search
4641 * vt102 is now the default TERM for serial TTYs, upgraded from
4644 * systemd-logind now works on VT-less systems.
4646 * The build tree has been reorganized. The individual
4647 components now have directories of their own.
4649 * A new condition type ConditionPathIsReadWrite= is now available.
4651 * nspawn learned the new -C switch to create cgroups for the
4652 container in other hierarchies.
4654 * We now have support for hardware watchdogs, configurable in
4657 * The scheduled shutdown logic now has a public API.
4659 * We now mount /tmp as tmpfs by default, but this can be
4660 masked and /etc/fstab can override it.
4662 * Since udisks does not make use of /media anymore we are not
4663 mounting a tmpfs on it anymore.
4665 * journalctl gained a new --local switch to only interleave
4666 locally generated journal files.
4668 * We can now load the IMA policy at boot automatically.
4670 * The GTK tools have been split off into a systemd-ui.
4672 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
4673 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
4674 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
4675 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
4676 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
4677 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
4678 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
4679 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
4680 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
4685 * This is mostly a bugfix release
4687 * Support optional initialization of the machine ID from the
4688 KVM or container configured UUID.
4690 * Support immediate reboots with "systemctl reboot -ff"
4692 * Show /etc/os-release data in systemd-analyze output
4694 * Many bugfixes for the journal, including endianness fixes and
4695 ensuring that disk space enforcement works
4697 * sd-login.h is C++ comptaible again
4699 * Extend the /etc/os-release format on request of the Debian
4702 * We now refuse non-UTF8 strings used in various configuration
4703 and unit files. This is done to ensure we do not pass invalid
4704 data over D-Bus or expose it elsewhere.
4706 * Register Mimo USB Screens as suitable for automatic seat
4709 * Read SELinux client context from journal clients in a race
4712 * Reorder configuration file lookup order. /etc now always
4713 overrides /run in order to allow the administrator to always
4714 and unconditionally override vendor supplied or
4715 automatically generated data.
4717 * The various user visible bits of the journal now have man
4718 pages. We still lack man pages for the journal API calls
4721 * We now ship all man pages in HTML format again in the
4724 Contributions from: Dave Reisner, Dirk Eibach, Frederic
4725 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
4726 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
4731 * This is mostly a bugfix release
4733 * systems lacking /etc/os-release are no longer supported.
4735 * Various functionality updates to libsystemd-login.so
4737 * Track class of PAM logins to distuingish greeters from
4740 Contributions from: Kay Sievers, Lennart Poettering, Michael
4745 * This is an important bugfix release for v41.
4747 * Building man pages is now optional which should be useful
4748 for those building systemd from git but unwilling to install
4751 * Watchdog support for supervising services is now usable. In
4752 a future release support for hardware watchdogs
4753 (i.e. /dev/watchdog) will be added building on this.
4755 * Service start rate limiting is now configurable and can be
4756 turned off per service. When a start rate limit is hit a
4757 reboot can automatically be triggered.
4759 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
4761 Contributions from: Benjamin Franzke, Bill Nottingham,
4762 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
4763 Schmidt, Michał Górny, Piotr Drąg
4767 * The systemd binary is installed /usr/lib/systemd/systemd now;
4768 An existing /sbin/init symlink needs to be adapted with the
4771 * The code that loads kernel modules has been ported to invoke
4772 libkmod directly, instead of modprobe. This means we do not
4773 support systems with module-init-tools anymore.
4775 * Watchdog support is now already useful, but still not
4778 * A new kernel command line option systemd.setenv= is
4779 understood to set system wide environment variables
4780 dynamically at boot.
4782 * We now limit the set of capabilities of systemd-journald.
4784 * We now set SIGPIPE to ignore by default, since it only is
4785 useful in shell pipelines, and has little use in general
4786 code. This can be disabled with IgnoreSIPIPE=no in unit
4789 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
4790 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
4795 * This is mostly a bugfix release
4797 * We now expose the reason why a service failed in the
4798 "Result" D-Bus property.
4800 * Rudimentary service watchdog support (will be completed over
4801 the next few releases.)
4803 * When systemd forks off in order execute some service we will
4804 now immediately changes its argv[0] to reflect which process
4805 it will execute. This is useful to minimize the time window
4806 with a generic argv[0], which makes bootcharts more useful
4808 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
4809 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
4810 Mike Kazantsev, Ray Strode
4814 * This is mostly a test release, but incorporates many
4817 * New systemd-cgtop tool to show control groups by their
4820 * Linking against libacl for ACLs is optional again. If
4821 disabled, support tracking device access for active logins
4822 goes becomes unavailable, and so does access to the user
4823 journals by the respective users.
4825 * If a group "adm" exists, journal files are automatically
4826 owned by them, thus allow members of this group full access
4827 to the system journal as well as all user journals.
4829 * The journal now stores the SELinux context of the logging
4830 client for all entries.
4832 * Add C++ inclusion guards to all public headers
4834 * New output mode "cat" in the journal to print only text
4835 messages, without any meta data like date or time.
4837 * Include tiny X server wrapper as a temporary stop-gap to
4838 teach XOrg udev display enumeration. This is used by display
4839 managers such as gdm, and will go away as soon as XOrg
4840 learned native udev hotplugging for display devices.
4842 * Add new systemd-cat tool for executing arbitrary programs
4843 with STDERR/STDOUT connected to the journal. Can also act as
4844 BSD logger replacement, and does so by default.
4846 * Optionally store all locally generated coredumps in the
4847 journal along with meta data.
4849 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
4850 writing short strings to files (for usage for /sys), and for
4851 creating symlinks, character and block device nodes.
4853 * New unit file option ControlGroupPersistent= to make cgroups
4854 persistent, following the mechanisms outlined in
4855 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
4857 * Support multiple local RTCs in a sane way
4859 * No longer monopolize IO when replaying readahead data on
4860 rotating disks, since we might starve non-file-system IO to
4861 death, since fanotify() will not see accesses done by blkid,
4864 * Do not show kernel threads in systemd-cgls anymore, unless
4865 requested with new -k switch.
4867 Contributions from: Dan Horák, Kay Sievers, Lennart
4868 Poettering, Michal Schmidt
4872 * This is mostly a test release, but incorporates many
4875 * The git repository moved to:
4876 git://anongit.freedesktop.org/systemd/systemd
4877 ssh://git.freedesktop.org/git/systemd/systemd
4879 * First release with the journal
4880 http://0pointer.de/blog/projects/the-journal.html
4882 * The journal replaces both systemd-kmsg-syslogd and
4883 systemd-stdout-bridge.
4885 * New sd_pid_get_unit() API call in libsystemd-logind
4887 * Many systemadm clean-ups
4889 * Introduce remote-fs-pre.target which is ordered before all
4890 remote mounts and may be used to start services before all
4893 * Added Mageia support
4895 * Add bash completion for systemd-loginctl
4897 * Actively monitor PID file creation for daemons which exit in
4898 the parent process before having finished writing the PID
4899 file in the daemon process. Daemons which do this need to be
4900 fixed (i.e. PID file creation must have finished before the
4901 parent exits), but we now react a bit more gracefully to them.
4903 * Add colourful boot output, mimicking the well-known output
4904 of existing distributions.
4906 * New option PassCredentials= for socket units, for
4907 compatibility with a recent kernel ABI breakage.
4909 * /etc/rc.local is now hooked in via a generator binary, and
4910 thus will no longer act as synchronization point during
4913 * systemctl list-unit-files now supports --root=.
4915 * systemd-tmpfiles now understands two new commands: z, Z for
4916 relabelling files according to the SELinux database. This is
4917 useful to apply SELinux labels to specific files in /sys,
4920 * Output of SysV services is now forwarded to both the console
4921 and the journal by default, not only just the console.
4923 * New man pages for all APIs from libsystemd-login.
4925 * The build tree got reorganized and a the build system is a
4926 lot more modular allowing embedded setups to specifically
4927 select the components of systemd they are interested in.
4929 * Support for Linux systems lacking the kernel VT subsystem is
4932 * configure's --with-rootdir= got renamed to
4933 --with-rootprefix= to follow the naming used by udev and
4936 * Unless specified otherwise we will now install to /usr instead
4937 of /usr/local by default.
4939 * Processes with '@' in argv[0][0] are now excluded from the
4940 final shut-down killing spree, following the logic explained
4942 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
4944 * All processes remaining in a service cgroup when we enter
4945 the START or START_PRE states are now killed with
4946 SIGKILL. That means it is no longer possible to spawn
4947 background processes from ExecStart= lines (which was never
4948 supported anyway, and bad style).
4950 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
4951 reloading of units together.
4953 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
4954 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
4955 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
4956 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
4957 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek