1 systemd System and Service Manager
5 * The python-systemd code has been removed from the systemd repository.
6 A new repository has been created which accommodates the code from
7 now on, and we kindly ask distributions to create a separate package
8 for this: https://github.com/systemd/python-systemd
10 * The systemd daemon will now reload its main configuration
11 (/etc/systemd/system.conf) on daemon-reload.
13 * sd-dhcp now exposes vendor specific extensions via
14 sd_dhcp_lease_get_vendor_specific().
16 * systemd-networkd gained a number of new configuration options.
18 - A new boolean configuration option for TAP devices called
19 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
20 device, thus allowing to send and receive GSO packets.
22 - A new tunnel configuration option called 'CopyDSCP='.
23 If enabled, the DSCP field of ip6 tunnels is copied into the
26 - A set of boolean bridge configuration options were added.
27 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
28 and 'UnicastFlood=' are now parsed by networkd and applied to the
29 respective bridge link device via the respective IFLA_BRPORT_*
32 - A new string configuration option to override the hostname sent
33 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
34 is true, networkd will use the configured hostname instead of the
35 system hostname when sending DHCP requests.
37 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
38 networkd will configure the IPv6 flow-label of the tunnel device
41 - The 'macvtap' virtual network devices are now supported, similar to
42 the already supported 'macvlan' devices.
44 * systemd-resolved now implements RFC5452 to improve resilience against
45 cache poisoning. Additionally, source port randomization is enabled
46 by default to further protect against DNS spoofing attacks.
48 * nss-mymachines now supports translating UIDs and GIDs of running
49 containers with user-namespaces enabled. If a container 'foo'
50 translates a host uid 'UID' to the container uid 'TUID', then
51 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
52 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
53 mapped as 'vg-foo-TGID'.
55 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
56 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
57 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
58 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
59 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
60 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
61 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
62 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
63 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
64 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
70 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
71 There are no known issues with current sysfs, and udev does not need
72 or should be used to work around such bugs.
74 * udev does no longer enable USB HID power management. Several reports
75 indicate, that some devices cannot handle that setting.
77 * The udev accelerometer helper was removed. The functionality
78 is now fully included in iio-sensor-proxy. But this means,
79 older iio-sensor-proxy versions will no longer provide
80 accelerometer/orientation data with this systemd version.
81 Please upgrade iio-sensor-proxy to version 1.0.
83 * networkd gained a new configuration option IPv6PrivacyExtensions=
84 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
85 for Stateless Address") on selected networks.
87 * For the sake of fewer build-time dependencies and less code in the
88 main repository, the python bindings are about to be removed in the
89 next release. A new repository has been created which accommodates
90 the code from now on, and we kindly ask distributions to create a
91 separate package for this. The removal will take place in v223.
93 https://github.com/systemd/python-systemd
95 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
96 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
97 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
98 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
99 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
100 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
101 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
102 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
103 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
104 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
106 -- Berlin, 2015-07-07
110 * The sd-bus.h and sd-event.h APIs have now been declared
111 stable and have been added to the official interface of
112 libsystemd.so. sd-bus implements an alternative D-Bus client
113 library, that is relatively easy to use, very efficient and
114 supports both classic D-Bus as well as kdbus as transport
115 backend. sd-event is a generic event loop abstraction that
116 is built around Linux epoll, but adds features such as event
117 prioritization or efficient timer handling. Both APIs are good
118 choices for C programs looking for a bus and/or event loop
119 implementation that is minimal and does not have to be
120 portable to other kernels.
122 * kdbus support is no longer compile-time optional. It is now
123 always built-in. However, it can still be disabled at
124 runtime using the kdbus=0 kernel command line setting, and
125 that setting may be changed to default to off, by specifying
126 --disable-kdbus at build-time. Note though that the kernel
127 command line setting has no effect if the kdbus.ko kernel
128 module is not installed, in which case kdbus is (obviously)
129 also disabled. We encourage all downstream distributions to
130 begin testing kdbus by adding it to the kernel images in the
131 development distributions, and leaving kdbus support in
134 * The minimal required util-linux version has been bumped to
137 * Support for chkconfig (--enable-chkconfig) was removed in
138 favor of calling an abstraction tool
139 /lib/systemd/systemd-sysv-install. This needs to be
140 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
141 in README for details.
143 * If there's a systemd unit and a SysV init script for the
144 same service name, and the user executes "systemctl enable"
145 for it (or a related call), then this will now enable both
146 (or execute the related operation on both), not just the
149 * The libudev API documentation has been converted from gtkdoc
152 * gudev has been removed from the systemd tree, it is now an
155 * The systemd-cgtop tool learnt a new --raw switch to generate
156 "raw" (machine parsable) output.
158 * networkd's IPForwarding= .network file setting learnt the
159 new setting "kernel", which ensures that networkd does not
160 change the IP forwarding sysctl from the default kernel
163 * The systemd-logind bus API now exposes a new boolean
164 property "Docked" that reports whether logind considers the
165 system "docked", i.e. connected to a docking station or not.
167 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
168 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
169 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
170 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
171 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
172 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
173 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
174 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
175 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
176 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
177 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
178 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
179 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
180 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
181 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
182 Fink, Zbigniew Jędrzejewski-Szmek
184 -- Berlin, 2015-06-19
188 * The gudev library has been extracted into a separate repository
189 available at: https://git.gnome.org/browse/libgudev/
190 It is now managed as part of the Gnome project. Distributions
191 are recommended to pass --disable-gudev to systemd and use
192 gudev from the Gnome project instead. gudev is still included
193 in systemd, for now. It will be removed soon, though. Please
194 also see the announcement-thread on systemd-devel:
195 http://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
197 * systemd now exposes a CPUUsageNSec= property for each
198 service unit on the bus, that contains the overall consumed
199 CPU time of a service (the sum of what each process of the
200 service consumed). This value is only available if
201 CPUAccounting= is turned on for a service, and is then shown
202 in the "systemctl status" output.
204 * Support for configuring alternative mappings of the old SysV
205 runlevels to systemd targets has been removed. They are now
206 hardcoded in a way that runlevels 2, 3, 4 all map to
207 multi-user.target and 5 to graphical.target (which
208 previously was already the default behaviour).
210 * The auto-mounter logic gained support for mount point
211 expiry, using a new TimeoutIdleSec= setting in .automount
212 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
214 * The EFI System Partition (ESP) as mounted to /boot by
215 systemd-efi-boot-generator will now be unmounted
216 automatically after 2 minutes of not being used. This should
217 minimize the risk of ESP corruptions.
219 * New /etc/fstab options x-systemd.requires= and
220 x-systemd.requires-mounts-for= are now supported to express
221 additional dependencies for mounts. This is useful for
222 journalling file systems that support external journal
223 devices or overlay file systems that require underlying file
224 systems to be mounted.
226 * systemd does not support direct live-upgrades (via systemctl
227 daemon-reexec) from versions older than v44 anymore. As no
228 distribution we are aware of shipped such old versions in a
229 stable release this should not be problematic.
231 * When systemd forks off a new per-connection service instance
232 it will now set the $REMOTE_ADDR environment variable to the
233 remote IP address, and $REMOTE_PORT environment variable to
234 the remote IP port. This behaviour is similar to the
235 corresponding environment variables defined by CGI.
237 * systemd-networkd gained support for uplink failure
238 detection. The BindCarrier= option allows binding interface
239 configuration dynamically to the link sense of other
240 interfaces. This is useful to achieve behaviour like in
243 * systemd-networkd gained support for configuring the DHCP
244 client identifier to use when requesting leases.
246 * systemd-networkd now has a per-network UseNTP= option to
247 configure whether NTP server information acquired via DHCP
248 is passed on to services like systemd-timesyncd.
250 * systemd-networkd gained support for vti6 tunnels.
252 * Note that systemd-networkd manages the sysctl variable
253 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
254 it is configured for since v219. The variable controls IP
255 forwarding, and is a per-interface alternative to the global
256 /proc/sys/net/ipv[46]/ip_forward. This setting is
257 configurable in the IPForward= option, which defaults to
258 "no". This means if networkd is used for an interface it is
259 no longer sufficient to set the global sysctl option to turn
260 on IP forwarding! Instead, the .network file option
261 IPForward= needs to be turned on! Note that the
262 implementation of this behaviour was broken in v219 and has
265 * Many bonding and vxlan options are now configurable in
268 * systemd-nspawn gained a new --property= setting to set unit
269 properties for the container scope. This is useful for
270 setting resource parameters (e.g "CPUShares=500") on
271 containers started from the command line.
273 * systemd-nspawn gained a new --private-users= switch to make
274 use of user namespacing available on recent Linux kernels.
276 * systemd-nspawn may now be called as part of a shell pipeline
277 in which case the pipes used for stdin and stdout are passed
278 directly to the process invoked in the container, without
279 indirection via a pseudo tty.
281 * systemd-nspawn gained a new switch to control the UNIX
282 signal to use when killing the init process of the container
285 * systemd-nspawn gained a new --overlay= switch for mounting
286 overlay file systems into the container using the new kernel
289 * When a container image is imported via systemd-importd and
290 the host file system is not btrfs, a loopback block device
291 file is created in /var/lib/machines.raw with a btrfs file
292 system inside. It is then mounted to /var/lib/machines to
293 enable btrfs features for container management. The loopback
294 file and btrfs file system is grown as needed when container
295 images are imported via systemd-importd.
297 * systemd-machined/systemd-importd gained support for btrfs
298 quota, to enforce container disk space limits on disk. This
299 is exposed in "machinectl set-limit".
301 * systemd-importd now can import containers from local .tar,
302 .raw and .qcow2 images, and export them to .tar and .raw. It
303 can also import dkr v2 images now from the network (on top
306 * systemd-importd gained support for verifying downloaded
307 images with gpg2 (previously only gpg1 was supported).
309 * systemd-machined, systemd-logind, systemd: most bus calls
310 are now accessible to unprivileged processes via
311 PolicyKit. Also, systemd-logind will now allow users to kill
312 their own sessions without further privileges or
315 * systemd-shutdownd has been removed. This service was
316 previously responsible for implementing scheduled shutdowns
317 as exposed in /usr/bin/shutdown's time parameter. This
318 functionality has now been moved into systemd-logind and is
319 accessible via a bus interface.
321 * "systemctl reboot" gained a new switch --firmware-setup that
322 can be used to reboot into the EFI firmware setup, if that
323 is available. systemd-logind now exposes an API on the bus
324 to trigger such reboots, in case graphical desktop UIs want
325 to cover this functionality.
327 * "systemctl enable", "systemctl disable" and "systemctl mask"
328 now support a new "--now" switch. If specified the units
329 that are enabled will also be started, and the ones
330 disabled/masked also stopped.
332 * The Gummiboot EFI boot loader tool has been merged into
333 systemd, and renamed to "systemd-boot". The bootctl tool has been
334 updated to support systemd-boot.
336 * An EFI kernel stub has been added that may be used to create
337 kernel EFI binaries that contain not only the actual kernel,
338 but also an initrd, boot splash, command line and OS release
339 information. This combined binary can then be signed as a
340 single image, so that the firmware can verify it all in one
341 step. systemd-boot has special support for EFI binaries created
342 like this and can extract OS release information from them
343 and show them in the boot menu. This functionality is useful
344 to implement cryptographically verified boot schemes.
346 * Optional support has been added to systemd-fsck to pass
347 fsck's progress report to an AF_UNIX socket in the file
350 * udev will no longer create device symlinks for all block
351 devices by default. A blacklist for excluding special block
352 devices from this logic has been turned into a whitelist
353 that requires picking block devices explicitly that require
356 * A new (currently still internal) API sd-device.h has been
357 added to libsystemd. This modernized API is supposed to
358 replace libudev eventually. In fact, already much of libudev
359 is now just a wrapper around sd-device.h.
361 * A new hwdb database for storing metadata about pointing
362 stick devices has been added.
364 * systemd-tmpfiles gained support for setting file attributes
365 similar to the "chattr" tool with new 'h' and 'H' lines.
367 * systemd-journald will no longer unconditionally set the
368 btrfs NOCOW flag on new journal files. This is instead done
369 with tmpfiles snippet using the new 'h' line type. This
370 allows easy disabling of this logic, by masking the
371 journal-nocow.conf tmpfiles file.
373 * systemd-journald will now translate audit message types to
374 human readable identifiers when writing them to the
375 journal. This should improve readability of audit messages.
377 * The LUKS logic gained support for the offset= and skip=
378 options in /etc/crypttab, as previously implemented by
381 * /usr/lib/os-release gained a new optional field VARIANT= for
382 distributions that support multiple variants (such as a
383 desktop edition, a server edition, ...)
385 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
386 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
387 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
388 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
389 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
390 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
391 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
392 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
393 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
394 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
395 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
396 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
397 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
398 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
399 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
400 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
401 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
402 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
403 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
404 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
405 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
406 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
407 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
408 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
409 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
410 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
411 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
413 -- Berlin, 2015-05-22
417 * Introduce a new API "sd-hwdb.h" for querying the hardware
418 metadata database. With this minimal interface one can query
419 and enumerate the udev hwdb, decoupled from the old libudev
420 library. libudev's interface for this is now only a wrapper
421 around sd-hwdb. A new tool systemd-hwdb has been added to
422 interface with and update the database.
424 * When any of systemd's tools copies files (for example due to
425 tmpfiles' C lines) a btrfs reflink will attempted first,
426 before bytewise copying is done.
428 * systemd-nspawn gained a new --ephemeral switch. When
429 specified a btrfs snapshot is taken of the container's root
430 directory, and immediately removed when the container
431 terminates again. Thus, a container can be started whose
432 changes never alter the container's root directory, and are
433 lost on container termination. This switch can also be used
434 for starting a container off the root file system of the
435 host without affecting the host OS. This switch is only
436 available on btrfs file systems.
438 * systemd-nspawn gained a new --template= switch. It takes the
439 path to a container tree to use as template for the tree
440 specified via --directory=, should that directory be
441 missing. This allows instantiating containers dynamically,
442 on first run. This switch is only available on btrfs file
445 * When a .mount unit refers to a mount point on which multiple
446 mounts are stacked, and the .mount unit is stopped all of
447 the stacked mount points will now be unmounted until no
450 * systemd now has an explicit notion of supported and
451 unsupported unit types. Jobs enqueued for unsupported unit
452 types will now fail with an "unsupported" error code. More
453 specifically .swap, .automount and .device units are not
454 supported in containers, .busname units are not supported on
455 non-kdbus systems. .swap and .automount are also not
456 supported if their respective kernel compile time options
459 * machinectl gained support for two new "copy-from" and
460 "copy-to" commands for copying files from a running
461 container to the host or vice versa.
463 * machinectl gained support for a new "bind" command to bind
464 mount host directories into local containers. This is
465 currently only supported for nspawn containers.
467 * networkd gained support for configuring bridge forwarding
468 database entries (fdb) from .network files.
470 * A new tiny daemon "systemd-importd" has been added that can
471 download container images in tar, raw, qcow2 or dkr formats,
472 and make them available locally in /var/lib/machines, so
473 that they can run as nspawn containers. The daemon can GPG
474 verify the downloads (not supported for dkr, since it has no
475 provisions for verifying downloads). It will transparently
476 decompress bz2, xz, gzip compressed downloads if necessary,
477 and restore sparse files on disk. The daemon uses privilege
478 separation to ensure the actual download logic runs with
479 fewer privileges than the daemon itself. machinectl has
480 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
481 make the functionality of importd available to the
482 user. With this in place the Fedora and Ubuntu "Cloud"
483 images can be downloaded and booted as containers unmodified
484 (the Fedora images lack the appropriate GPG signature files
485 currently, so they cannot be verified, but this will change
486 soon, hopefully). Note that downloading images is currently
487 only fully supported on btrfs.
489 * machinectl is now able to list container images found in
490 /var/lib/machines, along with some metadata about sizes of
491 disk and similar. If the directory is located on btrfs and
492 quota is enabled, this includes quota display. A new command
493 "image-status" has been added that shows additional
494 information about images.
496 * machinectl is now able to clone container images
497 efficiently, if the underlying file system (btrfs) supports
498 it, with the new "machinectl list-images" command. It also
499 gained commands for renaming and removing images, as well as
500 marking them read-only or read-write (supported also on
501 legacy file systems).
503 * networkd gained support for collecting LLDP network
504 announcements, from hardware that supports this. This is
505 shown in networkctl output.
507 * systemd-run gained support for a new -t (--pty) switch for
508 invoking a binary on a pty whose input and output is
509 connected to the invoking terminal. This allows executing
510 processes as system services while interactively
511 communicating with them via the terminal. Most interestingly
512 this is supported across container boundaries. Invoking
513 "systemd-run -t /bin/bash" is an alternative to running a
514 full login session, the difference being that the former
515 will not register a session, nor go through the PAM session
518 * tmpfiles gained support for a new "v" line type for creating
519 btrfs subvolumes. If the underlying file system is a legacy
520 file system, this automatically degrades to creating a
521 normal directory. Among others /var/lib/machines is now
522 created like this at boot, should it be missing.
524 * The directory /var/lib/containers/ has been deprecated and
525 been replaced by /var/lib/machines. The term "machines" has
526 been used in the systemd context as generic term for both
527 VMs and containers, and hence appears more appropriate for
528 this, as the directory can also contain raw images bootable
531 * systemd-nspawn when invoked with -M but without --directory=
532 or --image= is now capable of searching for the container
533 root directory, subvolume or disk image automatically, in
534 /var/lib/machines. systemd-nspawn@.service has been updated
535 to make use of this, thus allowing it to be used for raw
538 * A new machines.target unit has been introduced that is
539 supposed to group all containers/VMs invoked as services on
540 the system. systemd-nspawn@.service has been updated to
543 * machinectl gained a new "start" command, for invoking a
544 container as a service. "machinectl start foo" is mostly
545 equivalent to "systemctl start systemd-nspawn@foo.service",
546 but handles escaping in a nicer way.
548 * systemd-nspawn will now mount most of the cgroupfs tree
549 read-only into each container, with the exception of the
550 container's own subtree in the name=systemd hierarchy.
552 * journald now sets the special FS_NOCOW file flag for its
553 journal files. This should improve performance on btrfs, by
554 avoiding heavy fragmentation when journald's write-pattern
555 is used on COW file systems. It degrades btrfs' data
556 integrity guarantees for the files to the same levels as for
557 ext3/ext4 however. This should be OK though as journald does
558 its own data integrity checks and all its objects are
559 checksummed on disk. Also, journald should handle btrfs disk
560 full events a lot more gracefully now, by processing SIGBUS
561 errors, and not relying on fallocate() anymore.
563 * When journald detects that journal files it is writing to
564 have been deleted it will immediately start new journal
567 * systemd now provides a way to store file descriptors
568 per-service in PID 1.This is useful for daemons to ensure
569 that fds they require are not lost during a daemon
570 restart. The fds are passed to the daemon on the next
571 invocation in the same way socket activation fds are
572 passed. This is now used by journald to ensure that the
573 various sockets connected to all the system's stdout/stderr
574 are not lost when journald is restarted. File descriptors
575 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
576 an extension to sd_notify(). Note that a limit is enforced
577 on the number of fds a service can store in PID 1, and it
578 defaults to 0, so that no fds may be stored, unless this is
579 explicitly turned on.
581 * The default TERM variable to use for units connected to a
582 terminal, when no other value is explicitly is set is now
583 vt220 rather than vt102. This should be fairly safe still,
584 but allows PgUp/PgDn work.
586 * The /etc/crypttab option header= as known from Debian is now
589 * "loginctl user-status" and "loginctl session-status" will
590 now show the last 10 lines of log messages of the
591 user/session following the status output. Similar,
592 "machinectl status" will show the last 10 log lines
593 associated with a virtual machine or container
594 service. (Note that this is usually not the log messages
595 done in the VM/container itself, but simply what the
596 container manager logs. For nspawn this includes all console
599 * "loginctl session-status" without further argument will now
600 show the status of the session of the caller. Similar,
601 "lock-session", "unlock-session", "activate",
602 "enable-linger", "disable-linger" may now be called without
603 session/user parameter in which case they apply to the
604 caller's session/user.
606 * An X11 session scriptlet is now shipped that uploads
607 $DISPLAY and $XAUTHORITY into the environment of the systemd
608 --user daemon if a session begins. This should improve
609 compatibility with X11 enabled applications run as systemd
612 * Generators are now subject to masking via /etc and /run, the
613 same way as unit files.
615 * networkd .network files gained support for configuring
616 per-link IPv4/IPv6 packet forwarding as well as IPv4
617 masquerading. This is by default turned on for veth links to
618 containers, as registered by systemd-nspawn. This means that
619 nspawn containers run with --network-veth will now get
620 automatic routed access to the host's networks without any
621 further configuration or setup, as long as networkd runs on
624 * systemd-nspawn gained the --port= (-p) switch to expose TCP
625 or UDP posts of a container on the host. With this in place
626 it is possible to run containers with private veth links
627 (--network-veth), and have their functionality exposed on
628 the host as if their services were running directly on the
631 * systemd-nspawn's --network-veth switch now gained a short
632 version "-n", since with the changes above it is now truly
633 useful out-of-the-box. The systemd-nspawn@.service has been
634 updated to make use of it too by default.
636 * systemd-nspawn will now maintain a per-image R/W lock, to
637 ensure that the same image is not started more than once
638 writable. (It's OK to run an image multiple times
639 simultaneously in read-only mode.)
641 * systemd-nspawn's --image= option is now capable of
642 dissecting and booting MBR and GPT disk images that contain
643 only a single active Linux partition. Previously it
644 supported only GPT disk images with proper GPT type
645 IDs. This allows running cloud images from major
646 distributions directly with systemd-nspawn, without
649 * In addition to collecting mouse dpi data in the udev
650 hardware database, there's now support for collecting angle
651 information for mouse scroll wheels. The database is
652 supposed to guarantee similar scrolling behavior on mice
653 that it knows about. There's also support for collecting
654 information about Touchpad types.
656 * udev's input_id built-in will now also collect touch screen
657 dimension data and attach it to probed devices.
659 * /etc/os-release gained support for a Distribution Privacy
662 * networkd gained support for creating "ipvlan", "gretap",
663 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
665 * systemd-tmpfiles gained support for "a" lines for setting
668 * systemd-nspawn will now mount /tmp in the container to
669 tmpfs, automatically.
671 * systemd now exposes the memory.usage_in_bytes cgroup
672 attribute and shows it for each service in the "systemctl
673 status" output, if available.
675 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
676 immediate reboot is triggered. This useful if shutdown is
677 hung and is unable to complete, to expedite the
678 operation. Note that this kind of reboot will still unmount
679 all file systems, and hence should not result in fsck being
682 * A .device unit for an optical block device will now be
683 considered active only when a medium is in the drive. Also,
684 mount units are now bound to their backing devices thus
685 triggering automatic unmounting when devices become
686 unavailable. With this in place systemd will now
687 automatically unmount left-over mounts when a CD-ROM is
688 ejected or an USB stick is yanked from the system.
690 * networkd-wait-online now has support for waiting for
691 specific interfaces only (with globbing), and for giving up
692 after a configurable timeout.
694 * networkd now exits when idle. It will be automatically
695 restarted as soon as interfaces show up, are removed or
696 change state. networkd will stay around as long as there is
697 at least one DHCP state machine or similar around, that keep
700 * networkd may now configure IPv6 link-local addressing in
701 addition to IPv4 link-local addressing.
703 * The IPv6 "token" for use in SLAAC may now be configured for
704 each .network interface in networkd.
706 * Routes configured with networkd may now be assigned a scope
709 * networkd's [Match] sections now support globbing and lists
710 of multiple space-separated matches per item.
712 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
713 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
714 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
715 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
716 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
717 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
718 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
719 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
720 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
721 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
722 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
723 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
724 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
725 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
726 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
727 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
728 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
729 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
730 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
731 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
732 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
733 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
734 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
735 Hoffmann, Zbigniew Jędrzejewski-Szmek
737 -- Berlin, 2015-02-16
741 * When querying unit file enablement status (for example via
742 "systemctl is-enabled"), a new state "indirect" is now known
743 which indicates that a unit might not be enabled itself, but
744 another unit listed in its Alias= setting might be.
746 * Similar to the various existing ConditionXYZ= settings for
747 units there are now matching AssertXYZ= settings. While
748 failing conditions cause a unit to be skipped, but its job
749 to succeed, failing assertions declared like this will cause
750 a unit start operation and its job to fail.
752 * hostnamed now knows a new chassis type "embedded".
754 * systemctl gained a new "edit" command. When used on a unit
755 file this allows extending unit files with .d/ drop-in
756 configuration snippets or editing the full file (after
757 copying it from /usr/lib to /etc). This will invoke the
758 user's editor (as configured with $EDITOR), and reload the
759 modified configuration after editing.
761 * "systemctl status" now shows the suggested enablement state
762 for a unit, as declared in the (usually vendor-supplied)
765 * nss-myhostname will now resolve the single-label host name
766 "gateway" to the locally configured default IP routing
767 gateways, ordered by their metrics. This assigns a stable
768 name to the used gateways, regardless which ones are
769 currently configured. Note that the name will only be
770 resolved after all other name sources (if nss-myhostname is
771 configured properly) and should hence not negatively impact
772 systems that use the single-label host name "gateway" in
775 * systemd-inhibit now allows filtering by mode when listing
778 * Scope and service units gained a new "Delegate" boolean
779 property, which when set allows processes running inside the
780 unit to further partition resources. This is primarily
781 useful for systemd user instances as well as container
784 * journald will now pick up audit messages directly from
785 the kernel, and log them like any other log message. The
786 audit fields are split up and fully indexed. This means that
787 journalctl in many ways is now a (nicer!) alternative to
788 ausearch, the traditional audit client. Note that this
789 implements only a minimal audit client, if you want the
790 special audit modes like reboot-on-log-overflow, please use
791 the traditional auditd instead, which can be used in
792 parallel to journald.
794 * The ConditionSecurity= unit file option now understands the
795 special string "audit" to check whether auditing is
798 * journalctl gained two new commands --vacuum-size= and
799 --vacuum-time= to delete old journal files until the
800 remaining ones take up no more the specified size on disk,
801 or are not older than the specified time.
803 * A new, native PPPoE library has been added to sd-network,
804 systemd's library of light-weight networking protocols. This
805 library will be used in a future version of networkd to
806 enable PPPoE communication without an external pppd daemon.
808 * The busctl tool now understands a new "capture" verb that
809 works similar to "monitor", but writes a packet capture
810 trace to STDOUT that can be redirected to a file which is
811 compatible with libcap's capture file format. This can then
812 be loaded in Wireshark and similar tools to inspect bus
815 * The busctl tool now understands a new "tree" verb that shows
816 the object trees of a specific service on the bus, or of all
819 * The busctl tool now understands a new "introspect" verb that
820 shows all interfaces and members of objects on the bus,
821 including their signature and values. This is particularly
822 useful to get more information about bus objects shown by
823 the new "busctl tree" command.
825 * The busctl tool now understands new verbs "call",
826 "set-property" and "get-property" for invoking bus method
827 calls, setting and getting bus object properties in a
830 * busctl gained a new --augment-creds= argument that controls
831 whether the tool shall augment credential information it
832 gets from the bus with data from /proc, in a possibly
835 * nspawn's --link-journal= switch gained two new values
836 "try-guest" and "try-host" that work like "guest" and
837 "host", but do not fail if the host has no persistent
838 journalling enabled. -j is now equivalent to
839 --link-journal=try-guest.
841 * macvlan network devices created by nspawn will now have
842 stable MAC addresses.
844 * A new SmackProcessLabel= unit setting has been added, which
845 controls the SMACK security label processes forked off by
846 the respective unit shall use.
848 * If compiled with --enable-xkbcommon, systemd-localed will
849 verify x11 keymap settings by compiling the given keymap. It
850 will spew out warnings if the compilation fails. This
851 requires libxkbcommon to be installed.
853 * When a coredump is collected a larger number of metadata
854 fields is now collected and included in the journal records
855 created for it. More specifically control group membership,
856 environment variables, memory maps, working directory,
857 chroot directory, /proc/$PID/status, and a list of open file
858 descriptors is now stored in the log entry.
860 * The udev hwdb now contains DPI information for mice. For
863 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
865 * All systemd programs that read standalone configuration
866 files in /etc now also support a corresponding series of
867 .conf.d configuration directories in /etc/, /run/,
868 /usr/local/lib/, /usr/lib/, and (if configured with
869 --enable-split-usr) /lib/. In particular, the following
870 configuration files now have corresponding configuration
871 directories: system.conf user.conf, logind.conf,
872 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
873 resolved.conf, timesyncd.conf, journal-remote.conf, and
874 journal-upload.conf. Note that distributions should use the
875 configuration directories in /usr/lib/; the directories in
876 /etc/ are reserved for the system administrator.
878 * systemd-rfkill will no longer take the rfkill device name
879 into account when storing rfkill state on disk, as the name
880 might be dynamically assigned and not stable. Instead, the
881 ID_PATH udev variable combined with the rfkill type (wlan,
882 bluetooth, ...) is used.
884 * A new service systemd-machine-id-commit.service has been
885 added. When used on systems where /etc is read-only during
886 boot, and /etc/machine-id is not initialized (but an empty
887 file), this service will copy the temporary machine ID
888 created as replacement into /etc after the system is fully
889 booted up. This is useful for systems that are freshly
890 installed with a non-initialized machine ID, but should get
891 a fixed machine ID for subsequent boots.
893 * networkd's .netdev files now provide a large set of
894 configuration parameters for VXLAN devices. Similar, the
895 bridge port cost parameter is now configurable in .network
896 files. There's also new support for configuring IP source
897 routing. networkd .link files gained support for a new
898 OriginalName= match that is useful to match against the
899 original interface name the kernel assigned. .network files
900 may include MTU= and MACAddress= fields for altering the MTU
901 and MAC address while being connected to a specific network
904 * The LUKS logic gained supported for configuring
905 UUID-specific key files. There's also new support for naming
906 LUKS device from the kernel command line, using the new
909 * Timer units may now be transiently created via the bus API
910 (this was previously already available for scope and service
911 units). In addition it is now possible to create multiple
912 transient units at the same time with a single bus call. The
913 "systemd-run" tool has been updated to make use of this for
914 running commands on a specified time, in at(1)-style.
916 * tmpfiles gained support for "t" lines, for assigning
917 extended attributes to files. Among other uses this may be
918 used to assign SMACK labels to files.
920 Contributions from: Alin Rauta, Alison Chaiken, Andrej
921 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
922 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
923 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
924 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
925 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
926 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
927 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
928 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
929 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
930 Michael Biebl, Michael Chapman, Michael Marineau, Michal
931 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
932 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
933 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
934 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
935 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
936 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
937 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
939 -- Berlin, 2014-12-10
943 * journalctl gained the new options -t/--identifier= to match
944 on the syslog identifier (aka "tag"), as well as --utc to
945 show log timestamps in the UTC timezone. journalctl now also
946 accepts -n/--lines=all to disable line capping in a pager.
948 * journalctl gained a new switch, --flush, that synchronously
949 flushes logs from /run/log/journal to /var/log/journal if
950 persistent storage is enabled. systemd-journal-flush.service
951 now waits until the operation is complete.
953 * Services can notify the manager before they start a reload
954 (by sending RELOADING=1) or shutdown (by sending
955 STOPPING=1). This allows the manager to track and show the
956 internal state of daemons and closes a race condition when
957 the process is still running but has closed its D-Bus
960 * Services with Type=oneshot do not have to have any ExecStart
963 * User units are now loaded also from
964 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
965 /run/systemd/user directory that was already previously
966 supported, but is under the control of the user.
968 * Job timeouts (i.e. time-outs on the time a job that is
969 queued stays in the run queue) can now optionally result in
970 immediate reboot or power-off actions (JobTimeoutAction= and
971 JobTimeoutRebootArgument=). This is useful on ".target"
972 units, to limit the maximum time a target remains
973 undispatched in the run queue, and to trigger an emergency
974 operation in such a case. This is now used by default to
975 turn off the system if boot-up (as defined by everything in
976 basic.target) hangs and does not complete for at least
977 15min. Also, if power-off or reboot hang for at least 30min
978 an immediate power-off/reboot operation is triggered. This
979 functionality is particularly useful to increase reliability
980 on embedded devices, but also on laptops which might
981 accidentally get powered on when carried in a backpack and
982 whose boot stays stuck in a hard disk encryption passphrase
985 * systemd-logind can be configured to also handle lid switch
986 events even when the machine is docked or multiple displays
987 are attached (HandleLidSwitchDocked= option).
989 * A helper binary and a service have been added which can be
990 used to resume from hibernation in the initramfs. A
991 generator will parse the resume= option on the kernel
992 command line to trigger resume.
994 * A user console daemon systemd-consoled has been
995 added. Currently, it is a preview, and will so far open a
996 single terminal on each session of the user marked as
997 Desktop=systemd-console.
999 * Route metrics can be specified for DHCP routes added by
1002 * The SELinux context of socket-activated services can be set
1003 from the information provided by the networking stack
1004 (SELinuxContextFromNet= option).
1006 * Userspace firmware loading support has been removed and
1007 the minimum supported kernel version is thus bumped to 3.7.
1009 * Timeout for udev workers has been increased from 1 to 3
1010 minutes, but a warning will be printed after 1 minute to
1011 help diagnose kernel modules that take a long time to load.
1013 * Udev rules can now remove tags on devices with TAG-="foobar".
1015 * systemd's readahead implementation has been removed. In many
1016 circumstances it didn't give expected benefits even for
1017 rotational disk drives and was becoming less relevant in the
1018 age of SSDs. As none of the developers has been using
1019 rotating media anymore, and nobody stepped up to actively
1020 maintain this component of systemd it has now been removed.
1022 * Swap units can use Options= to specify discard options.
1023 Discard options specified for swaps in /etc/fstab are now
1026 * Docker containers are now detected as a separate type of
1029 * The Password Agent protocol gained support for queries where
1030 the user input is shown, useful e.g. for user names.
1031 systemd-ask-password gained a new --echo option to turn that
1034 * The default sysctl.d/ snippets will now set:
1036 net.core.default_qdisc = fq_codel
1038 This selects Fair Queuing Controlled Delay as the default
1039 queuing discipline for network interfaces. fq_codel helps
1040 fight the network bufferbloat problem. It is believed to be
1041 a good default with no tuning required for most workloads.
1042 Downstream distributions may override this choice. On 10Gbit
1043 servers that do not do forwarding, "fq" may perform better.
1044 Systems without a good clocksource should use "pfifo_fast".
1046 * If kdbus is enabled during build a new option BusPolicy= is
1047 available for service units, that allows locking all service
1048 processes into a stricter bus policy, in order to limit
1049 access to various bus services, or even hide most of them
1050 from the service's view entirely.
1052 * networkctl will now show the .network and .link file
1053 networkd has applied to a specific interface.
1055 * sd-login gained a new API call sd_session_get_desktop() to
1056 query which desktop environment has been selected for a
1059 * UNIX utmp support is now compile-time optional to support
1060 legacy-free systems.
1062 * systemctl gained two new commands "add-wants" and
1063 "add-requires" for pulling in units from specific targets
1066 * If the word "rescue" is specified on the kernel command line
1067 the system will now boot into rescue mode (aka
1068 rescue.target), which was previously available only by
1069 specifying "1" or "systemd.unit=rescue.target" on the kernel
1070 command line. This new kernel command line option nicely
1071 mirrors the already existing "emergency" kernel command line
1074 * New kernel command line options mount.usr=, mount.usrflags=,
1075 mount.usrfstype= have been added that match root=, rootflags=,
1076 rootfstype= but allow mounting a specific file system to
1079 * The $NOTIFY_SOCKET is now also passed to control processes of
1080 services, not only the main process.
1082 * This version reenables support for fsck's -l switch. This
1083 means at least version v2.25 of util-linux is required for
1084 operation, otherwise dead-locks on device nodes may
1085 occur. Again: you need to update util-linux to at least
1086 v2.25 when updating systemd to v217.
1088 * The "multi-seat-x" tool has been removed from systemd, as
1089 its functionality has been integrated into X servers 1.16,
1090 and the tool is hence redundant. It is recommended to update
1091 display managers invoking this tool to simply invoke X
1092 directly from now on, again.
1094 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
1095 message flag has been added for all of systemd's PolicyKit
1096 authenticated method calls has been added. In particular
1097 this now allows optional interactive authorization via
1098 PolicyKit for many of PID1's privileged operations such as
1099 unit file enabling and disabling.
1101 * "udevadm hwdb --update" learnt a new switch "--usr" for
1102 placing the rebuilt hardware database in /usr instead of
1103 /etc. When used only hardware database entries stored in
1104 /usr will be used, and any user database entries in /etc are
1105 ignored. This functionality is useful for vendors to ship a
1106 pre-built database on systems where local configuration is
1107 unnecessary or unlikely.
1109 * Calendar time specifications in .timer units now also
1110 understand the strings "semi-annually", "quarterly" and
1111 "minutely" as shortcuts (in addition to the preexisting
1112 "anually", "hourly", ...).
1114 * systemd-tmpfiles will now correctly create files in /dev
1115 at boot which are marked for creation only at boot. It is
1116 recommended to always create static device nodes with 'c!'
1117 and 'b!', so that they are created only at boot and not
1118 overwritten at runtime.
1120 * When the watchdog logic is used for a service (WatchdogSec=)
1121 and the watchdog timeout is hit the service will now be
1122 terminated with SIGABRT (instead of just SIGTERM), in order
1123 to make sure a proper coredump and backtrace is
1124 generated. This ensures that hanging services will result in
1125 similar coredump/backtrace behaviour as services that hit a
1128 Contributions from: Andreas Henriksson, Andrei Borzenkov,
1129 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
1130 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
1131 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
1132 Herrmann, David Sommerseth, David Strauss, Emil Renner
1133 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
1134 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
1135 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
1136 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
1137 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
1138 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
1139 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
1140 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
1141 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
1142 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
1143 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
1144 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
1145 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
1146 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
1147 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
1148 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
1151 -- Berlin, 2014-10-28
1155 * timedated no longer reads NTP implementation unit names from
1156 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
1157 implementations should add a
1159 Conflicts=systemd-timesyncd.service
1161 to their unit files to take over and replace systemd's NTP
1162 default functionality.
1164 * systemd-sysusers gained a new line type "r" for configuring
1165 which UID/GID ranges to allocate system users/groups
1166 from. Lines of type "u" may now add an additional column
1167 that specifies the home directory for the system user to be
1168 created. Also, systemd-sysusers may now optionally read user
1169 information from STDIN instead of a file. This is useful for
1170 invoking it from RPM preinst scriptlets that need to create
1171 users before the first RPM file is installed since these
1172 files might need to be owned by them. A new
1173 %sysusers_create_inline RPM macro has been introduced to do
1174 just that. systemd-sysusers now updates the shadow files as
1175 well as the user/group databases, which should enhance
1176 compatibility with certain tools like grpck.
1178 * A number of bus APIs of PID 1 now optionally consult
1179 PolicyKit to permit access for otherwise unprivileged
1180 clients under certain conditions. Note that this currently
1181 doesn't support interactive authentication yet, but this is
1182 expected to be added eventually, too.
1184 * /etc/machine-info now has new fields for configuring the
1185 deployment environment of the machine, as well as the
1186 location of the machine. hostnamectl has been updated with
1187 new command to update these fields.
1189 * systemd-timesyncd has been updated to automatically acquire
1190 NTP server information from systemd-networkd, which might
1191 have been discovered via DHCP.
1193 * systemd-resolved now includes a caching DNS stub resolver
1194 and a complete LLMNR name resolution implementation. A new
1195 NSS module "nss-resolve" has been added which can be used
1196 instead of glibc's own "nss-dns" to resolve hostnames via
1197 systemd-resolved. Hostnames, addresses and arbitrary RRs may
1198 be resolved via systemd-resolved D-Bus APIs. In contrast to
1199 the glibc internal resolver systemd-resolved is aware of
1200 multi-homed system, and keeps DNS server and caches separate
1201 and per-interface. Queries are sent simultaneously on all
1202 interfaces that have DNS servers configured, in order to
1203 properly handle VPNs and local LANs which might resolve
1204 separate sets of domain names. systemd-resolved may acquire
1205 DNS server information from systemd-networkd automatically,
1206 which in turn might have discovered them via DHCP. A tool
1207 "systemd-resolve-host" has been added that may be used to
1208 query the DNS logic in resolved. systemd-resolved implements
1209 IDNA and automatically uses IDNA or UTF-8 encoding depending
1210 on whether classic DNS or LLMNR is used as transport. In the
1211 next releases we intend to add a DNSSEC and mDNS/DNS-SD
1212 implementation to systemd-resolved.
1214 * A new NSS module nss-mymachines has been added, that
1215 automatically resolves the names of all local registered
1216 containers to their respective IP addresses.
1218 * A new client tool "networkctl" for systemd-networkd has been
1219 added. It currently is entirely passive and will query
1220 networking configuration from udev, rtnetlink and networkd,
1221 and present it to the user in a very friendly
1222 way. Eventually, we hope to extend it to become a full
1223 control utility for networkd.
1225 * .socket units gained a new DeferAcceptSec= setting that
1226 controls the kernels' TCP_DEFER_ACCEPT sockopt for
1227 TCP. Similar, support for controlling TCP keep-alive
1228 settings has been added (KeepAliveTimeSec=,
1229 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
1230 turning off Nagle's algorithm on TCP has been added
1233 * logind learned a new session type "web", for use in projects
1234 like Cockpit which register web clients as PAM sessions.
1236 * timer units with at least one OnCalendar= setting will now
1237 be started only after timer-sync.target has been
1238 reached. This way they will not elapse before the system
1239 clock has been corrected by a local NTP client or
1240 similar. This is particular useful on RTC-less embedded
1241 machines, that come up with an invalid system clock.
1243 * systemd-nspawn's --network-veth= switch should now result in
1244 stable MAC addresses for both the outer and the inner side
1247 * systemd-nspawn gained a new --volatile= switch for running
1248 container instances with /etc or /var unpopulated.
1250 * The kdbus client code has been updated to use the new Linux
1251 3.17 memfd subsystem instead of the old kdbus-specific one.
1253 * systemd-networkd's DHCP client and server now support
1254 FORCERENEW. There are also new configuration options to
1255 configure the vendor client identifier and broadcast mode
1258 * systemd will no longer inform the kernel about the current
1259 timezone, as this is necessarily incorrect and racy as the
1260 kernel has no understanding of DST and similar
1261 concepts. This hence means FAT timestamps will be always
1262 considered UTC, similar to what Android is already
1263 doing. Also, when the RTC is configured to the local time
1264 (rather than UTC) systemd will never synchronize back to it,
1265 as this might confuse Windows at a later boot.
1267 * systemd-analyze gained a new command "verify" for offline
1268 validation of unit files.
1270 * systemd-networkd gained support for a couple of additional
1271 settings for bonding networking setups. Also, the metric for
1272 statically configured routes may now be configured. For
1273 network interfaces where this is appropriate the peer IP
1274 address may now be configured.
1276 * systemd-networkd's DHCP client will no longer request
1277 broadcasting by default, as this tripped up some networks.
1278 For hardware where broadcast is required the feature should
1279 be switched back on using RequestBroadcast=yes.
1281 * systemd-networkd will now set up IPv4LL addresses (when
1282 enabled) even if DHCP is configured successfully.
1284 * udev will now default to respect network device names given
1285 by the kernel when the kernel indicates that these are
1286 predictable. This behavior can be tweaked by changing
1287 NamePolicy= in the relevant .link file.
1289 * A new library systemd-terminal has been added that
1290 implements full TTY stream parsing and rendering. This
1291 library is supposed to be used later on for implementing a
1292 full userspace VT subsystem, replacing the current kernel
1295 * A new tool systemd-journal-upload has been added to push
1296 journal data to a remote system running
1297 systemd-journal-remote.
1299 * journald will no longer forward all local data to another
1300 running syslog daemon. This change has been made because
1301 rsyslog (which appears to be the most commonly used syslog
1302 implementation these days) no longer makes use of this, and
1303 instead pulls the data out of the journal on its own. Since
1304 forwarding the messages to a non-existent syslog server is
1305 more expensive than we assumed we have now turned this
1306 off. If you run a syslog server that is not a recent rsyslog
1307 version, you have to turn this option on again
1308 (ForwardToSyslog= in journald.conf).
1310 * journald now optionally supports the LZ4 compressor for
1311 larger journal fields. This compressor should perform much
1312 better than XZ which was the previous default.
1314 * machinectl now shows the IP addresses of local containers,
1315 if it knows them, plus the interface name of the container.
1317 * A new tool "systemd-escape" has been added that makes it
1318 easy to escape strings to build unit names and similar.
1320 * sd_notify() messages may now include a new ERRNO= field
1321 which is parsed and collected by systemd and shown among the
1322 "systemctl status" output for a service.
1324 * A new component "systemd-firstboot" has been added that
1325 queries the most basic systemd information (timezone,
1326 hostname, root password) interactively on first
1327 boot. Alternatively it may also be used to provision these
1328 things offline on OS images installed into directories.
1330 * The default sysctl.d/ snippets will now set
1332 net.ipv4.conf.default.promote_secondaries=1
1334 This has the benefit of no flushing secondary IP addresses
1335 when primary addresses are removed.
1337 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
1338 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
1339 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
1340 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
1341 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
1342 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
1343 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
1344 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
1345 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
1346 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
1347 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
1348 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
1349 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
1350 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
1351 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
1353 -- Berlin, 2014-08-19
1357 * A new tool systemd-sysusers has been added. This tool
1358 creates system users and groups in /etc/passwd and
1359 /etc/group, based on static declarative system user/group
1360 definitions in /usr/lib/sysusers.d/. This is useful to
1361 enable factory resets and volatile systems that boot up with
1362 an empty /etc directory, and thus need system users and
1363 groups created during early boot. systemd now also ships
1364 with two default sysusers.d/ files for the most basic
1365 users and groups systemd and the core operating system
1368 * A new tmpfiles snippet has been added that rebuilds the
1369 essential files in /etc on boot, should they be missing.
1371 * A directive for ensuring automatic clean-up of
1372 /var/cache/man/ has been removed from the default
1373 configuration. This line should now be shipped by the man
1374 implementation. The necessary change has been made to the
1375 man-db implementation. Note that you need to update your man
1376 implementation to one that ships this line, otherwise no
1377 automatic clean-up of /var/cache/man will take place.
1379 * A new condition ConditionNeedsUpdate= has been added that
1380 may conditionalize services to only run when /etc or /var
1381 are "older" than the vendor operating system resources in
1382 /usr. This is useful for reconstructing or updating /etc
1383 after an offline update of /usr or a factory reset, on the
1384 next reboot. Services that want to run once after such an
1385 update or reset should use this condition and order
1386 themselves before the new systemd-update-done.service, which
1387 will mark the two directories as fully updated. A number of
1388 service files have been added making use of this, to rebuild
1389 the udev hardware database, the journald message catalog and
1390 dynamic loader cache (ldconfig). The systemd-sysusers tool
1391 described above also makes use of this now. With this in
1392 place it is now possible to start up a minimal operating
1393 system with /etc empty cleanly. For more information on the
1394 concepts involved see this recent blog story:
1396 http://0pointer.de/blog/projects/stateless.html
1398 * A new system group "input" has been introduced, and all
1399 input device nodes get this group assigned. This is useful
1400 for system-level software to get access to input devices. It
1401 complements what is already done for "audio" and "video".
1403 * systemd-networkd learnt minimal DHCPv4 server support in
1404 addition to the existing DHCPv4 client support. It also
1405 learnt DHCPv6 client and IPv6 Router Solicitation client
1406 support. The DHCPv4 client gained support for static routes
1407 passed in from the server. Note that the [DHCPv4] section
1408 known in older systemd-networkd versions has been renamed to
1409 [DHCP] and is now also used by the DHCPv6 client. Existing
1410 .network files using settings of this section should be
1411 updated, though compatibility is maintained. Optionally, the
1412 client hostname may now be sent to the DHCP server.
1414 * networkd gained support for vxlan virtual networks as well
1415 as tun/tap and dummy devices.
1417 * networkd gained support for automatic allocation of address
1418 ranges for interfaces from a system-wide pool of
1419 addresses. This is useful for dynamically managing a large
1420 number of interfaces with a single network configuration
1421 file. In particular this is useful to easily assign
1422 appropriate IP addresses to the veth links of a large number
1423 of nspawn instances.
1425 * RPM macros for processing sysusers, sysctl and binfmt
1426 drop-in snippets at package installation time have been
1429 * The /etc/os-release file should now be placed in
1430 /usr/lib/os-release. The old location is automatically
1431 created as symlink. /usr/lib is the more appropriate
1432 location of this file, since it shall actually describe the
1433 vendor operating system shipped in /usr, and not the
1434 configuration stored in /etc.
1436 * .mount units gained a new boolean SloppyOptions= setting
1437 that maps to mount(8)'s -s option which enables permissive
1438 parsing of unknown mount options.
1440 * tmpfiles learnt a new "L+" directive which creates a symlink
1441 but (unlike "L") deletes a pre-existing file first, should
1442 it already exist and not already be the correct
1443 symlink. Similar, "b+", "c+" and "p+" directives have been
1444 added as well, which create block and character devices, as
1445 well as fifos in the filesystem, possibly removing any
1446 pre-existing files of different types.
1448 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
1449 'argument' field (which so far specified the source to
1450 symlink/copy the files from) is now optional. If omitted the
1451 same file os copied from /usr/share/factory/ suffixed by the
1452 full destination path. This is useful for populating /etc
1453 with essential files, by copying them from vendor defaults
1454 shipped in /usr/share/factory/etc.
1456 * A new command "systemctl preset-all" has been added that
1457 applies the service preset settings to all installed unit
1458 files. A new switch --preset-mode= has been added that
1459 controls whether only enable or only disable operations
1462 * A new command "systemctl is-system-running" has been added
1463 that allows checking the overall state of the system, for
1464 example whether it is fully up and running.
1466 * When the system boots up with an empty /etc, the equivalent
1467 to "systemctl preset-all" is executed during early boot, to
1468 make sure all default services are enabled after a factory
1471 * systemd now contains a minimal preset file that enables the
1472 most basic services systemd ships by default.
1474 * Unit files' [Install] section gained a new DefaultInstance=
1475 field for defining the default instance to create if a
1476 template unit is enabled with no instance specified.
1478 * A new passive target cryptsetup-pre.target has been added
1479 that may be used by services that need to make they run and
1480 finish before the first LUKS cryptographic device is set up.
1482 * The /dev/loop-control and /dev/btrfs-control device nodes
1483 are now owned by the "disk" group by default, opening up
1484 access to this group.
1486 * systemd-coredump will now automatically generate a
1487 stack trace of all core dumps taking place on the system,
1488 based on elfutils' libdw library. This stack trace is logged
1491 * systemd-coredump may now optionally store coredumps directly
1492 on disk (in /var/lib/systemd/coredump, possibly compressed),
1493 instead of storing them unconditionally in the journal. This
1494 mode is the new default. A new configuration file
1495 /etc/systemd/coredump.conf has been added to configure this
1496 and other parameters of systemd-coredump.
1498 * coredumpctl gained a new "info" verb to show details about a
1499 specific coredump. A new switch "-1" has also been added
1500 that makes sure to only show information about the most
1501 recent entry instead of all entries. Also, as the tool is
1502 generally useful now the "systemd-" prefix of the binary
1503 name has been removed. Distributions that want to maintain
1504 compatibility with the old name should add a symlink from
1505 the old name to the new name.
1507 * journald's SplitMode= now defaults to "uid". This makes sure
1508 that unprivileged users can access their own coredumps with
1509 coredumpctl without restrictions.
1511 * New kernel command line options "systemd.wants=" (for
1512 pulling an additional unit during boot), "systemd.mask="
1513 (for masking a specific unit for the boot), and
1514 "systemd.debug-shell" (for enabling the debug shell on tty9)
1515 have been added. This is implemented in the new generator
1516 "systemd-debug-generator".
1518 * systemd-nspawn will now by default filter a couple of
1519 syscalls for containers, among them those required for
1520 kernel module loading, direct x86 IO port access, swap
1521 management, and kexec. Most importantly though
1522 open_by_handle_at() is now prohibited for containers,
1523 closing a hole similar to a recently discussed vulnerability
1524 in docker regarding access to files on file hierarchies the
1525 container should normally not have access to. Note that for
1526 nspawn we generally make no security claims anyway (and
1527 this is explicitly documented in the man page), so this is
1528 just a fix for one of the most obvious problems.
1530 * A new man page file-hierarchy(7) has been added that
1531 contains a minimized, modernized version of the file system
1532 layout systemd expects, similar in style to the FHS
1533 specification or hier(5). A new tool systemd-path(1) has
1534 been added to query many of these paths for the local
1537 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
1538 longer done. Since the directory now has a per-user size
1539 limit, and is cleaned on logout this appears unnecessary,
1540 in particular since this now brings the lifecycle of this
1541 directory closer in line with how IPC objects are handled.
1543 * systemd.pc now exports a number of additional directories,
1544 including $libdir (which is useful to identify the library
1545 path for the primary architecture of the system), and a
1546 couple of drop-in directories.
1548 * udev's predictable network interface names now use the dev_port
1549 sysfs attribute, introduced in linux 3.15 instead of dev_id to
1550 distinguish between ports of the same PCI function. dev_id should
1551 only be used for ports using the same HW address, hence the need
1554 * machined has been updated to export the OS version of a
1555 container (read from /etc/os-release and
1556 /usr/lib/os-release) on the bus. This is now shown in
1557 "machinectl status" for a machine.
1559 * A new service setting RestartForceExitStatus= has been
1560 added. If configured to a set of exit signals or process
1561 return values, the service will be restarted when the main
1562 daemon process exits with any of them, regardless of the
1565 * systemctl's -H switch for connecting to remote systemd
1566 machines has been extended so that it may be used to
1567 directly connect to a specific container on the
1568 host. "systemctl -H root@foobar:waldi" will now connect as
1569 user "root" to host "foobar", and then proceed directly to
1570 the container named "waldi". Note that currently you have to
1571 authenticate as user "root" for this to work, as entering
1572 containers is a privileged operation.
1574 Contributions from: Andreas Henriksson, Benjamin Steinwender,
1575 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
1576 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
1577 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
1578 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
1579 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
1580 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
1581 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
1582 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
1583 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
1584 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
1585 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
1587 -- Berlin, 2014-07-03
1591 * As an experimental feature, udev now tries to lock the
1592 disk device node (flock(LOCK_SH|LOCK_NB)) while it
1593 executes events for the disk or any of its partitions.
1594 Applications like partitioning programs can lock the
1595 disk device node (flock(LOCK_EX)) and claim temporary
1596 device ownership that way; udev will entirely skip all event
1597 handling for this disk and its partitions. If the disk
1598 was opened for writing, the close will trigger a partition
1599 table rescan in udev's "watch" facility, and if needed
1600 synthesize "change" events for the disk and all its partitions.
1601 This is now unconditionally enabled, and if it turns out to
1602 cause major problems, we might turn it on only for specific
1603 devices, or might need to disable it entirely. Device Mapper
1604 devices are excluded from this logic.
1606 * We temporarily dropped the "-l" switch for fsck invocations,
1607 since they collide with the flock() logic above. util-linux
1608 upstream has been changed already to avoid this conflict,
1609 and we will readd "-l" as soon as util-linux with this
1610 change has been released.
1612 * The dependency on libattr has been removed. Since a long
1613 time, the extended attribute calls have moved to glibc, and
1614 libattr is thus unnecessary.
1616 * Virtualization detection works without priviliges now. This
1617 means the systemd-detect-virt binary no longer requires
1618 CAP_SYS_PTRACE file capabilities, and our daemons can run
1619 with fewer privileges.
1621 * systemd-networkd now runs under its own "systemd-network"
1622 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
1623 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
1624 loses the ability to write to files owned by root this way.
1626 * Similar, systemd-resolved now runs under its own
1627 "systemd-resolve" user with no capabilities remaining.
1629 * Similar, systemd-bus-proxyd now runs under its own
1630 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
1632 * systemd-networkd gained support for setting up "veth"
1633 virtual ethernet devices for container connectivity, as well
1634 as GRE and VTI tunnels.
1636 * systemd-networkd will no longer automatically attempt to
1637 manually load kernel modules necessary for certain tunnel
1638 transports. Instead, it is assumed the kernel loads them
1639 automatically when required. This only works correctly on
1640 very new kernels. On older kernels, please consider adding
1641 the kernel modules to /etc/modules-load.d/ as a work-around.
1643 * The resolv.conf file systemd-resolved generates has been
1644 moved to /run/systemd/resolve/. If you have a symlink from
1645 /etc/resolv.conf, it might be necessary to correct it.
1647 * Two new service settings, ProtectHome= and ProtectSystem=,
1648 have been added. When enabled, they will make the user data
1649 (such as /home) inaccessible or read-only and the system
1650 (such as /usr) read-only, for specific services. This allows
1651 very light-weight per-service sandboxing to avoid
1652 modifications of user data or system files from
1653 services. These two new switches have been enabled for all
1654 of systemd's long-running services, where appropriate.
1656 * Socket units gained new SocketUser= and SocketGroup=
1657 settings to set the owner user and group of AF_UNIX sockets
1658 and FIFOs in the file system.
1660 * Socket units gained a new RemoveOnStop= setting. If enabled,
1661 all FIFOS and sockets in the file system will be removed
1662 when the specific socket unit is stopped.
1664 * Socket units gained a new Symlinks= setting. It takes a list
1665 of symlinks to create to file system sockets or FIFOs
1666 created by the specific Unix sockets. This is useful to
1667 manage symlinks to socket nodes with the same life-cycle as
1670 * The /dev/log socket and /dev/initctl FIFO have been moved to
1671 /run, and have been replaced by symlinks. This allows
1672 connecting to these facilities even if PrivateDevices=yes is
1673 used for a service (which makes /dev/log itself unavailable,
1674 but /run is left). This also has the benefit of ensuring
1675 that /dev only contains device nodes, directories and
1676 symlinks, and nothing else.
1678 * sd-daemon gained two new calls sd_pid_notify() and
1679 sd_pid_notifyf(). They are similar to sd_notify() and
1680 sd_notifyf(), but allow overriding of the source PID of
1681 notification messages if permissions permit this. This is
1682 useful to send notify messages on behalf of a different
1683 process (for example, the parent process). The
1684 systemd-notify tool has been updated to make use of this
1685 when sending messages (so that notification messages now
1686 originate from the shell script invoking systemd-notify and
1687 not the systemd-notify process itself. This should minimize
1688 a race where systemd fails to associate notification
1689 messages to services when the originating process already
1692 * A new "on-abnormal" setting for Restart= has been added. If
1693 set, it will result in automatic restarts on all "abnormal"
1694 reasons for a process to exit, which includes unclean
1695 signals, core dumps, timeouts and watchdog timeouts, but
1696 does not include clean and unclean exit codes or clean
1697 signals. Restart=on-abnormal is an alternative for
1698 Restart=on-failure for services that shall be able to
1699 terminate and avoid restarts on certain errors, by
1700 indicating so with an unclean exit code. Restart=on-failure
1701 or Restart=on-abnormal is now the recommended setting for
1702 all long-running services.
1704 * If the InaccessibleDirectories= service setting points to a
1705 mount point (or if there are any submounts contained within
1706 it), it is now attempted to completely unmount it, to make
1707 the file systems truly unavailable for the respective
1710 * The ReadOnlyDirectories= service setting and
1711 systemd-nspawn's --read-only parameter are now recursively
1712 applied to all submounts, too.
1714 * Mount units may now be created transiently via the bus APIs.
1716 * The support for SysV and LSB init scripts has been removed
1717 from the systemd daemon itself. Instead, it is now
1718 implemented as a generator that creates native systemd units
1719 from these scripts when needed. This enables us to remove a
1720 substantial amount of legacy code from PID 1, following the
1721 fact that many distributions only ship a very small number
1722 of LSB/SysV init scripts nowadays.
1724 * Privileged Xen (dom0) domains are not considered
1725 virtualization anymore by the virtualization detection
1726 logic. After all, they generally have unrestricted access to
1727 the hardware and usually are used to manage the unprivileged
1730 * systemd-tmpfiles gained a new "C" line type, for copying
1731 files or entire directories.
1733 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
1734 lines. So far, they have been non-globbing versions of the
1735 latter, and have thus been redundant. In future, it is
1736 recommended to only use "z". "m" has hence been removed
1737 from the documentation, even though it stays supported.
1739 * A tmpfiles snippet to recreate the most basic structure in
1740 /var has been added. This is enough to create the /var/run →
1741 /run symlink and create a couple of structural
1742 directories. This allows systems to boot up with an empty or
1743 volatile /var. Of course, while with this change, the core OS
1744 now is capable with dealing with a volatile /var, not all
1745 user services are ready for it. However, we hope that sooner
1746 or later, many service daemons will be changed upstream so
1747 that they are able to automatically create their necessary
1748 directories in /var at boot, should they be missing. This is
1749 the first step to allow state-less systems that only require
1750 the vendor image for /usr to boot.
1752 * systemd-nspawn has gained a new --tmpfs= switch to mount an
1753 empty tmpfs instance to a specific directory. This is
1754 particularly useful for making use of the automatic
1755 reconstruction of /var (see above), by passing --tmpfs=/var.
1757 * Access modes specified in tmpfiles snippets may now be
1758 prefixed with "~", which indicates that they shall be masked
1759 by whether the existing file or directory is currently
1760 writable, readable or executable at all. Also, if specified,
1761 the sgid/suid/sticky bits will be masked for all
1764 * A new passive target unit "network-pre.target" has been
1765 added which is useful for services that shall run before any
1766 network is configured, for example firewall scripts.
1768 * The "floppy" group that previously owned the /dev/fd*
1769 devices is no longer used. The "disk" group is now used
1770 instead. Distributions should probably deprecate usage of
1773 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
1774 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
1775 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
1776 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
1777 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
1778 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
1781 -- Berlin, 2014-06-11
1785 * A new "systemd-timesyncd" daemon has been added for
1786 synchronizing the system clock across the network. It
1787 implements an SNTP client. In contrast to NTP
1788 implementations such as chrony or the NTP reference server,
1789 this only implements a client side, and does not bother with
1790 the full NTP complexity, focusing only on querying time from
1791 one remote server and synchronizing the local clock to
1792 it. Unless you intend to serve NTP to networked clients or
1793 want to connect to local hardware clocks, this simple NTP
1794 client should be more than appropriate for most
1795 installations. The daemon runs with minimal privileges, and
1796 has been hooked up with networkd to only operate when
1797 network connectivity is available. The daemon saves the
1798 current clock to disk every time a new NTP sync has been
1799 acquired, and uses this to possibly correct the system clock
1800 early at bootup, in order to accommodate for systems that
1801 lack an RTC such as the Raspberry Pi and embedded devices,
1802 and to make sure that time monotonically progresses on these
1803 systems, even if it is not always correct. To make use of
1804 this daemon, a new system user and group "systemd-timesync"
1805 needs to be created on installation of systemd.
1807 * The queue "seqnum" interface of libudev has been disabled, as
1808 it was generally incompatible with device namespacing as
1809 sequence numbers of devices go "missing" if the devices are
1810 part of a different namespace.
1812 * "systemctl list-timers" and "systemctl list-sockets" gained
1813 a --recursive switch for showing units of these types also
1814 for all local containers, similar in style to the already
1815 supported --recursive switch for "systemctl list-units".
1817 * A new RebootArgument= setting has been added for service
1818 units, which may be used to specify a kernel reboot argument
1819 to use when triggering reboots with StartLimitAction=.
1821 * A new FailureAction= setting has been added for service
1822 units which may be used to specify an operation to trigger
1823 when a service fails. This works similarly to
1824 StartLimitAction=, but unlike it, controls what is done
1825 immediately rather than only after several attempts to
1826 restart the service in question.
1828 * hostnamed got updated to also expose the kernel name,
1829 release, and version on the bus. This is useful for
1830 executing commands like hostnamectl with the -H switch.
1831 systemd-analyze makes use of this to properly display
1832 details when running non-locally.
1834 * The bootchart tool can now show cgroup information in the
1835 graphs it generates.
1837 * The CFS CPU quota cgroup attribute is now exposed for
1838 services. The new CPUQuota= switch has been added for this
1839 which takes a percentage value. Setting this will have the
1840 result that a service may never get more CPU time than the
1841 specified percentage, even if the machine is otherwise idle.
1843 * systemd-networkd learned IPIP and SIT tunnel support.
1845 * LSB init scripts exposing a dependency on $network will now
1846 get a dependency on network-online.target rather than simply
1847 network.target. This should bring LSB handling closer to
1848 what it was on SysV systems.
1850 * A new fsck.repair= kernel option has been added to control
1851 how fsck shall deal with unclean file systems at boot.
1853 * The (.ini) configuration file parser will now silently
1854 ignore sections whose name begins with "X-". This may be
1855 used to maintain application-specific extension sections in unit
1858 * machined gained a new API to query the IP addresses of
1859 registered containers. "machinectl status" has been updated
1860 to show these addresses in its output.
1862 * A new call sd_uid_get_display() has been added to the
1863 sd-login APIs for querying the "primary" session of a
1864 user. The "primary" session of the user is elected from the
1865 user's sessions and generally a graphical session is
1866 preferred over a text one.
1868 * A minimal systemd-resolved daemon has been added. It
1869 currently simply acts as a companion to systemd-networkd and
1870 manages resolv.conf based on per-interface DNS
1871 configuration, possibly supplied via DHCP. In the long run
1872 we hope to extend this into a local DNSSEC enabled DNS and
1875 * The systemd-networkd-wait-online tool is now enabled by
1876 default. It will delay network-online.target until a network
1877 connection has been configured. The tool primarily integrates
1878 with networkd, but will also make a best effort to make sense
1879 of network configuration performed in some other way.
1881 * Two new service options StartupCPUShares= and
1882 StartupBlockIOWeight= have been added that work similarly to
1883 CPUShares= and BlockIOWeight= however only apply during
1884 system startup. This is useful to prioritize certain services
1885 differently during bootup than during normal runtime.
1887 * hostnamed has been changed to prefer the statically
1888 configured hostname in /etc/hostname (unless set to
1889 'localhost' or empty) over any dynamic one supplied by
1890 dhcp. With this change, the rules for picking the hostname
1891 match more closely the rules of other configuration settings
1892 where the local administrator's configuration in /etc always
1893 overrides any other settings.
1895 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
1896 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
1897 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
1898 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
1899 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
1900 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
1901 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
1902 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
1903 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
1904 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
1905 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
1906 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
1907 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
1908 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
1909 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
1910 Lindskog, WaLyong Cho, Will Woods, Zbigniew
1913 -- Beijing, 2014-05-28
1917 * When restoring the screen brightness at boot, stay away from
1918 the darkest setting or from the lowest 5% of the available
1919 range, depending on which is the larger value of both. This
1920 should effectively protect the user from rebooting into a
1921 black screen, should the brightness have been set to minimum
1924 * sd-login gained a new sd_machine_get_class() call to
1925 determine the class ("vm" or "container") of a machine
1926 registered with machined.
1928 * sd-login gained new calls
1929 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
1930 to query the identity of the peer of a local AF_UNIX
1931 connection. They operate similarly to their sd_pid_get_xyz()
1934 * PID 1 will now maintain a system-wide system state engine
1935 with the states "starting", "running", "degraded",
1936 "maintenance", "stopping". These states are bound to system
1937 startup, normal runtime, runtime with at least one failed
1938 service, rescue/emergency mode and system shutdown. This
1939 state is shown in the "systemctl status" output when no unit
1940 name is passed. It is useful to determine system state, in
1941 particularly when doing so for many systems or containers at
1944 * A new command "list-machines" has been added to "systemctl"
1945 that lists all local OS containers and shows their system
1946 state (see above), if systemd runs inside of them.
1948 * systemctl gained a new "-r" switch to recursively enumerate
1949 units on all local containers, when used with the
1950 "list-unit" command (which is the default one that is
1951 executed when no parameters are specified).
1953 * The GPT automatic partition discovery logic will now honour
1954 two GPT partition flags: one may be set on a partition to
1955 cause it to be mounted read-only, and the other may be set
1956 on a partition to ignore it during automatic discovery.
1958 * Two new GPT type UUIDs have been added for automatic root
1959 partition discovery, for 32-bit and 64-bit ARM. This is not
1960 particularly useful for discovering the root directory on
1961 these architectures during bare-metal boots (since UEFI is
1962 not common there), but still very useful to allow booting of
1963 ARM disk images in nspawn with the -i option.
1965 * MAC addresses of interfaces created with nspawn's
1966 --network-interface= switch will now be generated from the
1967 machine name, and thus be stable between multiple invocations
1970 * logind will now automatically remove all IPC objects owned
1971 by a user if she or he fully logs out. This makes sure that
1972 users who are logged out cannot continue to consume IPC
1973 resources. This covers SysV memory, semaphores and message
1974 queues as well as POSIX shared memory and message
1975 queues. Traditionally, SysV and POSIX IPC had no life-cycle
1976 limits. With this functionality, that is corrected. This may
1977 be turned off by using the RemoveIPC= switch of logind.conf.
1979 * The systemd-machine-id-setup and tmpfiles tools gained a
1980 --root= switch to operate on a specific root directory,
1983 * journald can now forward logged messages to the TTYs of all
1984 logged in users ("wall"). This is the default for all
1985 emergency messages now.
1987 * A new tool systemd-journal-remote has been added to stream
1988 journal log messages across the network.
1990 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
1991 controller trees are mounted into it. Note that the
1992 directories mounted beneath it are not read-only. This is a
1993 security measure and is particularly useful because glibc
1994 actually includes a search logic to pick any tmpfs it can
1995 find to implement shm_open() if /dev/shm is not available
1996 (which it might very well be in namespaced setups).
1998 * machinectl gained a new "poweroff" command to cleanly power
1999 down a local OS container.
2001 * The PrivateDevices= unit file setting will now also drop the
2002 CAP_MKNOD capability from the capability bound set, and
2003 imply DevicePolicy=closed.
2005 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
2006 comprehensively on all long-running systemd services where
2007 this is appropriate.
2009 * systemd-udevd will now run in a disassociated mount
2010 namespace. To mount directories from udev rules, make sure to
2011 pull in mount units via SYSTEMD_WANTS properties.
2013 * The kdbus support gained support for uploading policy into
2014 the kernel. sd-bus gained support for creating "monitoring"
2015 connections that can eavesdrop into all bus communication
2016 for debugging purposes.
2018 * Timestamps may now be specified in seconds since the UNIX
2019 epoch Jan 1st, 1970 by specifying "@" followed by the value
2022 * Native tcpwrap support in systemd has been removed. tcpwrap
2023 is old code, not really maintained anymore and has serious
2024 shortcomings, and better options such as firewalls
2025 exist. For setups that require tcpwrap usage, please
2026 consider invoking your socket-activated service via tcpd,
2027 like on traditional inetd.
2029 * A new system.conf configuration option
2030 DefaultTimerAccuracySec= has been added that controls the
2031 default AccuracySec= setting of .timer units.
2033 * Timer units gained a new WakeSystem= switch. If enabled,
2034 timers configured this way will cause the system to resume
2035 from system suspend (if the system supports that, which most
2038 * Timer units gained a new Persistent= switch. If enabled,
2039 timers configured this way will save to disk when they have
2040 been last triggered. This information is then used on next
2041 reboot to possible execute overdue timer events, that
2042 could not take place because the system was powered off.
2043 This enables simple anacron-like behaviour for timer units.
2045 * systemctl's "list-timers" will now also list the time a
2046 timer unit was last triggered in addition to the next time
2047 it will be triggered.
2049 * systemd-networkd will now assign predictable IPv4LL
2050 addresses to its local interfaces.
2052 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
2053 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
2054 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
2055 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
2056 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
2057 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
2058 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
2059 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
2062 -- Berlin, 2014-03-25
2066 * A new unit file setting RestrictAddressFamilies= has been
2067 added to restrict which socket address families unit
2068 processes gain access to. This takes address family names
2069 like "AF_INET" or "AF_UNIX", and is useful to minimize the
2070 attack surface of services via exotic protocol stacks. This
2071 is built on seccomp system call filters.
2073 * Two new unit file settings RuntimeDirectory= and
2074 RuntimeDirectoryMode= have been added that may be used to
2075 manage a per-daemon runtime directories below /run. This is
2076 an alternative for setting up directory permissions with
2077 tmpfiles snippets, and has the advantage that the runtime
2078 directory's lifetime is bound to the daemon runtime and that
2079 the daemon starts up with an empty directory each time. This
2080 is particularly useful when writing services that drop
2081 privileges using the User= or Group= setting.
2083 * The DeviceAllow= unit setting now supports globbing for
2084 matching against device group names.
2086 * The systemd configuration file system.conf gained new
2087 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
2088 DefaultMemoryAccounting= to globally turn on/off accounting
2089 for specific resources (cgroups) for all units. These
2090 settings may still be overridden individually in each unit
2093 * systemd-gpt-auto-generator is now able to discover /srv and
2094 root partitions in addition to /home and swap partitions. It
2095 also supports LUKS-encrypted partitions now. With this in
2096 place, automatic discovery of partitions to mount following
2097 the Discoverable Partitions Specification
2098 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
2099 is now a lot more complete. This allows booting without
2100 /etc/fstab and without root= on the kernel command line on
2101 systems prepared appropriately.
2103 * systemd-nspawn gained a new --image= switch which allows
2104 booting up disk images and Linux installations on any block
2105 device that follow the Discoverable Partitions Specification
2106 (see above). This means that installations made with
2107 appropriately updated installers may now be started and
2108 deployed using container managers, completely
2109 unmodified. (We hope that libvirt-lxc will add support for
2110 this feature soon, too.)
2112 * systemd-nspawn gained a new --network-macvlan= setting to
2113 set up a private macvlan interface for the
2114 container. Similarly, systemd-networkd gained a new
2115 Kind=macvlan setting in .netdev files.
2117 * systemd-networkd now supports configuring local addresses
2120 * A new tool systemd-network-wait-online has been added to
2121 synchronously wait for network connectivity using
2124 * The sd-bus.h bus API gained a new sd_bus_track object for
2125 tracking the life-cycle of bus peers. Note that sd-bus.h is
2126 still not a public API though (unless you specify
2127 --enable-kdbus on the configure command line, which however
2128 voids your warranty and you get no API stability guarantee).
2130 * The $XDG_RUNTIME_DIR runtime directories for each user are
2131 now individual tmpfs instances, which has the benefit of
2132 introducing separate pools for each user, with individual
2133 size limits, and thus making sure that unprivileged clients
2134 can no longer negatively impact the system or other users by
2135 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
2136 RuntimeDirectorySize= has been introduced that allows
2137 controlling the default size limit for all users. It
2138 defaults to 10% of the available physical memory. This is no
2139 replacement for quotas on tmpfs though (which the kernel
2140 still does not support), as /dev/shm and /tmp are still
2141 shared resources used by both the system and unprivileged
2144 * logind will now automatically turn off automatic suspending
2145 on laptop lid close when more than one display is
2146 connected. This was previously expected to be implemented
2147 individually in desktop environments (such as GNOME),
2148 however has been added to logind now, in order to fix a
2149 boot-time race where a desktop environment might not have
2150 been started yet and thus not been able to take an inhibitor
2151 lock at the time where logind already suspends the system
2152 due to a closed lid.
2154 * logind will now wait at least 30s after each system
2155 suspend/resume cycle, and 3min after system boot before
2156 suspending the system due to a closed laptop lid. This
2157 should give USB docking stations and similar enough time to
2158 be probed and configured after system resume and boot in
2159 order to then act as suspend blocker.
2161 * systemd-run gained a new --property= setting which allows
2162 initialization of resource control properties (and others)
2163 for the created scope or service unit. Example: "systemd-run
2164 --property=BlockIOWeight=10 updatedb" may be used to run
2165 updatedb at a low block IO scheduling weight.
2167 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
2168 now also work in --scope mode.
2170 * When systemd is compiled with kdbus support, basic support
2171 for enforced policies is now in place. (Note that enabling
2172 kdbus still voids your warranty and no API compatibility
2175 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
2176 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
2177 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
2178 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
2179 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
2180 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
2181 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
2182 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
2183 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
2184 Zbigniew Jędrzejewski-Szmek
2186 -- Berlin, 2014-03-12
2190 * systemd will now relabel /dev after loading the SMACK policy
2191 according to SMACK rules.
2193 * A new unit file option AppArmorProfile= has been added to
2194 set the AppArmor profile for the processes of a unit.
2196 * A new condition check ConditionArchitecture= has been added
2197 to conditionalize units based on the system architecture, as
2198 reported by uname()'s "machine" field.
2200 * systemd-networkd now supports matching on the system
2201 virtualization, architecture, kernel command line, host name
2204 * logind is now a lot more aggressive when suspending the
2205 machine due to a closed laptop lid. Instead of acting only
2206 on the lid close action, it will continuously watch the lid
2207 status and act on it. This is useful for laptops where the
2208 power button is on the outside of the chassis so that it can
2209 be reached without opening the lid (such as the Lenovo
2210 Yoga). On those machines, logind will now immediately
2211 re-suspend the machine if the power button has been
2212 accidentally pressed while the laptop was suspended and in a
2213 backpack or similar.
2215 * logind will now watch SW_DOCK switches and inhibit reaction
2216 to the lid switch if it is pressed. This means that logind
2217 will not suspend the machine anymore if the lid is closed
2218 and the system is docked, if the laptop supports SW_DOCK
2219 notifications via the input layer. Note that ACPI docking
2220 stations do not generate this currently. Also note that this
2221 logic is usually not fully sufficient and Desktop
2222 Environments should take a lid switch inhibitor lock when an
2223 external display is connected, as systemd will not watch
2226 * nspawn will now make use of the devices cgroup controller by
2227 default, and only permit creation of and access to the usual
2228 API device nodes like /dev/null or /dev/random, as well as
2229 access to (but not creation of) the pty devices.
2231 * We will now ship a default .network file for
2232 systemd-networkd that automatically configures DHCP for
2233 network interfaces created by nspawn's --network-veth or
2234 --network-bridge= switches.
2236 * systemd will now understand the usual M, K, G, T suffixes
2237 according to SI conventions (i.e. to the base 1000) when
2238 referring to throughput and hardware metrics. It will stay
2239 with IEC conventions (i.e. to the base 1024) for software
2240 metrics, according to what is customary according to
2241 Wikipedia. We explicitly document which base applies for
2242 each configuration option.
2244 * The DeviceAllow= setting in unit files now supports a syntax
2245 to whitelist an entire group of devices node majors at once,
2246 based on the /proc/devices listing. For example, with the
2247 string "char-pts", it is now possible to whitelist all
2248 current and future pseudo-TTYs at once.
2250 * sd-event learned a new "post" event source. Event sources of
2251 this type are triggered by the dispatching of any event
2252 source of a type that is not "post". This is useful for
2253 implementing clean-up and check event sources that are
2254 triggered by other work being done in the program.
2256 * systemd-networkd is no longer statically enabled, but uses
2257 the usual [Install] sections so that it can be
2258 enabled/disabled using systemctl. It still is enabled by
2261 * When creating a veth interface pair with systemd-nspawn, the
2262 host side will now be prefixed with "vb-" if
2263 --network-bridge= is used, and with "ve-" if --network-veth
2264 is used. This way, it is easy to distinguish these cases on
2265 the host, for example to apply different configuration to
2266 them with systemd-networkd.
2268 * The compatibility libraries for libsystemd-journal.so,
2269 libsystem-id128.so, libsystemd-login.so and
2270 libsystemd-daemon.so do not make use of IFUNC
2271 anymore. Instead, we now build libsystemd.so multiple times
2272 under these alternative names. This means that the footprint
2273 is drastically increased, but given that these are
2274 transitional compatibility libraries, this should not matter
2275 much. This change has been made necessary to support the ARM
2276 platform for these compatibility libraries, as the ARM
2277 toolchain is not really at the same level as the toolchain
2278 for other architectures like x86 and does not support
2279 IFUNC. Please make sure to use --enable-compat-libs only
2280 during a transitional period!
2282 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
2283 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
2284 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
2285 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
2286 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
2287 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
2288 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
2289 Zbigniew Jędrzejewski-Szmek
2291 -- Berlin, 2014-02-24
2295 * A new component "systemd-networkd" has been added that can
2296 be used to configure local network interfaces statically or
2297 via DHCP. It is capable of bringing up bridges, VLANs, and
2298 bonding. Currently, no hook-ups for interactive network
2299 configuration are provided. Use this for your initrd,
2300 container, embedded, or server setup if you need a simple,
2301 yet powerful, network configuration solution. This
2302 configuration subsystem is quite nifty, as it allows wildcard
2303 hotplug matching in interfaces. For example, with a single
2304 configuration snippet, you can configure that all Ethernet
2305 interfaces showing up are automatically added to a bridge,
2306 or similar. It supports link-sensing and more.
2308 * A new tool "systemd-socket-proxyd" has been added which can
2309 act as a bidirectional proxy for TCP sockets. This is
2310 useful for adding socket activation support to services that
2311 do not actually support socket activation, including virtual
2312 machines and the like.
2314 * Add a new tool to save/restore rfkill state on
2317 * Save/restore state of keyboard backlights in addition to
2318 display backlights on shutdown/boot.
2320 * udev learned a new SECLABEL{} construct to label device
2321 nodes with a specific security label when they appear. For
2322 now, only SECLABEL{selinux} is supported, but the syntax is
2323 prepared for additional security frameworks.
2325 * udev gained a new scheme to configure link-level attributes
2326 from files in /etc/systemd/network/*.link. These files can
2327 match against MAC address, device path, driver name and type,
2328 and will apply attributes like the naming policy, link speed,
2329 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
2330 address assignment policy (randomized, ...).
2332 * The configuration of network interface naming rules for
2333 "permanent interface names" has changed: a new NamePolicy=
2334 setting in the [Link] section of .link files determines the
2335 priority of possible naming schemes (onboard, slot, mac,
2336 path). The default value of this setting is determined by
2337 /usr/lib/net/links/99-default.link. Old
2338 80-net-name-slot.rules udev configuration file has been
2339 removed, so local configuration overriding this file should
2340 be adapated to override 99-default.link instead.
2342 * When the User= switch is used in a unit file, also
2343 initialize $SHELL= based on the user database entry.
2345 * systemd no longer depends on libdbus. All communication is
2346 now done with sd-bus, systemd's low-level bus library
2349 * kdbus support has been added to PID 1 itself. When kdbus is
2350 enabled, this causes PID 1 to set up the system bus and
2351 enable support for a new ".busname" unit type that
2352 encapsulates bus name activation on kdbus. It works a little
2353 bit like ".socket" units, except for bus names. A new
2354 generator has been added that converts classic dbus1 service
2355 activation files automatically into native systemd .busname
2358 * sd-bus: add a light-weight vtable implementation that allows
2359 defining objects on the bus with a simple static const
2360 vtable array of its methods, signals and properties.
2362 * systemd will not generate or install static dbus
2363 introspection data anymore to /usr/share/dbus-1/interfaces,
2364 as the precise format of these files is unclear, and
2365 nothing makes use of it.
2367 * A proxy daemon is now provided to proxy clients connecting
2368 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
2369 compatibility with classic D-Bus.
2371 * A bus driver implementation has been added that supports the
2372 classic D-Bus bus driver calls on kdbus, also for
2373 compatibility purposes.
2375 * A new API "sd-event.h" has been added that implements a
2376 minimal event loop API built around epoll. It provides a
2377 couple of features that direct epoll usage is lacking:
2378 prioritization of events, scales to large numbers of timer
2379 events, per-event timer slack (accuracy), system-wide
2380 coalescing of timer events, exit handlers, watchdog
2381 supervision support using systemd's sd_notify() API, child
2384 * A new API "sd-rntl.h" has been added that provides an API
2385 around the route netlink interface of the kernel, similar in
2386 style to "sd-bus.h".
2388 * A new API "sd-dhcp-client.h" has been added that provides a
2389 small DHCPv4 client-side implementation. This is used by
2392 * There is a new kernel command line option
2393 "systemd.restore_state=0|1". When set to "0", none of the
2394 systemd tools will restore saved runtime state to hardware
2395 devices. More specifically, the rfkill and backlight states
2398 * The FsckPassNo= compatibility option in mount/service units
2399 has been removed. The fstab generator will now add the
2400 necessary dependencies automatically, and does not require
2401 PID1's support for that anymore.
2403 * journalctl gained a new switch, --list-boots, that lists
2404 recent boots with their times and boot IDs.
2406 * The various tools like systemctl, loginctl, timedatectl,
2407 busctl, systemd-run, ... have gained a new switch "-M" to
2408 connect to a specific, local OS container (as direct
2409 connection, without requiring SSH). This works on any
2410 container that is registered with machined, such as those
2411 created by libvirt-lxc or nspawn.
2413 * systemd-run and systemd-analyze also gained support for "-H"
2414 to connect to remote hosts via SSH. This is particularly
2415 useful for systemd-run because it enables queuing of jobs
2416 onto remote systems.
2418 * machinectl gained a new command "login" to open a getty
2419 login in any local container. This works with any container
2420 that is registered with machined (such as those created by
2421 libvirt-lxc or nspawn), and which runs systemd inside.
2423 * machinectl gained a new "reboot" command that may be used to
2424 trigger a reboot on a specific container that is registered
2425 with machined. This works on any container that runs an init
2426 system of some kind.
2428 * systemctl gained a new "list-timers" command to print a nice
2429 listing of installed timer units with the times they elapse
2432 * Alternative reboot() parameters may now be specified on the
2433 "systemctl reboot" command line and are passed to the
2434 reboot() system call.
2436 * systemctl gained a new --job-mode= switch to configure the
2437 mode to queue a job with. This is a more generic version of
2438 --fail, --irreversible, and --ignore-dependencies, which are
2439 still available but not advertised anymore.
2441 * /etc/systemd/system.conf gained new settings to configure
2442 various default timeouts of units, as well as the default
2443 start limit interval and burst. These may still be overridden
2446 * PID1 will now export on the bus profile data of the security
2447 policy upload process (such as the SELinux policy upload to
2450 * journald: when forwarding logs to the console, include
2451 timestamps (following the setting in
2452 /sys/module/printk/parameters/time).
2454 * OnCalendar= in timer units now understands the special
2455 strings "yearly" and "annually". (Both are equivalent)
2457 * The accuracy of timer units is now configurable with the new
2458 AccuracySec= setting. It defaults to 1min.
2460 * A new dependency type JoinsNamespaceOf= has been added that
2461 allows running two services within the same /tmp and network
2462 namespace, if PrivateNetwork= or PrivateTmp= are used.
2464 * A new command "cat" has been added to systemctl. It outputs
2465 the original unit file of a unit, and concatenates the
2466 contents of additional "drop-in" unit file snippets, so that
2467 the full configuration is shown.
2469 * systemctl now supports globbing on the various "list-xyz"
2470 commands, like "list-units" or "list-sockets", as well as on
2471 those commands which take multiple unit names.
2473 * journalctl's --unit= switch gained support for globbing.
2475 * All systemd daemons now make use of the watchdog logic so
2476 that systemd automatically notices when they hang.
2478 * If the $container_ttys environment variable is set,
2479 getty-generator will automatically spawn a getty for each
2480 listed tty. This is useful for container managers to request
2481 login gettys to be spawned on as many ttys as needed.
2483 * %h, %s, %U specifier support is not available anymore when
2484 used in unit files for PID 1. This is because NSS calls are
2485 not safe from PID 1. They stay available for --user
2486 instances of systemd, and as special case for the root user.
2488 * loginctl gained a new "--no-legend" switch to turn off output
2491 * The "sd-login.h" API gained three new calls:
2492 sd_session_is_remote(), sd_session_get_remote_user(),
2493 sd_session_get_remote_host() to query information about
2496 * The udev hardware database now also carries vendor/product
2497 information of SDIO devices.
2499 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
2500 determine whether watchdog notifications are requested by
2503 * Socket-activated per-connection services now include a
2504 short description of the connection parameters in the
2507 * tmpfiles gained a new "--boot" option. When this is not used,
2508 only lines where the command character is not suffixed with
2509 "!" are executed. When this option is specified, those
2510 options are executed too. This partitions tmpfiles
2511 directives into those that can be safely executed at any
2512 time, and those which should be run only at boot (for
2513 example, a line that creates /run/nologin).
2515 * A new API "sd-resolve.h" has been added which provides a simple
2516 asynchronous wrapper around glibc NSS host name resolution
2517 calls, such as getaddrinfo(). In contrast to glibc's
2518 getaddrinfo_a(), it does not use signals. In contrast to most
2519 other asynchronous name resolution libraries, this one does
2520 not reimplement DNS, but reuses NSS, so that alternate
2521 host name resolution systems continue to work, such as mDNS,
2522 LDAP, etc. This API is based on libasyncns, but it has been
2523 cleaned up for inclusion in systemd.
2525 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
2526 "sd-daemon.h" are no longer found in individual libraries
2527 libsystemd-journal.so, libsystemd-login.so,
2528 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
2529 merged them into a single library, libsystemd.so, which
2530 provides all symbols. The reason for this is cyclic
2531 dependencies, as these libraries tend to use each other's
2532 symbols. So far, we have managed to workaround that by linking
2533 a copy of a good part of our code into each of these
2534 libraries again and again, which, however, makes certain
2535 things hard to do, like sharing static variables. Also, it
2536 substantially increases footprint. With this change, there
2537 is only one library for the basic APIs systemd
2538 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
2539 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
2540 library as well, however are subject to the --enable-kdbus
2541 switch (see below). Note that "sd-dhcp-client.h" is not part
2542 of this library (this is because it only consumes, never
2543 provides, services of/to other APIs). To make the transition
2544 easy from the separate libraries to the unified one, we
2545 provide the --enable-compat-libs compile-time switch which
2546 will generate stub libraries that are compatible with the
2547 old ones but redirect all calls to the new one.
2549 * All of the kdbus logic and the new APIs "sd-bus.h",
2550 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
2551 and "sd-utf8.h" are compile-time optional via the
2552 "--enable-kdbus" switch, and they are not compiled in by
2553 default. To make use of kdbus, you have to explicitly enable
2554 the switch. Note however, that neither the kernel nor the
2555 userspace API for all of this is considered stable yet. We
2556 want to maintain the freedom to still change the APIs for
2557 now. By specifying this build-time switch, you acknowledge
2558 that you are aware of the instability of the current
2561 * Also, note that while kdbus is pretty much complete,
2562 it lacks one thing: proper policy support. This means you
2563 can build a fully working system with all features; however,
2564 it will be highly insecure. Policy support will be added in
2565 one of the next releases, at the same time that we will
2566 declare the APIs stable.
2568 * When the kernel command line argument "kdbus" is specified,
2569 systemd will automatically load the kdbus.ko kernel module. At
2570 this stage of development, it is only useful for testing kdbus
2571 and should not be used in production. Note: if "--enable-kdbus"
2572 is specified, and the kdbus.ko kernel module is available, and
2573 "kdbus" is added to the kernel command line, the entire system
2574 runs with kdbus instead of dbus-daemon, with the above mentioned
2575 problem of missing the system policy enforcement. Also a future
2576 version of kdbus.ko or a newer systemd will not be compatible with
2577 each other, and will unlikely be able to boot the machine if only
2578 one of them is updated.
2580 * systemctl gained a new "import-environment" command which
2581 uploads the caller's environment (or parts thereof) into the
2582 service manager so that it is inherited by services started
2583 by the manager. This is useful to upload variables like
2584 $DISPLAY into the user service manager.
2586 * A new PrivateDevices= switch has been added to service units
2587 which allows running a service with a namespaced /dev
2588 directory that does not contain any device nodes for
2589 physical devices. More specifically, it only includes devices
2590 such as /dev/null, /dev/urandom, and /dev/zero which are API
2593 * logind has been extended to support behaviour like VT
2594 switching on seats that do not support a VT. This makes
2595 multi-session available on seats that are not the first seat
2596 (seat0), and on systems where kernel support for VTs has
2597 been disabled at compile-time.
2599 * If a process holds a delay lock for system sleep or shutdown
2600 and fails to release it in time, we will now log its
2601 identity. This makes it easier to identify processes that
2602 cause slow suspends or power-offs.
2604 * When parsing /etc/crypttab, support for a new key-slot=
2605 option as supported by Debian is added. It allows indicating
2606 which LUKS slot to use on disk, speeding up key loading.
2608 * The sd_journald_sendv() API call has been checked and
2609 officially declared to be async-signal-safe so that it may
2610 be invoked from signal handlers for logging purposes.
2612 * Boot-time status output is now enabled automatically after a
2613 short timeout if boot does not progress, in order to give
2614 the user an indication what she or he is waiting for.
2616 * The boot-time output has been improved to show how much time
2617 remains until jobs expire.
2619 * The KillMode= switch in service units gained a new possible
2620 value "mixed". If set, and the unit is shut down, then the
2621 initial SIGTERM signal is sent only to the main daemon
2622 process, while the following SIGKILL signal is sent to
2623 all remaining processes of the service.
2625 * When a scope unit is registered, a new property "Controller"
2626 may be set. If set to a valid bus name, systemd will send a
2627 RequestStop() signal to this name when it would like to shut
2628 down the scope. This may be used to hook manager logic into
2629 the shutdown logic of scope units. Also, scope units may now
2630 be put in a special "abandoned" state, in which case the
2631 manager process which created them takes no further
2632 responsibilities for it.
2634 * When reading unit files, systemd will now verify
2635 the access mode of these files, and warn about certain
2636 suspicious combinations. This has been added to make it
2637 easier to track down packaging bugs where unit files are
2638 marked executable or world-writable.
2640 * systemd-nspawn gained a new "--setenv=" switch to set
2641 container-wide environment variables. The similar option in
2642 systemd-activate was renamed from "--environment=" to
2643 "--setenv=" for consistency.
2645 * systemd-nspawn has been updated to create a new kdbus domain
2646 for each container that is invoked, thus allowing each
2647 container to have its own set of system and user buses,
2648 independent of the host.
2650 * systemd-nspawn gained a new --drop-capability= switch to run
2651 the container with less capabilities than the default. Both
2652 --drop-capability= and --capability= now take the special
2653 string "all" for dropping or keeping all capabilities.
2655 * systemd-nspawn gained new switches for executing containers
2656 with specific SELinux labels set.
2658 * systemd-nspawn gained a new --quiet switch to not generate
2659 any additional output but the container's own console
2662 * systemd-nspawn gained a new --share-system switch to run a
2663 container without PID namespacing enabled.
2665 * systemd-nspawn gained a new --register= switch to control
2666 whether the container is registered with systemd-machined or
2667 not. This is useful for containers that do not run full
2668 OS images, but only specific apps.
2670 * systemd-nspawn gained a new --keep-unit which may be used
2671 when invoked as the only program from a service unit, and
2672 results in registration of the unit service itself in
2673 systemd-machined, instead of a newly opened scope unit.
2675 * systemd-nspawn gained a new --network-interface= switch for
2676 moving arbitrary interfaces to the container. The new
2677 --network-veth switch creates a virtual Ethernet connection
2678 between host and container. The new --network-bridge=
2679 switch then allows assigning the host side of this virtual
2680 Ethernet connection to a bridge device.
2682 * systemd-nspawn gained a new --personality= switch for
2683 setting the kernel personality for the container. This is
2684 useful when running a 32-bit container on a 64-bit host. A
2685 similar option Personality= is now also available for service
2688 * logind will now also track a "Desktop" identifier for each
2689 session which encodes the desktop environment of it. This is
2690 useful for desktop environments that want to identify
2691 multiple running sessions of itself easily.
2693 * A new SELinuxContext= setting for service units has been
2694 added that allows setting a specific SELinux execution
2695 context for a service.
2697 * Most systemd client tools will now honour $SYSTEMD_LESS for
2698 settings of the "less" pager. By default, these tools will
2699 override $LESS to allow certain operations to work, such as
2700 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
2701 influence this logic.
2703 * systemd's "seccomp" hook-up has been changed to make use of
2704 the libseccomp library instead of using its own
2705 implementation. This has benefits for portability among
2708 * For usage together with SystemCallFilter=, a new
2709 SystemCallErrorNumber= setting has been introduced that
2710 allows configuration of a system error number to be returned
2711 on filtered system calls, instead of immediately killing the
2712 process. Also, SystemCallArchitectures= has been added to
2713 limit access to system calls of a particular architecture
2714 (in order to turn off support for unused secondary
2715 architectures). There is also a global
2716 SystemCallArchitectures= setting in system.conf now to turn
2717 off support for non-native system calls system-wide.
2719 * systemd requires a kernel with a working name_to_handle_at(),
2720 please see the kernel config requirements in the README file.
2722 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
2723 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
2724 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
2725 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
2726 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
2727 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
2728 Elia Pinto, Florian Weimer, George McCollister, Goffredo
2729 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
2730 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
2731 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
2732 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
2733 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
2734 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
2735 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
2736 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
2737 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
2738 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
2739 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
2740 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
2741 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
2742 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
2743 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
2744 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
2745 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
2747 -- Berlin, 2014-02-20
2751 * logind has gained support for facilitating privileged input
2752 and drm device access for unprivileged clients. This work is
2753 useful to allow Wayland display servers (and similar
2754 programs, such as kmscon) to run under the user's ID and
2755 access input and drm devices which are normally
2756 protected. When this is used (and the kernel is new enough)
2757 logind will "mute" IO on the file descriptors passed to
2758 Wayland as long as it is in the background and "unmute" it
2759 if it returns into the foreground. This allows secure
2760 session switching without allowing background sessions to
2761 eavesdrop on input and display data. This also introduces
2762 session switching support if VT support is turned off in the
2763 kernel, and on seats that are not seat0.
2765 * A new kernel command line option luks.options= is understood
2766 now which allows specifying LUKS options for usage for LUKS
2767 encrypted partitions specified with luks.uuid=.
2769 * tmpfiles.d(5) snippets may now use specifier expansion in
2770 path names. More specifically %m, %b, %H, %v, are now
2771 replaced by the local machine id, boot id, hostname, and
2772 kernel version number.
2774 * A new tmpfiles.d(5) command "m" has been introduced which
2775 may be used to change the owner/group/access mode of a file
2776 or directory if it exists, but do nothing if it does not.
2778 * This release removes high-level support for the
2779 MemorySoftLimit= cgroup setting. The underlying kernel
2780 cgroup attribute memory.soft_limit= is currently badly
2781 designed and likely to be removed from the kernel API in its
2782 current form, hence we should not expose it for now.
2784 * The memory.use_hierarchy cgroup attribute is now enabled for
2785 all cgroups systemd creates in the memory cgroup
2786 hierarchy. This option is likely to be come the built-in
2787 default in the kernel anyway, and the non-hierarchical mode
2788 never made much sense in the intrinsically hierarchical
2791 * A new field _SYSTEMD_SLICE= is logged along with all journal
2792 messages containing the slice a message was generated
2793 from. This is useful to allow easy per-customer filtering of
2794 logs among other things.
2796 * systemd-journald will no longer adjust the group of journal
2797 files it creates to the "systemd-journal" group. Instead we
2798 rely on the journal directory to be owned by the
2799 "systemd-journal" group, and its setgid bit set, so that the
2800 kernel file system layer will automatically enforce that
2801 journal files inherit this group assignment. The reason for
2802 this change is that we cannot allow NSS look-ups from
2803 journald which would be necessary to resolve
2804 "systemd-journal" to a numeric GID, because this might
2805 create deadlocks if NSS involves synchronous queries to
2806 other daemons (such as nscd, or sssd) which in turn are
2807 logging clients of journald and might block on it, which
2808 would then dead lock. A tmpfiles.d(5) snippet included in
2809 systemd will make sure the setgid bit and group are
2810 properly set on the journal directory if it exists on every
2811 boot. However, we recommend adjusting it manually after
2812 upgrades too (or from RPM scriptlets), so that the change is
2813 not delayed until next reboot.
2815 * Backlight and random seed files in /var/lib/ have moved into
2816 the /var/lib/systemd/ directory, in order to centralize all
2817 systemd generated files in one directory.
2819 * Boot time performance measurements (as displayed by
2820 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
2821 performance information if that's available to determine how
2822 much time BIOS and boot loader initialization required. With
2823 a sufficiently new BIOS you hence no longer need to boot
2824 with Gummiboot to get access to such information.
2826 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
2827 Cristian Rodríguez, Dave Reisner, David Herrmann, David
2828 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
2829 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
2830 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
2831 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
2832 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
2834 -- Berlin, 2013-10-02
2838 * The Restart= option for services now understands a new
2839 on-watchdog setting, which will restart the service
2840 automatically if the service stops sending out watchdog keep
2841 alive messages (as configured with WatchdogSec=).
2843 * The getty generator (which is responsible for bringing up a
2844 getty on configured serial consoles) will no longer only
2845 start a getty on the primary kernel console but on all
2846 others, too. This makes the order in which console= is
2847 specified on the kernel command line less important.
2849 * libsystemd-logind gained a new sd_session_get_vt() call to
2850 retrieve the VT number of a session.
2852 * If the option "tries=0" is set for an entry of /etc/crypttab
2853 its passphrase is queried indefinitely instead of any
2854 maximum number of tries.
2856 * If a service with a configure PID file terminates its PID
2857 file will now be removed automatically if it still exists
2858 afterwards. This should put an end to stale PID files.
2860 * systemd-run will now also take relative binary path names
2861 for execution and no longer insists on absolute paths.
2863 * InaccessibleDirectories= and ReadOnlyDirectories= now take
2864 paths that are optionally prefixed with "-" to indicate that
2865 it should not be considered a failure if they do not exist.
2867 * journalctl -o (and similar commands) now understands a new
2868 output mode "short-precise", it is similar to "short" but
2869 shows timestamps with usec accuracy.
2871 * The option "discard" (as known from Debian) is now
2872 synonymous to "allow-discards" in /etc/crypttab. In fact,
2873 "discard" is preferred now (since it is easier to remember
2876 * Some licensing clean-ups were made, so that more code is now
2877 LGPL-2.1 licensed than before.
2879 * A minimal tool to save/restore the display backlight
2880 brightness across reboots has been added. It will store the
2881 backlight setting as late as possible at shutdown, and
2882 restore it as early as possible during reboot.
2884 * A logic to automatically discover and enable home and swap
2885 partitions on GPT disks has been added. With this in place
2886 /etc/fstab becomes optional for many setups as systemd can
2887 discover certain partitions located on the root disk
2888 automatically. Home partitions are recognized under their
2889 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
2890 partitions are recognized under their GPT type ID
2891 0657fd6da4ab43c484e50933c84b4f4f.
2893 * systemd will no longer pass any environment from the kernel
2894 or initrd to system services. If you want to set an
2895 environment for all services, do so via the kernel command
2896 line systemd.setenv= assignment.
2898 * The systemd-sysctl tool no longer natively reads the file
2899 /etc/sysctl.conf. If desired, the file should be symlinked
2900 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
2901 legacy support by a symlink rather than built-in code, it
2902 also makes the otherwise hidden order of application of the
2903 different files visible. (Note that this partly reverts to a
2904 pre-198 application order of sysctl knobs!)
2906 * The "systemctl set-log-level" and "systemctl dump" commands
2907 have been moved to systemd-analyze.
2909 * systemd-run learned the new --remain-after-exit switch,
2910 which causes the scope unit not to be cleaned up
2911 automatically after the process terminated.
2913 * tmpfiles learned a new --exclude-prefix= switch to exclude
2914 certain paths from operation.
2916 * journald will now automatically flush all messages to disk
2917 as soon as a message at the log level CRIT, ALERT or EMERG
2920 Contributions from: Andrew Cook, Brandon Philips, Christian
2921 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
2922 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
2923 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
2924 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
2925 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
2926 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
2927 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
2928 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
2929 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
2930 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
2931 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
2932 William Giokas, Zbigniew Jędrzejewski-Szmek
2934 -- Berlin, 2013-09-13
2938 * The documentation has been updated to cover the various new
2939 concepts introduced with 205.
2941 * Unit files now understand the new %v specifier which
2942 resolves to the kernel version string as returned by "uname
2945 * systemctl now supports filtering the unit list output by
2946 load state, active state and sub state, using the new
2949 * "systemctl status" will now show the results of the
2950 condition checks (like ConditionPathExists= and similar) of
2951 the last start attempts of the unit. They are also logged to
2954 * "journalctl -b" may now be used to look for boot output of a
2955 specific boot. Try "journalctl -b -1" for the previous boot,
2956 but the syntax is substantially more powerful.
2958 * "journalctl --show-cursor" has been added which prints the
2959 cursor string the last shown log line. This may then be used
2960 with the new "journalctl --after-cursor=" switch to continue
2961 browsing logs from that point on.
2963 * "journalctl --force" may now be used to force regeneration
2966 * Creation of "dead" device nodes has been moved from udev
2967 into kmod and tmpfiles. Previously, udev would read the kmod
2968 databases to pre-generate dead device nodes based on meta
2969 information contained in kernel modules, so that these would
2970 be auto-loaded on access rather then at boot. As this
2971 does not really have much to do with the exposing actual
2972 kernel devices to userspace this has always been slightly
2973 alien in the udev codebase. Following the new scheme kmod
2974 will now generate a runtime snippet for tmpfiles from the
2975 module meta information and it now is tmpfiles' job to the
2976 create the nodes. This also allows overriding access and
2977 other parameters for the nodes using the usual tmpfiles
2978 facilities. As side effect this allows us to remove the
2979 CAP_SYS_MKNOD capability bit from udevd entirely.
2981 * logind's device ACLs may now be applied to these "dead"
2982 devices nodes too, thus finally allowing managed access to
2983 devices such as /dev/snd/sequencer whithout loading the
2984 backing module right-away.
2986 * A new RPM macro has been added that may be used to apply
2987 tmpfiles configuration during package installation.
2989 * systemd-detect-virt and ConditionVirtualization= now can
2990 detect User-Mode-Linux machines (UML).
2992 * journald will now implicitly log the effective capabilities
2993 set of processes in the message metadata.
2995 * systemd-cryptsetup has gained support for TrueCrypt volumes.
2997 * The initrd interface has been simplified (more specifically,
2998 support for passing performance data via environment
2999 variables and fsck results via files in /run has been
3000 removed). These features were non-essential, and are
3001 nowadays available in a much nicer way by having systemd in
3002 the initrd serialize its state and have the hosts systemd
3003 deserialize it again.
3005 * The udev "keymap" data files and tools to apply keyboard
3006 specific mappings of scan to key codes, and force-release
3007 scan code lists have been entirely replaced by a udev
3008 "keyboard" builtin and a hwdb data file.
3010 * systemd will now honour the kernel's "quiet" command line
3011 argument also during late shutdown, resulting in a
3012 completely silent shutdown when used.
3014 * There's now an option to control the SO_REUSEPORT socket
3015 option in .socket units.
3017 * Instance units will now automatically get a per-template
3018 subslice of system.slice unless something else is explicitly
3019 configured. For example, instances of sshd@.service will now
3020 implicitly be placed in system-sshd.slice rather than
3021 system.slice as before.
3023 * Test coverage support may now be enabled at build time.
3025 Contributions from: Dave Reisner, Frederic Crozat, Harald
3026 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
3027 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
3028 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
3029 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
3030 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
3031 Giokas, Zbigniew Jędrzejewski-Szmek
3033 -- Berlin, 2013-07-23
3037 * Two new unit types have been introduced:
3039 Scope units are very similar to service units, however, are
3040 created out of pre-existing processes -- instead of PID 1
3041 forking off the processes. By using scope units it is
3042 possible for system services and applications to group their
3043 own child processes (worker processes) in a powerful way
3044 which then maybe used to organize them, or kill them
3045 together, or apply resource limits on them.
3047 Slice units may be used to partition system resources in an
3048 hierarchical fashion and then assign other units to them. By
3049 default there are now three slices: system.slice (for all
3050 system services), user.slice (for all user sessions),
3051 machine.slice (for VMs and containers).
3053 Slices and scopes have been introduced primarily in
3054 context of the work to move cgroup handling to a
3055 single-writer scheme, where only PID 1
3056 creates/removes/manages cgroups.
3058 * There's a new concept of "transient" units. In contrast to
3059 normal units these units are created via an API at runtime,
3060 not from configuration from disk. More specifically this
3061 means it is now possible to run arbitrary programs as
3062 independent services, with all execution parameters passed
3063 in via bus APIs rather than read from disk. Transient units
3064 make systemd substantially more dynamic then it ever was,
3065 and useful as a general batch manager.
3067 * logind has been updated to make use of scope and slice units
3068 for managing user sessions. As a user logs in he will get
3069 his own private slice unit, to which all sessions are added
3070 as scope units. We also added support for automatically
3071 adding an instance of user@.service for the user into the
3072 slice. Effectively logind will no longer create cgroup
3073 hierarchies on its own now, it will defer entirely to PID 1
3074 for this by means of scope, service and slice units. Since
3075 user sessions this way become entities managed by PID 1
3076 the output of "systemctl" is now a lot more comprehensive.
3078 * A new mini-daemon "systemd-machined" has been added which
3079 may be used by virtualization managers to register local
3080 VMs/containers. nspawn has been updated accordingly, and
3081 libvirt will be updated shortly. machined will collect a bit
3082 of meta information about the VMs/containers, and assign
3083 them their own scope unit (see above). The collected
3084 meta-data is then made available via the "machinectl" tool,
3085 and exposed in "ps" and similar tools. machined/machinectl
3086 is compile-time optional.
3088 * As discussed earlier, the low-level cgroup configuration
3089 options ControlGroup=, ControlGroupModify=,
3090 ControlGroupPersistent=, ControlGroupAttribute= have been
3091 removed. Please use high-level attribute settings instead as
3092 well as slice units.
3094 * A new bus call SetUnitProperties() has been added to alter
3095 various runtime parameters of a unit. This is primarily
3096 useful to alter cgroup parameters dynamically in a nice way,
3097 but will be extended later on to make more properties
3098 modifiable at runtime. systemctl gained a new set-properties
3099 command that wraps this call.
3101 * A new tool "systemd-run" has been added which can be used to
3102 run arbitrary command lines as transient services or scopes,
3103 while configuring a number of settings via the command
3104 line. This tool is currently very basic, however already
3105 very useful. We plan to extend this tool to even allow
3106 queuing of execution jobs with time triggers from the
3107 command line, similar in fashion to "at".
3109 * nspawn will now inform the user explicitly that kernels with
3110 audit enabled break containers, and suggest the user to turn
3113 * Support for detecting the IMA and AppArmor security
3114 frameworks with ConditionSecurity= has been added.
3116 * journalctl gained a new "-k" switch for showing only kernel
3117 messages, mimicking dmesg output; in addition to "--user"
3118 and "--system" switches for showing only user's own logs
3121 * systemd-delta can now show information about drop-in
3122 snippets extending unit files.
3124 * libsystemd-bus has been substantially updated but is still
3125 not available as public API.
3127 * systemd will now look for the "debug" argument on the kernel
3128 command line and enable debug logging, similar to what
3129 "systemd.log_level=debug" already did before.
3131 * "systemctl set-default", "systemctl get-default" has been
3132 added to configure the default.target symlink, which
3133 controls what to boot into by default.
3135 * "systemctl set-log-level" has been added as a convenient
3136 way to raise and lower systemd logging threshold.
3138 * "systemd-analyze plot" will now show the time the various
3139 generators needed for execution, as well as information
3140 about the unit file loading.
3142 * libsystemd-journal gained a new sd_journal_open_files() call
3143 for opening specific journal files. journactl also gained a
3144 new switch to expose this new functionality. Previously we
3145 only supported opening all files from a directory, or all
3146 files from the system, as opening individual files only is
3147 racy due to journal file rotation.
3149 * systemd gained the new DefaultEnvironment= setting in
3150 /etc/systemd/system.conf to set environment variables for
3153 * If a privileged process logs a journal message with the
3154 OBJECT_PID= field set, then journald will automatically
3155 augment this with additional OBJECT_UID=, OBJECT_GID=,
3156 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
3157 system services want to log events about specific client
3158 processes. journactl/systemctl has been updated to make use
3159 of this information if all log messages regarding a specific
3162 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
3163 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
3164 Reisner, David Coppa, David King, David Strauss, Eelco
3165 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
3166 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
3167 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
3168 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
3169 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
3170 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
3171 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
3172 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
3173 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
3174 Łukasz Stelmach, 장동준
3178 * The Python bindings gained some minimal support for the APIs
3179 exposed by libsystemd-logind.
3181 * ConditionSecurity= gained support for detecting SMACK. Since
3182 this condition already supports SELinux and AppArmor we only
3183 miss IMA for this. Patches welcome!
3185 Contributions from: Karol Lewandowski, Lennart Poettering,
3186 Zbigniew Jędrzejewski-Szmek
3190 * systemd-nspawn will now create /etc/resolv.conf if
3191 necessary, before bind-mounting the host's file onto it.
3193 * systemd-nspawn will now store meta information about a
3194 container on the container's cgroup as extended attribute
3195 fields, including the root directory.
3197 * The cgroup hierarchy has been reworked in many ways. All
3198 objects any of the components systemd creates in the cgroup
3199 tree are now suffixed. More specifically, user sessions are
3200 now placed in cgroups suffixed with ".session", users in
3201 cgroups suffixed with ".user", and nspawn containers in
3202 cgroups suffixed with ".nspawn". Furthermore, all cgroup
3203 names are now escaped in a simple scheme to avoid collision
3204 of userspace object names with kernel filenames. This work
3205 is preparation for making these objects relocatable in the
3206 cgroup tree, in order to allow easy resource partitioning of
3207 these objects without causing naming conflicts.
3209 * systemctl list-dependencies gained the new switches
3210 --plain, --reverse, --after and --before.
3212 * systemd-inhibit now shows the process name of processes that
3213 have taken an inhibitor lock.
3215 * nss-myhostname will now also resolve "localhost"
3216 implicitly. This makes /etc/hosts an optional file and
3217 nicely handles that on IPv6 ::1 maps to both "localhost" and
3220 * libsystemd-logind.so gained a new call
3221 sd_get_machine_names() to enumerate running containers and
3222 VMs (currently only supported by very new libvirt and
3223 nspawn). sd_login_monitor can now be used to watch
3224 VMs/containers coming and going.
3226 * .include is not allowed recursively anymore, and only in
3227 unit files. Usually it is better to use drop-in snippets in
3228 .d/*.conf anyway, as introduced with systemd 198.
3230 * systemd-analyze gained a new "critical-chain" command that
3231 determines the slowest chain of units run during system
3232 boot-up. It is very useful for tracking down where
3233 optimizing boot time is the most beneficial.
3235 * systemd will no longer allow manipulating service paths in
3236 the name=systemd:/system cgroup tree using ControlGroup= in
3237 units. (But is still fine with it in all other dirs.)
3239 * There's a new systemd-nspawn@.service service file that may
3240 be used to easily run nspawn containers as system
3241 services. With the container's root directory in
3242 /var/lib/container/foobar it is now sufficient to run
3243 "systemctl start systemd-nspawn@foobar.service" to boot it.
3245 * systemd-cgls gained a new parameter "--machine" to list only
3246 the processes within a certain container.
3248 * ConditionSecurity= now can check for "apparmor". We still
3249 are lacking checks for SMACK and IMA for this condition
3250 check though. Patches welcome!
3252 * A new configuration file /etc/systemd/sleep.conf has been
3253 added that may be used to configure which kernel operation
3254 systemd is supposed to execute when "suspend", "hibernate"
3255 or "hybrid-sleep" is requested. This makes the new kernel
3256 "freeze" state accessible to the user.
3258 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
3259 the passed argument if applicable.
3261 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
3262 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
3263 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
3264 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
3265 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
3266 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
3267 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
3272 * The output of 'systemctl list-jobs' got some polishing. The
3273 '--type=' argument may now be passed more than once. A new
3274 command 'systemctl list-sockets' has been added which shows
3275 a list of kernel sockets systemd is listening on with the
3276 socket units they belong to, plus the units these socket
3279 * The experimental libsystemd-bus library got substantial
3280 updates to work in conjunction with the (also experimental)
3281 kdbus kernel project. It works well enough to exchange
3282 messages with some sophistication. Note that kdbus is not
3283 ready yet, and the library is mostly an elaborate test case
3284 for now, and not installable.
3286 * systemd gained a new unit 'systemd-static-nodes.service'
3287 that generates static device nodes earlier during boot, and
3288 can run in conjunction with udev.
3290 * libsystemd-login gained a new call sd_pid_get_user_unit()
3291 to retrieve the user systemd unit a process is running
3292 in. This is useful for systems where systemd is used as
3295 * systemd-nspawn now places all containers in the new /machine
3296 top-level cgroup directory in the name=systemd
3297 hierarchy. libvirt will soon do the same, so that we get a
3298 uniform separation of /system, /user and /machine for system
3299 services, user processes and containers/virtual
3300 machines. This new cgroup hierarchy is also useful to stick
3301 stable names to specific container instances, which can be
3302 recognized later this way (this name may be controlled
3303 via systemd-nspawn's new -M switch). libsystemd-login also
3304 gained a new call sd_pid_get_machine_name() to retrieve the
3305 name of the container/VM a specific process belongs to.
3307 * bootchart can now store its data in the journal.
3309 * libsystemd-journal gained a new call
3310 sd_journal_add_conjunction() for AND expressions to the
3311 matching logic. This can be used to express more complex
3312 logical expressions.
3314 * journactl can now take multiple --unit= and --user-unit=
3317 * The cryptsetup logic now understands the "luks.key=" kernel
3318 command line switch for specifying a file to read the
3319 decryption key from. Also, if a configured key file is not
3320 found the tool will now automatically fall back to prompting
3323 * Python systemd.journal module was updated to wrap recently
3324 added functions from libsystemd-journal. The interface was
3325 changed to bring the low level interface in s.j._Reader
3326 closer to the C API, and the high level interface in
3327 s.j.Reader was updated to wrap and convert all data about
3330 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
3331 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
3332 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
3333 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
3334 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
3335 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
3339 * journalctl --update-catalog now understands a new --root=
3340 option to operate on catalogs found in a different root
3343 * During shutdown after systemd has terminated all running
3344 services a final killing loop kills all remaining left-over
3345 processes. We will now print the name of these processes
3346 when we send SIGKILL to them, since this usually indicates a
3349 * If /etc/crypttab refers to password files stored on
3350 configured mount points automatic dependencies will now be
3351 generated to ensure the specific mount is established first
3352 before the key file is attempted to be read.
3354 * 'systemctl status' will now show information about the
3355 network sockets a socket unit is listening on.
3357 * 'systemctl status' will also shown information about any
3358 drop-in configuration file for units. (Drop-In configuration
3359 files in this context are files such as
3360 /etc/systemd/systemd/foobar.service.d/*.conf)
3362 * systemd-cgtop now optionally shows summed up CPU times of
3363 cgroups. Press '%' while running cgtop to switch between
3364 percentage and absolute mode. This is useful to determine
3365 which cgroups use up the most CPU time over the entire
3366 runtime of the system. systemd-cgtop has also been updated
3367 to be 'pipeable' for processing with further shell tools.
3369 * 'hostnamectl set-hostname' will now allow setting of FQDN
3372 * The formatting and parsing of time span values has been
3373 changed. The parser now understands fractional expressions
3374 such as "5.5h". The formatter will now output fractional
3375 expressions for all time spans under 1min, i.e. "5.123456s"
3376 rather than "5s 123ms 456us". For time spans under 1s
3377 millisecond values are shown, for those under 1ms
3378 microsecond values are shown. This should greatly improve
3379 all time-related output of systemd.
3381 * libsystemd-login and libsystemd-journal gained new
3382 functions for querying the poll() events mask and poll()
3383 timeout value for integration into arbitrary event
3386 * localectl gained the ability to list available X11 keymaps
3387 (models, layouts, variants, options).
3389 * 'systemd-analyze dot' gained the ability to filter for
3390 specific units via shell-style globs, to create smaller,
3391 more useful graphs. I.e. it is now possible to create simple
3392 graphs of all the dependencies between only target units, or
3393 of all units that Avahi has dependencies with.
3395 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
3396 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
3397 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
3398 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
3399 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
3400 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
3401 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
3405 * The boot-time readahead implementation for rotating media
3406 will now read the read-ahead data in multiple passes which
3407 consist of all read requests made in equidistant time
3408 intervals. This means instead of strictly reading read-ahead
3409 data in its physical order on disk we now try to find a
3410 middle ground between physical and access time order.
3412 * /etc/os-release files gained a new BUILD_ID= field for usage
3413 on operating systems that provide continuous builds of OS
3416 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
3417 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
3418 William Douglas, Zbigniew Jędrzejewski-Szmek
3422 * systemd-python gained an API exposing libsystemd-daemon.
3424 * The SMACK setup logic gained support for uploading CIPSO
3427 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
3428 ReadOnlyDirectories= and InaccessibleDirectories= has
3429 changed. The private /tmp and /var/tmp directories are now
3430 shared by all processes of a service (which means
3431 ExecStartPre= may now leave data in /tmp that ExecStart= of
3432 the same service can still access). When a service is
3433 stopped its temporary directories are immediately deleted
3434 (normal clean-up with tmpfiles is still done in addition to
3437 * By default, systemd will now set a couple of sysctl
3438 variables in the kernel: the safe sysrq options are turned
3439 on, IP route verification is turned on, and source routing
3440 disabled. The recently added hardlink and softlink
3441 protection of the kernel is turned on. These settings should
3442 be reasonably safe, and good defaults for all new systems.
3444 * The predictable network naming logic may now be turned off
3445 with a new kernel command line switch: net.ifnames=0.
3447 * A new libsystemd-bus module has been added that implements a
3448 pretty complete D-Bus client library. For details see:
3450 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
3452 * journald will now explicitly flush the journal files to disk
3453 at the latest 5min after each write. The file will then also
3454 be marked offline until the next write. This should increase
3455 reliability in case of a crash. The synchronization delay
3456 can be configured via SyncIntervalSec= in journald.conf.
3458 * There's a new remote-fs-setup.target unit that can be used
3459 to pull in specific services when at least one remote file
3460 system is to be mounted.
3462 * There are new targets timers.target and paths.target as
3463 canonical targets to pull user timer and path units in
3464 from. This complements sockets.target with a similar
3465 purpose for socket units.
3467 * libudev gained a new call udev_device_set_attribute_value()
3468 to set sysfs attributes of a device.
3470 * The udev daemon now sets the default number of worker
3471 processes executed in parallel based on the number of available
3472 CPUs instead of the amount of available RAM. This is supposed
3473 to provide a more reliable default and limit a too aggressive
3474 paralellism for setups with 1000s of devices connected.
3476 Contributions from: Auke Kok, Colin Walters, Cristian
3477 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
3478 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
3479 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
3480 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
3481 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
3482 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
3483 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
3484 Zbigniew Jędrzejewski-Szmek
3488 * Configuration of unit files may now be extended via drop-in
3489 files without having to edit/override the unit files
3490 themselves. More specifically, if the administrator wants to
3491 change one value for a service file foobar.service he can
3492 now do so by dropping in a configuration snippet into
3493 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
3494 will load all these snippets and apply them on top of the
3495 main unit configuration file, possibly extending or
3496 overriding its settings. Using these drop-in snippets is
3497 generally nicer than the two earlier options for changing
3498 unit files locally: copying the files from
3499 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
3500 them there; or creating a new file in /etc/systemd/system/
3501 that incorporates the original one via ".include". Drop-in
3502 snippets into these .d/ directories can be placed in any
3503 directory systemd looks for units in, and the usual
3504 overriding semantics between /usr/lib, /etc and /run apply
3507 * Most unit file settings which take lists of items can now be
3508 reset by assigning the empty string to them. For example,
3509 normally, settings such as Environment=FOO=BAR append a new
3510 environment variable assignment to the environment block,
3511 each time they are used. By assigning Environment= the empty
3512 string the environment block can be reset to empty. This is
3513 particularly useful with the .d/*.conf drop-in snippets
3514 mentioned above, since this adds the ability to reset list
3515 settings from vendor unit files via these drop-ins.
3517 * systemctl gained a new "list-dependencies" command for
3518 listing the dependencies of a unit recursively.
3520 * Inhibitors are now honored and listed by "systemctl
3521 suspend", "systemctl poweroff" (and similar) too, not only
3522 GNOME. These commands will also list active sessions by
3525 * Resource limits (as exposed by the various control group
3526 controllers) can now be controlled dynamically at runtime
3527 for all units. More specifically, you can now use a command
3528 like "systemctl set-cgroup-attr foobar.service cpu.shares
3529 2000" to alter the CPU shares a specific service gets. These
3530 settings are stored persistently on disk, and thus allow the
3531 administrator to easily adjust the resource usage of
3532 services with a few simple commands. This dynamic resource
3533 management logic is also available to other programs via the
3534 bus. Almost any kernel cgroup attribute and controller is
3537 * systemd-vconsole-setup will now copy all font settings to
3538 all allocated VTs, where it previously applied them only to
3541 * libsystemd-login gained the new sd_session_get_tty() API
3544 * This release drops support for a few legacy or
3545 distribution-specific LSB facility names when parsing init
3546 scripts: $x-display-manager, $mail-transfer-agent,
3547 $mail-transport-agent, $mail-transfer-agent, $smtp,
3548 $null. Also, the mail-transfer-agent.target unit backing
3549 this has been removed. Distributions which want to retain
3550 compatibility with this should carry the burden for
3551 supporting this themselves and patch support for these back
3552 in, if they really need to. Also, the facilities $syslog and
3553 $local_fs are now ignored, since systemd does not support
3554 early-boot LSB init scripts anymore, and these facilities
3555 are implied anyway for normal services. syslog.target has
3558 * There are new bus calls on PID1's Manager object for
3559 cancelling jobs, and removing snapshot units. Previously,
3560 both calls were only available on the Job and Snapshot
3563 * systemd-journal-gatewayd gained SSL support.
3565 * The various "environment" files, such as /etc/locale.conf
3566 now support continuation lines with a backslash ("\") as
3567 last character in the line, similarly in style (but different)
3568 to how this is supported in shells.
3570 * For normal user processes the _SYSTEMD_USER_UNIT= field is
3571 now implicitly appended to every log entry logged. systemctl
3572 has been updated to filter by this field when operating on a
3573 user systemd instance.
3575 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
3576 CAP_AUDIT_CONTROL capabilities to the capabilities set for
3577 the container. This makes it easier to boot unmodified
3578 Fedora systems in a container, which however still requires
3579 audit=0 to be passed on the kernel command line. Auditing in
3580 kernel and userspace is unfortunately still too broken in
3581 context of containers, hence we recommend compiling it out
3582 of the kernel or using audit=0. Hopefully this will be fixed
3583 one day for good in the kernel.
3585 * nspawn gained the new --bind= and --bind-ro= parameters to
3586 bind mount specific directories from the host into the
3589 * nspawn will now mount its own devpts file system instance
3590 into the container, in order not to leak pty devices from
3591 the host into the container.
3593 * systemd will now read the firmware boot time performance
3594 information from the EFI variables, if the used boot loader
3595 supports this, and takes it into account for boot performance
3596 analysis via "systemd-analyze". This is currently supported
3597 only in conjunction with Gummiboot, but could be supported
3598 by other boot loaders too. For details see:
3600 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
3602 * A new generator has been added that automatically mounts the
3603 EFI System Partition (ESP) to /boot, if that directory
3604 exists, is empty, and no other file system has been
3605 configured to be mounted there.
3607 * logind will now send out PrepareForSleep(false) out
3608 unconditionally, after coming back from suspend. This may be
3609 used by applications as asynchronous notification for
3610 system resume events.
3612 * "systemctl unlock-sessions" has been added, that allows
3613 unlocking the screens of all user sessions at once, similar
3614 to how "systemctl lock-sessions" already locked all users
3615 sessions. This is backed by a new D-Bus call UnlockSessions().
3617 * "loginctl seat-status" will now show the master device of a
3618 seat. (i.e. the device of a seat that needs to be around for
3619 the seat to be considered available, usually the graphics
3622 * tmpfiles gained a new "X" line type, that allows
3623 configuration of files and directories (with wildcards) that
3624 shall be excluded from automatic cleanup ("aging").
3626 * udev default rules set the device node permissions now only
3627 at "add" events, and do not change them any longer with a
3628 later "change" event.
3630 * The log messages for lid events and power/sleep keypresses
3631 now carry a message ID.
3633 * We now have a substantially larger unit test suite, but this
3634 continues to be work in progress.
3636 * udevadm hwdb gained a new --root= parameter to change the
3637 root directory to operate relative to.
3639 * logind will now issue a background sync() request to the kernel
3640 early at shutdown, so that dirty buffers are flushed to disk early
3641 instead of at the last moment, in order to optimize shutdown
3644 * A new bootctl tool has been added that is an interface for
3645 certain boot loader operations. This is currently a preview
3646 and is likely to be extended into a small mechanism daemon
3647 like timedated, localed, hostnamed, and can be used by
3648 graphical UIs to enumerate available boot options, and
3649 request boot into firmware operations.
3651 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
3652 the rest of the package. It also has been updated to work
3653 correctly in initrds.
3655 * Policykit previously has been runtime optional, and is now
3656 also compile time optional via a configure switch.
3658 * systemd-analyze has been reimplemented in C. Also "systemctl
3659 dot" has moved into systemd-analyze.
3661 * "systemctl status" with no further parameters will now print
3662 the status of all active or failed units.
3664 * Operations such as "systemctl start" can now be executed
3665 with a new mode "--irreversible" which may be used to queue
3666 operations that cannot accidentally be reversed by a later
3667 job queuing. This is by default used to make shutdown
3668 requests more robust.
3670 * The Python API of systemd now gained a new module for
3671 reading journal files.
3673 * A new tool kernel-install has been added that can install
3674 kernel images according to the Boot Loader Specification:
3676 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
3678 * Boot time console output has been improved to provide
3679 animated boot time output for hanging jobs.
3681 * A new tool systemd-activate has been added which can be used
3682 to test socket activation with, directly from the command
3683 line. This should make it much easier to test and debug
3684 socket activation in daemons.
3686 * journalctl gained a new "--reverse" (or -r) option to show
3687 journal output in reverse order (i.e. newest line first).
3689 * journalctl gained a new "--pager-end" (or -e) option to jump
3690 to immediately jump to the end of the journal in the
3691 pager. This is only supported in conjunction with "less".
3693 * journalctl gained a new "--user-unit=" option, that works
3694 similarly to "--unit=" but filters for user units rather than
3697 * A number of unit files to ease adoption of systemd in
3698 initrds has been added. This moves some minimal logic from
3699 the various initrd implementations into systemd proper.
3701 * The journal files are now owned by a new group
3702 "systemd-journal", which exists specifically to allow access
3703 to the journal, and nothing else. Previously, we used the
3704 "adm" group for that, which however possibly covers more
3705 than just journal/log file access. This new group is now
3706 already used by systemd-journal-gatewayd to ensure this
3707 daemon gets access to the journal files and as little else
3708 as possible. Note that "make install" will also set FS ACLs
3709 up for /var/log/journal to give "adm" and "wheel" read
3710 access to it, in addition to "systemd-journal" which owns
3711 the journal files. We recommend that packaging scripts also
3712 add read access to "adm" + "wheel" to /var/log/journal, and
3713 all existing/future journal files. To normal users and
3714 administrators little changes, however packagers need to
3715 ensure to create the "systemd-journal" system group at
3716 package installation time.
3718 * The systemd-journal-gatewayd now runs as unprivileged user
3719 systemd-journal-gateway:systemd-journal-gateway. Packaging
3720 scripts need to create these system user/group at
3723 * timedated now exposes a new boolean property CanNTP that
3724 indicates whether a local NTP service is available or not.
3726 * systemd-detect-virt will now also detect xen PVs
3728 * The pstore file system is now mounted by default, if it is
3731 * In addition to the SELinux and IMA policies we will now also
3732 load SMACK policies at early boot.
3734 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
3735 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
3736 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
3737 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
3738 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
3739 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
3740 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
3741 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
3742 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
3743 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
3744 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
3745 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
3746 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
3747 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
3751 * Timer units now support calendar time events in addition to
3752 monotonic time events. That means you can now trigger a unit
3753 based on a calendar time specification such as "Thu,Fri
3754 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
3755 or fifth day of any month of the year 2013, given that it is
3756 a thursday or friday. This brings timer event support
3757 considerably closer to cron's capabilities. For details on
3758 the supported calendar time specification language see
3761 * udev now supports a number of different naming policies for
3762 network interfaces for predictable names, and a combination
3763 of these policies is now the default. Please see this wiki
3764 document for details:
3766 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
3768 * Auke Kok's bootchart implementation has been added to the
3769 systemd tree. It is an optional component that can graph the
3770 boot in quite some detail. It is one of the best bootchart
3771 implementations around and minimal in its code and
3774 * nss-myhostname has been integrated into the systemd source
3775 tree. nss-myhostname guarantees that the local hostname
3776 always stays resolvable via NSS. It has been a weak
3777 requirement of systemd-hostnamed since a long time, and
3778 since its code is actually trivial we decided to just
3779 include it in systemd's source tree. It can be turned off
3780 with a configure switch.
3782 * The read-ahead logic is now capable of properly detecting
3783 whether a btrfs file system is on SSD or rotating media, in
3784 order to optimize the read-ahead scheme. Previously, it was
3785 only capable of detecting this on traditional file systems
3788 * In udev, additional device properties are now read from the
3789 IAB in addition to the OUI database. Also, Bluetooth company
3790 identities are attached to the devices as well.
3792 * In service files %U may be used as specifier that is
3793 replaced by the configured user name of the service.
3795 * nspawn may now be invoked without a controlling TTY. This
3796 makes it suitable for invocation as its own service. This
3797 may be used to set up a simple containerized server system
3798 using only core OS tools.
3800 * systemd and nspawn can now accept socket file descriptors
3801 when they are started for socket activation. This enables
3802 implementation of socket activated nspawn
3803 containers. i.e. think about autospawning an entire OS image
3804 when the first SSH or HTTP connection is received. We expect
3805 that similar functionality will also be added to libvirt-lxc
3808 * journalctl will now suppress ANSI color codes when
3809 presenting log data.
3811 * systemctl will no longer show control group information for
3812 a unit if a the control group is empty anyway.
3814 * logind can now automatically suspend/hibernate/shutdown the
3817 * /etc/machine-info and hostnamed now also expose the chassis
3818 type of the system. This can be used to determine whether
3819 the local system is a laptop, desktop, handset or
3820 tablet. This information may either be configured by the
3821 user/vendor or is automatically determined from ACPI and DMI
3822 information if possible.
3824 * A number of PolicyKit actions are now bound together with
3825 "imply" rules. This should simplify creating UIs because
3826 many actions will now authenticate similar ones as well.
3828 * Unit files learnt a new condition ConditionACPower= which
3829 may be used to conditionalize a unit depending on whether an
3830 AC power source is connected or not, of whether the system
3831 is running on battery power.
3833 * systemctl gained a new "is-failed" verb that may be used in
3834 shell scripts and suchlike to check whether a specific unit
3835 is in the "failed" state.
3837 * The EnvironmentFile= setting in unit files now supports file
3838 globbing, and can hence be used to easily read a number of
3839 environment files at once.
3841 * systemd will no longer detect and recognize specific
3842 distributions. All distribution-specific #ifdeffery has been
3843 removed, systemd is now fully generic and
3844 distribution-agnostic. Effectively, not too much is lost as
3845 a lot of the code is still accessible via explicit configure
3846 switches. However, support for some distribution specific
3847 legacy configuration file formats has been dropped. We
3848 recommend distributions to simply adopt the configuration
3849 files everybody else uses now and convert the old
3850 configuration from packaging scripts. Most distributions
3851 already did that. If that's not possible or desirable,
3852 distributions are welcome to forward port the specific
3853 pieces of code locally from the git history.
3855 * When logging a message about a unit systemd will now always
3856 log the unit name in the message meta data.
3858 * localectl will now also discover system locale data that is
3859 not stored in locale archives, but directly unpacked.
3861 * logind will no longer unconditionally use framebuffer
3862 devices as seat masters, i.e. as devices that are required
3863 to be existing before a seat is considered preset. Instead,
3864 it will now look for all devices that are tagged as
3865 "seat-master" in udev. By default framebuffer devices will
3866 be marked as such, but depending on local systems other
3867 devices might be marked as well. This may be used to
3868 integrate graphics cards using closed source drivers (such
3869 as NVidia ones) more nicely into logind. Note however, that
3870 we recommend using the open source NVidia drivers instead,
3871 and no udev rules for the closed-source drivers will be
3872 shipped from us upstream.
3874 Contributions from: Adam Williamson, Alessandro Crismani, Auke
3875 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
3876 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
3877 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
3878 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
3879 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
3880 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
3881 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
3882 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
3883 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
3884 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
3885 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
3890 * udev gained support for loading additional device properties
3891 from an indexed database that is keyed by vendor/product IDs
3892 and similar device identifiers. For the beginning this
3893 "hwdb" is populated with data from the well-known PCI and
3894 USB database, but also includes PNP, ACPI and OID data. In
3895 the longer run this indexed database shall grow into
3896 becoming the one central database for non-essential
3897 userspace device metadata. Previously, data from the PCI/USB
3898 database was only attached to select devices, since the
3899 lookup was a relatively expensive operation due to O(n) time
3900 complexity (with n being the number of entries in the
3901 database). Since this is now O(1), we decided to add in this
3902 data for all devices where this is available, by
3903 default. Note that the indexed database needs to be rebuilt
3904 when new data files are installed. To achieve this you need
3905 to update your packaging scripts to invoke "udevadm hwdb
3906 --update" after installation of hwdb data files. For
3907 RPM-based distributions we introduced the new
3908 %udev_hwdb_update macro for this purpose.
3910 * The Journal gained support for the "Message Catalog", an
3911 indexed database to link up additional information with
3912 journal entries. For further details please check:
3914 http://www.freedesktop.org/wiki/Software/systemd/catalog
3916 The indexed message catalog database also needs to be
3917 rebuilt after installation of message catalog files. Use
3918 "journalctl --update-catalog" for this. For RPM-based
3919 distributions we introduced the %journal_catalog_update
3920 macro for this purpose.
3922 * The Python Journal bindings gained support for the standard
3923 Python logging framework.
3925 * The Journal API gained new functions for checking whether
3926 the underlying file system of a journal file is capable of
3927 properly reporting file change notifications, or whether
3928 applications that want to reflect journal changes "live"
3929 need to recheck journal files continuously in appropriate
3932 * It is now possible to set the "age" field for tmpfiles
3933 entries to 0, indicating that files matching this entry
3934 shall always be removed when the directories are cleaned up.
3936 * coredumpctl gained a new "gdb" verb which invokes gdb
3937 right-away on the selected coredump.
3939 * There's now support for "hybrid sleep" on kernels that
3940 support this, in addition to "suspend" and "hibernate". Use
3941 "systemctl hybrid-sleep" to make use of this.
3943 * logind's HandleSuspendKey= setting (and related settings)
3944 now gained support for a new "lock" setting to simply
3945 request the screen lock on all local sessions, instead of
3946 actually executing a suspend or hibernation.
3948 * systemd will now mount the EFI variables file system by
3951 * Socket units now gained support for configuration of the
3952 SMACK security label.
3954 * timedatectl will now output the time of the last and next
3955 daylight saving change.
3957 * We dropped support for various legacy and distro-specific
3958 concepts, such as insserv, early-boot SysV services
3959 (i.e. those for non-standard runlevels such as 'b' or 'S')
3960 or ArchLinux /etc/rc.conf support. We recommend the
3961 distributions who still need support this to either continue
3962 to maintain the necessary patches downstream, or find a
3963 different solution. (Talk to us if you have questions!)
3965 * Various systemd components will now bypass PolicyKit checks
3966 for root and otherwise handle properly if PolicyKit is not
3967 found to be around. This should fix most issues for
3968 PolicyKit-less systems. Quite frankly this should have been
3969 this way since day one. It is absolutely our intention to
3970 make systemd work fine on PolicyKit-less systems, and we
3971 consider it a bug if something does not work as it should if
3972 PolicyKit is not around.
3974 * For embedded systems it is now possible to build udev and
3975 systemd without blkid and/or kmod support.
3977 * "systemctl switch-root" is now capable of switching root
3978 more than once. I.e. in addition to transitions from the
3979 initrd to the host OS it is now possible to transition to
3980 further OS images from the host. This is useful to implement
3981 offline updating tools.
3983 * Various other additions have been made to the RPM macros
3984 shipped with systemd. Use %udev_rules_update() after
3985 installing new udev rules files. %_udevhwdbdir,
3986 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
3987 %_sysctldir are now available which resolve to the right
3988 directories for packages to place various data files in.
3990 * journalctl gained the new --full switch (in addition to
3991 --all, to disable ellipsation for long messages.
3993 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
3994 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
3995 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
3996 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
3997 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
3998 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
3999 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
4000 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
4001 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
4005 * journalctl gained new --since= and --until= switches to
4006 filter by time. It also now supports nice filtering for
4007 units via --unit=/-u.
4009 * Type=oneshot services may use ExecReload= and do the
4012 * The journal daemon now supports time-based rotation and
4013 vacuuming, in addition to the usual disk-space based
4016 * The journal will now index the available field values for
4017 each field name. This enables clients to show pretty drop
4018 downs of available match values when filtering. The bash
4019 completion of journalctl has been updated
4020 accordingly. journalctl gained a new switch -F to list all
4021 values a certain field takes in the journal database.
4023 * More service events are now written as structured messages
4024 to the journal, and made recognizable via message IDs.
4026 * The timedated, localed and hostnamed mini-services which
4027 previously only provided support for changing time, locale
4028 and hostname settings from graphical DEs such as GNOME now
4029 also have a minimal (but very useful) text-based client
4030 utility each. This is probably the nicest way to changing
4031 these settings from the command line now, especially since
4032 it lists available options and is fully integrated with bash
4035 * There's now a new tool "systemd-coredumpctl" to list and
4036 extract coredumps from the journal.
4038 * We now install a README each in /var/log/ and
4039 /etc/rc.d/init.d explaining where the system logs and init
4040 scripts went. This hopefully should help folks who go to
4041 that dirs and look into the otherwise now empty void and
4042 scratch their heads.
4044 * When user-services are invoked (by systemd --user) the
4045 $MANAGERPID env var is set to the PID of systemd.
4047 * SIGRTMIN+24 when sent to a --user instance will now result
4048 in immediate termination of systemd.
4050 * gatewayd received numerous feature additions such as a
4051 "follow" mode, for live syncing and filtering.
4053 * browse.html now allows filtering and showing detailed
4054 information on specific entries. Keyboard navigation and
4055 mouse screen support has been added.
4057 * gatewayd/journalctl now supports HTML5/JSON
4058 Server-Sent-Events as output.
4060 * The SysV init script compatibility logic will now
4061 heuristically determine whether a script supports the
4062 "reload" verb, and only then make this available as
4065 * "systemctl status --follow" has been removed, use "journalctl
4068 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
4069 have been removed since they are hardly useful to be
4072 * And I'd like to take the opportunity to specifically mention
4073 Zbigniew for his great contributions. Zbigniew, you rock!
4075 Contributions from: Andrew Eikum, Christian Hesse, Colin
4076 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
4077 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
4078 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
4079 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
4080 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
4081 Jędrzejewski-Szmek, Сковорода Никита Андреевич
4085 * If /etc/vconsole.conf is non-existent or empty we will no
4086 longer load any console font or key map at boot by
4087 default. Instead the kernel defaults will be left
4088 intact. This is definitely the right thing to do, as no
4089 configuration should mean no configuration, and hard-coding
4090 font names that are different on all archs is probably a bad
4091 idea. Also, the kernel default key map and font should be
4092 good enough for most cases anyway, and mostly identical to
4093 the userspace fonts/key maps we previously overloaded them
4094 with. If distributions want to continue to default to a
4095 non-kernel font or key map they should ship a default
4096 /etc/vconsole.conf with the appropriate contents.
4098 Contributions from: Colin Walters, Daniel J Walsh, Dave
4099 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
4100 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
4104 * journalctl gained a new --cursor= switch to show entries
4105 starting from the specified location in the journal.
4107 * We now enforce a size limit on journal entry fields exported
4108 with "-o json" in journalctl. Fields larger than 4K will be
4109 assigned null. This can be turned off with --all.
4111 * An (optional) journal gateway daemon is now available as
4112 "systemd-journal-gatewayd.service". This service provides
4113 access to the journal via HTTP and JSON. This functionality
4114 will be used to implement live log synchronization in both
4115 pull and push modes, but has various other users too, such
4116 as easy log access for debugging of embedded devices. Right
4117 now it is already useful to retrieve the journal via HTTP:
4119 # systemctl start systemd-journal-gatewayd.service
4120 # wget http://localhost:19531/entries
4122 This will download the journal contents in a
4123 /var/log/messages compatible format. The same as JSON:
4125 # curl -H"Accept: application/json" http://localhost:19531/entries
4127 This service is also accessible via a web browser where a
4128 single static HTML5 app is served that uses the JSON logic
4129 to enable the user to do some basic browsing of the
4130 journal. This will be extended later on. Here's an example
4131 screenshot of this app in its current state:
4133 http://0pointer.de/public/journal-gatewayd
4135 Contributions from: Kay Sievers, Lennart Poettering, Robert
4136 Milasan, Tom Gundersen
4140 * The bash completion logic is now available for journalctl
4143 * We do not mount the "cpuset" controller anymore together with
4144 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
4145 started if no parameters are assigned to it. "cpuset" hence
4146 broke code that assumed it it could create "cpu" groups and
4149 * journalctl -f will now subscribe to terminal size changes,
4150 and line break accordingly.
4152 Contributions from: Dave Reisner, Kay Sievers, Lennart
4153 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
4157 * nspawn will now create a symlink /etc/localtime in the
4158 container environment, copying the host's timezone
4159 setting. Previously this has been done via a bind mount, but
4160 since symlinks cannot be bind mounted this has now been
4161 changed to create/update the appropriate symlink.
4163 * journalctl -n's line number argument is now optional, and
4164 will default to 10 if omitted.
4166 * journald will now log the maximum size the journal files may
4167 take up on disk. This is particularly useful if the default
4168 built-in logic of determining this parameter from the file
4169 system size is used. Use "systemctl status
4170 systemd-journald.service" to see this information.
4172 * The multi-seat X wrapper tool has been stripped down. As X
4173 is now capable of enumerating graphics devices via udev in a
4174 seat-aware way the wrapper is not strictly necessary
4175 anymore. A stripped down temporary stop-gap is still shipped
4176 until the upstream display managers have been updated to
4177 fully support the new X logic. Expect this wrapper to be
4178 removed entirely in one of the next releases.
4180 * HandleSleepKey= in logind.conf has been split up into
4181 HandleSuspendKey= and HandleHibernateKey=. The old setting
4182 is not available anymore. X11 and the kernel are
4183 distinguishing between these keys and we should too. This
4184 also means the inhibition lock for these keys has been split
4187 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
4188 Poettering, Lukas Nykryn, Václav Pavlín
4192 * Whenever a unit changes state we will now log this to the
4193 journal and show along the unit's own log output in
4196 * ConditionPathIsMountPoint= can now properly detect bind
4197 mount points too. (Previously, a bind mount of one file
4198 system to another place in the same file system could not be
4199 detected as mount, since they shared struct stat's st_dev
4202 * We will now mount the cgroup controllers cpu, cpuacct,
4203 cpuset and the controllers net_cls, net_prio together by
4206 * nspawn containers will now have a virtualized boot
4207 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
4208 over with a randomized ID at container initialization). This
4209 has the effect of making "journalctl -b" do the right thing
4212 * The JSON output journal serialization has been updated not
4213 to generate "endless" list objects anymore, but rather one
4214 JSON object per line. This is more in line how most JSON
4215 parsers expect JSON objects. The new output mode
4216 "json-pretty" has been added to provide similar output, but
4217 neatly aligned for readability by humans.
4219 * We dropped all explicit sync() invocations in the shutdown
4220 code. The kernel does this implicitly anyway in the kernel
4221 reboot() syscall. halt(8)'s -n option is now a compatibility
4224 * We now support virtualized reboot() in containers, as
4225 supported by newer kernels. We will fall back to exit() if
4226 CAP_SYS_REBOOT is not available to the container. Also,
4227 nspawn makes use of this now and will actually reboot the
4228 container if the containerized OS asks for that.
4230 * journalctl will only show local log output by default
4231 now. Use --merge (-m) to show remote log output, too.
4233 * libsystemd-journal gained the new sd_journal_get_usage()
4234 call to determine the current disk usage of all journal
4235 files. This is exposed in the new "journalctl --disk-usage"
4238 * journald gained a new configuration setting SplitMode= in
4239 journald.conf which may be used to control how user journals
4240 are split off. See journald.conf(5) for details.
4242 * A new condition type ConditionFileNotEmpty= has been added.
4244 * tmpfiles' "w" lines now support file globbing, to write
4245 multiple files at once.
4247 * We added Python bindings for the journal submission
4248 APIs. More Python APIs for a number of selected APIs will
4249 likely follow. Note that we intend to add native bindings
4250 only for the Python language, as we consider it common
4251 enough to deserve bindings shipped within systemd. There are
4252 various projects outside of systemd that provide bindings
4253 for languages such as PHP or Lua.
4255 * Many conditions will now resolve specifiers such as %i. In
4256 addition, PathChanged= and related directives of .path units
4257 now support specifiers as well.
4259 * There's now a new RPM macro definition for the system preset
4262 * journald will now warn if it ca not forward a message to the
4263 syslog daemon because its socket is full.
4265 * timedated will no longer write or process /etc/timezone,
4266 except on Debian. As we do not support late mounted /usr
4267 anymore /etc/localtime always being a symlink is now safe,
4268 and hence the information in /etc/timezone is not necessary
4271 * logind will now always reserve one VT for a text getty (VT6
4272 by default). Previously if more than 6 X sessions where
4273 started they took up all the VTs with auto-spawned gettys,
4274 so that no text gettys were available anymore.
4276 * udev will now automatically inform the btrfs kernel logic
4277 about btrfs RAID components showing up. This should make
4278 simple hotplug based btrfs RAID assembly work.
4280 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
4281 (but not for its children which will stay at the kernel
4282 default). This should allow setups with a lot more listening
4285 * systemd will now always pass the configured timezone to the
4286 kernel at boot. timedated will do the same when the timezone
4289 * logind's inhibition logic has been updated. By default,
4290 logind will now handle the lid switch, the power and sleep
4291 keys all the time, even in graphical sessions. If DEs want
4292 to handle these events on their own they should take the new
4293 handle-power-key, handle-sleep-key and handle-lid-switch
4294 inhibitors during their runtime. A simple way to achieve
4295 that is to invoke the DE wrapped in an invocation of:
4297 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
4299 * Access to unit operations is now checked via SELinux taking
4300 the unit file label and client process label into account.
4302 * systemd will now notify the administrator in the journal
4303 when he over-mounts a non-empty directory.
4305 * There are new specifiers that are resolved in unit files,
4306 for the host name (%H), the machine ID (%m) and the boot ID
4309 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
4310 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
4311 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
4312 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
4313 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
4314 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
4315 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
4319 * Support for reading structured kernel messages from
4320 /dev/kmsg has now been added and is enabled by default.
4322 * Support for reading kernel messages from /proc/kmsg has now
4323 been removed. If you want kernel messages in the journal
4324 make sure to run a recent kernel (>= 3.5) that supports
4325 reading structured messages from /dev/kmsg (see
4326 above). /proc/kmsg is now exclusive property of classic
4327 syslog daemons again.
4329 * The libudev API gained the new
4330 udev_device_new_from_device_id() call.
4332 * The logic for file system namespace (ReadOnlyDirectory=,
4333 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
4334 require pivot_root() anymore. This means fewer temporary
4335 directories are created below /tmp for this feature.
4337 * nspawn containers will now see and receive all submounts
4338 made on the host OS below the root file system of the
4341 * Forward Secure Sealing is now supported for Journal files,
4342 which provide cryptographical sealing of journal files so
4343 that attackers cannot alter log history anymore without this
4344 being detectable. Lennart will soon post a blog story about
4345 this explaining it in more detail.
4347 * There are two new service settings RestartPreventExitStatus=
4348 and SuccessExitStatus= which allow configuration of exit
4349 status (exit code or signal) which will be excepted from the
4350 restart logic, resp. consider successful.
4352 * journalctl gained the new --verify switch that can be used
4353 to check the integrity of the structure of journal files and
4354 (if Forward Secure Sealing is enabled) the contents of
4357 * nspawn containers will now be run with /dev/stdin, /dev/fd/
4358 and similar symlinks pre-created. This makes running shells
4359 as container init process a lot more fun.
4361 * The fstab support can now handle PARTUUID= and PARTLABEL=
4364 * A new ConditionHost= condition has been added to match
4365 against the hostname (with globs) and machine ID. This is
4366 useful for clusters where a single OS image is used to
4367 provision a large number of hosts which shall run slightly
4368 different sets of services.
4370 * Services which hit the restart limit will now be placed in a
4373 Contributions from: Bertram Poettering, Dave Reisner, Huang
4374 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
4375 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
4379 * When running in --user mode systemd will now become a
4380 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
4381 tree a lot more organized.
4383 * A new PartOf= unit dependency type has been introduced that
4384 may be used to group services in a natural way.
4386 * "systemctl enable" may now be used to enable instances of
4389 * journalctl now prints error log levels in red, and
4390 warning/notice log levels in bright white. It also supports
4391 filtering by log level now.
4393 * cgtop gained a new -n switch (similar to top), to configure
4394 the maximum number of iterations to run for. It also gained
4395 -b, to run in batch mode (accepting no input).
4397 * The suffix ".service" may now be omitted on most systemctl
4398 command lines involving service unit names.
4400 * There's a new bus call in logind to lock all sessions, as
4401 well as a loginctl verb for it "lock-sessions".
4403 * libsystemd-logind.so gained a new call sd_journal_perror()
4404 that works similar to libc perror() but logs to the journal
4405 and encodes structured information about the error number.
4407 * /etc/crypttab entries now understand the new keyfile-size=
4410 * shutdown(8) now can send a (configurable) wall message when
4411 a shutdown is cancelled.
4413 * The mount propagation mode for the root file system will now
4414 default to "shared", which is useful to make containers work
4415 nicely out-of-the-box so that they receive new mounts from
4416 the host. This can be undone locally by running "mount
4417 --make-rprivate /" if needed.
4419 * The prefdm.service file has been removed. Distributions
4420 should maintain this unit downstream if they intend to keep
4421 it around. However, we recommend writing normal unit files
4422 for display managers instead.
4424 * Since systemd is a crucial part of the OS we will now
4425 default to a number of compiler switches that improve
4426 security (hardening) such as read-only relocations, stack
4427 protection, and suchlike.
4429 * The TimeoutSec= setting for services is now split into
4430 TimeoutStartSec= and TimeoutStopSec= to allow configuration
4431 of individual time outs for the start and the stop phase of
4434 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
4435 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
4436 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
4437 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
4438 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
4439 Gundersen, Zbigniew Jędrzejewski-Szmek
4443 * The journal and id128 C APIs are now fully documented as man
4446 * Extra safety checks have been added when transitioning from
4447 the initial RAM disk to the main system to avoid accidental
4450 * /etc/crypttab entries now understand the new keyfile-offset=
4453 * systemctl -t can now be used to filter by unit load state.
4455 * The journal C API gained the new sd_journal_wait() call to
4456 make writing synchronous journal clients easier.
4458 * journalctl gained the new -D switch to show journals from a
4461 * journalctl now displays a special marker between log
4462 messages of two different boots.
4464 * The journal is now explicitly flushed to /var via a service
4465 systemd-journal-flush.service, rather than implicitly simply
4466 by seeing /var/log/journal to be writable.
4468 * journalctl (and the journal C APIs) can now match for much
4469 more complex expressions, with alternatives and
4472 * When transitioning from the initial RAM disk to the main
4473 system we will now kill all processes in a killing spree to
4474 ensure no processes stay around by accident.
4476 * Three new specifiers may be used in unit files: %u, %h, %s
4477 resolve to the user name, user home directory resp. user
4478 shell. This is useful for running systemd user instances.
4480 * We now automatically rotate journal files if their data
4481 object hash table gets a fill level > 75%. We also size the
4482 hash table based on the configured maximum file size. This
4483 together should lower hash collisions drastically and thus
4484 speed things up a bit.
4486 * journalctl gained the new "--header" switch to introspect
4487 header data of journal files.
4489 * A new setting SystemCallFilters= has been added to services
4490 which may be used to apply blacklists or whitelists to
4491 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
4493 * nspawn gained a new --link-journal= switch (and quicker: -j)
4494 to link the container journal with the host. This makes it
4495 very easy to centralize log viewing on the host for all
4496 guests while still keeping the journal files separated.
4498 * Many bugfixes and optimizations
4500 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
4501 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
4502 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
4507 * Several tools now understand kernel command line arguments,
4508 which are only read when run in an initial RAM disk. They
4509 usually follow closely their normal counterparts, but are
4512 * There's a new tool to analyze the readahead files that are
4513 automatically generated at boot. Use:
4515 /usr/lib/systemd/systemd-readahead analyze /.readahead
4517 * We now provide an early debug shell on tty9 if this enabled. Use:
4519 systemctl enable debug-shell.service
4521 * All plymouth related units have been moved into the Plymouth
4522 package. Please make sure to upgrade your Plymouth version
4525 * systemd-tmpfiles now supports getting passed the basename of
4526 a configuration file only, in which case it will look for it
4527 in all appropriate directories automatically.
4529 * udevadm info now takes a /dev or /sys path as argument, and
4530 does the right thing. Example:
4532 udevadm info /dev/sda
4533 udevadm info /sys/class/block/sda
4535 * systemctl now prints a warning if a unit is stopped but a
4536 unit that might trigger it continues to run. Example: a
4537 service is stopped but the socket that activates it is left
4540 * "systemctl status" will now mention if the log output was
4541 shortened due to rotation since a service has been started.
4543 * The journal API now exposes functions to determine the
4544 "cutoff" times due to rotation.
4546 * journald now understands SIGUSR1 and SIGUSR2 for triggering
4547 immediately flushing of runtime logs to /var if possible,
4548 resp. for triggering immediate rotation of the journal
4551 * It is now considered an error if a service is attempted to
4552 be stopped that is not loaded.
4554 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
4556 * systemd-analyze now supports Python 3
4558 * tmpfiles now supports cleaning up directories via aging
4559 where the first level dirs are always kept around but
4560 directories beneath it automatically aged. This is enabled
4561 by prefixing the age field with '~'.
4563 * Seat objects now expose CanGraphical, CanTTY properties
4564 which is required to deal with very fast bootups where the
4565 display manager might be running before the graphics drivers
4566 completed initialization.
4568 * Seat objects now expose a State property.
4570 * We now include RPM macros for service enabling/disabling
4571 based on the preset logic. We recommend RPM based
4572 distributions to make use of these macros if possible. This
4573 makes it simpler to reuse RPM spec files across
4576 * We now make sure that the collected systemd unit name is
4577 always valid when services log to the journal via
4580 * There's a new man page kernel-command-line(7) detailing all
4581 command line options we understand.
4583 * The fstab generator may now be disabled at boot by passing
4584 fstab=0 on the kernel command line.
4586 * A new kernel command line option modules-load= is now understood
4587 to load a specific kernel module statically, early at boot.
4589 * Unit names specified on the systemctl command line are now
4590 automatically escaped as needed. Also, if file system or
4591 device paths are specified they are automatically turned
4592 into the appropriate mount or device unit names. Example:
4594 systemctl status /home
4595 systemctl status /dev/sda
4597 * The SysVConsole= configuration option has been removed from
4598 system.conf parsing.
4600 * The SysV search path is no longer exported on the D-Bus
4603 * The Names= option is been removed from unit file parsing.
4605 * There's a new man page bootup(7) detailing the boot process.
4607 * Every unit and every generator we ship with systemd now
4608 comes with full documentation. The self-explanatory boot is
4611 * A couple of services gained "systemd-" prefixes in their
4612 name if they wrap systemd code, rather than only external
4613 code. Among them fsck@.service which is now
4614 systemd-fsck@.service.
4616 * The HaveWatchdog property has been removed from the D-Bus
4619 * systemd.confirm_spawn= on the kernel command line should now
4622 * There's a new man page crypttab(5) which details all options
4623 we actually understand.
4625 * systemd-nspawn gained a new --capability= switch to pass
4626 additional capabilities to the container.
4628 * timedated will now read known NTP implementation unit names
4629 from /usr/lib/systemd/ntp-units.d/*.list,
4630 systemd-timedated-ntp.target has been removed.
4632 * journalctl gained a new switch "-b" that lists log data of
4633 the current boot only.
4635 * The notify socket is in the abstract namespace again, in
4636 order to support daemons which chroot() at start-up.
4638 * There is a new Storage= configuration option for journald
4639 which allows configuration of where log data should go. This
4640 also provides a way to disable journal logging entirely, so
4641 that data collected is only forwarded to the console, the
4642 kernel log buffer or another syslog implementation.
4644 * Many bugfixes and optimizations
4646 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
4647 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
4648 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
4649 Shawn Landden, Tom Gundersen
4653 * "systemctl help <unit>" now shows the man page if one is
4656 * Several new man pages have been added.
4658 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
4659 MaxLevelConsole= can now be specified in
4660 journald.conf. These options allow reducing the amount of
4661 data stored on disk or forwarded by the log level.
4663 * TimerSlackNSec= can now be specified in system.conf for
4664 PID1. This allows system-wide power savings.
4666 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
4667 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
4672 * logind is now capable of (optionally) handling power and
4673 sleep keys as well as the lid switch.
4675 * journalctl now understands the syntax "journalctl
4676 /usr/bin/avahi-daemon" to get all log output of a specific
4679 * CapabilityBoundingSet= in system.conf now also influences
4680 the capability bound set of usermode helpers of the kernel.
4682 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
4683 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
4684 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
4685 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
4689 * Note that we skipped 139 releases here in order to set the
4690 new version to something that is greater than both udev's
4691 and systemd's most recent version number.
4693 * udev: all udev sources are merged into the systemd source tree now.
4694 All future udev development will happen in the systemd tree. It
4695 is still fully supported to use the udev daemon and tools without
4696 systemd running, like in initramfs or other init systems. Building
4697 udev though, will require the *build* of the systemd tree, but
4698 udev can be properly *run* without systemd.
4700 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
4701 should be used to create dead device nodes as workarounds for broken
4704 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
4705 no longer supported. udev_monitor_new_from_netlink() needs to be
4706 used to subscribe to events.
4708 * udev: when udevd is started by systemd, processes which are left
4709 behind by forking them off of udev rules, are unconditionally cleaned
4710 up and killed now after the event handling has finished. Services or
4711 daemons must be started as systemd services. Services can be
4712 pulled-in by udev to get started, but they can no longer be directly
4713 forked by udev rules.
4715 * udev: the daemon binary is called systemd-udevd now and installed
4716 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
4717 to adapt to that, create symlink, or rename the binary after building
4720 * libudev no longer provides these symbols:
4721 udev_monitor_from_socket()
4722 udev_queue_get_failed_list_entry()
4723 udev_get_{dev,sys,run}_path()
4724 The versions number was bumped and symbol versioning introduced.
4726 * systemd-loginctl and systemd-journalctl have been renamed
4727 to loginctl and journalctl to match systemctl.
4729 * The config files: /etc/systemd/systemd-logind.conf and
4730 /etc/systemd/systemd-journald.conf have been renamed to
4731 logind.conf and journald.conf. Package updates should rename
4732 the files to the new names on upgrade.
4734 * For almost all files the license is now LGPL2.1+, changed
4735 from the previous GPL2.0+. Exceptions are some minor stuff
4736 of udev (which will be changed to LGPL2.1 eventually, too),
4737 and the MIT licensed sd-daemon.[ch] library that is suitable
4738 to be used as drop-in files.
4740 * systemd and logind now handle system sleep states, in
4741 particular suspending and hibernating.
4743 * logind now implements a sleep/shutdown/idle inhibiting logic
4744 suitable for a variety of uses. Soonishly Lennart will blog
4745 about this in more detail.
4747 * var-run.mount and var-lock.mount are no longer provided
4748 (which prevously bind mounted these directories to their new
4749 places). Distributions which have not converted these
4750 directories to symlinks should consider stealing these files
4751 from git history and add them downstream.
4753 * We introduced the Documentation= field for units and added
4754 this to all our shipped units. This is useful to make it
4755 easier to explore the boot and the purpose of the various
4758 * All smaller setup units (such as
4759 systemd-vconsole-setup.service) now detect properly if they
4760 are run in a container and are skipped when
4761 appropriate. This guarantees an entirely noise-free boot in
4762 Linux container environments such as systemd-nspawn.
4764 * A framework for implementing offline system updates is now
4765 integrated, for details see:
4766 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
4768 * A new service type Type=idle is available now which helps us
4769 avoiding ugly interleaving of getty output and boot status
4772 * There's now a system-wide CapabilityBoundingSet= option to
4773 globally reduce the set of capabilities for the
4774 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
4775 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
4776 even CAP_NET_ADMIN system-wide for secure systems.
4778 * There are now system-wide DefaultLimitXXX= options to
4779 globally change the defaults of the various resource limits
4780 for all units started by PID 1.
4782 * Harald Hoyer's systemd test suite has been integrated into
4783 systemd which allows easy testing of systemd builds in qemu
4784 and nspawn. (This is really awesome! Ask us for details!)
4786 * The fstab parser is now implemented as generator, not inside
4789 * systemctl will now warn you if .mount units generated from
4790 /etc/fstab are out of date due to changes in fstab that
4791 have not been read by systemd yet.
4793 * systemd is now suitable for usage in initrds. Dracut has
4794 already been updated to make use of this. With this in place
4795 initrds get a slight bit faster but primarily are much
4796 easier to introspect and debug since "systemctl status" in
4797 the host system can be used to introspect initrd services,
4798 and the journal from the initrd is kept around too.
4800 * systemd-delta has been added, a tool to explore differences
4801 between user/admin configuration and vendor defaults.
4803 * PrivateTmp= now affects both /tmp and /var/tmp.
4805 * Boot time status messages are now much prettier and feature
4806 proper english language. Booting up systemd has never been
4809 * Read-ahead pack files now include the inode number of all
4810 files to pre-cache. When the inode changes the pre-caching
4811 is not attempted. This should be nicer to deal with updated
4812 packages which might result in changes of read-ahead
4815 * We now temporaritly lower the kernel's read_ahead_kb variable
4816 when collecting read-ahead data to ensure the kernel's
4817 built-in read-ahead does not add noise to our measurements
4818 of necessary blocks to pre-cache.
4820 * There's now RequiresMountsFor= to add automatic dependencies
4821 for all mounts necessary for a specific file system path.
4823 * MountAuto= and SwapAuto= have been removed from
4824 system.conf. Mounting file systems at boot has to take place
4827 * nspawn now learned a new switch --uuid= to set the machine
4828 ID on the command line.
4830 * nspawn now learned the -b switch to automatically search
4833 * vt102 is now the default TERM for serial TTYs, upgraded from
4836 * systemd-logind now works on VT-less systems.
4838 * The build tree has been reorganized. The individual
4839 components now have directories of their own.
4841 * A new condition type ConditionPathIsReadWrite= is now available.
4843 * nspawn learned the new -C switch to create cgroups for the
4844 container in other hierarchies.
4846 * We now have support for hardware watchdogs, configurable in
4849 * The scheduled shutdown logic now has a public API.
4851 * We now mount /tmp as tmpfs by default, but this can be
4852 masked and /etc/fstab can override it.
4854 * Since udisks does not make use of /media anymore we are not
4855 mounting a tmpfs on it anymore.
4857 * journalctl gained a new --local switch to only interleave
4858 locally generated journal files.
4860 * We can now load the IMA policy at boot automatically.
4862 * The GTK tools have been split off into a systemd-ui.
4864 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
4865 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
4866 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
4867 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
4868 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
4869 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
4870 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
4871 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
4872 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
4877 * This is mostly a bugfix release
4879 * Support optional initialization of the machine ID from the
4880 KVM or container configured UUID.
4882 * Support immediate reboots with "systemctl reboot -ff"
4884 * Show /etc/os-release data in systemd-analyze output
4886 * Many bugfixes for the journal, including endianness fixes and
4887 ensuring that disk space enforcement works
4889 * sd-login.h is C++ comptaible again
4891 * Extend the /etc/os-release format on request of the Debian
4894 * We now refuse non-UTF8 strings used in various configuration
4895 and unit files. This is done to ensure we do not pass invalid
4896 data over D-Bus or expose it elsewhere.
4898 * Register Mimo USB Screens as suitable for automatic seat
4901 * Read SELinux client context from journal clients in a race
4904 * Reorder configuration file lookup order. /etc now always
4905 overrides /run in order to allow the administrator to always
4906 and unconditionally override vendor supplied or
4907 automatically generated data.
4909 * The various user visible bits of the journal now have man
4910 pages. We still lack man pages for the journal API calls
4913 * We now ship all man pages in HTML format again in the
4916 Contributions from: Dave Reisner, Dirk Eibach, Frederic
4917 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
4918 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
4923 * This is mostly a bugfix release
4925 * systems lacking /etc/os-release are no longer supported.
4927 * Various functionality updates to libsystemd-login.so
4929 * Track class of PAM logins to distinguish greeters from
4932 Contributions from: Kay Sievers, Lennart Poettering, Michael
4937 * This is an important bugfix release for v41.
4939 * Building man pages is now optional which should be useful
4940 for those building systemd from git but unwilling to install
4943 * Watchdog support for supervising services is now usable. In
4944 a future release support for hardware watchdogs
4945 (i.e. /dev/watchdog) will be added building on this.
4947 * Service start rate limiting is now configurable and can be
4948 turned off per service. When a start rate limit is hit a
4949 reboot can automatically be triggered.
4951 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
4953 Contributions from: Benjamin Franzke, Bill Nottingham,
4954 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
4955 Schmidt, Michał Górny, Piotr Drąg
4959 * The systemd binary is installed /usr/lib/systemd/systemd now;
4960 An existing /sbin/init symlink needs to be adapted with the
4963 * The code that loads kernel modules has been ported to invoke
4964 libkmod directly, instead of modprobe. This means we do not
4965 support systems with module-init-tools anymore.
4967 * Watchdog support is now already useful, but still not
4970 * A new kernel command line option systemd.setenv= is
4971 understood to set system wide environment variables
4972 dynamically at boot.
4974 * We now limit the set of capabilities of systemd-journald.
4976 * We now set SIGPIPE to ignore by default, since it only is
4977 useful in shell pipelines, and has little use in general
4978 code. This can be disabled with IgnoreSIPIPE=no in unit
4981 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
4982 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
4987 * This is mostly a bugfix release
4989 * We now expose the reason why a service failed in the
4990 "Result" D-Bus property.
4992 * Rudimentary service watchdog support (will be completed over
4993 the next few releases.)
4995 * When systemd forks off in order execute some service we will
4996 now immediately changes its argv[0] to reflect which process
4997 it will execute. This is useful to minimize the time window
4998 with a generic argv[0], which makes bootcharts more useful
5000 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
5001 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
5002 Mike Kazantsev, Ray Strode
5006 * This is mostly a test release, but incorporates many
5009 * New systemd-cgtop tool to show control groups by their
5012 * Linking against libacl for ACLs is optional again. If
5013 disabled, support tracking device access for active logins
5014 goes becomes unavailable, and so does access to the user
5015 journals by the respective users.
5017 * If a group "adm" exists, journal files are automatically
5018 owned by them, thus allow members of this group full access
5019 to the system journal as well as all user journals.
5021 * The journal now stores the SELinux context of the logging
5022 client for all entries.
5024 * Add C++ inclusion guards to all public headers
5026 * New output mode "cat" in the journal to print only text
5027 messages, without any meta data like date or time.
5029 * Include tiny X server wrapper as a temporary stop-gap to
5030 teach XOrg udev display enumeration. This is used by display
5031 managers such as gdm, and will go away as soon as XOrg
5032 learned native udev hotplugging for display devices.
5034 * Add new systemd-cat tool for executing arbitrary programs
5035 with STDERR/STDOUT connected to the journal. Can also act as
5036 BSD logger replacement, and does so by default.
5038 * Optionally store all locally generated coredumps in the
5039 journal along with meta data.
5041 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
5042 writing short strings to files (for usage for /sys), and for
5043 creating symlinks, character and block device nodes.
5045 * New unit file option ControlGroupPersistent= to make cgroups
5046 persistent, following the mechanisms outlined in
5047 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
5049 * Support multiple local RTCs in a sane way
5051 * No longer monopolize IO when replaying readahead data on
5052 rotating disks, since we might starve non-file-system IO to
5053 death, since fanotify() will not see accesses done by blkid,
5056 * Do not show kernel threads in systemd-cgls anymore, unless
5057 requested with new -k switch.
5059 Contributions from: Dan Horák, Kay Sievers, Lennart
5060 Poettering, Michal Schmidt
5064 * This is mostly a test release, but incorporates many
5067 * The git repository moved to:
5068 git://anongit.freedesktop.org/systemd/systemd
5069 ssh://git.freedesktop.org/git/systemd/systemd
5071 * First release with the journal
5072 http://0pointer.de/blog/projects/the-journal.html
5074 * The journal replaces both systemd-kmsg-syslogd and
5075 systemd-stdout-bridge.
5077 * New sd_pid_get_unit() API call in libsystemd-logind
5079 * Many systemadm clean-ups
5081 * Introduce remote-fs-pre.target which is ordered before all
5082 remote mounts and may be used to start services before all
5085 * Added Mageia support
5087 * Add bash completion for systemd-loginctl
5089 * Actively monitor PID file creation for daemons which exit in
5090 the parent process before having finished writing the PID
5091 file in the daemon process. Daemons which do this need to be
5092 fixed (i.e. PID file creation must have finished before the
5093 parent exits), but we now react a bit more gracefully to them.
5095 * Add colourful boot output, mimicking the well-known output
5096 of existing distributions.
5098 * New option PassCredentials= for socket units, for
5099 compatibility with a recent kernel ABI breakage.
5101 * /etc/rc.local is now hooked in via a generator binary, and
5102 thus will no longer act as synchronization point during
5105 * systemctl list-unit-files now supports --root=.
5107 * systemd-tmpfiles now understands two new commands: z, Z for
5108 relabelling files according to the SELinux database. This is
5109 useful to apply SELinux labels to specific files in /sys,
5112 * Output of SysV services is now forwarded to both the console
5113 and the journal by default, not only just the console.
5115 * New man pages for all APIs from libsystemd-login.
5117 * The build tree got reorganized and a the build system is a
5118 lot more modular allowing embedded setups to specifically
5119 select the components of systemd they are interested in.
5121 * Support for Linux systems lacking the kernel VT subsystem is
5124 * configure's --with-rootdir= got renamed to
5125 --with-rootprefix= to follow the naming used by udev and
5128 * Unless specified otherwise we will now install to /usr instead
5129 of /usr/local by default.
5131 * Processes with '@' in argv[0][0] are now excluded from the
5132 final shut-down killing spree, following the logic explained
5134 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
5136 * All processes remaining in a service cgroup when we enter
5137 the START or START_PRE states are now killed with
5138 SIGKILL. That means it is no longer possible to spawn
5139 background processes from ExecStart= lines (which was never
5140 supported anyway, and bad style).
5142 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
5143 reloading of units together.
5145 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
5146 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
5147 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
5148 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
5149 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek