1 systemd System and Service Manager
3 CHANGES WITH 234 in spe
5 * The code to call libnss_dns as a fallback from libnss_resolve when
6 the communication with systemd-resolved fails was removed. This
7 fallback was redundant and interfered with the [!UNAVAIL=return]
8 suffix. See nss-resolve(8) for the recommended configuration.
12 * This version requires at least gperf 3.1 for building, 3.0 is not
15 * The "hybrid" control group mode has been modified to improve
16 compatibility with "legacy" cgroups-v1 setups. Specifically, the
17 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
18 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
19 cgroups-v1 hierarchy), the only externally visible change being that
20 the cgroups-v2 hierarchy is also mounted, to
21 /sys/fs/cgroup/unified. This should provide a large degree of
22 compatibility with "legacy" cgroups-v1, while taking benefit of the
23 better management capabilities of cgroups-v2.
25 * The default control group setup mode may be selected both a boot-time
26 via a set of kernel command line parameters (specifically:
27 systemd.unified_cgroup_hierarchy= and
28 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
29 default selected on the configure command line
30 (--with-default-hierarchy=). The upstream default is "hybrid"
31 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
32 this will change in a future systemd version to be "unified" (pure
33 cgroups-v2 mode). The third option for the compile time option is
34 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
35 distributions to default to "hybrid" mode for release distributions,
36 starting with v233. We recommend "unified" for development
37 distributions (specifically: distributions such as Fedora's rawhide)
38 as that's where things are headed in the long run. Use "legacy" for
39 greatest stability and compatibility only.
41 * Note one current limitation of "unified" and "hybrid" control group
42 setup modes: the kernel currently does not permit the systemd --user
43 instance (i.e. unprivileged code) to migrate processes between two
44 disconnected cgroup subtrees, even if both are managed and owned by
45 the user. This effectively means "systemd-run --user --scope" doesn't
46 work when invoked from outside of any "systemd --user" service or
47 scope. Specifically, it is not supported from session scopes. We are
48 working on fixing this in a future systemd version. (See #3388 for
49 further details about this.)
51 * DBus policy files are now installed into /usr rather than /etc. Make
52 sure your system has dbus >= 1.9.18 running before upgrading to this
53 version, or override the install path with --with-dbuspolicydir= .
55 * All python scripts shipped with systemd (specifically: the various
56 tests written in Python) now require Python 3.
58 * systemd unit tests can now run standalone (without the source or
59 build directories), and can be installed into /usr/lib/systemd/tests/
60 with 'make install-tests'.
62 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
63 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
66 * Support for the %c, %r, %R specifiers in unit files has been
67 removed. Specifiers are not supposed to be dependent on configuration
68 in the unit file itself (so that they resolve the same regardless
69 where used in the unit files), but these specifiers were influenced
72 * The shell invoked by debug-shell.service now defaults to /bin/sh in
73 all cases. If distributions want to use a different shell for this
74 purpose (for example Fedora's /sbin/sushell) they need to specify
75 this explicitly at configure time using --with-debug-shell=.
77 * The confirmation spawn prompt has been reworked to offer the
80 (c)ontinue, proceed without asking anymore
81 (D)ump, show the state of the unit
82 (f)ail, don't execute the command and pretend it failed
84 (i)nfo, show a short summary of the unit
85 (j)obs, show jobs that are in progress
86 (s)kip, don't execute the command and pretend it succeeded
87 (y)es, execute the command
89 The 'n' choice for the confirmation spawn prompt has been removed,
90 because its meaning was confusing.
92 The prompt may now also be redirected to an alternative console by
93 specifying the console as parameter to systemd.confirm_spawn=.
95 * Services of Type=notify require a READY=1 notification to be sent
96 during startup. If no such message is sent, the service now fails,
97 even if the main process exited with a successful exit code.
99 * Services that fail to start up correctly now always have their
100 ExecStopPost= commands executed. Previously, they'd enter "failed"
101 state directly, without executing these commands.
103 * The option MulticastDNS= of network configuration files has acquired
104 an actual implementation. With MulticastDNS=yes a host can resolve
105 names of remote hosts and reply to mDNS A and AAAA requests.
107 * When units are about to be started an additional check is now done to
108 ensure that all dependencies of type BindsTo= (when used in
109 combination with After=) have been started.
111 * systemd-analyze gained a new verb "syscall-filter" which shows which
112 system call groups are defined for the SystemCallFilter= unit file
113 setting, and which system calls they contain.
115 * A new system call filter group "@filesystem" has been added,
116 consisting of various file system related system calls. Group
117 "@reboot" has been added, covering reboot, kexec and shutdown related
118 calls. Finally, group "@swap" has been added covering swap
119 configuration related calls.
121 * A new unit file option RestrictNamespaces= has been added that may be
122 used to restrict access to the various process namespace types the
123 Linux kernel provides. Specifically, it may be used to take away the
124 right for a service unit to create additional file system, network,
125 user, and other namespaces. This sandboxing option is particularly
126 relevant due to the high amount of recently discovered namespacing
127 related vulnerabilities in the kernel.
129 * systemd-udev's .link files gained support for a new AutoNegotiation=
130 setting for configuring Ethernet auto-negotiation.
132 * systemd-networkd's .network files gained support for a new
133 ListenPort= setting in the [DHCP] section to explicitly configure the
134 UDP client port the DHCP client shall listen on.
136 * .network files gained a new Unmanaged= boolean setting for explicitly
137 excluding one or more interfaces from management by systemd-networkd.
139 * The systemd-networkd ProxyARP= option has been renamed to
140 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
141 renamed to ReduceARPProxy=. The old names continue to be available
144 * systemd-networkd gained support for configuring IPv6 Proxy NDP
145 addresses via the new IPv6ProxyNDPAddress= .network file setting.
147 * systemd-networkd's bonding device support gained support for two new
148 configuration options ActiveSlave= and PrimarySlave=.
150 * The various options in the [Match] section of .network files gained
151 support for negative matching.
153 * New systemd-specific mount options are now understood in /etc/fstab:
155 x-systemd.mount-timeout= may be used to configure the maximum
156 permitted runtime of the mount command.
158 x-systemd.device-bound may be set to bind a mount point to its
159 backing device unit, in order to automatically remove a mount point
160 if its backing device is unplugged. This option may also be
161 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
162 on the block device, which is now automatically set for all CDROM
163 drives, so that mounted CDs are automatically unmounted when they are
164 removed from the drive.
166 x-systemd.after= and x-systemd.before= may be used to explicitly
167 order a mount after or before another unit or mount point.
169 * Enqueued start jobs for device units are now automatically garbage
170 collected if there are no jobs waiting for them anymore.
172 * systemctl list-jobs gained two new switches: with --after, for every
173 queued job the jobs it's waiting for are shown; with --before the
174 jobs which it's blocking are shown.
176 * systemd-nspawn gained support for ephemeral boots from disk images
177 (or in other words: --ephemeral and --image= may now be
178 combined). Moreover, ephemeral boots are now supported for normal
179 directories, even if the backing file system is not btrfs. Of course,
180 if the file system does not support file system snapshots or
181 reflinks, the initial copy operation will be relatively expensive, but
182 this should still be suitable for many use cases.
184 * Calendar time specifications in .timer units now support
185 specifications relative to the end of a month by using "~" instead of
186 "-" as separator between month and day. For example, "*-02~03" means
187 "the third last day in February". In addition a new syntax for
188 repeated events has been added using the "/" character. For example,
189 "9..17/2:00" means "every two hours from 9am to 5pm".
191 * systemd-socket-proxyd gained a new parameter --connections-max= for
192 configuring the maximum number of concurrent connections.
194 * sd-id128 gained a new API for generating unique IDs for the host in a
195 way that does not leak the machine ID. Specifically,
196 sd_id128_get_machine_app_specific() derives an ID based on the
197 machine ID a in well-defined, non-reversible, stable way. This is
198 useful whenever an identifier for the host is needed but where the
199 identifier shall not be useful to identify the system beyond the
200 scope of the application itself. (Internally this uses HMAC-SHA256 as
201 keyed hash function using the machine ID as input.)
203 * NotifyAccess= gained a new supported value "exec". When set
204 notifications are accepted from all processes systemd itself invoked,
205 including all control processes.
207 * .nspawn files gained support for defining overlay mounts using the
208 Overlay= and OverlayReadOnly= options. Previously this functionality
209 was only available on the systemd-nspawn command line.
211 * systemd-nspawn's --bind= and --overlay= options gained support for
212 bind/overlay mounts whose source lies within the container tree by
213 prefixing the source path with "+".
215 * systemd-nspawn's --bind= and --overlay= options gained support for
216 automatically allocating a temporary source directory in /var/tmp
217 that is removed when the container dies. Specifically, if the source
218 directory is specified as empty string this mechanism is selected. An
219 example usage is --overlay=+/var::/var, which creates an overlay
220 mount based on the original /var contained in the image, overlayed
221 with a temporary directory in the host's /var/tmp. This way changes
222 to /var are automatically flushed when the container shuts down.
224 * systemd-nspawn --image= option does now permit raw file system block
225 devices (in addition to images containing partition tables, as
228 * The disk image dissection logic in systemd-nspawn gained support for
229 automatically setting up LUKS encrypted as well as Verity protected
230 partitions. When a container is booted from an encrypted image the
231 passphrase is queried at start-up time. When a container with Verity
232 data is started, the root hash is search in a ".roothash" file
233 accompanying the disk image (alternatively, pass the root hash via
234 the new --root-hash= command line option).
236 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
237 be used to dissect disk images the same way as systemd-nspawn does
238 it, following the Bootable Partition Specification. It may even be
239 used to mount disk images with complex partition setups (including
240 LUKS and Verity partitions) to a local host directory, in order to
241 inspect them. This tool is not considered public API (yet), and is
242 thus not installed into /usr/bin. Please do not rely on its
243 existence, since it might go away or be changed in later systemd
246 * A new generator "systemd-verity-generator" has been added, similar in
247 style to "systemd-cryptsetup-generator", permitting automatic setup of
248 Verity root partitions when systemd boots up. In order to make use of
249 this your partition setup should follow the Discoverable Partitions
250 Specification, and the GPT partition ID of the root file system
251 partition should be identical to the upper 128bit of the Verity root
252 hash. The GPT partition ID of the Verity partition protecting it
253 should be the lower 128bit of the Verity root hash. If the partition
254 image follows this model it is sufficient to specify a single
255 "roothash=" kernel command line argument to both configure which root
256 image and verity partition to use as well as the root hash for
257 it. Note that systemd-nspawn's Verity support follows the same
258 semantics, meaning that disk images with proper Verity data in place
259 may be booted in containers with systemd-nspawn as well as on
260 physical systems via the verity generator. Also note that the "mkosi"
261 tool available at https://github.com/systemd/mkosi has been updated
262 to generate Verity protected disk images following this scheme. In
263 fact, it has been updated to generate disk images that optionally
264 implement a complete UEFI SecureBoot trust chain, involving a signed
265 kernel and initrd image that incorporates such a root hash as well as
266 a Verity-enabled root partition.
268 * The hardware database (hwdb) udev supports has been updated to carry
269 accelerometer quirks.
271 * All system services are now run with a fresh kernel keyring set up
272 for them. The invocation ID is stored by default in it, thus
273 providing a safe, non-overridable way to determine the invocation
276 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
277 options for bind mounting arbitrary paths in a service-specific
278 way. When these options are used, arbitrary host or service files and
279 directories may be mounted to arbitrary locations in the service's
282 * Documentation has been added that lists all of systemd's low-level
283 environment variables:
285 https://github.com/systemd/systemd/blob/master/ENVIRONMENT.md
287 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
288 whether a specific socket file descriptor matches a specified socket
291 * systemd-firstboot has been updated to check for the
292 systemd.firstboot= kernel command line option. It accepts a boolean
293 and when set to false the first boot questions are skipped.
295 * systemd-fstab-generator has been updated to check for the
296 systemd.volatile= kernel command line option, which either takes an
297 optional boolean parameter or the special value "state". If used the
298 system may be booted in a "volatile" boot mode. Specifically,
299 "systemd.volatile" is used, the root directory will be mounted as
300 tmpfs, and only /usr is mounted from the actual root file system. If
301 "systemd.volatile=state" is used, the root directory will be mounted
302 as usual, but /var is mounted as tmpfs. This concept provides similar
303 functionality as systemd-nspawn's --volatile= option, but provides it
304 on physical boots. Use this option for implementing stateless
305 systems, or testing systems with all state and/or configuration reset
306 to the defaults. (Note though that many distributions are not
307 prepared to boot up without a populated /etc or /var, though.)
309 * systemd-gpt-auto-generator gained support for LUKS encrypted root
310 partitions. Previously it only supported LUKS encrypted partitions
311 for all other uses, except for the root partition itself.
313 * Socket units gained support for listening on AF_VSOCK sockets for
314 communication in virtualized QEMU environments.
316 * The "configure" script gained a new option --with-fallback-hostname=
317 for specifying the fallback hostname to use if none is configured in
318 /etc/hostname. For example, by specifying
319 --with-fallback-hostname=fedora it is possible to default to a
320 hostname of "fedora" on pristine installations.
322 * systemd-cgls gained support for a new --unit= switch for listing only
323 the control groups of a specific unit. Similar --user-unit= has been
324 added for listing only the control groups of a specific user unit.
326 * systemd-mount gained a new --umount switch for unmounting a mount or
327 automount point (and all mount/automount points below it).
329 * systemd will now refuse full configuration reloads (via systemctl
330 daemon-reload and related calls) unless at least 16MiB of free space
331 are available in /run. This is a safety precaution in order to ensure
332 that generators can safely operate after the reload completed.
334 * A new unit file option RootImage= has been added, which has a similar
335 effect as RootDirectory= but mounts the service's root directory from
336 a disk image instead of plain directory. This logic reuses the same
337 image dissection and mount logic that systemd-nspawn already uses,
338 and hence supports any disk images systemd-nspawn supports, including
339 those following the Discoverable Partition Specification, as well as
340 Verity enabled images. This option enables systemd to run system
341 services directly off disk images acting as resource bundles,
342 possibly even including full integrity data.
344 * A new MountAPIVFS= unit file option has been added, taking a boolean
345 argument. If enabled /proc, /sys and /dev (collectively called the
346 "API VFS") will be mounted for the service. This is only relevant if
347 RootDirectory= or RootImage= is used for the service, as these mounts
348 are of course in place in the host mount namespace anyway.
350 * systemd-nspawn gained support for a new --pivot-root= switch. If
351 specified the root directory within the container image is pivoted to
352 the specified mount point, while the original root disk is moved to a
353 different place. This option enables booting of ostree images
354 directly with systemd-nspawn.
356 * The systemd build scripts will no longer complain if the NTP server
357 addresses are not changed from the defaults. Google now supports
358 these NTP servers officially. We still recommend downstreams to
359 properly register an NTP pool with the NTP pool project though.
361 * coredumpctl gained a new "--reverse" option for printing the list
362 of coredumps in reverse order.
364 * coredumpctl will now show additional information about truncated and
365 inaccessible coredumps, as well as coredumps that are still being
366 processed. It also gained a new --quiet switch for suppressing
367 additional informational message in its output.
369 * coredumpctl gained support for only showing coredumps newer and/or
370 older than specific timestamps, using the new --since= and --until=
371 options, reminiscent of journalctl's options by the same name.
373 * The systemd-coredump logic has been improved so that it may be reused
374 to collect backtraces in non-compiled languages, for example in
375 scripting languages such as Python.
377 * machinectl will now show the UID shift of local containers, if user
378 namespacing is enabled for them.
380 * systemd will now optionally run "environment generator" binaries at
381 configuration load time. They may be used to add environment
382 variables to the environment block passed to services invoked. One
383 user environment generator is shipped by default that sets up
384 environment variables based on files dropped into /etc/environment.d
385 and ~/.config/environment.d/.
387 * systemd-resolved now includes the new, recently published 2017 DNSSEC
390 * hostnamed has been updated to report a new chassis type of
391 "convertible" to cover "foldable" laptops that can both act as a
392 tablet and as a laptop, such as various Lenovo Yoga devices.
394 Contributions from: Adrián López, Alexander Galanin, Alexander
395 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
396 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
397 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
398 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
399 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
400 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
401 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
402 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
403 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
404 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
405 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
406 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
407 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
408 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
409 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
410 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
411 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
412 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
413 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
414 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
415 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
416 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
417 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
418 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
419 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
420 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
421 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
428 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
429 RestrictAddressFamilies= enabled. These sandboxing options should
430 generally be compatible with the various external udev call-out
431 binaries we are aware of, however there may be exceptions, in
432 particular when exotic languages for these call-outs are used. In
433 this case, consider turning off these settings locally.
435 * The new RemoveIPC= option can be used to remove IPC objects owned by
436 the user or group of a service when that service exits.
438 * The new ProtectKernelModules= option can be used to disable explicit
439 load and unload operations of kernel modules by a service. In
440 addition access to /usr/lib/modules is removed if this option is set.
442 * ProtectSystem= option gained a new value "strict", which causes the
443 whole file system tree with the exception of /dev, /proc, and /sys,
444 to be remounted read-only for a service.
446 * The new ProtectKernelTunables= option can be used to disable
447 modification of configuration files in /sys and /proc by a service.
448 Various directories and files are remounted read-only, so access is
449 restricted even if the file permissions would allow it.
451 * The new ProtectControlGroups= option can be used to disable write
452 access by a service to /sys/fs/cgroup.
454 * Various systemd services have been hardened with
455 ProtectKernelTunables=yes, ProtectControlGroups=yes,
456 RestrictAddressFamilies=.
458 * Support for dynamically creating users for the lifetime of a service
459 has been added. If DynamicUser=yes is specified, user and group IDs
460 will be allocated from the range 61184..65519 for the lifetime of the
461 service. They can be resolved using the new nss-systemd.so NSS
462 module. The module must be enabled in /etc/nsswitch.conf. Services
463 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
464 any resources allocated by the service will be cleaned up when the
465 service exits. They also have ProtectHome=read-only and
466 ProtectSystem=strict enabled, so they are not able to make any
467 permanent modifications to the system.
469 * The nss-systemd module also always resolves root and nobody, making
470 it possible to have no /etc/passwd or /etc/group files in minimal
471 container or chroot environments.
473 * Services may be started with their own user namespace using the new
474 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
475 under which the service is running are mapped. All other users are
478 * Support for the cgroup namespace has been added to systemd-nspawn. If
479 supported by kernel, the container system started by systemd-nspawn
480 will have its own view of the cgroup hierarchy. This new behaviour
481 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
483 * The new MemorySwapMax= option can be used to limit the maximum swap
484 usage under the unified cgroup hierarchy.
486 * Support for the CPU controller in the unified cgroup hierarchy has
487 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
488 options. This controller requires out-of-tree patches for the kernel
489 and the support is provisional.
491 * Mount and automount units may now be created transiently
492 (i.e. dynamically at runtime via the bus API, instead of requiring
493 unit files in the file system).
495 * systemd-mount is a new tool which may mount file systems – much like
496 mount(8), optionally pulling in additional dependencies through
497 transient .mount and .automount units. For example, this tool
498 automatically runs fsck on a backing block device before mounting,
499 and allows the automount logic to be used dynamically from the
500 command line for establishing mount points. This tool is particularly
501 useful when dealing with removable media, as it will ensure fsck is
502 run – if necessary – before the first access and that the file system
503 is quickly unmounted after each access by utilizing the automount
504 logic. This maximizes the chance that the file system on the
505 removable media stays in a clean state, and if it isn't in a clean
506 state is fixed automatically.
508 * LazyUnmount=yes option for mount units has been added to expose the
509 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
512 * /efi will be used as the mount point of the EFI boot partition, if
513 the directory is present, and the mount point was not configured
514 through other means (e.g. fstab). If /efi directory does not exist,
515 /boot will be used as before. This makes it easier to automatically
516 mount the EFI partition on systems where /boot is used for something
519 * When operating on GPT disk images for containers, systemd-nspawn will
520 now mount the ESP to /boot or /efi according to the same rules as PID
521 1 running on a host. This allows tools like "bootctl" to operate
522 correctly within such containers, in order to make container images
523 bootable on physical systems.
525 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
527 * Two new user session targets have been added to support running
528 graphical sessions under the systemd --user instance:
529 graphical-session.target and graphical-session-pre.target. See
530 systemd.special(7) for a description of how those targets should be
533 * The vconsole initialization code has been significantly reworked to
534 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
535 support unicode keymaps. Font and keymap configuration will now be
536 copied to all allocated virtual consoles.
538 * FreeBSD's bhyve virtualization is now detected.
540 * Information recorded in the journal for core dumps now includes the
541 contents of /proc/mountinfo and the command line of the process at
542 the top of the process hierarchy (which is usually the init process
545 * systemd-journal-gatewayd learned the --directory= option to serve
546 files from the specified location.
548 * journalctl --root=… can be used to peruse the journal in the
549 /var/log/ directories inside of a container tree. This is similar to
550 the existing --machine= option, but does not require the container to
553 * The hardware database has been extended to support
554 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
557 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
558 specify the click rate for mice which include a horizontal wheel with
559 a click rate that is different than the one for the vertical wheel.
561 * systemd-run gained a new --wait option that makes service execution
562 synchronous. (Specifically, the command will not return until the
563 specified service binary exited.)
565 * systemctl gained a new --wait option that causes the start command to
566 wait until the units being started have terminated again.
568 * A new journal output mode "short-full" has been added which displays
569 timestamps with abbreviated English day names and adds a timezone
570 suffix. Those timestamps include more information than the default
571 "short" output mode, and can be passed directly to journalctl's
572 --since= and --until= options.
574 * /etc/resolv.conf will be bind-mounted into containers started by
575 systemd-nspawn, if possible, so any changes to resolv.conf contents
576 are automatically propagated to the container.
578 * The number of instances for socket-activated services originating
579 from a single IP address can be limited with
580 MaxConnectionsPerSource=, extending the existing setting of
583 * systemd-networkd gained support for vcan ("Virtual CAN") interface
586 * .netdev and .network configuration can now be extended through
589 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
590 Segmentation Offload, Generic Receive Offload, Large Receive Offload
591 can be enabled and disabled using the new UDPSegmentationOffload=,
592 TCPSegmentationOffload=, GenericSegmentationOffload=,
593 GenericReceiveOffload=, LargeReceiveOffload= options in the
594 [Link] section of .link files.
596 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
597 Port VLAN ID can be configured for bridge devices using the new STP=,
598 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
599 section of .netdev files.
601 * The route table to which routes received over DHCP or RA should be
602 added can be configured with the new RouteTable= option in the [DHCP]
603 and [IPv6AcceptRA] sections of .network files.
605 * The Address Resolution Protocol can be disabled on links managed by
606 systemd-networkd using the ARP=no setting in the [Link] section of
609 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
610 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
611 encode information about the result and exit codes of the current
612 service runtime cycle.
614 * systemd-sysctl will now configure kernel parameters in the order
615 they occur in the configuration files. This matches what sysctl
616 has been traditionally doing.
618 * kernel-install "plugins" that are executed to perform various
619 tasks after a new kernel is added and before an old one is removed
620 can now return a special value to terminate the procedure and
621 prevent any later plugins from running.
623 * Journald's SplitMode=login setting has been deprecated. It has been
624 removed from documentation, and its use is discouraged. In a future
625 release it will be completely removed, and made equivalent to current
626 default of SplitMode=uid.
628 * Storage=both option setting in /etc/systemd/coredump.conf has been
629 removed. With fast LZ4 compression storing the core dump twice is not
632 * The --share-system systemd-nspawn option has been replaced with an
633 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
634 this functionality is discouraged. In addition the variables
635 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
636 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
637 individual namespaces.
639 * "machinectl list" now shows the IP address of running containers in
640 the output, as well as OS release information.
642 * "loginctl list" now shows the TTY of each session in the output.
644 * sd-bus gained new API calls sd_bus_track_set_recursive(),
645 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
646 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
647 tracking objects in a "recursive" mode, where a single client can be
648 counted multiple times, if it takes multiple references.
650 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
651 sd_bus_get_exit_on_disconnect(). They may be used to to make a
652 process using sd-bus automatically exit if the bus connection is
655 * Bus clients of the service manager may now "pin" loaded units into
656 memory, by taking an explicit reference on them. This is useful to
657 ensure the client can retrieve runtime data about the service even
658 after the service completed execution. Taking such a reference is
659 available only for privileged clients and should be helpful to watch
660 running services in a race-free manner, and in particular collect
661 information about exit statuses and results.
663 * The nss-resolve module has been changed to strictly return UNAVAIL
664 when communication via D-Bus with resolved failed, and NOTFOUND when
665 a lookup completed but was negative. This means it is now possible to
666 neatly configure fallbacks using nsswitch.conf result checking
667 expressions. Taking benefit of this, the new recommended
668 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
670 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
672 * A new setting CtrlAltDelBurstAction= has been added to
673 /etc/systemd/system.conf which may be used to configure the precise
674 behaviour if the user on the console presses Ctrl-Alt-Del more often
675 than 7 times in 2s. Previously this would unconditionally result in
676 an expedited, immediate reboot. With this new setting the precise
677 operation may be configured in more detail, and also turned off
680 * In .netdev files two new settings RemoteChecksumTx= and
681 RemoteChecksumRx= are now understood that permit configuring the
682 remote checksumming logic for VXLAN networks.
684 * The service manager learnt a new "invocation ID" concept for invoked
685 services. Each runtime cycle of a service will get a new invocation
686 ID (a 128bit random UUID) assigned that identifies the current
687 run of the service uniquely and globally. A new invocation ID
688 is generated each time a service starts up. The journal will store
689 the invocation ID of a service along with any logged messages, thus
690 making the invocation ID useful for matching the online runtime of a
691 service with the offline log data it generated in a safe way without
692 relying on synchronized timestamps. In many ways this new service
693 invocation ID concept is similar to the kernel's boot ID concept that
694 uniquely and globally identifies the runtime of each boot. The
695 invocation ID of a service is passed to the service itself via an
696 environment variable ($INVOCATION_ID). A new bus call
697 GetUnitByInvocationID() has been added that is similar to GetUnit()
698 but instead of retrieving the bus path for a unit by its name
699 retrieves it by its invocation ID. The returned path is valid only as
700 long as the passed invocation ID is current.
702 * systemd-resolved gained a new "DNSStubListener" setting in
703 resolved.conf. It either takes a boolean value or the special values
704 "udp" and "tcp", and configures whether to enable the stub DNS
705 listener on 127.0.0.53:53.
707 * IP addresses configured via networkd may now carry additional
708 configuration settings supported by the kernel. New options include:
709 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
710 PrefixRoute=, AutoJoin=.
712 * The PAM configuration fragment file for "user@.service" shipped with
713 systemd (i.e. the --user instance of systemd) has been stripped to
714 the minimum necessary to make the system boot. Previously, it
715 contained Fedora-specific stanzas that did not apply to other
716 distributions. It is expected that downstream distributions add
717 additional configuration lines, matching their needs to this file,
718 using it only as rough template of what systemd itself needs. Note
719 that this reduced fragment does not even include an invocation of
720 pam_limits which most distributions probably want to add, even though
721 systemd itself does not need it. (There's also the new build time
722 option --with-pamconfdir=no to disable installation of the PAM
725 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
726 capability is now also dropped from its set (in addition to
727 CAP_SYS_MKNOD as before).
729 * In service unit files it is now possible to connect a specific named
730 file descriptor with stdin/stdout/stdout of an executed service. The
731 name may be specified in matching .socket units using the
732 FileDescriptorName= setting.
734 * A number of journal settings may now be configured on the kernel
735 command line. Specifically, the following options are now understood:
736 systemd.journald.max_level_console=,
737 systemd.journald.max_level_store=,
738 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
739 systemd.journald.max_level_wall=.
741 * "systemctl is-enabled --full" will now show by which symlinks a unit
742 file is enabled in the unit dependency tree.
744 * Support for VeraCrypt encrypted partitions has been added to the
745 "cryptsetup" logic and /etc/crypttab.
747 * systemd-detect-virt gained support for a new --private-users switch
748 that checks whether the invoking processes are running inside a user
749 namespace. Similar, a new special value "private-users" for the
750 existing ConditionVirtualization= setting has been added, permitting
751 skipping of specific units in user namespace environments.
753 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
754 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
755 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
756 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
757 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
758 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
759 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
760 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
761 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
762 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
763 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
764 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
765 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
766 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
767 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
768 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
769 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
770 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
771 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
772 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
773 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
774 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
775 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
776 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
777 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
778 Jędrzejewski-Szmek, Zeal Jagannatha
780 — Santa Fe, 2016-11-03
784 * In service units the various ExecXYZ= settings have been extended
785 with an additional special character as first argument of the
786 assigned value: if the character '+' is used the specified command
787 line it will be run with full privileges, regardless of User=,
788 Group=, CapabilityBoundingSet= and similar options. The effect is
789 similar to the existing PermissionsStartOnly= option, but allows
790 configuration of this concept for each executed command line
793 * Services may now alter the service watchdog timeout at runtime by
794 sending a WATCHDOG_USEC= message via sd_notify().
796 * MemoryLimit= and related unit settings now optionally take percentage
797 specifications. The percentage is taken relative to the amount of
798 physical memory in the system (or in case of containers, the assigned
799 amount of memory). This allows scaling service resources neatly with
800 the amount of RAM available on the system. Similarly, systemd-logind's
801 RuntimeDirectorySize= option now also optionally takes percentage
804 * In similar fashion TasksMax= takes percentage values now, too. The
805 value is taken relative to the configured maximum number of processes
806 on the system. The per-service task maximum has been changed to 15%
807 using this functionality. (Effectively this is an increase of 512 →
808 4915 for service units, given the kernel's default pid_max setting.)
810 * Calendar time specifications in .timer units now understand a ".."
811 syntax for time ranges. Example: "4..7:10" may now be used for
812 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
815 * The InaccessableDirectories=, ReadOnlyDirectories= and
816 ReadWriteDirectories= unit file settings have been renamed to
817 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
818 applied to all kinds of file nodes, and not just directories, with
819 the exception of symlinks. Specifically these settings may now be
820 used on block and character device nodes, UNIX sockets and FIFOS as
821 well as regular files. The old names of these settings remain
822 available for compatibility.
824 * systemd will now log about all service processes it kills forcibly
825 (using SIGKILL) because they remained after the clean shutdown phase
826 of the service completed. This should help identifying services that
827 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
828 systemd-logind's configuration a similar log message is generated for
829 processes killed at the end of each session due to this setting.
831 * systemd will now set the $JOURNAL_STREAM environment variable for all
832 services whose stdout/stderr are connected to the Journal (which
833 effectively means by default: all services). The variable contains
834 the device and inode number of the file descriptor used for
835 stdout/stderr. This may be used by invoked programs to detect whether
836 their stdout/stderr is connected to the Journal, in which case they
837 can switch over to direct Journal communication, thus being able to
838 pass extended, structured metadata along with their log messages. As
839 one example, this is now used by glib's logging primitives.
841 * When using systemd's default tmp.mount unit for /tmp, the mount point
842 will now be established with the "nosuid" and "nodev" options. This
843 avoids privilege escalation attacks that put traps and exploits into
844 /tmp. However, this might cause problems if you e. g. put container
845 images or overlays into /tmp; if you need this, override tmp.mount's
846 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
849 * systemd now supports the "memory" cgroup controller also on
852 * The systemd-cgtop tool now optionally takes a control group path as
853 command line argument. If specified, the control group list shown is
854 limited to subgroups of that group.
856 * The SystemCallFilter= unit file setting gained support for
857 pre-defined, named system call filter sets. For example
858 SystemCallFilter=@clock is now an effective way to make all clock
859 changing-related system calls unavailable to a service. A number of
860 similar pre-defined groups are defined. Writing system call filters
861 for system services is simplified substantially with this new
862 concept. Accordingly, all of systemd's own, long-running services now
863 enable system call filtering based on this, by default.
865 * A new service setting MemoryDenyWriteExecute= has been added, taking
866 a boolean value. If turned on, a service may no longer create memory
867 mappings that are writable and executable at the same time. This
868 enhances security for services where this is enabled as it becomes
869 harder to dynamically write and then execute memory in exploited
870 service processes. This option has been enabled for all of systemd's
871 own long-running services.
873 * A new RestrictRealtime= service setting has been added, taking a
874 boolean argument. If set the service's processes may no longer
875 acquire realtime scheduling. This improves security as realtime
876 scheduling may otherwise be used to easily freeze the system.
878 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
879 value. This may be used for requesting that the system manager inside
880 of the container reports start-up completion to nspawn which then
881 propagates this notification further to the service manager
882 supervising nspawn itself. A related option NotifyReady= in .nspawn
883 files has been added too. This functionality allows ordering of the
884 start-up of multiple containers using the usual systemd ordering
887 * machinectl gained a new command "stop" that is an alias for
890 * systemd-resolved gained support for contacting DNS servers on
891 link-local IPv6 addresses.
893 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
894 its caches. A method call for requesting the same operation has been
895 added to the bus API too, and is made available via "systemd-resolve
898 * systemd-resolve gained a new --status switch. If passed a brief
899 summary of the used DNS configuration with per-interface information
902 * resolved.conf gained a new Cache= boolean option, defaulting to
903 on. If turned off local DNS caching is disabled. This comes with a
904 performance penalty in particular when DNSSEC is enabled. Note that
905 resolved disables its internal caching implicitly anyway, when the
906 configured DNS server is on a host-local IP address such as ::1 or
907 127.0.0.1, thus automatically avoiding double local caching.
909 * systemd-resolved now listens on the local IP address 127.0.0.53:53
910 for DNS requests. This improves compatibility with local programs
911 that do not use the libc NSS or systemd-resolved's bus APIs for name
912 resolution. This minimal DNS service is only available to local
913 programs and does not implement the full DNS protocol, but enough to
914 cover local DNS clients. A new, static resolv.conf file, listing just
915 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
916 now recommended to make /etc/resolv.conf a symlink to this file in
917 order to route all DNS lookups to systemd-resolved, regardless if
918 done via NSS, the bus API or raw DNS packets. Note that this local
919 DNS service is not as fully featured as the libc NSS or
920 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
921 used to deliver link-local address information (as this implies
922 sending a local interface index along), LLMNR/mDNS support via this
923 interface is severely restricted. It is thus strongly recommended for
924 all applications to use the libc NSS API or native systemd-resolved
927 * systemd-networkd's bridge support learned a new setting
928 VLANFiltering= for controlling VLAN filtering. Moreover a new section
929 in .network files has been added for configuring VLAN bridging in
930 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
932 * systemd-networkd's IPv6 Router Advertisement code now makes use of
933 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
934 now be acquired without relying on DHCPv6. Two new options
935 UseDomains= and UseDNS= have been added to configure this behaviour.
937 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
938 renamed IPv6AcceptRA=, without altering its behaviour. The old
939 setting name remains available for compatibility reasons.
941 * The systemd-networkd VTI/VTI6 tunneling support gained new options
942 Key=, InputKey= and OutputKey=.
944 * systemd-networkd gained support for VRF ("Virtual Routing Function")
945 interface configuration.
947 * "systemctl edit" may now be used to create new unit files by
948 specifying the --force switch.
950 * sd-event gained a new function sd_event_get_iteration() for
951 requesting the current iteration counter of the event loop. It starts
952 at zero and is increased by one with each event loop iteration.
954 * A new rpm macro %systemd_ordering is provided by the macros.systemd
955 file. It can be used in lieu of %systemd_requires in packages which
956 don't use any systemd functionality and are intended to be installed
957 in minimal containers without systemd present. This macro provides
958 ordering dependencies to ensure that if the package is installed in
959 the same rpm transaction as systemd, systemd will be installed before
960 the scriptlets for the package are executed, allowing unit presets
963 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
964 been added to simplify packaging of generators.
966 * The os-release file gained VERSION_CODENAME field for the
967 distribution nickname (e.g. VERSION_CODENAME=woody).
969 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
970 can be set to disable parsing of metadata and the creation
971 of persistent symlinks for that device.
973 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
974 to make them available to logged-in users has been reverted.
976 * Much of the common code of the various systemd components is now
977 built into an internal shared library libsystemd-shared-231.so
978 (incorporating the systemd version number in the name, to be updated
979 with future releases) that the components link to. This should
980 decrease systemd footprint both in memory during runtime and on
981 disk. Note that the shared library is not for public use, and is
982 neither API not ABI stable, but is likely to change with every new
983 released update. Packagers need to make sure that binaries
984 linking to libsystemd-shared.so are updated in step with the
987 * Configuration for "mkosi" is now part of the systemd
988 repository. mkosi is a tool to easily build legacy-free OS images,
989 and is available on github: https://github.com/systemd/mkosi. If
990 "mkosi" is invoked in the build tree a new raw OS image is generated
991 incorporating the systemd sources currently being worked on and a
992 clean, fresh distribution installation. The generated OS image may be
993 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
994 UEFI PC. This functionality is particularly useful to easily test
995 local changes made to systemd in a pristine, defined environment. See
998 * configure learned the --with-support-url= option to specify the
999 distribution's bugtracker.
1001 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
1002 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
1003 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
1004 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
1005 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
1006 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
1007 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
1008 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
1009 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
1010 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
1011 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
1012 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
1013 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
1014 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
1015 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
1016 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
1017 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
1018 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
1019 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
1021 — Berlin, 2016-07-25
1025 * DNSSEC is now turned on by default in systemd-resolved (in
1026 "allow-downgrade" mode), but may be turned off during compile time by
1027 passing "--with-default-dnssec=no" to "configure" (and of course,
1028 during runtime with DNSSEC= in resolved.conf). We recommend
1029 downstreams to leave this on at least during development cycles and
1030 report any issues with the DNSSEC logic upstream. We are very
1031 interested in collecting feedback about the DNSSEC validator and its
1032 limitations in the wild. Note however, that DNSSEC support is
1033 probably nothing downstreams should turn on in stable distros just
1034 yet, as it might create incompatibilities with a few DNS servers and
1035 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
1036 automatically whenever we detect such incompatible setups, but there
1037 might be systems we do not cover yet. Hence: please help us testing
1038 the DNSSEC code, leave this on where you can, report back, but then
1039 again don't consider turning this on in your stable, LTS or
1040 production release just yet. (Note that you have to enable
1041 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
1042 and its DNSSEC mode for host name resolution from local
1045 * systemd-resolve conveniently resolves DANE records with the --tlsa
1046 option and OPENPGPKEY records with the --openpgp option. It also
1047 supports dumping raw DNS record data via the new --raw= switch.
1049 * systemd-logind will now by default terminate user processes that are
1050 part of the user session scope unit (session-XX.scope) when the user
1051 logs out. This behavior is controlled by the KillUserProcesses=
1052 setting in logind.conf, and the previous default of "no" is now
1053 changed to "yes". This means that user sessions will be properly
1054 cleaned up after, but additional steps are necessary to allow
1055 intentionally long-running processes to survive logout.
1057 While the user is logged in at least once, user@.service is running,
1058 and any service that should survive the end of any individual login
1059 session can be started at a user service or scope using systemd-run.
1060 systemd-run(1) man page has been extended with an example which shows
1061 how to run screen in a scope unit underneath user@.service. The same
1062 command works for tmux.
1064 After the user logs out of all sessions, user@.service will be
1065 terminated too, by default, unless the user has "lingering" enabled.
1066 To effectively allow users to run long-term tasks even if they are
1067 logged out, lingering must be enabled for them. See loginctl(1) for
1068 details. The default polkit policy was modified to allow users to
1069 set lingering for themselves without authentication.
1071 Previous defaults can be restored at compile time by the
1072 --without-kill-user-processes option to "configure".
1074 * systemd-logind gained new configuration settings SessionsMax= and
1075 InhibitorsMax=, both with a default of 8192. It will not register new
1076 user sessions or inhibitors above this limit.
1078 * systemd-logind will now reload configuration on SIGHUP.
1080 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
1081 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
1082 enable. Also, support for the "io" cgroup controller in the unified
1083 hierarchy has been added, so that the "memory", "pids" and "io" are
1084 now the controllers that are supported on the unified hierarchy.
1086 WARNING: it is not possible to use previous systemd versions with
1087 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
1088 is necessary to also update systemd in the initramfs if using the
1089 unified hierarchy. An updated SELinux policy is also required.
1091 * LLDP support has been extended, and both passive (receive-only) and
1092 active (sender) modes are supported. Passive mode ("routers-only") is
1093 enabled by default in systemd-networkd. Active LLDP mode is enabled
1094 by default for containers on the internal network. The "networkctl
1095 lldp" command may be used to list information gathered. "networkctl
1096 status" will also show basic LLDP information on connected peers now.
1098 * The IAID and DUID unique identifier sent in DHCP requests may now be
1099 configured for the system and each .network file managed by
1100 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
1102 * systemd-networkd gained support for configuring proxy ARP support for
1103 each interface, via the ProxyArp= setting in .network files. It also
1104 gained support for configuring the multicast querier feature of
1105 bridge devices, via the new MulticastQuerier= setting in .netdev
1106 files. Similarly, snooping on the IGMP traffic can be controlled
1107 via the new setting MulticastSnooping=.
1109 A new setting PreferredLifetime= has been added for addresses
1110 configured in .network file to configure the lifetime intended for an
1113 The systemd-networkd DHCP server gained the option EmitRouter=, which
1114 defaults to yes, to configure whether the DHCP Option 3 (Router)
1117 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
1118 systemd-socket-activate and installed into /usr/bin. It is now fully
1121 * systemd-journald now uses separate threads to flush changes to disk
1122 when closing journal files, thus reducing impact of slow disk I/O on
1123 logging performance.
1125 * The sd-journal API gained two new calls
1126 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
1127 can be used to open journal files using file descriptors instead of
1128 file or directory paths. sd_journal_open_container() has been
1129 deprecated, sd_journal_open_directory_fd() should be used instead
1130 with the flag SD_JOURNAL_OS_ROOT.
1132 * journalctl learned a new output mode "-o short-unix" that outputs log
1133 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
1134 UTC). It also gained support for a new --no-hostname setting to
1135 suppress the hostname column in the family of "short" output modes.
1137 * systemd-ask-password now optionally skips printing of the password to
1138 stdout with --no-output which can be useful in scripts.
1140 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
1141 (devices tagged with ID_MAKER_TOOL) are now tagged with
1142 "uaccess" and are available to logged in users.
1144 * The DeviceAllow= unit setting now supports specifiers (with "%").
1146 * "systemctl show" gained a new --value switch, which allows print a
1147 only the contents of a specific unit property, without also printing
1148 the property's name. Similar support was added to "show*" verbs
1149 of loginctl and machinectl that output "key=value" lists.
1151 * A new unit type "generated" was added for files dynamically generated
1152 by generator tools. Similarly, a new unit type "transient" is used
1153 for unit files created using the runtime API. "systemctl enable" will
1154 refuse to operate on such files.
1156 * A new command "systemctl revert" has been added that may be used to
1157 revert to the vendor version of a unit file, in case local changes
1158 have been made by adding drop-ins or overriding the unit file.
1160 * "machinectl clean" gained a new verb to automatically remove all or
1161 just hidden container images.
1163 * systemd-tmpfiles gained support for a new line type "e" for emptying
1164 directories, if they exist, without creating them if they don't.
1166 * systemd-nspawn gained support for automatically patching the UID/GIDs
1167 of the owners and the ACLs of all files and directories in a
1168 container tree to match the UID/GID user namespacing range selected
1169 for the container invocation. This mode is enabled via the new
1170 --private-users-chown switch. It also gained support for
1171 automatically choosing a free, previously unused UID/GID range when
1172 starting a container, via the new --private-users=pick setting (which
1173 implies --private-users-chown). Together, these options for the first
1174 time make user namespacing for nspawn containers fully automatic and
1175 thus deployable. The systemd-nspawn@.service template unit file has
1176 been changed to use this functionality by default.
1178 * systemd-nspawn gained a new --network-zone= switch, that allows
1179 creating ad-hoc virtual Ethernet links between multiple containers,
1180 that only exist as long as at least one container referencing them is
1181 running. This allows easy connecting of multiple containers with a
1182 common link that implements an Ethernet broadcast domain. Each of
1183 these network "zones" may be named relatively freely by the user, and
1184 may be referenced by any number of containers, but each container may
1185 only reference one of these "zones". On the lower level, this is
1186 implemented by an automatically managed bridge network interface for
1187 each zone, that is created when the first container referencing its
1188 zone is created and removed when the last one referencing its zone
1191 * The default start timeout may now be configured on the kernel command
1192 line via systemd.default_timeout_start_sec=. It was already
1193 configurable via the DefaultTimeoutStartSec= option in
1194 /etc/systemd/system.conf.
1196 * Socket units gained a new TriggerLimitIntervalSec= and
1197 TriggerLimitBurst= setting to configure a limit on the activation
1198 rate of the socket unit.
1200 * The LimitNICE= setting now optionally takes normal UNIX nice values
1201 in addition to the raw integer limit value. If the specified
1202 parameter is prefixed with "+" or "-" and is in the range -20..19 the
1203 value is understood as UNIX nice value. If not prefixed like this it
1204 is understood as raw RLIMIT_NICE limit.
1206 * Note that the effect of the PrivateDevices= unit file setting changed
1207 slightly with this release: the per-device /dev file system will be
1208 mounted read-only from this version on, and will have "noexec"
1209 set. This (minor) change of behavior might cause some (exceptional)
1210 legacy software to break, when PrivateDevices=yes is set for its
1211 service. Please leave PrivateDevices= off if you run into problems
1214 * systemd-bootchart has been split out to a separate repository:
1215 https://github.com/systemd/systemd-bootchart
1217 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
1218 merged into the kernel in its current form.
1220 * The compatibility libraries libsystemd-daemon.so,
1221 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
1222 which have been deprecated since systemd-209 have been removed along
1223 with the corresponding pkg-config files. All symbols provided by
1224 those libraries are provided by libsystemd.so.
1226 * The Capabilities= unit file setting has been removed (it is ignored
1227 for backwards compatibility). AmbientCapabilities= and
1228 CapabilityBoundingSet= should be used instead.
1230 * A new special target has been added, initrd-root-device.target,
1231 which creates a synchronization point for dependencies of the root
1232 device in early userspace. Initramfs builders must ensure that this
1233 target is now included in early userspace.
1235 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
1236 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
1237 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
1238 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
1239 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
1240 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
1241 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
1242 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
1243 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
1244 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
1245 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
1246 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
1247 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
1248 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
1249 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
1250 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
1251 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
1252 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
1253 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
1254 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
1255 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
1256 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
1257 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
1258 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
1259 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
1262 — Fairfax, 2016-05-21
1266 * The systemd-resolved DNS resolver service has gained a substantial
1267 set of new features, most prominently it may now act as a DNSSEC
1268 validating stub resolver. DNSSEC mode is currently turned off by
1269 default, but is expected to be turned on by default in one of the
1270 next releases. For now, we invite everybody to test the DNSSEC logic
1271 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
1272 service also gained a full set of D-Bus interfaces, including calls
1273 to configure DNS and DNSSEC settings per link (for use by external
1274 network management software). systemd-resolved and systemd-networkd
1275 now distinguish between "search" and "routing" domains. The former
1276 are used to qualify single-label names, the latter are used purely
1277 for routing lookups within certain domains to specific links.
1278 resolved now also synthesizes RRs for all entries from /etc/hosts.
1280 * The systemd-resolve tool (which is a client utility for
1281 systemd-resolved) has been improved considerably and is now fully
1282 supported and documented. Hence it has moved from /usr/lib/systemd to
1285 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
1288 * The coredump collection logic has been reworked: when a coredump is
1289 collected it is now written to disk, compressed and processed
1290 (including stacktrace extraction) from a new instantiated service
1291 systemd-coredump@.service, instead of directly from the
1292 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
1293 processing large coredumps can take up a substantial amount of
1294 resources and time, and this previously happened entirely outside of
1295 systemd's service supervision. With the new logic the core_pattern
1296 hook only does minimal metadata collection before passing off control
1297 to the new instantiated service, which is configured with a time
1298 limit, a nice level and other settings to minimize negative impact on
1299 the rest of the system. Also note that the new logic will honour the
1300 RLIMIT_CORE setting of the crashed process, which now allows users
1301 and processes to turn off coredumping for their processes by setting
1304 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
1305 and all forked processes by default. Previously, PID 1 would leave
1306 the setting at "0" for all processes, as set by the kernel. Note that
1307 the resource limit traditionally has no effect on the generated
1308 coredumps on the system if the /proc/sys/kernel/core_pattern hook
1309 logic is used. Since the limit is now honoured (see above) its
1310 default has been changed so that the coredumping logic is enabled by
1311 default for all processes, while allowing specific opt-out.
1313 * When the stacktrace is extracted from processes of system users, this
1314 is now done as "systemd-coredump" user, in order to sandbox this
1315 potentially security sensitive parsing operation. (Note that when
1316 processing coredumps of normal users this is done under the user ID
1317 of process that crashed, as before.) Packagers should take notice
1318 that it is now necessary to create the "systemd-coredump" system user
1319 and group at package installation time.
1321 * The systemd-activate socket activation testing tool gained support
1322 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
1323 and --seqpacket switches. It also has been extended to support both
1324 new-style and inetd-style file descriptor passing. Use the new
1325 --inetd switch to request inetd-style file descriptor passing.
1327 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
1328 variable, which takes a boolean value. If set to false, ANSI color
1329 output is disabled in the tools even when run on a terminal that
1332 * The VXLAN support in networkd now supports two new settings
1333 DestinationPort= and PortRange=.
1335 * A new systemd.machine_id= kernel command line switch has been added,
1336 that may be used to set the machine ID in /etc/machine-id if it is
1337 not initialized yet. This command line option has no effect if the
1338 file is already initialized.
1340 * systemd-nspawn gained a new --as-pid2 switch that invokes any
1341 specified command line as PID 2 rather than PID 1 in the
1342 container. In this mode PID 1 is a minimal stub init process that
1343 implements the special POSIX and Linux semantics of PID 1 regarding
1344 signal and child process management. Note that this stub init process
1345 is implemented in nspawn itself and requires no support from the
1346 container image. This new logic is useful to support running
1347 arbitrary commands in the container, as normal processes are
1348 generally not prepared to run as PID 1.
1350 * systemd-nspawn gained a new --chdir= switch for setting the current
1351 working directory for the process started in the container.
1353 * "journalctl /dev/sda" will now output all kernel log messages for
1354 specified device from the current boot, in addition to all devices
1355 that are parents of it. This should make log output about devices
1356 pretty useful, as long as kernel drivers attach enough metadata to
1357 the log messages. (The usual SATA drivers do.)
1359 * The sd-journal API gained two new calls
1360 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
1361 that report whether log data from /run or /var has been found.
1363 * journalctl gained a new switch "--fields" that prints all journal
1364 record field names currently in use in the journal. This is backed
1365 by two new sd-journal API calls sd_journal_enumerate_fields() and
1366 sd_journal_restart_fields().
1368 * Most configurable timeouts in systemd now expect an argument of
1369 "infinity" to turn them off, instead of "0" as before. The semantics
1370 from now on is that a timeout of "0" means "now", and "infinity"
1371 means "never". To maintain backwards compatibility, "0" continues to
1372 turn off previously existing timeout settings.
1374 * "systemctl reload-or-try-restart" has been renamed to "systemctl
1375 try-reload-or-restart" to clarify what it actually does: the "try"
1376 logic applies to both reloading and restarting, not just restarting.
1377 The old name continues to be accepted for compatibility.
1379 * On boot-up, when PID 1 detects that the system clock is behind the
1380 release date of the systemd version in use, the clock is now set
1381 to the latter. Previously, this was already done in timesyncd, in order
1382 to avoid running with clocks set to the various clock epochs such as
1383 1902, 1938 or 1970. With this change the logic is now done in PID 1
1384 in addition to timesyncd during early boot-up, so that it is enforced
1385 before the first process is spawned by systemd. Note that the logic
1386 in timesyncd remains, as it is more comprehensive and ensures
1387 clock monotonicity by maintaining a persistent timestamp file in
1388 /var. Since /var is generally not available in earliest boot or the
1389 initrd, this part of the logic remains in timesyncd, and is not done
1392 * Support for tweaking details in net_cls.class_id through the
1393 NetClass= configuration directive has been removed, as the kernel
1394 people have decided to deprecate that controller in cgroup v2.
1395 Userspace tools such as nftables are moving over to setting rules
1396 that are specific to the full cgroup path of a task, which obsoletes
1397 these controllers anyway. The NetClass= directive is kept around for
1398 legacy compatibility reasons. For a more in-depth description of the
1399 kernel change, please refer to the respective upstream commit:
1401 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
1403 * A new service setting RuntimeMaxSec= has been added that may be used
1404 to specify a maximum runtime for a service. If the timeout is hit, the
1405 service is terminated and put into a failure state.
1407 * A new service setting AmbientCapabilities= has been added. It allows
1408 configuration of additional Linux process capabilities that are
1409 passed to the activated processes. This is only available on very
1412 * The process resource limit settings in service units may now be used
1413 to configure hard and soft limits individually.
1415 * The various libsystemd APIs such as sd-bus or sd-event now publicly
1416 expose support for gcc's __attribute__((cleanup())) C extension.
1417 Specifically, for many object destructor functions alternative
1418 versions have been added that have names suffixed with "p" and take a
1419 pointer to a pointer to the object to destroy, instead of just a
1420 pointer to the object itself. This is useful because these destructor
1421 functions may be used directly as parameters to the cleanup
1422 construct. Internally, systemd has been a heavy user of this GCC
1423 extension for a long time, and with this change similar support is
1424 now available to consumers of the library outside of systemd. Note
1425 that by using this extension in your sources compatibility with old
1426 and strictly ANSI compatible C compilers is lost. However, all gcc or
1427 LLVM versions of recent years support this extension.
1429 * Timer units gained support for a new setting RandomizedDelaySec= that
1430 allows configuring some additional randomized delay to the configured
1431 time. This is useful to spread out timer events to avoid load peaks in
1432 clusters or larger setups.
1434 * Calendar time specifications now support sub-second accuracy.
1436 * Socket units now support listening on SCTP and UDP-lite protocol
1439 * The sd-event API now comes with a full set of man pages.
1441 * Older versions of systemd contained experimental support for
1442 compressing journal files and coredumps with the LZ4 compressor that
1443 was not compatible with the lz4 binary (due to API limitations of the
1444 lz4 library). This support has been removed; only support for files
1445 compatible with the lz4 binary remains. This LZ4 logic is now
1446 officially supported and no longer considered experimental.
1448 * The dkr image import logic has been removed again from importd. dkr's
1449 micro-services focus doesn't fit into the machine image focus of
1450 importd, and quickly got out of date with the upstream dkr API.
1452 * Creation of the /run/lock/lockdev/ directory was dropped from
1453 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
1454 been available for many years. If you still need this, you need to
1455 create your own tmpfiles.d config file with:
1457 d /run/lock/lockdev 0775 root lock -
1459 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
1460 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
1461 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
1462 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
1463 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
1464 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
1465 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
1466 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
1467 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
1468 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
1469 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
1470 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
1471 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
1472 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
1473 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
1474 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
1475 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
1476 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
1477 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
1479 — Berlin, 2016-02-11
1483 * A number of properties previously only settable in unit
1484 files are now also available as properties to set when
1485 creating transient units programmatically via the bus, as it
1486 is exposed with systemd-run's --property=
1487 setting. Specifically, these are: SyslogIdentifier=,
1488 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
1489 EnvironmentFile=, ReadWriteDirectories=,
1490 ReadOnlyDirectories=, InaccessibleDirectories=,
1491 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
1493 * When creating transient services via the bus API it is now
1494 possible to pass in a set of file descriptors to use as
1495 STDIN/STDOUT/STDERR for the invoked process.
1497 * Slice units may now be created transiently via the bus APIs,
1498 similar to the way service and scope units may already be
1499 created transiently.
1501 * Wherever systemd expects a calendar timestamp specification
1502 (like in journalctl's --since= and --until= switches) UTC
1503 timestamps are now supported. Timestamps suffixed with "UTC"
1504 are now considered to be in Universal Time Coordinated
1505 instead of the local timezone. Also, timestamps may now
1506 optionally be specified with sub-second accuracy. Both of
1507 these additions also apply to recurring calendar event
1508 specification, such as OnCalendar= in timer units.
1510 * journalctl gained a new "--sync" switch that asks the
1511 journal daemon to write all so far unwritten log messages to
1512 disk and sync the files, before returning.
1514 * systemd-tmpfiles learned two new line types "q" and "Q" that
1515 operate like "v", but also set up a basic btrfs quota
1516 hierarchy when used on a btrfs file system with quota
1519 * tmpfiles' "v", "q" and "Q" will now create a plain directory
1520 instead of a subvolume (even on a btrfs file system) if the
1521 root directory is a plain directory, and not a
1522 subvolume. This should simplify things with certain chroot()
1523 environments which are not aware of the concept of btrfs
1526 * systemd-detect-virt gained a new --chroot switch to detect
1527 whether execution takes place in a chroot() environment.
1529 * CPUAffinity= now takes CPU index ranges in addition to
1532 * The various memory-related resource limit settings (such as
1533 LimitAS=) now understand the usual K, M, G, ... suffixes to
1534 the base of 1024 (IEC). Similar, the time-related resource
1535 limit settings understand the usual min, h, day, ...
1538 * There's a new system.conf setting DefaultTasksMax= to
1539 control the default TasksMax= setting for services and
1540 scopes running on the system. (TasksMax= is the primary
1541 setting that exposes the "pids" cgroup controller on systemd
1542 and was introduced in the previous systemd release.) The
1543 setting now defaults to 512, which means services that are
1544 not explicitly configured otherwise will only be able to
1545 create 512 processes or threads at maximum, from this
1546 version on. Note that this means that thread- or
1547 process-heavy services might need to be reconfigured to set
1548 TasksMax= to a higher value. It is sufficient to set
1549 TasksMax= in these specific unit files to a higher value, or
1550 even "infinity". Similar, there's now a logind.conf setting
1551 UserTasksMax= that defaults to 4096 and limits the total
1552 number of processes or tasks each user may own
1553 concurrently. nspawn containers also have the TasksMax=
1554 value set by default now, to 8192. Note that all of this
1555 only has an effect if the "pids" cgroup controller is
1556 enabled in the kernel. The general benefit of these changes
1557 should be a more robust and safer system, that provides a
1558 certain amount of per-service fork() bomb protection.
1560 * systemd-nspawn gained the new --network-veth-extra= switch
1561 to define additional and arbitrarily-named virtual Ethernet
1562 links between the host and the container.
1564 * A new service execution setting PassEnvironment= has been
1565 added that allows importing select environment variables
1566 from PID1's environment block into the environment block of
1569 * Timer units gained support for a new RemainAfterElapse=
1570 setting which takes a boolean argument. It defaults to on,
1571 exposing behaviour unchanged to previous releases. If set to
1572 off, timer units are unloaded after they elapsed if they
1573 cannot elapse again. This is particularly useful for
1574 transient timer units, which shall not stay around longer
1575 than until they first elapse.
1577 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
1578 default now (the kernel default is 16). This is beneficial
1579 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
1580 allows substantially larger numbers of queued
1581 datagrams. This should increase the capability of systemd to
1582 parallelize boot-up, as logging and sd_notify() are unlikely
1583 to stall execution anymore. If you need to change the value
1584 from the new defaults, use the usual sysctl.d/ snippets.
1586 * The compression framing format used by the journal or
1587 coredump processing has changed to be in line with what the
1588 official LZ4 tools generate. LZ4 compression support in
1589 systemd was considered unsupported previously, as the format
1590 was not compatible with the normal tools. With this release
1591 this has changed now, and it is hence safe for downstream
1592 distributions to turn it on. While not compressing as well
1593 as the XZ, LZ4 is substantially faster, which makes
1594 it a good default choice for the compression logic in the
1595 journal and in coredump handling.
1597 * Any reference to /etc/mtab has been dropped from
1598 systemd. The file has been obsolete since a while, but
1599 systemd refused to work on systems where it was incorrectly
1600 set up (it should be a symlink or non-existent). Please make
1601 sure to update to util-linux 2.27.1 or newer in conjunction
1602 with this systemd release, which also drops any reference to
1603 /etc/mtab. If you maintain a distribution make sure that no
1604 software you package still references it, as this is a
1605 likely source of bugs. There's also a glibc bug pending,
1606 asking for removal of any reference to this obsolete file:
1608 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
1610 Note that only util-linux versions built with
1611 --enable-libmount-force-mountinfo are supported.
1613 * Support for the ".snapshot" unit type has been removed. This
1614 feature turned out to be little useful and little used, and
1615 has now been removed from the core and from systemctl.
1617 * The dependency types RequiresOverridable= and
1618 RequisiteOverridable= have been removed from systemd. They
1619 have been used only very sparingly to our knowledge and
1620 other options that provide a similar effect (such as
1621 systemctl --mode=ignore-dependencies) are much more useful
1622 and commonly used. Moreover, they were only half-way
1623 implemented as the option to control behaviour regarding
1624 these dependencies was never added to systemctl. By removing
1625 these dependency types the execution engine becomes a bit
1626 simpler. Unit files that use these dependencies should be
1627 changed to use the non-Overridable dependency types
1628 instead. In fact, when parsing unit files with these
1629 options, that's what systemd will automatically convert them
1630 too, but it will also warn, asking users to fix the unit
1631 files accordingly. Removal of these dependency types should
1632 only affect a negligible number of unit files in the wild.
1634 * Behaviour of networkd's IPForward= option changed
1635 (again). It will no longer maintain a per-interface setting,
1636 but propagate one way from interfaces where this is enabled
1637 to the global kernel setting. The global setting will be
1638 enabled when requested by a network that is set up, but
1639 never be disabled again. This change was made to make sure
1640 IPv4 and IPv6 behaviour regarding packet forwarding is
1641 similar (as the Linux IPv6 stack does not support
1642 per-interface control of this setting) and to minimize
1645 * In unit files the behaviour of %u, %U, %h, %s has
1646 changed. These specifiers will now unconditionally resolve
1647 to the various user database fields of the user that the
1648 systemd instance is running as, instead of the user
1649 configured in the specific unit via User=. Note that this
1650 effectively doesn't change much, as resolving of these
1651 specifiers was already turned off in the --system instance
1652 of systemd, as we cannot do NSS lookups from PID 1. In the
1653 --user instance of systemd these specifiers where correctly
1654 resolved, but hardly made any sense, since the user instance
1655 lacks privileges to do user switches anyway, and User= is
1656 hence useless. Moreover, even in the --user instance of
1657 systemd behaviour was awkward as it would only take settings
1658 from User= assignment placed before the specifier into
1659 account. In order to unify and simplify the logic around
1660 this the specifiers will now always resolve to the
1661 credentials of the user invoking the manager (which in case
1662 of PID 1 is the root user).
1664 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
1665 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
1666 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
1667 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
1668 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
1669 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
1670 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
1671 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
1672 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
1673 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
1674 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
1675 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
1676 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
1677 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
1680 — Berlin, 2015-11-18
1684 * systemd now depends on util-linux v2.27. More specifically,
1685 the newly added mount monitor feature in libmount now
1686 replaces systemd's former own implementation.
1688 * libmount mandates /etc/mtab not to be regular file, and
1689 systemd now enforces this condition at early boot.
1690 /etc/mtab has been deprecated and warned about for a very
1691 long time, so systems running systemd should already have
1692 stopped having this file around as anything else than a
1693 symlink to /proc/self/mounts.
1695 * Support for the "pids" cgroup controller has been added. It
1696 allows accounting the number of tasks in a cgroup and
1697 enforcing limits on it. This adds two new setting
1698 TasksAccounting= and TasksMax= to each unit, as well as a
1699 global option DefaultTasksAccounting=.
1701 * Support for the "net_cls" cgroup controller has been added.
1702 It allows assigning a net class ID to each task in the
1703 cgroup, which can then be used in firewall rules and traffic
1704 shaping configurations. Note that the kernel netfilter net
1705 class code does not currently work reliably for ingress
1706 packets on unestablished sockets.
1708 This adds a new config directive called NetClass= to CGroup
1709 enabled units. Allowed values are positive numbers for fixed
1710 assignments and "auto" for picking a free value
1713 * 'systemctl is-system-running' now returns 'offline' if the
1714 system is not booted with systemd. This command can now be
1715 used as a substitute for 'systemd-notify --booted'.
1717 * Watchdog timeouts have been increased to 3 minutes for all
1718 in-tree service files. Apparently, disk IO issues are more
1719 frequent than we hoped, and user reported >1 minute waiting
1722 * 'machine-id-commit' functionality has been merged into
1723 'machine-id-setup --commit'. The separate binary has been
1726 * The WorkingDirectory= directive in unit files may now be set
1727 to the special value '~'. In this case, the working
1728 directory is set to the home directory of the user
1729 configured in User=.
1731 * "machinectl shell" will now open the shell in the home
1732 directory of the selected user by default.
1734 * The CrashChVT= configuration file setting is renamed to
1735 CrashChangeVT=, following our usual logic of not
1736 abbreviating unnecessarily. The old directive is still
1737 supported for compat reasons. Also, this directive now takes
1738 an integer value between 1 and 63, or a boolean value. The
1739 formerly supported '-1' value for disabling stays around for
1742 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
1743 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
1744 RootDirectory= properties can now be set for transient
1747 * The systemd-analyze tool gained a new "set-log-target" verb
1748 to change the logging target the system manager logs to
1749 dynamically during runtime. This is similar to how
1750 "systemd-analyze set-log-level" already changes the log
1753 * In nspawn /sys is now mounted as tmpfs, with only a selected
1754 set of subdirectories mounted in from the real sysfs. This
1755 enhances security slightly, and is useful for ensuring user
1756 namespaces work correctly.
1758 * Support for USB FunctionFS activation has been added. This
1759 allows implementation of USB gadget services that are
1760 activated as soon as they are requested, so that they don't
1761 have to run continuously, similar to classic socket
1764 * The "systemctl exit" command now optionally takes an
1765 additional parameter that sets the exit code to return from
1766 the systemd manager when exiting. This is only relevant when
1767 running the systemd user instance, or when running the
1768 system instance in a container.
1770 * sd-bus gained the new API calls sd_bus_path_encode_many()
1771 and sd_bus_path_decode_many() that allow easy encoding and
1772 decoding of multiple identifier strings inside a D-Bus
1773 object path. Another new call sd_bus_default_flush_close()
1774 has been added to flush and close per-thread default
1777 * systemd-cgtop gained support for a -M/--machine= switch to
1778 show the control groups within a certain container only.
1780 * "systemctl kill" gained support for an optional --fail
1781 switch. If specified the requested operation will fail of no
1782 processes have been killed, because the unit had no
1783 processes attached, or similar.
1785 * A new systemd.crash_reboot=1 kernel command line option has
1786 been added that triggers a reboot after crashing. This can
1787 also be set through CrashReboot= in systemd.conf.
1789 * The RuntimeDirectory= setting now understands unit
1790 specifiers like %i or %f.
1792 * A new (still internal) library API sd-ipv4acd has been added,
1793 that implements address conflict detection for IPv4. It's
1794 based on code from sd-ipv4ll, and will be useful for
1795 detecting DHCP address conflicts.
1797 * File descriptors passed during socket activation may now be
1798 named. A new API sd_listen_fds_with_names() is added to
1799 access the names. The default names may be overridden,
1800 either in the .socket file using the FileDescriptorName=
1801 parameter, or by passing FDNAME= when storing the file
1802 descriptors using sd_notify().
1804 * systemd-networkd gained support for:
1806 - Setting the IPv6 Router Advertisement settings via
1807 IPv6AcceptRouterAdvertisements= in .network files.
1809 - Configuring the HelloTimeSec=, MaxAgeSec= and
1810 ForwardDelaySec= bridge parameters in .netdev files.
1812 - Configuring PreferredSource= for static routes in
1815 * The "ask-password" framework used to query for LUKS harddisk
1816 passwords or SSL passwords during boot gained support for
1817 caching passwords in the kernel keyring, if it is
1818 available. This makes sure that the user only has to type in
1819 a passphrase once if there are multiple objects to unlock
1820 with the same one. Previously, such password caching was
1821 available only when Plymouth was used; this moves the
1822 caching logic into the systemd codebase itself. The
1823 "systemd-ask-password" utility gained a new --keyname=
1824 switch to control which kernel keyring key to use for
1825 caching a password in. This functionality is also useful for
1826 enabling display managers such as gdm to automatically
1827 unlock the user's GNOME keyring if its passphrase, the
1828 user's password and the harddisk password are the same, if
1829 gdm-autologin is used.
1831 * When downloading tar or raw images using "machinectl
1832 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
1833 file is now also downloaded, if it is available and stored
1834 next to the image file.
1836 * Units of type ".socket" gained a new boolean setting
1837 Writable= which is only useful in conjunction with
1838 ListenSpecial=. If true, enables opening the specified
1839 special file in O_RDWR mode rather than O_RDONLY mode.
1841 * systemd-rfkill has been reworked to become a singleton
1842 service that is activated through /dev/rfkill on each rfkill
1843 state change and saves the settings to disk. This way,
1844 systemd-rfkill is now compatible with devices that exist
1845 only intermittendly, and even restores state if the previous
1846 system shutdown was abrupt rather than clean.
1848 * The journal daemon gained support for vacuuming old journal
1849 files controlled by the number of files that shall remain,
1850 in addition to the already existing control by size and by
1851 date. This is useful as journal interleaving performance
1852 degrades with too many separate journal files, and allows
1853 putting an effective limit on them. The new setting defaults
1854 to 100, but this may be changed by setting SystemMaxFiles=
1855 and RuntimeMaxFiles= in journald.conf. Also, the
1856 "journalctl" tool gained the new --vacuum-files= switch to
1857 manually vacuum journal files to leave only the specified
1858 number of files in place.
1860 * udev will now create /dev/disk/by-path links for ATA devices
1861 on kernels where that is supported.
1863 * Galician, Serbian, Turkish and Korean translations were added.
1865 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
1866 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
1867 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
1868 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
1869 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
1870 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
1871 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
1872 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
1873 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
1874 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
1875 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
1876 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
1877 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
1878 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
1879 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
1880 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
1881 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
1882 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
1884 — Berlin, 2015-10-07
1888 * The DHCP implementation of systemd-networkd gained a set of
1891 - The DHCP server now supports emitting DNS and NTP
1892 information. It may be enabled and configured via
1893 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
1894 and NTP information is enabled, but no servers are
1895 configured, the corresponding uplink information (if there
1896 is any) is propagated.
1898 - Server and client now support transmission and reception
1899 of timezone information. It can be configured via the
1900 newly introduced network options UseTimezone=,
1901 EmitTimezone=, and Timezone=. Transmission of timezone
1902 information is enabled between host and containers by
1903 default now: the container will change its local timezone
1904 to what the host has set.
1906 - Lease timeouts can now be configured via
1907 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
1909 - The DHCP server improved on the stability of
1910 leases. Clients are more likely to get the same lease
1911 information back, even if the server loses state.
1913 - The DHCP server supports two new configuration options to
1914 control the lease address pool metrics, PoolOffset= and
1917 * The encapsulation limit of tunnels in systemd-networkd may
1918 now be configured via 'EncapsulationLimit='. It allows
1919 modifying the maximum additional levels of encapsulation
1920 that are permitted to be prepended to a packet.
1922 * systemd now supports the concept of user buses replacing
1923 session buses, if used with dbus-1.10 (and enabled via dbus
1924 --enable-user-session). It previously only supported this on
1925 kdbus-enabled systems, and this release expands this to
1926 'dbus-daemon' systems.
1928 * systemd-networkd now supports predictable interface names
1931 * systemd now optionally supports the new Linux kernel
1932 "unified" control group hierarchy. If enabled via the kernel
1933 command-line option 'systemd.unified_cgroup_hierarchy=1',
1934 systemd will try to mount the unified cgroup hierarchy
1935 directly on /sys/fs/cgroup. If not enabled, or not
1936 available, systemd will fall back to the legacy cgroup
1937 hierarchy setup, as before. Host system and containers can
1938 mix and match legacy and unified hierarchies as they
1939 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
1940 environment variable to individually select the hierarchy to
1941 use for executed containers. By default, nspawn will use the
1942 unified hierarchy for the containers if the host uses the
1943 unified hierarchy, and the legacy hierarchy otherwise.
1944 Please note that at this point the unified hierarchy is an
1945 experimental kernel feature and is likely to change in one
1946 of the next kernel releases. Therefore, it should not be
1947 enabled by default in downstream distributions yet. The
1948 minimum required kernel version for the unified hierarchy to
1949 work is 4.2. Note that when the unified hierarchy is used
1950 for the first time delegated access to controllers is
1951 safe. Because of this systemd-nspawn containers will get
1952 access to controllers now, as will systemd user
1953 sessions. This means containers and user sessions may now
1954 manage their own resources, partitioning up what the system
1957 * A new special scope unit "init.scope" has been introduced
1958 that encapsulates PID 1 of the system. It may be used to
1959 determine resource usage and enforce resource limits on PID
1960 1 itself. PID 1 hence moved out of the root of the control
1963 * The cgtop tool gained support for filtering out kernel
1964 threads when counting tasks in a control group. Also, the
1965 count of processes is now recursively summed up by
1966 default. Two options -k and --recursive= have been added to
1967 revert to old behaviour. The tool has also been updated to
1968 work correctly in containers now.
1970 * systemd-nspawn's --bind= and --bind-ro= options have been
1971 extended to allow creation of non-recursive bind mounts.
1973 * libsystemd gained two new calls sd_pid_get_cgroup() and
1974 sd_peer_get_cgroup() which return the control group path of
1975 a process or peer of a connected AF_UNIX socket. This
1976 function call is particularly useful when implementing
1977 delegated subtrees support in the control group hierarchy.
1979 * The "sd-event" event loop API of libsystemd now supports
1980 correct dequeuing of real-time signals, without losing
1983 * When systemd requests a PolicyKit decision when managing
1984 units it will now add additional fields to the request,
1985 including unit name and desired operation. This enables more
1986 powerful PolicyKit policies, that make decisions depending
1987 on these parameters.
1989 * nspawn learnt support for .nspawn settings files, that may
1990 accompany the image files or directories of containers, and
1991 may contain additional settings for the container. This is
1992 an alternative to configuring container parameters via the
1993 nspawn command line.
1995 Contributions from: Cristian Rodríguez, Daniel Mack, David
1996 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
1997 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
1998 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
1999 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
2000 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
2001 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
2002 Andersen, Tom Gundersen, Torstein Husebø
2004 — Berlin, 2015-09-08
2008 * machinectl gained a new verb 'shell' which opens a fresh
2009 shell on the target container or the host. It is similar to
2010 the existing 'login' command of machinectl, but spawns the
2011 shell directly without prompting for username or
2012 password. The pseudo machine '.host' now refers to the local
2013 host and is used by default. Hence, 'machinectl shell' can
2014 be used as replacement for 'su -' which spawns a session as
2015 a fresh systemd unit in a way that is fully isolated from
2016 the originating session.
2018 * systemd-networkd learned to cope with private-zone DHCP
2019 options and allows other programs to query the values.
2021 * SELinux access control when enabling/disabling units is no
2022 longer enforced with this release. The previous
2023 implementation was incorrect, and a new corrected
2024 implementation is not yet available. As unit file operations
2025 are still protected via PolicyKit and D-Bus policy this is
2026 not a security problem. Yet, distributions which care about
2027 optimal SELinux support should probably not stabilize on
2030 * sd-bus gained support for matches of type "arg0has=", that
2031 test for membership of strings in string arrays sent in bus
2034 * systemd-resolved now dumps the contents of its DNS and LLMNR
2035 caches to the logs on reception of the SIGUSR1 signal. This
2036 is useful to debug DNS behaviour.
2038 * The coredumpctl tool gained a new --directory= option to
2039 operate on journal files in a specific directory.
2041 * "systemctl reboot" and related commands gained a new
2042 "--message=" option which may be used to set a free-text
2043 wall message when shutting down or rebooting the
2044 system. This message is also logged, which is useful for
2045 figuring out the reason for a reboot or shutdown a
2048 * The "systemd-resolve-host" tool's -i switch now takes
2049 network interface numbers as alternative to interface names.
2051 * A new unit file setting for services has been introduced:
2052 UtmpMode= allows configuration of how precisely systemd
2053 handles utmp and wtmp entries for the service if this is
2054 enabled. This allows writing services that appear similar to
2055 user sessions in the output of the "w", "who", "last" and
2058 * systemd-resolved will now locally synthesize DNS resource
2059 records for the "localhost" and "gateway" domains as well as
2060 the local hostname. This should ensure that clients querying
2061 RRs via resolved will get similar results as those going via
2062 NSS, if nss-myhostname is enabled.
2064 Contributions from: Alastair Hughes, Alex Crawford, Daniel
2065 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
2066 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
2067 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
2068 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
2069 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
2070 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
2071 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
2072 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
2073 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
2074 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
2075 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2077 — Berlin, 2015-08-27
2081 * The systemd-efi-boot-generator functionality was merged into
2082 systemd-gpt-auto-generator.
2084 * systemd-networkd now supports Group Policy for vxlan
2085 devices. It can be enabled via the new boolean configuration
2086 option called 'GroupPolicyExtension='.
2088 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
2089 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
2090 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
2092 — Berlin, 2015-07-31
2096 * The python-systemd code has been removed from the systemd repository.
2097 A new repository has been created which accommodates the code from
2098 now on, and we kindly ask distributions to create a separate package
2099 for this: https://github.com/systemd/python-systemd
2101 * The systemd daemon will now reload its main configuration
2102 (/etc/systemd/system.conf) on daemon-reload.
2104 * sd-dhcp now exposes vendor specific extensions via
2105 sd_dhcp_lease_get_vendor_specific().
2107 * systemd-networkd gained a number of new configuration options.
2109 - A new boolean configuration option for TAP devices called
2110 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
2111 device, thus allowing to send and receive GSO packets.
2113 - A new tunnel configuration option called 'CopyDSCP='.
2114 If enabled, the DSCP field of ip6 tunnels is copied into the
2115 decapsulated packet.
2117 - A set of boolean bridge configuration options were added.
2118 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
2119 and 'UnicastFlood=' are now parsed by networkd and applied to the
2120 respective bridge link device via the respective IFLA_BRPORT_*
2123 - A new string configuration option to override the hostname sent
2124 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
2125 is true, networkd will use the configured hostname instead of the
2126 system hostname when sending DHCP requests.
2128 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
2129 networkd will configure the IPv6 flow-label of the tunnel device
2130 according to RFC2460.
2132 - The 'macvtap' virtual network devices are now supported, similar to
2133 the already supported 'macvlan' devices.
2135 * systemd-resolved now implements RFC5452 to improve resilience against
2136 cache poisoning. Additionally, source port randomization is enabled
2137 by default to further protect against DNS spoofing attacks.
2139 * nss-mymachines now supports translating UIDs and GIDs of running
2140 containers with user-namespaces enabled. If a container 'foo'
2141 translates a host uid 'UID' to the container uid 'TUID', then
2142 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
2143 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
2144 mapped as 'vg-foo-TGID'.
2146 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
2147 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
2148 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
2149 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
2150 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
2151 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
2152 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
2153 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
2154 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
2155 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
2157 — Berlin, 2015-07-29
2161 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
2162 There are no known issues with current sysfs, and udev does not need
2163 or should be used to work around such bugs.
2165 * udev does no longer enable USB HID power management. Several reports
2166 indicate, that some devices cannot handle that setting.
2168 * The udev accelerometer helper was removed. The functionality
2169 is now fully included in iio-sensor-proxy. But this means,
2170 older iio-sensor-proxy versions will no longer provide
2171 accelerometer/orientation data with this systemd version.
2172 Please upgrade iio-sensor-proxy to version 1.0.
2174 * networkd gained a new configuration option IPv6PrivacyExtensions=
2175 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
2176 for Stateless Address") on selected networks.
2178 * For the sake of fewer build-time dependencies and less code in the
2179 main repository, the python bindings are about to be removed in the
2180 next release. A new repository has been created which accommodates
2181 the code from now on, and we kindly ask distributions to create a
2182 separate package for this. The removal will take place in v223.
2184 https://github.com/systemd/python-systemd
2186 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
2187 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
2188 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
2189 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
2190 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
2191 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
2192 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
2193 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
2194 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
2195 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2197 — Berlin, 2015-07-07
2201 * The sd-bus.h and sd-event.h APIs have now been declared
2202 stable and have been added to the official interface of
2203 libsystemd.so. sd-bus implements an alternative D-Bus client
2204 library, that is relatively easy to use, very efficient and
2205 supports both classic D-Bus as well as kdbus as transport
2206 backend. sd-event is a generic event loop abstraction that
2207 is built around Linux epoll, but adds features such as event
2208 prioritization or efficient timer handling. Both APIs are good
2209 choices for C programs looking for a bus and/or event loop
2210 implementation that is minimal and does not have to be
2211 portable to other kernels.
2213 * kdbus support is no longer compile-time optional. It is now
2214 always built-in. However, it can still be disabled at
2215 runtime using the kdbus=0 kernel command line setting, and
2216 that setting may be changed to default to off, by specifying
2217 --disable-kdbus at build-time. Note though that the kernel
2218 command line setting has no effect if the kdbus.ko kernel
2219 module is not installed, in which case kdbus is (obviously)
2220 also disabled. We encourage all downstream distributions to
2221 begin testing kdbus by adding it to the kernel images in the
2222 development distributions, and leaving kdbus support in
2225 * The minimal required util-linux version has been bumped to
2228 * Support for chkconfig (--enable-chkconfig) was removed in
2229 favor of calling an abstraction tool
2230 /lib/systemd/systemd-sysv-install. This needs to be
2231 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
2232 in README for details.
2234 * If there's a systemd unit and a SysV init script for the
2235 same service name, and the user executes "systemctl enable"
2236 for it (or a related call), then this will now enable both
2237 (or execute the related operation on both), not just the
2240 * The libudev API documentation has been converted from gtkdoc
2243 * gudev has been removed from the systemd tree, it is now an
2246 * The systemd-cgtop tool learnt a new --raw switch to generate
2247 "raw" (machine parsable) output.
2249 * networkd's IPForwarding= .network file setting learnt the
2250 new setting "kernel", which ensures that networkd does not
2251 change the IP forwarding sysctl from the default kernel
2254 * The systemd-logind bus API now exposes a new boolean
2255 property "Docked" that reports whether logind considers the
2256 system "docked", i.e. connected to a docking station or not.
2258 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
2259 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
2260 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
2261 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
2262 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
2263 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
2264 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
2265 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
2266 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
2267 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
2268 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
2269 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
2270 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
2271 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
2272 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
2273 Fink, Zbigniew Jędrzejewski-Szmek
2275 — Berlin, 2015-06-19
2279 * The gudev library has been extracted into a separate repository
2280 available at: https://git.gnome.org/browse/libgudev/
2281 It is now managed as part of the Gnome project. Distributions
2282 are recommended to pass --disable-gudev to systemd and use
2283 gudev from the Gnome project instead. gudev is still included
2284 in systemd, for now. It will be removed soon, though. Please
2285 also see the announcement-thread on systemd-devel:
2286 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
2288 * systemd now exposes a CPUUsageNSec= property for each
2289 service unit on the bus, that contains the overall consumed
2290 CPU time of a service (the sum of what each process of the
2291 service consumed). This value is only available if
2292 CPUAccounting= is turned on for a service, and is then shown
2293 in the "systemctl status" output.
2295 * Support for configuring alternative mappings of the old SysV
2296 runlevels to systemd targets has been removed. They are now
2297 hardcoded in a way that runlevels 2, 3, 4 all map to
2298 multi-user.target and 5 to graphical.target (which
2299 previously was already the default behaviour).
2301 * The auto-mounter logic gained support for mount point
2302 expiry, using a new TimeoutIdleSec= setting in .automount
2303 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
2305 * The EFI System Partition (ESP) as mounted to /boot by
2306 systemd-efi-boot-generator will now be unmounted
2307 automatically after 2 minutes of not being used. This should
2308 minimize the risk of ESP corruptions.
2310 * New /etc/fstab options x-systemd.requires= and
2311 x-systemd.requires-mounts-for= are now supported to express
2312 additional dependencies for mounts. This is useful for
2313 journalling file systems that support external journal
2314 devices or overlay file systems that require underlying file
2315 systems to be mounted.
2317 * systemd does not support direct live-upgrades (via systemctl
2318 daemon-reexec) from versions older than v44 anymore. As no
2319 distribution we are aware of shipped such old versions in a
2320 stable release this should not be problematic.
2322 * When systemd forks off a new per-connection service instance
2323 it will now set the $REMOTE_ADDR environment variable to the
2324 remote IP address, and $REMOTE_PORT environment variable to
2325 the remote IP port. This behaviour is similar to the
2326 corresponding environment variables defined by CGI.
2328 * systemd-networkd gained support for uplink failure
2329 detection. The BindCarrier= option allows binding interface
2330 configuration dynamically to the link sense of other
2331 interfaces. This is useful to achieve behaviour like in
2334 * systemd-networkd gained support for configuring the DHCP
2335 client identifier to use when requesting leases.
2337 * systemd-networkd now has a per-network UseNTP= option to
2338 configure whether NTP server information acquired via DHCP
2339 is passed on to services like systemd-timesyncd.
2341 * systemd-networkd gained support for vti6 tunnels.
2343 * Note that systemd-networkd manages the sysctl variable
2344 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
2345 it is configured for since v219. The variable controls IP
2346 forwarding, and is a per-interface alternative to the global
2347 /proc/sys/net/ipv[46]/ip_forward. This setting is
2348 configurable in the IPForward= option, which defaults to
2349 "no". This means if networkd is used for an interface it is
2350 no longer sufficient to set the global sysctl option to turn
2351 on IP forwarding! Instead, the .network file option
2352 IPForward= needs to be turned on! Note that the
2353 implementation of this behaviour was broken in v219 and has
2356 * Many bonding and vxlan options are now configurable in
2359 * systemd-nspawn gained a new --property= setting to set unit
2360 properties for the container scope. This is useful for
2361 setting resource parameters (e.g. "CPUShares=500") on
2362 containers started from the command line.
2364 * systemd-nspawn gained a new --private-users= switch to make
2365 use of user namespacing available on recent Linux kernels.
2367 * systemd-nspawn may now be called as part of a shell pipeline
2368 in which case the pipes used for stdin and stdout are passed
2369 directly to the process invoked in the container, without
2370 indirection via a pseudo tty.
2372 * systemd-nspawn gained a new switch to control the UNIX
2373 signal to use when killing the init process of the container
2376 * systemd-nspawn gained a new --overlay= switch for mounting
2377 overlay file systems into the container using the new kernel
2380 * When a container image is imported via systemd-importd and
2381 the host file system is not btrfs, a loopback block device
2382 file is created in /var/lib/machines.raw with a btrfs file
2383 system inside. It is then mounted to /var/lib/machines to
2384 enable btrfs features for container management. The loopback
2385 file and btrfs file system is grown as needed when container
2386 images are imported via systemd-importd.
2388 * systemd-machined/systemd-importd gained support for btrfs
2389 quota, to enforce container disk space limits on disk. This
2390 is exposed in "machinectl set-limit".
2392 * systemd-importd now can import containers from local .tar,
2393 .raw and .qcow2 images, and export them to .tar and .raw. It
2394 can also import dkr v2 images now from the network (on top
2397 * systemd-importd gained support for verifying downloaded
2398 images with gpg2 (previously only gpg1 was supported).
2400 * systemd-machined, systemd-logind, systemd: most bus calls
2401 are now accessible to unprivileged processes via
2402 PolicyKit. Also, systemd-logind will now allow users to kill
2403 their own sessions without further privileges or
2406 * systemd-shutdownd has been removed. This service was
2407 previously responsible for implementing scheduled shutdowns
2408 as exposed in /usr/bin/shutdown's time parameter. This
2409 functionality has now been moved into systemd-logind and is
2410 accessible via a bus interface.
2412 * "systemctl reboot" gained a new switch --firmware-setup that
2413 can be used to reboot into the EFI firmware setup, if that
2414 is available. systemd-logind now exposes an API on the bus
2415 to trigger such reboots, in case graphical desktop UIs want
2416 to cover this functionality.
2418 * "systemctl enable", "systemctl disable" and "systemctl mask"
2419 now support a new "--now" switch. If specified the units
2420 that are enabled will also be started, and the ones
2421 disabled/masked also stopped.
2423 * The Gummiboot EFI boot loader tool has been merged into
2424 systemd, and renamed to "systemd-boot". The bootctl tool has been
2425 updated to support systemd-boot.
2427 * An EFI kernel stub has been added that may be used to create
2428 kernel EFI binaries that contain not only the actual kernel,
2429 but also an initrd, boot splash, command line and OS release
2430 information. This combined binary can then be signed as a
2431 single image, so that the firmware can verify it all in one
2432 step. systemd-boot has special support for EFI binaries created
2433 like this and can extract OS release information from them
2434 and show them in the boot menu. This functionality is useful
2435 to implement cryptographically verified boot schemes.
2437 * Optional support has been added to systemd-fsck to pass
2438 fsck's progress report to an AF_UNIX socket in the file
2441 * udev will no longer create device symlinks for all block
2442 devices by default. A blacklist for excluding special block
2443 devices from this logic has been turned into a whitelist
2444 that requires picking block devices explicitly that require
2447 * A new (currently still internal) API sd-device.h has been
2448 added to libsystemd. This modernized API is supposed to
2449 replace libudev eventually. In fact, already much of libudev
2450 is now just a wrapper around sd-device.h.
2452 * A new hwdb database for storing metadata about pointing
2453 stick devices has been added.
2455 * systemd-tmpfiles gained support for setting file attributes
2456 similar to the "chattr" tool with new 'h' and 'H' lines.
2458 * systemd-journald will no longer unconditionally set the
2459 btrfs NOCOW flag on new journal files. This is instead done
2460 with tmpfiles snippet using the new 'h' line type. This
2461 allows easy disabling of this logic, by masking the
2462 journal-nocow.conf tmpfiles file.
2464 * systemd-journald will now translate audit message types to
2465 human readable identifiers when writing them to the
2466 journal. This should improve readability of audit messages.
2468 * The LUKS logic gained support for the offset= and skip=
2469 options in /etc/crypttab, as previously implemented by
2472 * /usr/lib/os-release gained a new optional field VARIANT= for
2473 distributions that support multiple variants (such as a
2474 desktop edition, a server edition, ...)
2476 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
2477 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
2478 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
2479 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
2480 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
2481 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
2482 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
2483 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
2484 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
2485 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
2486 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
2487 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
2488 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
2489 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
2490 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
2491 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
2492 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
2493 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
2494 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
2495 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
2496 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
2497 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
2498 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
2499 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
2500 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
2501 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
2502 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
2504 — Berlin, 2015-05-22
2508 * Introduce a new API "sd-hwdb.h" for querying the hardware
2509 metadata database. With this minimal interface one can query
2510 and enumerate the udev hwdb, decoupled from the old libudev
2511 library. libudev's interface for this is now only a wrapper
2512 around sd-hwdb. A new tool systemd-hwdb has been added to
2513 interface with and update the database.
2515 * When any of systemd's tools copies files (for example due to
2516 tmpfiles' C lines) a btrfs reflink will attempted first,
2517 before bytewise copying is done.
2519 * systemd-nspawn gained a new --ephemeral switch. When
2520 specified a btrfs snapshot is taken of the container's root
2521 directory, and immediately removed when the container
2522 terminates again. Thus, a container can be started whose
2523 changes never alter the container's root directory, and are
2524 lost on container termination. This switch can also be used
2525 for starting a container off the root file system of the
2526 host without affecting the host OS. This switch is only
2527 available on btrfs file systems.
2529 * systemd-nspawn gained a new --template= switch. It takes the
2530 path to a container tree to use as template for the tree
2531 specified via --directory=, should that directory be
2532 missing. This allows instantiating containers dynamically,
2533 on first run. This switch is only available on btrfs file
2536 * When a .mount unit refers to a mount point on which multiple
2537 mounts are stacked, and the .mount unit is stopped all of
2538 the stacked mount points will now be unmounted until no
2539 mount point remains.
2541 * systemd now has an explicit notion of supported and
2542 unsupported unit types. Jobs enqueued for unsupported unit
2543 types will now fail with an "unsupported" error code. More
2544 specifically .swap, .automount and .device units are not
2545 supported in containers, .busname units are not supported on
2546 non-kdbus systems. .swap and .automount are also not
2547 supported if their respective kernel compile time options
2550 * machinectl gained support for two new "copy-from" and
2551 "copy-to" commands for copying files from a running
2552 container to the host or vice versa.
2554 * machinectl gained support for a new "bind" command to bind
2555 mount host directories into local containers. This is
2556 currently only supported for nspawn containers.
2558 * networkd gained support for configuring bridge forwarding
2559 database entries (fdb) from .network files.
2561 * A new tiny daemon "systemd-importd" has been added that can
2562 download container images in tar, raw, qcow2 or dkr formats,
2563 and make them available locally in /var/lib/machines, so
2564 that they can run as nspawn containers. The daemon can GPG
2565 verify the downloads (not supported for dkr, since it has no
2566 provisions for verifying downloads). It will transparently
2567 decompress bz2, xz, gzip compressed downloads if necessary,
2568 and restore sparse files on disk. The daemon uses privilege
2569 separation to ensure the actual download logic runs with
2570 fewer privileges than the daemon itself. machinectl has
2571 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
2572 make the functionality of importd available to the
2573 user. With this in place the Fedora and Ubuntu "Cloud"
2574 images can be downloaded and booted as containers unmodified
2575 (the Fedora images lack the appropriate GPG signature files
2576 currently, so they cannot be verified, but this will change
2577 soon, hopefully). Note that downloading images is currently
2578 only fully supported on btrfs.
2580 * machinectl is now able to list container images found in
2581 /var/lib/machines, along with some metadata about sizes of
2582 disk and similar. If the directory is located on btrfs and
2583 quota is enabled, this includes quota display. A new command
2584 "image-status" has been added that shows additional
2585 information about images.
2587 * machinectl is now able to clone container images
2588 efficiently, if the underlying file system (btrfs) supports
2589 it, with the new "machinectl clone" command. It also
2590 gained commands for renaming and removing images, as well as
2591 marking them read-only or read-write (supported also on
2592 legacy file systems).
2594 * networkd gained support for collecting LLDP network
2595 announcements, from hardware that supports this. This is
2596 shown in networkctl output.
2598 * systemd-run gained support for a new -t (--pty) switch for
2599 invoking a binary on a pty whose input and output is
2600 connected to the invoking terminal. This allows executing
2601 processes as system services while interactively
2602 communicating with them via the terminal. Most interestingly
2603 this is supported across container boundaries. Invoking
2604 "systemd-run -t /bin/bash" is an alternative to running a
2605 full login session, the difference being that the former
2606 will not register a session, nor go through the PAM session
2609 * tmpfiles gained support for a new "v" line type for creating
2610 btrfs subvolumes. If the underlying file system is a legacy
2611 file system, this automatically degrades to creating a
2612 normal directory. Among others /var/lib/machines is now
2613 created like this at boot, should it be missing.
2615 * The directory /var/lib/containers/ has been deprecated and
2616 been replaced by /var/lib/machines. The term "machines" has
2617 been used in the systemd context as generic term for both
2618 VMs and containers, and hence appears more appropriate for
2619 this, as the directory can also contain raw images bootable
2622 * systemd-nspawn when invoked with -M but without --directory=
2623 or --image= is now capable of searching for the container
2624 root directory, subvolume or disk image automatically, in
2625 /var/lib/machines. systemd-nspawn@.service has been updated
2626 to make use of this, thus allowing it to be used for raw
2629 * A new machines.target unit has been introduced that is
2630 supposed to group all containers/VMs invoked as services on
2631 the system. systemd-nspawn@.service has been updated to
2632 integrate with that.
2634 * machinectl gained a new "start" command, for invoking a
2635 container as a service. "machinectl start foo" is mostly
2636 equivalent to "systemctl start systemd-nspawn@foo.service",
2637 but handles escaping in a nicer way.
2639 * systemd-nspawn will now mount most of the cgroupfs tree
2640 read-only into each container, with the exception of the
2641 container's own subtree in the name=systemd hierarchy.
2643 * journald now sets the special FS_NOCOW file flag for its
2644 journal files. This should improve performance on btrfs, by
2645 avoiding heavy fragmentation when journald's write-pattern
2646 is used on COW file systems. It degrades btrfs' data
2647 integrity guarantees for the files to the same levels as for
2648 ext3/ext4 however. This should be OK though as journald does
2649 its own data integrity checks and all its objects are
2650 checksummed on disk. Also, journald should handle btrfs disk
2651 full events a lot more gracefully now, by processing SIGBUS
2652 errors, and not relying on fallocate() anymore.
2654 * When journald detects that journal files it is writing to
2655 have been deleted it will immediately start new journal
2658 * systemd now provides a way to store file descriptors
2659 per-service in PID 1. This is useful for daemons to ensure
2660 that fds they require are not lost during a daemon
2661 restart. The fds are passed to the daemon on the next
2662 invocation in the same way socket activation fds are
2663 passed. This is now used by journald to ensure that the
2664 various sockets connected to all the system's stdout/stderr
2665 are not lost when journald is restarted. File descriptors
2666 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
2667 an extension to sd_notify(). Note that a limit is enforced
2668 on the number of fds a service can store in PID 1, and it
2669 defaults to 0, so that no fds may be stored, unless this is
2670 explicitly turned on.
2672 * The default TERM variable to use for units connected to a
2673 terminal, when no other value is explicitly is set is now
2674 vt220 rather than vt102. This should be fairly safe still,
2675 but allows PgUp/PgDn work.
2677 * The /etc/crypttab option header= as known from Debian is now
2680 * "loginctl user-status" and "loginctl session-status" will
2681 now show the last 10 lines of log messages of the
2682 user/session following the status output. Similar,
2683 "machinectl status" will show the last 10 log lines
2684 associated with a virtual machine or container
2685 service. (Note that this is usually not the log messages
2686 done in the VM/container itself, but simply what the
2687 container manager logs. For nspawn this includes all console
2690 * "loginctl session-status" without further argument will now
2691 show the status of the session of the caller. Similar,
2692 "lock-session", "unlock-session", "activate",
2693 "enable-linger", "disable-linger" may now be called without
2694 session/user parameter in which case they apply to the
2695 caller's session/user.
2697 * An X11 session scriptlet is now shipped that uploads
2698 $DISPLAY and $XAUTHORITY into the environment of the systemd
2699 --user daemon if a session begins. This should improve
2700 compatibility with X11 enabled applications run as systemd
2703 * Generators are now subject to masking via /etc and /run, the
2704 same way as unit files.
2706 * networkd .network files gained support for configuring
2707 per-link IPv4/IPv6 packet forwarding as well as IPv4
2708 masquerading. This is by default turned on for veth links to
2709 containers, as registered by systemd-nspawn. This means that
2710 nspawn containers run with --network-veth will now get
2711 automatic routed access to the host's networks without any
2712 further configuration or setup, as long as networkd runs on
2715 * systemd-nspawn gained the --port= (-p) switch to expose TCP
2716 or UDP posts of a container on the host. With this in place
2717 it is possible to run containers with private veth links
2718 (--network-veth), and have their functionality exposed on
2719 the host as if their services were running directly on the
2722 * systemd-nspawn's --network-veth switch now gained a short
2723 version "-n", since with the changes above it is now truly
2724 useful out-of-the-box. The systemd-nspawn@.service has been
2725 updated to make use of it too by default.
2727 * systemd-nspawn will now maintain a per-image R/W lock, to
2728 ensure that the same image is not started more than once
2729 writable. (It's OK to run an image multiple times
2730 simultaneously in read-only mode.)
2732 * systemd-nspawn's --image= option is now capable of
2733 dissecting and booting MBR and GPT disk images that contain
2734 only a single active Linux partition. Previously it
2735 supported only GPT disk images with proper GPT type
2736 IDs. This allows running cloud images from major
2737 distributions directly with systemd-nspawn, without
2740 * In addition to collecting mouse dpi data in the udev
2741 hardware database, there's now support for collecting angle
2742 information for mouse scroll wheels. The database is
2743 supposed to guarantee similar scrolling behavior on mice
2744 that it knows about. There's also support for collecting
2745 information about Touchpad types.
2747 * udev's input_id built-in will now also collect touch screen
2748 dimension data and attach it to probed devices.
2750 * /etc/os-release gained support for a Distribution Privacy
2753 * networkd gained support for creating "ipvlan", "gretap",
2754 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
2756 * systemd-tmpfiles gained support for "a" lines for setting
2759 * systemd-nspawn will now mount /tmp in the container to
2760 tmpfs, automatically.
2762 * systemd now exposes the memory.usage_in_bytes cgroup
2763 attribute and shows it for each service in the "systemctl
2764 status" output, if available.
2766 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
2767 immediate reboot is triggered. This useful if shutdown is
2768 hung and is unable to complete, to expedite the
2769 operation. Note that this kind of reboot will still unmount
2770 all file systems, and hence should not result in fsck being
2773 * A .device unit for an optical block device will now be
2774 considered active only when a medium is in the drive. Also,
2775 mount units are now bound to their backing devices thus
2776 triggering automatic unmounting when devices become
2777 unavailable. With this in place systemd will now
2778 automatically unmount left-over mounts when a CD-ROM is
2779 ejected or an USB stick is yanked from the system.
2781 * networkd-wait-online now has support for waiting for
2782 specific interfaces only (with globbing), and for giving up
2783 after a configurable timeout.
2785 * networkd now exits when idle. It will be automatically
2786 restarted as soon as interfaces show up, are removed or
2787 change state. networkd will stay around as long as there is
2788 at least one DHCP state machine or similar around, that keep
2791 * networkd may now configure IPv6 link-local addressing in
2792 addition to IPv4 link-local addressing.
2794 * The IPv6 "token" for use in SLAAC may now be configured for
2795 each .network interface in networkd.
2797 * Routes configured with networkd may now be assigned a scope
2800 * networkd's [Match] sections now support globbing and lists
2801 of multiple space-separated matches per item.
2803 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
2804 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
2805 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
2806 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
2807 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
2808 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
2809 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
2810 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
2811 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
2812 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
2813 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
2814 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
2815 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
2816 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
2817 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
2818 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
2819 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
2820 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
2821 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
2822 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
2823 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
2824 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
2825 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
2826 Hoffmann, Zbigniew Jędrzejewski-Szmek
2828 — Berlin, 2015-02-16
2832 * When querying unit file enablement status (for example via
2833 "systemctl is-enabled"), a new state "indirect" is now known
2834 which indicates that a unit might not be enabled itself, but
2835 another unit listed in its Also= setting might be.
2837 * Similar to the various existing ConditionXYZ= settings for
2838 units, there are now matching AssertXYZ= settings. While
2839 failing conditions cause a unit to be skipped, but its job
2840 to succeed, failing assertions declared like this will cause
2841 a unit start operation and its job to fail.
2843 * hostnamed now knows a new chassis type "embedded".
2845 * systemctl gained a new "edit" command. When used on a unit
2846 file, this allows extending unit files with .d/ drop-in
2847 configuration snippets or editing the full file (after
2848 copying it from /usr/lib to /etc). This will invoke the
2849 user's editor (as configured with $EDITOR), and reload the
2850 modified configuration after editing.
2852 * "systemctl status" now shows the suggested enablement state
2853 for a unit, as declared in the (usually vendor-supplied)
2854 system preset files.
2856 * nss-myhostname will now resolve the single-label host name
2857 "gateway" to the locally configured default IP routing
2858 gateways, ordered by their metrics. This assigns a stable
2859 name to the used gateways, regardless which ones are
2860 currently configured. Note that the name will only be
2861 resolved after all other name sources (if nss-myhostname is
2862 configured properly) and should hence not negatively impact
2863 systems that use the single-label host name "gateway" in
2866 * systemd-inhibit now allows filtering by mode when listing
2869 * Scope and service units gained a new "Delegate" boolean
2870 property, which, when set, allows processes running inside the
2871 unit to further partition resources. This is primarily
2872 useful for systemd user instances as well as container
2875 * journald will now pick up audit messages directly from
2876 the kernel, and log them like any other log message. The
2877 audit fields are split up and fully indexed. This means that
2878 journalctl in many ways is now a (nicer!) alternative to
2879 ausearch, the traditional audit client. Note that this
2880 implements only a minimal audit client. If you want the
2881 special audit modes like reboot-on-log-overflow, please use
2882 the traditional auditd instead, which can be used in
2883 parallel to journald.
2885 * The ConditionSecurity= unit file option now understands the
2886 special string "audit" to check whether auditing is
2889 * journalctl gained two new commands --vacuum-size= and
2890 --vacuum-time= to delete old journal files until the
2891 remaining ones take up no more than the specified size on disk,
2892 or are not older than the specified time.
2894 * A new, native PPPoE library has been added to sd-network,
2895 systemd's library of light-weight networking protocols. This
2896 library will be used in a future version of networkd to
2897 enable PPPoE communication without an external pppd daemon.
2899 * The busctl tool now understands a new "capture" verb that
2900 works similar to "monitor", but writes a packet capture
2901 trace to STDOUT that can be redirected to a file which is
2902 compatible with libcap's capture file format. This can then
2903 be loaded in Wireshark and similar tools to inspect bus
2906 * The busctl tool now understands a new "tree" verb that shows
2907 the object trees of a specific service on the bus, or of all
2910 * The busctl tool now understands a new "introspect" verb that
2911 shows all interfaces and members of objects on the bus,
2912 including their signature and values. This is particularly
2913 useful to get more information about bus objects shown by
2914 the new "busctl tree" command.
2916 * The busctl tool now understands new verbs "call",
2917 "set-property" and "get-property" for invoking bus method
2918 calls, setting and getting bus object properties in a
2921 * busctl gained a new --augment-creds= argument that controls
2922 whether the tool shall augment credential information it
2923 gets from the bus with data from /proc, in a possibly
2926 * nspawn's --link-journal= switch gained two new values
2927 "try-guest" and "try-host" that work like "guest" and
2928 "host", but do not fail if the host has no persistent
2929 journalling enabled. -j is now equivalent to
2930 --link-journal=try-guest.
2932 * macvlan network devices created by nspawn will now have
2933 stable MAC addresses.
2935 * A new SmackProcessLabel= unit setting has been added, which
2936 controls the SMACK security label processes forked off by
2937 the respective unit shall use.
2939 * If compiled with --enable-xkbcommon, systemd-localed will
2940 verify x11 keymap settings by compiling the given keymap. It
2941 will spew out warnings if the compilation fails. This
2942 requires libxkbcommon to be installed.
2944 * When a coredump is collected, a larger number of metadata
2945 fields is now collected and included in the journal records
2946 created for it. More specifically, control group membership,
2947 environment variables, memory maps, working directory,
2948 chroot directory, /proc/$PID/status, and a list of open file
2949 descriptors is now stored in the log entry.
2951 * The udev hwdb now contains DPI information for mice. For
2954 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
2956 * All systemd programs that read standalone configuration
2957 files in /etc now also support a corresponding series of
2958 .conf.d configuration directories in /etc/, /run/,
2959 /usr/local/lib/, /usr/lib/, and (if configured with
2960 --enable-split-usr) /lib/. In particular, the following
2961 configuration files now have corresponding configuration
2962 directories: system.conf user.conf, logind.conf,
2963 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
2964 resolved.conf, timesyncd.conf, journal-remote.conf, and
2965 journal-upload.conf. Note that distributions should use the
2966 configuration directories in /usr/lib/; the directories in
2967 /etc/ are reserved for the system administrator.
2969 * systemd-rfkill will no longer take the rfkill device name
2970 into account when storing rfkill state on disk, as the name
2971 might be dynamically assigned and not stable. Instead, the
2972 ID_PATH udev variable combined with the rfkill type (wlan,
2973 bluetooth, ...) is used.
2975 * A new service systemd-machine-id-commit.service has been
2976 added. When used on systems where /etc is read-only during
2977 boot, and /etc/machine-id is not initialized (but an empty
2978 file), this service will copy the temporary machine ID
2979 created as replacement into /etc after the system is fully
2980 booted up. This is useful for systems that are freshly
2981 installed with a non-initialized machine ID, but should get
2982 a fixed machine ID for subsequent boots.
2984 * networkd's .netdev files now provide a large set of
2985 configuration parameters for VXLAN devices. Similarly, the
2986 bridge port cost parameter is now configurable in .network
2987 files. There's also new support for configuring IP source
2988 routing. networkd .link files gained support for a new
2989 OriginalName= match that is useful to match against the
2990 original interface name the kernel assigned. .network files
2991 may include MTU= and MACAddress= fields for altering the MTU
2992 and MAC address while being connected to a specific network
2995 * The LUKS logic gained supported for configuring
2996 UUID-specific key files. There's also new support for naming
2997 LUKS device from the kernel command line, using the new
2998 luks.name= argument.
3000 * Timer units may now be transiently created via the bus API
3001 (this was previously already available for scope and service
3002 units). In addition it is now possible to create multiple
3003 transient units at the same time with a single bus call. The
3004 "systemd-run" tool has been updated to make use of this for
3005 running commands on a specified time, in at(1)-style.
3007 * tmpfiles gained support for "t" lines, for assigning
3008 extended attributes to files. Among other uses this may be
3009 used to assign SMACK labels to files.
3011 Contributions from: Alin Rauta, Alison Chaiken, Andrej
3012 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
3013 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
3014 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
3015 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
3016 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
3017 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
3018 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
3019 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
3020 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
3021 Michael Biebl, Michael Chapman, Michael Marineau, Michal
3022 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
3023 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
3024 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
3025 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
3026 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
3027 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
3028 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
3030 — Berlin, 2014-12-10
3034 * journalctl gained the new options -t/--identifier= to match
3035 on the syslog identifier (aka "tag"), as well as --utc to
3036 show log timestamps in the UTC timezone. journalctl now also
3037 accepts -n/--lines=all to disable line capping in a pager.
3039 * journalctl gained a new switch, --flush, that synchronously
3040 flushes logs from /run/log/journal to /var/log/journal if
3041 persistent storage is enabled. systemd-journal-flush.service
3042 now waits until the operation is complete.
3044 * Services can notify the manager before they start a reload
3045 (by sending RELOADING=1) or shutdown (by sending
3046 STOPPING=1). This allows the manager to track and show the
3047 internal state of daemons and closes a race condition when
3048 the process is still running but has closed its D-Bus
3051 * Services with Type=oneshot do not have to have any ExecStart
3054 * User units are now loaded also from
3055 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
3056 /run/systemd/user directory that was already previously
3057 supported, but is under the control of the user.
3059 * Job timeouts (i.e. time-outs on the time a job that is
3060 queued stays in the run queue) can now optionally result in
3061 immediate reboot or power-off actions (JobTimeoutAction= and
3062 JobTimeoutRebootArgument=). This is useful on ".target"
3063 units, to limit the maximum time a target remains
3064 undispatched in the run queue, and to trigger an emergency
3065 operation in such a case. This is now used by default to
3066 turn off the system if boot-up (as defined by everything in
3067 basic.target) hangs and does not complete for at least
3068 15min. Also, if power-off or reboot hang for at least 30min
3069 an immediate power-off/reboot operation is triggered. This
3070 functionality is particularly useful to increase reliability
3071 on embedded devices, but also on laptops which might
3072 accidentally get powered on when carried in a backpack and
3073 whose boot stays stuck in a hard disk encryption passphrase
3076 * systemd-logind can be configured to also handle lid switch
3077 events even when the machine is docked or multiple displays
3078 are attached (HandleLidSwitchDocked= option).
3080 * A helper binary and a service have been added which can be
3081 used to resume from hibernation in the initramfs. A
3082 generator will parse the resume= option on the kernel
3083 command line to trigger resume.
3085 * A user console daemon systemd-consoled has been
3086 added. Currently, it is a preview, and will so far open a
3087 single terminal on each session of the user marked as
3088 Desktop=systemd-console.
3090 * Route metrics can be specified for DHCP routes added by
3093 * The SELinux context of socket-activated services can be set
3094 from the information provided by the networking stack
3095 (SELinuxContextFromNet= option).
3097 * Userspace firmware loading support has been removed and
3098 the minimum supported kernel version is thus bumped to 3.7.
3100 * Timeout for udev workers has been increased from 1 to 3
3101 minutes, but a warning will be printed after 1 minute to
3102 help diagnose kernel modules that take a long time to load.
3104 * Udev rules can now remove tags on devices with TAG-="foobar".
3106 * systemd's readahead implementation has been removed. In many
3107 circumstances it didn't give expected benefits even for
3108 rotational disk drives and was becoming less relevant in the
3109 age of SSDs. As none of the developers has been using
3110 rotating media anymore, and nobody stepped up to actively
3111 maintain this component of systemd it has now been removed.
3113 * Swap units can use Options= to specify discard options.
3114 Discard options specified for swaps in /etc/fstab are now
3117 * Docker containers are now detected as a separate type of
3120 * The Password Agent protocol gained support for queries where
3121 the user input is shown, useful e.g. for user names.
3122 systemd-ask-password gained a new --echo option to turn that
3125 * The default sysctl.d/ snippets will now set:
3127 net.core.default_qdisc = fq_codel
3129 This selects Fair Queuing Controlled Delay as the default
3130 queuing discipline for network interfaces. fq_codel helps
3131 fight the network bufferbloat problem. It is believed to be
3132 a good default with no tuning required for most workloads.
3133 Downstream distributions may override this choice. On 10Gbit
3134 servers that do not do forwarding, "fq" may perform better.
3135 Systems without a good clocksource should use "pfifo_fast".
3137 * If kdbus is enabled during build a new option BusPolicy= is
3138 available for service units, that allows locking all service
3139 processes into a stricter bus policy, in order to limit
3140 access to various bus services, or even hide most of them
3141 from the service's view entirely.
3143 * networkctl will now show the .network and .link file
3144 networkd has applied to a specific interface.
3146 * sd-login gained a new API call sd_session_get_desktop() to
3147 query which desktop environment has been selected for a
3150 * UNIX utmp support is now compile-time optional to support
3151 legacy-free systems.
3153 * systemctl gained two new commands "add-wants" and
3154 "add-requires" for pulling in units from specific targets
3157 * If the word "rescue" is specified on the kernel command line
3158 the system will now boot into rescue mode (aka
3159 rescue.target), which was previously available only by
3160 specifying "1" or "systemd.unit=rescue.target" on the kernel
3161 command line. This new kernel command line option nicely
3162 mirrors the already existing "emergency" kernel command line
3165 * New kernel command line options mount.usr=, mount.usrflags=,
3166 mount.usrfstype= have been added that match root=, rootflags=,
3167 rootfstype= but allow mounting a specific file system to
3170 * The $NOTIFY_SOCKET is now also passed to control processes of
3171 services, not only the main process.
3173 * This version reenables support for fsck's -l switch. This
3174 means at least version v2.25 of util-linux is required for
3175 operation, otherwise dead-locks on device nodes may
3176 occur. Again: you need to update util-linux to at least
3177 v2.25 when updating systemd to v217.
3179 * The "multi-seat-x" tool has been removed from systemd, as
3180 its functionality has been integrated into X servers 1.16,
3181 and the tool is hence redundant. It is recommended to update
3182 display managers invoking this tool to simply invoke X
3183 directly from now on, again.
3185 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
3186 message flag has been added for all of systemd's PolicyKit
3187 authenticated method calls has been added. In particular
3188 this now allows optional interactive authorization via
3189 PolicyKit for many of PID1's privileged operations such as
3190 unit file enabling and disabling.
3192 * "udevadm hwdb --update" learnt a new switch "--usr" for
3193 placing the rebuilt hardware database in /usr instead of
3194 /etc. When used only hardware database entries stored in
3195 /usr will be used, and any user database entries in /etc are
3196 ignored. This functionality is useful for vendors to ship a
3197 pre-built database on systems where local configuration is
3198 unnecessary or unlikely.
3200 * Calendar time specifications in .timer units now also
3201 understand the strings "semi-annually", "quarterly" and
3202 "minutely" as shortcuts (in addition to the preexisting
3203 "anually", "hourly", ...).
3205 * systemd-tmpfiles will now correctly create files in /dev
3206 at boot which are marked for creation only at boot. It is
3207 recommended to always create static device nodes with 'c!'
3208 and 'b!', so that they are created only at boot and not
3209 overwritten at runtime.
3211 * When the watchdog logic is used for a service (WatchdogSec=)
3212 and the watchdog timeout is hit the service will now be
3213 terminated with SIGABRT (instead of just SIGTERM), in order
3214 to make sure a proper coredump and backtrace is
3215 generated. This ensures that hanging services will result in
3216 similar coredump/backtrace behaviour as services that hit a
3219 Contributions from: Andreas Henriksson, Andrei Borzenkov,
3220 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
3221 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
3222 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
3223 Herrmann, David Sommerseth, David Strauss, Emil Renner
3224 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
3225 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
3226 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
3227 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
3228 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
3229 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
3230 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
3231 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
3232 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
3233 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
3234 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
3235 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
3236 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
3237 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
3238 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
3239 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
3242 — Berlin, 2014-10-28
3246 * timedated no longer reads NTP implementation unit names from
3247 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
3248 implementations should add a
3250 Conflicts=systemd-timesyncd.service
3252 to their unit files to take over and replace systemd's NTP
3253 default functionality.
3255 * systemd-sysusers gained a new line type "r" for configuring
3256 which UID/GID ranges to allocate system users/groups
3257 from. Lines of type "u" may now add an additional column
3258 that specifies the home directory for the system user to be
3259 created. Also, systemd-sysusers may now optionally read user
3260 information from STDIN instead of a file. This is useful for
3261 invoking it from RPM preinst scriptlets that need to create
3262 users before the first RPM file is installed since these
3263 files might need to be owned by them. A new
3264 %sysusers_create_inline RPM macro has been introduced to do
3265 just that. systemd-sysusers now updates the shadow files as
3266 well as the user/group databases, which should enhance
3267 compatibility with certain tools like grpck.
3269 * A number of bus APIs of PID 1 now optionally consult
3270 PolicyKit to permit access for otherwise unprivileged
3271 clients under certain conditions. Note that this currently
3272 doesn't support interactive authentication yet, but this is
3273 expected to be added eventually, too.
3275 * /etc/machine-info now has new fields for configuring the
3276 deployment environment of the machine, as well as the
3277 location of the machine. hostnamectl has been updated with
3278 new command to update these fields.
3280 * systemd-timesyncd has been updated to automatically acquire
3281 NTP server information from systemd-networkd, which might
3282 have been discovered via DHCP.
3284 * systemd-resolved now includes a caching DNS stub resolver
3285 and a complete LLMNR name resolution implementation. A new
3286 NSS module "nss-resolve" has been added which can be used
3287 instead of glibc's own "nss-dns" to resolve hostnames via
3288 systemd-resolved. Hostnames, addresses and arbitrary RRs may
3289 be resolved via systemd-resolved D-Bus APIs. In contrast to
3290 the glibc internal resolver systemd-resolved is aware of
3291 multi-homed system, and keeps DNS server and caches separate
3292 and per-interface. Queries are sent simultaneously on all
3293 interfaces that have DNS servers configured, in order to
3294 properly handle VPNs and local LANs which might resolve
3295 separate sets of domain names. systemd-resolved may acquire
3296 DNS server information from systemd-networkd automatically,
3297 which in turn might have discovered them via DHCP. A tool
3298 "systemd-resolve-host" has been added that may be used to
3299 query the DNS logic in resolved. systemd-resolved implements
3300 IDNA and automatically uses IDNA or UTF-8 encoding depending
3301 on whether classic DNS or LLMNR is used as transport. In the
3302 next releases we intend to add a DNSSEC and mDNS/DNS-SD
3303 implementation to systemd-resolved.
3305 * A new NSS module nss-mymachines has been added, that
3306 automatically resolves the names of all local registered
3307 containers to their respective IP addresses.
3309 * A new client tool "networkctl" for systemd-networkd has been
3310 added. It currently is entirely passive and will query
3311 networking configuration from udev, rtnetlink and networkd,
3312 and present it to the user in a very friendly
3313 way. Eventually, we hope to extend it to become a full
3314 control utility for networkd.
3316 * .socket units gained a new DeferAcceptSec= setting that
3317 controls the kernels' TCP_DEFER_ACCEPT sockopt for
3318 TCP. Similarly, support for controlling TCP keep-alive
3319 settings has been added (KeepAliveTimeSec=,
3320 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
3321 turning off Nagle's algorithm on TCP has been added
3324 * logind learned a new session type "web", for use in projects
3325 like Cockpit which register web clients as PAM sessions.
3327 * timer units with at least one OnCalendar= setting will now
3328 be started only after time-sync.target has been
3329 reached. This way they will not elapse before the system
3330 clock has been corrected by a local NTP client or
3331 similar. This is particular useful on RTC-less embedded
3332 machines, that come up with an invalid system clock.
3334 * systemd-nspawn's --network-veth= switch should now result in
3335 stable MAC addresses for both the outer and the inner side
3338 * systemd-nspawn gained a new --volatile= switch for running
3339 container instances with /etc or /var unpopulated.
3341 * The kdbus client code has been updated to use the new Linux
3342 3.17 memfd subsystem instead of the old kdbus-specific one.
3344 * systemd-networkd's DHCP client and server now support
3345 FORCERENEW. There are also new configuration options to
3346 configure the vendor client identifier and broadcast mode
3349 * systemd will no longer inform the kernel about the current
3350 timezone, as this is necessarily incorrect and racy as the
3351 kernel has no understanding of DST and similar
3352 concepts. This hence means FAT timestamps will be always
3353 considered UTC, similar to what Android is already
3354 doing. Also, when the RTC is configured to the local time
3355 (rather than UTC) systemd will never synchronize back to it,
3356 as this might confuse Windows at a later boot.
3358 * systemd-analyze gained a new command "verify" for offline
3359 validation of unit files.
3361 * systemd-networkd gained support for a couple of additional
3362 settings for bonding networking setups. Also, the metric for
3363 statically configured routes may now be configured. For
3364 network interfaces where this is appropriate the peer IP
3365 address may now be configured.
3367 * systemd-networkd's DHCP client will no longer request
3368 broadcasting by default, as this tripped up some networks.
3369 For hardware where broadcast is required the feature should
3370 be switched back on using RequestBroadcast=yes.
3372 * systemd-networkd will now set up IPv4LL addresses (when
3373 enabled) even if DHCP is configured successfully.
3375 * udev will now default to respect network device names given
3376 by the kernel when the kernel indicates that these are
3377 predictable. This behavior can be tweaked by changing
3378 NamePolicy= in the relevant .link file.
3380 * A new library systemd-terminal has been added that
3381 implements full TTY stream parsing and rendering. This
3382 library is supposed to be used later on for implementing a
3383 full userspace VT subsystem, replacing the current kernel
3386 * A new tool systemd-journal-upload has been added to push
3387 journal data to a remote system running
3388 systemd-journal-remote.
3390 * journald will no longer forward all local data to another
3391 running syslog daemon. This change has been made because
3392 rsyslog (which appears to be the most commonly used syslog
3393 implementation these days) no longer makes use of this, and
3394 instead pulls the data out of the journal on its own. Since
3395 forwarding the messages to a non-existent syslog server is
3396 more expensive than we assumed we have now turned this
3397 off. If you run a syslog server that is not a recent rsyslog
3398 version, you have to turn this option on again
3399 (ForwardToSyslog= in journald.conf).
3401 * journald now optionally supports the LZ4 compressor for
3402 larger journal fields. This compressor should perform much
3403 better than XZ which was the previous default.
3405 * machinectl now shows the IP addresses of local containers,
3406 if it knows them, plus the interface name of the container.
3408 * A new tool "systemd-escape" has been added that makes it
3409 easy to escape strings to build unit names and similar.
3411 * sd_notify() messages may now include a new ERRNO= field
3412 which is parsed and collected by systemd and shown among the
3413 "systemctl status" output for a service.
3415 * A new component "systemd-firstboot" has been added that
3416 queries the most basic systemd information (timezone,
3417 hostname, root password) interactively on first
3418 boot. Alternatively it may also be used to provision these
3419 things offline on OS images installed into directories.
3421 * The default sysctl.d/ snippets will now set
3423 net.ipv4.conf.default.promote_secondaries=1
3425 This has the benefit of no flushing secondary IP addresses
3426 when primary addresses are removed.
3428 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
3429 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
3430 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
3431 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
3432 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
3433 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
3434 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
3435 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
3436 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
3437 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
3438 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
3439 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
3440 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
3441 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
3442 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
3444 — Berlin, 2014-08-19
3448 * A new tool systemd-sysusers has been added. This tool
3449 creates system users and groups in /etc/passwd and
3450 /etc/group, based on static declarative system user/group
3451 definitions in /usr/lib/sysusers.d/. This is useful to
3452 enable factory resets and volatile systems that boot up with
3453 an empty /etc directory, and thus need system users and
3454 groups created during early boot. systemd now also ships
3455 with two default sysusers.d/ files for the most basic
3456 users and groups systemd and the core operating system
3459 * A new tmpfiles snippet has been added that rebuilds the
3460 essential files in /etc on boot, should they be missing.
3462 * A directive for ensuring automatic clean-up of
3463 /var/cache/man/ has been removed from the default
3464 configuration. This line should now be shipped by the man
3465 implementation. The necessary change has been made to the
3466 man-db implementation. Note that you need to update your man
3467 implementation to one that ships this line, otherwise no
3468 automatic clean-up of /var/cache/man will take place.
3470 * A new condition ConditionNeedsUpdate= has been added that
3471 may conditionalize services to only run when /etc or /var
3472 are "older" than the vendor operating system resources in
3473 /usr. This is useful for reconstructing or updating /etc
3474 after an offline update of /usr or a factory reset, on the
3475 next reboot. Services that want to run once after such an
3476 update or reset should use this condition and order
3477 themselves before the new systemd-update-done.service, which
3478 will mark the two directories as fully updated. A number of
3479 service files have been added making use of this, to rebuild
3480 the udev hardware database, the journald message catalog and
3481 dynamic loader cache (ldconfig). The systemd-sysusers tool
3482 described above also makes use of this now. With this in
3483 place it is now possible to start up a minimal operating
3484 system with /etc empty cleanly. For more information on the
3485 concepts involved see this recent blog story:
3487 http://0pointer.de/blog/projects/stateless.html
3489 * A new system group "input" has been introduced, and all
3490 input device nodes get this group assigned. This is useful
3491 for system-level software to get access to input devices. It
3492 complements what is already done for "audio" and "video".
3494 * systemd-networkd learnt minimal DHCPv4 server support in
3495 addition to the existing DHCPv4 client support. It also
3496 learnt DHCPv6 client and IPv6 Router Solicitation client
3497 support. The DHCPv4 client gained support for static routes
3498 passed in from the server. Note that the [DHCPv4] section
3499 known in older systemd-networkd versions has been renamed to
3500 [DHCP] and is now also used by the DHCPv6 client. Existing
3501 .network files using settings of this section should be
3502 updated, though compatibility is maintained. Optionally, the
3503 client hostname may now be sent to the DHCP server.
3505 * networkd gained support for vxlan virtual networks as well
3506 as tun/tap and dummy devices.
3508 * networkd gained support for automatic allocation of address
3509 ranges for interfaces from a system-wide pool of
3510 addresses. This is useful for dynamically managing a large
3511 number of interfaces with a single network configuration
3512 file. In particular this is useful to easily assign
3513 appropriate IP addresses to the veth links of a large number
3514 of nspawn instances.
3516 * RPM macros for processing sysusers, sysctl and binfmt
3517 drop-in snippets at package installation time have been
3520 * The /etc/os-release file should now be placed in
3521 /usr/lib/os-release. The old location is automatically
3522 created as symlink. /usr/lib is the more appropriate
3523 location of this file, since it shall actually describe the
3524 vendor operating system shipped in /usr, and not the
3525 configuration stored in /etc.
3527 * .mount units gained a new boolean SloppyOptions= setting
3528 that maps to mount(8)'s -s option which enables permissive
3529 parsing of unknown mount options.
3531 * tmpfiles learnt a new "L+" directive which creates a symlink
3532 but (unlike "L") deletes a pre-existing file first, should
3533 it already exist and not already be the correct
3534 symlink. Similarly, "b+", "c+" and "p+" directives have been
3535 added as well, which create block and character devices, as
3536 well as fifos in the filesystem, possibly removing any
3537 pre-existing files of different types.
3539 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
3540 'argument' field (which so far specified the source to
3541 symlink/copy the files from) is now optional. If omitted the
3542 same file os copied from /usr/share/factory/ suffixed by the
3543 full destination path. This is useful for populating /etc
3544 with essential files, by copying them from vendor defaults
3545 shipped in /usr/share/factory/etc.
3547 * A new command "systemctl preset-all" has been added that
3548 applies the service preset settings to all installed unit
3549 files. A new switch --preset-mode= has been added that
3550 controls whether only enable or only disable operations
3553 * A new command "systemctl is-system-running" has been added
3554 that allows checking the overall state of the system, for
3555 example whether it is fully up and running.
3557 * When the system boots up with an empty /etc, the equivalent
3558 to "systemctl preset-all" is executed during early boot, to
3559 make sure all default services are enabled after a factory
3562 * systemd now contains a minimal preset file that enables the
3563 most basic services systemd ships by default.
3565 * Unit files' [Install] section gained a new DefaultInstance=
3566 field for defining the default instance to create if a
3567 template unit is enabled with no instance specified.
3569 * A new passive target cryptsetup-pre.target has been added
3570 that may be used by services that need to make they run and
3571 finish before the first LUKS cryptographic device is set up.
3573 * The /dev/loop-control and /dev/btrfs-control device nodes
3574 are now owned by the "disk" group by default, opening up
3575 access to this group.
3577 * systemd-coredump will now automatically generate a
3578 stack trace of all core dumps taking place on the system,
3579 based on elfutils' libdw library. This stack trace is logged
3582 * systemd-coredump may now optionally store coredumps directly
3583 on disk (in /var/lib/systemd/coredump, possibly compressed),
3584 instead of storing them unconditionally in the journal. This
3585 mode is the new default. A new configuration file
3586 /etc/systemd/coredump.conf has been added to configure this
3587 and other parameters of systemd-coredump.
3589 * coredumpctl gained a new "info" verb to show details about a
3590 specific coredump. A new switch "-1" has also been added
3591 that makes sure to only show information about the most
3592 recent entry instead of all entries. Also, as the tool is
3593 generally useful now the "systemd-" prefix of the binary
3594 name has been removed. Distributions that want to maintain
3595 compatibility with the old name should add a symlink from
3596 the old name to the new name.
3598 * journald's SplitMode= now defaults to "uid". This makes sure
3599 that unprivileged users can access their own coredumps with
3600 coredumpctl without restrictions.
3602 * New kernel command line options "systemd.wants=" (for
3603 pulling an additional unit during boot), "systemd.mask="
3604 (for masking a specific unit for the boot), and
3605 "systemd.debug-shell" (for enabling the debug shell on tty9)
3606 have been added. This is implemented in the new generator
3607 "systemd-debug-generator".
3609 * systemd-nspawn will now by default filter a couple of
3610 syscalls for containers, among them those required for
3611 kernel module loading, direct x86 IO port access, swap
3612 management, and kexec. Most importantly though
3613 open_by_handle_at() is now prohibited for containers,
3614 closing a hole similar to a recently discussed vulnerability
3615 in docker regarding access to files on file hierarchies the
3616 container should normally not have access to. Note that, for
3617 nspawn, we generally make no security claims anyway (and
3618 this is explicitly documented in the man page), so this is
3619 just a fix for one of the most obvious problems.
3621 * A new man page file-hierarchy(7) has been added that
3622 contains a minimized, modernized version of the file system
3623 layout systemd expects, similar in style to the FHS
3624 specification or hier(5). A new tool systemd-path(1) has
3625 been added to query many of these paths for the local
3628 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
3629 longer done. Since the directory now has a per-user size
3630 limit, and is cleaned on logout this appears unnecessary,
3631 in particular since this now brings the lifecycle of this
3632 directory closer in line with how IPC objects are handled.
3634 * systemd.pc now exports a number of additional directories,
3635 including $libdir (which is useful to identify the library
3636 path for the primary architecture of the system), and a
3637 couple of drop-in directories.
3639 * udev's predictable network interface names now use the dev_port
3640 sysfs attribute, introduced in linux 3.15 instead of dev_id to
3641 distinguish between ports of the same PCI function. dev_id should
3642 only be used for ports using the same HW address, hence the need
3645 * machined has been updated to export the OS version of a
3646 container (read from /etc/os-release and
3647 /usr/lib/os-release) on the bus. This is now shown in
3648 "machinectl status" for a machine.
3650 * A new service setting RestartForceExitStatus= has been
3651 added. If configured to a set of exit signals or process
3652 return values, the service will be restarted when the main
3653 daemon process exits with any of them, regardless of the
3656 * systemctl's -H switch for connecting to remote systemd
3657 machines has been extended so that it may be used to
3658 directly connect to a specific container on the
3659 host. "systemctl -H root@foobar:waldi" will now connect as
3660 user "root" to host "foobar", and then proceed directly to
3661 the container named "waldi". Note that currently you have to
3662 authenticate as user "root" for this to work, as entering
3663 containers is a privileged operation.
3665 Contributions from: Andreas Henriksson, Benjamin Steinwender,
3666 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
3667 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
3668 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
3669 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
3670 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
3671 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
3672 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
3673 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
3674 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
3675 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
3676 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
3678 — Berlin, 2014-07-03
3682 * As an experimental feature, udev now tries to lock the
3683 disk device node (flock(LOCK_SH|LOCK_NB)) while it
3684 executes events for the disk or any of its partitions.
3685 Applications like partitioning programs can lock the
3686 disk device node (flock(LOCK_EX)) and claim temporary
3687 device ownership that way; udev will entirely skip all event
3688 handling for this disk and its partitions. If the disk
3689 was opened for writing, the close will trigger a partition
3690 table rescan in udev's "watch" facility, and if needed
3691 synthesize "change" events for the disk and all its partitions.
3692 This is now unconditionally enabled, and if it turns out to
3693 cause major problems, we might turn it on only for specific
3694 devices, or might need to disable it entirely. Device Mapper
3695 devices are excluded from this logic.
3697 * We temporarily dropped the "-l" switch for fsck invocations,
3698 since they collide with the flock() logic above. util-linux
3699 upstream has been changed already to avoid this conflict,
3700 and we will readd "-l" as soon as util-linux with this
3701 change has been released.
3703 * The dependency on libattr has been removed. Since a long
3704 time, the extended attribute calls have moved to glibc, and
3705 libattr is thus unnecessary.
3707 * Virtualization detection works without privileges now. This
3708 means the systemd-detect-virt binary no longer requires
3709 CAP_SYS_PTRACE file capabilities, and our daemons can run
3710 with fewer privileges.
3712 * systemd-networkd now runs under its own "systemd-network"
3713 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
3714 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
3715 loses the ability to write to files owned by root this way.
3717 * Similarly, systemd-resolved now runs under its own
3718 "systemd-resolve" user with no capabilities remaining.
3720 * Similarly, systemd-bus-proxyd now runs under its own
3721 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
3723 * systemd-networkd gained support for setting up "veth"
3724 virtual Ethernet devices for container connectivity, as well
3725 as GRE and VTI tunnels.
3727 * systemd-networkd will no longer automatically attempt to
3728 manually load kernel modules necessary for certain tunnel
3729 transports. Instead, it is assumed the kernel loads them
3730 automatically when required. This only works correctly on
3731 very new kernels. On older kernels, please consider adding
3732 the kernel modules to /etc/modules-load.d/ as a work-around.
3734 * The resolv.conf file systemd-resolved generates has been
3735 moved to /run/systemd/resolve/. If you have a symlink from
3736 /etc/resolv.conf, it might be necessary to correct it.
3738 * Two new service settings, ProtectHome= and ProtectSystem=,
3739 have been added. When enabled, they will make the user data
3740 (such as /home) inaccessible or read-only and the system
3741 (such as /usr) read-only, for specific services. This allows
3742 very light-weight per-service sandboxing to avoid
3743 modifications of user data or system files from
3744 services. These two new switches have been enabled for all
3745 of systemd's long-running services, where appropriate.
3747 * Socket units gained new SocketUser= and SocketGroup=
3748 settings to set the owner user and group of AF_UNIX sockets
3749 and FIFOs in the file system.
3751 * Socket units gained a new RemoveOnStop= setting. If enabled,
3752 all FIFOS and sockets in the file system will be removed
3753 when the specific socket unit is stopped.
3755 * Socket units gained a new Symlinks= setting. It takes a list
3756 of symlinks to create to file system sockets or FIFOs
3757 created by the specific Unix sockets. This is useful to
3758 manage symlinks to socket nodes with the same life-cycle as
3761 * The /dev/log socket and /dev/initctl FIFO have been moved to
3762 /run, and have been replaced by symlinks. This allows
3763 connecting to these facilities even if PrivateDevices=yes is
3764 used for a service (which makes /dev/log itself unavailable,
3765 but /run is left). This also has the benefit of ensuring
3766 that /dev only contains device nodes, directories and
3767 symlinks, and nothing else.
3769 * sd-daemon gained two new calls sd_pid_notify() and
3770 sd_pid_notifyf(). They are similar to sd_notify() and
3771 sd_notifyf(), but allow overriding of the source PID of
3772 notification messages if permissions permit this. This is
3773 useful to send notify messages on behalf of a different
3774 process (for example, the parent process). The
3775 systemd-notify tool has been updated to make use of this
3776 when sending messages (so that notification messages now
3777 originate from the shell script invoking systemd-notify and
3778 not the systemd-notify process itself. This should minimize
3779 a race where systemd fails to associate notification
3780 messages to services when the originating process already
3783 * A new "on-abnormal" setting for Restart= has been added. If
3784 set, it will result in automatic restarts on all "abnormal"
3785 reasons for a process to exit, which includes unclean
3786 signals, core dumps, timeouts and watchdog timeouts, but
3787 does not include clean and unclean exit codes or clean
3788 signals. Restart=on-abnormal is an alternative for
3789 Restart=on-failure for services that shall be able to
3790 terminate and avoid restarts on certain errors, by
3791 indicating so with an unclean exit code. Restart=on-failure
3792 or Restart=on-abnormal is now the recommended setting for
3793 all long-running services.
3795 * If the InaccessibleDirectories= service setting points to a
3796 mount point (or if there are any submounts contained within
3797 it), it is now attempted to completely unmount it, to make
3798 the file systems truly unavailable for the respective
3801 * The ReadOnlyDirectories= service setting and
3802 systemd-nspawn's --read-only parameter are now recursively
3803 applied to all submounts, too.
3805 * Mount units may now be created transiently via the bus APIs.
3807 * The support for SysV and LSB init scripts has been removed
3808 from the systemd daemon itself. Instead, it is now
3809 implemented as a generator that creates native systemd units
3810 from these scripts when needed. This enables us to remove a
3811 substantial amount of legacy code from PID 1, following the
3812 fact that many distributions only ship a very small number
3813 of LSB/SysV init scripts nowadays.
3815 * Privileged Xen (dom0) domains are not considered
3816 virtualization anymore by the virtualization detection
3817 logic. After all, they generally have unrestricted access to
3818 the hardware and usually are used to manage the unprivileged
3821 * systemd-tmpfiles gained a new "C" line type, for copying
3822 files or entire directories.
3824 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
3825 lines. So far, they have been non-globbing versions of the
3826 latter, and have thus been redundant. In future, it is
3827 recommended to only use "z". "m" has hence been removed
3828 from the documentation, even though it stays supported.
3830 * A tmpfiles snippet to recreate the most basic structure in
3831 /var has been added. This is enough to create the /var/run →
3832 /run symlink and create a couple of structural
3833 directories. This allows systems to boot up with an empty or
3834 volatile /var. Of course, while with this change, the core OS
3835 now is capable with dealing with a volatile /var, not all
3836 user services are ready for it. However, we hope that sooner
3837 or later, many service daemons will be changed upstream so
3838 that they are able to automatically create their necessary
3839 directories in /var at boot, should they be missing. This is
3840 the first step to allow state-less systems that only require
3841 the vendor image for /usr to boot.
3843 * systemd-nspawn has gained a new --tmpfs= switch to mount an
3844 empty tmpfs instance to a specific directory. This is
3845 particularly useful for making use of the automatic
3846 reconstruction of /var (see above), by passing --tmpfs=/var.
3848 * Access modes specified in tmpfiles snippets may now be
3849 prefixed with "~", which indicates that they shall be masked
3850 by whether the existing file or directory is currently
3851 writable, readable or executable at all. Also, if specified,
3852 the sgid/suid/sticky bits will be masked for all
3855 * A new passive target unit "network-pre.target" has been
3856 added which is useful for services that shall run before any
3857 network is configured, for example firewall scripts.
3859 * The "floppy" group that previously owned the /dev/fd*
3860 devices is no longer used. The "disk" group is now used
3861 instead. Distributions should probably deprecate usage of
3864 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
3865 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
3866 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
3867 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
3868 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
3869 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
3872 — Berlin, 2014-06-11
3876 * A new "systemd-timesyncd" daemon has been added for
3877 synchronizing the system clock across the network. It
3878 implements an SNTP client. In contrast to NTP
3879 implementations such as chrony or the NTP reference server,
3880 this only implements a client side, and does not bother with
3881 the full NTP complexity, focusing only on querying time from
3882 one remote server and synchronizing the local clock to
3883 it. Unless you intend to serve NTP to networked clients or
3884 want to connect to local hardware clocks, this simple NTP
3885 client should be more than appropriate for most
3886 installations. The daemon runs with minimal privileges, and
3887 has been hooked up with networkd to only operate when
3888 network connectivity is available. The daemon saves the
3889 current clock to disk every time a new NTP sync has been
3890 acquired, and uses this to possibly correct the system clock
3891 early at bootup, in order to accommodate for systems that
3892 lack an RTC such as the Raspberry Pi and embedded devices,
3893 and to make sure that time monotonically progresses on these
3894 systems, even if it is not always correct. To make use of
3895 this daemon, a new system user and group "systemd-timesync"
3896 needs to be created on installation of systemd.
3898 * The queue "seqnum" interface of libudev has been disabled, as
3899 it was generally incompatible with device namespacing as
3900 sequence numbers of devices go "missing" if the devices are
3901 part of a different namespace.
3903 * "systemctl list-timers" and "systemctl list-sockets" gained
3904 a --recursive switch for showing units of these types also
3905 for all local containers, similar in style to the already
3906 supported --recursive switch for "systemctl list-units".
3908 * A new RebootArgument= setting has been added for service
3909 units, which may be used to specify a kernel reboot argument
3910 to use when triggering reboots with StartLimitAction=.
3912 * A new FailureAction= setting has been added for service
3913 units which may be used to specify an operation to trigger
3914 when a service fails. This works similarly to
3915 StartLimitAction=, but unlike it, controls what is done
3916 immediately rather than only after several attempts to
3917 restart the service in question.
3919 * hostnamed got updated to also expose the kernel name,
3920 release, and version on the bus. This is useful for
3921 executing commands like hostnamectl with the -H switch.
3922 systemd-analyze makes use of this to properly display
3923 details when running non-locally.
3925 * The bootchart tool can now show cgroup information in the
3926 graphs it generates.
3928 * The CFS CPU quota cgroup attribute is now exposed for
3929 services. The new CPUQuota= switch has been added for this
3930 which takes a percentage value. Setting this will have the
3931 result that a service may never get more CPU time than the
3932 specified percentage, even if the machine is otherwise idle.
3934 * systemd-networkd learned IPIP and SIT tunnel support.
3936 * LSB init scripts exposing a dependency on $network will now
3937 get a dependency on network-online.target rather than simply
3938 network.target. This should bring LSB handling closer to
3939 what it was on SysV systems.
3941 * A new fsck.repair= kernel option has been added to control
3942 how fsck shall deal with unclean file systems at boot.
3944 * The (.ini) configuration file parser will now silently
3945 ignore sections whose name begins with "X-". This may be
3946 used to maintain application-specific extension sections in unit
3949 * machined gained a new API to query the IP addresses of
3950 registered containers. "machinectl status" has been updated
3951 to show these addresses in its output.
3953 * A new call sd_uid_get_display() has been added to the
3954 sd-login APIs for querying the "primary" session of a
3955 user. The "primary" session of the user is elected from the
3956 user's sessions and generally a graphical session is
3957 preferred over a text one.
3959 * A minimal systemd-resolved daemon has been added. It
3960 currently simply acts as a companion to systemd-networkd and
3961 manages resolv.conf based on per-interface DNS
3962 configuration, possibly supplied via DHCP. In the long run
3963 we hope to extend this into a local DNSSEC enabled DNS and
3966 * The systemd-networkd-wait-online tool is now enabled by
3967 default. It will delay network-online.target until a network
3968 connection has been configured. The tool primarily integrates
3969 with networkd, but will also make a best effort to make sense
3970 of network configuration performed in some other way.
3972 * Two new service options StartupCPUShares= and
3973 StartupBlockIOWeight= have been added that work similarly to
3974 CPUShares= and BlockIOWeight= however only apply during
3975 system startup. This is useful to prioritize certain services
3976 differently during bootup than during normal runtime.
3978 * hostnamed has been changed to prefer the statically
3979 configured hostname in /etc/hostname (unless set to
3980 'localhost' or empty) over any dynamic one supplied by
3981 dhcp. With this change, the rules for picking the hostname
3982 match more closely the rules of other configuration settings
3983 where the local administrator's configuration in /etc always
3984 overrides any other settings.
3986 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
3987 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
3988 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
3989 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
3990 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
3991 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
3992 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
3993 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
3994 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
3995 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
3996 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
3997 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
3998 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
3999 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
4000 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
4001 Lindskog, WaLyong Cho, Will Woods, Zbigniew
4004 — Beijing, 2014-05-28
4008 * When restoring the screen brightness at boot, stay away from
4009 the darkest setting or from the lowest 5% of the available
4010 range, depending on which is the larger value of both. This
4011 should effectively protect the user from rebooting into a
4012 black screen, should the brightness have been set to minimum
4015 * sd-login gained a new sd_machine_get_class() call to
4016 determine the class ("vm" or "container") of a machine
4017 registered with machined.
4019 * sd-login gained new calls
4020 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
4021 to query the identity of the peer of a local AF_UNIX
4022 connection. They operate similarly to their sd_pid_get_xyz()
4025 * PID 1 will now maintain a system-wide system state engine
4026 with the states "starting", "running", "degraded",
4027 "maintenance", "stopping". These states are bound to system
4028 startup, normal runtime, runtime with at least one failed
4029 service, rescue/emergency mode and system shutdown. This
4030 state is shown in the "systemctl status" output when no unit
4031 name is passed. It is useful to determine system state, in
4032 particularly when doing so for many systems or containers at
4035 * A new command "list-machines" has been added to "systemctl"
4036 that lists all local OS containers and shows their system
4037 state (see above), if systemd runs inside of them.
4039 * systemctl gained a new "-r" switch to recursively enumerate
4040 units on all local containers, when used with the
4041 "list-unit" command (which is the default one that is
4042 executed when no parameters are specified).
4044 * The GPT automatic partition discovery logic will now honour
4045 two GPT partition flags: one may be set on a partition to
4046 cause it to be mounted read-only, and the other may be set
4047 on a partition to ignore it during automatic discovery.
4049 * Two new GPT type UUIDs have been added for automatic root
4050 partition discovery, for 32-bit and 64-bit ARM. This is not
4051 particularly useful for discovering the root directory on
4052 these architectures during bare-metal boots (since UEFI is
4053 not common there), but still very useful to allow booting of
4054 ARM disk images in nspawn with the -i option.
4056 * MAC addresses of interfaces created with nspawn's
4057 --network-interface= switch will now be generated from the
4058 machine name, and thus be stable between multiple invocations
4061 * logind will now automatically remove all IPC objects owned
4062 by a user if she or he fully logs out. This makes sure that
4063 users who are logged out cannot continue to consume IPC
4064 resources. This covers SysV memory, semaphores and message
4065 queues as well as POSIX shared memory and message
4066 queues. Traditionally, SysV and POSIX IPC had no life-cycle
4067 limits. With this functionality, that is corrected. This may
4068 be turned off by using the RemoveIPC= switch of logind.conf.
4070 * The systemd-machine-id-setup and tmpfiles tools gained a
4071 --root= switch to operate on a specific root directory,
4074 * journald can now forward logged messages to the TTYs of all
4075 logged in users ("wall"). This is the default for all
4076 emergency messages now.
4078 * A new tool systemd-journal-remote has been added to stream
4079 journal log messages across the network.
4081 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
4082 controller trees are mounted into it. Note that the
4083 directories mounted beneath it are not read-only. This is a
4084 security measure and is particularly useful because glibc
4085 actually includes a search logic to pick any tmpfs it can
4086 find to implement shm_open() if /dev/shm is not available
4087 (which it might very well be in namespaced setups).
4089 * machinectl gained a new "poweroff" command to cleanly power
4090 down a local OS container.
4092 * The PrivateDevices= unit file setting will now also drop the
4093 CAP_MKNOD capability from the capability bound set, and
4094 imply DevicePolicy=closed.
4096 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
4097 comprehensively on all long-running systemd services where
4098 this is appropriate.
4100 * systemd-udevd will now run in a disassociated mount
4101 namespace. To mount directories from udev rules, make sure to
4102 pull in mount units via SYSTEMD_WANTS properties.
4104 * The kdbus support gained support for uploading policy into
4105 the kernel. sd-bus gained support for creating "monitoring"
4106 connections that can eavesdrop into all bus communication
4107 for debugging purposes.
4109 * Timestamps may now be specified in seconds since the UNIX
4110 epoch Jan 1st, 1970 by specifying "@" followed by the value
4113 * Native tcpwrap support in systemd has been removed. tcpwrap
4114 is old code, not really maintained anymore and has serious
4115 shortcomings, and better options such as firewalls
4116 exist. For setups that require tcpwrap usage, please
4117 consider invoking your socket-activated service via tcpd,
4118 like on traditional inetd.
4120 * A new system.conf configuration option
4121 DefaultTimerAccuracySec= has been added that controls the
4122 default AccuracySec= setting of .timer units.
4124 * Timer units gained a new WakeSystem= switch. If enabled,
4125 timers configured this way will cause the system to resume
4126 from system suspend (if the system supports that, which most
4129 * Timer units gained a new Persistent= switch. If enabled,
4130 timers configured this way will save to disk when they have
4131 been last triggered. This information is then used on next
4132 reboot to possible execute overdue timer events, that
4133 could not take place because the system was powered off.
4134 This enables simple anacron-like behaviour for timer units.
4136 * systemctl's "list-timers" will now also list the time a
4137 timer unit was last triggered in addition to the next time
4138 it will be triggered.
4140 * systemd-networkd will now assign predictable IPv4LL
4141 addresses to its local interfaces.
4143 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
4144 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
4145 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
4146 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
4147 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
4148 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
4149 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
4150 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
4153 — Berlin, 2014-03-25
4157 * A new unit file setting RestrictAddressFamilies= has been
4158 added to restrict which socket address families unit
4159 processes gain access to. This takes address family names
4160 like "AF_INET" or "AF_UNIX", and is useful to minimize the
4161 attack surface of services via exotic protocol stacks. This
4162 is built on seccomp system call filters.
4164 * Two new unit file settings RuntimeDirectory= and
4165 RuntimeDirectoryMode= have been added that may be used to
4166 manage a per-daemon runtime directories below /run. This is
4167 an alternative for setting up directory permissions with
4168 tmpfiles snippets, and has the advantage that the runtime
4169 directory's lifetime is bound to the daemon runtime and that
4170 the daemon starts up with an empty directory each time. This
4171 is particularly useful when writing services that drop
4172 privileges using the User= or Group= setting.
4174 * The DeviceAllow= unit setting now supports globbing for
4175 matching against device group names.
4177 * The systemd configuration file system.conf gained new
4178 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
4179 DefaultMemoryAccounting= to globally turn on/off accounting
4180 for specific resources (cgroups) for all units. These
4181 settings may still be overridden individually in each unit
4184 * systemd-gpt-auto-generator is now able to discover /srv and
4185 root partitions in addition to /home and swap partitions. It
4186 also supports LUKS-encrypted partitions now. With this in
4187 place, automatic discovery of partitions to mount following
4188 the Discoverable Partitions Specification
4189 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
4190 is now a lot more complete. This allows booting without
4191 /etc/fstab and without root= on the kernel command line on
4192 systems prepared appropriately.
4194 * systemd-nspawn gained a new --image= switch which allows
4195 booting up disk images and Linux installations on any block
4196 device that follow the Discoverable Partitions Specification
4197 (see above). This means that installations made with
4198 appropriately updated installers may now be started and
4199 deployed using container managers, completely
4200 unmodified. (We hope that libvirt-lxc will add support for
4201 this feature soon, too.)
4203 * systemd-nspawn gained a new --network-macvlan= setting to
4204 set up a private macvlan interface for the
4205 container. Similarly, systemd-networkd gained a new
4206 Kind=macvlan setting in .netdev files.
4208 * systemd-networkd now supports configuring local addresses
4211 * A new tool systemd-network-wait-online has been added to
4212 synchronously wait for network connectivity using
4215 * The sd-bus.h bus API gained a new sd_bus_track object for
4216 tracking the life-cycle of bus peers. Note that sd-bus.h is
4217 still not a public API though (unless you specify
4218 --enable-kdbus on the configure command line, which however
4219 voids your warranty and you get no API stability guarantee).
4221 * The $XDG_RUNTIME_DIR runtime directories for each user are
4222 now individual tmpfs instances, which has the benefit of
4223 introducing separate pools for each user, with individual
4224 size limits, and thus making sure that unprivileged clients
4225 can no longer negatively impact the system or other users by
4226 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
4227 RuntimeDirectorySize= has been introduced that allows
4228 controlling the default size limit for all users. It
4229 defaults to 10% of the available physical memory. This is no
4230 replacement for quotas on tmpfs though (which the kernel
4231 still does not support), as /dev/shm and /tmp are still
4232 shared resources used by both the system and unprivileged
4235 * logind will now automatically turn off automatic suspending
4236 on laptop lid close when more than one display is
4237 connected. This was previously expected to be implemented
4238 individually in desktop environments (such as GNOME),
4239 however has been added to logind now, in order to fix a
4240 boot-time race where a desktop environment might not have
4241 been started yet and thus not been able to take an inhibitor
4242 lock at the time where logind already suspends the system
4243 due to a closed lid.
4245 * logind will now wait at least 30s after each system
4246 suspend/resume cycle, and 3min after system boot before
4247 suspending the system due to a closed laptop lid. This
4248 should give USB docking stations and similar enough time to
4249 be probed and configured after system resume and boot in
4250 order to then act as suspend blocker.
4252 * systemd-run gained a new --property= setting which allows
4253 initialization of resource control properties (and others)
4254 for the created scope or service unit. Example: "systemd-run
4255 --property=BlockIOWeight=10 updatedb" may be used to run
4256 updatedb at a low block IO scheduling weight.
4258 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
4259 now also work in --scope mode.
4261 * When systemd is compiled with kdbus support, basic support
4262 for enforced policies is now in place. (Note that enabling
4263 kdbus still voids your warranty and no API compatibility
4266 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
4267 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
4268 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
4269 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
4270 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
4271 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
4272 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
4273 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
4274 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
4275 Zbigniew Jędrzejewski-Szmek
4277 — Berlin, 2014-03-12
4281 * systemd will now relabel /dev after loading the SMACK policy
4282 according to SMACK rules.
4284 * A new unit file option AppArmorProfile= has been added to
4285 set the AppArmor profile for the processes of a unit.
4287 * A new condition check ConditionArchitecture= has been added
4288 to conditionalize units based on the system architecture, as
4289 reported by uname()'s "machine" field.
4291 * systemd-networkd now supports matching on the system
4292 virtualization, architecture, kernel command line, host name
4295 * logind is now a lot more aggressive when suspending the
4296 machine due to a closed laptop lid. Instead of acting only
4297 on the lid close action, it will continuously watch the lid
4298 status and act on it. This is useful for laptops where the
4299 power button is on the outside of the chassis so that it can
4300 be reached without opening the lid (such as the Lenovo
4301 Yoga). On those machines, logind will now immediately
4302 re-suspend the machine if the power button has been
4303 accidentally pressed while the laptop was suspended and in a
4304 backpack or similar.
4306 * logind will now watch SW_DOCK switches and inhibit reaction
4307 to the lid switch if it is pressed. This means that logind
4308 will not suspend the machine anymore if the lid is closed
4309 and the system is docked, if the laptop supports SW_DOCK
4310 notifications via the input layer. Note that ACPI docking
4311 stations do not generate this currently. Also note that this
4312 logic is usually not fully sufficient and Desktop
4313 Environments should take a lid switch inhibitor lock when an
4314 external display is connected, as systemd will not watch
4317 * nspawn will now make use of the devices cgroup controller by
4318 default, and only permit creation of and access to the usual
4319 API device nodes like /dev/null or /dev/random, as well as
4320 access to (but not creation of) the pty devices.
4322 * We will now ship a default .network file for
4323 systemd-networkd that automatically configures DHCP for
4324 network interfaces created by nspawn's --network-veth or
4325 --network-bridge= switches.
4327 * systemd will now understand the usual M, K, G, T suffixes
4328 according to SI conventions (i.e. to the base 1000) when
4329 referring to throughput and hardware metrics. It will stay
4330 with IEC conventions (i.e. to the base 1024) for software
4331 metrics, according to what is customary according to
4332 Wikipedia. We explicitly document which base applies for
4333 each configuration option.
4335 * The DeviceAllow= setting in unit files now supports a syntax
4336 to whitelist an entire group of devices node majors at once,
4337 based on the /proc/devices listing. For example, with the
4338 string "char-pts", it is now possible to whitelist all
4339 current and future pseudo-TTYs at once.
4341 * sd-event learned a new "post" event source. Event sources of
4342 this type are triggered by the dispatching of any event
4343 source of a type that is not "post". This is useful for
4344 implementing clean-up and check event sources that are
4345 triggered by other work being done in the program.
4347 * systemd-networkd is no longer statically enabled, but uses
4348 the usual [Install] sections so that it can be
4349 enabled/disabled using systemctl. It still is enabled by
4352 * When creating a veth interface pair with systemd-nspawn, the
4353 host side will now be prefixed with "vb-" if
4354 --network-bridge= is used, and with "ve-" if --network-veth
4355 is used. This way, it is easy to distinguish these cases on
4356 the host, for example to apply different configuration to
4357 them with systemd-networkd.
4359 * The compatibility libraries for libsystemd-journal.so,
4360 libsystem-id128.so, libsystemd-login.so and
4361 libsystemd-daemon.so do not make use of IFUNC
4362 anymore. Instead, we now build libsystemd.so multiple times
4363 under these alternative names. This means that the footprint
4364 is drastically increased, but given that these are
4365 transitional compatibility libraries, this should not matter
4366 much. This change has been made necessary to support the ARM
4367 platform for these compatibility libraries, as the ARM
4368 toolchain is not really at the same level as the toolchain
4369 for other architectures like x86 and does not support
4370 IFUNC. Please make sure to use --enable-compat-libs only
4371 during a transitional period!
4373 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
4374 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
4375 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
4376 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
4377 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
4378 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
4379 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
4380 Zbigniew Jędrzejewski-Szmek
4382 — Berlin, 2014-02-24
4386 * A new component "systemd-networkd" has been added that can
4387 be used to configure local network interfaces statically or
4388 via DHCP. It is capable of bringing up bridges, VLANs, and
4389 bonding. Currently, no hook-ups for interactive network
4390 configuration are provided. Use this for your initrd,
4391 container, embedded, or server setup if you need a simple,
4392 yet powerful, network configuration solution. This
4393 configuration subsystem is quite nifty, as it allows wildcard
4394 hotplug matching in interfaces. For example, with a single
4395 configuration snippet, you can configure that all Ethernet
4396 interfaces showing up are automatically added to a bridge,
4397 or similar. It supports link-sensing and more.
4399 * A new tool "systemd-socket-proxyd" has been added which can
4400 act as a bidirectional proxy for TCP sockets. This is
4401 useful for adding socket activation support to services that
4402 do not actually support socket activation, including virtual
4403 machines and the like.
4405 * Add a new tool to save/restore rfkill state on
4408 * Save/restore state of keyboard backlights in addition to
4409 display backlights on shutdown/boot.
4411 * udev learned a new SECLABEL{} construct to label device
4412 nodes with a specific security label when they appear. For
4413 now, only SECLABEL{selinux} is supported, but the syntax is
4414 prepared for additional security frameworks.
4416 * udev gained a new scheme to configure link-level attributes
4417 from files in /etc/systemd/network/*.link. These files can
4418 match against MAC address, device path, driver name and type,
4419 and will apply attributes like the naming policy, link speed,
4420 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
4421 address assignment policy (randomized, ...).
4423 * The configuration of network interface naming rules for
4424 "permanent interface names" has changed: a new NamePolicy=
4425 setting in the [Link] section of .link files determines the
4426 priority of possible naming schemes (onboard, slot, MAC,
4427 path). The default value of this setting is determined by
4428 /usr/lib/net/links/99-default.link. Old
4429 80-net-name-slot.rules udev configuration file has been
4430 removed, so local configuration overriding this file should
4431 be adapted to override 99-default.link instead.
4433 * When the User= switch is used in a unit file, also
4434 initialize $SHELL= based on the user database entry.
4436 * systemd no longer depends on libdbus. All communication is
4437 now done with sd-bus, systemd's low-level bus library
4440 * kdbus support has been added to PID 1 itself. When kdbus is
4441 enabled, this causes PID 1 to set up the system bus and
4442 enable support for a new ".busname" unit type that
4443 encapsulates bus name activation on kdbus. It works a little
4444 bit like ".socket" units, except for bus names. A new
4445 generator has been added that converts classic dbus1 service
4446 activation files automatically into native systemd .busname
4449 * sd-bus: add a light-weight vtable implementation that allows
4450 defining objects on the bus with a simple static const
4451 vtable array of its methods, signals and properties.
4453 * systemd will not generate or install static dbus
4454 introspection data anymore to /usr/share/dbus-1/interfaces,
4455 as the precise format of these files is unclear, and
4456 nothing makes use of it.
4458 * A proxy daemon is now provided to proxy clients connecting
4459 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
4460 compatibility with classic D-Bus.
4462 * A bus driver implementation has been added that supports the
4463 classic D-Bus bus driver calls on kdbus, also for
4464 compatibility purposes.
4466 * A new API "sd-event.h" has been added that implements a
4467 minimal event loop API built around epoll. It provides a
4468 couple of features that direct epoll usage is lacking:
4469 prioritization of events, scales to large numbers of timer
4470 events, per-event timer slack (accuracy), system-wide
4471 coalescing of timer events, exit handlers, watchdog
4472 supervision support using systemd's sd_notify() API, child
4475 * A new API "sd-rntl.h" has been added that provides an API
4476 around the route netlink interface of the kernel, similar in
4477 style to "sd-bus.h".
4479 * A new API "sd-dhcp-client.h" has been added that provides a
4480 small DHCPv4 client-side implementation. This is used by
4483 * There is a new kernel command line option
4484 "systemd.restore_state=0|1". When set to "0", none of the
4485 systemd tools will restore saved runtime state to hardware
4486 devices. More specifically, the rfkill and backlight states
4489 * The FsckPassNo= compatibility option in mount/service units
4490 has been removed. The fstab generator will now add the
4491 necessary dependencies automatically, and does not require
4492 PID1's support for that anymore.
4494 * journalctl gained a new switch, --list-boots, that lists
4495 recent boots with their times and boot IDs.
4497 * The various tools like systemctl, loginctl, timedatectl,
4498 busctl, systemd-run, ... have gained a new switch "-M" to
4499 connect to a specific, local OS container (as direct
4500 connection, without requiring SSH). This works on any
4501 container that is registered with machined, such as those
4502 created by libvirt-lxc or nspawn.
4504 * systemd-run and systemd-analyze also gained support for "-H"
4505 to connect to remote hosts via SSH. This is particularly
4506 useful for systemd-run because it enables queuing of jobs
4507 onto remote systems.
4509 * machinectl gained a new command "login" to open a getty
4510 login in any local container. This works with any container
4511 that is registered with machined (such as those created by
4512 libvirt-lxc or nspawn), and which runs systemd inside.
4514 * machinectl gained a new "reboot" command that may be used to
4515 trigger a reboot on a specific container that is registered
4516 with machined. This works on any container that runs an init
4517 system of some kind.
4519 * systemctl gained a new "list-timers" command to print a nice
4520 listing of installed timer units with the times they elapse
4523 * Alternative reboot() parameters may now be specified on the
4524 "systemctl reboot" command line and are passed to the
4525 reboot() system call.
4527 * systemctl gained a new --job-mode= switch to configure the
4528 mode to queue a job with. This is a more generic version of
4529 --fail, --irreversible, and --ignore-dependencies, which are
4530 still available but not advertised anymore.
4532 * /etc/systemd/system.conf gained new settings to configure
4533 various default timeouts of units, as well as the default
4534 start limit interval and burst. These may still be overridden
4537 * PID1 will now export on the bus profile data of the security
4538 policy upload process (such as the SELinux policy upload to
4541 * journald: when forwarding logs to the console, include
4542 timestamps (following the setting in
4543 /sys/module/printk/parameters/time).
4545 * OnCalendar= in timer units now understands the special
4546 strings "yearly" and "annually". (Both are equivalent)
4548 * The accuracy of timer units is now configurable with the new
4549 AccuracySec= setting. It defaults to 1min.
4551 * A new dependency type JoinsNamespaceOf= has been added that
4552 allows running two services within the same /tmp and network
4553 namespace, if PrivateNetwork= or PrivateTmp= are used.
4555 * A new command "cat" has been added to systemctl. It outputs
4556 the original unit file of a unit, and concatenates the
4557 contents of additional "drop-in" unit file snippets, so that
4558 the full configuration is shown.
4560 * systemctl now supports globbing on the various "list-xyz"
4561 commands, like "list-units" or "list-sockets", as well as on
4562 those commands which take multiple unit names.
4564 * journalctl's --unit= switch gained support for globbing.
4566 * All systemd daemons now make use of the watchdog logic so
4567 that systemd automatically notices when they hang.
4569 * If the $container_ttys environment variable is set,
4570 getty-generator will automatically spawn a getty for each
4571 listed tty. This is useful for container managers to request
4572 login gettys to be spawned on as many ttys as needed.
4574 * %h, %s, %U specifier support is not available anymore when
4575 used in unit files for PID 1. This is because NSS calls are
4576 not safe from PID 1. They stay available for --user
4577 instances of systemd, and as special case for the root user.
4579 * loginctl gained a new "--no-legend" switch to turn off output
4582 * The "sd-login.h" API gained three new calls:
4583 sd_session_is_remote(), sd_session_get_remote_user(),
4584 sd_session_get_remote_host() to query information about
4587 * The udev hardware database now also carries vendor/product
4588 information of SDIO devices.
4590 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
4591 determine whether watchdog notifications are requested by
4594 * Socket-activated per-connection services now include a
4595 short description of the connection parameters in the
4598 * tmpfiles gained a new "--boot" option. When this is not used,
4599 only lines where the command character is not suffixed with
4600 "!" are executed. When this option is specified, those
4601 options are executed too. This partitions tmpfiles
4602 directives into those that can be safely executed at any
4603 time, and those which should be run only at boot (for
4604 example, a line that creates /run/nologin).
4606 * A new API "sd-resolve.h" has been added which provides a simple
4607 asynchronous wrapper around glibc NSS host name resolution
4608 calls, such as getaddrinfo(). In contrast to glibc's
4609 getaddrinfo_a(), it does not use signals. In contrast to most
4610 other asynchronous name resolution libraries, this one does
4611 not reimplement DNS, but reuses NSS, so that alternate
4612 host name resolution systems continue to work, such as mDNS,
4613 LDAP, etc. This API is based on libasyncns, but it has been
4614 cleaned up for inclusion in systemd.
4616 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
4617 "sd-daemon.h" are no longer found in individual libraries
4618 libsystemd-journal.so, libsystemd-login.so,
4619 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
4620 merged them into a single library, libsystemd.so, which
4621 provides all symbols. The reason for this is cyclic
4622 dependencies, as these libraries tend to use each other's
4623 symbols. So far, we have managed to workaround that by linking
4624 a copy of a good part of our code into each of these
4625 libraries again and again, which, however, makes certain
4626 things hard to do, like sharing static variables. Also, it
4627 substantially increases footprint. With this change, there
4628 is only one library for the basic APIs systemd
4629 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
4630 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
4631 library as well, however are subject to the --enable-kdbus
4632 switch (see below). Note that "sd-dhcp-client.h" is not part
4633 of this library (this is because it only consumes, never
4634 provides, services of/to other APIs). To make the transition
4635 easy from the separate libraries to the unified one, we
4636 provide the --enable-compat-libs compile-time switch which
4637 will generate stub libraries that are compatible with the
4638 old ones but redirect all calls to the new one.
4640 * All of the kdbus logic and the new APIs "sd-bus.h",
4641 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
4642 and "sd-utf8.h" are compile-time optional via the
4643 "--enable-kdbus" switch, and they are not compiled in by
4644 default. To make use of kdbus, you have to explicitly enable
4645 the switch. Note however, that neither the kernel nor the
4646 userspace API for all of this is considered stable yet. We
4647 want to maintain the freedom to still change the APIs for
4648 now. By specifying this build-time switch, you acknowledge
4649 that you are aware of the instability of the current
4652 * Also, note that while kdbus is pretty much complete,
4653 it lacks one thing: proper policy support. This means you
4654 can build a fully working system with all features; however,
4655 it will be highly insecure. Policy support will be added in
4656 one of the next releases, at the same time that we will
4657 declare the APIs stable.
4659 * When the kernel command line argument "kdbus" is specified,
4660 systemd will automatically load the kdbus.ko kernel module. At
4661 this stage of development, it is only useful for testing kdbus
4662 and should not be used in production. Note: if "--enable-kdbus"
4663 is specified, and the kdbus.ko kernel module is available, and
4664 "kdbus" is added to the kernel command line, the entire system
4665 runs with kdbus instead of dbus-daemon, with the above mentioned
4666 problem of missing the system policy enforcement. Also a future
4667 version of kdbus.ko or a newer systemd will not be compatible with
4668 each other, and will unlikely be able to boot the machine if only
4669 one of them is updated.
4671 * systemctl gained a new "import-environment" command which
4672 uploads the caller's environment (or parts thereof) into the
4673 service manager so that it is inherited by services started
4674 by the manager. This is useful to upload variables like
4675 $DISPLAY into the user service manager.
4677 * A new PrivateDevices= switch has been added to service units
4678 which allows running a service with a namespaced /dev
4679 directory that does not contain any device nodes for
4680 physical devices. More specifically, it only includes devices
4681 such as /dev/null, /dev/urandom, and /dev/zero which are API
4684 * logind has been extended to support behaviour like VT
4685 switching on seats that do not support a VT. This makes
4686 multi-session available on seats that are not the first seat
4687 (seat0), and on systems where kernel support for VTs has
4688 been disabled at compile-time.
4690 * If a process holds a delay lock for system sleep or shutdown
4691 and fails to release it in time, we will now log its
4692 identity. This makes it easier to identify processes that
4693 cause slow suspends or power-offs.
4695 * When parsing /etc/crypttab, support for a new key-slot=
4696 option as supported by Debian is added. It allows indicating
4697 which LUKS slot to use on disk, speeding up key loading.
4699 * The sd_journald_sendv() API call has been checked and
4700 officially declared to be async-signal-safe so that it may
4701 be invoked from signal handlers for logging purposes.
4703 * Boot-time status output is now enabled automatically after a
4704 short timeout if boot does not progress, in order to give
4705 the user an indication what she or he is waiting for.
4707 * The boot-time output has been improved to show how much time
4708 remains until jobs expire.
4710 * The KillMode= switch in service units gained a new possible
4711 value "mixed". If set, and the unit is shut down, then the
4712 initial SIGTERM signal is sent only to the main daemon
4713 process, while the following SIGKILL signal is sent to
4714 all remaining processes of the service.
4716 * When a scope unit is registered, a new property "Controller"
4717 may be set. If set to a valid bus name, systemd will send a
4718 RequestStop() signal to this name when it would like to shut
4719 down the scope. This may be used to hook manager logic into
4720 the shutdown logic of scope units. Also, scope units may now
4721 be put in a special "abandoned" state, in which case the
4722 manager process which created them takes no further
4723 responsibilities for it.
4725 * When reading unit files, systemd will now verify
4726 the access mode of these files, and warn about certain
4727 suspicious combinations. This has been added to make it
4728 easier to track down packaging bugs where unit files are
4729 marked executable or world-writable.
4731 * systemd-nspawn gained a new "--setenv=" switch to set
4732 container-wide environment variables. The similar option in
4733 systemd-activate was renamed from "--environment=" to
4734 "--setenv=" for consistency.
4736 * systemd-nspawn has been updated to create a new kdbus domain
4737 for each container that is invoked, thus allowing each
4738 container to have its own set of system and user buses,
4739 independent of the host.
4741 * systemd-nspawn gained a new --drop-capability= switch to run
4742 the container with less capabilities than the default. Both
4743 --drop-capability= and --capability= now take the special
4744 string "all" for dropping or keeping all capabilities.
4746 * systemd-nspawn gained new switches for executing containers
4747 with specific SELinux labels set.
4749 * systemd-nspawn gained a new --quiet switch to not generate
4750 any additional output but the container's own console
4753 * systemd-nspawn gained a new --share-system switch to run a
4754 container without PID namespacing enabled.
4756 * systemd-nspawn gained a new --register= switch to control
4757 whether the container is registered with systemd-machined or
4758 not. This is useful for containers that do not run full
4759 OS images, but only specific apps.
4761 * systemd-nspawn gained a new --keep-unit which may be used
4762 when invoked as the only program from a service unit, and
4763 results in registration of the unit service itself in
4764 systemd-machined, instead of a newly opened scope unit.
4766 * systemd-nspawn gained a new --network-interface= switch for
4767 moving arbitrary interfaces to the container. The new
4768 --network-veth switch creates a virtual Ethernet connection
4769 between host and container. The new --network-bridge=
4770 switch then allows assigning the host side of this virtual
4771 Ethernet connection to a bridge device.
4773 * systemd-nspawn gained a new --personality= switch for
4774 setting the kernel personality for the container. This is
4775 useful when running a 32-bit container on a 64-bit host. A
4776 similar option Personality= is now also available for service
4779 * logind will now also track a "Desktop" identifier for each
4780 session which encodes the desktop environment of it. This is
4781 useful for desktop environments that want to identify
4782 multiple running sessions of itself easily.
4784 * A new SELinuxContext= setting for service units has been
4785 added that allows setting a specific SELinux execution
4786 context for a service.
4788 * Most systemd client tools will now honour $SYSTEMD_LESS for
4789 settings of the "less" pager. By default, these tools will
4790 override $LESS to allow certain operations to work, such as
4791 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
4792 influence this logic.
4794 * systemd's "seccomp" hook-up has been changed to make use of
4795 the libseccomp library instead of using its own
4796 implementation. This has benefits for portability among
4799 * For usage together with SystemCallFilter=, a new
4800 SystemCallErrorNumber= setting has been introduced that
4801 allows configuration of a system error number to be returned
4802 on filtered system calls, instead of immediately killing the
4803 process. Also, SystemCallArchitectures= has been added to
4804 limit access to system calls of a particular architecture
4805 (in order to turn off support for unused secondary
4806 architectures). There is also a global
4807 SystemCallArchitectures= setting in system.conf now to turn
4808 off support for non-native system calls system-wide.
4810 * systemd requires a kernel with a working name_to_handle_at(),
4811 please see the kernel config requirements in the README file.
4813 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
4814 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
4815 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
4816 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
4817 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
4818 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
4819 Elia Pinto, Florian Weimer, George McCollister, Goffredo
4820 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
4821 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
4822 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
4823 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
4824 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
4825 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
4826 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
4827 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
4828 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
4829 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
4830 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
4831 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
4832 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
4833 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
4834 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
4835 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
4836 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
4838 — Berlin, 2014-02-20
4842 * logind has gained support for facilitating privileged input
4843 and drm device access for unprivileged clients. This work is
4844 useful to allow Wayland display servers (and similar
4845 programs, such as kmscon) to run under the user's ID and
4846 access input and drm devices which are normally
4847 protected. When this is used (and the kernel is new enough)
4848 logind will "mute" IO on the file descriptors passed to
4849 Wayland as long as it is in the background and "unmute" it
4850 if it returns into the foreground. This allows secure
4851 session switching without allowing background sessions to
4852 eavesdrop on input and display data. This also introduces
4853 session switching support if VT support is turned off in the
4854 kernel, and on seats that are not seat0.
4856 * A new kernel command line option luks.options= is understood
4857 now which allows specifying LUKS options for usage for LUKS
4858 encrypted partitions specified with luks.uuid=.
4860 * tmpfiles.d(5) snippets may now use specifier expansion in
4861 path names. More specifically %m, %b, %H, %v, are now
4862 replaced by the local machine id, boot id, hostname, and
4863 kernel version number.
4865 * A new tmpfiles.d(5) command "m" has been introduced which
4866 may be used to change the owner/group/access mode of a file
4867 or directory if it exists, but do nothing if it does not.
4869 * This release removes high-level support for the
4870 MemorySoftLimit= cgroup setting. The underlying kernel
4871 cgroup attribute memory.soft_limit= is currently badly
4872 designed and likely to be removed from the kernel API in its
4873 current form, hence we should not expose it for now.
4875 * The memory.use_hierarchy cgroup attribute is now enabled for
4876 all cgroups systemd creates in the memory cgroup
4877 hierarchy. This option is likely to be come the built-in
4878 default in the kernel anyway, and the non-hierarchical mode
4879 never made much sense in the intrinsically hierarchical
4882 * A new field _SYSTEMD_SLICE= is logged along with all journal
4883 messages containing the slice a message was generated
4884 from. This is useful to allow easy per-customer filtering of
4885 logs among other things.
4887 * systemd-journald will no longer adjust the group of journal
4888 files it creates to the "systemd-journal" group. Instead we
4889 rely on the journal directory to be owned by the
4890 "systemd-journal" group, and its setgid bit set, so that the
4891 kernel file system layer will automatically enforce that
4892 journal files inherit this group assignment. The reason for
4893 this change is that we cannot allow NSS look-ups from
4894 journald which would be necessary to resolve
4895 "systemd-journal" to a numeric GID, because this might
4896 create deadlocks if NSS involves synchronous queries to
4897 other daemons (such as nscd, or sssd) which in turn are
4898 logging clients of journald and might block on it, which
4899 would then dead lock. A tmpfiles.d(5) snippet included in
4900 systemd will make sure the setgid bit and group are
4901 properly set on the journal directory if it exists on every
4902 boot. However, we recommend adjusting it manually after
4903 upgrades too (or from RPM scriptlets), so that the change is
4904 not delayed until next reboot.
4906 * Backlight and random seed files in /var/lib/ have moved into
4907 the /var/lib/systemd/ directory, in order to centralize all
4908 systemd generated files in one directory.
4910 * Boot time performance measurements (as displayed by
4911 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
4912 performance information if that's available to determine how
4913 much time BIOS and boot loader initialization required. With
4914 a sufficiently new BIOS you hence no longer need to boot
4915 with Gummiboot to get access to such information.
4917 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
4918 Cristian Rodríguez, Dave Reisner, David Herrmann, David
4919 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
4920 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
4921 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
4922 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
4923 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
4925 — Berlin, 2013-10-02
4929 * The Restart= option for services now understands a new
4930 on-watchdog setting, which will restart the service
4931 automatically if the service stops sending out watchdog keep
4932 alive messages (as configured with WatchdogSec=).
4934 * The getty generator (which is responsible for bringing up a
4935 getty on configured serial consoles) will no longer only
4936 start a getty on the primary kernel console but on all
4937 others, too. This makes the order in which console= is
4938 specified on the kernel command line less important.
4940 * libsystemd-logind gained a new sd_session_get_vt() call to
4941 retrieve the VT number of a session.
4943 * If the option "tries=0" is set for an entry of /etc/crypttab
4944 its passphrase is queried indefinitely instead of any
4945 maximum number of tries.
4947 * If a service with a configure PID file terminates its PID
4948 file will now be removed automatically if it still exists
4949 afterwards. This should put an end to stale PID files.
4951 * systemd-run will now also take relative binary path names
4952 for execution and no longer insists on absolute paths.
4954 * InaccessibleDirectories= and ReadOnlyDirectories= now take
4955 paths that are optionally prefixed with "-" to indicate that
4956 it should not be considered a failure if they do not exist.
4958 * journalctl -o (and similar commands) now understands a new
4959 output mode "short-precise", it is similar to "short" but
4960 shows timestamps with usec accuracy.
4962 * The option "discard" (as known from Debian) is now
4963 synonymous to "allow-discards" in /etc/crypttab. In fact,
4964 "discard" is preferred now (since it is easier to remember
4967 * Some licensing clean-ups were made, so that more code is now
4968 LGPL-2.1 licensed than before.
4970 * A minimal tool to save/restore the display backlight
4971 brightness across reboots has been added. It will store the
4972 backlight setting as late as possible at shutdown, and
4973 restore it as early as possible during reboot.
4975 * A logic to automatically discover and enable home and swap
4976 partitions on GPT disks has been added. With this in place
4977 /etc/fstab becomes optional for many setups as systemd can
4978 discover certain partitions located on the root disk
4979 automatically. Home partitions are recognized under their
4980 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
4981 partitions are recognized under their GPT type ID
4982 0657fd6da4ab43c484e50933c84b4f4f.
4984 * systemd will no longer pass any environment from the kernel
4985 or initrd to system services. If you want to set an
4986 environment for all services, do so via the kernel command
4987 line systemd.setenv= assignment.
4989 * The systemd-sysctl tool no longer natively reads the file
4990 /etc/sysctl.conf. If desired, the file should be symlinked
4991 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
4992 legacy support by a symlink rather than built-in code, it
4993 also makes the otherwise hidden order of application of the
4994 different files visible. (Note that this partly reverts to a
4995 pre-198 application order of sysctl knobs!)
4997 * The "systemctl set-log-level" and "systemctl dump" commands
4998 have been moved to systemd-analyze.
5000 * systemd-run learned the new --remain-after-exit switch,
5001 which causes the scope unit not to be cleaned up
5002 automatically after the process terminated.
5004 * tmpfiles learned a new --exclude-prefix= switch to exclude
5005 certain paths from operation.
5007 * journald will now automatically flush all messages to disk
5008 as soon as a message at the log level CRIT, ALERT or EMERG
5011 Contributions from: Andrew Cook, Brandon Philips, Christian
5012 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
5013 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
5014 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
5015 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
5016 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
5017 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
5018 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
5019 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
5020 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
5021 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
5022 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
5023 William Giokas, Zbigniew Jędrzejewski-Szmek
5025 — Berlin, 2013-09-13
5029 * The documentation has been updated to cover the various new
5030 concepts introduced with 205.
5032 * Unit files now understand the new %v specifier which
5033 resolves to the kernel version string as returned by "uname
5036 * systemctl now supports filtering the unit list output by
5037 load state, active state and sub state, using the new
5040 * "systemctl status" will now show the results of the
5041 condition checks (like ConditionPathExists= and similar) of
5042 the last start attempts of the unit. They are also logged to
5045 * "journalctl -b" may now be used to look for boot output of a
5046 specific boot. Try "journalctl -b -1" for the previous boot,
5047 but the syntax is substantially more powerful.
5049 * "journalctl --show-cursor" has been added which prints the
5050 cursor string the last shown log line. This may then be used
5051 with the new "journalctl --after-cursor=" switch to continue
5052 browsing logs from that point on.
5054 * "journalctl --force" may now be used to force regeneration
5057 * Creation of "dead" device nodes has been moved from udev
5058 into kmod and tmpfiles. Previously, udev would read the kmod
5059 databases to pre-generate dead device nodes based on meta
5060 information contained in kernel modules, so that these would
5061 be auto-loaded on access rather then at boot. As this
5062 does not really have much to do with the exposing actual
5063 kernel devices to userspace this has always been slightly
5064 alien in the udev codebase. Following the new scheme kmod
5065 will now generate a runtime snippet for tmpfiles from the
5066 module meta information and it now is tmpfiles' job to the
5067 create the nodes. This also allows overriding access and
5068 other parameters for the nodes using the usual tmpfiles
5069 facilities. As side effect this allows us to remove the
5070 CAP_SYS_MKNOD capability bit from udevd entirely.
5072 * logind's device ACLs may now be applied to these "dead"
5073 devices nodes too, thus finally allowing managed access to
5074 devices such as /dev/snd/sequencer without loading the
5075 backing module right-away.
5077 * A new RPM macro has been added that may be used to apply
5078 tmpfiles configuration during package installation.
5080 * systemd-detect-virt and ConditionVirtualization= now can
5081 detect User-Mode-Linux machines (UML).
5083 * journald will now implicitly log the effective capabilities
5084 set of processes in the message metadata.
5086 * systemd-cryptsetup has gained support for TrueCrypt volumes.
5088 * The initrd interface has been simplified (more specifically,
5089 support for passing performance data via environment
5090 variables and fsck results via files in /run has been
5091 removed). These features were non-essential, and are
5092 nowadays available in a much nicer way by having systemd in
5093 the initrd serialize its state and have the hosts systemd
5094 deserialize it again.
5096 * The udev "keymap" data files and tools to apply keyboard
5097 specific mappings of scan to key codes, and force-release
5098 scan code lists have been entirely replaced by a udev
5099 "keyboard" builtin and a hwdb data file.
5101 * systemd will now honour the kernel's "quiet" command line
5102 argument also during late shutdown, resulting in a
5103 completely silent shutdown when used.
5105 * There's now an option to control the SO_REUSEPORT socket
5106 option in .socket units.
5108 * Instance units will now automatically get a per-template
5109 subslice of system.slice unless something else is explicitly
5110 configured. For example, instances of sshd@.service will now
5111 implicitly be placed in system-sshd.slice rather than
5112 system.slice as before.
5114 * Test coverage support may now be enabled at build time.
5116 Contributions from: Dave Reisner, Frederic Crozat, Harald
5117 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
5118 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
5119 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
5120 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
5121 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
5122 Giokas, Zbigniew Jędrzejewski-Szmek
5124 — Berlin, 2013-07-23
5128 * Two new unit types have been introduced:
5130 Scope units are very similar to service units, however, are
5131 created out of pre-existing processes — instead of PID 1
5132 forking off the processes. By using scope units it is
5133 possible for system services and applications to group their
5134 own child processes (worker processes) in a powerful way
5135 which then maybe used to organize them, or kill them
5136 together, or apply resource limits on them.
5138 Slice units may be used to partition system resources in an
5139 hierarchical fashion and then assign other units to them. By
5140 default there are now three slices: system.slice (for all
5141 system services), user.slice (for all user sessions),
5142 machine.slice (for VMs and containers).
5144 Slices and scopes have been introduced primarily in
5145 context of the work to move cgroup handling to a
5146 single-writer scheme, where only PID 1
5147 creates/removes/manages cgroups.
5149 * There's a new concept of "transient" units. In contrast to
5150 normal units these units are created via an API at runtime,
5151 not from configuration from disk. More specifically this
5152 means it is now possible to run arbitrary programs as
5153 independent services, with all execution parameters passed
5154 in via bus APIs rather than read from disk. Transient units
5155 make systemd substantially more dynamic then it ever was,
5156 and useful as a general batch manager.
5158 * logind has been updated to make use of scope and slice units
5159 for managing user sessions. As a user logs in he will get
5160 his own private slice unit, to which all sessions are added
5161 as scope units. We also added support for automatically
5162 adding an instance of user@.service for the user into the
5163 slice. Effectively logind will no longer create cgroup
5164 hierarchies on its own now, it will defer entirely to PID 1
5165 for this by means of scope, service and slice units. Since
5166 user sessions this way become entities managed by PID 1
5167 the output of "systemctl" is now a lot more comprehensive.
5169 * A new mini-daemon "systemd-machined" has been added which
5170 may be used by virtualization managers to register local
5171 VMs/containers. nspawn has been updated accordingly, and
5172 libvirt will be updated shortly. machined will collect a bit
5173 of meta information about the VMs/containers, and assign
5174 them their own scope unit (see above). The collected
5175 meta-data is then made available via the "machinectl" tool,
5176 and exposed in "ps" and similar tools. machined/machinectl
5177 is compile-time optional.
5179 * As discussed earlier, the low-level cgroup configuration
5180 options ControlGroup=, ControlGroupModify=,
5181 ControlGroupPersistent=, ControlGroupAttribute= have been
5182 removed. Please use high-level attribute settings instead as
5183 well as slice units.
5185 * A new bus call SetUnitProperties() has been added to alter
5186 various runtime parameters of a unit. This is primarily
5187 useful to alter cgroup parameters dynamically in a nice way,
5188 but will be extended later on to make more properties
5189 modifiable at runtime. systemctl gained a new set-properties
5190 command that wraps this call.
5192 * A new tool "systemd-run" has been added which can be used to
5193 run arbitrary command lines as transient services or scopes,
5194 while configuring a number of settings via the command
5195 line. This tool is currently very basic, however already
5196 very useful. We plan to extend this tool to even allow
5197 queuing of execution jobs with time triggers from the
5198 command line, similar in fashion to "at".
5200 * nspawn will now inform the user explicitly that kernels with
5201 audit enabled break containers, and suggest the user to turn
5204 * Support for detecting the IMA and AppArmor security
5205 frameworks with ConditionSecurity= has been added.
5207 * journalctl gained a new "-k" switch for showing only kernel
5208 messages, mimicking dmesg output; in addition to "--user"
5209 and "--system" switches for showing only user's own logs
5212 * systemd-delta can now show information about drop-in
5213 snippets extending unit files.
5215 * libsystemd-bus has been substantially updated but is still
5216 not available as public API.
5218 * systemd will now look for the "debug" argument on the kernel
5219 command line and enable debug logging, similar to what
5220 "systemd.log_level=debug" already did before.
5222 * "systemctl set-default", "systemctl get-default" has been
5223 added to configure the default.target symlink, which
5224 controls what to boot into by default.
5226 * "systemctl set-log-level" has been added as a convenient
5227 way to raise and lower systemd logging threshold.
5229 * "systemd-analyze plot" will now show the time the various
5230 generators needed for execution, as well as information
5231 about the unit file loading.
5233 * libsystemd-journal gained a new sd_journal_open_files() call
5234 for opening specific journal files. journactl also gained a
5235 new switch to expose this new functionality. Previously we
5236 only supported opening all files from a directory, or all
5237 files from the system, as opening individual files only is
5238 racy due to journal file rotation.
5240 * systemd gained the new DefaultEnvironment= setting in
5241 /etc/systemd/system.conf to set environment variables for
5244 * If a privileged process logs a journal message with the
5245 OBJECT_PID= field set, then journald will automatically
5246 augment this with additional OBJECT_UID=, OBJECT_GID=,
5247 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
5248 system services want to log events about specific client
5249 processes. journactl/systemctl has been updated to make use
5250 of this information if all log messages regarding a specific
5253 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
5254 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
5255 Reisner, David Coppa, David King, David Strauss, Eelco
5256 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
5257 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
5258 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
5259 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
5260 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
5261 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
5262 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
5263 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
5264 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
5265 Łukasz Stelmach, 장동준
5269 * The Python bindings gained some minimal support for the APIs
5270 exposed by libsystemd-logind.
5272 * ConditionSecurity= gained support for detecting SMACK. Since
5273 this condition already supports SELinux and AppArmor we only
5274 miss IMA for this. Patches welcome!
5276 Contributions from: Karol Lewandowski, Lennart Poettering,
5277 Zbigniew Jędrzejewski-Szmek
5281 * systemd-nspawn will now create /etc/resolv.conf if
5282 necessary, before bind-mounting the host's file onto it.
5284 * systemd-nspawn will now store meta information about a
5285 container on the container's cgroup as extended attribute
5286 fields, including the root directory.
5288 * The cgroup hierarchy has been reworked in many ways. All
5289 objects any of the components systemd creates in the cgroup
5290 tree are now suffixed. More specifically, user sessions are
5291 now placed in cgroups suffixed with ".session", users in
5292 cgroups suffixed with ".user", and nspawn containers in
5293 cgroups suffixed with ".nspawn". Furthermore, all cgroup
5294 names are now escaped in a simple scheme to avoid collision
5295 of userspace object names with kernel filenames. This work
5296 is preparation for making these objects relocatable in the
5297 cgroup tree, in order to allow easy resource partitioning of
5298 these objects without causing naming conflicts.
5300 * systemctl list-dependencies gained the new switches
5301 --plain, --reverse, --after and --before.
5303 * systemd-inhibit now shows the process name of processes that
5304 have taken an inhibitor lock.
5306 * nss-myhostname will now also resolve "localhost"
5307 implicitly. This makes /etc/hosts an optional file and
5308 nicely handles that on IPv6 ::1 maps to both "localhost" and
5311 * libsystemd-logind.so gained a new call
5312 sd_get_machine_names() to enumerate running containers and
5313 VMs (currently only supported by very new libvirt and
5314 nspawn). sd_login_monitor can now be used to watch
5315 VMs/containers coming and going.
5317 * .include is not allowed recursively anymore, and only in
5318 unit files. Usually it is better to use drop-in snippets in
5319 .d/*.conf anyway, as introduced with systemd 198.
5321 * systemd-analyze gained a new "critical-chain" command that
5322 determines the slowest chain of units run during system
5323 boot-up. It is very useful for tracking down where
5324 optimizing boot time is the most beneficial.
5326 * systemd will no longer allow manipulating service paths in
5327 the name=systemd:/system cgroup tree using ControlGroup= in
5328 units. (But is still fine with it in all other dirs.)
5330 * There's a new systemd-nspawn@.service service file that may
5331 be used to easily run nspawn containers as system
5332 services. With the container's root directory in
5333 /var/lib/container/foobar it is now sufficient to run
5334 "systemctl start systemd-nspawn@foobar.service" to boot it.
5336 * systemd-cgls gained a new parameter "--machine" to list only
5337 the processes within a certain container.
5339 * ConditionSecurity= now can check for "apparmor". We still
5340 are lacking checks for SMACK and IMA for this condition
5341 check though. Patches welcome!
5343 * A new configuration file /etc/systemd/sleep.conf has been
5344 added that may be used to configure which kernel operation
5345 systemd is supposed to execute when "suspend", "hibernate"
5346 or "hybrid-sleep" is requested. This makes the new kernel
5347 "freeze" state accessible to the user.
5349 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
5350 the passed argument if applicable.
5352 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
5353 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
5354 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
5355 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
5356 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
5357 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
5358 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
5363 * The output of 'systemctl list-jobs' got some polishing. The
5364 '--type=' argument may now be passed more than once. A new
5365 command 'systemctl list-sockets' has been added which shows
5366 a list of kernel sockets systemd is listening on with the
5367 socket units they belong to, plus the units these socket
5370 * The experimental libsystemd-bus library got substantial
5371 updates to work in conjunction with the (also experimental)
5372 kdbus kernel project. It works well enough to exchange
5373 messages with some sophistication. Note that kdbus is not
5374 ready yet, and the library is mostly an elaborate test case
5375 for now, and not installable.
5377 * systemd gained a new unit 'systemd-static-nodes.service'
5378 that generates static device nodes earlier during boot, and
5379 can run in conjunction with udev.
5381 * libsystemd-login gained a new call sd_pid_get_user_unit()
5382 to retrieve the user systemd unit a process is running
5383 in. This is useful for systems where systemd is used as
5386 * systemd-nspawn now places all containers in the new /machine
5387 top-level cgroup directory in the name=systemd
5388 hierarchy. libvirt will soon do the same, so that we get a
5389 uniform separation of /system, /user and /machine for system
5390 services, user processes and containers/virtual
5391 machines. This new cgroup hierarchy is also useful to stick
5392 stable names to specific container instances, which can be
5393 recognized later this way (this name may be controlled
5394 via systemd-nspawn's new -M switch). libsystemd-login also
5395 gained a new call sd_pid_get_machine_name() to retrieve the
5396 name of the container/VM a specific process belongs to.
5398 * bootchart can now store its data in the journal.
5400 * libsystemd-journal gained a new call
5401 sd_journal_add_conjunction() for AND expressions to the
5402 matching logic. This can be used to express more complex
5403 logical expressions.
5405 * journactl can now take multiple --unit= and --user-unit=
5408 * The cryptsetup logic now understands the "luks.key=" kernel
5409 command line switch for specifying a file to read the
5410 decryption key from. Also, if a configured key file is not
5411 found the tool will now automatically fall back to prompting
5414 * Python systemd.journal module was updated to wrap recently
5415 added functions from libsystemd-journal. The interface was
5416 changed to bring the low level interface in s.j._Reader
5417 closer to the C API, and the high level interface in
5418 s.j.Reader was updated to wrap and convert all data about
5421 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
5422 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
5423 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
5424 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
5425 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
5426 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5430 * journalctl --update-catalog now understands a new --root=
5431 option to operate on catalogs found in a different root
5434 * During shutdown after systemd has terminated all running
5435 services a final killing loop kills all remaining left-over
5436 processes. We will now print the name of these processes
5437 when we send SIGKILL to them, since this usually indicates a
5440 * If /etc/crypttab refers to password files stored on
5441 configured mount points automatic dependencies will now be
5442 generated to ensure the specific mount is established first
5443 before the key file is attempted to be read.
5445 * 'systemctl status' will now show information about the
5446 network sockets a socket unit is listening on.
5448 * 'systemctl status' will also shown information about any
5449 drop-in configuration file for units. (Drop-In configuration
5450 files in this context are files such as
5451 /etc/systemd/systemd/foobar.service.d/*.conf)
5453 * systemd-cgtop now optionally shows summed up CPU times of
5454 cgroups. Press '%' while running cgtop to switch between
5455 percentage and absolute mode. This is useful to determine
5456 which cgroups use up the most CPU time over the entire
5457 runtime of the system. systemd-cgtop has also been updated
5458 to be 'pipeable' for processing with further shell tools.
5460 * 'hostnamectl set-hostname' will now allow setting of FQDN
5463 * The formatting and parsing of time span values has been
5464 changed. The parser now understands fractional expressions
5465 such as "5.5h". The formatter will now output fractional
5466 expressions for all time spans under 1min, i.e. "5.123456s"
5467 rather than "5s 123ms 456us". For time spans under 1s
5468 millisecond values are shown, for those under 1ms
5469 microsecond values are shown. This should greatly improve
5470 all time-related output of systemd.
5472 * libsystemd-login and libsystemd-journal gained new
5473 functions for querying the poll() events mask and poll()
5474 timeout value for integration into arbitrary event
5477 * localectl gained the ability to list available X11 keymaps
5478 (models, layouts, variants, options).
5480 * 'systemd-analyze dot' gained the ability to filter for
5481 specific units via shell-style globs, to create smaller,
5482 more useful graphs. I.e. it is now possible to create simple
5483 graphs of all the dependencies between only target units, or
5484 of all units that Avahi has dependencies with.
5486 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
5487 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
5488 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
5489 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
5490 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
5491 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
5492 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
5496 * The boot-time readahead implementation for rotating media
5497 will now read the read-ahead data in multiple passes which
5498 consist of all read requests made in equidistant time
5499 intervals. This means instead of strictly reading read-ahead
5500 data in its physical order on disk we now try to find a
5501 middle ground between physical and access time order.
5503 * /etc/os-release files gained a new BUILD_ID= field for usage
5504 on operating systems that provide continuous builds of OS
5507 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
5508 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
5509 William Douglas, Zbigniew Jędrzejewski-Szmek
5513 * systemd-python gained an API exposing libsystemd-daemon.
5515 * The SMACK setup logic gained support for uploading CIPSO
5518 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
5519 ReadOnlyDirectories= and InaccessibleDirectories= has
5520 changed. The private /tmp and /var/tmp directories are now
5521 shared by all processes of a service (which means
5522 ExecStartPre= may now leave data in /tmp that ExecStart= of
5523 the same service can still access). When a service is
5524 stopped its temporary directories are immediately deleted
5525 (normal clean-up with tmpfiles is still done in addition to
5528 * By default, systemd will now set a couple of sysctl
5529 variables in the kernel: the safe sysrq options are turned
5530 on, IP route verification is turned on, and source routing
5531 disabled. The recently added hardlink and softlink
5532 protection of the kernel is turned on. These settings should
5533 be reasonably safe, and good defaults for all new systems.
5535 * The predictable network naming logic may now be turned off
5536 with a new kernel command line switch: net.ifnames=0.
5538 * A new libsystemd-bus module has been added that implements a
5539 pretty complete D-Bus client library. For details see:
5541 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
5543 * journald will now explicitly flush the journal files to disk
5544 at the latest 5min after each write. The file will then also
5545 be marked offline until the next write. This should increase
5546 reliability in case of a crash. The synchronization delay
5547 can be configured via SyncIntervalSec= in journald.conf.
5549 * There's a new remote-fs-setup.target unit that can be used
5550 to pull in specific services when at least one remote file
5551 system is to be mounted.
5553 * There are new targets timers.target and paths.target as
5554 canonical targets to pull user timer and path units in
5555 from. This complements sockets.target with a similar
5556 purpose for socket units.
5558 * libudev gained a new call udev_device_set_attribute_value()
5559 to set sysfs attributes of a device.
5561 * The udev daemon now sets the default number of worker
5562 processes executed in parallel based on the number of available
5563 CPUs instead of the amount of available RAM. This is supposed
5564 to provide a more reliable default and limit a too aggressive
5565 parallelism for setups with 1000s of devices connected.
5567 Contributions from: Auke Kok, Colin Walters, Cristian
5568 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
5569 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
5570 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
5571 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
5572 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
5573 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
5574 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
5575 Zbigniew Jędrzejewski-Szmek
5579 * Configuration of unit files may now be extended via drop-in
5580 files without having to edit/override the unit files
5581 themselves. More specifically, if the administrator wants to
5582 change one value for a service file foobar.service he can
5583 now do so by dropping in a configuration snippet into
5584 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
5585 will load all these snippets and apply them on top of the
5586 main unit configuration file, possibly extending or
5587 overriding its settings. Using these drop-in snippets is
5588 generally nicer than the two earlier options for changing
5589 unit files locally: copying the files from
5590 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
5591 them there; or creating a new file in /etc/systemd/system/
5592 that incorporates the original one via ".include". Drop-in
5593 snippets into these .d/ directories can be placed in any
5594 directory systemd looks for units in, and the usual
5595 overriding semantics between /usr/lib, /etc and /run apply
5598 * Most unit file settings which take lists of items can now be
5599 reset by assigning the empty string to them. For example,
5600 normally, settings such as Environment=FOO=BAR append a new
5601 environment variable assignment to the environment block,
5602 each time they are used. By assigning Environment= the empty
5603 string the environment block can be reset to empty. This is
5604 particularly useful with the .d/*.conf drop-in snippets
5605 mentioned above, since this adds the ability to reset list
5606 settings from vendor unit files via these drop-ins.
5608 * systemctl gained a new "list-dependencies" command for
5609 listing the dependencies of a unit recursively.
5611 * Inhibitors are now honored and listed by "systemctl
5612 suspend", "systemctl poweroff" (and similar) too, not only
5613 GNOME. These commands will also list active sessions by
5616 * Resource limits (as exposed by the various control group
5617 controllers) can now be controlled dynamically at runtime
5618 for all units. More specifically, you can now use a command
5619 like "systemctl set-cgroup-attr foobar.service cpu.shares
5620 2000" to alter the CPU shares a specific service gets. These
5621 settings are stored persistently on disk, and thus allow the
5622 administrator to easily adjust the resource usage of
5623 services with a few simple commands. This dynamic resource
5624 management logic is also available to other programs via the
5625 bus. Almost any kernel cgroup attribute and controller is
5628 * systemd-vconsole-setup will now copy all font settings to
5629 all allocated VTs, where it previously applied them only to
5632 * libsystemd-login gained the new sd_session_get_tty() API
5635 * This release drops support for a few legacy or
5636 distribution-specific LSB facility names when parsing init
5637 scripts: $x-display-manager, $mail-transfer-agent,
5638 $mail-transport-agent, $mail-transfer-agent, $smtp,
5639 $null. Also, the mail-transfer-agent.target unit backing
5640 this has been removed. Distributions which want to retain
5641 compatibility with this should carry the burden for
5642 supporting this themselves and patch support for these back
5643 in, if they really need to. Also, the facilities $syslog and
5644 $local_fs are now ignored, since systemd does not support
5645 early-boot LSB init scripts anymore, and these facilities
5646 are implied anyway for normal services. syslog.target has
5649 * There are new bus calls on PID1's Manager object for
5650 cancelling jobs, and removing snapshot units. Previously,
5651 both calls were only available on the Job and Snapshot
5654 * systemd-journal-gatewayd gained SSL support.
5656 * The various "environment" files, such as /etc/locale.conf
5657 now support continuation lines with a backslash ("\") as
5658 last character in the line, similarly in style (but different)
5659 to how this is supported in shells.
5661 * For normal user processes the _SYSTEMD_USER_UNIT= field is
5662 now implicitly appended to every log entry logged. systemctl
5663 has been updated to filter by this field when operating on a
5664 user systemd instance.
5666 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
5667 CAP_AUDIT_CONTROL capabilities to the capabilities set for
5668 the container. This makes it easier to boot unmodified
5669 Fedora systems in a container, which however still requires
5670 audit=0 to be passed on the kernel command line. Auditing in
5671 kernel and userspace is unfortunately still too broken in
5672 context of containers, hence we recommend compiling it out
5673 of the kernel or using audit=0. Hopefully this will be fixed
5674 one day for good in the kernel.
5676 * nspawn gained the new --bind= and --bind-ro= parameters to
5677 bind mount specific directories from the host into the
5680 * nspawn will now mount its own devpts file system instance
5681 into the container, in order not to leak pty devices from
5682 the host into the container.
5684 * systemd will now read the firmware boot time performance
5685 information from the EFI variables, if the used boot loader
5686 supports this, and takes it into account for boot performance
5687 analysis via "systemd-analyze". This is currently supported
5688 only in conjunction with Gummiboot, but could be supported
5689 by other boot loaders too. For details see:
5691 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
5693 * A new generator has been added that automatically mounts the
5694 EFI System Partition (ESP) to /boot, if that directory
5695 exists, is empty, and no other file system has been
5696 configured to be mounted there.
5698 * logind will now send out PrepareForSleep(false) out
5699 unconditionally, after coming back from suspend. This may be
5700 used by applications as asynchronous notification for
5701 system resume events.
5703 * "systemctl unlock-sessions" has been added, that allows
5704 unlocking the screens of all user sessions at once, similar
5705 to how "systemctl lock-sessions" already locked all users
5706 sessions. This is backed by a new D-Bus call UnlockSessions().
5708 * "loginctl seat-status" will now show the master device of a
5709 seat. (i.e. the device of a seat that needs to be around for
5710 the seat to be considered available, usually the graphics
5713 * tmpfiles gained a new "X" line type, that allows
5714 configuration of files and directories (with wildcards) that
5715 shall be excluded from automatic cleanup ("aging").
5717 * udev default rules set the device node permissions now only
5718 at "add" events, and do not change them any longer with a
5719 later "change" event.
5721 * The log messages for lid events and power/sleep keypresses
5722 now carry a message ID.
5724 * We now have a substantially larger unit test suite, but this
5725 continues to be work in progress.
5727 * udevadm hwdb gained a new --root= parameter to change the
5728 root directory to operate relative to.
5730 * logind will now issue a background sync() request to the kernel
5731 early at shutdown, so that dirty buffers are flushed to disk early
5732 instead of at the last moment, in order to optimize shutdown
5735 * A new bootctl tool has been added that is an interface for
5736 certain boot loader operations. This is currently a preview
5737 and is likely to be extended into a small mechanism daemon
5738 like timedated, localed, hostnamed, and can be used by
5739 graphical UIs to enumerate available boot options, and
5740 request boot into firmware operations.
5742 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
5743 the rest of the package. It also has been updated to work
5744 correctly in initrds.
5746 * Policykit previously has been runtime optional, and is now
5747 also compile time optional via a configure switch.
5749 * systemd-analyze has been reimplemented in C. Also "systemctl
5750 dot" has moved into systemd-analyze.
5752 * "systemctl status" with no further parameters will now print
5753 the status of all active or failed units.
5755 * Operations such as "systemctl start" can now be executed
5756 with a new mode "--irreversible" which may be used to queue
5757 operations that cannot accidentally be reversed by a later
5758 job queuing. This is by default used to make shutdown
5759 requests more robust.
5761 * The Python API of systemd now gained a new module for
5762 reading journal files.
5764 * A new tool kernel-install has been added that can install
5765 kernel images according to the Boot Loader Specification:
5767 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
5769 * Boot time console output has been improved to provide
5770 animated boot time output for hanging jobs.
5772 * A new tool systemd-activate has been added which can be used
5773 to test socket activation with, directly from the command
5774 line. This should make it much easier to test and debug
5775 socket activation in daemons.
5777 * journalctl gained a new "--reverse" (or -r) option to show
5778 journal output in reverse order (i.e. newest line first).
5780 * journalctl gained a new "--pager-end" (or -e) option to jump
5781 to immediately jump to the end of the journal in the
5782 pager. This is only supported in conjunction with "less".
5784 * journalctl gained a new "--user-unit=" option, that works
5785 similarly to "--unit=" but filters for user units rather than
5788 * A number of unit files to ease adoption of systemd in
5789 initrds has been added. This moves some minimal logic from
5790 the various initrd implementations into systemd proper.
5792 * The journal files are now owned by a new group
5793 "systemd-journal", which exists specifically to allow access
5794 to the journal, and nothing else. Previously, we used the
5795 "adm" group for that, which however possibly covers more
5796 than just journal/log file access. This new group is now
5797 already used by systemd-journal-gatewayd to ensure this
5798 daemon gets access to the journal files and as little else
5799 as possible. Note that "make install" will also set FS ACLs
5800 up for /var/log/journal to give "adm" and "wheel" read
5801 access to it, in addition to "systemd-journal" which owns
5802 the journal files. We recommend that packaging scripts also
5803 add read access to "adm" + "wheel" to /var/log/journal, and
5804 all existing/future journal files. To normal users and
5805 administrators little changes, however packagers need to
5806 ensure to create the "systemd-journal" system group at
5807 package installation time.
5809 * The systemd-journal-gatewayd now runs as unprivileged user
5810 systemd-journal-gateway:systemd-journal-gateway. Packaging
5811 scripts need to create these system user/group at
5814 * timedated now exposes a new boolean property CanNTP that
5815 indicates whether a local NTP service is available or not.
5817 * systemd-detect-virt will now also detect xen PVs
5819 * The pstore file system is now mounted by default, if it is
5822 * In addition to the SELinux and IMA policies we will now also
5823 load SMACK policies at early boot.
5825 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
5826 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
5827 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
5828 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
5829 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
5830 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
5831 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
5832 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
5833 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
5834 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
5835 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
5836 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
5837 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
5838 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
5842 * Timer units now support calendar time events in addition to
5843 monotonic time events. That means you can now trigger a unit
5844 based on a calendar time specification such as "Thu,Fri
5845 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
5846 or fifth day of any month of the year 2013, given that it is
5847 a thursday or friday. This brings timer event support
5848 considerably closer to cron's capabilities. For details on
5849 the supported calendar time specification language see
5852 * udev now supports a number of different naming policies for
5853 network interfaces for predictable names, and a combination
5854 of these policies is now the default. Please see this wiki
5855 document for details:
5857 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
5859 * Auke Kok's bootchart implementation has been added to the
5860 systemd tree. It is an optional component that can graph the
5861 boot in quite some detail. It is one of the best bootchart
5862 implementations around and minimal in its code and
5865 * nss-myhostname has been integrated into the systemd source
5866 tree. nss-myhostname guarantees that the local hostname
5867 always stays resolvable via NSS. It has been a weak
5868 requirement of systemd-hostnamed since a long time, and
5869 since its code is actually trivial we decided to just
5870 include it in systemd's source tree. It can be turned off
5871 with a configure switch.
5873 * The read-ahead logic is now capable of properly detecting
5874 whether a btrfs file system is on SSD or rotating media, in
5875 order to optimize the read-ahead scheme. Previously, it was
5876 only capable of detecting this on traditional file systems
5879 * In udev, additional device properties are now read from the
5880 IAB in addition to the OUI database. Also, Bluetooth company
5881 identities are attached to the devices as well.
5883 * In service files %U may be used as specifier that is
5884 replaced by the configured user name of the service.
5886 * nspawn may now be invoked without a controlling TTY. This
5887 makes it suitable for invocation as its own service. This
5888 may be used to set up a simple containerized server system
5889 using only core OS tools.
5891 * systemd and nspawn can now accept socket file descriptors
5892 when they are started for socket activation. This enables
5893 implementation of socket activated nspawn
5894 containers. i.e. think about autospawning an entire OS image
5895 when the first SSH or HTTP connection is received. We expect
5896 that similar functionality will also be added to libvirt-lxc
5899 * journalctl will now suppress ANSI color codes when
5900 presenting log data.
5902 * systemctl will no longer show control group information for
5903 a unit if the control group is empty anyway.
5905 * logind can now automatically suspend/hibernate/shutdown the
5908 * /etc/machine-info and hostnamed now also expose the chassis
5909 type of the system. This can be used to determine whether
5910 the local system is a laptop, desktop, handset or
5911 tablet. This information may either be configured by the
5912 user/vendor or is automatically determined from ACPI and DMI
5913 information if possible.
5915 * A number of PolicyKit actions are now bound together with
5916 "imply" rules. This should simplify creating UIs because
5917 many actions will now authenticate similar ones as well.
5919 * Unit files learnt a new condition ConditionACPower= which
5920 may be used to conditionalize a unit depending on whether an
5921 AC power source is connected or not, of whether the system
5922 is running on battery power.
5924 * systemctl gained a new "is-failed" verb that may be used in
5925 shell scripts and suchlike to check whether a specific unit
5926 is in the "failed" state.
5928 * The EnvironmentFile= setting in unit files now supports file
5929 globbing, and can hence be used to easily read a number of
5930 environment files at once.
5932 * systemd will no longer detect and recognize specific
5933 distributions. All distribution-specific #ifdeffery has been
5934 removed, systemd is now fully generic and
5935 distribution-agnostic. Effectively, not too much is lost as
5936 a lot of the code is still accessible via explicit configure
5937 switches. However, support for some distribution specific
5938 legacy configuration file formats has been dropped. We
5939 recommend distributions to simply adopt the configuration
5940 files everybody else uses now and convert the old
5941 configuration from packaging scripts. Most distributions
5942 already did that. If that's not possible or desirable,
5943 distributions are welcome to forward port the specific
5944 pieces of code locally from the git history.
5946 * When logging a message about a unit systemd will now always
5947 log the unit name in the message meta data.
5949 * localectl will now also discover system locale data that is
5950 not stored in locale archives, but directly unpacked.
5952 * logind will no longer unconditionally use framebuffer
5953 devices as seat masters, i.e. as devices that are required
5954 to be existing before a seat is considered preset. Instead,
5955 it will now look for all devices that are tagged as
5956 "seat-master" in udev. By default, framebuffer devices will
5957 be marked as such, but depending on local systems, other
5958 devices might be marked as well. This may be used to
5959 integrate graphics cards using closed source drivers (such
5960 as NVidia ones) more nicely into logind. Note however, that
5961 we recommend using the open source NVidia drivers instead,
5962 and no udev rules for the closed-source drivers will be
5963 shipped from us upstream.
5965 Contributions from: Adam Williamson, Alessandro Crismani, Auke
5966 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
5967 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
5968 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
5969 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
5970 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
5971 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
5972 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
5973 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
5974 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
5975 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
5976 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
5981 * udev gained support for loading additional device properties
5982 from an indexed database that is keyed by vendor/product IDs
5983 and similar device identifiers. For the beginning this
5984 "hwdb" is populated with data from the well-known PCI and
5985 USB database, but also includes PNP, ACPI and OID data. In
5986 the longer run this indexed database shall grow into
5987 becoming the one central database for non-essential
5988 userspace device metadata. Previously, data from the PCI/USB
5989 database was only attached to select devices, since the
5990 lookup was a relatively expensive operation due to O(n) time
5991 complexity (with n being the number of entries in the
5992 database). Since this is now O(1), we decided to add in this
5993 data for all devices where this is available, by
5994 default. Note that the indexed database needs to be rebuilt
5995 when new data files are installed. To achieve this you need
5996 to update your packaging scripts to invoke "udevadm hwdb
5997 --update" after installation of hwdb data files. For
5998 RPM-based distributions we introduced the new
5999 %udev_hwdb_update macro for this purpose.
6001 * The Journal gained support for the "Message Catalog", an
6002 indexed database to link up additional information with
6003 journal entries. For further details please check:
6005 https://www.freedesktop.org/wiki/Software/systemd/catalog
6007 The indexed message catalog database also needs to be
6008 rebuilt after installation of message catalog files. Use
6009 "journalctl --update-catalog" for this. For RPM-based
6010 distributions we introduced the %journal_catalog_update
6011 macro for this purpose.
6013 * The Python Journal bindings gained support for the standard
6014 Python logging framework.
6016 * The Journal API gained new functions for checking whether
6017 the underlying file system of a journal file is capable of
6018 properly reporting file change notifications, or whether
6019 applications that want to reflect journal changes "live"
6020 need to recheck journal files continuously in appropriate
6023 * It is now possible to set the "age" field for tmpfiles
6024 entries to 0, indicating that files matching this entry
6025 shall always be removed when the directories are cleaned up.
6027 * coredumpctl gained a new "gdb" verb which invokes gdb
6028 right-away on the selected coredump.
6030 * There's now support for "hybrid sleep" on kernels that
6031 support this, in addition to "suspend" and "hibernate". Use
6032 "systemctl hybrid-sleep" to make use of this.
6034 * logind's HandleSuspendKey= setting (and related settings)
6035 now gained support for a new "lock" setting to simply
6036 request the screen lock on all local sessions, instead of
6037 actually executing a suspend or hibernation.
6039 * systemd will now mount the EFI variables file system by
6042 * Socket units now gained support for configuration of the
6043 SMACK security label.
6045 * timedatectl will now output the time of the last and next
6046 daylight saving change.
6048 * We dropped support for various legacy and distro-specific
6049 concepts, such as insserv, early-boot SysV services
6050 (i.e. those for non-standard runlevels such as 'b' or 'S')
6051 or ArchLinux /etc/rc.conf support. We recommend the
6052 distributions who still need support this to either continue
6053 to maintain the necessary patches downstream, or find a
6054 different solution. (Talk to us if you have questions!)
6056 * Various systemd components will now bypass PolicyKit checks
6057 for root and otherwise handle properly if PolicyKit is not
6058 found to be around. This should fix most issues for
6059 PolicyKit-less systems. Quite frankly this should have been
6060 this way since day one. It is absolutely our intention to
6061 make systemd work fine on PolicyKit-less systems, and we
6062 consider it a bug if something does not work as it should if
6063 PolicyKit is not around.
6065 * For embedded systems it is now possible to build udev and
6066 systemd without blkid and/or kmod support.
6068 * "systemctl switch-root" is now capable of switching root
6069 more than once. I.e. in addition to transitions from the
6070 initrd to the host OS it is now possible to transition to
6071 further OS images from the host. This is useful to implement
6072 offline updating tools.
6074 * Various other additions have been made to the RPM macros
6075 shipped with systemd. Use %udev_rules_update() after
6076 installing new udev rules files. %_udevhwdbdir,
6077 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
6078 %_sysctldir are now available which resolve to the right
6079 directories for packages to place various data files in.
6081 * journalctl gained the new --full switch (in addition to
6082 --all, to disable ellipsation for long messages.
6084 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
6085 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
6086 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
6087 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
6088 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
6089 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
6090 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
6091 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
6092 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
6096 * journalctl gained new --since= and --until= switches to
6097 filter by time. It also now supports nice filtering for
6098 units via --unit=/-u.
6100 * Type=oneshot services may use ExecReload= and do the
6103 * The journal daemon now supports time-based rotation and
6104 vacuuming, in addition to the usual disk-space based
6107 * The journal will now index the available field values for
6108 each field name. This enables clients to show pretty drop
6109 downs of available match values when filtering. The bash
6110 completion of journalctl has been updated
6111 accordingly. journalctl gained a new switch -F to list all
6112 values a certain field takes in the journal database.
6114 * More service events are now written as structured messages
6115 to the journal, and made recognizable via message IDs.
6117 * The timedated, localed and hostnamed mini-services which
6118 previously only provided support for changing time, locale
6119 and hostname settings from graphical DEs such as GNOME now
6120 also have a minimal (but very useful) text-based client
6121 utility each. This is probably the nicest way to changing
6122 these settings from the command line now, especially since
6123 it lists available options and is fully integrated with bash
6126 * There's now a new tool "systemd-coredumpctl" to list and
6127 extract coredumps from the journal.
6129 * We now install a README each in /var/log/ and
6130 /etc/rc.d/init.d explaining where the system logs and init
6131 scripts went. This hopefully should help folks who go to
6132 that dirs and look into the otherwise now empty void and
6133 scratch their heads.
6135 * When user-services are invoked (by systemd --user) the
6136 $MANAGERPID env var is set to the PID of systemd.
6138 * SIGRTMIN+24 when sent to a --user instance will now result
6139 in immediate termination of systemd.
6141 * gatewayd received numerous feature additions such as a
6142 "follow" mode, for live syncing and filtering.
6144 * browse.html now allows filtering and showing detailed
6145 information on specific entries. Keyboard navigation and
6146 mouse screen support has been added.
6148 * gatewayd/journalctl now supports HTML5/JSON
6149 Server-Sent-Events as output.
6151 * The SysV init script compatibility logic will now
6152 heuristically determine whether a script supports the
6153 "reload" verb, and only then make this available as
6156 * "systemctl status --follow" has been removed, use "journalctl
6159 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
6160 have been removed since they are hardly useful to be
6163 * And I'd like to take the opportunity to specifically mention
6164 Zbigniew for his great contributions. Zbigniew, you rock!
6166 Contributions from: Andrew Eikum, Christian Hesse, Colin
6167 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
6168 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
6169 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
6170 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
6171 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
6172 Jędrzejewski-Szmek, Сковорода Никита Андреевич
6176 * If /etc/vconsole.conf is non-existent or empty we will no
6177 longer load any console font or key map at boot by
6178 default. Instead the kernel defaults will be left
6179 intact. This is definitely the right thing to do, as no
6180 configuration should mean no configuration, and hard-coding
6181 font names that are different on all archs is probably a bad
6182 idea. Also, the kernel default key map and font should be
6183 good enough for most cases anyway, and mostly identical to
6184 the userspace fonts/key maps we previously overloaded them
6185 with. If distributions want to continue to default to a
6186 non-kernel font or key map they should ship a default
6187 /etc/vconsole.conf with the appropriate contents.
6189 Contributions from: Colin Walters, Daniel J Walsh, Dave
6190 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
6191 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6195 * journalctl gained a new --cursor= switch to show entries
6196 starting from the specified location in the journal.
6198 * We now enforce a size limit on journal entry fields exported
6199 with "-o json" in journalctl. Fields larger than 4K will be
6200 assigned null. This can be turned off with --all.
6202 * An (optional) journal gateway daemon is now available as
6203 "systemd-journal-gatewayd.service". This service provides
6204 access to the journal via HTTP and JSON. This functionality
6205 will be used to implement live log synchronization in both
6206 pull and push modes, but has various other users too, such
6207 as easy log access for debugging of embedded devices. Right
6208 now it is already useful to retrieve the journal via HTTP:
6210 # systemctl start systemd-journal-gatewayd.service
6211 # wget http://localhost:19531/entries
6213 This will download the journal contents in a
6214 /var/log/messages compatible format. The same as JSON:
6216 # curl -H"Accept: application/json" http://localhost:19531/entries
6218 This service is also accessible via a web browser where a
6219 single static HTML5 app is served that uses the JSON logic
6220 to enable the user to do some basic browsing of the
6221 journal. This will be extended later on. Here's an example
6222 screenshot of this app in its current state:
6224 http://0pointer.de/public/journal-gatewayd
6226 Contributions from: Kay Sievers, Lennart Poettering, Robert
6227 Milasan, Tom Gundersen
6231 * The bash completion logic is now available for journalctl
6234 * We do not mount the "cpuset" controller anymore together with
6235 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
6236 started if no parameters are assigned to it. "cpuset" hence
6237 broke code that assumed it could create "cpu" groups and
6240 * journalctl -f will now subscribe to terminal size changes,
6241 and line break accordingly.
6243 Contributions from: Dave Reisner, Kay Sievers, Lennart
6244 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
6248 * nspawn will now create a symlink /etc/localtime in the
6249 container environment, copying the host's timezone
6250 setting. Previously this has been done via a bind mount, but
6251 since symlinks cannot be bind mounted this has now been
6252 changed to create/update the appropriate symlink.
6254 * journalctl -n's line number argument is now optional, and
6255 will default to 10 if omitted.
6257 * journald will now log the maximum size the journal files may
6258 take up on disk. This is particularly useful if the default
6259 built-in logic of determining this parameter from the file
6260 system size is used. Use "systemctl status
6261 systemd-journald.service" to see this information.
6263 * The multi-seat X wrapper tool has been stripped down. As X
6264 is now capable of enumerating graphics devices via udev in a
6265 seat-aware way the wrapper is not strictly necessary
6266 anymore. A stripped down temporary stop-gap is still shipped
6267 until the upstream display managers have been updated to
6268 fully support the new X logic. Expect this wrapper to be
6269 removed entirely in one of the next releases.
6271 * HandleSleepKey= in logind.conf has been split up into
6272 HandleSuspendKey= and HandleHibernateKey=. The old setting
6273 is not available anymore. X11 and the kernel are
6274 distinguishing between these keys and we should too. This
6275 also means the inhibition lock for these keys has been split
6278 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
6279 Poettering, Lukas Nykryn, Václav Pavlín
6283 * Whenever a unit changes state we will now log this to the
6284 journal and show along the unit's own log output in
6287 * ConditionPathIsMountPoint= can now properly detect bind
6288 mount points too. (Previously, a bind mount of one file
6289 system to another place in the same file system could not be
6290 detected as mount, since they shared struct stat's st_dev
6293 * We will now mount the cgroup controllers cpu, cpuacct,
6294 cpuset and the controllers net_cls, net_prio together by
6297 * nspawn containers will now have a virtualized boot
6298 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
6299 over with a randomized ID at container initialization). This
6300 has the effect of making "journalctl -b" do the right thing
6303 * The JSON output journal serialization has been updated not
6304 to generate "endless" list objects anymore, but rather one
6305 JSON object per line. This is more in line how most JSON
6306 parsers expect JSON objects. The new output mode
6307 "json-pretty" has been added to provide similar output, but
6308 neatly aligned for readability by humans.
6310 * We dropped all explicit sync() invocations in the shutdown
6311 code. The kernel does this implicitly anyway in the kernel
6312 reboot() syscall. halt(8)'s -n option is now a compatibility
6315 * We now support virtualized reboot() in containers, as
6316 supported by newer kernels. We will fall back to exit() if
6317 CAP_SYS_REBOOT is not available to the container. Also,
6318 nspawn makes use of this now and will actually reboot the
6319 container if the containerized OS asks for that.
6321 * journalctl will only show local log output by default
6322 now. Use --merge (-m) to show remote log output, too.
6324 * libsystemd-journal gained the new sd_journal_get_usage()
6325 call to determine the current disk usage of all journal
6326 files. This is exposed in the new "journalctl --disk-usage"
6329 * journald gained a new configuration setting SplitMode= in
6330 journald.conf which may be used to control how user journals
6331 are split off. See journald.conf(5) for details.
6333 * A new condition type ConditionFileNotEmpty= has been added.
6335 * tmpfiles' "w" lines now support file globbing, to write
6336 multiple files at once.
6338 * We added Python bindings for the journal submission
6339 APIs. More Python APIs for a number of selected APIs will
6340 likely follow. Note that we intend to add native bindings
6341 only for the Python language, as we consider it common
6342 enough to deserve bindings shipped within systemd. There are
6343 various projects outside of systemd that provide bindings
6344 for languages such as PHP or Lua.
6346 * Many conditions will now resolve specifiers such as %i. In
6347 addition, PathChanged= and related directives of .path units
6348 now support specifiers as well.
6350 * There's now a new RPM macro definition for the system preset
6353 * journald will now warn if it ca not forward a message to the
6354 syslog daemon because its socket is full.
6356 * timedated will no longer write or process /etc/timezone,
6357 except on Debian. As we do not support late mounted /usr
6358 anymore /etc/localtime always being a symlink is now safe,
6359 and hence the information in /etc/timezone is not necessary
6362 * logind will now always reserve one VT for a text getty (VT6
6363 by default). Previously if more than 6 X sessions where
6364 started they took up all the VTs with auto-spawned gettys,
6365 so that no text gettys were available anymore.
6367 * udev will now automatically inform the btrfs kernel logic
6368 about btrfs RAID components showing up. This should make
6369 simple hotplug based btrfs RAID assembly work.
6371 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
6372 (but not for its children which will stay at the kernel
6373 default). This should allow setups with a lot more listening
6376 * systemd will now always pass the configured timezone to the
6377 kernel at boot. timedated will do the same when the timezone
6380 * logind's inhibition logic has been updated. By default,
6381 logind will now handle the lid switch, the power and sleep
6382 keys all the time, even in graphical sessions. If DEs want
6383 to handle these events on their own they should take the new
6384 handle-power-key, handle-sleep-key and handle-lid-switch
6385 inhibitors during their runtime. A simple way to achieve
6386 that is to invoke the DE wrapped in an invocation of:
6388 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
6390 * Access to unit operations is now checked via SELinux taking
6391 the unit file label and client process label into account.
6393 * systemd will now notify the administrator in the journal
6394 when he over-mounts a non-empty directory.
6396 * There are new specifiers that are resolved in unit files,
6397 for the host name (%H), the machine ID (%m) and the boot ID
6400 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
6401 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
6402 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
6403 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
6404 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
6405 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
6406 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
6410 * Support for reading structured kernel messages from
6411 /dev/kmsg has now been added and is enabled by default.
6413 * Support for reading kernel messages from /proc/kmsg has now
6414 been removed. If you want kernel messages in the journal
6415 make sure to run a recent kernel (>= 3.5) that supports
6416 reading structured messages from /dev/kmsg (see
6417 above). /proc/kmsg is now exclusive property of classic
6418 syslog daemons again.
6420 * The libudev API gained the new
6421 udev_device_new_from_device_id() call.
6423 * The logic for file system namespace (ReadOnlyDirectory=,
6424 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
6425 require pivot_root() anymore. This means fewer temporary
6426 directories are created below /tmp for this feature.
6428 * nspawn containers will now see and receive all submounts
6429 made on the host OS below the root file system of the
6432 * Forward Secure Sealing is now supported for Journal files,
6433 which provide cryptographical sealing of journal files so
6434 that attackers cannot alter log history anymore without this
6435 being detectable. Lennart will soon post a blog story about
6436 this explaining it in more detail.
6438 * There are two new service settings RestartPreventExitStatus=
6439 and SuccessExitStatus= which allow configuration of exit
6440 status (exit code or signal) which will be excepted from the
6441 restart logic, resp. consider successful.
6443 * journalctl gained the new --verify switch that can be used
6444 to check the integrity of the structure of journal files and
6445 (if Forward Secure Sealing is enabled) the contents of
6448 * nspawn containers will now be run with /dev/stdin, /dev/fd/
6449 and similar symlinks pre-created. This makes running shells
6450 as container init process a lot more fun.
6452 * The fstab support can now handle PARTUUID= and PARTLABEL=
6455 * A new ConditionHost= condition has been added to match
6456 against the hostname (with globs) and machine ID. This is
6457 useful for clusters where a single OS image is used to
6458 provision a large number of hosts which shall run slightly
6459 different sets of services.
6461 * Services which hit the restart limit will now be placed in a
6464 Contributions from: Bertram Poettering, Dave Reisner, Huang
6465 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
6466 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
6470 * When running in --user mode systemd will now become a
6471 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
6472 tree a lot more organized.
6474 * A new PartOf= unit dependency type has been introduced that
6475 may be used to group services in a natural way.
6477 * "systemctl enable" may now be used to enable instances of
6480 * journalctl now prints error log levels in red, and
6481 warning/notice log levels in bright white. It also supports
6482 filtering by log level now.
6484 * cgtop gained a new -n switch (similar to top), to configure
6485 the maximum number of iterations to run for. It also gained
6486 -b, to run in batch mode (accepting no input).
6488 * The suffix ".service" may now be omitted on most systemctl
6489 command lines involving service unit names.
6491 * There's a new bus call in logind to lock all sessions, as
6492 well as a loginctl verb for it "lock-sessions".
6494 * libsystemd-logind.so gained a new call sd_journal_perror()
6495 that works similar to libc perror() but logs to the journal
6496 and encodes structured information about the error number.
6498 * /etc/crypttab entries now understand the new keyfile-size=
6501 * shutdown(8) now can send a (configurable) wall message when
6502 a shutdown is cancelled.
6504 * The mount propagation mode for the root file system will now
6505 default to "shared", which is useful to make containers work
6506 nicely out-of-the-box so that they receive new mounts from
6507 the host. This can be undone locally by running "mount
6508 --make-rprivate /" if needed.
6510 * The prefdm.service file has been removed. Distributions
6511 should maintain this unit downstream if they intend to keep
6512 it around. However, we recommend writing normal unit files
6513 for display managers instead.
6515 * Since systemd is a crucial part of the OS we will now
6516 default to a number of compiler switches that improve
6517 security (hardening) such as read-only relocations, stack
6518 protection, and suchlike.
6520 * The TimeoutSec= setting for services is now split into
6521 TimeoutStartSec= and TimeoutStopSec= to allow configuration
6522 of individual time outs for the start and the stop phase of
6525 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
6526 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
6527 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
6528 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
6529 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
6530 Gundersen, Zbigniew Jędrzejewski-Szmek
6534 * The journal and id128 C APIs are now fully documented as man
6537 * Extra safety checks have been added when transitioning from
6538 the initial RAM disk to the main system to avoid accidental
6541 * /etc/crypttab entries now understand the new keyfile-offset=
6544 * systemctl -t can now be used to filter by unit load state.
6546 * The journal C API gained the new sd_journal_wait() call to
6547 make writing synchronous journal clients easier.
6549 * journalctl gained the new -D switch to show journals from a
6552 * journalctl now displays a special marker between log
6553 messages of two different boots.
6555 * The journal is now explicitly flushed to /var via a service
6556 systemd-journal-flush.service, rather than implicitly simply
6557 by seeing /var/log/journal to be writable.
6559 * journalctl (and the journal C APIs) can now match for much
6560 more complex expressions, with alternatives and
6563 * When transitioning from the initial RAM disk to the main
6564 system we will now kill all processes in a killing spree to
6565 ensure no processes stay around by accident.
6567 * Three new specifiers may be used in unit files: %u, %h, %s
6568 resolve to the user name, user home directory resp. user
6569 shell. This is useful for running systemd user instances.
6571 * We now automatically rotate journal files if their data
6572 object hash table gets a fill level > 75%. We also size the
6573 hash table based on the configured maximum file size. This
6574 together should lower hash collisions drastically and thus
6575 speed things up a bit.
6577 * journalctl gained the new "--header" switch to introspect
6578 header data of journal files.
6580 * A new setting SystemCallFilters= has been added to services
6581 which may be used to apply blacklists or whitelists to
6582 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
6584 * nspawn gained a new --link-journal= switch (and quicker: -j)
6585 to link the container journal with the host. This makes it
6586 very easy to centralize log viewing on the host for all
6587 guests while still keeping the journal files separated.
6589 * Many bugfixes and optimizations
6591 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
6592 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
6593 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
6598 * Several tools now understand kernel command line arguments,
6599 which are only read when run in an initial RAM disk. They
6600 usually follow closely their normal counterparts, but are
6603 * There's a new tool to analyze the readahead files that are
6604 automatically generated at boot. Use:
6606 /usr/lib/systemd/systemd-readahead analyze /.readahead
6608 * We now provide an early debug shell on tty9 if this enabled. Use:
6610 systemctl enable debug-shell.service
6612 * All plymouth related units have been moved into the Plymouth
6613 package. Please make sure to upgrade your Plymouth version
6616 * systemd-tmpfiles now supports getting passed the basename of
6617 a configuration file only, in which case it will look for it
6618 in all appropriate directories automatically.
6620 * udevadm info now takes a /dev or /sys path as argument, and
6621 does the right thing. Example:
6623 udevadm info /dev/sda
6624 udevadm info /sys/class/block/sda
6626 * systemctl now prints a warning if a unit is stopped but a
6627 unit that might trigger it continues to run. Example: a
6628 service is stopped but the socket that activates it is left
6631 * "systemctl status" will now mention if the log output was
6632 shortened due to rotation since a service has been started.
6634 * The journal API now exposes functions to determine the
6635 "cutoff" times due to rotation.
6637 * journald now understands SIGUSR1 and SIGUSR2 for triggering
6638 immediately flushing of runtime logs to /var if possible,
6639 resp. for triggering immediate rotation of the journal
6642 * It is now considered an error if a service is attempted to
6643 be stopped that is not loaded.
6645 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
6647 * systemd-analyze now supports Python 3
6649 * tmpfiles now supports cleaning up directories via aging
6650 where the first level dirs are always kept around but
6651 directories beneath it automatically aged. This is enabled
6652 by prefixing the age field with '~'.
6654 * Seat objects now expose CanGraphical, CanTTY properties
6655 which is required to deal with very fast bootups where the
6656 display manager might be running before the graphics drivers
6657 completed initialization.
6659 * Seat objects now expose a State property.
6661 * We now include RPM macros for service enabling/disabling
6662 based on the preset logic. We recommend RPM based
6663 distributions to make use of these macros if possible. This
6664 makes it simpler to reuse RPM spec files across
6667 * We now make sure that the collected systemd unit name is
6668 always valid when services log to the journal via
6671 * There's a new man page kernel-command-line(7) detailing all
6672 command line options we understand.
6674 * The fstab generator may now be disabled at boot by passing
6675 fstab=0 on the kernel command line.
6677 * A new kernel command line option modules-load= is now understood
6678 to load a specific kernel module statically, early at boot.
6680 * Unit names specified on the systemctl command line are now
6681 automatically escaped as needed. Also, if file system or
6682 device paths are specified they are automatically turned
6683 into the appropriate mount or device unit names. Example:
6685 systemctl status /home
6686 systemctl status /dev/sda
6688 * The SysVConsole= configuration option has been removed from
6689 system.conf parsing.
6691 * The SysV search path is no longer exported on the D-Bus
6694 * The Names= option has been removed from unit file parsing.
6696 * There's a new man page bootup(7) detailing the boot process.
6698 * Every unit and every generator we ship with systemd now
6699 comes with full documentation. The self-explanatory boot is
6702 * A couple of services gained "systemd-" prefixes in their
6703 name if they wrap systemd code, rather than only external
6704 code. Among them fsck@.service which is now
6705 systemd-fsck@.service.
6707 * The HaveWatchdog property has been removed from the D-Bus
6710 * systemd.confirm_spawn= on the kernel command line should now
6713 * There's a new man page crypttab(5) which details all options
6714 we actually understand.
6716 * systemd-nspawn gained a new --capability= switch to pass
6717 additional capabilities to the container.
6719 * timedated will now read known NTP implementation unit names
6720 from /usr/lib/systemd/ntp-units.d/*.list,
6721 systemd-timedated-ntp.target has been removed.
6723 * journalctl gained a new switch "-b" that lists log data of
6724 the current boot only.
6726 * The notify socket is in the abstract namespace again, in
6727 order to support daemons which chroot() at start-up.
6729 * There is a new Storage= configuration option for journald
6730 which allows configuration of where log data should go. This
6731 also provides a way to disable journal logging entirely, so
6732 that data collected is only forwarded to the console, the
6733 kernel log buffer or another syslog implementation.
6735 * Many bugfixes and optimizations
6737 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
6738 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
6739 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
6740 Shawn Landden, Tom Gundersen
6744 * "systemctl help <unit>" now shows the man page if one is
6747 * Several new man pages have been added.
6749 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
6750 MaxLevelConsole= can now be specified in
6751 journald.conf. These options allow reducing the amount of
6752 data stored on disk or forwarded by the log level.
6754 * TimerSlackNSec= can now be specified in system.conf for
6755 PID1. This allows system-wide power savings.
6757 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
6758 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
6763 * logind is now capable of (optionally) handling power and
6764 sleep keys as well as the lid switch.
6766 * journalctl now understands the syntax "journalctl
6767 /usr/bin/avahi-daemon" to get all log output of a specific
6770 * CapabilityBoundingSet= in system.conf now also influences
6771 the capability bound set of usermode helpers of the kernel.
6773 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
6774 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
6775 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
6776 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
6780 * Note that we skipped 139 releases here in order to set the
6781 new version to something that is greater than both udev's
6782 and systemd's most recent version number.
6784 * udev: all udev sources are merged into the systemd source tree now.
6785 All future udev development will happen in the systemd tree. It
6786 is still fully supported to use the udev daemon and tools without
6787 systemd running, like in initramfs or other init systems. Building
6788 udev though, will require the *build* of the systemd tree, but
6789 udev can be properly *run* without systemd.
6791 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
6792 should be used to create dead device nodes as workarounds for broken
6795 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
6796 no longer supported. udev_monitor_new_from_netlink() needs to be
6797 used to subscribe to events.
6799 * udev: when udevd is started by systemd, processes which are left
6800 behind by forking them off of udev rules, are unconditionally cleaned
6801 up and killed now after the event handling has finished. Services or
6802 daemons must be started as systemd services. Services can be
6803 pulled-in by udev to get started, but they can no longer be directly
6804 forked by udev rules.
6806 * udev: the daemon binary is called systemd-udevd now and installed
6807 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
6808 to adapt to that, create symlink, or rename the binary after building
6811 * libudev no longer provides these symbols:
6812 udev_monitor_from_socket()
6813 udev_queue_get_failed_list_entry()
6814 udev_get_{dev,sys,run}_path()
6815 The versions number was bumped and symbol versioning introduced.
6817 * systemd-loginctl and systemd-journalctl have been renamed
6818 to loginctl and journalctl to match systemctl.
6820 * The config files: /etc/systemd/systemd-logind.conf and
6821 /etc/systemd/systemd-journald.conf have been renamed to
6822 logind.conf and journald.conf. Package updates should rename
6823 the files to the new names on upgrade.
6825 * For almost all files the license is now LGPL2.1+, changed
6826 from the previous GPL2.0+. Exceptions are some minor stuff
6827 of udev (which will be changed to LGPL2.1 eventually, too),
6828 and the MIT licensed sd-daemon.[ch] library that is suitable
6829 to be used as drop-in files.
6831 * systemd and logind now handle system sleep states, in
6832 particular suspending and hibernating.
6834 * logind now implements a sleep/shutdown/idle inhibiting logic
6835 suitable for a variety of uses. Soonishly Lennart will blog
6836 about this in more detail.
6838 * var-run.mount and var-lock.mount are no longer provided
6839 (which previously bind mounted these directories to their new
6840 places). Distributions which have not converted these
6841 directories to symlinks should consider stealing these files
6842 from git history and add them downstream.
6844 * We introduced the Documentation= field for units and added
6845 this to all our shipped units. This is useful to make it
6846 easier to explore the boot and the purpose of the various
6849 * All smaller setup units (such as
6850 systemd-vconsole-setup.service) now detect properly if they
6851 are run in a container and are skipped when
6852 appropriate. This guarantees an entirely noise-free boot in
6853 Linux container environments such as systemd-nspawn.
6855 * A framework for implementing offline system updates is now
6856 integrated, for details see:
6857 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
6859 * A new service type Type=idle is available now which helps us
6860 avoiding ugly interleaving of getty output and boot status
6863 * There's now a system-wide CapabilityBoundingSet= option to
6864 globally reduce the set of capabilities for the
6865 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
6866 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
6867 even CAP_NET_ADMIN system-wide for secure systems.
6869 * There are now system-wide DefaultLimitXXX= options to
6870 globally change the defaults of the various resource limits
6871 for all units started by PID 1.
6873 * Harald Hoyer's systemd test suite has been integrated into
6874 systemd which allows easy testing of systemd builds in qemu
6875 and nspawn. (This is really awesome! Ask us for details!)
6877 * The fstab parser is now implemented as generator, not inside
6880 * systemctl will now warn you if .mount units generated from
6881 /etc/fstab are out of date due to changes in fstab that
6882 have not been read by systemd yet.
6884 * systemd is now suitable for usage in initrds. Dracut has
6885 already been updated to make use of this. With this in place
6886 initrds get a slight bit faster but primarily are much
6887 easier to introspect and debug since "systemctl status" in
6888 the host system can be used to introspect initrd services,
6889 and the journal from the initrd is kept around too.
6891 * systemd-delta has been added, a tool to explore differences
6892 between user/admin configuration and vendor defaults.
6894 * PrivateTmp= now affects both /tmp and /var/tmp.
6896 * Boot time status messages are now much prettier and feature
6897 proper english language. Booting up systemd has never been
6900 * Read-ahead pack files now include the inode number of all
6901 files to pre-cache. When the inode changes the pre-caching
6902 is not attempted. This should be nicer to deal with updated
6903 packages which might result in changes of read-ahead
6906 * We now temporaritly lower the kernel's read_ahead_kb variable
6907 when collecting read-ahead data to ensure the kernel's
6908 built-in read-ahead does not add noise to our measurements
6909 of necessary blocks to pre-cache.
6911 * There's now RequiresMountsFor= to add automatic dependencies
6912 for all mounts necessary for a specific file system path.
6914 * MountAuto= and SwapAuto= have been removed from
6915 system.conf. Mounting file systems at boot has to take place
6918 * nspawn now learned a new switch --uuid= to set the machine
6919 ID on the command line.
6921 * nspawn now learned the -b switch to automatically search
6924 * vt102 is now the default TERM for serial TTYs, upgraded from
6927 * systemd-logind now works on VT-less systems.
6929 * The build tree has been reorganized. The individual
6930 components now have directories of their own.
6932 * A new condition type ConditionPathIsReadWrite= is now available.
6934 * nspawn learned the new -C switch to create cgroups for the
6935 container in other hierarchies.
6937 * We now have support for hardware watchdogs, configurable in
6940 * The scheduled shutdown logic now has a public API.
6942 * We now mount /tmp as tmpfs by default, but this can be
6943 masked and /etc/fstab can override it.
6945 * Since udisks does not make use of /media anymore we are not
6946 mounting a tmpfs on it anymore.
6948 * journalctl gained a new --local switch to only interleave
6949 locally generated journal files.
6951 * We can now load the IMA policy at boot automatically.
6953 * The GTK tools have been split off into a systemd-ui.
6955 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
6956 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
6957 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
6958 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
6959 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
6960 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
6961 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
6962 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
6963 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
6968 * This is mostly a bugfix release
6970 * Support optional initialization of the machine ID from the
6971 KVM or container configured UUID.
6973 * Support immediate reboots with "systemctl reboot -ff"
6975 * Show /etc/os-release data in systemd-analyze output
6977 * Many bugfixes for the journal, including endianness fixes and
6978 ensuring that disk space enforcement works
6980 * sd-login.h is C++ compatible again
6982 * Extend the /etc/os-release format on request of the Debian
6985 * We now refuse non-UTF8 strings used in various configuration
6986 and unit files. This is done to ensure we do not pass invalid
6987 data over D-Bus or expose it elsewhere.
6989 * Register Mimo USB Screens as suitable for automatic seat
6992 * Read SELinux client context from journal clients in a race
6995 * Reorder configuration file lookup order. /etc now always
6996 overrides /run in order to allow the administrator to always
6997 and unconditionally override vendor-supplied or
6998 automatically generated data.
7000 * The various user visible bits of the journal now have man
7001 pages. We still lack man pages for the journal API calls
7004 * We now ship all man pages in HTML format again in the
7007 Contributions from: Dave Reisner, Dirk Eibach, Frederic
7008 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
7009 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
7014 * This is mostly a bugfix release
7016 * systems lacking /etc/os-release are no longer supported.
7018 * Various functionality updates to libsystemd-login.so
7020 * Track class of PAM logins to distinguish greeters from
7023 Contributions from: Kay Sievers, Lennart Poettering, Michael
7028 * This is an important bugfix release for v41.
7030 * Building man pages is now optional which should be useful
7031 for those building systemd from git but unwilling to install
7034 * Watchdog support for supervising services is now usable. In
7035 a future release support for hardware watchdogs
7036 (i.e. /dev/watchdog) will be added building on this.
7038 * Service start rate limiting is now configurable and can be
7039 turned off per service. When a start rate limit is hit a
7040 reboot can automatically be triggered.
7042 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
7044 Contributions from: Benjamin Franzke, Bill Nottingham,
7045 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
7046 Schmidt, Michał Górny, Piotr Drąg
7050 * The systemd binary is installed /usr/lib/systemd/systemd now;
7051 An existing /sbin/init symlink needs to be adapted with the
7054 * The code that loads kernel modules has been ported to invoke
7055 libkmod directly, instead of modprobe. This means we do not
7056 support systems with module-init-tools anymore.
7058 * Watchdog support is now already useful, but still not
7061 * A new kernel command line option systemd.setenv= is
7062 understood to set system wide environment variables
7063 dynamically at boot.
7065 * We now limit the set of capabilities of systemd-journald.
7067 * We now set SIGPIPE to ignore by default, since it only is
7068 useful in shell pipelines, and has little use in general
7069 code. This can be disabled with IgnoreSIPIPE=no in unit
7072 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
7073 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
7078 * This is mostly a bugfix release
7080 * We now expose the reason why a service failed in the
7081 "Result" D-Bus property.
7083 * Rudimentary service watchdog support (will be completed over
7084 the next few releases.)
7086 * When systemd forks off in order execute some service we will
7087 now immediately changes its argv[0] to reflect which process
7088 it will execute. This is useful to minimize the time window
7089 with a generic argv[0], which makes bootcharts more useful
7091 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
7092 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
7093 Mike Kazantsev, Ray Strode
7097 * This is mostly a test release, but incorporates many
7100 * New systemd-cgtop tool to show control groups by their
7103 * Linking against libacl for ACLs is optional again. If
7104 disabled, support tracking device access for active logins
7105 goes becomes unavailable, and so does access to the user
7106 journals by the respective users.
7108 * If a group "adm" exists, journal files are automatically
7109 owned by them, thus allow members of this group full access
7110 to the system journal as well as all user journals.
7112 * The journal now stores the SELinux context of the logging
7113 client for all entries.
7115 * Add C++ inclusion guards to all public headers
7117 * New output mode "cat" in the journal to print only text
7118 messages, without any meta data like date or time.
7120 * Include tiny X server wrapper as a temporary stop-gap to
7121 teach XOrg udev display enumeration. This is used by display
7122 managers such as gdm, and will go away as soon as XOrg
7123 learned native udev hotplugging for display devices.
7125 * Add new systemd-cat tool for executing arbitrary programs
7126 with STDERR/STDOUT connected to the journal. Can also act as
7127 BSD logger replacement, and does so by default.
7129 * Optionally store all locally generated coredumps in the
7130 journal along with meta data.
7132 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
7133 writing short strings to files (for usage for /sys), and for
7134 creating symlinks, character and block device nodes.
7136 * New unit file option ControlGroupPersistent= to make cgroups
7137 persistent, following the mechanisms outlined in
7138 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
7140 * Support multiple local RTCs in a sane way
7142 * No longer monopolize IO when replaying readahead data on
7143 rotating disks, since we might starve non-file-system IO to
7144 death, since fanotify() will not see accesses done by blkid,
7147 * Do not show kernel threads in systemd-cgls anymore, unless
7148 requested with new -k switch.
7150 Contributions from: Dan Horák, Kay Sievers, Lennart
7151 Poettering, Michal Schmidt
7155 * This is mostly a test release, but incorporates many
7158 * The git repository moved to:
7159 git://anongit.freedesktop.org/systemd/systemd
7160 ssh://git.freedesktop.org/git/systemd/systemd
7162 * First release with the journal
7163 http://0pointer.de/blog/projects/the-journal.html
7165 * The journal replaces both systemd-kmsg-syslogd and
7166 systemd-stdout-bridge.
7168 * New sd_pid_get_unit() API call in libsystemd-logind
7170 * Many systemadm clean-ups
7172 * Introduce remote-fs-pre.target which is ordered before all
7173 remote mounts and may be used to start services before all
7176 * Added Mageia support
7178 * Add bash completion for systemd-loginctl
7180 * Actively monitor PID file creation for daemons which exit in
7181 the parent process before having finished writing the PID
7182 file in the daemon process. Daemons which do this need to be
7183 fixed (i.e. PID file creation must have finished before the
7184 parent exits), but we now react a bit more gracefully to them.
7186 * Add colourful boot output, mimicking the well-known output
7187 of existing distributions.
7189 * New option PassCredentials= for socket units, for
7190 compatibility with a recent kernel ABI breakage.
7192 * /etc/rc.local is now hooked in via a generator binary, and
7193 thus will no longer act as synchronization point during
7196 * systemctl list-unit-files now supports --root=.
7198 * systemd-tmpfiles now understands two new commands: z, Z for
7199 relabelling files according to the SELinux database. This is
7200 useful to apply SELinux labels to specific files in /sys,
7203 * Output of SysV services is now forwarded to both the console
7204 and the journal by default, not only just the console.
7206 * New man pages for all APIs from libsystemd-login.
7208 * The build tree got reorganized and the build system is a
7209 lot more modular allowing embedded setups to specifically
7210 select the components of systemd they are interested in.
7212 * Support for Linux systems lacking the kernel VT subsystem is
7215 * configure's --with-rootdir= got renamed to
7216 --with-rootprefix= to follow the naming used by udev and
7219 * Unless specified otherwise we will now install to /usr instead
7220 of /usr/local by default.
7222 * Processes with '@' in argv[0][0] are now excluded from the
7223 final shut-down killing spree, following the logic explained
7225 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
7227 * All processes remaining in a service cgroup when we enter
7228 the START or START_PRE states are now killed with
7229 SIGKILL. That means it is no longer possible to spawn
7230 background processes from ExecStart= lines (which was never
7231 supported anyway, and bad style).
7233 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
7234 reloading of units together.
7236 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
7237 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
7238 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
7239 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
7240 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek