1 systemd System and Service Manager
5 * The "hybrid" control group mode has been modified to improve
6 compatibility with "legacy" cgroups-v1 setups. Specifically, the
7 "hybrid" setup of /sys/fs/cgroup is now pretty much identical to
8 "legacy" (including /sys/fs/cgroup/systemd as "name=systemd" named
9 cgroups-v1 hierarchy), the only externally visible change being that
10 the cgroups-v2 hierarchy is also mounted, to
11 /sys/fs/cgroup/unified. This should provide a large degree of
12 compatibility with "legacy" cgroups-v1, while taking benefit of the
13 better management capabilities of cgroups-v2.
15 * The default control group setup mode may be selected both a boot-time
16 via a set of kernel command line parameters (specifically:
17 systemd.unified_cgroup_hierarchy= and
18 systemd.legacy_systemd_cgroup_controller=), as well as a compile-time
19 default selected on the configure command line
20 (--with-default-hierarchy=). The upstream default is "hybrid"
21 (i.e. the cgroups-v1 + cgroups-v2 mixture discussed above) now, but
22 this will change in a future systemd version to be "unified" (pure
23 cgroups-v2 mode). The third option for the compile time option is
24 "legacy", to enter pure cgroups-v1 mode. We recommend downstream
25 distributions to default to "hybrid" mode for release distributions,
26 starting with v233. We recommend "unified" for development
27 distributions (specifically: distributions such as Fedora's rawhide)
28 as that's where things are headed in the long run. Use "legacy" for
29 greatest stability and compatibility only.
31 * Note one current limitation of "unified" and "hybrid" control group
32 setup modes: the kernel currently does not permit the systemd --user
33 instance (i.e. unprivileged code) to migrate processes between two
34 disconnected cgroup subtrees, even if both are managed and owned by
35 the user. This effectively means "systemd-run --user --scope" doesn't
36 work when invoked from outside of any "systemd --user" service or
37 scope. Specifically, it is not supported from session scopes. We are
38 working on fixing this in a future systemd version. (See #3388 for
39 further details about this.)
41 * DBus policy files are now installed into /usr rather than /etc. Make
42 sure your system has dbus >= 1.9.18 running before upgrading to this
43 version, or override the install path with --with-dbuspolicydir= .
45 * All python scripts shipped with systemd (specifically: the various
46 tests written in Python) now require Python 3.
48 * systemd unit tests can now run standalone (without the source or
49 build directories), and can be installed into /usr/lib/systemd/tests/
50 with 'make install-tests'.
52 * Note that from this version on, CONFIG_CRYPTO_USER_API_HASH,
53 CONFIG_CRYPTO_HMAC and CONFIG_CRYPTO_SHA256 need to be enabled in the
56 * Support for the %c, %r, %R specifiers in unit files has been
57 removed. Specifiers are not supposed to be dependent on configuration
58 in the unit file itself (so that they resolve the same regardless
59 where used in the unit files), but these specifiers were influenced
62 * The shell invoked by debug-shell.service now defaults to /bin/sh in
63 all cases. If distributions want to use a different shell for this
64 purpose (for example Fedora's /sbin/sushell) they need to specify
65 this explicitly at configure time using --with-debug-shell=.
67 * The confirmation spawn prompt has been reworked to offer the
70 (c)ontinue, proceed without asking anymore
71 (D)ump, show the state of the unit
72 (f)ail, don't execute the command and pretend it failed
74 (i)nfo, show a short summary of the unit
75 (j)obs, show jobs that are in progress
76 (s)kip, don't execute the command and pretend it succeeded
77 (y)es, execute the command
79 The 'n' choice for the confirmation spawn prompt has been removed,
80 because its meaning was confusing.
82 The prompt may now also be redirected to an alternative console by
83 specifying the console as parameter to systemd.confirm_spawn=.
85 * Services of Type=notify require a READY=1 notification to be sent
86 during startup. If no such message is sent, the service now fails,
87 even if the main process exited with a successful exit code.
89 * Services that fail to start up correctly now always have their
90 ExecStopPost= commands executed. Previously, they'd enter "failed"
91 state directly, without executing these commands.
93 * The option MulticastDNS= of network configuration files has acquired
94 an actual implementation. With MulticastDNS=yes a host can resolve
95 names of remote hosts and reply to mDNS A and AAAA requests.
97 * When units are about to be started an additional check is now done to
98 ensure that all dependencies of type BindsTo= (when used in
99 combination with After=) have been started.
101 * systemd-analyze gained a new verb "syscall-filter" which shows which
102 system call groups are defined for the SystemCallFilter= unit file
103 setting, and which system calls they contain.
105 * A new system call filter group "@filesystem" has been added,
106 consisting of various file system related system calls. Group
107 "@reboot" has been added, covering reboot, kexec and shutdown related
108 calls. Finally, group "@swap" has been added covering swap
109 configuration related calls.
111 * A new unit file option RestrictNamespaces= has been added that may be
112 used to restrict access to the various process namespace types the
113 Linux kernel provides. Specifically, it may be used to take away the
114 right for a service unit to create additional file system, network,
115 user, and other namespaces. This sandboxing option is particularly
116 relevant due to the high amount of recently discovered namespacing
117 related vulnerabilities in the kernel.
119 * systemd-udev's .link files gained support for a new AutoNegotiation=
120 setting for configuring Ethernet auto-negotiation.
122 * systemd-networkd's .network files gained support for a new
123 ListenPort= setting in the [DHCP] section to explicitly configure the
124 UDP client port the DHCP client shall listen on.
126 * .network files gained a new Unmanaged= boolean setting for explicitly
127 excluding one or more interfaces from management by systemd-networkd.
129 * The systemd-networkd ProxyARP= option has been renamed to
130 IPV4ProxyARP=. Similarly, VXLAN-specific option ARPProxy= has been
131 renamed to ReduceARPProxy=. The old names continue to be available
134 * systemd-networkd gained support for configuring IPv6 Proxy NDP
135 addresses via the new IPv6ProxyNDPAddress= .network file setting.
137 * systemd-networkd's bonding device support gained support for two new
138 configuration options ActiveSlave= and PrimarySlave=.
140 * The various options in the [Match] section of .network files gained
141 support for negative matching.
143 * New systemd-specific mount options are now understood in /etc/fstab:
145 x-systemd.mount-timeout= may be used to configure the maximum
146 permitted runtime of the mount command.
148 x-systemd.device-bound may be set to bind a mount point to its
149 backing device unit, in order to automatically remove a mount point
150 if its backing device is unplugged. This option may also be
151 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
152 on the block device, which is now automatically set for all CDROM
153 drives, so that mounted CDs are automatically unmounted when they are
154 removed from the drive.
156 x-systemd.after= and x-systemd.before= may be used to explicitly
157 order a mount after or before another unit or mount point.
159 * Enqueued start jobs for device units are now automatically garbage
160 collected if there are no jobs waiting for them anymore.
162 * systemctl list-jobs gained two new switches: with --after, for every
163 queued job the jobs it's waiting for are shown; with --before the
164 jobs which it's blocking are shown.
166 * systemd-nspawn gained support for ephemeral boots from disk images
167 (or in other words: --ephemeral and --image= may now be
168 combined). Moreover, ephemeral boots are now supported for normal
169 directories, even if the backing file system is not btrfs. Of course,
170 if the file system does not support file system snapshots or
171 reflinks, the initial copy operation will be relatively expensive, but
172 this should still be suitable for many use cases.
174 * Calendar time specifications in .timer units now support
175 specifications relative to the end of a month by using "~" instead of
176 "-" as separator between month and day. For example, "*-02~03" means
177 "the third last day in February". In addition a new syntax for
178 repeated events has been added using the "/" character. For example,
179 "9..17/2:00" means "every two hours from 9am to 5pm".
181 * systemd-socket-proxyd gained a new parameter --connections-max= for
182 configuring the maximum number of concurrent connections.
184 * sd-id128 gained a new API for generating unique IDs for the host in a
185 way that does not leak the machine ID. Specifically,
186 sd_id128_get_machine_app_specific() derives an ID based on the
187 machine ID a in well-defined, non-reversible, stable way. This is
188 useful whenever an identifier for the host is needed but where the
189 identifier shall not be useful to identify the system beyond the
190 scope of the application itself. (Internally this uses HMAC-SHA256 as
191 keyed hash function using the machine ID as input.)
193 * NotifyAccess= gained a new supported value "exec". When set
194 notifications are accepted from all processes systemd itself invoked,
195 including all control processes.
197 * .nspawn files gained support for defining overlay mounts using the
198 Overlay= and OverlayReadOnly= options. Previously this functionality
199 was only available on the systemd-nspawn command line.
201 * systemd-nspawn's --bind= and --overlay= options gained support for
202 bind/overlay mounts whose source lies within the container tree by
203 prefixing the source path with "+".
205 * systemd-nspawn's --bind= and --overlay= options gained support for
206 automatically allocating a temporary source directory in /var/tmp
207 that is removed when the container dies. Specifically, if the source
208 directory is specified as empty string this mechanism is selected. An
209 example usage is --overlay=+/var::/var, which creates an overlay
210 mount based on the original /var contained in the image, overlayed
211 with a temporary directory in the host's /var/tmp. This way changes
212 to /var are automatically flushed when the container shuts down.
214 * systemd-nspawn --image= option does now permit raw file system block
215 devices (in addition to images containing partition tables, as
218 * The disk image dissection logic in systemd-nspawn gained support for
219 automatically setting up LUKS encrypted as well as Verity protected
220 partitions. When a container is booted from an encrypted image the
221 passphrase is queried at start-up time. When a container with Verity
222 data is started, the root hash is search in a ".roothash" file
223 accompanying the disk image (alternatively, pass the root hash via
224 the new --root-hash= command line option).
226 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
227 be used to dissect disk images the same way as systemd-nspawn does
228 it, following the Bootable Partition Specification. It may even be
229 used to mount disk images with complex partition setups (including
230 LUKS and Verity partitions) to a local host directory, in order to
231 inspect them. This tool is not considered public API (yet), and is
232 thus not installed into /usr/bin. Please do not rely on its
233 existence, since it might go away or be changed in later systemd
236 * A new generator "systemd-verity-generator" has been added, similar in
237 style to "systemd-cryptsetup-generator", permitting automatic setup of
238 Verity root partitions when systemd boots up. In order to make use of
239 this your partition setup should follow the Discoverable Partitions
240 Specification, and the GPT partition ID of the root file system
241 partition should be identical to the upper 128bit of the Verity root
242 hash. The GPT partition ID of the Verity partition protecting it
243 should be the lower 128bit of the Verity root hash. If the partition
244 image follows this model it is sufficient to specify a single
245 "roothash=" kernel command line argument to both configure which root
246 image and verity partition to use as well as the root hash for
247 it. Note that systemd-nspawn's Verity support follows the same
248 semantics, meaning that disk images with proper Verity data in place
249 may be booted in containers with systemd-nspawn as well as on
250 physical systems via the verity generator. Also note that the "mkosi"
251 tool available at https://github.com/systemd/mkosi has been updated
252 to generate Verity protected disk images following this scheme. In
253 fact, it has been updated to generate disk images that optionally
254 implement a complete UEFI SecureBoot trust chain, involving a signed
255 kernel and initrd image that incorporates such a root hash as well as
256 a Verity-enabled root partition.
258 * The hardware database (hwdb) udev supports has been updated to carry
259 accelerometer quirks.
261 * All system services are now run with a fresh kernel keyring set up
262 for them. The invocation ID is stored by default in it, thus
263 providing a safe, non-overridable way to determine the invocation
266 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
267 options for bind mounting arbitrary paths in a service-specific
268 way. When these options are used, arbitrary host or service files and
269 directories may be mounted to arbitrary locations in the service's
272 * Documentation has been added that lists all of systemd's low-level
273 environment variables:
275 https://github.com/systemd/systemd/blob/master/ENVIRONMENT.md
277 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
278 whether a specific socket file descriptor matches a specified socket
281 * systemd-firstboot has been updated to check for the
282 systemd.firstboot= kernel command line option. It accepts a boolean
283 and when set to false the first boot questions are skipped.
285 * systemd-fstab-generator has been updated to check for the
286 systemd.volatile= kernel command line option, which either takes an
287 optional boolean parameter or the special value "state". If used the
288 system may be booted in a "volatile" boot mode. Specifically,
289 "systemd.volatile" is used, the root directory will be mounted as
290 tmpfs, and only /usr is mounted from the actual root file system. If
291 "systemd.volatile=state" is used, the root directory will be mounted
292 as usual, but /var is mounted as tmpfs. This concept provides similar
293 functionality as systemd-nspawn's --volatile= option, but provides it
294 on physical boots. Use this option for implementing stateless
295 systems, or testing systems with all state and/or configuration reset
296 to the defaults. (Note though that many distributions are not
297 prepared to boot up without a populated /etc or /var, though.)
299 * systemd-gpt-auto-generator gained support for LUKS encrypted root
300 partitions. Previously it only supported LUKS encrypted partitions
301 for all other uses, except for the root partition itself.
303 * Socket units gained support for listening on AF_VSOCK sockets for
304 communication in virtualized QEMU environments.
306 * The "configure" script gained a new option --with-fallback-hostname=
307 for specifying the fallback hostname to use if none is configured in
308 /etc/hostname. For example, by specifying
309 --with-fallback-hostname=fedora it is possible to default to a
310 hostname of "fedora" on pristine installations.
312 * systemd-cgls gained support for a new --unit= switch for listing only
313 the control groups of a specific unit. Similar --user-unit= has been
314 added for listing only the control groups of a specific user unit.
316 * systemd-mount gained a new --umount switch for unmounting a mount or
317 automount point (and all mount/automount points below it).
319 * systemd will now refuse full configuration reloads (via systemctl
320 daemon-reload and related calls) unless at least 16MiB of free space
321 are available in /run. This is a safety precaution in order to ensure
322 that generators can safely operate after the reload completed.
324 * A new unit file option RootImage= has been added, which has a similar
325 effect as RootDirectory= but mounts the service's root directory from
326 a disk image instead of plain directory. This logic reuses the same
327 image dissection and mount logic that systemd-nspawn already uses,
328 and hence supports any disk images systemd-nspawn supports, including
329 those following the Discoverable Partition Specification, as well as
330 Verity enabled images. This option enables systemd to run system
331 services directly off disk images acting as resource bundles,
332 possibly even including full integrity data.
334 * A new MountAPIVFS= unit file option has been added, taking a boolean
335 argument. If enabled /proc, /sys and /dev (collectively called the
336 "API VFS") will be mounted for the service. This is only relevant if
337 RootDirectory= or RootImage= is used for the service, as these mounts
338 are of course in place in the host mount namespace anyway.
340 * systemd-nspawn gained support for a new --pivot-root= switch. If
341 specified the root directory within the container image is pivoted to
342 the specified mount point, while the original root disk is moved to a
343 different place. This option enables booting of ostree images
344 directly with systemd-nspawn.
346 * The systemd build scripts will no longer complain if the NTP server
347 addresses are not changed from the defaults. Google now supports
348 these NTP servers officially. We still recommend downstreams to
349 properly register an NTP pool with the NTP pool project though.
351 * coredumpctl gained new new "--reverse" option for printing the list
352 of coredumps in reverse order.
354 * coredumpctl will now show additional information about truncated and
355 inaccessible coredumps, as well as coredumps that are still being
356 processed. It also gained a new --quiet switch for suppressing
357 additional informational message in its output.
359 * coredumpctl gained support for only showing coredumps newer and/or
360 older than specific timestamps, using the new --since= and --until=
361 options, reminiscent of journalctl's options by the same name.
363 * The systemd-coredump logic has been improved so that it may be reused
364 to collect backtraces in non-compiled languages, for example in
365 scripting languages such as Python.
367 * machinectl will now show the UID shift of local containers, if user
368 namespacing is enabled for them.
370 * systemd will now optionally run "environment generator" binaries at
371 configuration load time. They may be used to add environment
372 variables to the environment block passed to services invoked. One
373 user environment generator is shipped by default that sets up
374 environment variables based on files dropped into /etc/environment.d
375 and ~/.config/environment.d/.
377 * systemd-resolved now includes the new, recently published 2017 DNSSEC
380 * hostnamed has been updated to report a new chassis type of
381 "convertible" to cover "foldable" laptops that can both act as a
382 tablet and as a laptop, such as various Lenovo Yoga devices.
384 Contributions from: Adrián López, Alexander Galanin, Alexander
385 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
386 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
387 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
388 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
389 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
390 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
391 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
392 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
393 Gianluca Boiano, Giedrius Statkevičius, Graeme Lawes, Hans de Goede,
394 Harald Hoyer, Ian Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan
395 Synacek, Jason Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen,
396 Karl Kraus, Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart
397 Poettering, Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de
398 Vries, Maks Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry,
399 Mark Stosberg, Martin Ejdestig, Martin Pitt, Mauricio Faria de
400 Oliveira, micah, Michael Biebl, Michael Shields, Michal Schmidt, Michal
401 Sekletar, Michel Kraus, Mike Gilbert, Mikko Ylinen, Mirza Krak,
402 Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter Körner, Philip
403 Withnall, Piotr Drąg, Ray Strode, Reverend Homer, Rike-Benjamin
404 Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan Bilovol, sammynx,
405 Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger, Stefan Hajnoczi,
406 Stefan Schweter, Stuart McLaren, Susant Sahani, Sylvain Plantefève,
407 Taylor Smock, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tibor
408 Nagy, Tobias Stoeckmann, Tom Gundersen, Torstein Husebø, Viktar
409 Vaŭčkievič, Viktor Mihajlovski, Vitaly Sulimov, Waldemar Brodkorb,
410 Walter Garcia-Fontes, Wim de With, Yassine Imounachen, Yi EungJun,
411 YunQiang Su, Yu Watanabe, Zbigniew Jędrzejewski-Szmek, Александр
418 * udev now runs with MemoryDenyWriteExecute=, RestrictRealtime= and
419 RestrictAddressFamilies= enabled. These sandboxing options should
420 generally be compatible with the various external udev call-out
421 binaries we are aware of, however there may be exceptions, in
422 particular when exotic languages for these call-outs are used. In
423 this case, consider turning off these settings locally.
425 * The new RemoveIPC= option can be used to remove IPC objects owned by
426 the user or group of a service when that service exits.
428 * The new ProtectKernelModules= option can be used to disable explicit
429 load and unload operations of kernel modules by a service. In
430 addition access to /usr/lib/modules is removed if this option is set.
432 * ProtectSystem= option gained a new value "strict", which causes the
433 whole file system tree with the exception of /dev, /proc, and /sys,
434 to be remounted read-only for a service.
436 * The new ProtectKernelTunables= option can be used to disable
437 modification of configuration files in /sys and /proc by a service.
438 Various directories and files are remounted read-only, so access is
439 restricted even if the file permissions would allow it.
441 * The new ProtectControlGroups= option can be used to disable write
442 access by a service to /sys/fs/cgroup.
444 * Various systemd services have been hardened with
445 ProtectKernelTunables=yes, ProtectControlGroups=yes,
446 RestrictAddressFamilies=.
448 * Support for dynamically creating users for the lifetime of a service
449 has been added. If DynamicUser=yes is specified, user and group IDs
450 will be allocated from the range 61184..65519 for the lifetime of the
451 service. They can be resolved using the new nss-systemd.so NSS
452 module. The module must be enabled in /etc/nsswitch.conf. Services
453 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
454 any resources allocated by the service will be cleaned up when the
455 service exits. They also have ProtectHome=read-only and
456 ProtectSystem=strict enabled, so they are not able to make any
457 permanent modifications to the system.
459 * The nss-systemd module also always resolves root and nobody, making
460 it possible to have no /etc/passwd or /etc/group files in minimal
461 container or chroot environments.
463 * Services may be started with their own user namespace using the new
464 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
465 under which the service is running are mapped. All other users are
468 * Support for the cgroup namespace has been added to systemd-nspawn. If
469 supported by kernel, the container system started by systemd-nspawn
470 will have its own view of the cgroup hierarchy. This new behaviour
471 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
473 * The new MemorySwapMax= option can be used to limit the maximum swap
474 usage under the unified cgroup hierarchy.
476 * Support for the CPU controller in the unified cgroup hierarchy has
477 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
478 options. This controller requires out-of-tree patches for the kernel
479 and the support is provisional.
481 * Mount and automount units may now be created transiently
482 (i.e. dynamically at runtime via the bus API, instead of requiring
483 unit files in the file system).
485 * systemd-mount is a new tool which may mount file systems – much like
486 mount(8), optionally pulling in additional dependencies through
487 transient .mount and .automount units. For example, this tool
488 automatically runs fsck on a backing block device before mounting,
489 and allows the automount logic to be used dynamically from the
490 command line for establishing mount points. This tool is particularly
491 useful when dealing with removable media, as it will ensure fsck is
492 run – if necessary – before the first access and that the file system
493 is quickly unmounted after each access by utilizing the automount
494 logic. This maximizes the chance that the file system on the
495 removable media stays in a clean state, and if it isn't in a clean
496 state is fixed automatically.
498 * LazyUnmount=yes option for mount units has been added to expose the
499 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
502 * /efi will be used as the mount point of the EFI boot partition, if
503 the directory is present, and the mount point was not configured
504 through other means (e.g. fstab). If /efi directory does not exist,
505 /boot will be used as before. This makes it easier to automatically
506 mount the EFI partition on systems where /boot is used for something
509 * When operating on GPT disk images for containers, systemd-nspawn will
510 now mount the ESP to /boot or /efi according to the same rules as PID
511 1 running on a host. This allows tools like "bootctl" to operate
512 correctly within such containers, in order to make container images
513 bootable on physical systems.
515 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
517 * Two new user session targets have been added to support running
518 graphical sessions under the systemd --user instance:
519 graphical-session.target and graphical-session-pre.target. See
520 systemd.special(7) for a description of how those targets should be
523 * The vconsole initialization code has been significantly reworked to
524 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
525 support unicode keymaps. Font and keymap configuration will now be
526 copied to all allocated virtual consoles.
528 * FreeBSD's bhyve virtualization is now detected.
530 * Information recorded in the journal for core dumps now includes the
531 contents of /proc/mountinfo and the command line of the process at
532 the top of the process hierarchy (which is usually the init process
535 * systemd-journal-gatewayd learned the --directory= option to serve
536 files from the specified location.
538 * journalctl --root=… can be used to peruse the journal in the
539 /var/log/ directories inside of a container tree. This is similar to
540 the existing --machine= option, but does not require the container to
543 * The hardware database has been extended to support
544 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
547 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
548 specify the click rate for mice which include a horizontal wheel with
549 a click rate that is different than the one for the vertical wheel.
551 * systemd-run gained a new --wait option that makes service execution
552 synchronous. (Specifically, the command will not return until the
553 specified service binary exited.)
555 * systemctl gained a new --wait option that causes the start command to
556 wait until the units being started have terminated again.
558 * A new journal output mode "short-full" has been added which displays
559 timestamps with abbreviated English day names and adds a timezone
560 suffix. Those timestamps include more information than the default
561 "short" output mode, and can be passed directly to journalctl's
562 --since= and --until= options.
564 * /etc/resolv.conf will be bind-mounted into containers started by
565 systemd-nspawn, if possible, so any changes to resolv.conf contents
566 are automatically propagated to the container.
568 * The number of instances for socket-activated services originating
569 from a single IP address can be limited with
570 MaxConnectionsPerSource=, extending the existing setting of
573 * systemd-networkd gained support for vcan ("Virtual CAN") interface
576 * .netdev and .network configuration can now be extended through
579 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
580 Segmentation Offload, Generic Receive Offload, Large Receive Offload
581 can be enabled and disabled using the new UDPSegmentationOffload=,
582 TCPSegmentationOffload=, GenericSegmentationOffload=,
583 GenericReceiveOffload=, LargeReceiveOffload= options in the
584 [Link] section of .link files.
586 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
587 Port VLAN ID can be configured for bridge devices using the new STP=,
588 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
589 section of .netdev files.
591 * The route table to which routes received over DHCP or RA should be
592 added can be configured with the new RouteTable= option in the [DHCP]
593 and [IPv6AcceptRA] sections of .network files.
595 * The Address Resolution Protocol can be disabled on links managed by
596 systemd-networkd using the ARP=no setting in the [Link] section of
599 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
600 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
601 encode information about the result and exit codes of the current
602 service runtime cycle.
604 * systemd-sysctl will now configure kernel parameters in the order
605 they occur in the configuration files. This matches what sysctl
606 has been traditionally doing.
608 * kernel-install "plugins" that are executed to perform various
609 tasks after a new kernel is added and before an old one is removed
610 can now return a special value to terminate the procedure and
611 prevent any later plugins from running.
613 * Journald's SplitMode=login setting has been deprecated. It has been
614 removed from documentation, and its use is discouraged. In a future
615 release it will be completely removed, and made equivalent to current
616 default of SplitMode=uid.
618 * Storage=both option setting in /etc/systemd/coredump.conf has been
619 removed. With fast LZ4 compression storing the core dump twice is not
622 * The --share-system systemd-nspawn option has been replaced with an
623 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
624 this functionality is discouraged. In addition the variables
625 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
626 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
627 individual namespaces.
629 * "machinectl list" now shows the IP address of running containers in
630 the output, as well as OS release information.
632 * "loginctl list" now shows the TTY of each session in the output.
634 * sd-bus gained new API calls sd_bus_track_set_recursive(),
635 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
636 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
637 tracking objects in a "recursive" mode, where a single client can be
638 counted multiple times, if it takes multiple references.
640 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
641 sd_bus_get_exit_on_disconnect(). They may be used to to make a
642 process using sd-bus automatically exit if the bus connection is
645 * Bus clients of the service manager may now "pin" loaded units into
646 memory, by taking an explicit reference on them. This is useful to
647 ensure the client can retrieve runtime data about the service even
648 after the service completed execution. Taking such a reference is
649 available only for privileged clients and should be helpful to watch
650 running services in a race-free manner, and in particular collect
651 information about exit statuses and results.
653 * The nss-resolve module has been changed to strictly return UNAVAIL
654 when communication via D-Bus with resolved failed, and NOTFOUND when
655 a lookup completed but was negative. This means it is now possible to
656 neatly configure fallbacks using nsswitch.conf result checking
657 expressions. Taking benefit of this, the new recommended
658 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
660 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
662 * A new setting CtrlAltDelBurstAction= has been added to
663 /etc/systemd/system.conf which may be used to configure the precise
664 behaviour if the user on the console presses Ctrl-Alt-Del more often
665 than 7 times in 2s. Previously this would unconditionally result in
666 an expedited, immediate reboot. With this new setting the precise
667 operation may be configured in more detail, and also turned off
670 * In .netdev files two new settings RemoteChecksumTx= and
671 RemoteChecksumRx= are now understood that permit configuring the
672 remote checksumming logic for VXLAN networks.
674 * The service manager learnt a new "invocation ID" concept for invoked
675 services. Each runtime cycle of a service will get a new invocation
676 ID (a 128bit random UUID) assigned that identifies the current
677 run of the service uniquely and globally. A new invocation ID
678 is generated each time a service starts up. The journal will store
679 the invocation ID of a service along with any logged messages, thus
680 making the invocation ID useful for matching the online runtime of a
681 service with the offline log data it generated in a safe way without
682 relying on synchronized timestamps. In many ways this new service
683 invocation ID concept is similar to the kernel's boot ID concept that
684 uniquely and globally identifies the runtime of each boot. The
685 invocation ID of a service is passed to the service itself via an
686 environment variable ($INVOCATION_ID). A new bus call
687 GetUnitByInvocationID() has been added that is similar to GetUnit()
688 but instead of retrieving the bus path for a unit by its name
689 retrieves it by its invocation ID. The returned path is valid only as
690 long as the passed invocation ID is current.
692 * systemd-resolved gained a new "DNSStubListener" setting in
693 resolved.conf. It either takes a boolean value or the special values
694 "udp" and "tcp", and configures whether to enable the stub DNS
695 listener on 127.0.0.53:53.
697 * IP addresses configured via networkd may now carry additional
698 configuration settings supported by the kernel. New options include:
699 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
700 PrefixRoute=, AutoJoin=.
702 * The PAM configuration fragment file for "user@.service" shipped with
703 systemd (i.e. the --user instance of systemd) has been stripped to
704 the minimum necessary to make the system boot. Previously, it
705 contained Fedora-specific stanzas that did not apply to other
706 distributions. It is expected that downstream distributions add
707 additional configuration lines, matching their needs to this file,
708 using it only as rough template of what systemd itself needs. Note
709 that this reduced fragment does not even include an invocation of
710 pam_limits which most distributions probably want to add, even though
711 systemd itself does not need it. (There's also the new build time
712 option --with-pamconfdir=no to disable installation of the PAM
715 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
716 capability is now also dropped from its set (in addition to
717 CAP_SYS_MKNOD as before).
719 * In service unit files it is now possible to connect a specific named
720 file descriptor with stdin/stdout/stdout of an executed service. The
721 name may be specified in matching .socket units using the
722 FileDescriptorName= setting.
724 * A number of journal settings may now be configured on the kernel
725 command line. Specifically, the following options are now understood:
726 systemd.journald.max_level_console=,
727 systemd.journald.max_level_store=,
728 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
729 systemd.journald.max_level_wall=.
731 * "systemctl is-enabled --full" will now show by which symlinks a unit
732 file is enabled in the unit dependency tree.
734 * Support for VeraCrypt encrypted partitions has been added to the
735 "cryptsetup" logic and /etc/crypttab.
737 * systemd-detect-virt gained support for a new --private-users switch
738 that checks whether the invoking processes are running inside a user
739 namespace. Similar, a new special value "private-users" for the
740 existing ConditionVirtualization= setting has been added, permitting
741 skipping of specific units in user namespace environments.
743 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
744 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
745 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
746 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
747 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
748 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
749 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
750 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
751 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
752 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
753 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
754 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
755 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
756 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
757 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
758 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
759 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
760 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
761 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
762 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
763 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
764 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
765 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
766 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
767 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
768 Jędrzejewski-Szmek, Zeal Jagannatha
770 — Santa Fe, 2016-11-03
774 * In service units the various ExecXYZ= settings have been extended
775 with an additional special character as first argument of the
776 assigned value: if the character '+' is used the specified command
777 line it will be run with full privileges, regardless of User=,
778 Group=, CapabilityBoundingSet= and similar options. The effect is
779 similar to the existing PermissionsStartOnly= option, but allows
780 configuration of this concept for each executed command line
783 * Services may now alter the service watchdog timeout at runtime by
784 sending a WATCHDOG_USEC= message via sd_notify().
786 * MemoryLimit= and related unit settings now optionally take percentage
787 specifications. The percentage is taken relative to the amount of
788 physical memory in the system (or in case of containers, the assigned
789 amount of memory). This allows scaling service resources neatly with
790 the amount of RAM available on the system. Similarly, systemd-logind's
791 RuntimeDirectorySize= option now also optionally takes percentage
794 * In similar fashion TasksMax= takes percentage values now, too. The
795 value is taken relative to the configured maximum number of processes
796 on the system. The per-service task maximum has been changed to 15%
797 using this functionality. (Effectively this is an increase of 512 →
798 4915 for service units, given the kernel's default pid_max setting.)
800 * Calendar time specifications in .timer units now understand a ".."
801 syntax for time ranges. Example: "4..7:10" may now be used for
802 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
805 * The InaccessableDirectories=, ReadOnlyDirectories= and
806 ReadWriteDirectories= unit file settings have been renamed to
807 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
808 applied to all kinds of file nodes, and not just directories, with
809 the exception of symlinks. Specifically these settings may now be
810 used on block and character device nodes, UNIX sockets and FIFOS as
811 well as regular files. The old names of these settings remain
812 available for compatibility.
814 * systemd will now log about all service processes it kills forcibly
815 (using SIGKILL) because they remained after the clean shutdown phase
816 of the service completed. This should help identifying services that
817 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
818 systemd-logind's configuration a similar log message is generated for
819 processes killed at the end of each session due to this setting.
821 * systemd will now set the $JOURNAL_STREAM environment variable for all
822 services whose stdout/stderr are connected to the Journal (which
823 effectively means by default: all services). The variable contains
824 the device and inode number of the file descriptor used for
825 stdout/stderr. This may be used by invoked programs to detect whether
826 their stdout/stderr is connected to the Journal, in which case they
827 can switch over to direct Journal communication, thus being able to
828 pass extended, structured metadata along with their log messages. As
829 one example, this is now used by glib's logging primitives.
831 * When using systemd's default tmp.mount unit for /tmp, the mount point
832 will now be established with the "nosuid" and "nodev" options. This
833 avoids privilege escalation attacks that put traps and exploits into
834 /tmp. However, this might cause problems if you e. g. put container
835 images or overlays into /tmp; if you need this, override tmp.mount's
836 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
839 * systemd now supports the "memory" cgroup controller also on
842 * The systemd-cgtop tool now optionally takes a control group path as
843 command line argument. If specified, the control group list shown is
844 limited to subgroups of that group.
846 * The SystemCallFilter= unit file setting gained support for
847 pre-defined, named system call filter sets. For example
848 SystemCallFilter=@clock is now an effective way to make all clock
849 changing-related system calls unavailable to a service. A number of
850 similar pre-defined groups are defined. Writing system call filters
851 for system services is simplified substantially with this new
852 concept. Accordingly, all of systemd's own, long-running services now
853 enable system call filtering based on this, by default.
855 * A new service setting MemoryDenyWriteExecute= has been added, taking
856 a boolean value. If turned on, a service may no longer create memory
857 mappings that are writable and executable at the same time. This
858 enhances security for services where this is enabled as it becomes
859 harder to dynamically write and then execute memory in exploited
860 service processes. This option has been enabled for all of systemd's
861 own long-running services.
863 * A new RestrictRealtime= service setting has been added, taking a
864 boolean argument. If set the service's processes may no longer
865 acquire realtime scheduling. This improves security as realtime
866 scheduling may otherwise be used to easily freeze the system.
868 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
869 value. This may be used for requesting that the system manager inside
870 of the container reports start-up completion to nspawn which then
871 propagates this notification further to the service manager
872 supervising nspawn itself. A related option NotifyReady= in .nspawn
873 files has been added too. This functionality allows ordering of the
874 start-up of multiple containers using the usual systemd ordering
877 * machinectl gained a new command "stop" that is an alias for
880 * systemd-resolved gained support for contacting DNS servers on
881 link-local IPv6 addresses.
883 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
884 its caches. A method call for requesting the same operation has been
885 added to the bus API too, and is made available via "systemd-resolve
888 * systemd-resolve gained a new --status switch. If passed a brief
889 summary of the used DNS configuration with per-interface information
892 * resolved.conf gained a new Cache= boolean option, defaulting to
893 on. If turned off local DNS caching is disabled. This comes with a
894 performance penalty in particular when DNSSEC is enabled. Note that
895 resolved disables its internal caching implicitly anyway, when the
896 configured DNS server is on a host-local IP address such as ::1 or
897 127.0.0.1, thus automatically avoiding double local caching.
899 * systemd-resolved now listens on the local IP address 127.0.0.53:53
900 for DNS requests. This improves compatibility with local programs
901 that do not use the libc NSS or systemd-resolved's bus APIs for name
902 resolution. This minimal DNS service is only available to local
903 programs and does not implement the full DNS protocol, but enough to
904 cover local DNS clients. A new, static resolv.conf file, listing just
905 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
906 now recommended to make /etc/resolv.conf a symlink to this file in
907 order to route all DNS lookups to systemd-resolved, regardless if
908 done via NSS, the bus API or raw DNS packets. Note that this local
909 DNS service is not as fully featured as the libc NSS or
910 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
911 used to deliver link-local address information (as this implies
912 sending a local interface index along), LLMNR/mDNS support via this
913 interface is severely restricted. It is thus strongly recommended for
914 all applications to use the libc NSS API or native systemd-resolved
917 * systemd-networkd's bridge support learned a new setting
918 VLANFiltering= for controlling VLAN filtering. Moreover a new section
919 in .network files has been added for configuring VLAN bridging in
920 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
922 * systemd-networkd's IPv6 Router Advertisement code now makes use of
923 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
924 now be acquired without relying on DHCPv6. Two new options
925 UseDomains= and UseDNS= have been added to configure this behaviour.
927 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
928 renamed IPv6AcceptRA=, without altering its behaviour. The old
929 setting name remains available for compatibility reasons.
931 * The systemd-networkd VTI/VTI6 tunneling support gained new options
932 Key=, InputKey= and OutputKey=.
934 * systemd-networkd gained support for VRF ("Virtual Routing Function")
935 interface configuration.
937 * "systemctl edit" may now be used to create new unit files by
938 specifying the --force switch.
940 * sd-event gained a new function sd_event_get_iteration() for
941 requesting the current iteration counter of the event loop. It starts
942 at zero and is increased by one with each event loop iteration.
944 * A new rpm macro %systemd_ordering is provided by the macros.systemd
945 file. It can be used in lieu of %systemd_requires in packages which
946 don't use any systemd functionality and are intended to be installed
947 in minimal containers without systemd present. This macro provides
948 ordering dependencies to ensure that if the package is installed in
949 the same rpm transaction as systemd, systemd will be installed before
950 the scriptlets for the package are executed, allowing unit presets
953 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
954 been added to simplify packaging of generators.
956 * The os-release file gained VERSION_CODENAME field for the
957 distribution nickname (e.g. VERSION_CODENAME=woody).
959 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
960 can be set to disable parsing of metadata and the creation
961 of persistent symlinks for that device.
963 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
964 to make them available to logged-in users has been reverted.
966 * Much of the common code of the various systemd components is now
967 built into an internal shared library libsystemd-shared-231.so
968 (incorporating the systemd version number in the name, to be updated
969 with future releases) that the components link to. This should
970 decrease systemd footprint both in memory during runtime and on
971 disk. Note that the shared library is not for public use, and is
972 neither API not ABI stable, but is likely to change with every new
973 released update. Packagers need to make sure that binaries
974 linking to libsystemd-shared.so are updated in step with the
977 * Configuration for "mkosi" is now part of the systemd
978 repository. mkosi is a tool to easily build legacy-free OS images,
979 and is available on github: https://github.com/systemd/mkosi. If
980 "mkosi" is invoked in the build tree a new raw OS image is generated
981 incorporating the systemd sources currently being worked on and a
982 clean, fresh distribution installation. The generated OS image may be
983 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
984 UEFI PC. This functionality is particularly useful to easily test
985 local changes made to systemd in a pristine, defined environment. See
988 * configure learned the --with-support-url= option to specify the
989 distribution's bugtracker.
991 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
992 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
993 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
994 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
995 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
996 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
997 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
998 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
999 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
1000 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
1001 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
1002 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
1003 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
1004 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
1005 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
1006 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
1007 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
1008 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
1009 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
1011 — Berlin, 2016-07-25
1015 * DNSSEC is now turned on by default in systemd-resolved (in
1016 "allow-downgrade" mode), but may be turned off during compile time by
1017 passing "--with-default-dnssec=no" to "configure" (and of course,
1018 during runtime with DNSSEC= in resolved.conf). We recommend
1019 downstreams to leave this on at least during development cycles and
1020 report any issues with the DNSSEC logic upstream. We are very
1021 interested in collecting feedback about the DNSSEC validator and its
1022 limitations in the wild. Note however, that DNSSEC support is
1023 probably nothing downstreams should turn on in stable distros just
1024 yet, as it might create incompatibilities with a few DNS servers and
1025 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
1026 automatically whenever we detect such incompatible setups, but there
1027 might be systems we do not cover yet. Hence: please help us testing
1028 the DNSSEC code, leave this on where you can, report back, but then
1029 again don't consider turning this on in your stable, LTS or
1030 production release just yet. (Note that you have to enable
1031 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
1032 and its DNSSEC mode for host name resolution from local
1035 * systemd-resolve conveniently resolves DANE records with the --tlsa
1036 option and OPENPGPKEY records with the --openpgp option. It also
1037 supports dumping raw DNS record data via the new --raw= switch.
1039 * systemd-logind will now by default terminate user processes that are
1040 part of the user session scope unit (session-XX.scope) when the user
1041 logs out. This behavior is controlled by the KillUserProcesses=
1042 setting in logind.conf, and the previous default of "no" is now
1043 changed to "yes". This means that user sessions will be properly
1044 cleaned up after, but additional steps are necessary to allow
1045 intentionally long-running processes to survive logout.
1047 While the user is logged in at least once, user@.service is running,
1048 and any service that should survive the end of any individual login
1049 session can be started at a user service or scope using systemd-run.
1050 systemd-run(1) man page has been extended with an example which shows
1051 how to run screen in a scope unit underneath user@.service. The same
1052 command works for tmux.
1054 After the user logs out of all sessions, user@.service will be
1055 terminated too, by default, unless the user has "lingering" enabled.
1056 To effectively allow users to run long-term tasks even if they are
1057 logged out, lingering must be enabled for them. See loginctl(1) for
1058 details. The default polkit policy was modified to allow users to
1059 set lingering for themselves without authentication.
1061 Previous defaults can be restored at compile time by the
1062 --without-kill-user-processes option to "configure".
1064 * systemd-logind gained new configuration settings SessionsMax= and
1065 InhibitorsMax=, both with a default of 8192. It will not register new
1066 user sessions or inhibitors above this limit.
1068 * systemd-logind will now reload configuration on SIGHUP.
1070 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
1071 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
1072 enable. Also, support for the "io" cgroup controller in the unified
1073 hierarchy has been added, so that the "memory", "pids" and "io" are
1074 now the controllers that are supported on the unified hierarchy.
1076 WARNING: it is not possible to use previous systemd versions with
1077 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
1078 is necessary to also update systemd in the initramfs if using the
1079 unified hierarchy. An updated SELinux policy is also required.
1081 * LLDP support has been extended, and both passive (receive-only) and
1082 active (sender) modes are supported. Passive mode ("routers-only") is
1083 enabled by default in systemd-networkd. Active LLDP mode is enabled
1084 by default for containers on the internal network. The "networkctl
1085 lldp" command may be used to list information gathered. "networkctl
1086 status" will also show basic LLDP information on connected peers now.
1088 * The IAID and DUID unique identifier sent in DHCP requests may now be
1089 configured for the system and each .network file managed by
1090 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
1092 * systemd-networkd gained support for configuring proxy ARP support for
1093 each interface, via the ProxyArp= setting in .network files. It also
1094 gained support for configuring the multicast querier feature of
1095 bridge devices, via the new MulticastQuerier= setting in .netdev
1096 files. Similarly, snooping on the IGMP traffic can be controlled
1097 via the new setting MulticastSnooping=.
1099 A new setting PreferredLifetime= has been added for addresses
1100 configured in .network file to configure the lifetime intended for an
1103 The systemd-networkd DHCP server gained the option EmitRouter=, which
1104 defaults to yes, to configure whether the DHCP Option 3 (Router)
1107 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
1108 systemd-socket-activate and installed into /usr/bin. It is now fully
1111 * systemd-journald now uses separate threads to flush changes to disk
1112 when closing journal files, thus reducing impact of slow disk I/O on
1113 logging performance.
1115 * The sd-journal API gained two new calls
1116 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
1117 can be used to open journal files using file descriptors instead of
1118 file or directory paths. sd_journal_open_container() has been
1119 deprecated, sd_journal_open_directory_fd() should be used instead
1120 with the flag SD_JOURNAL_OS_ROOT.
1122 * journalctl learned a new output mode "-o short-unix" that outputs log
1123 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
1124 UTC). It also gained support for a new --no-hostname setting to
1125 suppress the hostname column in the family of "short" output modes.
1127 * systemd-ask-password now optionally skips printing of the password to
1128 stdout with --no-output which can be useful in scripts.
1130 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
1131 (devices tagged with ID_MAKER_TOOL) are now tagged with
1132 "uaccess" and are available to logged in users.
1134 * The DeviceAllow= unit setting now supports specifiers (with "%").
1136 * "systemctl show" gained a new --value switch, which allows print a
1137 only the contents of a specific unit property, without also printing
1138 the property's name. Similar support was added to "show*" verbs
1139 of loginctl and machinectl that output "key=value" lists.
1141 * A new unit type "generated" was added for files dynamically generated
1142 by generator tools. Similarly, a new unit type "transient" is used
1143 for unit files created using the runtime API. "systemctl enable" will
1144 refuse to operate on such files.
1146 * A new command "systemctl revert" has been added that may be used to
1147 revert to the vendor version of a unit file, in case local changes
1148 have been made by adding drop-ins or overriding the unit file.
1150 * "machinectl clean" gained a new verb to automatically remove all or
1151 just hidden container images.
1153 * systemd-tmpfiles gained support for a new line type "e" for emptying
1154 directories, if they exist, without creating them if they don't.
1156 * systemd-nspawn gained support for automatically patching the UID/GIDs
1157 of the owners and the ACLs of all files and directories in a
1158 container tree to match the UID/GID user namespacing range selected
1159 for the container invocation. This mode is enabled via the new
1160 --private-users-chown switch. It also gained support for
1161 automatically choosing a free, previously unused UID/GID range when
1162 starting a container, via the new --private-users=pick setting (which
1163 implies --private-users-chown). Together, these options for the first
1164 time make user namespacing for nspawn containers fully automatic and
1165 thus deployable. The systemd-nspawn@.service template unit file has
1166 been changed to use this functionality by default.
1168 * systemd-nspawn gained a new --network-zone= switch, that allows
1169 creating ad-hoc virtual Ethernet links between multiple containers,
1170 that only exist as long as at least one container referencing them is
1171 running. This allows easy connecting of multiple containers with a
1172 common link that implements an Ethernet broadcast domain. Each of
1173 these network "zones" may be named relatively freely by the user, and
1174 may be referenced by any number of containers, but each container may
1175 only reference one of these "zones". On the lower level, this is
1176 implemented by an automatically managed bridge network interface for
1177 each zone, that is created when the first container referencing its
1178 zone is created and removed when the last one referencing its zone
1181 * The default start timeout may now be configured on the kernel command
1182 line via systemd.default_timeout_start_sec=. It was already
1183 configurable via the DefaultTimeoutStartSec= option in
1184 /etc/systemd/system.conf.
1186 * Socket units gained a new TriggerLimitIntervalSec= and
1187 TriggerLimitBurst= setting to configure a limit on the activation
1188 rate of the socket unit.
1190 * The LimitNICE= setting now optionally takes normal UNIX nice values
1191 in addition to the raw integer limit value. If the specified
1192 parameter is prefixed with "+" or "-" and is in the range -20..19 the
1193 value is understood as UNIX nice value. If not prefixed like this it
1194 is understood as raw RLIMIT_NICE limit.
1196 * Note that the effect of the PrivateDevices= unit file setting changed
1197 slightly with this release: the per-device /dev file system will be
1198 mounted read-only from this version on, and will have "noexec"
1199 set. This (minor) change of behavior might cause some (exceptional)
1200 legacy software to break, when PrivateDevices=yes is set for its
1201 service. Please leave PrivateDevices= off if you run into problems
1204 * systemd-bootchart has been split out to a separate repository:
1205 https://github.com/systemd/systemd-bootchart
1207 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
1208 merged into the kernel in its current form.
1210 * The compatibility libraries libsystemd-daemon.so,
1211 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
1212 which have been deprecated since systemd-209 have been removed along
1213 with the corresponding pkg-config files. All symbols provided by
1214 those libraries are provided by libsystemd.so.
1216 * The Capabilities= unit file setting has been removed (it is ignored
1217 for backwards compatibility). AmbientCapabilities= and
1218 CapabilityBoundingSet= should be used instead.
1220 * A new special target has been added, initrd-root-device.target,
1221 which creates a synchronization point for dependencies of the root
1222 device in early userspace. Initramfs builders must ensure that this
1223 target is now included in early userspace.
1225 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
1226 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
1227 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
1228 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
1229 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
1230 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
1231 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
1232 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
1233 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
1234 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
1235 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
1236 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
1237 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
1238 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
1239 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
1240 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
1241 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
1242 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
1243 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
1244 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
1245 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
1246 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
1247 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
1248 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
1249 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
1252 — Fairfax, 2016-05-21
1256 * The systemd-resolved DNS resolver service has gained a substantial
1257 set of new features, most prominently it may now act as a DNSSEC
1258 validating stub resolver. DNSSEC mode is currently turned off by
1259 default, but is expected to be turned on by default in one of the
1260 next releases. For now, we invite everybody to test the DNSSEC logic
1261 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
1262 service also gained a full set of D-Bus interfaces, including calls
1263 to configure DNS and DNSSEC settings per link (for use by external
1264 network management software). systemd-resolved and systemd-networkd
1265 now distinguish between "search" and "routing" domains. The former
1266 are used to qualify single-label names, the latter are used purely
1267 for routing lookups within certain domains to specific links.
1268 resolved now also synthesizes RRs for all entries from /etc/hosts.
1270 * The systemd-resolve tool (which is a client utility for
1271 systemd-resolved) has been improved considerably and is now fully
1272 supported and documented. Hence it has moved from /usr/lib/systemd to
1275 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
1278 * The coredump collection logic has been reworked: when a coredump is
1279 collected it is now written to disk, compressed and processed
1280 (including stacktrace extraction) from a new instantiated service
1281 systemd-coredump@.service, instead of directly from the
1282 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
1283 processing large coredumps can take up a substantial amount of
1284 resources and time, and this previously happened entirely outside of
1285 systemd's service supervision. With the new logic the core_pattern
1286 hook only does minimal metadata collection before passing off control
1287 to the new instantiated service, which is configured with a time
1288 limit, a nice level and other settings to minimize negative impact on
1289 the rest of the system. Also note that the new logic will honour the
1290 RLIMIT_CORE setting of the crashed process, which now allows users
1291 and processes to turn off coredumping for their processes by setting
1294 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
1295 and all forked processes by default. Previously, PID 1 would leave
1296 the setting at "0" for all processes, as set by the kernel. Note that
1297 the resource limit traditionally has no effect on the generated
1298 coredumps on the system if the /proc/sys/kernel/core_pattern hook
1299 logic is used. Since the limit is now honoured (see above) its
1300 default has been changed so that the coredumping logic is enabled by
1301 default for all processes, while allowing specific opt-out.
1303 * When the stacktrace is extracted from processes of system users, this
1304 is now done as "systemd-coredump" user, in order to sandbox this
1305 potentially security sensitive parsing operation. (Note that when
1306 processing coredumps of normal users this is done under the user ID
1307 of process that crashed, as before.) Packagers should take notice
1308 that it is now necessary to create the "systemd-coredump" system user
1309 and group at package installation time.
1311 * The systemd-activate socket activation testing tool gained support
1312 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
1313 and --seqpacket switches. It also has been extended to support both
1314 new-style and inetd-style file descriptor passing. Use the new
1315 --inetd switch to request inetd-style file descriptor passing.
1317 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
1318 variable, which takes a boolean value. If set to false, ANSI color
1319 output is disabled in the tools even when run on a terminal that
1322 * The VXLAN support in networkd now supports two new settings
1323 DestinationPort= and PortRange=.
1325 * A new systemd.machine_id= kernel command line switch has been added,
1326 that may be used to set the machine ID in /etc/machine-id if it is
1327 not initialized yet. This command line option has no effect if the
1328 file is already initialized.
1330 * systemd-nspawn gained a new --as-pid2 switch that invokes any
1331 specified command line as PID 2 rather than PID 1 in the
1332 container. In this mode PID 1 is a minimal stub init process that
1333 implements the special POSIX and Linux semantics of PID 1 regarding
1334 signal and child process management. Note that this stub init process
1335 is implemented in nspawn itself and requires no support from the
1336 container image. This new logic is useful to support running
1337 arbitrary commands in the container, as normal processes are
1338 generally not prepared to run as PID 1.
1340 * systemd-nspawn gained a new --chdir= switch for setting the current
1341 working directory for the process started in the container.
1343 * "journalctl /dev/sda" will now output all kernel log messages for
1344 specified device from the current boot, in addition to all devices
1345 that are parents of it. This should make log output about devices
1346 pretty useful, as long as kernel drivers attach enough metadata to
1347 the log messages. (The usual SATA drivers do.)
1349 * The sd-journal API gained two new calls
1350 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
1351 that report whether log data from /run or /var has been found.
1353 * journalctl gained a new switch "--fields" that prints all journal
1354 record field names currently in use in the journal. This is backed
1355 by two new sd-journal API calls sd_journal_enumerate_fields() and
1356 sd_journal_restart_fields().
1358 * Most configurable timeouts in systemd now expect an argument of
1359 "infinity" to turn them off, instead of "0" as before. The semantics
1360 from now on is that a timeout of "0" means "now", and "infinity"
1361 means "never". To maintain backwards compatibility, "0" continues to
1362 turn off previously existing timeout settings.
1364 * "systemctl reload-or-try-restart" has been renamed to "systemctl
1365 try-reload-or-restart" to clarify what it actually does: the "try"
1366 logic applies to both reloading and restarting, not just restarting.
1367 The old name continues to be accepted for compatibility.
1369 * On boot-up, when PID 1 detects that the system clock is behind the
1370 release date of the systemd version in use, the clock is now set
1371 to the latter. Previously, this was already done in timesyncd, in order
1372 to avoid running with clocks set to the various clock epochs such as
1373 1902, 1938 or 1970. With this change the logic is now done in PID 1
1374 in addition to timesyncd during early boot-up, so that it is enforced
1375 before the first process is spawned by systemd. Note that the logic
1376 in timesyncd remains, as it is more comprehensive and ensures
1377 clock monotonicity by maintaining a persistent timestamp file in
1378 /var. Since /var is generally not available in earliest boot or the
1379 initrd, this part of the logic remains in timesyncd, and is not done
1382 * Support for tweaking details in net_cls.class_id through the
1383 NetClass= configuration directive has been removed, as the kernel
1384 people have decided to deprecate that controller in cgroup v2.
1385 Userspace tools such as nftables are moving over to setting rules
1386 that are specific to the full cgroup path of a task, which obsoletes
1387 these controllers anyway. The NetClass= directive is kept around for
1388 legacy compatibility reasons. For a more in-depth description of the
1389 kernel change, please refer to the respective upstream commit:
1391 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
1393 * A new service setting RuntimeMaxSec= has been added that may be used
1394 to specify a maximum runtime for a service. If the timeout is hit, the
1395 service is terminated and put into a failure state.
1397 * A new service setting AmbientCapabilities= has been added. It allows
1398 configuration of additional Linux process capabilities that are
1399 passed to the activated processes. This is only available on very
1402 * The process resource limit settings in service units may now be used
1403 to configure hard and soft limits individually.
1405 * The various libsystemd APIs such as sd-bus or sd-event now publicly
1406 expose support for gcc's __attribute__((cleanup())) C extension.
1407 Specifically, for many object destructor functions alternative
1408 versions have been added that have names suffixed with "p" and take a
1409 pointer to a pointer to the object to destroy, instead of just a
1410 pointer to the object itself. This is useful because these destructor
1411 functions may be used directly as parameters to the cleanup
1412 construct. Internally, systemd has been a heavy user of this GCC
1413 extension for a long time, and with this change similar support is
1414 now available to consumers of the library outside of systemd. Note
1415 that by using this extension in your sources compatibility with old
1416 and strictly ANSI compatible C compilers is lost. However, all gcc or
1417 LLVM versions of recent years support this extension.
1419 * Timer units gained support for a new setting RandomizedDelaySec= that
1420 allows configuring some additional randomized delay to the configured
1421 time. This is useful to spread out timer events to avoid load peaks in
1422 clusters or larger setups.
1424 * Calendar time specifications now support sub-second accuracy.
1426 * Socket units now support listening on SCTP and UDP-lite protocol
1429 * The sd-event API now comes with a full set of man pages.
1431 * Older versions of systemd contained experimental support for
1432 compressing journal files and coredumps with the LZ4 compressor that
1433 was not compatible with the lz4 binary (due to API limitations of the
1434 lz4 library). This support has been removed; only support for files
1435 compatible with the lz4 binary remains. This LZ4 logic is now
1436 officially supported and no longer considered experimental.
1438 * The dkr image import logic has been removed again from importd. dkr's
1439 micro-services focus doesn't fit into the machine image focus of
1440 importd, and quickly got out of date with the upstream dkr API.
1442 * Creation of the /run/lock/lockdev/ directory was dropped from
1443 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
1444 been available for many years. If you still need this, you need to
1445 create your own tmpfiles.d config file with:
1447 d /run/lock/lockdev 0775 root lock -
1449 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
1450 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
1451 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
1452 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
1453 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
1454 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
1455 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
1456 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
1457 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
1458 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
1459 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
1460 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
1461 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
1462 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
1463 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
1464 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
1465 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
1466 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
1467 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
1469 — Berlin, 2016-02-11
1473 * A number of properties previously only settable in unit
1474 files are now also available as properties to set when
1475 creating transient units programmatically via the bus, as it
1476 is exposed with systemd-run's --property=
1477 setting. Specifically, these are: SyslogIdentifier=,
1478 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
1479 EnvironmentFile=, ReadWriteDirectories=,
1480 ReadOnlyDirectories=, InaccessibleDirectories=,
1481 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
1483 * When creating transient services via the bus API it is now
1484 possible to pass in a set of file descriptors to use as
1485 STDIN/STDOUT/STDERR for the invoked process.
1487 * Slice units may now be created transiently via the bus APIs,
1488 similar to the way service and scope units may already be
1489 created transiently.
1491 * Wherever systemd expects a calendar timestamp specification
1492 (like in journalctl's --since= and --until= switches) UTC
1493 timestamps are now supported. Timestamps suffixed with "UTC"
1494 are now considered to be in Universal Time Coordinated
1495 instead of the local timezone. Also, timestamps may now
1496 optionally be specified with sub-second accuracy. Both of
1497 these additions also apply to recurring calendar event
1498 specification, such as OnCalendar= in timer units.
1500 * journalctl gained a new "--sync" switch that asks the
1501 journal daemon to write all so far unwritten log messages to
1502 disk and sync the files, before returning.
1504 * systemd-tmpfiles learned two new line types "q" and "Q" that
1505 operate like "v", but also set up a basic btrfs quota
1506 hierarchy when used on a btrfs file system with quota
1509 * tmpfiles' "v", "q" and "Q" will now create a plain directory
1510 instead of a subvolume (even on a btrfs file system) if the
1511 root directory is a plain directory, and not a
1512 subvolume. This should simplify things with certain chroot()
1513 environments which are not aware of the concept of btrfs
1516 * systemd-detect-virt gained a new --chroot switch to detect
1517 whether execution takes place in a chroot() environment.
1519 * CPUAffinity= now takes CPU index ranges in addition to
1522 * The various memory-related resource limit settings (such as
1523 LimitAS=) now understand the usual K, M, G, ... suffixes to
1524 the base of 1024 (IEC). Similar, the time-related resource
1525 limit settings understand the usual min, h, day, ...
1528 * There's a new system.conf setting DefaultTasksMax= to
1529 control the default TasksMax= setting for services and
1530 scopes running on the system. (TasksMax= is the primary
1531 setting that exposes the "pids" cgroup controller on systemd
1532 and was introduced in the previous systemd release.) The
1533 setting now defaults to 512, which means services that are
1534 not explicitly configured otherwise will only be able to
1535 create 512 processes or threads at maximum, from this
1536 version on. Note that this means that thread- or
1537 process-heavy services might need to be reconfigured to set
1538 TasksMax= to a higher value. It is sufficient to set
1539 TasksMax= in these specific unit files to a higher value, or
1540 even "infinity". Similar, there's now a logind.conf setting
1541 UserTasksMax= that defaults to 4096 and limits the total
1542 number of processes or tasks each user may own
1543 concurrently. nspawn containers also have the TasksMax=
1544 value set by default now, to 8192. Note that all of this
1545 only has an effect if the "pids" cgroup controller is
1546 enabled in the kernel. The general benefit of these changes
1547 should be a more robust and safer system, that provides a
1548 certain amount of per-service fork() bomb protection.
1550 * systemd-nspawn gained the new --network-veth-extra= switch
1551 to define additional and arbitrarily-named virtual Ethernet
1552 links between the host and the container.
1554 * A new service execution setting PassEnvironment= has been
1555 added that allows importing select environment variables
1556 from PID1's environment block into the environment block of
1559 * Timer units gained support for a new RemainAfterElapse=
1560 setting which takes a boolean argument. It defaults to on,
1561 exposing behaviour unchanged to previous releases. If set to
1562 off, timer units are unloaded after they elapsed if they
1563 cannot elapse again. This is particularly useful for
1564 transient timer units, which shall not stay around longer
1565 than until they first elapse.
1567 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
1568 default now (the kernel default is 16). This is beneficial
1569 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
1570 allows substantially larger numbers of queued
1571 datagrams. This should increase the capability of systemd to
1572 parallelize boot-up, as logging and sd_notify() are unlikely
1573 to stall execution anymore. If you need to change the value
1574 from the new defaults, use the usual sysctl.d/ snippets.
1576 * The compression framing format used by the journal or
1577 coredump processing has changed to be in line with what the
1578 official LZ4 tools generate. LZ4 compression support in
1579 systemd was considered unsupported previously, as the format
1580 was not compatible with the normal tools. With this release
1581 this has changed now, and it is hence safe for downstream
1582 distributions to turn it on. While not compressing as well
1583 as the XZ, LZ4 is substantially faster, which makes
1584 it a good default choice for the compression logic in the
1585 journal and in coredump handling.
1587 * Any reference to /etc/mtab has been dropped from
1588 systemd. The file has been obsolete since a while, but
1589 systemd refused to work on systems where it was incorrectly
1590 set up (it should be a symlink or non-existent). Please make
1591 sure to update to util-linux 2.27.1 or newer in conjunction
1592 with this systemd release, which also drops any reference to
1593 /etc/mtab. If you maintain a distribution make sure that no
1594 software you package still references it, as this is a
1595 likely source of bugs. There's also a glibc bug pending,
1596 asking for removal of any reference to this obsolete file:
1598 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
1600 Note that only util-linux versions built with
1601 --enable-libmount-force-mountinfo are supported.
1603 * Support for the ".snapshot" unit type has been removed. This
1604 feature turned out to be little useful and little used, and
1605 has now been removed from the core and from systemctl.
1607 * The dependency types RequiresOverridable= and
1608 RequisiteOverridable= have been removed from systemd. They
1609 have been used only very sparingly to our knowledge and
1610 other options that provide a similar effect (such as
1611 systemctl --mode=ignore-dependencies) are much more useful
1612 and commonly used. Moreover, they were only half-way
1613 implemented as the option to control behaviour regarding
1614 these dependencies was never added to systemctl. By removing
1615 these dependency types the execution engine becomes a bit
1616 simpler. Unit files that use these dependencies should be
1617 changed to use the non-Overridable dependency types
1618 instead. In fact, when parsing unit files with these
1619 options, that's what systemd will automatically convert them
1620 too, but it will also warn, asking users to fix the unit
1621 files accordingly. Removal of these dependency types should
1622 only affect a negligible number of unit files in the wild.
1624 * Behaviour of networkd's IPForward= option changed
1625 (again). It will no longer maintain a per-interface setting,
1626 but propagate one way from interfaces where this is enabled
1627 to the global kernel setting. The global setting will be
1628 enabled when requested by a network that is set up, but
1629 never be disabled again. This change was made to make sure
1630 IPv4 and IPv6 behaviour regarding packet forwarding is
1631 similar (as the Linux IPv6 stack does not support
1632 per-interface control of this setting) and to minimize
1635 * In unit files the behaviour of %u, %U, %h, %s has
1636 changed. These specifiers will now unconditionally resolve
1637 to the various user database fields of the user that the
1638 systemd instance is running as, instead of the user
1639 configured in the specific unit via User=. Note that this
1640 effectively doesn't change much, as resolving of these
1641 specifiers was already turned off in the --system instance
1642 of systemd, as we cannot do NSS lookups from PID 1. In the
1643 --user instance of systemd these specifiers where correctly
1644 resolved, but hardly made any sense, since the user instance
1645 lacks privileges to do user switches anyway, and User= is
1646 hence useless. Moreover, even in the --user instance of
1647 systemd behaviour was awkward as it would only take settings
1648 from User= assignment placed before the specifier into
1649 account. In order to unify and simplify the logic around
1650 this the specifiers will now always resolve to the
1651 credentials of the user invoking the manager (which in case
1652 of PID 1 is the root user).
1654 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
1655 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
1656 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
1657 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
1658 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
1659 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
1660 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
1661 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
1662 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
1663 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
1664 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
1665 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
1666 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
1667 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
1670 — Berlin, 2015-11-18
1674 * systemd now depends on util-linux v2.27. More specifically,
1675 the newly added mount monitor feature in libmount now
1676 replaces systemd's former own implementation.
1678 * libmount mandates /etc/mtab not to be regular file, and
1679 systemd now enforces this condition at early boot.
1680 /etc/mtab has been deprecated and warned about for a very
1681 long time, so systems running systemd should already have
1682 stopped having this file around as anything else than a
1683 symlink to /proc/self/mounts.
1685 * Support for the "pids" cgroup controller has been added. It
1686 allows accounting the number of tasks in a cgroup and
1687 enforcing limits on it. This adds two new setting
1688 TasksAccounting= and TasksMax= to each unit, as well as a
1689 global option DefaultTasksAccounting=.
1691 * Support for the "net_cls" cgroup controller has been added.
1692 It allows assigning a net class ID to each task in the
1693 cgroup, which can then be used in firewall rules and traffic
1694 shaping configurations. Note that the kernel netfilter net
1695 class code does not currently work reliably for ingress
1696 packets on unestablished sockets.
1698 This adds a new config directive called NetClass= to CGroup
1699 enabled units. Allowed values are positive numbers for fixed
1700 assignments and "auto" for picking a free value
1703 * 'systemctl is-system-running' now returns 'offline' if the
1704 system is not booted with systemd. This command can now be
1705 used as a substitute for 'systemd-notify --booted'.
1707 * Watchdog timeouts have been increased to 3 minutes for all
1708 in-tree service files. Apparently, disk IO issues are more
1709 frequent than we hoped, and user reported >1 minute waiting
1712 * 'machine-id-commit' functionality has been merged into
1713 'machine-id-setup --commit'. The separate binary has been
1716 * The WorkingDirectory= directive in unit files may now be set
1717 to the special value '~'. In this case, the working
1718 directory is set to the home directory of the user
1719 configured in User=.
1721 * "machinectl shell" will now open the shell in the home
1722 directory of the selected user by default.
1724 * The CrashChVT= configuration file setting is renamed to
1725 CrashChangeVT=, following our usual logic of not
1726 abbreviating unnecessarily. The old directive is still
1727 supported for compat reasons. Also, this directive now takes
1728 an integer value between 1 and 63, or a boolean value. The
1729 formerly supported '-1' value for disabling stays around for
1732 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
1733 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
1734 RootDirectory= properties can now be set for transient
1737 * The systemd-analyze tool gained a new "set-log-target" verb
1738 to change the logging target the system manager logs to
1739 dynamically during runtime. This is similar to how
1740 "systemd-analyze set-log-level" already changes the log
1743 * In nspawn /sys is now mounted as tmpfs, with only a selected
1744 set of subdirectories mounted in from the real sysfs. This
1745 enhances security slightly, and is useful for ensuring user
1746 namespaces work correctly.
1748 * Support for USB FunctionFS activation has been added. This
1749 allows implementation of USB gadget services that are
1750 activated as soon as they are requested, so that they don't
1751 have to run continuously, similar to classic socket
1754 * The "systemctl exit" command now optionally takes an
1755 additional parameter that sets the exit code to return from
1756 the systemd manager when exiting. This is only relevant when
1757 running the systemd user instance, or when running the
1758 system instance in a container.
1760 * sd-bus gained the new API calls sd_bus_path_encode_many()
1761 and sd_bus_path_decode_many() that allow easy encoding and
1762 decoding of multiple identifier strings inside a D-Bus
1763 object path. Another new call sd_bus_default_flush_close()
1764 has been added to flush and close per-thread default
1767 * systemd-cgtop gained support for a -M/--machine= switch to
1768 show the control groups within a certain container only.
1770 * "systemctl kill" gained support for an optional --fail
1771 switch. If specified the requested operation will fail of no
1772 processes have been killed, because the unit had no
1773 processes attached, or similar.
1775 * A new systemd.crash_reboot=1 kernel command line option has
1776 been added that triggers a reboot after crashing. This can
1777 also be set through CrashReboot= in systemd.conf.
1779 * The RuntimeDirectory= setting now understands unit
1780 specifiers like %i or %f.
1782 * A new (still internal) library API sd-ipv4acd has been added,
1783 that implements address conflict detection for IPv4. It's
1784 based on code from sd-ipv4ll, and will be useful for
1785 detecting DHCP address conflicts.
1787 * File descriptors passed during socket activation may now be
1788 named. A new API sd_listen_fds_with_names() is added to
1789 access the names. The default names may be overridden,
1790 either in the .socket file using the FileDescriptorName=
1791 parameter, or by passing FDNAME= when storing the file
1792 descriptors using sd_notify().
1794 * systemd-networkd gained support for:
1796 - Setting the IPv6 Router Advertisement settings via
1797 IPv6AcceptRouterAdvertisements= in .network files.
1799 - Configuring the HelloTimeSec=, MaxAgeSec= and
1800 ForwardDelaySec= bridge parameters in .netdev files.
1802 - Configuring PreferredSource= for static routes in
1805 * The "ask-password" framework used to query for LUKS harddisk
1806 passwords or SSL passwords during boot gained support for
1807 caching passwords in the kernel keyring, if it is
1808 available. This makes sure that the user only has to type in
1809 a passphrase once if there are multiple objects to unlock
1810 with the same one. Previously, such password caching was
1811 available only when Plymouth was used; this moves the
1812 caching logic into the systemd codebase itself. The
1813 "systemd-ask-password" utility gained a new --keyname=
1814 switch to control which kernel keyring key to use for
1815 caching a password in. This functionality is also useful for
1816 enabling display managers such as gdm to automatically
1817 unlock the user's GNOME keyring if its passphrase, the
1818 user's password and the harddisk password are the same, if
1819 gdm-autologin is used.
1821 * When downloading tar or raw images using "machinectl
1822 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
1823 file is now also downloaded, if it is available and stored
1824 next to the image file.
1826 * Units of type ".socket" gained a new boolean setting
1827 Writable= which is only useful in conjunction with
1828 ListenSpecial=. If true, enables opening the specified
1829 special file in O_RDWR mode rather than O_RDONLY mode.
1831 * systemd-rfkill has been reworked to become a singleton
1832 service that is activated through /dev/rfkill on each rfkill
1833 state change and saves the settings to disk. This way,
1834 systemd-rfkill is now compatible with devices that exist
1835 only intermittendly, and even restores state if the previous
1836 system shutdown was abrupt rather than clean.
1838 * The journal daemon gained support for vacuuming old journal
1839 files controlled by the number of files that shall remain,
1840 in addition to the already existing control by size and by
1841 date. This is useful as journal interleaving performance
1842 degrades with too many separate journal files, and allows
1843 putting an effective limit on them. The new setting defaults
1844 to 100, but this may be changed by setting SystemMaxFiles=
1845 and RuntimeMaxFiles= in journald.conf. Also, the
1846 "journalctl" tool gained the new --vacuum-files= switch to
1847 manually vacuum journal files to leave only the specified
1848 number of files in place.
1850 * udev will now create /dev/disk/by-path links for ATA devices
1851 on kernels where that is supported.
1853 * Galician, Serbian, Turkish and Korean translations were added.
1855 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
1856 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
1857 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
1858 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
1859 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
1860 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
1861 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
1862 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
1863 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
1864 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
1865 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
1866 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
1867 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
1868 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
1869 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
1870 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
1871 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
1872 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
1874 — Berlin, 2015-10-07
1878 * The DHCP implementation of systemd-networkd gained a set of
1881 - The DHCP server now supports emitting DNS and NTP
1882 information. It may be enabled and configured via
1883 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
1884 and NTP information is enabled, but no servers are
1885 configured, the corresponding uplink information (if there
1886 is any) is propagated.
1888 - Server and client now support transmission and reception
1889 of timezone information. It can be configured via the
1890 newly introduced network options UseTimezone=,
1891 EmitTimezone=, and Timezone=. Transmission of timezone
1892 information is enabled between host and containers by
1893 default now: the container will change its local timezone
1894 to what the host has set.
1896 - Lease timeouts can now be configured via
1897 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
1899 - The DHCP server improved on the stability of
1900 leases. Clients are more likely to get the same lease
1901 information back, even if the server loses state.
1903 - The DHCP server supports two new configuration options to
1904 control the lease address pool metrics, PoolOffset= and
1907 * The encapsulation limit of tunnels in systemd-networkd may
1908 now be configured via 'EncapsulationLimit='. It allows
1909 modifying the maximum additional levels of encapsulation
1910 that are permitted to be prepended to a packet.
1912 * systemd now supports the concept of user buses replacing
1913 session buses, if used with dbus-1.10 (and enabled via dbus
1914 --enable-user-session). It previously only supported this on
1915 kdbus-enabled systems, and this release expands this to
1916 'dbus-daemon' systems.
1918 * systemd-networkd now supports predictable interface names
1921 * systemd now optionally supports the new Linux kernel
1922 "unified" control group hierarchy. If enabled via the kernel
1923 command-line option 'systemd.unified_cgroup_hierarchy=1',
1924 systemd will try to mount the unified cgroup hierarchy
1925 directly on /sys/fs/cgroup. If not enabled, or not
1926 available, systemd will fall back to the legacy cgroup
1927 hierarchy setup, as before. Host system and containers can
1928 mix and match legacy and unified hierarchies as they
1929 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
1930 environment variable to individually select the hierarchy to
1931 use for executed containers. By default, nspawn will use the
1932 unified hierarchy for the containers if the host uses the
1933 unified hierarchy, and the legacy hierarchy otherwise.
1934 Please note that at this point the unified hierarchy is an
1935 experimental kernel feature and is likely to change in one
1936 of the next kernel releases. Therefore, it should not be
1937 enabled by default in downstream distributions yet. The
1938 minimum required kernel version for the unified hierarchy to
1939 work is 4.2. Note that when the unified hierarchy is used
1940 for the first time delegated access to controllers is
1941 safe. Because of this systemd-nspawn containers will get
1942 access to controllers now, as will systemd user
1943 sessions. This means containers and user sessions may now
1944 manage their own resources, partitioning up what the system
1947 * A new special scope unit "init.scope" has been introduced
1948 that encapsulates PID 1 of the system. It may be used to
1949 determine resource usage and enforce resource limits on PID
1950 1 itself. PID 1 hence moved out of the root of the control
1953 * The cgtop tool gained support for filtering out kernel
1954 threads when counting tasks in a control group. Also, the
1955 count of processes is now recursively summed up by
1956 default. Two options -k and --recursive= have been added to
1957 revert to old behaviour. The tool has also been updated to
1958 work correctly in containers now.
1960 * systemd-nspawn's --bind= and --bind-ro= options have been
1961 extended to allow creation of non-recursive bind mounts.
1963 * libsystemd gained two new calls sd_pid_get_cgroup() and
1964 sd_peer_get_cgroup() which return the control group path of
1965 a process or peer of a connected AF_UNIX socket. This
1966 function call is particularly useful when implementing
1967 delegated subtrees support in the control group hierarchy.
1969 * The "sd-event" event loop API of libsystemd now supports
1970 correct dequeuing of real-time signals, without losing
1973 * When systemd requests a PolicyKit decision when managing
1974 units it will now add additional fields to the request,
1975 including unit name and desired operation. This enables more
1976 powerful PolicyKit policies, that make decisions depending
1977 on these parameters.
1979 * nspawn learnt support for .nspawn settings files, that may
1980 accompany the image files or directories of containers, and
1981 may contain additional settings for the container. This is
1982 an alternative to configuring container parameters via the
1983 nspawn command line.
1985 Contributions from: Cristian Rodríguez, Daniel Mack, David
1986 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
1987 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
1988 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
1989 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
1990 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
1991 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
1992 Andersen, Tom Gundersen, Torstein Husebø
1994 — Berlin, 2015-09-08
1998 * machinectl gained a new verb 'shell' which opens a fresh
1999 shell on the target container or the host. It is similar to
2000 the existing 'login' command of machinectl, but spawns the
2001 shell directly without prompting for username or
2002 password. The pseudo machine '.host' now refers to the local
2003 host and is used by default. Hence, 'machinectl shell' can
2004 be used as replacement for 'su -' which spawns a session as
2005 a fresh systemd unit in a way that is fully isolated from
2006 the originating session.
2008 * systemd-networkd learned to cope with private-zone DHCP
2009 options and allows other programs to query the values.
2011 * SELinux access control when enabling/disabling units is no
2012 longer enforced with this release. The previous
2013 implementation was incorrect, and a new corrected
2014 implementation is not yet available. As unit file operations
2015 are still protected via PolicyKit and D-Bus policy this is
2016 not a security problem. Yet, distributions which care about
2017 optimal SELinux support should probably not stabilize on
2020 * sd-bus gained support for matches of type "arg0has=", that
2021 test for membership of strings in string arrays sent in bus
2024 * systemd-resolved now dumps the contents of its DNS and LLMNR
2025 caches to the logs on reception of the SIGUSR1 signal. This
2026 is useful to debug DNS behaviour.
2028 * The coredumpctl tool gained a new --directory= option to
2029 operate on journal files in a specific directory.
2031 * "systemctl reboot" and related commands gained a new
2032 "--message=" option which may be used to set a free-text
2033 wall message when shutting down or rebooting the
2034 system. This message is also logged, which is useful for
2035 figuring out the reason for a reboot or shutdown a
2038 * The "systemd-resolve-host" tool's -i switch now takes
2039 network interface numbers as alternative to interface names.
2041 * A new unit file setting for services has been introduced:
2042 UtmpMode= allows configuration of how precisely systemd
2043 handles utmp and wtmp entries for the service if this is
2044 enabled. This allows writing services that appear similar to
2045 user sessions in the output of the "w", "who", "last" and
2048 * systemd-resolved will now locally synthesize DNS resource
2049 records for the "localhost" and "gateway" domains as well as
2050 the local hostname. This should ensure that clients querying
2051 RRs via resolved will get similar results as those going via
2052 NSS, if nss-myhostname is enabled.
2054 Contributions from: Alastair Hughes, Alex Crawford, Daniel
2055 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
2056 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
2057 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
2058 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
2059 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
2060 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
2061 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
2062 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
2063 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
2064 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
2065 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2067 — Berlin, 2015-08-27
2071 * The systemd-efi-boot-generator functionality was merged into
2072 systemd-gpt-auto-generator.
2074 * systemd-networkd now supports Group Policy for vxlan
2075 devices. It can be enabled via the new boolean configuration
2076 option called 'GroupPolicyExtension='.
2078 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
2079 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
2080 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
2082 — Berlin, 2015-07-31
2086 * The python-systemd code has been removed from the systemd repository.
2087 A new repository has been created which accommodates the code from
2088 now on, and we kindly ask distributions to create a separate package
2089 for this: https://github.com/systemd/python-systemd
2091 * The systemd daemon will now reload its main configuration
2092 (/etc/systemd/system.conf) on daemon-reload.
2094 * sd-dhcp now exposes vendor specific extensions via
2095 sd_dhcp_lease_get_vendor_specific().
2097 * systemd-networkd gained a number of new configuration options.
2099 - A new boolean configuration option for TAP devices called
2100 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
2101 device, thus allowing to send and receive GSO packets.
2103 - A new tunnel configuration option called 'CopyDSCP='.
2104 If enabled, the DSCP field of ip6 tunnels is copied into the
2105 decapsulated packet.
2107 - A set of boolean bridge configuration options were added.
2108 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
2109 and 'UnicastFlood=' are now parsed by networkd and applied to the
2110 respective bridge link device via the respective IFLA_BRPORT_*
2113 - A new string configuration option to override the hostname sent
2114 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
2115 is true, networkd will use the configured hostname instead of the
2116 system hostname when sending DHCP requests.
2118 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
2119 networkd will configure the IPv6 flow-label of the tunnel device
2120 according to RFC2460.
2122 - The 'macvtap' virtual network devices are now supported, similar to
2123 the already supported 'macvlan' devices.
2125 * systemd-resolved now implements RFC5452 to improve resilience against
2126 cache poisoning. Additionally, source port randomization is enabled
2127 by default to further protect against DNS spoofing attacks.
2129 * nss-mymachines now supports translating UIDs and GIDs of running
2130 containers with user-namespaces enabled. If a container 'foo'
2131 translates a host uid 'UID' to the container uid 'TUID', then
2132 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
2133 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
2134 mapped as 'vg-foo-TGID'.
2136 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
2137 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
2138 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
2139 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
2140 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
2141 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
2142 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
2143 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
2144 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
2145 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
2147 — Berlin, 2015-07-29
2151 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
2152 There are no known issues with current sysfs, and udev does not need
2153 or should be used to work around such bugs.
2155 * udev does no longer enable USB HID power management. Several reports
2156 indicate, that some devices cannot handle that setting.
2158 * The udev accelerometer helper was removed. The functionality
2159 is now fully included in iio-sensor-proxy. But this means,
2160 older iio-sensor-proxy versions will no longer provide
2161 accelerometer/orientation data with this systemd version.
2162 Please upgrade iio-sensor-proxy to version 1.0.
2164 * networkd gained a new configuration option IPv6PrivacyExtensions=
2165 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
2166 for Stateless Address") on selected networks.
2168 * For the sake of fewer build-time dependencies and less code in the
2169 main repository, the python bindings are about to be removed in the
2170 next release. A new repository has been created which accommodates
2171 the code from now on, and we kindly ask distributions to create a
2172 separate package for this. The removal will take place in v223.
2174 https://github.com/systemd/python-systemd
2176 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
2177 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
2178 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
2179 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
2180 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
2181 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
2182 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
2183 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
2184 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
2185 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2187 — Berlin, 2015-07-07
2191 * The sd-bus.h and sd-event.h APIs have now been declared
2192 stable and have been added to the official interface of
2193 libsystemd.so. sd-bus implements an alternative D-Bus client
2194 library, that is relatively easy to use, very efficient and
2195 supports both classic D-Bus as well as kdbus as transport
2196 backend. sd-event is a generic event loop abstraction that
2197 is built around Linux epoll, but adds features such as event
2198 prioritization or efficient timer handling. Both APIs are good
2199 choices for C programs looking for a bus and/or event loop
2200 implementation that is minimal and does not have to be
2201 portable to other kernels.
2203 * kdbus support is no longer compile-time optional. It is now
2204 always built-in. However, it can still be disabled at
2205 runtime using the kdbus=0 kernel command line setting, and
2206 that setting may be changed to default to off, by specifying
2207 --disable-kdbus at build-time. Note though that the kernel
2208 command line setting has no effect if the kdbus.ko kernel
2209 module is not installed, in which case kdbus is (obviously)
2210 also disabled. We encourage all downstream distributions to
2211 begin testing kdbus by adding it to the kernel images in the
2212 development distributions, and leaving kdbus support in
2215 * The minimal required util-linux version has been bumped to
2218 * Support for chkconfig (--enable-chkconfig) was removed in
2219 favor of calling an abstraction tool
2220 /lib/systemd/systemd-sysv-install. This needs to be
2221 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
2222 in README for details.
2224 * If there's a systemd unit and a SysV init script for the
2225 same service name, and the user executes "systemctl enable"
2226 for it (or a related call), then this will now enable both
2227 (or execute the related operation on both), not just the
2230 * The libudev API documentation has been converted from gtkdoc
2233 * gudev has been removed from the systemd tree, it is now an
2236 * The systemd-cgtop tool learnt a new --raw switch to generate
2237 "raw" (machine parsable) output.
2239 * networkd's IPForwarding= .network file setting learnt the
2240 new setting "kernel", which ensures that networkd does not
2241 change the IP forwarding sysctl from the default kernel
2244 * The systemd-logind bus API now exposes a new boolean
2245 property "Docked" that reports whether logind considers the
2246 system "docked", i.e. connected to a docking station or not.
2248 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
2249 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
2250 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
2251 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
2252 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
2253 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
2254 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
2255 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
2256 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
2257 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
2258 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
2259 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
2260 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
2261 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
2262 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
2263 Fink, Zbigniew Jędrzejewski-Szmek
2265 — Berlin, 2015-06-19
2269 * The gudev library has been extracted into a separate repository
2270 available at: https://git.gnome.org/browse/libgudev/
2271 It is now managed as part of the Gnome project. Distributions
2272 are recommended to pass --disable-gudev to systemd and use
2273 gudev from the Gnome project instead. gudev is still included
2274 in systemd, for now. It will be removed soon, though. Please
2275 also see the announcement-thread on systemd-devel:
2276 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
2278 * systemd now exposes a CPUUsageNSec= property for each
2279 service unit on the bus, that contains the overall consumed
2280 CPU time of a service (the sum of what each process of the
2281 service consumed). This value is only available if
2282 CPUAccounting= is turned on for a service, and is then shown
2283 in the "systemctl status" output.
2285 * Support for configuring alternative mappings of the old SysV
2286 runlevels to systemd targets has been removed. They are now
2287 hardcoded in a way that runlevels 2, 3, 4 all map to
2288 multi-user.target and 5 to graphical.target (which
2289 previously was already the default behaviour).
2291 * The auto-mounter logic gained support for mount point
2292 expiry, using a new TimeoutIdleSec= setting in .automount
2293 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
2295 * The EFI System Partition (ESP) as mounted to /boot by
2296 systemd-efi-boot-generator will now be unmounted
2297 automatically after 2 minutes of not being used. This should
2298 minimize the risk of ESP corruptions.
2300 * New /etc/fstab options x-systemd.requires= and
2301 x-systemd.requires-mounts-for= are now supported to express
2302 additional dependencies for mounts. This is useful for
2303 journalling file systems that support external journal
2304 devices or overlay file systems that require underlying file
2305 systems to be mounted.
2307 * systemd does not support direct live-upgrades (via systemctl
2308 daemon-reexec) from versions older than v44 anymore. As no
2309 distribution we are aware of shipped such old versions in a
2310 stable release this should not be problematic.
2312 * When systemd forks off a new per-connection service instance
2313 it will now set the $REMOTE_ADDR environment variable to the
2314 remote IP address, and $REMOTE_PORT environment variable to
2315 the remote IP port. This behaviour is similar to the
2316 corresponding environment variables defined by CGI.
2318 * systemd-networkd gained support for uplink failure
2319 detection. The BindCarrier= option allows binding interface
2320 configuration dynamically to the link sense of other
2321 interfaces. This is useful to achieve behaviour like in
2324 * systemd-networkd gained support for configuring the DHCP
2325 client identifier to use when requesting leases.
2327 * systemd-networkd now has a per-network UseNTP= option to
2328 configure whether NTP server information acquired via DHCP
2329 is passed on to services like systemd-timesyncd.
2331 * systemd-networkd gained support for vti6 tunnels.
2333 * Note that systemd-networkd manages the sysctl variable
2334 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
2335 it is configured for since v219. The variable controls IP
2336 forwarding, and is a per-interface alternative to the global
2337 /proc/sys/net/ipv[46]/ip_forward. This setting is
2338 configurable in the IPForward= option, which defaults to
2339 "no". This means if networkd is used for an interface it is
2340 no longer sufficient to set the global sysctl option to turn
2341 on IP forwarding! Instead, the .network file option
2342 IPForward= needs to be turned on! Note that the
2343 implementation of this behaviour was broken in v219 and has
2346 * Many bonding and vxlan options are now configurable in
2349 * systemd-nspawn gained a new --property= setting to set unit
2350 properties for the container scope. This is useful for
2351 setting resource parameters (e.g. "CPUShares=500") on
2352 containers started from the command line.
2354 * systemd-nspawn gained a new --private-users= switch to make
2355 use of user namespacing available on recent Linux kernels.
2357 * systemd-nspawn may now be called as part of a shell pipeline
2358 in which case the pipes used for stdin and stdout are passed
2359 directly to the process invoked in the container, without
2360 indirection via a pseudo tty.
2362 * systemd-nspawn gained a new switch to control the UNIX
2363 signal to use when killing the init process of the container
2366 * systemd-nspawn gained a new --overlay= switch for mounting
2367 overlay file systems into the container using the new kernel
2370 * When a container image is imported via systemd-importd and
2371 the host file system is not btrfs, a loopback block device
2372 file is created in /var/lib/machines.raw with a btrfs file
2373 system inside. It is then mounted to /var/lib/machines to
2374 enable btrfs features for container management. The loopback
2375 file and btrfs file system is grown as needed when container
2376 images are imported via systemd-importd.
2378 * systemd-machined/systemd-importd gained support for btrfs
2379 quota, to enforce container disk space limits on disk. This
2380 is exposed in "machinectl set-limit".
2382 * systemd-importd now can import containers from local .tar,
2383 .raw and .qcow2 images, and export them to .tar and .raw. It
2384 can also import dkr v2 images now from the network (on top
2387 * systemd-importd gained support for verifying downloaded
2388 images with gpg2 (previously only gpg1 was supported).
2390 * systemd-machined, systemd-logind, systemd: most bus calls
2391 are now accessible to unprivileged processes via
2392 PolicyKit. Also, systemd-logind will now allow users to kill
2393 their own sessions without further privileges or
2396 * systemd-shutdownd has been removed. This service was
2397 previously responsible for implementing scheduled shutdowns
2398 as exposed in /usr/bin/shutdown's time parameter. This
2399 functionality has now been moved into systemd-logind and is
2400 accessible via a bus interface.
2402 * "systemctl reboot" gained a new switch --firmware-setup that
2403 can be used to reboot into the EFI firmware setup, if that
2404 is available. systemd-logind now exposes an API on the bus
2405 to trigger such reboots, in case graphical desktop UIs want
2406 to cover this functionality.
2408 * "systemctl enable", "systemctl disable" and "systemctl mask"
2409 now support a new "--now" switch. If specified the units
2410 that are enabled will also be started, and the ones
2411 disabled/masked also stopped.
2413 * The Gummiboot EFI boot loader tool has been merged into
2414 systemd, and renamed to "systemd-boot". The bootctl tool has been
2415 updated to support systemd-boot.
2417 * An EFI kernel stub has been added that may be used to create
2418 kernel EFI binaries that contain not only the actual kernel,
2419 but also an initrd, boot splash, command line and OS release
2420 information. This combined binary can then be signed as a
2421 single image, so that the firmware can verify it all in one
2422 step. systemd-boot has special support for EFI binaries created
2423 like this and can extract OS release information from them
2424 and show them in the boot menu. This functionality is useful
2425 to implement cryptographically verified boot schemes.
2427 * Optional support has been added to systemd-fsck to pass
2428 fsck's progress report to an AF_UNIX socket in the file
2431 * udev will no longer create device symlinks for all block
2432 devices by default. A blacklist for excluding special block
2433 devices from this logic has been turned into a whitelist
2434 that requires picking block devices explicitly that require
2437 * A new (currently still internal) API sd-device.h has been
2438 added to libsystemd. This modernized API is supposed to
2439 replace libudev eventually. In fact, already much of libudev
2440 is now just a wrapper around sd-device.h.
2442 * A new hwdb database for storing metadata about pointing
2443 stick devices has been added.
2445 * systemd-tmpfiles gained support for setting file attributes
2446 similar to the "chattr" tool with new 'h' and 'H' lines.
2448 * systemd-journald will no longer unconditionally set the
2449 btrfs NOCOW flag on new journal files. This is instead done
2450 with tmpfiles snippet using the new 'h' line type. This
2451 allows easy disabling of this logic, by masking the
2452 journal-nocow.conf tmpfiles file.
2454 * systemd-journald will now translate audit message types to
2455 human readable identifiers when writing them to the
2456 journal. This should improve readability of audit messages.
2458 * The LUKS logic gained support for the offset= and skip=
2459 options in /etc/crypttab, as previously implemented by
2462 * /usr/lib/os-release gained a new optional field VARIANT= for
2463 distributions that support multiple variants (such as a
2464 desktop edition, a server edition, ...)
2466 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
2467 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
2468 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
2469 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
2470 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
2471 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
2472 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
2473 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
2474 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
2475 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
2476 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
2477 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
2478 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
2479 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
2480 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
2481 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
2482 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
2483 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
2484 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
2485 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
2486 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
2487 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
2488 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
2489 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
2490 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
2491 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
2492 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
2494 — Berlin, 2015-05-22
2498 * Introduce a new API "sd-hwdb.h" for querying the hardware
2499 metadata database. With this minimal interface one can query
2500 and enumerate the udev hwdb, decoupled from the old libudev
2501 library. libudev's interface for this is now only a wrapper
2502 around sd-hwdb. A new tool systemd-hwdb has been added to
2503 interface with and update the database.
2505 * When any of systemd's tools copies files (for example due to
2506 tmpfiles' C lines) a btrfs reflink will attempted first,
2507 before bytewise copying is done.
2509 * systemd-nspawn gained a new --ephemeral switch. When
2510 specified a btrfs snapshot is taken of the container's root
2511 directory, and immediately removed when the container
2512 terminates again. Thus, a container can be started whose
2513 changes never alter the container's root directory, and are
2514 lost on container termination. This switch can also be used
2515 for starting a container off the root file system of the
2516 host without affecting the host OS. This switch is only
2517 available on btrfs file systems.
2519 * systemd-nspawn gained a new --template= switch. It takes the
2520 path to a container tree to use as template for the tree
2521 specified via --directory=, should that directory be
2522 missing. This allows instantiating containers dynamically,
2523 on first run. This switch is only available on btrfs file
2526 * When a .mount unit refers to a mount point on which multiple
2527 mounts are stacked, and the .mount unit is stopped all of
2528 the stacked mount points will now be unmounted until no
2529 mount point remains.
2531 * systemd now has an explicit notion of supported and
2532 unsupported unit types. Jobs enqueued for unsupported unit
2533 types will now fail with an "unsupported" error code. More
2534 specifically .swap, .automount and .device units are not
2535 supported in containers, .busname units are not supported on
2536 non-kdbus systems. .swap and .automount are also not
2537 supported if their respective kernel compile time options
2540 * machinectl gained support for two new "copy-from" and
2541 "copy-to" commands for copying files from a running
2542 container to the host or vice versa.
2544 * machinectl gained support for a new "bind" command to bind
2545 mount host directories into local containers. This is
2546 currently only supported for nspawn containers.
2548 * networkd gained support for configuring bridge forwarding
2549 database entries (fdb) from .network files.
2551 * A new tiny daemon "systemd-importd" has been added that can
2552 download container images in tar, raw, qcow2 or dkr formats,
2553 and make them available locally in /var/lib/machines, so
2554 that they can run as nspawn containers. The daemon can GPG
2555 verify the downloads (not supported for dkr, since it has no
2556 provisions for verifying downloads). It will transparently
2557 decompress bz2, xz, gzip compressed downloads if necessary,
2558 and restore sparse files on disk. The daemon uses privilege
2559 separation to ensure the actual download logic runs with
2560 fewer privileges than the daemon itself. machinectl has
2561 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
2562 make the functionality of importd available to the
2563 user. With this in place the Fedora and Ubuntu "Cloud"
2564 images can be downloaded and booted as containers unmodified
2565 (the Fedora images lack the appropriate GPG signature files
2566 currently, so they cannot be verified, but this will change
2567 soon, hopefully). Note that downloading images is currently
2568 only fully supported on btrfs.
2570 * machinectl is now able to list container images found in
2571 /var/lib/machines, along with some metadata about sizes of
2572 disk and similar. If the directory is located on btrfs and
2573 quota is enabled, this includes quota display. A new command
2574 "image-status" has been added that shows additional
2575 information about images.
2577 * machinectl is now able to clone container images
2578 efficiently, if the underlying file system (btrfs) supports
2579 it, with the new "machinectl clone" command. It also
2580 gained commands for renaming and removing images, as well as
2581 marking them read-only or read-write (supported also on
2582 legacy file systems).
2584 * networkd gained support for collecting LLDP network
2585 announcements, from hardware that supports this. This is
2586 shown in networkctl output.
2588 * systemd-run gained support for a new -t (--pty) switch for
2589 invoking a binary on a pty whose input and output is
2590 connected to the invoking terminal. This allows executing
2591 processes as system services while interactively
2592 communicating with them via the terminal. Most interestingly
2593 this is supported across container boundaries. Invoking
2594 "systemd-run -t /bin/bash" is an alternative to running a
2595 full login session, the difference being that the former
2596 will not register a session, nor go through the PAM session
2599 * tmpfiles gained support for a new "v" line type for creating
2600 btrfs subvolumes. If the underlying file system is a legacy
2601 file system, this automatically degrades to creating a
2602 normal directory. Among others /var/lib/machines is now
2603 created like this at boot, should it be missing.
2605 * The directory /var/lib/containers/ has been deprecated and
2606 been replaced by /var/lib/machines. The term "machines" has
2607 been used in the systemd context as generic term for both
2608 VMs and containers, and hence appears more appropriate for
2609 this, as the directory can also contain raw images bootable
2612 * systemd-nspawn when invoked with -M but without --directory=
2613 or --image= is now capable of searching for the container
2614 root directory, subvolume or disk image automatically, in
2615 /var/lib/machines. systemd-nspawn@.service has been updated
2616 to make use of this, thus allowing it to be used for raw
2619 * A new machines.target unit has been introduced that is
2620 supposed to group all containers/VMs invoked as services on
2621 the system. systemd-nspawn@.service has been updated to
2622 integrate with that.
2624 * machinectl gained a new "start" command, for invoking a
2625 container as a service. "machinectl start foo" is mostly
2626 equivalent to "systemctl start systemd-nspawn@foo.service",
2627 but handles escaping in a nicer way.
2629 * systemd-nspawn will now mount most of the cgroupfs tree
2630 read-only into each container, with the exception of the
2631 container's own subtree in the name=systemd hierarchy.
2633 * journald now sets the special FS_NOCOW file flag for its
2634 journal files. This should improve performance on btrfs, by
2635 avoiding heavy fragmentation when journald's write-pattern
2636 is used on COW file systems. It degrades btrfs' data
2637 integrity guarantees for the files to the same levels as for
2638 ext3/ext4 however. This should be OK though as journald does
2639 its own data integrity checks and all its objects are
2640 checksummed on disk. Also, journald should handle btrfs disk
2641 full events a lot more gracefully now, by processing SIGBUS
2642 errors, and not relying on fallocate() anymore.
2644 * When journald detects that journal files it is writing to
2645 have been deleted it will immediately start new journal
2648 * systemd now provides a way to store file descriptors
2649 per-service in PID 1. This is useful for daemons to ensure
2650 that fds they require are not lost during a daemon
2651 restart. The fds are passed to the daemon on the next
2652 invocation in the same way socket activation fds are
2653 passed. This is now used by journald to ensure that the
2654 various sockets connected to all the system's stdout/stderr
2655 are not lost when journald is restarted. File descriptors
2656 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
2657 an extension to sd_notify(). Note that a limit is enforced
2658 on the number of fds a service can store in PID 1, and it
2659 defaults to 0, so that no fds may be stored, unless this is
2660 explicitly turned on.
2662 * The default TERM variable to use for units connected to a
2663 terminal, when no other value is explicitly is set is now
2664 vt220 rather than vt102. This should be fairly safe still,
2665 but allows PgUp/PgDn work.
2667 * The /etc/crypttab option header= as known from Debian is now
2670 * "loginctl user-status" and "loginctl session-status" will
2671 now show the last 10 lines of log messages of the
2672 user/session following the status output. Similar,
2673 "machinectl status" will show the last 10 log lines
2674 associated with a virtual machine or container
2675 service. (Note that this is usually not the log messages
2676 done in the VM/container itself, but simply what the
2677 container manager logs. For nspawn this includes all console
2680 * "loginctl session-status" without further argument will now
2681 show the status of the session of the caller. Similar,
2682 "lock-session", "unlock-session", "activate",
2683 "enable-linger", "disable-linger" may now be called without
2684 session/user parameter in which case they apply to the
2685 caller's session/user.
2687 * An X11 session scriptlet is now shipped that uploads
2688 $DISPLAY and $XAUTHORITY into the environment of the systemd
2689 --user daemon if a session begins. This should improve
2690 compatibility with X11 enabled applications run as systemd
2693 * Generators are now subject to masking via /etc and /run, the
2694 same way as unit files.
2696 * networkd .network files gained support for configuring
2697 per-link IPv4/IPv6 packet forwarding as well as IPv4
2698 masquerading. This is by default turned on for veth links to
2699 containers, as registered by systemd-nspawn. This means that
2700 nspawn containers run with --network-veth will now get
2701 automatic routed access to the host's networks without any
2702 further configuration or setup, as long as networkd runs on
2705 * systemd-nspawn gained the --port= (-p) switch to expose TCP
2706 or UDP posts of a container on the host. With this in place
2707 it is possible to run containers with private veth links
2708 (--network-veth), and have their functionality exposed on
2709 the host as if their services were running directly on the
2712 * systemd-nspawn's --network-veth switch now gained a short
2713 version "-n", since with the changes above it is now truly
2714 useful out-of-the-box. The systemd-nspawn@.service has been
2715 updated to make use of it too by default.
2717 * systemd-nspawn will now maintain a per-image R/W lock, to
2718 ensure that the same image is not started more than once
2719 writable. (It's OK to run an image multiple times
2720 simultaneously in read-only mode.)
2722 * systemd-nspawn's --image= option is now capable of
2723 dissecting and booting MBR and GPT disk images that contain
2724 only a single active Linux partition. Previously it
2725 supported only GPT disk images with proper GPT type
2726 IDs. This allows running cloud images from major
2727 distributions directly with systemd-nspawn, without
2730 * In addition to collecting mouse dpi data in the udev
2731 hardware database, there's now support for collecting angle
2732 information for mouse scroll wheels. The database is
2733 supposed to guarantee similar scrolling behavior on mice
2734 that it knows about. There's also support for collecting
2735 information about Touchpad types.
2737 * udev's input_id built-in will now also collect touch screen
2738 dimension data and attach it to probed devices.
2740 * /etc/os-release gained support for a Distribution Privacy
2743 * networkd gained support for creating "ipvlan", "gretap",
2744 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
2746 * systemd-tmpfiles gained support for "a" lines for setting
2749 * systemd-nspawn will now mount /tmp in the container to
2750 tmpfs, automatically.
2752 * systemd now exposes the memory.usage_in_bytes cgroup
2753 attribute and shows it for each service in the "systemctl
2754 status" output, if available.
2756 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
2757 immediate reboot is triggered. This useful if shutdown is
2758 hung and is unable to complete, to expedite the
2759 operation. Note that this kind of reboot will still unmount
2760 all file systems, and hence should not result in fsck being
2763 * A .device unit for an optical block device will now be
2764 considered active only when a medium is in the drive. Also,
2765 mount units are now bound to their backing devices thus
2766 triggering automatic unmounting when devices become
2767 unavailable. With this in place systemd will now
2768 automatically unmount left-over mounts when a CD-ROM is
2769 ejected or an USB stick is yanked from the system.
2771 * networkd-wait-online now has support for waiting for
2772 specific interfaces only (with globbing), and for giving up
2773 after a configurable timeout.
2775 * networkd now exits when idle. It will be automatically
2776 restarted as soon as interfaces show up, are removed or
2777 change state. networkd will stay around as long as there is
2778 at least one DHCP state machine or similar around, that keep
2781 * networkd may now configure IPv6 link-local addressing in
2782 addition to IPv4 link-local addressing.
2784 * The IPv6 "token" for use in SLAAC may now be configured for
2785 each .network interface in networkd.
2787 * Routes configured with networkd may now be assigned a scope
2790 * networkd's [Match] sections now support globbing and lists
2791 of multiple space-separated matches per item.
2793 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
2794 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
2795 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
2796 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
2797 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
2798 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
2799 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
2800 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
2801 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
2802 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
2803 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
2804 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
2805 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
2806 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
2807 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
2808 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
2809 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
2810 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
2811 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
2812 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
2813 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
2814 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
2815 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
2816 Hoffmann, Zbigniew Jędrzejewski-Szmek
2818 — Berlin, 2015-02-16
2822 * When querying unit file enablement status (for example via
2823 "systemctl is-enabled"), a new state "indirect" is now known
2824 which indicates that a unit might not be enabled itself, but
2825 another unit listed in its Also= setting might be.
2827 * Similar to the various existing ConditionXYZ= settings for
2828 units, there are now matching AssertXYZ= settings. While
2829 failing conditions cause a unit to be skipped, but its job
2830 to succeed, failing assertions declared like this will cause
2831 a unit start operation and its job to fail.
2833 * hostnamed now knows a new chassis type "embedded".
2835 * systemctl gained a new "edit" command. When used on a unit
2836 file, this allows extending unit files with .d/ drop-in
2837 configuration snippets or editing the full file (after
2838 copying it from /usr/lib to /etc). This will invoke the
2839 user's editor (as configured with $EDITOR), and reload the
2840 modified configuration after editing.
2842 * "systemctl status" now shows the suggested enablement state
2843 for a unit, as declared in the (usually vendor-supplied)
2844 system preset files.
2846 * nss-myhostname will now resolve the single-label host name
2847 "gateway" to the locally configured default IP routing
2848 gateways, ordered by their metrics. This assigns a stable
2849 name to the used gateways, regardless which ones are
2850 currently configured. Note that the name will only be
2851 resolved after all other name sources (if nss-myhostname is
2852 configured properly) and should hence not negatively impact
2853 systems that use the single-label host name "gateway" in
2856 * systemd-inhibit now allows filtering by mode when listing
2859 * Scope and service units gained a new "Delegate" boolean
2860 property, which, when set, allows processes running inside the
2861 unit to further partition resources. This is primarily
2862 useful for systemd user instances as well as container
2865 * journald will now pick up audit messages directly from
2866 the kernel, and log them like any other log message. The
2867 audit fields are split up and fully indexed. This means that
2868 journalctl in many ways is now a (nicer!) alternative to
2869 ausearch, the traditional audit client. Note that this
2870 implements only a minimal audit client. If you want the
2871 special audit modes like reboot-on-log-overflow, please use
2872 the traditional auditd instead, which can be used in
2873 parallel to journald.
2875 * The ConditionSecurity= unit file option now understands the
2876 special string "audit" to check whether auditing is
2879 * journalctl gained two new commands --vacuum-size= and
2880 --vacuum-time= to delete old journal files until the
2881 remaining ones take up no more than the specified size on disk,
2882 or are not older than the specified time.
2884 * A new, native PPPoE library has been added to sd-network,
2885 systemd's library of light-weight networking protocols. This
2886 library will be used in a future version of networkd to
2887 enable PPPoE communication without an external pppd daemon.
2889 * The busctl tool now understands a new "capture" verb that
2890 works similar to "monitor", but writes a packet capture
2891 trace to STDOUT that can be redirected to a file which is
2892 compatible with libcap's capture file format. This can then
2893 be loaded in Wireshark and similar tools to inspect bus
2896 * The busctl tool now understands a new "tree" verb that shows
2897 the object trees of a specific service on the bus, or of all
2900 * The busctl tool now understands a new "introspect" verb that
2901 shows all interfaces and members of objects on the bus,
2902 including their signature and values. This is particularly
2903 useful to get more information about bus objects shown by
2904 the new "busctl tree" command.
2906 * The busctl tool now understands new verbs "call",
2907 "set-property" and "get-property" for invoking bus method
2908 calls, setting and getting bus object properties in a
2911 * busctl gained a new --augment-creds= argument that controls
2912 whether the tool shall augment credential information it
2913 gets from the bus with data from /proc, in a possibly
2916 * nspawn's --link-journal= switch gained two new values
2917 "try-guest" and "try-host" that work like "guest" and
2918 "host", but do not fail if the host has no persistent
2919 journalling enabled. -j is now equivalent to
2920 --link-journal=try-guest.
2922 * macvlan network devices created by nspawn will now have
2923 stable MAC addresses.
2925 * A new SmackProcessLabel= unit setting has been added, which
2926 controls the SMACK security label processes forked off by
2927 the respective unit shall use.
2929 * If compiled with --enable-xkbcommon, systemd-localed will
2930 verify x11 keymap settings by compiling the given keymap. It
2931 will spew out warnings if the compilation fails. This
2932 requires libxkbcommon to be installed.
2934 * When a coredump is collected, a larger number of metadata
2935 fields is now collected and included in the journal records
2936 created for it. More specifically, control group membership,
2937 environment variables, memory maps, working directory,
2938 chroot directory, /proc/$PID/status, and a list of open file
2939 descriptors is now stored in the log entry.
2941 * The udev hwdb now contains DPI information for mice. For
2944 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
2946 * All systemd programs that read standalone configuration
2947 files in /etc now also support a corresponding series of
2948 .conf.d configuration directories in /etc/, /run/,
2949 /usr/local/lib/, /usr/lib/, and (if configured with
2950 --enable-split-usr) /lib/. In particular, the following
2951 configuration files now have corresponding configuration
2952 directories: system.conf user.conf, logind.conf,
2953 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
2954 resolved.conf, timesyncd.conf, journal-remote.conf, and
2955 journal-upload.conf. Note that distributions should use the
2956 configuration directories in /usr/lib/; the directories in
2957 /etc/ are reserved for the system administrator.
2959 * systemd-rfkill will no longer take the rfkill device name
2960 into account when storing rfkill state on disk, as the name
2961 might be dynamically assigned and not stable. Instead, the
2962 ID_PATH udev variable combined with the rfkill type (wlan,
2963 bluetooth, ...) is used.
2965 * A new service systemd-machine-id-commit.service has been
2966 added. When used on systems where /etc is read-only during
2967 boot, and /etc/machine-id is not initialized (but an empty
2968 file), this service will copy the temporary machine ID
2969 created as replacement into /etc after the system is fully
2970 booted up. This is useful for systems that are freshly
2971 installed with a non-initialized machine ID, but should get
2972 a fixed machine ID for subsequent boots.
2974 * networkd's .netdev files now provide a large set of
2975 configuration parameters for VXLAN devices. Similarly, the
2976 bridge port cost parameter is now configurable in .network
2977 files. There's also new support for configuring IP source
2978 routing. networkd .link files gained support for a new
2979 OriginalName= match that is useful to match against the
2980 original interface name the kernel assigned. .network files
2981 may include MTU= and MACAddress= fields for altering the MTU
2982 and MAC address while being connected to a specific network
2985 * The LUKS logic gained supported for configuring
2986 UUID-specific key files. There's also new support for naming
2987 LUKS device from the kernel command line, using the new
2988 luks.name= argument.
2990 * Timer units may now be transiently created via the bus API
2991 (this was previously already available for scope and service
2992 units). In addition it is now possible to create multiple
2993 transient units at the same time with a single bus call. The
2994 "systemd-run" tool has been updated to make use of this for
2995 running commands on a specified time, in at(1)-style.
2997 * tmpfiles gained support for "t" lines, for assigning
2998 extended attributes to files. Among other uses this may be
2999 used to assign SMACK labels to files.
3001 Contributions from: Alin Rauta, Alison Chaiken, Andrej
3002 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
3003 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
3004 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
3005 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
3006 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
3007 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
3008 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
3009 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
3010 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
3011 Michael Biebl, Michael Chapman, Michael Marineau, Michal
3012 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
3013 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
3014 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
3015 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
3016 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
3017 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
3018 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
3020 — Berlin, 2014-12-10
3024 * journalctl gained the new options -t/--identifier= to match
3025 on the syslog identifier (aka "tag"), as well as --utc to
3026 show log timestamps in the UTC timezone. journalctl now also
3027 accepts -n/--lines=all to disable line capping in a pager.
3029 * journalctl gained a new switch, --flush, that synchronously
3030 flushes logs from /run/log/journal to /var/log/journal if
3031 persistent storage is enabled. systemd-journal-flush.service
3032 now waits until the operation is complete.
3034 * Services can notify the manager before they start a reload
3035 (by sending RELOADING=1) or shutdown (by sending
3036 STOPPING=1). This allows the manager to track and show the
3037 internal state of daemons and closes a race condition when
3038 the process is still running but has closed its D-Bus
3041 * Services with Type=oneshot do not have to have any ExecStart
3044 * User units are now loaded also from
3045 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
3046 /run/systemd/user directory that was already previously
3047 supported, but is under the control of the user.
3049 * Job timeouts (i.e. time-outs on the time a job that is
3050 queued stays in the run queue) can now optionally result in
3051 immediate reboot or power-off actions (JobTimeoutAction= and
3052 JobTimeoutRebootArgument=). This is useful on ".target"
3053 units, to limit the maximum time a target remains
3054 undispatched in the run queue, and to trigger an emergency
3055 operation in such a case. This is now used by default to
3056 turn off the system if boot-up (as defined by everything in
3057 basic.target) hangs and does not complete for at least
3058 15min. Also, if power-off or reboot hang for at least 30min
3059 an immediate power-off/reboot operation is triggered. This
3060 functionality is particularly useful to increase reliability
3061 on embedded devices, but also on laptops which might
3062 accidentally get powered on when carried in a backpack and
3063 whose boot stays stuck in a hard disk encryption passphrase
3066 * systemd-logind can be configured to also handle lid switch
3067 events even when the machine is docked or multiple displays
3068 are attached (HandleLidSwitchDocked= option).
3070 * A helper binary and a service have been added which can be
3071 used to resume from hibernation in the initramfs. A
3072 generator will parse the resume= option on the kernel
3073 command line to trigger resume.
3075 * A user console daemon systemd-consoled has been
3076 added. Currently, it is a preview, and will so far open a
3077 single terminal on each session of the user marked as
3078 Desktop=systemd-console.
3080 * Route metrics can be specified for DHCP routes added by
3083 * The SELinux context of socket-activated services can be set
3084 from the information provided by the networking stack
3085 (SELinuxContextFromNet= option).
3087 * Userspace firmware loading support has been removed and
3088 the minimum supported kernel version is thus bumped to 3.7.
3090 * Timeout for udev workers has been increased from 1 to 3
3091 minutes, but a warning will be printed after 1 minute to
3092 help diagnose kernel modules that take a long time to load.
3094 * Udev rules can now remove tags on devices with TAG-="foobar".
3096 * systemd's readahead implementation has been removed. In many
3097 circumstances it didn't give expected benefits even for
3098 rotational disk drives and was becoming less relevant in the
3099 age of SSDs. As none of the developers has been using
3100 rotating media anymore, and nobody stepped up to actively
3101 maintain this component of systemd it has now been removed.
3103 * Swap units can use Options= to specify discard options.
3104 Discard options specified for swaps in /etc/fstab are now
3107 * Docker containers are now detected as a separate type of
3110 * The Password Agent protocol gained support for queries where
3111 the user input is shown, useful e.g. for user names.
3112 systemd-ask-password gained a new --echo option to turn that
3115 * The default sysctl.d/ snippets will now set:
3117 net.core.default_qdisc = fq_codel
3119 This selects Fair Queuing Controlled Delay as the default
3120 queuing discipline for network interfaces. fq_codel helps
3121 fight the network bufferbloat problem. It is believed to be
3122 a good default with no tuning required for most workloads.
3123 Downstream distributions may override this choice. On 10Gbit
3124 servers that do not do forwarding, "fq" may perform better.
3125 Systems without a good clocksource should use "pfifo_fast".
3127 * If kdbus is enabled during build a new option BusPolicy= is
3128 available for service units, that allows locking all service
3129 processes into a stricter bus policy, in order to limit
3130 access to various bus services, or even hide most of them
3131 from the service's view entirely.
3133 * networkctl will now show the .network and .link file
3134 networkd has applied to a specific interface.
3136 * sd-login gained a new API call sd_session_get_desktop() to
3137 query which desktop environment has been selected for a
3140 * UNIX utmp support is now compile-time optional to support
3141 legacy-free systems.
3143 * systemctl gained two new commands "add-wants" and
3144 "add-requires" for pulling in units from specific targets
3147 * If the word "rescue" is specified on the kernel command line
3148 the system will now boot into rescue mode (aka
3149 rescue.target), which was previously available only by
3150 specifying "1" or "systemd.unit=rescue.target" on the kernel
3151 command line. This new kernel command line option nicely
3152 mirrors the already existing "emergency" kernel command line
3155 * New kernel command line options mount.usr=, mount.usrflags=,
3156 mount.usrfstype= have been added that match root=, rootflags=,
3157 rootfstype= but allow mounting a specific file system to
3160 * The $NOTIFY_SOCKET is now also passed to control processes of
3161 services, not only the main process.
3163 * This version reenables support for fsck's -l switch. This
3164 means at least version v2.25 of util-linux is required for
3165 operation, otherwise dead-locks on device nodes may
3166 occur. Again: you need to update util-linux to at least
3167 v2.25 when updating systemd to v217.
3169 * The "multi-seat-x" tool has been removed from systemd, as
3170 its functionality has been integrated into X servers 1.16,
3171 and the tool is hence redundant. It is recommended to update
3172 display managers invoking this tool to simply invoke X
3173 directly from now on, again.
3175 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
3176 message flag has been added for all of systemd's PolicyKit
3177 authenticated method calls has been added. In particular
3178 this now allows optional interactive authorization via
3179 PolicyKit for many of PID1's privileged operations such as
3180 unit file enabling and disabling.
3182 * "udevadm hwdb --update" learnt a new switch "--usr" for
3183 placing the rebuilt hardware database in /usr instead of
3184 /etc. When used only hardware database entries stored in
3185 /usr will be used, and any user database entries in /etc are
3186 ignored. This functionality is useful for vendors to ship a
3187 pre-built database on systems where local configuration is
3188 unnecessary or unlikely.
3190 * Calendar time specifications in .timer units now also
3191 understand the strings "semi-annually", "quarterly" and
3192 "minutely" as shortcuts (in addition to the preexisting
3193 "anually", "hourly", ...).
3195 * systemd-tmpfiles will now correctly create files in /dev
3196 at boot which are marked for creation only at boot. It is
3197 recommended to always create static device nodes with 'c!'
3198 and 'b!', so that they are created only at boot and not
3199 overwritten at runtime.
3201 * When the watchdog logic is used for a service (WatchdogSec=)
3202 and the watchdog timeout is hit the service will now be
3203 terminated with SIGABRT (instead of just SIGTERM), in order
3204 to make sure a proper coredump and backtrace is
3205 generated. This ensures that hanging services will result in
3206 similar coredump/backtrace behaviour as services that hit a
3209 Contributions from: Andreas Henriksson, Andrei Borzenkov,
3210 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
3211 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
3212 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
3213 Herrmann, David Sommerseth, David Strauss, Emil Renner
3214 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
3215 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
3216 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
3217 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
3218 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
3219 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
3220 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
3221 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
3222 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
3223 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
3224 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
3225 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
3226 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
3227 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
3228 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
3229 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
3232 — Berlin, 2014-10-28
3236 * timedated no longer reads NTP implementation unit names from
3237 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
3238 implementations should add a
3240 Conflicts=systemd-timesyncd.service
3242 to their unit files to take over and replace systemd's NTP
3243 default functionality.
3245 * systemd-sysusers gained a new line type "r" for configuring
3246 which UID/GID ranges to allocate system users/groups
3247 from. Lines of type "u" may now add an additional column
3248 that specifies the home directory for the system user to be
3249 created. Also, systemd-sysusers may now optionally read user
3250 information from STDIN instead of a file. This is useful for
3251 invoking it from RPM preinst scriptlets that need to create
3252 users before the first RPM file is installed since these
3253 files might need to be owned by them. A new
3254 %sysusers_create_inline RPM macro has been introduced to do
3255 just that. systemd-sysusers now updates the shadow files as
3256 well as the user/group databases, which should enhance
3257 compatibility with certain tools like grpck.
3259 * A number of bus APIs of PID 1 now optionally consult
3260 PolicyKit to permit access for otherwise unprivileged
3261 clients under certain conditions. Note that this currently
3262 doesn't support interactive authentication yet, but this is
3263 expected to be added eventually, too.
3265 * /etc/machine-info now has new fields for configuring the
3266 deployment environment of the machine, as well as the
3267 location of the machine. hostnamectl has been updated with
3268 new command to update these fields.
3270 * systemd-timesyncd has been updated to automatically acquire
3271 NTP server information from systemd-networkd, which might
3272 have been discovered via DHCP.
3274 * systemd-resolved now includes a caching DNS stub resolver
3275 and a complete LLMNR name resolution implementation. A new
3276 NSS module "nss-resolve" has been added which can be used
3277 instead of glibc's own "nss-dns" to resolve hostnames via
3278 systemd-resolved. Hostnames, addresses and arbitrary RRs may
3279 be resolved via systemd-resolved D-Bus APIs. In contrast to
3280 the glibc internal resolver systemd-resolved is aware of
3281 multi-homed system, and keeps DNS server and caches separate
3282 and per-interface. Queries are sent simultaneously on all
3283 interfaces that have DNS servers configured, in order to
3284 properly handle VPNs and local LANs which might resolve
3285 separate sets of domain names. systemd-resolved may acquire
3286 DNS server information from systemd-networkd automatically,
3287 which in turn might have discovered them via DHCP. A tool
3288 "systemd-resolve-host" has been added that may be used to
3289 query the DNS logic in resolved. systemd-resolved implements
3290 IDNA and automatically uses IDNA or UTF-8 encoding depending
3291 on whether classic DNS or LLMNR is used as transport. In the
3292 next releases we intend to add a DNSSEC and mDNS/DNS-SD
3293 implementation to systemd-resolved.
3295 * A new NSS module nss-mymachines has been added, that
3296 automatically resolves the names of all local registered
3297 containers to their respective IP addresses.
3299 * A new client tool "networkctl" for systemd-networkd has been
3300 added. It currently is entirely passive and will query
3301 networking configuration from udev, rtnetlink and networkd,
3302 and present it to the user in a very friendly
3303 way. Eventually, we hope to extend it to become a full
3304 control utility for networkd.
3306 * .socket units gained a new DeferAcceptSec= setting that
3307 controls the kernels' TCP_DEFER_ACCEPT sockopt for
3308 TCP. Similarly, support for controlling TCP keep-alive
3309 settings has been added (KeepAliveTimeSec=,
3310 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
3311 turning off Nagle's algorithm on TCP has been added
3314 * logind learned a new session type "web", for use in projects
3315 like Cockpit which register web clients as PAM sessions.
3317 * timer units with at least one OnCalendar= setting will now
3318 be started only after timer-sync.target has been
3319 reached. This way they will not elapse before the system
3320 clock has been corrected by a local NTP client or
3321 similar. This is particular useful on RTC-less embedded
3322 machines, that come up with an invalid system clock.
3324 * systemd-nspawn's --network-veth= switch should now result in
3325 stable MAC addresses for both the outer and the inner side
3328 * systemd-nspawn gained a new --volatile= switch for running
3329 container instances with /etc or /var unpopulated.
3331 * The kdbus client code has been updated to use the new Linux
3332 3.17 memfd subsystem instead of the old kdbus-specific one.
3334 * systemd-networkd's DHCP client and server now support
3335 FORCERENEW. There are also new configuration options to
3336 configure the vendor client identifier and broadcast mode
3339 * systemd will no longer inform the kernel about the current
3340 timezone, as this is necessarily incorrect and racy as the
3341 kernel has no understanding of DST and similar
3342 concepts. This hence means FAT timestamps will be always
3343 considered UTC, similar to what Android is already
3344 doing. Also, when the RTC is configured to the local time
3345 (rather than UTC) systemd will never synchronize back to it,
3346 as this might confuse Windows at a later boot.
3348 * systemd-analyze gained a new command "verify" for offline
3349 validation of unit files.
3351 * systemd-networkd gained support for a couple of additional
3352 settings for bonding networking setups. Also, the metric for
3353 statically configured routes may now be configured. For
3354 network interfaces where this is appropriate the peer IP
3355 address may now be configured.
3357 * systemd-networkd's DHCP client will no longer request
3358 broadcasting by default, as this tripped up some networks.
3359 For hardware where broadcast is required the feature should
3360 be switched back on using RequestBroadcast=yes.
3362 * systemd-networkd will now set up IPv4LL addresses (when
3363 enabled) even if DHCP is configured successfully.
3365 * udev will now default to respect network device names given
3366 by the kernel when the kernel indicates that these are
3367 predictable. This behavior can be tweaked by changing
3368 NamePolicy= in the relevant .link file.
3370 * A new library systemd-terminal has been added that
3371 implements full TTY stream parsing and rendering. This
3372 library is supposed to be used later on for implementing a
3373 full userspace VT subsystem, replacing the current kernel
3376 * A new tool systemd-journal-upload has been added to push
3377 journal data to a remote system running
3378 systemd-journal-remote.
3380 * journald will no longer forward all local data to another
3381 running syslog daemon. This change has been made because
3382 rsyslog (which appears to be the most commonly used syslog
3383 implementation these days) no longer makes use of this, and
3384 instead pulls the data out of the journal on its own. Since
3385 forwarding the messages to a non-existent syslog server is
3386 more expensive than we assumed we have now turned this
3387 off. If you run a syslog server that is not a recent rsyslog
3388 version, you have to turn this option on again
3389 (ForwardToSyslog= in journald.conf).
3391 * journald now optionally supports the LZ4 compressor for
3392 larger journal fields. This compressor should perform much
3393 better than XZ which was the previous default.
3395 * machinectl now shows the IP addresses of local containers,
3396 if it knows them, plus the interface name of the container.
3398 * A new tool "systemd-escape" has been added that makes it
3399 easy to escape strings to build unit names and similar.
3401 * sd_notify() messages may now include a new ERRNO= field
3402 which is parsed and collected by systemd and shown among the
3403 "systemctl status" output for a service.
3405 * A new component "systemd-firstboot" has been added that
3406 queries the most basic systemd information (timezone,
3407 hostname, root password) interactively on first
3408 boot. Alternatively it may also be used to provision these
3409 things offline on OS images installed into directories.
3411 * The default sysctl.d/ snippets will now set
3413 net.ipv4.conf.default.promote_secondaries=1
3415 This has the benefit of no flushing secondary IP addresses
3416 when primary addresses are removed.
3418 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
3419 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
3420 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
3421 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
3422 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
3423 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
3424 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
3425 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
3426 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
3427 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
3428 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
3429 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
3430 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
3431 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
3432 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
3434 — Berlin, 2014-08-19
3438 * A new tool systemd-sysusers has been added. This tool
3439 creates system users and groups in /etc/passwd and
3440 /etc/group, based on static declarative system user/group
3441 definitions in /usr/lib/sysusers.d/. This is useful to
3442 enable factory resets and volatile systems that boot up with
3443 an empty /etc directory, and thus need system users and
3444 groups created during early boot. systemd now also ships
3445 with two default sysusers.d/ files for the most basic
3446 users and groups systemd and the core operating system
3449 * A new tmpfiles snippet has been added that rebuilds the
3450 essential files in /etc on boot, should they be missing.
3452 * A directive for ensuring automatic clean-up of
3453 /var/cache/man/ has been removed from the default
3454 configuration. This line should now be shipped by the man
3455 implementation. The necessary change has been made to the
3456 man-db implementation. Note that you need to update your man
3457 implementation to one that ships this line, otherwise no
3458 automatic clean-up of /var/cache/man will take place.
3460 * A new condition ConditionNeedsUpdate= has been added that
3461 may conditionalize services to only run when /etc or /var
3462 are "older" than the vendor operating system resources in
3463 /usr. This is useful for reconstructing or updating /etc
3464 after an offline update of /usr or a factory reset, on the
3465 next reboot. Services that want to run once after such an
3466 update or reset should use this condition and order
3467 themselves before the new systemd-update-done.service, which
3468 will mark the two directories as fully updated. A number of
3469 service files have been added making use of this, to rebuild
3470 the udev hardware database, the journald message catalog and
3471 dynamic loader cache (ldconfig). The systemd-sysusers tool
3472 described above also makes use of this now. With this in
3473 place it is now possible to start up a minimal operating
3474 system with /etc empty cleanly. For more information on the
3475 concepts involved see this recent blog story:
3477 http://0pointer.de/blog/projects/stateless.html
3479 * A new system group "input" has been introduced, and all
3480 input device nodes get this group assigned. This is useful
3481 for system-level software to get access to input devices. It
3482 complements what is already done for "audio" and "video".
3484 * systemd-networkd learnt minimal DHCPv4 server support in
3485 addition to the existing DHCPv4 client support. It also
3486 learnt DHCPv6 client and IPv6 Router Solicitation client
3487 support. The DHCPv4 client gained support for static routes
3488 passed in from the server. Note that the [DHCPv4] section
3489 known in older systemd-networkd versions has been renamed to
3490 [DHCP] and is now also used by the DHCPv6 client. Existing
3491 .network files using settings of this section should be
3492 updated, though compatibility is maintained. Optionally, the
3493 client hostname may now be sent to the DHCP server.
3495 * networkd gained support for vxlan virtual networks as well
3496 as tun/tap and dummy devices.
3498 * networkd gained support for automatic allocation of address
3499 ranges for interfaces from a system-wide pool of
3500 addresses. This is useful for dynamically managing a large
3501 number of interfaces with a single network configuration
3502 file. In particular this is useful to easily assign
3503 appropriate IP addresses to the veth links of a large number
3504 of nspawn instances.
3506 * RPM macros for processing sysusers, sysctl and binfmt
3507 drop-in snippets at package installation time have been
3510 * The /etc/os-release file should now be placed in
3511 /usr/lib/os-release. The old location is automatically
3512 created as symlink. /usr/lib is the more appropriate
3513 location of this file, since it shall actually describe the
3514 vendor operating system shipped in /usr, and not the
3515 configuration stored in /etc.
3517 * .mount units gained a new boolean SloppyOptions= setting
3518 that maps to mount(8)'s -s option which enables permissive
3519 parsing of unknown mount options.
3521 * tmpfiles learnt a new "L+" directive which creates a symlink
3522 but (unlike "L") deletes a pre-existing file first, should
3523 it already exist and not already be the correct
3524 symlink. Similarly, "b+", "c+" and "p+" directives have been
3525 added as well, which create block and character devices, as
3526 well as fifos in the filesystem, possibly removing any
3527 pre-existing files of different types.
3529 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
3530 'argument' field (which so far specified the source to
3531 symlink/copy the files from) is now optional. If omitted the
3532 same file os copied from /usr/share/factory/ suffixed by the
3533 full destination path. This is useful for populating /etc
3534 with essential files, by copying them from vendor defaults
3535 shipped in /usr/share/factory/etc.
3537 * A new command "systemctl preset-all" has been added that
3538 applies the service preset settings to all installed unit
3539 files. A new switch --preset-mode= has been added that
3540 controls whether only enable or only disable operations
3543 * A new command "systemctl is-system-running" has been added
3544 that allows checking the overall state of the system, for
3545 example whether it is fully up and running.
3547 * When the system boots up with an empty /etc, the equivalent
3548 to "systemctl preset-all" is executed during early boot, to
3549 make sure all default services are enabled after a factory
3552 * systemd now contains a minimal preset file that enables the
3553 most basic services systemd ships by default.
3555 * Unit files' [Install] section gained a new DefaultInstance=
3556 field for defining the default instance to create if a
3557 template unit is enabled with no instance specified.
3559 * A new passive target cryptsetup-pre.target has been added
3560 that may be used by services that need to make they run and
3561 finish before the first LUKS cryptographic device is set up.
3563 * The /dev/loop-control and /dev/btrfs-control device nodes
3564 are now owned by the "disk" group by default, opening up
3565 access to this group.
3567 * systemd-coredump will now automatically generate a
3568 stack trace of all core dumps taking place on the system,
3569 based on elfutils' libdw library. This stack trace is logged
3572 * systemd-coredump may now optionally store coredumps directly
3573 on disk (in /var/lib/systemd/coredump, possibly compressed),
3574 instead of storing them unconditionally in the journal. This
3575 mode is the new default. A new configuration file
3576 /etc/systemd/coredump.conf has been added to configure this
3577 and other parameters of systemd-coredump.
3579 * coredumpctl gained a new "info" verb to show details about a
3580 specific coredump. A new switch "-1" has also been added
3581 that makes sure to only show information about the most
3582 recent entry instead of all entries. Also, as the tool is
3583 generally useful now the "systemd-" prefix of the binary
3584 name has been removed. Distributions that want to maintain
3585 compatibility with the old name should add a symlink from
3586 the old name to the new name.
3588 * journald's SplitMode= now defaults to "uid". This makes sure
3589 that unprivileged users can access their own coredumps with
3590 coredumpctl without restrictions.
3592 * New kernel command line options "systemd.wants=" (for
3593 pulling an additional unit during boot), "systemd.mask="
3594 (for masking a specific unit for the boot), and
3595 "systemd.debug-shell" (for enabling the debug shell on tty9)
3596 have been added. This is implemented in the new generator
3597 "systemd-debug-generator".
3599 * systemd-nspawn will now by default filter a couple of
3600 syscalls for containers, among them those required for
3601 kernel module loading, direct x86 IO port access, swap
3602 management, and kexec. Most importantly though
3603 open_by_handle_at() is now prohibited for containers,
3604 closing a hole similar to a recently discussed vulnerability
3605 in docker regarding access to files on file hierarchies the
3606 container should normally not have access to. Note that, for
3607 nspawn, we generally make no security claims anyway (and
3608 this is explicitly documented in the man page), so this is
3609 just a fix for one of the most obvious problems.
3611 * A new man page file-hierarchy(7) has been added that
3612 contains a minimized, modernized version of the file system
3613 layout systemd expects, similar in style to the FHS
3614 specification or hier(5). A new tool systemd-path(1) has
3615 been added to query many of these paths for the local
3618 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
3619 longer done. Since the directory now has a per-user size
3620 limit, and is cleaned on logout this appears unnecessary,
3621 in particular since this now brings the lifecycle of this
3622 directory closer in line with how IPC objects are handled.
3624 * systemd.pc now exports a number of additional directories,
3625 including $libdir (which is useful to identify the library
3626 path for the primary architecture of the system), and a
3627 couple of drop-in directories.
3629 * udev's predictable network interface names now use the dev_port
3630 sysfs attribute, introduced in linux 3.15 instead of dev_id to
3631 distinguish between ports of the same PCI function. dev_id should
3632 only be used for ports using the same HW address, hence the need
3635 * machined has been updated to export the OS version of a
3636 container (read from /etc/os-release and
3637 /usr/lib/os-release) on the bus. This is now shown in
3638 "machinectl status" for a machine.
3640 * A new service setting RestartForceExitStatus= has been
3641 added. If configured to a set of exit signals or process
3642 return values, the service will be restarted when the main
3643 daemon process exits with any of them, regardless of the
3646 * systemctl's -H switch for connecting to remote systemd
3647 machines has been extended so that it may be used to
3648 directly connect to a specific container on the
3649 host. "systemctl -H root@foobar:waldi" will now connect as
3650 user "root" to host "foobar", and then proceed directly to
3651 the container named "waldi". Note that currently you have to
3652 authenticate as user "root" for this to work, as entering
3653 containers is a privileged operation.
3655 Contributions from: Andreas Henriksson, Benjamin Steinwender,
3656 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
3657 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
3658 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
3659 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
3660 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
3661 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
3662 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
3663 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
3664 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
3665 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
3666 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
3668 — Berlin, 2014-07-03
3672 * As an experimental feature, udev now tries to lock the
3673 disk device node (flock(LOCK_SH|LOCK_NB)) while it
3674 executes events for the disk or any of its partitions.
3675 Applications like partitioning programs can lock the
3676 disk device node (flock(LOCK_EX)) and claim temporary
3677 device ownership that way; udev will entirely skip all event
3678 handling for this disk and its partitions. If the disk
3679 was opened for writing, the close will trigger a partition
3680 table rescan in udev's "watch" facility, and if needed
3681 synthesize "change" events for the disk and all its partitions.
3682 This is now unconditionally enabled, and if it turns out to
3683 cause major problems, we might turn it on only for specific
3684 devices, or might need to disable it entirely. Device Mapper
3685 devices are excluded from this logic.
3687 * We temporarily dropped the "-l" switch for fsck invocations,
3688 since they collide with the flock() logic above. util-linux
3689 upstream has been changed already to avoid this conflict,
3690 and we will readd "-l" as soon as util-linux with this
3691 change has been released.
3693 * The dependency on libattr has been removed. Since a long
3694 time, the extended attribute calls have moved to glibc, and
3695 libattr is thus unnecessary.
3697 * Virtualization detection works without privileges now. This
3698 means the systemd-detect-virt binary no longer requires
3699 CAP_SYS_PTRACE file capabilities, and our daemons can run
3700 with fewer privileges.
3702 * systemd-networkd now runs under its own "systemd-network"
3703 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
3704 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
3705 loses the ability to write to files owned by root this way.
3707 * Similarly, systemd-resolved now runs under its own
3708 "systemd-resolve" user with no capabilities remaining.
3710 * Similarly, systemd-bus-proxyd now runs under its own
3711 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
3713 * systemd-networkd gained support for setting up "veth"
3714 virtual Ethernet devices for container connectivity, as well
3715 as GRE and VTI tunnels.
3717 * systemd-networkd will no longer automatically attempt to
3718 manually load kernel modules necessary for certain tunnel
3719 transports. Instead, it is assumed the kernel loads them
3720 automatically when required. This only works correctly on
3721 very new kernels. On older kernels, please consider adding
3722 the kernel modules to /etc/modules-load.d/ as a work-around.
3724 * The resolv.conf file systemd-resolved generates has been
3725 moved to /run/systemd/resolve/. If you have a symlink from
3726 /etc/resolv.conf, it might be necessary to correct it.
3728 * Two new service settings, ProtectHome= and ProtectSystem=,
3729 have been added. When enabled, they will make the user data
3730 (such as /home) inaccessible or read-only and the system
3731 (such as /usr) read-only, for specific services. This allows
3732 very light-weight per-service sandboxing to avoid
3733 modifications of user data or system files from
3734 services. These two new switches have been enabled for all
3735 of systemd's long-running services, where appropriate.
3737 * Socket units gained new SocketUser= and SocketGroup=
3738 settings to set the owner user and group of AF_UNIX sockets
3739 and FIFOs in the file system.
3741 * Socket units gained a new RemoveOnStop= setting. If enabled,
3742 all FIFOS and sockets in the file system will be removed
3743 when the specific socket unit is stopped.
3745 * Socket units gained a new Symlinks= setting. It takes a list
3746 of symlinks to create to file system sockets or FIFOs
3747 created by the specific Unix sockets. This is useful to
3748 manage symlinks to socket nodes with the same life-cycle as
3751 * The /dev/log socket and /dev/initctl FIFO have been moved to
3752 /run, and have been replaced by symlinks. This allows
3753 connecting to these facilities even if PrivateDevices=yes is
3754 used for a service (which makes /dev/log itself unavailable,
3755 but /run is left). This also has the benefit of ensuring
3756 that /dev only contains device nodes, directories and
3757 symlinks, and nothing else.
3759 * sd-daemon gained two new calls sd_pid_notify() and
3760 sd_pid_notifyf(). They are similar to sd_notify() and
3761 sd_notifyf(), but allow overriding of the source PID of
3762 notification messages if permissions permit this. This is
3763 useful to send notify messages on behalf of a different
3764 process (for example, the parent process). The
3765 systemd-notify tool has been updated to make use of this
3766 when sending messages (so that notification messages now
3767 originate from the shell script invoking systemd-notify and
3768 not the systemd-notify process itself. This should minimize
3769 a race where systemd fails to associate notification
3770 messages to services when the originating process already
3773 * A new "on-abnormal" setting for Restart= has been added. If
3774 set, it will result in automatic restarts on all "abnormal"
3775 reasons for a process to exit, which includes unclean
3776 signals, core dumps, timeouts and watchdog timeouts, but
3777 does not include clean and unclean exit codes or clean
3778 signals. Restart=on-abnormal is an alternative for
3779 Restart=on-failure for services that shall be able to
3780 terminate and avoid restarts on certain errors, by
3781 indicating so with an unclean exit code. Restart=on-failure
3782 or Restart=on-abnormal is now the recommended setting for
3783 all long-running services.
3785 * If the InaccessibleDirectories= service setting points to a
3786 mount point (or if there are any submounts contained within
3787 it), it is now attempted to completely unmount it, to make
3788 the file systems truly unavailable for the respective
3791 * The ReadOnlyDirectories= service setting and
3792 systemd-nspawn's --read-only parameter are now recursively
3793 applied to all submounts, too.
3795 * Mount units may now be created transiently via the bus APIs.
3797 * The support for SysV and LSB init scripts has been removed
3798 from the systemd daemon itself. Instead, it is now
3799 implemented as a generator that creates native systemd units
3800 from these scripts when needed. This enables us to remove a
3801 substantial amount of legacy code from PID 1, following the
3802 fact that many distributions only ship a very small number
3803 of LSB/SysV init scripts nowadays.
3805 * Privileged Xen (dom0) domains are not considered
3806 virtualization anymore by the virtualization detection
3807 logic. After all, they generally have unrestricted access to
3808 the hardware and usually are used to manage the unprivileged
3811 * systemd-tmpfiles gained a new "C" line type, for copying
3812 files or entire directories.
3814 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
3815 lines. So far, they have been non-globbing versions of the
3816 latter, and have thus been redundant. In future, it is
3817 recommended to only use "z". "m" has hence been removed
3818 from the documentation, even though it stays supported.
3820 * A tmpfiles snippet to recreate the most basic structure in
3821 /var has been added. This is enough to create the /var/run →
3822 /run symlink and create a couple of structural
3823 directories. This allows systems to boot up with an empty or
3824 volatile /var. Of course, while with this change, the core OS
3825 now is capable with dealing with a volatile /var, not all
3826 user services are ready for it. However, we hope that sooner
3827 or later, many service daemons will be changed upstream so
3828 that they are able to automatically create their necessary
3829 directories in /var at boot, should they be missing. This is
3830 the first step to allow state-less systems that only require
3831 the vendor image for /usr to boot.
3833 * systemd-nspawn has gained a new --tmpfs= switch to mount an
3834 empty tmpfs instance to a specific directory. This is
3835 particularly useful for making use of the automatic
3836 reconstruction of /var (see above), by passing --tmpfs=/var.
3838 * Access modes specified in tmpfiles snippets may now be
3839 prefixed with "~", which indicates that they shall be masked
3840 by whether the existing file or directory is currently
3841 writable, readable or executable at all. Also, if specified,
3842 the sgid/suid/sticky bits will be masked for all
3845 * A new passive target unit "network-pre.target" has been
3846 added which is useful for services that shall run before any
3847 network is configured, for example firewall scripts.
3849 * The "floppy" group that previously owned the /dev/fd*
3850 devices is no longer used. The "disk" group is now used
3851 instead. Distributions should probably deprecate usage of
3854 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
3855 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
3856 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
3857 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
3858 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
3859 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
3862 — Berlin, 2014-06-11
3866 * A new "systemd-timesyncd" daemon has been added for
3867 synchronizing the system clock across the network. It
3868 implements an SNTP client. In contrast to NTP
3869 implementations such as chrony or the NTP reference server,
3870 this only implements a client side, and does not bother with
3871 the full NTP complexity, focusing only on querying time from
3872 one remote server and synchronizing the local clock to
3873 it. Unless you intend to serve NTP to networked clients or
3874 want to connect to local hardware clocks, this simple NTP
3875 client should be more than appropriate for most
3876 installations. The daemon runs with minimal privileges, and
3877 has been hooked up with networkd to only operate when
3878 network connectivity is available. The daemon saves the
3879 current clock to disk every time a new NTP sync has been
3880 acquired, and uses this to possibly correct the system clock
3881 early at bootup, in order to accommodate for systems that
3882 lack an RTC such as the Raspberry Pi and embedded devices,
3883 and to make sure that time monotonically progresses on these
3884 systems, even if it is not always correct. To make use of
3885 this daemon, a new system user and group "systemd-timesync"
3886 needs to be created on installation of systemd.
3888 * The queue "seqnum" interface of libudev has been disabled, as
3889 it was generally incompatible with device namespacing as
3890 sequence numbers of devices go "missing" if the devices are
3891 part of a different namespace.
3893 * "systemctl list-timers" and "systemctl list-sockets" gained
3894 a --recursive switch for showing units of these types also
3895 for all local containers, similar in style to the already
3896 supported --recursive switch for "systemctl list-units".
3898 * A new RebootArgument= setting has been added for service
3899 units, which may be used to specify a kernel reboot argument
3900 to use when triggering reboots with StartLimitAction=.
3902 * A new FailureAction= setting has been added for service
3903 units which may be used to specify an operation to trigger
3904 when a service fails. This works similarly to
3905 StartLimitAction=, but unlike it, controls what is done
3906 immediately rather than only after several attempts to
3907 restart the service in question.
3909 * hostnamed got updated to also expose the kernel name,
3910 release, and version on the bus. This is useful for
3911 executing commands like hostnamectl with the -H switch.
3912 systemd-analyze makes use of this to properly display
3913 details when running non-locally.
3915 * The bootchart tool can now show cgroup information in the
3916 graphs it generates.
3918 * The CFS CPU quota cgroup attribute is now exposed for
3919 services. The new CPUQuota= switch has been added for this
3920 which takes a percentage value. Setting this will have the
3921 result that a service may never get more CPU time than the
3922 specified percentage, even if the machine is otherwise idle.
3924 * systemd-networkd learned IPIP and SIT tunnel support.
3926 * LSB init scripts exposing a dependency on $network will now
3927 get a dependency on network-online.target rather than simply
3928 network.target. This should bring LSB handling closer to
3929 what it was on SysV systems.
3931 * A new fsck.repair= kernel option has been added to control
3932 how fsck shall deal with unclean file systems at boot.
3934 * The (.ini) configuration file parser will now silently
3935 ignore sections whose name begins with "X-". This may be
3936 used to maintain application-specific extension sections in unit
3939 * machined gained a new API to query the IP addresses of
3940 registered containers. "machinectl status" has been updated
3941 to show these addresses in its output.
3943 * A new call sd_uid_get_display() has been added to the
3944 sd-login APIs for querying the "primary" session of a
3945 user. The "primary" session of the user is elected from the
3946 user's sessions and generally a graphical session is
3947 preferred over a text one.
3949 * A minimal systemd-resolved daemon has been added. It
3950 currently simply acts as a companion to systemd-networkd and
3951 manages resolv.conf based on per-interface DNS
3952 configuration, possibly supplied via DHCP. In the long run
3953 we hope to extend this into a local DNSSEC enabled DNS and
3956 * The systemd-networkd-wait-online tool is now enabled by
3957 default. It will delay network-online.target until a network
3958 connection has been configured. The tool primarily integrates
3959 with networkd, but will also make a best effort to make sense
3960 of network configuration performed in some other way.
3962 * Two new service options StartupCPUShares= and
3963 StartupBlockIOWeight= have been added that work similarly to
3964 CPUShares= and BlockIOWeight= however only apply during
3965 system startup. This is useful to prioritize certain services
3966 differently during bootup than during normal runtime.
3968 * hostnamed has been changed to prefer the statically
3969 configured hostname in /etc/hostname (unless set to
3970 'localhost' or empty) over any dynamic one supplied by
3971 dhcp. With this change, the rules for picking the hostname
3972 match more closely the rules of other configuration settings
3973 where the local administrator's configuration in /etc always
3974 overrides any other settings.
3976 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
3977 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
3978 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
3979 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
3980 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
3981 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
3982 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
3983 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
3984 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
3985 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
3986 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
3987 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
3988 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
3989 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
3990 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
3991 Lindskog, WaLyong Cho, Will Woods, Zbigniew
3994 — Beijing, 2014-05-28
3998 * When restoring the screen brightness at boot, stay away from
3999 the darkest setting or from the lowest 5% of the available
4000 range, depending on which is the larger value of both. This
4001 should effectively protect the user from rebooting into a
4002 black screen, should the brightness have been set to minimum
4005 * sd-login gained a new sd_machine_get_class() call to
4006 determine the class ("vm" or "container") of a machine
4007 registered with machined.
4009 * sd-login gained new calls
4010 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
4011 to query the identity of the peer of a local AF_UNIX
4012 connection. They operate similarly to their sd_pid_get_xyz()
4015 * PID 1 will now maintain a system-wide system state engine
4016 with the states "starting", "running", "degraded",
4017 "maintenance", "stopping". These states are bound to system
4018 startup, normal runtime, runtime with at least one failed
4019 service, rescue/emergency mode and system shutdown. This
4020 state is shown in the "systemctl status" output when no unit
4021 name is passed. It is useful to determine system state, in
4022 particularly when doing so for many systems or containers at
4025 * A new command "list-machines" has been added to "systemctl"
4026 that lists all local OS containers and shows their system
4027 state (see above), if systemd runs inside of them.
4029 * systemctl gained a new "-r" switch to recursively enumerate
4030 units on all local containers, when used with the
4031 "list-unit" command (which is the default one that is
4032 executed when no parameters are specified).
4034 * The GPT automatic partition discovery logic will now honour
4035 two GPT partition flags: one may be set on a partition to
4036 cause it to be mounted read-only, and the other may be set
4037 on a partition to ignore it during automatic discovery.
4039 * Two new GPT type UUIDs have been added for automatic root
4040 partition discovery, for 32-bit and 64-bit ARM. This is not
4041 particularly useful for discovering the root directory on
4042 these architectures during bare-metal boots (since UEFI is
4043 not common there), but still very useful to allow booting of
4044 ARM disk images in nspawn with the -i option.
4046 * MAC addresses of interfaces created with nspawn's
4047 --network-interface= switch will now be generated from the
4048 machine name, and thus be stable between multiple invocations
4051 * logind will now automatically remove all IPC objects owned
4052 by a user if she or he fully logs out. This makes sure that
4053 users who are logged out cannot continue to consume IPC
4054 resources. This covers SysV memory, semaphores and message
4055 queues as well as POSIX shared memory and message
4056 queues. Traditionally, SysV and POSIX IPC had no life-cycle
4057 limits. With this functionality, that is corrected. This may
4058 be turned off by using the RemoveIPC= switch of logind.conf.
4060 * The systemd-machine-id-setup and tmpfiles tools gained a
4061 --root= switch to operate on a specific root directory,
4064 * journald can now forward logged messages to the TTYs of all
4065 logged in users ("wall"). This is the default for all
4066 emergency messages now.
4068 * A new tool systemd-journal-remote has been added to stream
4069 journal log messages across the network.
4071 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
4072 controller trees are mounted into it. Note that the
4073 directories mounted beneath it are not read-only. This is a
4074 security measure and is particularly useful because glibc
4075 actually includes a search logic to pick any tmpfs it can
4076 find to implement shm_open() if /dev/shm is not available
4077 (which it might very well be in namespaced setups).
4079 * machinectl gained a new "poweroff" command to cleanly power
4080 down a local OS container.
4082 * The PrivateDevices= unit file setting will now also drop the
4083 CAP_MKNOD capability from the capability bound set, and
4084 imply DevicePolicy=closed.
4086 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
4087 comprehensively on all long-running systemd services where
4088 this is appropriate.
4090 * systemd-udevd will now run in a disassociated mount
4091 namespace. To mount directories from udev rules, make sure to
4092 pull in mount units via SYSTEMD_WANTS properties.
4094 * The kdbus support gained support for uploading policy into
4095 the kernel. sd-bus gained support for creating "monitoring"
4096 connections that can eavesdrop into all bus communication
4097 for debugging purposes.
4099 * Timestamps may now be specified in seconds since the UNIX
4100 epoch Jan 1st, 1970 by specifying "@" followed by the value
4103 * Native tcpwrap support in systemd has been removed. tcpwrap
4104 is old code, not really maintained anymore and has serious
4105 shortcomings, and better options such as firewalls
4106 exist. For setups that require tcpwrap usage, please
4107 consider invoking your socket-activated service via tcpd,
4108 like on traditional inetd.
4110 * A new system.conf configuration option
4111 DefaultTimerAccuracySec= has been added that controls the
4112 default AccuracySec= setting of .timer units.
4114 * Timer units gained a new WakeSystem= switch. If enabled,
4115 timers configured this way will cause the system to resume
4116 from system suspend (if the system supports that, which most
4119 * Timer units gained a new Persistent= switch. If enabled,
4120 timers configured this way will save to disk when they have
4121 been last triggered. This information is then used on next
4122 reboot to possible execute overdue timer events, that
4123 could not take place because the system was powered off.
4124 This enables simple anacron-like behaviour for timer units.
4126 * systemctl's "list-timers" will now also list the time a
4127 timer unit was last triggered in addition to the next time
4128 it will be triggered.
4130 * systemd-networkd will now assign predictable IPv4LL
4131 addresses to its local interfaces.
4133 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
4134 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
4135 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
4136 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
4137 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
4138 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
4139 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
4140 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
4143 — Berlin, 2014-03-25
4147 * A new unit file setting RestrictAddressFamilies= has been
4148 added to restrict which socket address families unit
4149 processes gain access to. This takes address family names
4150 like "AF_INET" or "AF_UNIX", and is useful to minimize the
4151 attack surface of services via exotic protocol stacks. This
4152 is built on seccomp system call filters.
4154 * Two new unit file settings RuntimeDirectory= and
4155 RuntimeDirectoryMode= have been added that may be used to
4156 manage a per-daemon runtime directories below /run. This is
4157 an alternative for setting up directory permissions with
4158 tmpfiles snippets, and has the advantage that the runtime
4159 directory's lifetime is bound to the daemon runtime and that
4160 the daemon starts up with an empty directory each time. This
4161 is particularly useful when writing services that drop
4162 privileges using the User= or Group= setting.
4164 * The DeviceAllow= unit setting now supports globbing for
4165 matching against device group names.
4167 * The systemd configuration file system.conf gained new
4168 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
4169 DefaultMemoryAccounting= to globally turn on/off accounting
4170 for specific resources (cgroups) for all units. These
4171 settings may still be overridden individually in each unit
4174 * systemd-gpt-auto-generator is now able to discover /srv and
4175 root partitions in addition to /home and swap partitions. It
4176 also supports LUKS-encrypted partitions now. With this in
4177 place, automatic discovery of partitions to mount following
4178 the Discoverable Partitions Specification
4179 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
4180 is now a lot more complete. This allows booting without
4181 /etc/fstab and without root= on the kernel command line on
4182 systems prepared appropriately.
4184 * systemd-nspawn gained a new --image= switch which allows
4185 booting up disk images and Linux installations on any block
4186 device that follow the Discoverable Partitions Specification
4187 (see above). This means that installations made with
4188 appropriately updated installers may now be started and
4189 deployed using container managers, completely
4190 unmodified. (We hope that libvirt-lxc will add support for
4191 this feature soon, too.)
4193 * systemd-nspawn gained a new --network-macvlan= setting to
4194 set up a private macvlan interface for the
4195 container. Similarly, systemd-networkd gained a new
4196 Kind=macvlan setting in .netdev files.
4198 * systemd-networkd now supports configuring local addresses
4201 * A new tool systemd-network-wait-online has been added to
4202 synchronously wait for network connectivity using
4205 * The sd-bus.h bus API gained a new sd_bus_track object for
4206 tracking the life-cycle of bus peers. Note that sd-bus.h is
4207 still not a public API though (unless you specify
4208 --enable-kdbus on the configure command line, which however
4209 voids your warranty and you get no API stability guarantee).
4211 * The $XDG_RUNTIME_DIR runtime directories for each user are
4212 now individual tmpfs instances, which has the benefit of
4213 introducing separate pools for each user, with individual
4214 size limits, and thus making sure that unprivileged clients
4215 can no longer negatively impact the system or other users by
4216 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
4217 RuntimeDirectorySize= has been introduced that allows
4218 controlling the default size limit for all users. It
4219 defaults to 10% of the available physical memory. This is no
4220 replacement for quotas on tmpfs though (which the kernel
4221 still does not support), as /dev/shm and /tmp are still
4222 shared resources used by both the system and unprivileged
4225 * logind will now automatically turn off automatic suspending
4226 on laptop lid close when more than one display is
4227 connected. This was previously expected to be implemented
4228 individually in desktop environments (such as GNOME),
4229 however has been added to logind now, in order to fix a
4230 boot-time race where a desktop environment might not have
4231 been started yet and thus not been able to take an inhibitor
4232 lock at the time where logind already suspends the system
4233 due to a closed lid.
4235 * logind will now wait at least 30s after each system
4236 suspend/resume cycle, and 3min after system boot before
4237 suspending the system due to a closed laptop lid. This
4238 should give USB docking stations and similar enough time to
4239 be probed and configured after system resume and boot in
4240 order to then act as suspend blocker.
4242 * systemd-run gained a new --property= setting which allows
4243 initialization of resource control properties (and others)
4244 for the created scope or service unit. Example: "systemd-run
4245 --property=BlockIOWeight=10 updatedb" may be used to run
4246 updatedb at a low block IO scheduling weight.
4248 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
4249 now also work in --scope mode.
4251 * When systemd is compiled with kdbus support, basic support
4252 for enforced policies is now in place. (Note that enabling
4253 kdbus still voids your warranty and no API compatibility
4256 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
4257 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
4258 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
4259 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
4260 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
4261 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
4262 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
4263 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
4264 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
4265 Zbigniew Jędrzejewski-Szmek
4267 — Berlin, 2014-03-12
4271 * systemd will now relabel /dev after loading the SMACK policy
4272 according to SMACK rules.
4274 * A new unit file option AppArmorProfile= has been added to
4275 set the AppArmor profile for the processes of a unit.
4277 * A new condition check ConditionArchitecture= has been added
4278 to conditionalize units based on the system architecture, as
4279 reported by uname()'s "machine" field.
4281 * systemd-networkd now supports matching on the system
4282 virtualization, architecture, kernel command line, host name
4285 * logind is now a lot more aggressive when suspending the
4286 machine due to a closed laptop lid. Instead of acting only
4287 on the lid close action, it will continuously watch the lid
4288 status and act on it. This is useful for laptops where the
4289 power button is on the outside of the chassis so that it can
4290 be reached without opening the lid (such as the Lenovo
4291 Yoga). On those machines, logind will now immediately
4292 re-suspend the machine if the power button has been
4293 accidentally pressed while the laptop was suspended and in a
4294 backpack or similar.
4296 * logind will now watch SW_DOCK switches and inhibit reaction
4297 to the lid switch if it is pressed. This means that logind
4298 will not suspend the machine anymore if the lid is closed
4299 and the system is docked, if the laptop supports SW_DOCK
4300 notifications via the input layer. Note that ACPI docking
4301 stations do not generate this currently. Also note that this
4302 logic is usually not fully sufficient and Desktop
4303 Environments should take a lid switch inhibitor lock when an
4304 external display is connected, as systemd will not watch
4307 * nspawn will now make use of the devices cgroup controller by
4308 default, and only permit creation of and access to the usual
4309 API device nodes like /dev/null or /dev/random, as well as
4310 access to (but not creation of) the pty devices.
4312 * We will now ship a default .network file for
4313 systemd-networkd that automatically configures DHCP for
4314 network interfaces created by nspawn's --network-veth or
4315 --network-bridge= switches.
4317 * systemd will now understand the usual M, K, G, T suffixes
4318 according to SI conventions (i.e. to the base 1000) when
4319 referring to throughput and hardware metrics. It will stay
4320 with IEC conventions (i.e. to the base 1024) for software
4321 metrics, according to what is customary according to
4322 Wikipedia. We explicitly document which base applies for
4323 each configuration option.
4325 * The DeviceAllow= setting in unit files now supports a syntax
4326 to whitelist an entire group of devices node majors at once,
4327 based on the /proc/devices listing. For example, with the
4328 string "char-pts", it is now possible to whitelist all
4329 current and future pseudo-TTYs at once.
4331 * sd-event learned a new "post" event source. Event sources of
4332 this type are triggered by the dispatching of any event
4333 source of a type that is not "post". This is useful for
4334 implementing clean-up and check event sources that are
4335 triggered by other work being done in the program.
4337 * systemd-networkd is no longer statically enabled, but uses
4338 the usual [Install] sections so that it can be
4339 enabled/disabled using systemctl. It still is enabled by
4342 * When creating a veth interface pair with systemd-nspawn, the
4343 host side will now be prefixed with "vb-" if
4344 --network-bridge= is used, and with "ve-" if --network-veth
4345 is used. This way, it is easy to distinguish these cases on
4346 the host, for example to apply different configuration to
4347 them with systemd-networkd.
4349 * The compatibility libraries for libsystemd-journal.so,
4350 libsystem-id128.so, libsystemd-login.so and
4351 libsystemd-daemon.so do not make use of IFUNC
4352 anymore. Instead, we now build libsystemd.so multiple times
4353 under these alternative names. This means that the footprint
4354 is drastically increased, but given that these are
4355 transitional compatibility libraries, this should not matter
4356 much. This change has been made necessary to support the ARM
4357 platform for these compatibility libraries, as the ARM
4358 toolchain is not really at the same level as the toolchain
4359 for other architectures like x86 and does not support
4360 IFUNC. Please make sure to use --enable-compat-libs only
4361 during a transitional period!
4363 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
4364 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
4365 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
4366 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
4367 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
4368 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
4369 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
4370 Zbigniew Jędrzejewski-Szmek
4372 — Berlin, 2014-02-24
4376 * A new component "systemd-networkd" has been added that can
4377 be used to configure local network interfaces statically or
4378 via DHCP. It is capable of bringing up bridges, VLANs, and
4379 bonding. Currently, no hook-ups for interactive network
4380 configuration are provided. Use this for your initrd,
4381 container, embedded, or server setup if you need a simple,
4382 yet powerful, network configuration solution. This
4383 configuration subsystem is quite nifty, as it allows wildcard
4384 hotplug matching in interfaces. For example, with a single
4385 configuration snippet, you can configure that all Ethernet
4386 interfaces showing up are automatically added to a bridge,
4387 or similar. It supports link-sensing and more.
4389 * A new tool "systemd-socket-proxyd" has been added which can
4390 act as a bidirectional proxy for TCP sockets. This is
4391 useful for adding socket activation support to services that
4392 do not actually support socket activation, including virtual
4393 machines and the like.
4395 * Add a new tool to save/restore rfkill state on
4398 * Save/restore state of keyboard backlights in addition to
4399 display backlights on shutdown/boot.
4401 * udev learned a new SECLABEL{} construct to label device
4402 nodes with a specific security label when they appear. For
4403 now, only SECLABEL{selinux} is supported, but the syntax is
4404 prepared for additional security frameworks.
4406 * udev gained a new scheme to configure link-level attributes
4407 from files in /etc/systemd/network/*.link. These files can
4408 match against MAC address, device path, driver name and type,
4409 and will apply attributes like the naming policy, link speed,
4410 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
4411 address assignment policy (randomized, ...).
4413 * The configuration of network interface naming rules for
4414 "permanent interface names" has changed: a new NamePolicy=
4415 setting in the [Link] section of .link files determines the
4416 priority of possible naming schemes (onboard, slot, MAC,
4417 path). The default value of this setting is determined by
4418 /usr/lib/net/links/99-default.link. Old
4419 80-net-name-slot.rules udev configuration file has been
4420 removed, so local configuration overriding this file should
4421 be adapted to override 99-default.link instead.
4423 * When the User= switch is used in a unit file, also
4424 initialize $SHELL= based on the user database entry.
4426 * systemd no longer depends on libdbus. All communication is
4427 now done with sd-bus, systemd's low-level bus library
4430 * kdbus support has been added to PID 1 itself. When kdbus is
4431 enabled, this causes PID 1 to set up the system bus and
4432 enable support for a new ".busname" unit type that
4433 encapsulates bus name activation on kdbus. It works a little
4434 bit like ".socket" units, except for bus names. A new
4435 generator has been added that converts classic dbus1 service
4436 activation files automatically into native systemd .busname
4439 * sd-bus: add a light-weight vtable implementation that allows
4440 defining objects on the bus with a simple static const
4441 vtable array of its methods, signals and properties.
4443 * systemd will not generate or install static dbus
4444 introspection data anymore to /usr/share/dbus-1/interfaces,
4445 as the precise format of these files is unclear, and
4446 nothing makes use of it.
4448 * A proxy daemon is now provided to proxy clients connecting
4449 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
4450 compatibility with classic D-Bus.
4452 * A bus driver implementation has been added that supports the
4453 classic D-Bus bus driver calls on kdbus, also for
4454 compatibility purposes.
4456 * A new API "sd-event.h" has been added that implements a
4457 minimal event loop API built around epoll. It provides a
4458 couple of features that direct epoll usage is lacking:
4459 prioritization of events, scales to large numbers of timer
4460 events, per-event timer slack (accuracy), system-wide
4461 coalescing of timer events, exit handlers, watchdog
4462 supervision support using systemd's sd_notify() API, child
4465 * A new API "sd-rntl.h" has been added that provides an API
4466 around the route netlink interface of the kernel, similar in
4467 style to "sd-bus.h".
4469 * A new API "sd-dhcp-client.h" has been added that provides a
4470 small DHCPv4 client-side implementation. This is used by
4473 * There is a new kernel command line option
4474 "systemd.restore_state=0|1". When set to "0", none of the
4475 systemd tools will restore saved runtime state to hardware
4476 devices. More specifically, the rfkill and backlight states
4479 * The FsckPassNo= compatibility option in mount/service units
4480 has been removed. The fstab generator will now add the
4481 necessary dependencies automatically, and does not require
4482 PID1's support for that anymore.
4484 * journalctl gained a new switch, --list-boots, that lists
4485 recent boots with their times and boot IDs.
4487 * The various tools like systemctl, loginctl, timedatectl,
4488 busctl, systemd-run, ... have gained a new switch "-M" to
4489 connect to a specific, local OS container (as direct
4490 connection, without requiring SSH). This works on any
4491 container that is registered with machined, such as those
4492 created by libvirt-lxc or nspawn.
4494 * systemd-run and systemd-analyze also gained support for "-H"
4495 to connect to remote hosts via SSH. This is particularly
4496 useful for systemd-run because it enables queuing of jobs
4497 onto remote systems.
4499 * machinectl gained a new command "login" to open a getty
4500 login in any local container. This works with any container
4501 that is registered with machined (such as those created by
4502 libvirt-lxc or nspawn), and which runs systemd inside.
4504 * machinectl gained a new "reboot" command that may be used to
4505 trigger a reboot on a specific container that is registered
4506 with machined. This works on any container that runs an init
4507 system of some kind.
4509 * systemctl gained a new "list-timers" command to print a nice
4510 listing of installed timer units with the times they elapse
4513 * Alternative reboot() parameters may now be specified on the
4514 "systemctl reboot" command line and are passed to the
4515 reboot() system call.
4517 * systemctl gained a new --job-mode= switch to configure the
4518 mode to queue a job with. This is a more generic version of
4519 --fail, --irreversible, and --ignore-dependencies, which are
4520 still available but not advertised anymore.
4522 * /etc/systemd/system.conf gained new settings to configure
4523 various default timeouts of units, as well as the default
4524 start limit interval and burst. These may still be overridden
4527 * PID1 will now export on the bus profile data of the security
4528 policy upload process (such as the SELinux policy upload to
4531 * journald: when forwarding logs to the console, include
4532 timestamps (following the setting in
4533 /sys/module/printk/parameters/time).
4535 * OnCalendar= in timer units now understands the special
4536 strings "yearly" and "annually". (Both are equivalent)
4538 * The accuracy of timer units is now configurable with the new
4539 AccuracySec= setting. It defaults to 1min.
4541 * A new dependency type JoinsNamespaceOf= has been added that
4542 allows running two services within the same /tmp and network
4543 namespace, if PrivateNetwork= or PrivateTmp= are used.
4545 * A new command "cat" has been added to systemctl. It outputs
4546 the original unit file of a unit, and concatenates the
4547 contents of additional "drop-in" unit file snippets, so that
4548 the full configuration is shown.
4550 * systemctl now supports globbing on the various "list-xyz"
4551 commands, like "list-units" or "list-sockets", as well as on
4552 those commands which take multiple unit names.
4554 * journalctl's --unit= switch gained support for globbing.
4556 * All systemd daemons now make use of the watchdog logic so
4557 that systemd automatically notices when they hang.
4559 * If the $container_ttys environment variable is set,
4560 getty-generator will automatically spawn a getty for each
4561 listed tty. This is useful for container managers to request
4562 login gettys to be spawned on as many ttys as needed.
4564 * %h, %s, %U specifier support is not available anymore when
4565 used in unit files for PID 1. This is because NSS calls are
4566 not safe from PID 1. They stay available for --user
4567 instances of systemd, and as special case for the root user.
4569 * loginctl gained a new "--no-legend" switch to turn off output
4572 * The "sd-login.h" API gained three new calls:
4573 sd_session_is_remote(), sd_session_get_remote_user(),
4574 sd_session_get_remote_host() to query information about
4577 * The udev hardware database now also carries vendor/product
4578 information of SDIO devices.
4580 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
4581 determine whether watchdog notifications are requested by
4584 * Socket-activated per-connection services now include a
4585 short description of the connection parameters in the
4588 * tmpfiles gained a new "--boot" option. When this is not used,
4589 only lines where the command character is not suffixed with
4590 "!" are executed. When this option is specified, those
4591 options are executed too. This partitions tmpfiles
4592 directives into those that can be safely executed at any
4593 time, and those which should be run only at boot (for
4594 example, a line that creates /run/nologin).
4596 * A new API "sd-resolve.h" has been added which provides a simple
4597 asynchronous wrapper around glibc NSS host name resolution
4598 calls, such as getaddrinfo(). In contrast to glibc's
4599 getaddrinfo_a(), it does not use signals. In contrast to most
4600 other asynchronous name resolution libraries, this one does
4601 not reimplement DNS, but reuses NSS, so that alternate
4602 host name resolution systems continue to work, such as mDNS,
4603 LDAP, etc. This API is based on libasyncns, but it has been
4604 cleaned up for inclusion in systemd.
4606 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
4607 "sd-daemon.h" are no longer found in individual libraries
4608 libsystemd-journal.so, libsystemd-login.so,
4609 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
4610 merged them into a single library, libsystemd.so, which
4611 provides all symbols. The reason for this is cyclic
4612 dependencies, as these libraries tend to use each other's
4613 symbols. So far, we have managed to workaround that by linking
4614 a copy of a good part of our code into each of these
4615 libraries again and again, which, however, makes certain
4616 things hard to do, like sharing static variables. Also, it
4617 substantially increases footprint. With this change, there
4618 is only one library for the basic APIs systemd
4619 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
4620 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
4621 library as well, however are subject to the --enable-kdbus
4622 switch (see below). Note that "sd-dhcp-client.h" is not part
4623 of this library (this is because it only consumes, never
4624 provides, services of/to other APIs). To make the transition
4625 easy from the separate libraries to the unified one, we
4626 provide the --enable-compat-libs compile-time switch which
4627 will generate stub libraries that are compatible with the
4628 old ones but redirect all calls to the new one.
4630 * All of the kdbus logic and the new APIs "sd-bus.h",
4631 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
4632 and "sd-utf8.h" are compile-time optional via the
4633 "--enable-kdbus" switch, and they are not compiled in by
4634 default. To make use of kdbus, you have to explicitly enable
4635 the switch. Note however, that neither the kernel nor the
4636 userspace API for all of this is considered stable yet. We
4637 want to maintain the freedom to still change the APIs for
4638 now. By specifying this build-time switch, you acknowledge
4639 that you are aware of the instability of the current
4642 * Also, note that while kdbus is pretty much complete,
4643 it lacks one thing: proper policy support. This means you
4644 can build a fully working system with all features; however,
4645 it will be highly insecure. Policy support will be added in
4646 one of the next releases, at the same time that we will
4647 declare the APIs stable.
4649 * When the kernel command line argument "kdbus" is specified,
4650 systemd will automatically load the kdbus.ko kernel module. At
4651 this stage of development, it is only useful for testing kdbus
4652 and should not be used in production. Note: if "--enable-kdbus"
4653 is specified, and the kdbus.ko kernel module is available, and
4654 "kdbus" is added to the kernel command line, the entire system
4655 runs with kdbus instead of dbus-daemon, with the above mentioned
4656 problem of missing the system policy enforcement. Also a future
4657 version of kdbus.ko or a newer systemd will not be compatible with
4658 each other, and will unlikely be able to boot the machine if only
4659 one of them is updated.
4661 * systemctl gained a new "import-environment" command which
4662 uploads the caller's environment (or parts thereof) into the
4663 service manager so that it is inherited by services started
4664 by the manager. This is useful to upload variables like
4665 $DISPLAY into the user service manager.
4667 * A new PrivateDevices= switch has been added to service units
4668 which allows running a service with a namespaced /dev
4669 directory that does not contain any device nodes for
4670 physical devices. More specifically, it only includes devices
4671 such as /dev/null, /dev/urandom, and /dev/zero which are API
4674 * logind has been extended to support behaviour like VT
4675 switching on seats that do not support a VT. This makes
4676 multi-session available on seats that are not the first seat
4677 (seat0), and on systems where kernel support for VTs has
4678 been disabled at compile-time.
4680 * If a process holds a delay lock for system sleep or shutdown
4681 and fails to release it in time, we will now log its
4682 identity. This makes it easier to identify processes that
4683 cause slow suspends or power-offs.
4685 * When parsing /etc/crypttab, support for a new key-slot=
4686 option as supported by Debian is added. It allows indicating
4687 which LUKS slot to use on disk, speeding up key loading.
4689 * The sd_journald_sendv() API call has been checked and
4690 officially declared to be async-signal-safe so that it may
4691 be invoked from signal handlers for logging purposes.
4693 * Boot-time status output is now enabled automatically after a
4694 short timeout if boot does not progress, in order to give
4695 the user an indication what she or he is waiting for.
4697 * The boot-time output has been improved to show how much time
4698 remains until jobs expire.
4700 * The KillMode= switch in service units gained a new possible
4701 value "mixed". If set, and the unit is shut down, then the
4702 initial SIGTERM signal is sent only to the main daemon
4703 process, while the following SIGKILL signal is sent to
4704 all remaining processes of the service.
4706 * When a scope unit is registered, a new property "Controller"
4707 may be set. If set to a valid bus name, systemd will send a
4708 RequestStop() signal to this name when it would like to shut
4709 down the scope. This may be used to hook manager logic into
4710 the shutdown logic of scope units. Also, scope units may now
4711 be put in a special "abandoned" state, in which case the
4712 manager process which created them takes no further
4713 responsibilities for it.
4715 * When reading unit files, systemd will now verify
4716 the access mode of these files, and warn about certain
4717 suspicious combinations. This has been added to make it
4718 easier to track down packaging bugs where unit files are
4719 marked executable or world-writable.
4721 * systemd-nspawn gained a new "--setenv=" switch to set
4722 container-wide environment variables. The similar option in
4723 systemd-activate was renamed from "--environment=" to
4724 "--setenv=" for consistency.
4726 * systemd-nspawn has been updated to create a new kdbus domain
4727 for each container that is invoked, thus allowing each
4728 container to have its own set of system and user buses,
4729 independent of the host.
4731 * systemd-nspawn gained a new --drop-capability= switch to run
4732 the container with less capabilities than the default. Both
4733 --drop-capability= and --capability= now take the special
4734 string "all" for dropping or keeping all capabilities.
4736 * systemd-nspawn gained new switches for executing containers
4737 with specific SELinux labels set.
4739 * systemd-nspawn gained a new --quiet switch to not generate
4740 any additional output but the container's own console
4743 * systemd-nspawn gained a new --share-system switch to run a
4744 container without PID namespacing enabled.
4746 * systemd-nspawn gained a new --register= switch to control
4747 whether the container is registered with systemd-machined or
4748 not. This is useful for containers that do not run full
4749 OS images, but only specific apps.
4751 * systemd-nspawn gained a new --keep-unit which may be used
4752 when invoked as the only program from a service unit, and
4753 results in registration of the unit service itself in
4754 systemd-machined, instead of a newly opened scope unit.
4756 * systemd-nspawn gained a new --network-interface= switch for
4757 moving arbitrary interfaces to the container. The new
4758 --network-veth switch creates a virtual Ethernet connection
4759 between host and container. The new --network-bridge=
4760 switch then allows assigning the host side of this virtual
4761 Ethernet connection to a bridge device.
4763 * systemd-nspawn gained a new --personality= switch for
4764 setting the kernel personality for the container. This is
4765 useful when running a 32-bit container on a 64-bit host. A
4766 similar option Personality= is now also available for service
4769 * logind will now also track a "Desktop" identifier for each
4770 session which encodes the desktop environment of it. This is
4771 useful for desktop environments that want to identify
4772 multiple running sessions of itself easily.
4774 * A new SELinuxContext= setting for service units has been
4775 added that allows setting a specific SELinux execution
4776 context for a service.
4778 * Most systemd client tools will now honour $SYSTEMD_LESS for
4779 settings of the "less" pager. By default, these tools will
4780 override $LESS to allow certain operations to work, such as
4781 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
4782 influence this logic.
4784 * systemd's "seccomp" hook-up has been changed to make use of
4785 the libseccomp library instead of using its own
4786 implementation. This has benefits for portability among
4789 * For usage together with SystemCallFilter=, a new
4790 SystemCallErrorNumber= setting has been introduced that
4791 allows configuration of a system error number to be returned
4792 on filtered system calls, instead of immediately killing the
4793 process. Also, SystemCallArchitectures= has been added to
4794 limit access to system calls of a particular architecture
4795 (in order to turn off support for unused secondary
4796 architectures). There is also a global
4797 SystemCallArchitectures= setting in system.conf now to turn
4798 off support for non-native system calls system-wide.
4800 * systemd requires a kernel with a working name_to_handle_at(),
4801 please see the kernel config requirements in the README file.
4803 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
4804 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
4805 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
4806 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
4807 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
4808 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
4809 Elia Pinto, Florian Weimer, George McCollister, Goffredo
4810 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
4811 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
4812 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
4813 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
4814 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
4815 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
4816 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
4817 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
4818 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
4819 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
4820 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
4821 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
4822 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
4823 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
4824 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
4825 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
4826 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
4828 — Berlin, 2014-02-20
4832 * logind has gained support for facilitating privileged input
4833 and drm device access for unprivileged clients. This work is
4834 useful to allow Wayland display servers (and similar
4835 programs, such as kmscon) to run under the user's ID and
4836 access input and drm devices which are normally
4837 protected. When this is used (and the kernel is new enough)
4838 logind will "mute" IO on the file descriptors passed to
4839 Wayland as long as it is in the background and "unmute" it
4840 if it returns into the foreground. This allows secure
4841 session switching without allowing background sessions to
4842 eavesdrop on input and display data. This also introduces
4843 session switching support if VT support is turned off in the
4844 kernel, and on seats that are not seat0.
4846 * A new kernel command line option luks.options= is understood
4847 now which allows specifying LUKS options for usage for LUKS
4848 encrypted partitions specified with luks.uuid=.
4850 * tmpfiles.d(5) snippets may now use specifier expansion in
4851 path names. More specifically %m, %b, %H, %v, are now
4852 replaced by the local machine id, boot id, hostname, and
4853 kernel version number.
4855 * A new tmpfiles.d(5) command "m" has been introduced which
4856 may be used to change the owner/group/access mode of a file
4857 or directory if it exists, but do nothing if it does not.
4859 * This release removes high-level support for the
4860 MemorySoftLimit= cgroup setting. The underlying kernel
4861 cgroup attribute memory.soft_limit= is currently badly
4862 designed and likely to be removed from the kernel API in its
4863 current form, hence we should not expose it for now.
4865 * The memory.use_hierarchy cgroup attribute is now enabled for
4866 all cgroups systemd creates in the memory cgroup
4867 hierarchy. This option is likely to be come the built-in
4868 default in the kernel anyway, and the non-hierarchical mode
4869 never made much sense in the intrinsically hierarchical
4872 * A new field _SYSTEMD_SLICE= is logged along with all journal
4873 messages containing the slice a message was generated
4874 from. This is useful to allow easy per-customer filtering of
4875 logs among other things.
4877 * systemd-journald will no longer adjust the group of journal
4878 files it creates to the "systemd-journal" group. Instead we
4879 rely on the journal directory to be owned by the
4880 "systemd-journal" group, and its setgid bit set, so that the
4881 kernel file system layer will automatically enforce that
4882 journal files inherit this group assignment. The reason for
4883 this change is that we cannot allow NSS look-ups from
4884 journald which would be necessary to resolve
4885 "systemd-journal" to a numeric GID, because this might
4886 create deadlocks if NSS involves synchronous queries to
4887 other daemons (such as nscd, or sssd) which in turn are
4888 logging clients of journald and might block on it, which
4889 would then dead lock. A tmpfiles.d(5) snippet included in
4890 systemd will make sure the setgid bit and group are
4891 properly set on the journal directory if it exists on every
4892 boot. However, we recommend adjusting it manually after
4893 upgrades too (or from RPM scriptlets), so that the change is
4894 not delayed until next reboot.
4896 * Backlight and random seed files in /var/lib/ have moved into
4897 the /var/lib/systemd/ directory, in order to centralize all
4898 systemd generated files in one directory.
4900 * Boot time performance measurements (as displayed by
4901 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
4902 performance information if that's available to determine how
4903 much time BIOS and boot loader initialization required. With
4904 a sufficiently new BIOS you hence no longer need to boot
4905 with Gummiboot to get access to such information.
4907 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
4908 Cristian Rodríguez, Dave Reisner, David Herrmann, David
4909 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
4910 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
4911 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
4912 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
4913 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
4915 — Berlin, 2013-10-02
4919 * The Restart= option for services now understands a new
4920 on-watchdog setting, which will restart the service
4921 automatically if the service stops sending out watchdog keep
4922 alive messages (as configured with WatchdogSec=).
4924 * The getty generator (which is responsible for bringing up a
4925 getty on configured serial consoles) will no longer only
4926 start a getty on the primary kernel console but on all
4927 others, too. This makes the order in which console= is
4928 specified on the kernel command line less important.
4930 * libsystemd-logind gained a new sd_session_get_vt() call to
4931 retrieve the VT number of a session.
4933 * If the option "tries=0" is set for an entry of /etc/crypttab
4934 its passphrase is queried indefinitely instead of any
4935 maximum number of tries.
4937 * If a service with a configure PID file terminates its PID
4938 file will now be removed automatically if it still exists
4939 afterwards. This should put an end to stale PID files.
4941 * systemd-run will now also take relative binary path names
4942 for execution and no longer insists on absolute paths.
4944 * InaccessibleDirectories= and ReadOnlyDirectories= now take
4945 paths that are optionally prefixed with "-" to indicate that
4946 it should not be considered a failure if they do not exist.
4948 * journalctl -o (and similar commands) now understands a new
4949 output mode "short-precise", it is similar to "short" but
4950 shows timestamps with usec accuracy.
4952 * The option "discard" (as known from Debian) is now
4953 synonymous to "allow-discards" in /etc/crypttab. In fact,
4954 "discard" is preferred now (since it is easier to remember
4957 * Some licensing clean-ups were made, so that more code is now
4958 LGPL-2.1 licensed than before.
4960 * A minimal tool to save/restore the display backlight
4961 brightness across reboots has been added. It will store the
4962 backlight setting as late as possible at shutdown, and
4963 restore it as early as possible during reboot.
4965 * A logic to automatically discover and enable home and swap
4966 partitions on GPT disks has been added. With this in place
4967 /etc/fstab becomes optional for many setups as systemd can
4968 discover certain partitions located on the root disk
4969 automatically. Home partitions are recognized under their
4970 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
4971 partitions are recognized under their GPT type ID
4972 0657fd6da4ab43c484e50933c84b4f4f.
4974 * systemd will no longer pass any environment from the kernel
4975 or initrd to system services. If you want to set an
4976 environment for all services, do so via the kernel command
4977 line systemd.setenv= assignment.
4979 * The systemd-sysctl tool no longer natively reads the file
4980 /etc/sysctl.conf. If desired, the file should be symlinked
4981 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
4982 legacy support by a symlink rather than built-in code, it
4983 also makes the otherwise hidden order of application of the
4984 different files visible. (Note that this partly reverts to a
4985 pre-198 application order of sysctl knobs!)
4987 * The "systemctl set-log-level" and "systemctl dump" commands
4988 have been moved to systemd-analyze.
4990 * systemd-run learned the new --remain-after-exit switch,
4991 which causes the scope unit not to be cleaned up
4992 automatically after the process terminated.
4994 * tmpfiles learned a new --exclude-prefix= switch to exclude
4995 certain paths from operation.
4997 * journald will now automatically flush all messages to disk
4998 as soon as a message at the log level CRIT, ALERT or EMERG
5001 Contributions from: Andrew Cook, Brandon Philips, Christian
5002 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
5003 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
5004 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
5005 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
5006 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
5007 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
5008 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
5009 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
5010 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
5011 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
5012 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
5013 William Giokas, Zbigniew Jędrzejewski-Szmek
5015 — Berlin, 2013-09-13
5019 * The documentation has been updated to cover the various new
5020 concepts introduced with 205.
5022 * Unit files now understand the new %v specifier which
5023 resolves to the kernel version string as returned by "uname
5026 * systemctl now supports filtering the unit list output by
5027 load state, active state and sub state, using the new
5030 * "systemctl status" will now show the results of the
5031 condition checks (like ConditionPathExists= and similar) of
5032 the last start attempts of the unit. They are also logged to
5035 * "journalctl -b" may now be used to look for boot output of a
5036 specific boot. Try "journalctl -b -1" for the previous boot,
5037 but the syntax is substantially more powerful.
5039 * "journalctl --show-cursor" has been added which prints the
5040 cursor string the last shown log line. This may then be used
5041 with the new "journalctl --after-cursor=" switch to continue
5042 browsing logs from that point on.
5044 * "journalctl --force" may now be used to force regeneration
5047 * Creation of "dead" device nodes has been moved from udev
5048 into kmod and tmpfiles. Previously, udev would read the kmod
5049 databases to pre-generate dead device nodes based on meta
5050 information contained in kernel modules, so that these would
5051 be auto-loaded on access rather then at boot. As this
5052 does not really have much to do with the exposing actual
5053 kernel devices to userspace this has always been slightly
5054 alien in the udev codebase. Following the new scheme kmod
5055 will now generate a runtime snippet for tmpfiles from the
5056 module meta information and it now is tmpfiles' job to the
5057 create the nodes. This also allows overriding access and
5058 other parameters for the nodes using the usual tmpfiles
5059 facilities. As side effect this allows us to remove the
5060 CAP_SYS_MKNOD capability bit from udevd entirely.
5062 * logind's device ACLs may now be applied to these "dead"
5063 devices nodes too, thus finally allowing managed access to
5064 devices such as /dev/snd/sequencer without loading the
5065 backing module right-away.
5067 * A new RPM macro has been added that may be used to apply
5068 tmpfiles configuration during package installation.
5070 * systemd-detect-virt and ConditionVirtualization= now can
5071 detect User-Mode-Linux machines (UML).
5073 * journald will now implicitly log the effective capabilities
5074 set of processes in the message metadata.
5076 * systemd-cryptsetup has gained support for TrueCrypt volumes.
5078 * The initrd interface has been simplified (more specifically,
5079 support for passing performance data via environment
5080 variables and fsck results via files in /run has been
5081 removed). These features were non-essential, and are
5082 nowadays available in a much nicer way by having systemd in
5083 the initrd serialize its state and have the hosts systemd
5084 deserialize it again.
5086 * The udev "keymap" data files and tools to apply keyboard
5087 specific mappings of scan to key codes, and force-release
5088 scan code lists have been entirely replaced by a udev
5089 "keyboard" builtin and a hwdb data file.
5091 * systemd will now honour the kernel's "quiet" command line
5092 argument also during late shutdown, resulting in a
5093 completely silent shutdown when used.
5095 * There's now an option to control the SO_REUSEPORT socket
5096 option in .socket units.
5098 * Instance units will now automatically get a per-template
5099 subslice of system.slice unless something else is explicitly
5100 configured. For example, instances of sshd@.service will now
5101 implicitly be placed in system-sshd.slice rather than
5102 system.slice as before.
5104 * Test coverage support may now be enabled at build time.
5106 Contributions from: Dave Reisner, Frederic Crozat, Harald
5107 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
5108 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
5109 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
5110 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
5111 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
5112 Giokas, Zbigniew Jędrzejewski-Szmek
5114 — Berlin, 2013-07-23
5118 * Two new unit types have been introduced:
5120 Scope units are very similar to service units, however, are
5121 created out of pre-existing processes — instead of PID 1
5122 forking off the processes. By using scope units it is
5123 possible for system services and applications to group their
5124 own child processes (worker processes) in a powerful way
5125 which then maybe used to organize them, or kill them
5126 together, or apply resource limits on them.
5128 Slice units may be used to partition system resources in an
5129 hierarchical fashion and then assign other units to them. By
5130 default there are now three slices: system.slice (for all
5131 system services), user.slice (for all user sessions),
5132 machine.slice (for VMs and containers).
5134 Slices and scopes have been introduced primarily in
5135 context of the work to move cgroup handling to a
5136 single-writer scheme, where only PID 1
5137 creates/removes/manages cgroups.
5139 * There's a new concept of "transient" units. In contrast to
5140 normal units these units are created via an API at runtime,
5141 not from configuration from disk. More specifically this
5142 means it is now possible to run arbitrary programs as
5143 independent services, with all execution parameters passed
5144 in via bus APIs rather than read from disk. Transient units
5145 make systemd substantially more dynamic then it ever was,
5146 and useful as a general batch manager.
5148 * logind has been updated to make use of scope and slice units
5149 for managing user sessions. As a user logs in he will get
5150 his own private slice unit, to which all sessions are added
5151 as scope units. We also added support for automatically
5152 adding an instance of user@.service for the user into the
5153 slice. Effectively logind will no longer create cgroup
5154 hierarchies on its own now, it will defer entirely to PID 1
5155 for this by means of scope, service and slice units. Since
5156 user sessions this way become entities managed by PID 1
5157 the output of "systemctl" is now a lot more comprehensive.
5159 * A new mini-daemon "systemd-machined" has been added which
5160 may be used by virtualization managers to register local
5161 VMs/containers. nspawn has been updated accordingly, and
5162 libvirt will be updated shortly. machined will collect a bit
5163 of meta information about the VMs/containers, and assign
5164 them their own scope unit (see above). The collected
5165 meta-data is then made available via the "machinectl" tool,
5166 and exposed in "ps" and similar tools. machined/machinectl
5167 is compile-time optional.
5169 * As discussed earlier, the low-level cgroup configuration
5170 options ControlGroup=, ControlGroupModify=,
5171 ControlGroupPersistent=, ControlGroupAttribute= have been
5172 removed. Please use high-level attribute settings instead as
5173 well as slice units.
5175 * A new bus call SetUnitProperties() has been added to alter
5176 various runtime parameters of a unit. This is primarily
5177 useful to alter cgroup parameters dynamically in a nice way,
5178 but will be extended later on to make more properties
5179 modifiable at runtime. systemctl gained a new set-properties
5180 command that wraps this call.
5182 * A new tool "systemd-run" has been added which can be used to
5183 run arbitrary command lines as transient services or scopes,
5184 while configuring a number of settings via the command
5185 line. This tool is currently very basic, however already
5186 very useful. We plan to extend this tool to even allow
5187 queuing of execution jobs with time triggers from the
5188 command line, similar in fashion to "at".
5190 * nspawn will now inform the user explicitly that kernels with
5191 audit enabled break containers, and suggest the user to turn
5194 * Support for detecting the IMA and AppArmor security
5195 frameworks with ConditionSecurity= has been added.
5197 * journalctl gained a new "-k" switch for showing only kernel
5198 messages, mimicking dmesg output; in addition to "--user"
5199 and "--system" switches for showing only user's own logs
5202 * systemd-delta can now show information about drop-in
5203 snippets extending unit files.
5205 * libsystemd-bus has been substantially updated but is still
5206 not available as public API.
5208 * systemd will now look for the "debug" argument on the kernel
5209 command line and enable debug logging, similar to what
5210 "systemd.log_level=debug" already did before.
5212 * "systemctl set-default", "systemctl get-default" has been
5213 added to configure the default.target symlink, which
5214 controls what to boot into by default.
5216 * "systemctl set-log-level" has been added as a convenient
5217 way to raise and lower systemd logging threshold.
5219 * "systemd-analyze plot" will now show the time the various
5220 generators needed for execution, as well as information
5221 about the unit file loading.
5223 * libsystemd-journal gained a new sd_journal_open_files() call
5224 for opening specific journal files. journactl also gained a
5225 new switch to expose this new functionality. Previously we
5226 only supported opening all files from a directory, or all
5227 files from the system, as opening individual files only is
5228 racy due to journal file rotation.
5230 * systemd gained the new DefaultEnvironment= setting in
5231 /etc/systemd/system.conf to set environment variables for
5234 * If a privileged process logs a journal message with the
5235 OBJECT_PID= field set, then journald will automatically
5236 augment this with additional OBJECT_UID=, OBJECT_GID=,
5237 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
5238 system services want to log events about specific client
5239 processes. journactl/systemctl has been updated to make use
5240 of this information if all log messages regarding a specific
5243 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
5244 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
5245 Reisner, David Coppa, David King, David Strauss, Eelco
5246 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
5247 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
5248 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
5249 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
5250 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
5251 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
5252 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
5253 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
5254 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
5255 Łukasz Stelmach, 장동준
5259 * The Python bindings gained some minimal support for the APIs
5260 exposed by libsystemd-logind.
5262 * ConditionSecurity= gained support for detecting SMACK. Since
5263 this condition already supports SELinux and AppArmor we only
5264 miss IMA for this. Patches welcome!
5266 Contributions from: Karol Lewandowski, Lennart Poettering,
5267 Zbigniew Jędrzejewski-Szmek
5271 * systemd-nspawn will now create /etc/resolv.conf if
5272 necessary, before bind-mounting the host's file onto it.
5274 * systemd-nspawn will now store meta information about a
5275 container on the container's cgroup as extended attribute
5276 fields, including the root directory.
5278 * The cgroup hierarchy has been reworked in many ways. All
5279 objects any of the components systemd creates in the cgroup
5280 tree are now suffixed. More specifically, user sessions are
5281 now placed in cgroups suffixed with ".session", users in
5282 cgroups suffixed with ".user", and nspawn containers in
5283 cgroups suffixed with ".nspawn". Furthermore, all cgroup
5284 names are now escaped in a simple scheme to avoid collision
5285 of userspace object names with kernel filenames. This work
5286 is preparation for making these objects relocatable in the
5287 cgroup tree, in order to allow easy resource partitioning of
5288 these objects without causing naming conflicts.
5290 * systemctl list-dependencies gained the new switches
5291 --plain, --reverse, --after and --before.
5293 * systemd-inhibit now shows the process name of processes that
5294 have taken an inhibitor lock.
5296 * nss-myhostname will now also resolve "localhost"
5297 implicitly. This makes /etc/hosts an optional file and
5298 nicely handles that on IPv6 ::1 maps to both "localhost" and
5301 * libsystemd-logind.so gained a new call
5302 sd_get_machine_names() to enumerate running containers and
5303 VMs (currently only supported by very new libvirt and
5304 nspawn). sd_login_monitor can now be used to watch
5305 VMs/containers coming and going.
5307 * .include is not allowed recursively anymore, and only in
5308 unit files. Usually it is better to use drop-in snippets in
5309 .d/*.conf anyway, as introduced with systemd 198.
5311 * systemd-analyze gained a new "critical-chain" command that
5312 determines the slowest chain of units run during system
5313 boot-up. It is very useful for tracking down where
5314 optimizing boot time is the most beneficial.
5316 * systemd will no longer allow manipulating service paths in
5317 the name=systemd:/system cgroup tree using ControlGroup= in
5318 units. (But is still fine with it in all other dirs.)
5320 * There's a new systemd-nspawn@.service service file that may
5321 be used to easily run nspawn containers as system
5322 services. With the container's root directory in
5323 /var/lib/container/foobar it is now sufficient to run
5324 "systemctl start systemd-nspawn@foobar.service" to boot it.
5326 * systemd-cgls gained a new parameter "--machine" to list only
5327 the processes within a certain container.
5329 * ConditionSecurity= now can check for "apparmor". We still
5330 are lacking checks for SMACK and IMA for this condition
5331 check though. Patches welcome!
5333 * A new configuration file /etc/systemd/sleep.conf has been
5334 added that may be used to configure which kernel operation
5335 systemd is supposed to execute when "suspend", "hibernate"
5336 or "hybrid-sleep" is requested. This makes the new kernel
5337 "freeze" state accessible to the user.
5339 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
5340 the passed argument if applicable.
5342 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
5343 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
5344 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
5345 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
5346 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
5347 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
5348 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
5353 * The output of 'systemctl list-jobs' got some polishing. The
5354 '--type=' argument may now be passed more than once. A new
5355 command 'systemctl list-sockets' has been added which shows
5356 a list of kernel sockets systemd is listening on with the
5357 socket units they belong to, plus the units these socket
5360 * The experimental libsystemd-bus library got substantial
5361 updates to work in conjunction with the (also experimental)
5362 kdbus kernel project. It works well enough to exchange
5363 messages with some sophistication. Note that kdbus is not
5364 ready yet, and the library is mostly an elaborate test case
5365 for now, and not installable.
5367 * systemd gained a new unit 'systemd-static-nodes.service'
5368 that generates static device nodes earlier during boot, and
5369 can run in conjunction with udev.
5371 * libsystemd-login gained a new call sd_pid_get_user_unit()
5372 to retrieve the user systemd unit a process is running
5373 in. This is useful for systems where systemd is used as
5376 * systemd-nspawn now places all containers in the new /machine
5377 top-level cgroup directory in the name=systemd
5378 hierarchy. libvirt will soon do the same, so that we get a
5379 uniform separation of /system, /user and /machine for system
5380 services, user processes and containers/virtual
5381 machines. This new cgroup hierarchy is also useful to stick
5382 stable names to specific container instances, which can be
5383 recognized later this way (this name may be controlled
5384 via systemd-nspawn's new -M switch). libsystemd-login also
5385 gained a new call sd_pid_get_machine_name() to retrieve the
5386 name of the container/VM a specific process belongs to.
5388 * bootchart can now store its data in the journal.
5390 * libsystemd-journal gained a new call
5391 sd_journal_add_conjunction() for AND expressions to the
5392 matching logic. This can be used to express more complex
5393 logical expressions.
5395 * journactl can now take multiple --unit= and --user-unit=
5398 * The cryptsetup logic now understands the "luks.key=" kernel
5399 command line switch for specifying a file to read the
5400 decryption key from. Also, if a configured key file is not
5401 found the tool will now automatically fall back to prompting
5404 * Python systemd.journal module was updated to wrap recently
5405 added functions from libsystemd-journal. The interface was
5406 changed to bring the low level interface in s.j._Reader
5407 closer to the C API, and the high level interface in
5408 s.j.Reader was updated to wrap and convert all data about
5411 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
5412 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
5413 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
5414 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
5415 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
5416 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5420 * journalctl --update-catalog now understands a new --root=
5421 option to operate on catalogs found in a different root
5424 * During shutdown after systemd has terminated all running
5425 services a final killing loop kills all remaining left-over
5426 processes. We will now print the name of these processes
5427 when we send SIGKILL to them, since this usually indicates a
5430 * If /etc/crypttab refers to password files stored on
5431 configured mount points automatic dependencies will now be
5432 generated to ensure the specific mount is established first
5433 before the key file is attempted to be read.
5435 * 'systemctl status' will now show information about the
5436 network sockets a socket unit is listening on.
5438 * 'systemctl status' will also shown information about any
5439 drop-in configuration file for units. (Drop-In configuration
5440 files in this context are files such as
5441 /etc/systemd/systemd/foobar.service.d/*.conf)
5443 * systemd-cgtop now optionally shows summed up CPU times of
5444 cgroups. Press '%' while running cgtop to switch between
5445 percentage and absolute mode. This is useful to determine
5446 which cgroups use up the most CPU time over the entire
5447 runtime of the system. systemd-cgtop has also been updated
5448 to be 'pipeable' for processing with further shell tools.
5450 * 'hostnamectl set-hostname' will now allow setting of FQDN
5453 * The formatting and parsing of time span values has been
5454 changed. The parser now understands fractional expressions
5455 such as "5.5h". The formatter will now output fractional
5456 expressions for all time spans under 1min, i.e. "5.123456s"
5457 rather than "5s 123ms 456us". For time spans under 1s
5458 millisecond values are shown, for those under 1ms
5459 microsecond values are shown. This should greatly improve
5460 all time-related output of systemd.
5462 * libsystemd-login and libsystemd-journal gained new
5463 functions for querying the poll() events mask and poll()
5464 timeout value for integration into arbitrary event
5467 * localectl gained the ability to list available X11 keymaps
5468 (models, layouts, variants, options).
5470 * 'systemd-analyze dot' gained the ability to filter for
5471 specific units via shell-style globs, to create smaller,
5472 more useful graphs. I.e. it is now possible to create simple
5473 graphs of all the dependencies between only target units, or
5474 of all units that Avahi has dependencies with.
5476 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
5477 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
5478 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
5479 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
5480 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
5481 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
5482 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
5486 * The boot-time readahead implementation for rotating media
5487 will now read the read-ahead data in multiple passes which
5488 consist of all read requests made in equidistant time
5489 intervals. This means instead of strictly reading read-ahead
5490 data in its physical order on disk we now try to find a
5491 middle ground between physical and access time order.
5493 * /etc/os-release files gained a new BUILD_ID= field for usage
5494 on operating systems that provide continuous builds of OS
5497 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
5498 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
5499 William Douglas, Zbigniew Jędrzejewski-Szmek
5503 * systemd-python gained an API exposing libsystemd-daemon.
5505 * The SMACK setup logic gained support for uploading CIPSO
5508 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
5509 ReadOnlyDirectories= and InaccessibleDirectories= has
5510 changed. The private /tmp and /var/tmp directories are now
5511 shared by all processes of a service (which means
5512 ExecStartPre= may now leave data in /tmp that ExecStart= of
5513 the same service can still access). When a service is
5514 stopped its temporary directories are immediately deleted
5515 (normal clean-up with tmpfiles is still done in addition to
5518 * By default, systemd will now set a couple of sysctl
5519 variables in the kernel: the safe sysrq options are turned
5520 on, IP route verification is turned on, and source routing
5521 disabled. The recently added hardlink and softlink
5522 protection of the kernel is turned on. These settings should
5523 be reasonably safe, and good defaults for all new systems.
5525 * The predictable network naming logic may now be turned off
5526 with a new kernel command line switch: net.ifnames=0.
5528 * A new libsystemd-bus module has been added that implements a
5529 pretty complete D-Bus client library. For details see:
5531 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
5533 * journald will now explicitly flush the journal files to disk
5534 at the latest 5min after each write. The file will then also
5535 be marked offline until the next write. This should increase
5536 reliability in case of a crash. The synchronization delay
5537 can be configured via SyncIntervalSec= in journald.conf.
5539 * There's a new remote-fs-setup.target unit that can be used
5540 to pull in specific services when at least one remote file
5541 system is to be mounted.
5543 * There are new targets timers.target and paths.target as
5544 canonical targets to pull user timer and path units in
5545 from. This complements sockets.target with a similar
5546 purpose for socket units.
5548 * libudev gained a new call udev_device_set_attribute_value()
5549 to set sysfs attributes of a device.
5551 * The udev daemon now sets the default number of worker
5552 processes executed in parallel based on the number of available
5553 CPUs instead of the amount of available RAM. This is supposed
5554 to provide a more reliable default and limit a too aggressive
5555 parallelism for setups with 1000s of devices connected.
5557 Contributions from: Auke Kok, Colin Walters, Cristian
5558 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
5559 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
5560 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
5561 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
5562 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
5563 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
5564 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
5565 Zbigniew Jędrzejewski-Szmek
5569 * Configuration of unit files may now be extended via drop-in
5570 files without having to edit/override the unit files
5571 themselves. More specifically, if the administrator wants to
5572 change one value for a service file foobar.service he can
5573 now do so by dropping in a configuration snippet into
5574 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
5575 will load all these snippets and apply them on top of the
5576 main unit configuration file, possibly extending or
5577 overriding its settings. Using these drop-in snippets is
5578 generally nicer than the two earlier options for changing
5579 unit files locally: copying the files from
5580 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
5581 them there; or creating a new file in /etc/systemd/system/
5582 that incorporates the original one via ".include". Drop-in
5583 snippets into these .d/ directories can be placed in any
5584 directory systemd looks for units in, and the usual
5585 overriding semantics between /usr/lib, /etc and /run apply
5588 * Most unit file settings which take lists of items can now be
5589 reset by assigning the empty string to them. For example,
5590 normally, settings such as Environment=FOO=BAR append a new
5591 environment variable assignment to the environment block,
5592 each time they are used. By assigning Environment= the empty
5593 string the environment block can be reset to empty. This is
5594 particularly useful with the .d/*.conf drop-in snippets
5595 mentioned above, since this adds the ability to reset list
5596 settings from vendor unit files via these drop-ins.
5598 * systemctl gained a new "list-dependencies" command for
5599 listing the dependencies of a unit recursively.
5601 * Inhibitors are now honored and listed by "systemctl
5602 suspend", "systemctl poweroff" (and similar) too, not only
5603 GNOME. These commands will also list active sessions by
5606 * Resource limits (as exposed by the various control group
5607 controllers) can now be controlled dynamically at runtime
5608 for all units. More specifically, you can now use a command
5609 like "systemctl set-cgroup-attr foobar.service cpu.shares
5610 2000" to alter the CPU shares a specific service gets. These
5611 settings are stored persistently on disk, and thus allow the
5612 administrator to easily adjust the resource usage of
5613 services with a few simple commands. This dynamic resource
5614 management logic is also available to other programs via the
5615 bus. Almost any kernel cgroup attribute and controller is
5618 * systemd-vconsole-setup will now copy all font settings to
5619 all allocated VTs, where it previously applied them only to
5622 * libsystemd-login gained the new sd_session_get_tty() API
5625 * This release drops support for a few legacy or
5626 distribution-specific LSB facility names when parsing init
5627 scripts: $x-display-manager, $mail-transfer-agent,
5628 $mail-transport-agent, $mail-transfer-agent, $smtp,
5629 $null. Also, the mail-transfer-agent.target unit backing
5630 this has been removed. Distributions which want to retain
5631 compatibility with this should carry the burden for
5632 supporting this themselves and patch support for these back
5633 in, if they really need to. Also, the facilities $syslog and
5634 $local_fs are now ignored, since systemd does not support
5635 early-boot LSB init scripts anymore, and these facilities
5636 are implied anyway for normal services. syslog.target has
5639 * There are new bus calls on PID1's Manager object for
5640 cancelling jobs, and removing snapshot units. Previously,
5641 both calls were only available on the Job and Snapshot
5644 * systemd-journal-gatewayd gained SSL support.
5646 * The various "environment" files, such as /etc/locale.conf
5647 now support continuation lines with a backslash ("\") as
5648 last character in the line, similarly in style (but different)
5649 to how this is supported in shells.
5651 * For normal user processes the _SYSTEMD_USER_UNIT= field is
5652 now implicitly appended to every log entry logged. systemctl
5653 has been updated to filter by this field when operating on a
5654 user systemd instance.
5656 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
5657 CAP_AUDIT_CONTROL capabilities to the capabilities set for
5658 the container. This makes it easier to boot unmodified
5659 Fedora systems in a container, which however still requires
5660 audit=0 to be passed on the kernel command line. Auditing in
5661 kernel and userspace is unfortunately still too broken in
5662 context of containers, hence we recommend compiling it out
5663 of the kernel or using audit=0. Hopefully this will be fixed
5664 one day for good in the kernel.
5666 * nspawn gained the new --bind= and --bind-ro= parameters to
5667 bind mount specific directories from the host into the
5670 * nspawn will now mount its own devpts file system instance
5671 into the container, in order not to leak pty devices from
5672 the host into the container.
5674 * systemd will now read the firmware boot time performance
5675 information from the EFI variables, if the used boot loader
5676 supports this, and takes it into account for boot performance
5677 analysis via "systemd-analyze". This is currently supported
5678 only in conjunction with Gummiboot, but could be supported
5679 by other boot loaders too. For details see:
5681 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
5683 * A new generator has been added that automatically mounts the
5684 EFI System Partition (ESP) to /boot, if that directory
5685 exists, is empty, and no other file system has been
5686 configured to be mounted there.
5688 * logind will now send out PrepareForSleep(false) out
5689 unconditionally, after coming back from suspend. This may be
5690 used by applications as asynchronous notification for
5691 system resume events.
5693 * "systemctl unlock-sessions" has been added, that allows
5694 unlocking the screens of all user sessions at once, similar
5695 to how "systemctl lock-sessions" already locked all users
5696 sessions. This is backed by a new D-Bus call UnlockSessions().
5698 * "loginctl seat-status" will now show the master device of a
5699 seat. (i.e. the device of a seat that needs to be around for
5700 the seat to be considered available, usually the graphics
5703 * tmpfiles gained a new "X" line type, that allows
5704 configuration of files and directories (with wildcards) that
5705 shall be excluded from automatic cleanup ("aging").
5707 * udev default rules set the device node permissions now only
5708 at "add" events, and do not change them any longer with a
5709 later "change" event.
5711 * The log messages for lid events and power/sleep keypresses
5712 now carry a message ID.
5714 * We now have a substantially larger unit test suite, but this
5715 continues to be work in progress.
5717 * udevadm hwdb gained a new --root= parameter to change the
5718 root directory to operate relative to.
5720 * logind will now issue a background sync() request to the kernel
5721 early at shutdown, so that dirty buffers are flushed to disk early
5722 instead of at the last moment, in order to optimize shutdown
5725 * A new bootctl tool has been added that is an interface for
5726 certain boot loader operations. This is currently a preview
5727 and is likely to be extended into a small mechanism daemon
5728 like timedated, localed, hostnamed, and can be used by
5729 graphical UIs to enumerate available boot options, and
5730 request boot into firmware operations.
5732 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
5733 the rest of the package. It also has been updated to work
5734 correctly in initrds.
5736 * Policykit previously has been runtime optional, and is now
5737 also compile time optional via a configure switch.
5739 * systemd-analyze has been reimplemented in C. Also "systemctl
5740 dot" has moved into systemd-analyze.
5742 * "systemctl status" with no further parameters will now print
5743 the status of all active or failed units.
5745 * Operations such as "systemctl start" can now be executed
5746 with a new mode "--irreversible" which may be used to queue
5747 operations that cannot accidentally be reversed by a later
5748 job queuing. This is by default used to make shutdown
5749 requests more robust.
5751 * The Python API of systemd now gained a new module for
5752 reading journal files.
5754 * A new tool kernel-install has been added that can install
5755 kernel images according to the Boot Loader Specification:
5757 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
5759 * Boot time console output has been improved to provide
5760 animated boot time output for hanging jobs.
5762 * A new tool systemd-activate has been added which can be used
5763 to test socket activation with, directly from the command
5764 line. This should make it much easier to test and debug
5765 socket activation in daemons.
5767 * journalctl gained a new "--reverse" (or -r) option to show
5768 journal output in reverse order (i.e. newest line first).
5770 * journalctl gained a new "--pager-end" (or -e) option to jump
5771 to immediately jump to the end of the journal in the
5772 pager. This is only supported in conjunction with "less".
5774 * journalctl gained a new "--user-unit=" option, that works
5775 similarly to "--unit=" but filters for user units rather than
5778 * A number of unit files to ease adoption of systemd in
5779 initrds has been added. This moves some minimal logic from
5780 the various initrd implementations into systemd proper.
5782 * The journal files are now owned by a new group
5783 "systemd-journal", which exists specifically to allow access
5784 to the journal, and nothing else. Previously, we used the
5785 "adm" group for that, which however possibly covers more
5786 than just journal/log file access. This new group is now
5787 already used by systemd-journal-gatewayd to ensure this
5788 daemon gets access to the journal files and as little else
5789 as possible. Note that "make install" will also set FS ACLs
5790 up for /var/log/journal to give "adm" and "wheel" read
5791 access to it, in addition to "systemd-journal" which owns
5792 the journal files. We recommend that packaging scripts also
5793 add read access to "adm" + "wheel" to /var/log/journal, and
5794 all existing/future journal files. To normal users and
5795 administrators little changes, however packagers need to
5796 ensure to create the "systemd-journal" system group at
5797 package installation time.
5799 * The systemd-journal-gatewayd now runs as unprivileged user
5800 systemd-journal-gateway:systemd-journal-gateway. Packaging
5801 scripts need to create these system user/group at
5804 * timedated now exposes a new boolean property CanNTP that
5805 indicates whether a local NTP service is available or not.
5807 * systemd-detect-virt will now also detect xen PVs
5809 * The pstore file system is now mounted by default, if it is
5812 * In addition to the SELinux and IMA policies we will now also
5813 load SMACK policies at early boot.
5815 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
5816 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
5817 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
5818 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
5819 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
5820 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
5821 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
5822 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
5823 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
5824 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
5825 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
5826 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
5827 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
5828 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
5832 * Timer units now support calendar time events in addition to
5833 monotonic time events. That means you can now trigger a unit
5834 based on a calendar time specification such as "Thu,Fri
5835 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
5836 or fifth day of any month of the year 2013, given that it is
5837 a thursday or friday. This brings timer event support
5838 considerably closer to cron's capabilities. For details on
5839 the supported calendar time specification language see
5842 * udev now supports a number of different naming policies for
5843 network interfaces for predictable names, and a combination
5844 of these policies is now the default. Please see this wiki
5845 document for details:
5847 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
5849 * Auke Kok's bootchart implementation has been added to the
5850 systemd tree. It is an optional component that can graph the
5851 boot in quite some detail. It is one of the best bootchart
5852 implementations around and minimal in its code and
5855 * nss-myhostname has been integrated into the systemd source
5856 tree. nss-myhostname guarantees that the local hostname
5857 always stays resolvable via NSS. It has been a weak
5858 requirement of systemd-hostnamed since a long time, and
5859 since its code is actually trivial we decided to just
5860 include it in systemd's source tree. It can be turned off
5861 with a configure switch.
5863 * The read-ahead logic is now capable of properly detecting
5864 whether a btrfs file system is on SSD or rotating media, in
5865 order to optimize the read-ahead scheme. Previously, it was
5866 only capable of detecting this on traditional file systems
5869 * In udev, additional device properties are now read from the
5870 IAB in addition to the OUI database. Also, Bluetooth company
5871 identities are attached to the devices as well.
5873 * In service files %U may be used as specifier that is
5874 replaced by the configured user name of the service.
5876 * nspawn may now be invoked without a controlling TTY. This
5877 makes it suitable for invocation as its own service. This
5878 may be used to set up a simple containerized server system
5879 using only core OS tools.
5881 * systemd and nspawn can now accept socket file descriptors
5882 when they are started for socket activation. This enables
5883 implementation of socket activated nspawn
5884 containers. i.e. think about autospawning an entire OS image
5885 when the first SSH or HTTP connection is received. We expect
5886 that similar functionality will also be added to libvirt-lxc
5889 * journalctl will now suppress ANSI color codes when
5890 presenting log data.
5892 * systemctl will no longer show control group information for
5893 a unit if the control group is empty anyway.
5895 * logind can now automatically suspend/hibernate/shutdown the
5898 * /etc/machine-info and hostnamed now also expose the chassis
5899 type of the system. This can be used to determine whether
5900 the local system is a laptop, desktop, handset or
5901 tablet. This information may either be configured by the
5902 user/vendor or is automatically determined from ACPI and DMI
5903 information if possible.
5905 * A number of PolicyKit actions are now bound together with
5906 "imply" rules. This should simplify creating UIs because
5907 many actions will now authenticate similar ones as well.
5909 * Unit files learnt a new condition ConditionACPower= which
5910 may be used to conditionalize a unit depending on whether an
5911 AC power source is connected or not, of whether the system
5912 is running on battery power.
5914 * systemctl gained a new "is-failed" verb that may be used in
5915 shell scripts and suchlike to check whether a specific unit
5916 is in the "failed" state.
5918 * The EnvironmentFile= setting in unit files now supports file
5919 globbing, and can hence be used to easily read a number of
5920 environment files at once.
5922 * systemd will no longer detect and recognize specific
5923 distributions. All distribution-specific #ifdeffery has been
5924 removed, systemd is now fully generic and
5925 distribution-agnostic. Effectively, not too much is lost as
5926 a lot of the code is still accessible via explicit configure
5927 switches. However, support for some distribution specific
5928 legacy configuration file formats has been dropped. We
5929 recommend distributions to simply adopt the configuration
5930 files everybody else uses now and convert the old
5931 configuration from packaging scripts. Most distributions
5932 already did that. If that's not possible or desirable,
5933 distributions are welcome to forward port the specific
5934 pieces of code locally from the git history.
5936 * When logging a message about a unit systemd will now always
5937 log the unit name in the message meta data.
5939 * localectl will now also discover system locale data that is
5940 not stored in locale archives, but directly unpacked.
5942 * logind will no longer unconditionally use framebuffer
5943 devices as seat masters, i.e. as devices that are required
5944 to be existing before a seat is considered preset. Instead,
5945 it will now look for all devices that are tagged as
5946 "seat-master" in udev. By default, framebuffer devices will
5947 be marked as such, but depending on local systems, other
5948 devices might be marked as well. This may be used to
5949 integrate graphics cards using closed source drivers (such
5950 as NVidia ones) more nicely into logind. Note however, that
5951 we recommend using the open source NVidia drivers instead,
5952 and no udev rules for the closed-source drivers will be
5953 shipped from us upstream.
5955 Contributions from: Adam Williamson, Alessandro Crismani, Auke
5956 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
5957 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
5958 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
5959 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
5960 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
5961 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
5962 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
5963 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
5964 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
5965 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
5966 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
5971 * udev gained support for loading additional device properties
5972 from an indexed database that is keyed by vendor/product IDs
5973 and similar device identifiers. For the beginning this
5974 "hwdb" is populated with data from the well-known PCI and
5975 USB database, but also includes PNP, ACPI and OID data. In
5976 the longer run this indexed database shall grow into
5977 becoming the one central database for non-essential
5978 userspace device metadata. Previously, data from the PCI/USB
5979 database was only attached to select devices, since the
5980 lookup was a relatively expensive operation due to O(n) time
5981 complexity (with n being the number of entries in the
5982 database). Since this is now O(1), we decided to add in this
5983 data for all devices where this is available, by
5984 default. Note that the indexed database needs to be rebuilt
5985 when new data files are installed. To achieve this you need
5986 to update your packaging scripts to invoke "udevadm hwdb
5987 --update" after installation of hwdb data files. For
5988 RPM-based distributions we introduced the new
5989 %udev_hwdb_update macro for this purpose.
5991 * The Journal gained support for the "Message Catalog", an
5992 indexed database to link up additional information with
5993 journal entries. For further details please check:
5995 https://www.freedesktop.org/wiki/Software/systemd/catalog
5997 The indexed message catalog database also needs to be
5998 rebuilt after installation of message catalog files. Use
5999 "journalctl --update-catalog" for this. For RPM-based
6000 distributions we introduced the %journal_catalog_update
6001 macro for this purpose.
6003 * The Python Journal bindings gained support for the standard
6004 Python logging framework.
6006 * The Journal API gained new functions for checking whether
6007 the underlying file system of a journal file is capable of
6008 properly reporting file change notifications, or whether
6009 applications that want to reflect journal changes "live"
6010 need to recheck journal files continuously in appropriate
6013 * It is now possible to set the "age" field for tmpfiles
6014 entries to 0, indicating that files matching this entry
6015 shall always be removed when the directories are cleaned up.
6017 * coredumpctl gained a new "gdb" verb which invokes gdb
6018 right-away on the selected coredump.
6020 * There's now support for "hybrid sleep" on kernels that
6021 support this, in addition to "suspend" and "hibernate". Use
6022 "systemctl hybrid-sleep" to make use of this.
6024 * logind's HandleSuspendKey= setting (and related settings)
6025 now gained support for a new "lock" setting to simply
6026 request the screen lock on all local sessions, instead of
6027 actually executing a suspend or hibernation.
6029 * systemd will now mount the EFI variables file system by
6032 * Socket units now gained support for configuration of the
6033 SMACK security label.
6035 * timedatectl will now output the time of the last and next
6036 daylight saving change.
6038 * We dropped support for various legacy and distro-specific
6039 concepts, such as insserv, early-boot SysV services
6040 (i.e. those for non-standard runlevels such as 'b' or 'S')
6041 or ArchLinux /etc/rc.conf support. We recommend the
6042 distributions who still need support this to either continue
6043 to maintain the necessary patches downstream, or find a
6044 different solution. (Talk to us if you have questions!)
6046 * Various systemd components will now bypass PolicyKit checks
6047 for root and otherwise handle properly if PolicyKit is not
6048 found to be around. This should fix most issues for
6049 PolicyKit-less systems. Quite frankly this should have been
6050 this way since day one. It is absolutely our intention to
6051 make systemd work fine on PolicyKit-less systems, and we
6052 consider it a bug if something does not work as it should if
6053 PolicyKit is not around.
6055 * For embedded systems it is now possible to build udev and
6056 systemd without blkid and/or kmod support.
6058 * "systemctl switch-root" is now capable of switching root
6059 more than once. I.e. in addition to transitions from the
6060 initrd to the host OS it is now possible to transition to
6061 further OS images from the host. This is useful to implement
6062 offline updating tools.
6064 * Various other additions have been made to the RPM macros
6065 shipped with systemd. Use %udev_rules_update() after
6066 installing new udev rules files. %_udevhwdbdir,
6067 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
6068 %_sysctldir are now available which resolve to the right
6069 directories for packages to place various data files in.
6071 * journalctl gained the new --full switch (in addition to
6072 --all, to disable ellipsation for long messages.
6074 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
6075 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
6076 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
6077 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
6078 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
6079 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
6080 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
6081 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
6082 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
6086 * journalctl gained new --since= and --until= switches to
6087 filter by time. It also now supports nice filtering for
6088 units via --unit=/-u.
6090 * Type=oneshot services may use ExecReload= and do the
6093 * The journal daemon now supports time-based rotation and
6094 vacuuming, in addition to the usual disk-space based
6097 * The journal will now index the available field values for
6098 each field name. This enables clients to show pretty drop
6099 downs of available match values when filtering. The bash
6100 completion of journalctl has been updated
6101 accordingly. journalctl gained a new switch -F to list all
6102 values a certain field takes in the journal database.
6104 * More service events are now written as structured messages
6105 to the journal, and made recognizable via message IDs.
6107 * The timedated, localed and hostnamed mini-services which
6108 previously only provided support for changing time, locale
6109 and hostname settings from graphical DEs such as GNOME now
6110 also have a minimal (but very useful) text-based client
6111 utility each. This is probably the nicest way to changing
6112 these settings from the command line now, especially since
6113 it lists available options and is fully integrated with bash
6116 * There's now a new tool "systemd-coredumpctl" to list and
6117 extract coredumps from the journal.
6119 * We now install a README each in /var/log/ and
6120 /etc/rc.d/init.d explaining where the system logs and init
6121 scripts went. This hopefully should help folks who go to
6122 that dirs and look into the otherwise now empty void and
6123 scratch their heads.
6125 * When user-services are invoked (by systemd --user) the
6126 $MANAGERPID env var is set to the PID of systemd.
6128 * SIGRTMIN+24 when sent to a --user instance will now result
6129 in immediate termination of systemd.
6131 * gatewayd received numerous feature additions such as a
6132 "follow" mode, for live syncing and filtering.
6134 * browse.html now allows filtering and showing detailed
6135 information on specific entries. Keyboard navigation and
6136 mouse screen support has been added.
6138 * gatewayd/journalctl now supports HTML5/JSON
6139 Server-Sent-Events as output.
6141 * The SysV init script compatibility logic will now
6142 heuristically determine whether a script supports the
6143 "reload" verb, and only then make this available as
6146 * "systemctl status --follow" has been removed, use "journalctl
6149 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
6150 have been removed since they are hardly useful to be
6153 * And I'd like to take the opportunity to specifically mention
6154 Zbigniew for his great contributions. Zbigniew, you rock!
6156 Contributions from: Andrew Eikum, Christian Hesse, Colin
6157 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
6158 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
6159 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
6160 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
6161 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
6162 Jędrzejewski-Szmek, Сковорода Никита Андреевич
6166 * If /etc/vconsole.conf is non-existent or empty we will no
6167 longer load any console font or key map at boot by
6168 default. Instead the kernel defaults will be left
6169 intact. This is definitely the right thing to do, as no
6170 configuration should mean no configuration, and hard-coding
6171 font names that are different on all archs is probably a bad
6172 idea. Also, the kernel default key map and font should be
6173 good enough for most cases anyway, and mostly identical to
6174 the userspace fonts/key maps we previously overloaded them
6175 with. If distributions want to continue to default to a
6176 non-kernel font or key map they should ship a default
6177 /etc/vconsole.conf with the appropriate contents.
6179 Contributions from: Colin Walters, Daniel J Walsh, Dave
6180 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
6181 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6185 * journalctl gained a new --cursor= switch to show entries
6186 starting from the specified location in the journal.
6188 * We now enforce a size limit on journal entry fields exported
6189 with "-o json" in journalctl. Fields larger than 4K will be
6190 assigned null. This can be turned off with --all.
6192 * An (optional) journal gateway daemon is now available as
6193 "systemd-journal-gatewayd.service". This service provides
6194 access to the journal via HTTP and JSON. This functionality
6195 will be used to implement live log synchronization in both
6196 pull and push modes, but has various other users too, such
6197 as easy log access for debugging of embedded devices. Right
6198 now it is already useful to retrieve the journal via HTTP:
6200 # systemctl start systemd-journal-gatewayd.service
6201 # wget http://localhost:19531/entries
6203 This will download the journal contents in a
6204 /var/log/messages compatible format. The same as JSON:
6206 # curl -H"Accept: application/json" http://localhost:19531/entries
6208 This service is also accessible via a web browser where a
6209 single static HTML5 app is served that uses the JSON logic
6210 to enable the user to do some basic browsing of the
6211 journal. This will be extended later on. Here's an example
6212 screenshot of this app in its current state:
6214 http://0pointer.de/public/journal-gatewayd
6216 Contributions from: Kay Sievers, Lennart Poettering, Robert
6217 Milasan, Tom Gundersen
6221 * The bash completion logic is now available for journalctl
6224 * We do not mount the "cpuset" controller anymore together with
6225 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
6226 started if no parameters are assigned to it. "cpuset" hence
6227 broke code that assumed it could create "cpu" groups and
6230 * journalctl -f will now subscribe to terminal size changes,
6231 and line break accordingly.
6233 Contributions from: Dave Reisner, Kay Sievers, Lennart
6234 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
6238 * nspawn will now create a symlink /etc/localtime in the
6239 container environment, copying the host's timezone
6240 setting. Previously this has been done via a bind mount, but
6241 since symlinks cannot be bind mounted this has now been
6242 changed to create/update the appropriate symlink.
6244 * journalctl -n's line number argument is now optional, and
6245 will default to 10 if omitted.
6247 * journald will now log the maximum size the journal files may
6248 take up on disk. This is particularly useful if the default
6249 built-in logic of determining this parameter from the file
6250 system size is used. Use "systemctl status
6251 systemd-journald.service" to see this information.
6253 * The multi-seat X wrapper tool has been stripped down. As X
6254 is now capable of enumerating graphics devices via udev in a
6255 seat-aware way the wrapper is not strictly necessary
6256 anymore. A stripped down temporary stop-gap is still shipped
6257 until the upstream display managers have been updated to
6258 fully support the new X logic. Expect this wrapper to be
6259 removed entirely in one of the next releases.
6261 * HandleSleepKey= in logind.conf has been split up into
6262 HandleSuspendKey= and HandleHibernateKey=. The old setting
6263 is not available anymore. X11 and the kernel are
6264 distinguishing between these keys and we should too. This
6265 also means the inhibition lock for these keys has been split
6268 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
6269 Poettering, Lukas Nykryn, Václav Pavlín
6273 * Whenever a unit changes state we will now log this to the
6274 journal and show along the unit's own log output in
6277 * ConditionPathIsMountPoint= can now properly detect bind
6278 mount points too. (Previously, a bind mount of one file
6279 system to another place in the same file system could not be
6280 detected as mount, since they shared struct stat's st_dev
6283 * We will now mount the cgroup controllers cpu, cpuacct,
6284 cpuset and the controllers net_cls, net_prio together by
6287 * nspawn containers will now have a virtualized boot
6288 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
6289 over with a randomized ID at container initialization). This
6290 has the effect of making "journalctl -b" do the right thing
6293 * The JSON output journal serialization has been updated not
6294 to generate "endless" list objects anymore, but rather one
6295 JSON object per line. This is more in line how most JSON
6296 parsers expect JSON objects. The new output mode
6297 "json-pretty" has been added to provide similar output, but
6298 neatly aligned for readability by humans.
6300 * We dropped all explicit sync() invocations in the shutdown
6301 code. The kernel does this implicitly anyway in the kernel
6302 reboot() syscall. halt(8)'s -n option is now a compatibility
6305 * We now support virtualized reboot() in containers, as
6306 supported by newer kernels. We will fall back to exit() if
6307 CAP_SYS_REBOOT is not available to the container. Also,
6308 nspawn makes use of this now and will actually reboot the
6309 container if the containerized OS asks for that.
6311 * journalctl will only show local log output by default
6312 now. Use --merge (-m) to show remote log output, too.
6314 * libsystemd-journal gained the new sd_journal_get_usage()
6315 call to determine the current disk usage of all journal
6316 files. This is exposed in the new "journalctl --disk-usage"
6319 * journald gained a new configuration setting SplitMode= in
6320 journald.conf which may be used to control how user journals
6321 are split off. See journald.conf(5) for details.
6323 * A new condition type ConditionFileNotEmpty= has been added.
6325 * tmpfiles' "w" lines now support file globbing, to write
6326 multiple files at once.
6328 * We added Python bindings for the journal submission
6329 APIs. More Python APIs for a number of selected APIs will
6330 likely follow. Note that we intend to add native bindings
6331 only for the Python language, as we consider it common
6332 enough to deserve bindings shipped within systemd. There are
6333 various projects outside of systemd that provide bindings
6334 for languages such as PHP or Lua.
6336 * Many conditions will now resolve specifiers such as %i. In
6337 addition, PathChanged= and related directives of .path units
6338 now support specifiers as well.
6340 * There's now a new RPM macro definition for the system preset
6343 * journald will now warn if it ca not forward a message to the
6344 syslog daemon because its socket is full.
6346 * timedated will no longer write or process /etc/timezone,
6347 except on Debian. As we do not support late mounted /usr
6348 anymore /etc/localtime always being a symlink is now safe,
6349 and hence the information in /etc/timezone is not necessary
6352 * logind will now always reserve one VT for a text getty (VT6
6353 by default). Previously if more than 6 X sessions where
6354 started they took up all the VTs with auto-spawned gettys,
6355 so that no text gettys were available anymore.
6357 * udev will now automatically inform the btrfs kernel logic
6358 about btrfs RAID components showing up. This should make
6359 simple hotplug based btrfs RAID assembly work.
6361 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
6362 (but not for its children which will stay at the kernel
6363 default). This should allow setups with a lot more listening
6366 * systemd will now always pass the configured timezone to the
6367 kernel at boot. timedated will do the same when the timezone
6370 * logind's inhibition logic has been updated. By default,
6371 logind will now handle the lid switch, the power and sleep
6372 keys all the time, even in graphical sessions. If DEs want
6373 to handle these events on their own they should take the new
6374 handle-power-key, handle-sleep-key and handle-lid-switch
6375 inhibitors during their runtime. A simple way to achieve
6376 that is to invoke the DE wrapped in an invocation of:
6378 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
6380 * Access to unit operations is now checked via SELinux taking
6381 the unit file label and client process label into account.
6383 * systemd will now notify the administrator in the journal
6384 when he over-mounts a non-empty directory.
6386 * There are new specifiers that are resolved in unit files,
6387 for the host name (%H), the machine ID (%m) and the boot ID
6390 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
6391 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
6392 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
6393 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
6394 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
6395 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
6396 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
6400 * Support for reading structured kernel messages from
6401 /dev/kmsg has now been added and is enabled by default.
6403 * Support for reading kernel messages from /proc/kmsg has now
6404 been removed. If you want kernel messages in the journal
6405 make sure to run a recent kernel (>= 3.5) that supports
6406 reading structured messages from /dev/kmsg (see
6407 above). /proc/kmsg is now exclusive property of classic
6408 syslog daemons again.
6410 * The libudev API gained the new
6411 udev_device_new_from_device_id() call.
6413 * The logic for file system namespace (ReadOnlyDirectory=,
6414 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
6415 require pivot_root() anymore. This means fewer temporary
6416 directories are created below /tmp for this feature.
6418 * nspawn containers will now see and receive all submounts
6419 made on the host OS below the root file system of the
6422 * Forward Secure Sealing is now supported for Journal files,
6423 which provide cryptographical sealing of journal files so
6424 that attackers cannot alter log history anymore without this
6425 being detectable. Lennart will soon post a blog story about
6426 this explaining it in more detail.
6428 * There are two new service settings RestartPreventExitStatus=
6429 and SuccessExitStatus= which allow configuration of exit
6430 status (exit code or signal) which will be excepted from the
6431 restart logic, resp. consider successful.
6433 * journalctl gained the new --verify switch that can be used
6434 to check the integrity of the structure of journal files and
6435 (if Forward Secure Sealing is enabled) the contents of
6438 * nspawn containers will now be run with /dev/stdin, /dev/fd/
6439 and similar symlinks pre-created. This makes running shells
6440 as container init process a lot more fun.
6442 * The fstab support can now handle PARTUUID= and PARTLABEL=
6445 * A new ConditionHost= condition has been added to match
6446 against the hostname (with globs) and machine ID. This is
6447 useful for clusters where a single OS image is used to
6448 provision a large number of hosts which shall run slightly
6449 different sets of services.
6451 * Services which hit the restart limit will now be placed in a
6454 Contributions from: Bertram Poettering, Dave Reisner, Huang
6455 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
6456 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
6460 * When running in --user mode systemd will now become a
6461 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
6462 tree a lot more organized.
6464 * A new PartOf= unit dependency type has been introduced that
6465 may be used to group services in a natural way.
6467 * "systemctl enable" may now be used to enable instances of
6470 * journalctl now prints error log levels in red, and
6471 warning/notice log levels in bright white. It also supports
6472 filtering by log level now.
6474 * cgtop gained a new -n switch (similar to top), to configure
6475 the maximum number of iterations to run for. It also gained
6476 -b, to run in batch mode (accepting no input).
6478 * The suffix ".service" may now be omitted on most systemctl
6479 command lines involving service unit names.
6481 * There's a new bus call in logind to lock all sessions, as
6482 well as a loginctl verb for it "lock-sessions".
6484 * libsystemd-logind.so gained a new call sd_journal_perror()
6485 that works similar to libc perror() but logs to the journal
6486 and encodes structured information about the error number.
6488 * /etc/crypttab entries now understand the new keyfile-size=
6491 * shutdown(8) now can send a (configurable) wall message when
6492 a shutdown is cancelled.
6494 * The mount propagation mode for the root file system will now
6495 default to "shared", which is useful to make containers work
6496 nicely out-of-the-box so that they receive new mounts from
6497 the host. This can be undone locally by running "mount
6498 --make-rprivate /" if needed.
6500 * The prefdm.service file has been removed. Distributions
6501 should maintain this unit downstream if they intend to keep
6502 it around. However, we recommend writing normal unit files
6503 for display managers instead.
6505 * Since systemd is a crucial part of the OS we will now
6506 default to a number of compiler switches that improve
6507 security (hardening) such as read-only relocations, stack
6508 protection, and suchlike.
6510 * The TimeoutSec= setting for services is now split into
6511 TimeoutStartSec= and TimeoutStopSec= to allow configuration
6512 of individual time outs for the start and the stop phase of
6515 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
6516 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
6517 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
6518 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
6519 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
6520 Gundersen, Zbigniew Jędrzejewski-Szmek
6524 * The journal and id128 C APIs are now fully documented as man
6527 * Extra safety checks have been added when transitioning from
6528 the initial RAM disk to the main system to avoid accidental
6531 * /etc/crypttab entries now understand the new keyfile-offset=
6534 * systemctl -t can now be used to filter by unit load state.
6536 * The journal C API gained the new sd_journal_wait() call to
6537 make writing synchronous journal clients easier.
6539 * journalctl gained the new -D switch to show journals from a
6542 * journalctl now displays a special marker between log
6543 messages of two different boots.
6545 * The journal is now explicitly flushed to /var via a service
6546 systemd-journal-flush.service, rather than implicitly simply
6547 by seeing /var/log/journal to be writable.
6549 * journalctl (and the journal C APIs) can now match for much
6550 more complex expressions, with alternatives and
6553 * When transitioning from the initial RAM disk to the main
6554 system we will now kill all processes in a killing spree to
6555 ensure no processes stay around by accident.
6557 * Three new specifiers may be used in unit files: %u, %h, %s
6558 resolve to the user name, user home directory resp. user
6559 shell. This is useful for running systemd user instances.
6561 * We now automatically rotate journal files if their data
6562 object hash table gets a fill level > 75%. We also size the
6563 hash table based on the configured maximum file size. This
6564 together should lower hash collisions drastically and thus
6565 speed things up a bit.
6567 * journalctl gained the new "--header" switch to introspect
6568 header data of journal files.
6570 * A new setting SystemCallFilters= has been added to services
6571 which may be used to apply blacklists or whitelists to
6572 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
6574 * nspawn gained a new --link-journal= switch (and quicker: -j)
6575 to link the container journal with the host. This makes it
6576 very easy to centralize log viewing on the host for all
6577 guests while still keeping the journal files separated.
6579 * Many bugfixes and optimizations
6581 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
6582 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
6583 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
6588 * Several tools now understand kernel command line arguments,
6589 which are only read when run in an initial RAM disk. They
6590 usually follow closely their normal counterparts, but are
6593 * There's a new tool to analyze the readahead files that are
6594 automatically generated at boot. Use:
6596 /usr/lib/systemd/systemd-readahead analyze /.readahead
6598 * We now provide an early debug shell on tty9 if this enabled. Use:
6600 systemctl enable debug-shell.service
6602 * All plymouth related units have been moved into the Plymouth
6603 package. Please make sure to upgrade your Plymouth version
6606 * systemd-tmpfiles now supports getting passed the basename of
6607 a configuration file only, in which case it will look for it
6608 in all appropriate directories automatically.
6610 * udevadm info now takes a /dev or /sys path as argument, and
6611 does the right thing. Example:
6613 udevadm info /dev/sda
6614 udevadm info /sys/class/block/sda
6616 * systemctl now prints a warning if a unit is stopped but a
6617 unit that might trigger it continues to run. Example: a
6618 service is stopped but the socket that activates it is left
6621 * "systemctl status" will now mention if the log output was
6622 shortened due to rotation since a service has been started.
6624 * The journal API now exposes functions to determine the
6625 "cutoff" times due to rotation.
6627 * journald now understands SIGUSR1 and SIGUSR2 for triggering
6628 immediately flushing of runtime logs to /var if possible,
6629 resp. for triggering immediate rotation of the journal
6632 * It is now considered an error if a service is attempted to
6633 be stopped that is not loaded.
6635 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
6637 * systemd-analyze now supports Python 3
6639 * tmpfiles now supports cleaning up directories via aging
6640 where the first level dirs are always kept around but
6641 directories beneath it automatically aged. This is enabled
6642 by prefixing the age field with '~'.
6644 * Seat objects now expose CanGraphical, CanTTY properties
6645 which is required to deal with very fast bootups where the
6646 display manager might be running before the graphics drivers
6647 completed initialization.
6649 * Seat objects now expose a State property.
6651 * We now include RPM macros for service enabling/disabling
6652 based on the preset logic. We recommend RPM based
6653 distributions to make use of these macros if possible. This
6654 makes it simpler to reuse RPM spec files across
6657 * We now make sure that the collected systemd unit name is
6658 always valid when services log to the journal via
6661 * There's a new man page kernel-command-line(7) detailing all
6662 command line options we understand.
6664 * The fstab generator may now be disabled at boot by passing
6665 fstab=0 on the kernel command line.
6667 * A new kernel command line option modules-load= is now understood
6668 to load a specific kernel module statically, early at boot.
6670 * Unit names specified on the systemctl command line are now
6671 automatically escaped as needed. Also, if file system or
6672 device paths are specified they are automatically turned
6673 into the appropriate mount or device unit names. Example:
6675 systemctl status /home
6676 systemctl status /dev/sda
6678 * The SysVConsole= configuration option has been removed from
6679 system.conf parsing.
6681 * The SysV search path is no longer exported on the D-Bus
6684 * The Names= option has been removed from unit file parsing.
6686 * There's a new man page bootup(7) detailing the boot process.
6688 * Every unit and every generator we ship with systemd now
6689 comes with full documentation. The self-explanatory boot is
6692 * A couple of services gained "systemd-" prefixes in their
6693 name if they wrap systemd code, rather than only external
6694 code. Among them fsck@.service which is now
6695 systemd-fsck@.service.
6697 * The HaveWatchdog property has been removed from the D-Bus
6700 * systemd.confirm_spawn= on the kernel command line should now
6703 * There's a new man page crypttab(5) which details all options
6704 we actually understand.
6706 * systemd-nspawn gained a new --capability= switch to pass
6707 additional capabilities to the container.
6709 * timedated will now read known NTP implementation unit names
6710 from /usr/lib/systemd/ntp-units.d/*.list,
6711 systemd-timedated-ntp.target has been removed.
6713 * journalctl gained a new switch "-b" that lists log data of
6714 the current boot only.
6716 * The notify socket is in the abstract namespace again, in
6717 order to support daemons which chroot() at start-up.
6719 * There is a new Storage= configuration option for journald
6720 which allows configuration of where log data should go. This
6721 also provides a way to disable journal logging entirely, so
6722 that data collected is only forwarded to the console, the
6723 kernel log buffer or another syslog implementation.
6725 * Many bugfixes and optimizations
6727 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
6728 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
6729 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
6730 Shawn Landden, Tom Gundersen
6734 * "systemctl help <unit>" now shows the man page if one is
6737 * Several new man pages have been added.
6739 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
6740 MaxLevelConsole= can now be specified in
6741 journald.conf. These options allow reducing the amount of
6742 data stored on disk or forwarded by the log level.
6744 * TimerSlackNSec= can now be specified in system.conf for
6745 PID1. This allows system-wide power savings.
6747 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
6748 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
6753 * logind is now capable of (optionally) handling power and
6754 sleep keys as well as the lid switch.
6756 * journalctl now understands the syntax "journalctl
6757 /usr/bin/avahi-daemon" to get all log output of a specific
6760 * CapabilityBoundingSet= in system.conf now also influences
6761 the capability bound set of usermode helpers of the kernel.
6763 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
6764 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
6765 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
6766 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
6770 * Note that we skipped 139 releases here in order to set the
6771 new version to something that is greater than both udev's
6772 and systemd's most recent version number.
6774 * udev: all udev sources are merged into the systemd source tree now.
6775 All future udev development will happen in the systemd tree. It
6776 is still fully supported to use the udev daemon and tools without
6777 systemd running, like in initramfs or other init systems. Building
6778 udev though, will require the *build* of the systemd tree, but
6779 udev can be properly *run* without systemd.
6781 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
6782 should be used to create dead device nodes as workarounds for broken
6785 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
6786 no longer supported. udev_monitor_new_from_netlink() needs to be
6787 used to subscribe to events.
6789 * udev: when udevd is started by systemd, processes which are left
6790 behind by forking them off of udev rules, are unconditionally cleaned
6791 up and killed now after the event handling has finished. Services or
6792 daemons must be started as systemd services. Services can be
6793 pulled-in by udev to get started, but they can no longer be directly
6794 forked by udev rules.
6796 * udev: the daemon binary is called systemd-udevd now and installed
6797 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
6798 to adapt to that, create symlink, or rename the binary after building
6801 * libudev no longer provides these symbols:
6802 udev_monitor_from_socket()
6803 udev_queue_get_failed_list_entry()
6804 udev_get_{dev,sys,run}_path()
6805 The versions number was bumped and symbol versioning introduced.
6807 * systemd-loginctl and systemd-journalctl have been renamed
6808 to loginctl and journalctl to match systemctl.
6810 * The config files: /etc/systemd/systemd-logind.conf and
6811 /etc/systemd/systemd-journald.conf have been renamed to
6812 logind.conf and journald.conf. Package updates should rename
6813 the files to the new names on upgrade.
6815 * For almost all files the license is now LGPL2.1+, changed
6816 from the previous GPL2.0+. Exceptions are some minor stuff
6817 of udev (which will be changed to LGPL2.1 eventually, too),
6818 and the MIT licensed sd-daemon.[ch] library that is suitable
6819 to be used as drop-in files.
6821 * systemd and logind now handle system sleep states, in
6822 particular suspending and hibernating.
6824 * logind now implements a sleep/shutdown/idle inhibiting logic
6825 suitable for a variety of uses. Soonishly Lennart will blog
6826 about this in more detail.
6828 * var-run.mount and var-lock.mount are no longer provided
6829 (which previously bind mounted these directories to their new
6830 places). Distributions which have not converted these
6831 directories to symlinks should consider stealing these files
6832 from git history and add them downstream.
6834 * We introduced the Documentation= field for units and added
6835 this to all our shipped units. This is useful to make it
6836 easier to explore the boot and the purpose of the various
6839 * All smaller setup units (such as
6840 systemd-vconsole-setup.service) now detect properly if they
6841 are run in a container and are skipped when
6842 appropriate. This guarantees an entirely noise-free boot in
6843 Linux container environments such as systemd-nspawn.
6845 * A framework for implementing offline system updates is now
6846 integrated, for details see:
6847 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
6849 * A new service type Type=idle is available now which helps us
6850 avoiding ugly interleaving of getty output and boot status
6853 * There's now a system-wide CapabilityBoundingSet= option to
6854 globally reduce the set of capabilities for the
6855 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
6856 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
6857 even CAP_NET_ADMIN system-wide for secure systems.
6859 * There are now system-wide DefaultLimitXXX= options to
6860 globally change the defaults of the various resource limits
6861 for all units started by PID 1.
6863 * Harald Hoyer's systemd test suite has been integrated into
6864 systemd which allows easy testing of systemd builds in qemu
6865 and nspawn. (This is really awesome! Ask us for details!)
6867 * The fstab parser is now implemented as generator, not inside
6870 * systemctl will now warn you if .mount units generated from
6871 /etc/fstab are out of date due to changes in fstab that
6872 have not been read by systemd yet.
6874 * systemd is now suitable for usage in initrds. Dracut has
6875 already been updated to make use of this. With this in place
6876 initrds get a slight bit faster but primarily are much
6877 easier to introspect and debug since "systemctl status" in
6878 the host system can be used to introspect initrd services,
6879 and the journal from the initrd is kept around too.
6881 * systemd-delta has been added, a tool to explore differences
6882 between user/admin configuration and vendor defaults.
6884 * PrivateTmp= now affects both /tmp and /var/tmp.
6886 * Boot time status messages are now much prettier and feature
6887 proper english language. Booting up systemd has never been
6890 * Read-ahead pack files now include the inode number of all
6891 files to pre-cache. When the inode changes the pre-caching
6892 is not attempted. This should be nicer to deal with updated
6893 packages which might result in changes of read-ahead
6896 * We now temporaritly lower the kernel's read_ahead_kb variable
6897 when collecting read-ahead data to ensure the kernel's
6898 built-in read-ahead does not add noise to our measurements
6899 of necessary blocks to pre-cache.
6901 * There's now RequiresMountsFor= to add automatic dependencies
6902 for all mounts necessary for a specific file system path.
6904 * MountAuto= and SwapAuto= have been removed from
6905 system.conf. Mounting file systems at boot has to take place
6908 * nspawn now learned a new switch --uuid= to set the machine
6909 ID on the command line.
6911 * nspawn now learned the -b switch to automatically search
6914 * vt102 is now the default TERM for serial TTYs, upgraded from
6917 * systemd-logind now works on VT-less systems.
6919 * The build tree has been reorganized. The individual
6920 components now have directories of their own.
6922 * A new condition type ConditionPathIsReadWrite= is now available.
6924 * nspawn learned the new -C switch to create cgroups for the
6925 container in other hierarchies.
6927 * We now have support for hardware watchdogs, configurable in
6930 * The scheduled shutdown logic now has a public API.
6932 * We now mount /tmp as tmpfs by default, but this can be
6933 masked and /etc/fstab can override it.
6935 * Since udisks does not make use of /media anymore we are not
6936 mounting a tmpfs on it anymore.
6938 * journalctl gained a new --local switch to only interleave
6939 locally generated journal files.
6941 * We can now load the IMA policy at boot automatically.
6943 * The GTK tools have been split off into a systemd-ui.
6945 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
6946 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
6947 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
6948 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
6949 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
6950 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
6951 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
6952 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
6953 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
6958 * This is mostly a bugfix release
6960 * Support optional initialization of the machine ID from the
6961 KVM or container configured UUID.
6963 * Support immediate reboots with "systemctl reboot -ff"
6965 * Show /etc/os-release data in systemd-analyze output
6967 * Many bugfixes for the journal, including endianness fixes and
6968 ensuring that disk space enforcement works
6970 * sd-login.h is C++ compatible again
6972 * Extend the /etc/os-release format on request of the Debian
6975 * We now refuse non-UTF8 strings used in various configuration
6976 and unit files. This is done to ensure we do not pass invalid
6977 data over D-Bus or expose it elsewhere.
6979 * Register Mimo USB Screens as suitable for automatic seat
6982 * Read SELinux client context from journal clients in a race
6985 * Reorder configuration file lookup order. /etc now always
6986 overrides /run in order to allow the administrator to always
6987 and unconditionally override vendor-supplied or
6988 automatically generated data.
6990 * The various user visible bits of the journal now have man
6991 pages. We still lack man pages for the journal API calls
6994 * We now ship all man pages in HTML format again in the
6997 Contributions from: Dave Reisner, Dirk Eibach, Frederic
6998 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
6999 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
7004 * This is mostly a bugfix release
7006 * systems lacking /etc/os-release are no longer supported.
7008 * Various functionality updates to libsystemd-login.so
7010 * Track class of PAM logins to distinguish greeters from
7013 Contributions from: Kay Sievers, Lennart Poettering, Michael
7018 * This is an important bugfix release for v41.
7020 * Building man pages is now optional which should be useful
7021 for those building systemd from git but unwilling to install
7024 * Watchdog support for supervising services is now usable. In
7025 a future release support for hardware watchdogs
7026 (i.e. /dev/watchdog) will be added building on this.
7028 * Service start rate limiting is now configurable and can be
7029 turned off per service. When a start rate limit is hit a
7030 reboot can automatically be triggered.
7032 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
7034 Contributions from: Benjamin Franzke, Bill Nottingham,
7035 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
7036 Schmidt, Michał Górny, Piotr Drąg
7040 * The systemd binary is installed /usr/lib/systemd/systemd now;
7041 An existing /sbin/init symlink needs to be adapted with the
7044 * The code that loads kernel modules has been ported to invoke
7045 libkmod directly, instead of modprobe. This means we do not
7046 support systems with module-init-tools anymore.
7048 * Watchdog support is now already useful, but still not
7051 * A new kernel command line option systemd.setenv= is
7052 understood to set system wide environment variables
7053 dynamically at boot.
7055 * We now limit the set of capabilities of systemd-journald.
7057 * We now set SIGPIPE to ignore by default, since it only is
7058 useful in shell pipelines, and has little use in general
7059 code. This can be disabled with IgnoreSIPIPE=no in unit
7062 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
7063 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
7068 * This is mostly a bugfix release
7070 * We now expose the reason why a service failed in the
7071 "Result" D-Bus property.
7073 * Rudimentary service watchdog support (will be completed over
7074 the next few releases.)
7076 * When systemd forks off in order execute some service we will
7077 now immediately changes its argv[0] to reflect which process
7078 it will execute. This is useful to minimize the time window
7079 with a generic argv[0], which makes bootcharts more useful
7081 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
7082 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
7083 Mike Kazantsev, Ray Strode
7087 * This is mostly a test release, but incorporates many
7090 * New systemd-cgtop tool to show control groups by their
7093 * Linking against libacl for ACLs is optional again. If
7094 disabled, support tracking device access for active logins
7095 goes becomes unavailable, and so does access to the user
7096 journals by the respective users.
7098 * If a group "adm" exists, journal files are automatically
7099 owned by them, thus allow members of this group full access
7100 to the system journal as well as all user journals.
7102 * The journal now stores the SELinux context of the logging
7103 client for all entries.
7105 * Add C++ inclusion guards to all public headers
7107 * New output mode "cat" in the journal to print only text
7108 messages, without any meta data like date or time.
7110 * Include tiny X server wrapper as a temporary stop-gap to
7111 teach XOrg udev display enumeration. This is used by display
7112 managers such as gdm, and will go away as soon as XOrg
7113 learned native udev hotplugging for display devices.
7115 * Add new systemd-cat tool for executing arbitrary programs
7116 with STDERR/STDOUT connected to the journal. Can also act as
7117 BSD logger replacement, and does so by default.
7119 * Optionally store all locally generated coredumps in the
7120 journal along with meta data.
7122 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
7123 writing short strings to files (for usage for /sys), and for
7124 creating symlinks, character and block device nodes.
7126 * New unit file option ControlGroupPersistent= to make cgroups
7127 persistent, following the mechanisms outlined in
7128 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
7130 * Support multiple local RTCs in a sane way
7132 * No longer monopolize IO when replaying readahead data on
7133 rotating disks, since we might starve non-file-system IO to
7134 death, since fanotify() will not see accesses done by blkid,
7137 * Do not show kernel threads in systemd-cgls anymore, unless
7138 requested with new -k switch.
7140 Contributions from: Dan Horák, Kay Sievers, Lennart
7141 Poettering, Michal Schmidt
7145 * This is mostly a test release, but incorporates many
7148 * The git repository moved to:
7149 git://anongit.freedesktop.org/systemd/systemd
7150 ssh://git.freedesktop.org/git/systemd/systemd
7152 * First release with the journal
7153 http://0pointer.de/blog/projects/the-journal.html
7155 * The journal replaces both systemd-kmsg-syslogd and
7156 systemd-stdout-bridge.
7158 * New sd_pid_get_unit() API call in libsystemd-logind
7160 * Many systemadm clean-ups
7162 * Introduce remote-fs-pre.target which is ordered before all
7163 remote mounts and may be used to start services before all
7166 * Added Mageia support
7168 * Add bash completion for systemd-loginctl
7170 * Actively monitor PID file creation for daemons which exit in
7171 the parent process before having finished writing the PID
7172 file in the daemon process. Daemons which do this need to be
7173 fixed (i.e. PID file creation must have finished before the
7174 parent exits), but we now react a bit more gracefully to them.
7176 * Add colourful boot output, mimicking the well-known output
7177 of existing distributions.
7179 * New option PassCredentials= for socket units, for
7180 compatibility with a recent kernel ABI breakage.
7182 * /etc/rc.local is now hooked in via a generator binary, and
7183 thus will no longer act as synchronization point during
7186 * systemctl list-unit-files now supports --root=.
7188 * systemd-tmpfiles now understands two new commands: z, Z for
7189 relabelling files according to the SELinux database. This is
7190 useful to apply SELinux labels to specific files in /sys,
7193 * Output of SysV services is now forwarded to both the console
7194 and the journal by default, not only just the console.
7196 * New man pages for all APIs from libsystemd-login.
7198 * The build tree got reorganized and the build system is a
7199 lot more modular allowing embedded setups to specifically
7200 select the components of systemd they are interested in.
7202 * Support for Linux systems lacking the kernel VT subsystem is
7205 * configure's --with-rootdir= got renamed to
7206 --with-rootprefix= to follow the naming used by udev and
7209 * Unless specified otherwise we will now install to /usr instead
7210 of /usr/local by default.
7212 * Processes with '@' in argv[0][0] are now excluded from the
7213 final shut-down killing spree, following the logic explained
7215 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
7217 * All processes remaining in a service cgroup when we enter
7218 the START or START_PRE states are now killed with
7219 SIGKILL. That means it is no longer possible to spawn
7220 background processes from ExecStart= lines (which was never
7221 supported anyway, and bad style).
7223 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
7224 reloading of units together.
7226 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
7227 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
7228 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
7229 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
7230 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek