1 systemd System and Service Manager
3 CHANGES WITH 233 in spe
5 [ LIST FAR FROM COMPLETE YET ]
7 * DBus policy files are now installed into /usr rather than /etc. Make
8 sure your system has dbus >= 1.9.18 running before upgrading to this
9 version, or override the install path with --with-dbuspolicydir= .
11 * The shell invoked by debug-shell.service now defaults to /bin/sh in
12 all cases. If distributions want to use a different shell for this
13 purpose (for example Fedora's /sbin/sushell) they need to specify
14 this explicitly at configure time using --with-debug-shell=.
16 * The confirmation spawn prompt has been reworked to offer the
19 (c)ontinue, proceed without asking anymore
20 (D)ump, show the state of the unit
21 (f)ail, don't execute the command and pretend it failed
23 (i)nfo, show a short summary of the unit
24 (j)obs, show jobs that are in progress
25 (s)kip, don't execute the command and pretend it succeeded
26 (y)es, execute the command
28 The 'n' choice for the confirmation spawn prompt has been removed,
29 because its meaning was confusing.
31 The prompt may now also be redirected to an alternative console by
32 specifying the console as parameter to systemd.confirm_spawn=.
34 * Services of Type=notify require a READY=1 notification to be sent
35 during startup. If no such message is sent, the service now fails,
36 even if the main process exited with a successful exit code.
38 * Services that fail to start up correctly now always have their
39 ExecStopPost= commands executed. Previously, they'd enter "failed"
40 state directly, without executing these commands.
42 * The option MulticastDNS= of network configuration files has acquired
43 an actual implementation. With MulticastDNS=yes a host can resolve
44 names of remote hosts and to reply to mDNS's A and AAAA requests from
47 * When units are about to be started an additional check is now done to
48 ensure that all dependencies of type BindsTo= (when used in
49 combination with After=) have been started.
51 * systemd-analyze gained a new verb "syscall-filter" which shows which
52 system call groups are defined for the SystemCallFilter= unit file
53 setting, and which system calls they precisely contain.
55 * A new system call filter group "@filesystem" has been added,
56 consisting of various file system related system calls. A group
57 "@reboot" has been added, covering reboot, kexec and shutdown related
58 calls. Finally, a group "@swap" has been added covering swap
59 configuration related calls.
61 * A new unit file option RestrictNamespaces= has been added that may be
62 used to restrict access to the various process namespace types the
63 Linux kernel provides. Specifically, it may be used to take away the
64 right for specific service units to create additional file system,
65 networking, user, and other namespaces. This sandboxing option is
66 particularly relevant due to the high amount of recently discovered
67 namespacing related vulnerabilities in the kernel.
69 * .link gained support for a new AutoNegotiation= setting for
70 configuring Ethernet auto-negotiation.
72 * systemd-networkd's .network files gained support for a new
73 ListenPort= setting in the [DHCP] section to explicitly configure the
74 UDP client port the DHCP client shall listen on.
76 * New systemd-specific mount options are now understood in /etc/fstab:
78 x-systemd.mount-timeout= may be used to configure the maximum
79 permitted runtime of the mount command.
81 x-systemd.device-bound may be set to bind a mount point to its
82 backing device unit, in order to automatically remove a mount point
83 if its backing device is unplugged. This option may also be
84 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
85 on the block device, which is now automatically set for all CDROM
86 drives, so that mounted CDs are automatically unmounted when they are
87 removed from the drive.
89 x-systemd.after= and x-systemd.before= may be use to explicitly order
90 a mount after or before another unit or mount point.
92 * Enqueued start jobs for device units are now automatically garbage
93 collected if there are no jobs waiting for them anymore.
95 * systemctl list-jobs gained two new switches: With --after, every
96 queued job shows which other queued job is waiting for it; with
97 --before it shows which other jobs every job is waiting for.
99 * systemd-nspawn gained support for ephemeral boots from disk images
100 (or in other words: --ephemeral and --image= may now be
101 combined). Moreover, ephemeral boots are now supported for normal
102 directories, even if the backing file system is not btrfs. Of course,
103 if the file system does not support file system snapshots or
104 reflinks, the initial copy operation will be relatively expensive, but
105 this should still be suitable for many use cases.
107 * Calendar time specifications in .timer units now support
108 specifications relative to the end of a month by using "~" instead of
109 "-" as separator between month and day. For example, "*-02~03" means
110 "The third last day in February". In addition a new syntax for
111 repeated events has been added using the "/" character. For example,
112 "9..17/2:00" means "every two hours from 9am to 5pm".
114 * systemd-socket-proxyd gained a new parameter --connections-max= for
115 configuring the maximum number of concurrent connections.
117 * All python scripts shipped with systemd (specifically: the various
118 tests written in Python) now require Python 3.
120 * sd-id128 gained a new API for generating unique IDs for the host
121 that does not leak the machine ID. Specifically,
122 sd_id128_get_machine_app_specific() derives an ID based on the
123 machine ID a in well-defined, non-reversible, stable way. This is
124 useful whenever an identifier for the host is needed but where the
125 identifier shall not be useful to identify the system beyond the
126 scope of the application itself. (Internally this uses HMAC-SHA256 as
127 keyed hash function using the machine ID as input.)
129 * NotifyAccess= gained a new supported value "exec". When set
130 notifications are accepted from all processes systemd itself invoked,
131 including all control processes.
133 * .nspawn files gained support for defining overlay mounts using the
134 Overlay= and OverlayReadOnly= options. Previously this functionality
135 was only available on the systemd-nspawn command line.
137 * systemd-nspawn's --bind= and --overlay= options gained support for
138 bind/overlay mounts whose source lies within the container tree by
139 prefixing the source path with "+".
141 * systemd-nspawn's --bind= and --overlay= options gained support for
142 automatically allocating a temporary source directory in /var/tmp
143 that is removed when the container dies. Specifically, if the source
144 directory is specified as empty string this mechanism is selected. An
145 example usage is --overlay=+/var::/var, which creates an overlay
146 mount based on the original /var contained in the image, overlayed
147 with a temporary directory in the host's /var/tmp. This way changes
148 to /var are automatically flushed when the container shuts down.
150 * .network files gained a new Unmanaged= boolean setting for explicitly
151 excluding one or more interfaces from management by systemd-networkd.
153 * systemd-nspawn --image= option does now permit raw file system block
154 devices (in addition to images containing partition tables, as
157 * The disk image dissection logic in systemd-nspawn gained support for
158 automatically setting up LUKS encrypted as well as Verity protected
159 partitions. When a container is booted from an encrypted image the
160 passphrase is queried at start-up time. When a container with Verity
161 data is started, the root hash is search in a ".roothash" file
162 accompanying the disk image (alternatively, pass the root hash via
163 the new --root-hash= command line option).
165 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
166 be used to dissect disk images the same way as systemd-nspawn does
167 it, following the Bootable Partition Specification. It may even be
168 used to mount disk images with complex partition setups (including
169 LUKS and Verity partitions) to a local host directory, in order to
170 inspect them. This tool is not considered public API (yet), and is
171 thus not installed into /usr/bin. Please do not rely on its
172 existance, since it might go away or be changed in later systemd
175 * A new generator "systemd-verity-generator" has been added, similar in
176 style to "systemd-cryptsetup-generator", permitting automatic setup of
177 Verity root partitions when systemd boots up. In order to make use of
178 this your partition setup should follow the Discoverable Partitions
179 Specification, and the GPT partition ID of the root file system
180 partition should be identical to the upper 128bit of the Verity root
181 hash. The GPT partition ID of the Verity partition protecting it
182 should be the lower 128bit of the Verity root hash. If the partition
183 image follows this model it is sufficient to specify a single
184 "roothash=" kernel command line argument to both configure which root
185 image and verity partition to use as well as the root hash for
186 it. Note that systemd-nspawn's Verity support follows the same
187 semantics, meaning that disk images with proper Verity data in place
188 may be booted in containers with systemd-nspawn as well as on
189 physical systems via the verity generator. Also note that the "mkosi"
190 tool available at https://github.com/systemd/mkosi has been updated
191 to generate Verity protected disk images following this scheme. In
192 fact, it has been updated to generate disk images that optionally
193 implement a complete UEFI SecureBoot trust chain, involving a signed
194 kernel and initrd image that incorporates such a root hash as well as
195 a Verity-enabled root partition.
197 * Support for the %c, %r, %R specifiers in unit files has been
198 removed. Specifiers are not supposed to be dependent on configuration
199 of unit files themselves (so that they resolve to the same regardless
200 where used in the unit files), but these options were due to the
201 existence of the Slice= option.
203 * The various options in the [Match] section of .network files gained
204 support for negative matching.
206 * The hardware database (hwdb) udev supports has been updated to carry
207 accelerometer quirks.
209 * All system services are now run with a fresh kernel keyring set up
210 for them. The invocation ID is stored by default in it, thus
211 providing a safe, non-overridable way to determine the invocation
214 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
215 options for bind mounting arbitrary paths in a service-specific
216 way. When these options are used, arbitrary host or service files and
217 directories may be mounted to arbitrary locations in the service's
220 * Documentation has been added that lists all of systemd's low-level
221 environment variables:
223 https://github.com/systemd/systemd/blob/master/ENVIRONMENT.md
225 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
226 whether a specific socket file descriptor matches a specified socket
229 * systemd-firstboot has been updated to check for the
230 systemd.firstboot= kernel command line option. It accepts a boolean
231 and when set to false the first boot questions are skipped.
233 * The systemd-networkd ProxyARP= option has been renamed to
234 IPV4ProxyARP=. Similar, VXLAN-specific option ARPProxy= has been
235 renamed to ReduceARPProxy=. The old names continue to be available
238 * systemd-networkd's bonding device support gained support for two new
239 configuration options ActiveSlave= and PrimarySlave=.
241 * systemd-fstab-generator has been updated to check for the
242 systemd.volatile= kernel command line option, which either takes a
243 boolean parameter or the special value "state". If used the system
244 may be booted in a "volatile" boot mode. Specifically,
245 systemd.volatile=yes is used, the root directory will be mounted as
246 tmpfs, and only /usr is mounted from the actual root file system. If
247 systemd.volatile=state is used, the root directory will be mounted as
248 usual, but /var is mounted as tmpfs. This concept provides similar
249 functionality as systemd-nspawn's --volatile= option, but provides it
250 on physical boots. Use this option for implementing stateless
251 systems, or testing systems with all state and/or configuration reset
252 to the defaults. (Note though that many distributions are not
253 prepared to boot up without a populated /etc or /var, though)
255 * systemd-gpt-auto-generator gained support for LUKS encrypted root
256 partitions. Previously it only supported LUKS encrypted partitions
257 for all other uses, except for the root partition itself.
259 * Socket units gained support for listening on AF_VSOCK sockets for
260 communication in virtualized QEMU environments.
262 * The "configure" script gained a new option --with-fallback-hostname=
263 for specifying the fallback hostname to use if none is configured in
264 /etc/hostname. For example, by specifying
265 --with-fallback-hostname=fedora it is possible to default to a
266 hostname of "fedora" when the user didn't specify anything
269 * systemd-cgls gained support for a new --unit= switch for listing only
270 the control groups of a specific unit. Similar --user-unit= has been
271 added for listing only the control groups of a specific user unit.
273 * systemd-mount gained a new --umount switch for unmounting a mount or
274 automount point (and all mount/automount points below it).
276 * systemd will now refuse full configuration reloads (via systemctl
277 daemon-reload and related calls) unless at least 16MiB of free space
278 are available in /run. This is a safety precaution in order to ensure
279 that generators can safely operate after the reload completed.
281 * A new unit file option RootImage= has been added, which has a similar
282 effect as RootDirectory= but mounts the service's root directory from
283 a disk image instead of plain directory. This logic reuses the same
284 image dissection and mount logic that systemd-nspawn already uses,
285 and hence supports any disk images systemd-nspawn supports, including
286 those following the Discoverable Partition Specification, as well as
287 Verity enabled images. This option enables systemd to run system
288 services directly off disk images acting as resource bundles,
289 possibly even including full integrity data.
291 * A new MountAPIVFS= unit file option has been added, taking a boolean
292 argument. If enabled /proc, /sys and /dev (collectively called the
293 "API VFS") will be mounted for the service. This is only relevant if
294 RootDirectory= or RootImage= is used for the service, as these mounts
295 are of course in place in the host mount namespace anyway.
297 * systemd-nspawn gained support for a new --pivot-root= switch. If
298 specified the root directory within the container image is pivoted to
299 the specified mount point, while the original root disk is moved to a
300 different place. This option enables booting of ostree images
301 directly with systemd-nspawn.
303 * systemd-networkd gained support for configuring IPv6 Proxy NDP
304 addresses via the new IPv6ProxyNDPAddress= .network file setting.
306 * The systemd build scripts will no longer complain if the NTP server
307 addresses are not changed from the defaults. Google is now supporting
308 these NTP servers officially. We still recommend downstreams to
309 properly register an NTP pool with the NTP pool project though.
311 * coredumpctl gained new new "--reverse" option for printing the list
312 of coredumps in reverse order.
314 * The systemd-coredump logic has been improved so that it may be reused
315 for collecting backtraces in non-compiled languages, for example in
316 scripting languages such as Python.
318 * machinectl will now show the UID shift of local containers, if user
319 namespacing is enabled for them.
321 * systemd will now optionally run "environment generator" binaries at
322 configuration load time. They may be used to add environment
323 variables to the environment block passed to services invoked. One
324 user environment generator is shipped by default that sets up
325 environment variables based on files dropped into
326 ~/.config/environment.d/.
328 Contributions from: Adrián López, Alexander Galanin, Alexander
329 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
330 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
331 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
332 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
333 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
334 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
335 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
336 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
337 Gianluca Boiano, Graeme Lawes, Hans de Goede, Harald Hoyer, Ian
338 Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan Synacek, Jason
339 Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen, Karl Kraus,
340 Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart Poettering,
341 Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de Vries, Maks
342 Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry, Mark
343 Stosberg, Martin Ejdestig, Martin Pitt, micah, Michael Biebl, Michael
344 Shields, Michal Schmidt, Michal Sekletar, Michel Kraus, Mike Gilbert,
345 Mirza Krak, Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter
346 Körner, Philip Withnall, Piotr Drąg, Ray Strode, Reverend Homer,
347 Rike-Benjamin Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan
348 Bilovol, sammynx, Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger,
349 Stefan Hajnoczi, Stefan Schweter, Susant Sahani, Sylvain Plantefève,
350 Taylor Smock, Thomas Blume, Thomas H. P. Andersen, Tobias Stoeckmann,
351 Tom Gundersen, Torstein Husebø, Viktar Vaŭčkievič, Viktor Mihajlovski,
352 Waldemar Brodkorb, Walter Garcia-Fontes, Wim de With, Yassine
353 Imounachen, Yi EungJun, Yu Watanabe, Zbigniew Jędrzejewski-Szmek,
360 * The new RemoveIPC= option can be used to remove IPC objects owned by
361 the user or group of a service when that service exits.
363 * The new ProtectKernelModules= option can be used to disable explicit
364 load and unload operations of kernel modules by a service. In
365 addition access to /usr/lib/modules is removed if this option is set.
367 * ProtectSystem= option gained a new value "strict", which causes the
368 whole file system tree with the exception of /dev, /proc, and /sys,
369 to be remounted read-only for a service.
371 * The new ProtectKernelTunables= option can be used to disable
372 modification of configuration files in /sys and /proc by a service.
373 Various directories and files are remounted read-only, so access is
374 restricted even if the file permissions would allow it.
376 * The new ProtectControlGroups= option can be used to disable write
377 access by a service to /sys/fs/cgroup.
379 * Various systemd services have been hardened with
380 ProtectKernelTunables=yes, ProtectControlGroups=yes,
381 RestrictAddressFamilies=.
383 * Support for dynamically creating users for the lifetime of a service
384 has been added. If DynamicUser=yes is specified, user and group IDs
385 will be allocated from the range 61184..65519 for the lifetime of the
386 service. They can be resolved using the new nss-systemd.so NSS
387 module. The module must be enabled in /etc/nsswitch.conf. Services
388 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
389 any resources allocated by the service will be cleaned up when the
390 service exits. They also have ProtectHome=read-only and
391 ProtectSystem=strict enabled, so they are not able to make any
392 permanent modifications to the system.
394 * The nss-systemd module also always resolves root and nobody, making
395 it possible to have no /etc/passwd or /etc/group files in minimal
396 container or chroot environments.
398 * Services may be started with their own user namespace using the new
399 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
400 under which the service is running are mapped. All other users are
403 * Support for the cgroup namespace has been added to systemd-nspawn. If
404 supported by kernel, the container system started by systemd-nspawn
405 will have its own view of the cgroup hierarchy. This new behaviour
406 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
408 * The new MemorySwapMax= option can be used to limit the maximum swap
409 usage under the unified cgroup hierarchy.
411 * Support for the CPU controller in the unified cgroup hierarchy has
412 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
413 options. This controller requires out-of-tree patches for the kernel
414 and the support is provisional.
416 * Mount and automount units may now be created transiently
417 (i.e. dynamically at runtime via the bus API, instead of requiring
418 unit files in the file system).
420 * systemd-mount is a new tool which may mount file systems – much like
421 mount(8), optionally pulling in additional dependencies through
422 transient .mount and .automount units. For example, this tool
423 automatically runs fsck on a backing block device before mounting,
424 and allows the automount logic to be used dynamically from the
425 command line for establishing mount points. This tool is particularly
426 useful when dealing with removable media, as it will ensure fsck is
427 run – if necessary – before the first access and that the file system
428 is quickly unmounted after each access by utilizing the automount
429 logic. This maximizes the chance that the file system on the
430 removable media stays in a clean state, and if it isn't in a clean
431 state is fixed automatically.
433 * LazyUnmount=yes option for mount units has been added to expose the
434 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
437 * /efi will be used as the mount point of the EFI boot partition, if
438 the directory is present, and the mount point was not configured
439 through other means (e.g. fstab). If /efi directory does not exist,
440 /boot will be used as before. This makes it easier to automatically
441 mount the EFI partition on systems where /boot is used for something
444 * When operating on GPT disk images for containers, systemd-nspawn will
445 now mount the ESP to /boot or /efi according to the same rules as PID
446 1 running on a host. This allows tools like "bootctl" to operate
447 correctly within such containers, in order to make container images
448 bootable on physical systems.
450 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
452 * Two new user session targets have been added to support running
453 graphical sessions under the systemd --user instance:
454 graphical-session.target and graphical-session-pre.target. See
455 systemd.special(7) for a description of how those targets should be
458 * The vconsole initialization code has been significantly reworked to
459 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
460 support unicode keymaps. Font and keymap configuration will now be
461 copied to all allocated virtual consoles.
463 * FreeBSD's bhyve virtualization is now detected.
465 * Information recorded in the journal for core dumps now includes the
466 contents of /proc/mountinfo and the command line of the process at
467 the top of the process hierarchy (which is usually the init process
470 * systemd-journal-gatewayd learned the --directory= option to serve
471 files from the specified location.
473 * journalctl --root=… can be used to peruse the journal in the
474 /var/log/ directories inside of a container tree. This is similar to
475 the existing --machine= option, but does not require the container to
478 * The hardware database has been extended to support
479 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
482 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
483 specify the click rate for mice which include a horizontal wheel with
484 a click rate that is different than the one for the vertical wheel.
486 * systemd-run gained a new --wait option that makes service execution
487 synchronous. (Specifically, the command will not return until the
488 specified service binary exited.)
490 * systemctl gained a new --wait option that causes the start command to
491 wait until the units being started have terminated again.
493 * A new journal output mode "short-full" has been added which displays
494 timestamps with abbreviated English day names and adds a timezone
495 suffix. Those timestamps include more information than the default
496 "short" output mode, and can be passed directly to journalctl's
497 --since= and --until= options.
499 * /etc/resolv.conf will be bind-mounted into containers started by
500 systemd-nspawn, if possible, so any changes to resolv.conf contents
501 are automatically propagated to the container.
503 * The number of instances for socket-activated services originating
504 from a single IP address can be limited with
505 MaxConnectionsPerSource=, extending the existing setting of
508 * systemd-networkd gained support for vcan ("Virtual CAN") interface
511 * .netdev and .network configuration can now be extended through
514 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
515 Segmentation Offload, Generic Receive Offload, Large Receive Offload
516 can be enabled and disabled using the new UDPSegmentationOffload=,
517 TCPSegmentationOffload=, GenericSegmentationOffload=,
518 GenericReceiveOffload=, LargeReceiveOffload= options in the
519 [Link] section of .link files.
521 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
522 Port VLAN ID can be configured for bridge devices using the new STP=,
523 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
524 section of .netdev files.
526 * The route table to which routes received over DHCP or RA should be
527 added can be configured with the new RouteTable= option in the [DHCP]
528 and [IPv6AcceptRA] sections of .network files.
530 * The Address Resolution Protocol can be disabled on links managed by
531 systemd-networkd using the ARP=no setting in the [Link] section of
534 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
535 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
536 encode information about the result and exit codes of the current
537 service runtime cycle.
539 * systemd-sysctl will now configure kernel parameters in the order
540 they occur in the configuration files. This matches what sysctl
541 has been traditionally doing.
543 * kernel-install "plugins" that are executed to perform various
544 tasks after a new kernel is added and before an old one is removed
545 can now return a special value to terminate the procedure and
546 prevent any later plugins from running.
548 * Journald's SplitMode=login setting has been deprecated. It has been
549 removed from documentation, and its use is discouraged. In a future
550 release it will be completely removed, and made equivalent to current
551 default of SplitMode=uid.
553 * Storage=both option setting in /etc/systemd/coredump.conf has been
554 removed. With fast LZ4 compression storing the core dump twice is not
557 * The --share-system systemd-nspawn option has been replaced with an
558 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
559 this functionality is discouraged. In addition the variables
560 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
561 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
562 individual namespaces.
564 * "machinectl list" now shows the IP address of running containers in
565 the output, as well as OS release information.
567 * "loginctl list" now shows the TTY of each session in the output.
569 * sd-bus gained new API calls sd_bus_track_set_recursive(),
570 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
571 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
572 tracking objects in a "recursive" mode, where a single client can be
573 counted multiple times, if it takes multiple references.
575 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
576 sd_bus_get_exit_on_disconnect(). They may be used to to make a
577 process using sd-bus automatically exit if the bus connection is
580 * Bus clients of the service manager may now "pin" loaded units into
581 memory, by taking an explicit reference on them. This is useful to
582 ensure the client can retrieve runtime data about the service even
583 after the service completed execution. Taking such a reference is
584 available only for privileged clients and should be helpful to watch
585 running services in a race-free manner, and in particular collect
586 information about exit statuses and results.
588 * The nss-resolve module has been changed to strictly return UNAVAIL
589 when communication via D-Bus with resolved failed, and NOTFOUND when
590 a lookup completed but was negative. This means it is now possible to
591 neatly configure fallbacks using nsswitch.conf result checking
592 expressions. Taking benefit of this, the new recommended
593 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
595 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
597 * A new setting CtrlAltDelBurstAction= has been added to
598 /etc/systemd/system.conf which may be used to configure the precise
599 behaviour if the user on the console presses Ctrl-Alt-Del more often
600 than 7 times in 2s. Previously this would unconditionally result in
601 an expedited, immediate reboot. With this new setting the precise
602 operation may be configured in more detail, and also turned off
605 * In .netdev files two new settings RemoteChecksumTx= and
606 RemoteChecksumRx= are now understood that permit configuring the
607 remote checksumming logic for VXLAN networks.
609 * The service manager learnt a new "invocation ID" concept for invoked
610 services. Each runtime cycle of a service will get a new invocation
611 ID (a 128bit random UUID) assigned that identifies the current
612 run of the service uniquely and globally. A new invocation ID
613 is generated each time a service starts up. The journal will store
614 the invocation ID of a service along with any logged messages, thus
615 making the invocation ID useful for matching the online runtime of a
616 service with the offline log data it generated in a safe way without
617 relying on synchronized timestamps. In many ways this new service
618 invocation ID concept is similar to the kernel's boot ID concept that
619 uniquely and globally identifies the runtime of each boot. The
620 invocation ID of a service is passed to the service itself via an
621 environment variable ($INVOCATION_ID). A new bus call
622 GetUnitByInvocationID() has been added that is similar to GetUnit()
623 but instead of retrieving the bus path for a unit by its name
624 retrieves it by its invocation ID. The returned path is valid only as
625 long as the passed invocation ID is current.
627 * systemd-resolved gained a new "DNSStubListener" setting in
628 resolved.conf. It either takes a boolean value or the special values
629 "udp" and "tcp", and configures whether to enable the stub DNS
630 listener on 127.0.0.53:53.
632 * IP addresses configured via networkd may now carry additional
633 configuration settings supported by the kernel. New options include:
634 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
635 PrefixRoute=, AutoJoin=.
637 * The PAM configuration fragment file for "user@.service" shipped with
638 systemd (i.e. the --user instance of systemd) has been stripped to
639 the minimum necessary to make the system boot. Previously, it
640 contained Fedora-specific stanzas that did not apply to other
641 distributions. It is expected that downstream distributions add
642 additional configuration lines, matching their needs to this file,
643 using it only as rough template of what systemd itself needs. Note
644 that this reduced fragment does not even include an invocation of
645 pam_limits which most distributions probably want to add, even though
646 systemd itself does not need it. (There's also the new build time
647 option --with-pamconfdir=no to disable installation of the PAM
650 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
651 capability is now also dropped from its set (in addition to
652 CAP_SYS_MKNOD as before).
654 * In service unit files it is now possible to connect a specific named
655 file descriptor with stdin/stdout/stdout of an executed service. The
656 name may be specified in matching .socket units using the
657 FileDescriptorName= setting.
659 * A number of journal settings may now be configured on the kernel
660 command line. Specifically, the following options are now understood:
661 systemd.journald.max_level_console=,
662 systemd.journald.max_level_store=,
663 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
664 systemd.journald.max_level_wall=.
666 * "systemctl is-enabled --full" will now show by which symlinks a unit
667 file is enabled in the unit dependency tree.
669 * Support for VeraCrypt encrypted partitions has been added to the
670 "cryptsetup" logic and /etc/crypttab.
672 * systemd-detect-virt gained support for a new --private-users switch
673 that checks whether the invoking processes are running inside a user
674 namespace. Similar, a new special value "private-users" for the
675 existing ConditionVirtualization= setting has been added, permitting
676 skipping of specific units in user namespace environments.
678 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
679 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
680 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
681 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
682 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
683 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
684 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
685 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
686 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
687 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
688 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
689 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
690 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
691 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
692 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
693 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
694 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
695 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
696 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
697 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
698 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
699 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
700 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
701 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
702 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
703 Jędrzejewski-Szmek, Zeal Jagannatha
705 — Santa Fe, 2016-11-03
709 * In service units the various ExecXYZ= settings have been extended
710 with an additional special character as first argument of the
711 assigned value: if the character '+' is used the specified command
712 line it will be run with full privileges, regardless of User=,
713 Group=, CapabilityBoundingSet= and similar options. The effect is
714 similar to the existing PermissionsStartOnly= option, but allows
715 configuration of this concept for each executed command line
718 * Services may now alter the service watchdog timeout at runtime by
719 sending a WATCHDOG_USEC= message via sd_notify().
721 * MemoryLimit= and related unit settings now optionally take percentage
722 specifications. The percentage is taken relative to the amount of
723 physical memory in the system (or in case of containers, the assigned
724 amount of memory). This allows scaling service resources neatly with
725 the amount of RAM available on the system. Similarly, systemd-logind's
726 RuntimeDirectorySize= option now also optionally takes percentage
729 * In similar fashion TasksMax= takes percentage values now, too. The
730 value is taken relative to the configured maximum number of processes
731 on the system. The per-service task maximum has been changed to 15%
732 using this functionality. (Effectively this is an increase of 512 →
733 4915 for service units, given the kernel's default pid_max setting.)
735 * Calendar time specifications in .timer units now understand a ".."
736 syntax for time ranges. Example: "4..7:10" may now be used for
737 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
740 * The InaccessableDirectories=, ReadOnlyDirectories= and
741 ReadWriteDirectories= unit file settings have been renamed to
742 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
743 applied to all kinds of file nodes, and not just directories, with
744 the exception of symlinks. Specifically these settings may now be
745 used on block and character device nodes, UNIX sockets and FIFOS as
746 well as regular files. The old names of these settings remain
747 available for compatibility.
749 * systemd will now log about all service processes it kills forcibly
750 (using SIGKILL) because they remained after the clean shutdown phase
751 of the service completed. This should help identifying services that
752 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
753 systemd-logind's configuration a similar log message is generated for
754 processes killed at the end of each session due to this setting.
756 * systemd will now set the $JOURNAL_STREAM environment variable for all
757 services whose stdout/stderr are connected to the Journal (which
758 effectively means by default: all services). The variable contains
759 the device and inode number of the file descriptor used for
760 stdout/stderr. This may be used by invoked programs to detect whether
761 their stdout/stderr is connected to the Journal, in which case they
762 can switch over to direct Journal communication, thus being able to
763 pass extended, structured metadata along with their log messages. As
764 one example, this is now used by glib's logging primitives.
766 * When using systemd's default tmp.mount unit for /tmp, the mount point
767 will now be established with the "nosuid" and "nodev" options. This
768 avoids privilege escalation attacks that put traps and exploits into
769 /tmp. However, this might cause problems if you e. g. put container
770 images or overlays into /tmp; if you need this, override tmp.mount's
771 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
774 * systemd now supports the "memory" cgroup controller also on
777 * The systemd-cgtop tool now optionally takes a control group path as
778 command line argument. If specified, the control group list shown is
779 limited to subgroups of that group.
781 * The SystemCallFilter= unit file setting gained support for
782 pre-defined, named system call filter sets. For example
783 SystemCallFilter=@clock is now an effective way to make all clock
784 changing-related system calls unavailable to a service. A number of
785 similar pre-defined groups are defined. Writing system call filters
786 for system services is simplified substantially with this new
787 concept. Accordingly, all of systemd's own, long-running services now
788 enable system call filtering based on this, by default.
790 * A new service setting MemoryDenyWriteExecute= has been added, taking
791 a boolean value. If turned on, a service may no longer create memory
792 mappings that are writable and executable at the same time. This
793 enhances security for services where this is enabled as it becomes
794 harder to dynamically write and then execute memory in exploited
795 service processes. This option has been enabled for all of systemd's
796 own long-running services.
798 * A new RestrictRealtime= service setting has been added, taking a
799 boolean argument. If set the service's processes may no longer
800 acquire realtime scheduling. This improves security as realtime
801 scheduling may otherwise be used to easily freeze the system.
803 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
804 value. This may be used for requesting that the system manager inside
805 of the container reports start-up completion to nspawn which then
806 propagates this notification further to the service manager
807 supervising nspawn itself. A related option NotifyReady= in .nspawn
808 files has been added too. This functionality allows ordering of the
809 start-up of multiple containers using the usual systemd ordering
812 * machinectl gained a new command "stop" that is an alias for
815 * systemd-resolved gained support for contacting DNS servers on
816 link-local IPv6 addresses.
818 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
819 its caches. A method call for requesting the same operation has been
820 added to the bus API too, and is made available via "systemd-resolve
823 * systemd-resolve gained a new --status switch. If passed a brief
824 summary of the used DNS configuration with per-interface information
827 * resolved.conf gained a new Cache= boolean option, defaulting to
828 on. If turned off local DNS caching is disabled. This comes with a
829 performance penalty in particular when DNSSEC is enabled. Note that
830 resolved disables its internal caching implicitly anyway, when the
831 configured DNS server is on a host-local IP address such as ::1 or
832 127.0.0.1, thus automatically avoiding double local caching.
834 * systemd-resolved now listens on the local IP address 127.0.0.53:53
835 for DNS requests. This improves compatibility with local programs
836 that do not use the libc NSS or systemd-resolved's bus APIs for name
837 resolution. This minimal DNS service is only available to local
838 programs and does not implement the full DNS protocol, but enough to
839 cover local DNS clients. A new, static resolv.conf file, listing just
840 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
841 now recommended to make /etc/resolv.conf a symlink to this file in
842 order to route all DNS lookups to systemd-resolved, regardless if
843 done via NSS, the bus API or raw DNS packets. Note that this local
844 DNS service is not as fully featured as the libc NSS or
845 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
846 used to deliver link-local address information (as this implies
847 sending a local interface index along), LLMNR/mDNS support via this
848 interface is severely restricted. It is thus strongly recommended for
849 all applications to use the libc NSS API or native systemd-resolved
852 * systemd-networkd's bridge support learned a new setting
853 VLANFiltering= for controlling VLAN filtering. Moreover a new section
854 in .network files has been added for configuring VLAN bridging in
855 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
857 * systemd-networkd's IPv6 Router Advertisement code now makes use of
858 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
859 now be acquired without relying on DHCPv6. Two new options
860 UseDomains= and UseDNS= have been added to configure this behaviour.
862 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
863 renamed IPv6AcceptRA=, without altering its behaviour. The old
864 setting name remains available for compatibility reasons.
866 * The systemd-networkd VTI/VTI6 tunneling support gained new options
867 Key=, InputKey= and OutputKey=.
869 * systemd-networkd gained support for VRF ("Virtual Routing Function")
870 interface configuration.
872 * "systemctl edit" may now be used to create new unit files by
873 specifying the --force switch.
875 * sd-event gained a new function sd_event_get_iteration() for
876 requesting the current iteration counter of the event loop. It starts
877 at zero and is increased by one with each event loop iteration.
879 * A new rpm macro %systemd_ordering is provided by the macros.systemd
880 file. It can be used in lieu of %systemd_requires in packages which
881 don't use any systemd functionality and are intended to be installed
882 in minimal containers without systemd present. This macro provides
883 ordering dependencies to ensure that if the package is installed in
884 the same rpm transaction as systemd, systemd will be installed before
885 the scriptlets for the package are executed, allowing unit presets
888 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
889 been added to simplify packaging of generators.
891 * The os-release file gained VERSION_CODENAME field for the
892 distribution nickname (e.g. VERSION_CODENAME=woody).
894 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
895 can be set to disable parsing of metadata and the creation
896 of persistent symlinks for that device.
898 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
899 to make them available to logged-in users has been reverted.
901 * Much of the common code of the various systemd components is now
902 built into an internal shared library libsystemd-shared-231.so
903 (incorporating the systemd version number in the name, to be updated
904 with future releases) that the components link to. This should
905 decrease systemd footprint both in memory during runtime and on
906 disk. Note that the shared library is not for public use, and is
907 neither API not ABI stable, but is likely to change with every new
908 released update. Packagers need to make sure that binaries
909 linking to libsystemd-shared.so are updated in step with the
912 * Configuration for "mkosi" is now part of the systemd
913 repository. mkosi is a tool to easily build legacy-free OS images,
914 and is available on github: https://github.com/systemd/mkosi. If
915 "mkosi" is invoked in the build tree a new raw OS image is generated
916 incorporating the systemd sources currently being worked on and a
917 clean, fresh distribution installation. The generated OS image may be
918 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
919 UEFI PC. This functionality is particularly useful to easily test
920 local changes made to systemd in a pristine, defined environment. See
923 * configure learned the --with-support-url= option to specify the
924 distribution's bugtracker.
926 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
927 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
928 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
929 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
930 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
931 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
932 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
933 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
934 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
935 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
936 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
937 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
938 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
939 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
940 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
941 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
942 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
943 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
944 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
950 * DNSSEC is now turned on by default in systemd-resolved (in
951 "allow-downgrade" mode), but may be turned off during compile time by
952 passing "--with-default-dnssec=no" to "configure" (and of course,
953 during runtime with DNSSEC= in resolved.conf). We recommend
954 downstreams to leave this on at least during development cycles and
955 report any issues with the DNSSEC logic upstream. We are very
956 interested in collecting feedback about the DNSSEC validator and its
957 limitations in the wild. Note however, that DNSSEC support is
958 probably nothing downstreams should turn on in stable distros just
959 yet, as it might create incompatibilities with a few DNS servers and
960 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
961 automatically whenever we detect such incompatible setups, but there
962 might be systems we do not cover yet. Hence: please help us testing
963 the DNSSEC code, leave this on where you can, report back, but then
964 again don't consider turning this on in your stable, LTS or
965 production release just yet. (Note that you have to enable
966 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
967 and its DNSSEC mode for host name resolution from local
970 * systemd-resolve conveniently resolves DANE records with the --tlsa
971 option and OPENPGPKEY records with the --openpgp option. It also
972 supports dumping raw DNS record data via the new --raw= switch.
974 * systemd-logind will now by default terminate user processes that are
975 part of the user session scope unit (session-XX.scope) when the user
976 logs out. This behavior is controlled by the KillUserProcesses=
977 setting in logind.conf, and the previous default of "no" is now
978 changed to "yes". This means that user sessions will be properly
979 cleaned up after, but additional steps are necessary to allow
980 intentionally long-running processes to survive logout.
982 While the user is logged in at least once, user@.service is running,
983 and any service that should survive the end of any individual login
984 session can be started at a user service or scope using systemd-run.
985 systemd-run(1) man page has been extended with an example which shows
986 how to run screen in a scope unit underneath user@.service. The same
987 command works for tmux.
989 After the user logs out of all sessions, user@.service will be
990 terminated too, by default, unless the user has "lingering" enabled.
991 To effectively allow users to run long-term tasks even if they are
992 logged out, lingering must be enabled for them. See loginctl(1) for
993 details. The default polkit policy was modified to allow users to
994 set lingering for themselves without authentication.
996 Previous defaults can be restored at compile time by the
997 --without-kill-user-processes option to "configure".
999 * systemd-logind gained new configuration settings SessionsMax= and
1000 InhibitorsMax=, both with a default of 8192. It will not register new
1001 user sessions or inhibitors above this limit.
1003 * systemd-logind will now reload configuration on SIGHUP.
1005 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
1006 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
1007 enable. Also, support for the "io" cgroup controller in the unified
1008 hierarchy has been added, so that the "memory", "pids" and "io" are
1009 now the controllers that are supported on the unified hierarchy.
1011 WARNING: it is not possible to use previous systemd versions with
1012 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
1013 is necessary to also update systemd in the initramfs if using the
1014 unified hierarchy. An updated SELinux policy is also required.
1016 * LLDP support has been extended, and both passive (receive-only) and
1017 active (sender) modes are supported. Passive mode ("routers-only") is
1018 enabled by default in systemd-networkd. Active LLDP mode is enabled
1019 by default for containers on the internal network. The "networkctl
1020 lldp" command may be used to list information gathered. "networkctl
1021 status" will also show basic LLDP information on connected peers now.
1023 * The IAID and DUID unique identifier sent in DHCP requests may now be
1024 configured for the system and each .network file managed by
1025 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
1027 * systemd-networkd gained support for configuring proxy ARP support for
1028 each interface, via the ProxyArp= setting in .network files. It also
1029 gained support for configuring the multicast querier feature of
1030 bridge devices, via the new MulticastQuerier= setting in .netdev
1031 files. Similarly, snooping on the IGMP traffic can be controlled
1032 via the new setting MulticastSnooping=.
1034 A new setting PreferredLifetime= has been added for addresses
1035 configured in .network file to configure the lifetime intended for an
1038 The systemd-networkd DHCP server gained the option EmitRouter=, which
1039 defaults to yes, to configure whether the DHCP Option 3 (Router)
1042 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
1043 systemd-socket-activate and installed into /usr/bin. It is now fully
1046 * systemd-journald now uses separate threads to flush changes to disk
1047 when closing journal files, thus reducing impact of slow disk I/O on
1048 logging performance.
1050 * The sd-journal API gained two new calls
1051 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
1052 can be used to open journal files using file descriptors instead of
1053 file or directory paths. sd_journal_open_container() has been
1054 deprecated, sd_journal_open_directory_fd() should be used instead
1055 with the flag SD_JOURNAL_OS_ROOT.
1057 * journalctl learned a new output mode "-o short-unix" that outputs log
1058 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
1059 UTC). It also gained support for a new --no-hostname setting to
1060 suppress the hostname column in the family of "short" output modes.
1062 * systemd-ask-password now optionally skips printing of the password to
1063 stdout with --no-output which can be useful in scripts.
1065 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
1066 (devices tagged with ID_MAKER_TOOL) are now tagged with
1067 "uaccess" and are available to logged in users.
1069 * The DeviceAllow= unit setting now supports specifiers (with "%").
1071 * "systemctl show" gained a new --value switch, which allows print a
1072 only the contents of a specific unit property, without also printing
1073 the property's name. Similar support was added to "show*" verbs
1074 of loginctl and machinectl that output "key=value" lists.
1076 * A new unit type "generated" was added for files dynamically generated
1077 by generator tools. Similarly, a new unit type "transient" is used
1078 for unit files created using the runtime API. "systemctl enable" will
1079 refuse to operate on such files.
1081 * A new command "systemctl revert" has been added that may be used to
1082 revert to the vendor version of a unit file, in case local changes
1083 have been made by adding drop-ins or overriding the unit file.
1085 * "machinectl clean" gained a new verb to automatically remove all or
1086 just hidden container images.
1088 * systemd-tmpfiles gained support for a new line type "e" for emptying
1089 directories, if they exist, without creating them if they don't.
1091 * systemd-nspawn gained support for automatically patching the UID/GIDs
1092 of the owners and the ACLs of all files and directories in a
1093 container tree to match the UID/GID user namespacing range selected
1094 for the container invocation. This mode is enabled via the new
1095 --private-users-chown switch. It also gained support for
1096 automatically choosing a free, previously unused UID/GID range when
1097 starting a container, via the new --private-users=pick setting (which
1098 implies --private-users-chown). Together, these options for the first
1099 time make user namespacing for nspawn containers fully automatic and
1100 thus deployable. The systemd-nspawn@.service template unit file has
1101 been changed to use this functionality by default.
1103 * systemd-nspawn gained a new --network-zone= switch, that allows
1104 creating ad-hoc virtual Ethernet links between multiple containers,
1105 that only exist as long as at least one container referencing them is
1106 running. This allows easy connecting of multiple containers with a
1107 common link that implements an Ethernet broadcast domain. Each of
1108 these network "zones" may be named relatively freely by the user, and
1109 may be referenced by any number of containers, but each container may
1110 only reference one of these "zones". On the lower level, this is
1111 implemented by an automatically managed bridge network interface for
1112 each zone, that is created when the first container referencing its
1113 zone is created and removed when the last one referencing its zone
1116 * The default start timeout may now be configured on the kernel command
1117 line via systemd.default_timeout_start_sec=. It was already
1118 configurable via the DefaultTimeoutStartSec= option in
1119 /etc/systemd/system.conf.
1121 * Socket units gained a new TriggerLimitIntervalSec= and
1122 TriggerLimitBurst= setting to configure a limit on the activation
1123 rate of the socket unit.
1125 * The LimitNICE= setting now optionally takes normal UNIX nice values
1126 in addition to the raw integer limit value. If the specified
1127 parameter is prefixed with "+" or "-" and is in the range -20..19 the
1128 value is understood as UNIX nice value. If not prefixed like this it
1129 is understood as raw RLIMIT_NICE limit.
1131 * Note that the effect of the PrivateDevices= unit file setting changed
1132 slightly with this release: the per-device /dev file system will be
1133 mounted read-only from this version on, and will have "noexec"
1134 set. This (minor) change of behavior might cause some (exceptional)
1135 legacy software to break, when PrivateDevices=yes is set for its
1136 service. Please leave PrivateDevices= off if you run into problems
1139 * systemd-bootchart has been split out to a separate repository:
1140 https://github.com/systemd/systemd-bootchart
1142 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
1143 merged into the kernel in its current form.
1145 * The compatibility libraries libsystemd-daemon.so,
1146 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
1147 which have been deprecated since systemd-209 have been removed along
1148 with the corresponding pkg-config files. All symbols provided by
1149 those libraries are provided by libsystemd.so.
1151 * The Capabilities= unit file setting has been removed (it is ignored
1152 for backwards compatibility). AmbientCapabilities= and
1153 CapabilityBoundingSet= should be used instead.
1155 * A new special target has been added, initrd-root-device.target,
1156 which creates a synchronization point for dependencies of the root
1157 device in early userspace. Initramfs builders must ensure that this
1158 target is now included in early userspace.
1160 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
1161 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
1162 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
1163 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
1164 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
1165 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
1166 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
1167 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
1168 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
1169 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
1170 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
1171 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
1172 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
1173 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
1174 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
1175 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
1176 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
1177 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
1178 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
1179 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
1180 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
1181 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
1182 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
1183 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
1184 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
1187 — Fairfax, 2016-05-21
1191 * The systemd-resolved DNS resolver service has gained a substantial
1192 set of new features, most prominently it may now act as a DNSSEC
1193 validating stub resolver. DNSSEC mode is currently turned off by
1194 default, but is expected to be turned on by default in one of the
1195 next releases. For now, we invite everybody to test the DNSSEC logic
1196 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
1197 service also gained a full set of D-Bus interfaces, including calls
1198 to configure DNS and DNSSEC settings per link (for use by external
1199 network management software). systemd-resolved and systemd-networkd
1200 now distinguish between "search" and "routing" domains. The former
1201 are used to qualify single-label names, the latter are used purely
1202 for routing lookups within certain domains to specific links.
1203 resolved now also synthesizes RRs for all entries from /etc/hosts.
1205 * The systemd-resolve tool (which is a client utility for
1206 systemd-resolved) has been improved considerably and is now fully
1207 supported and documented. Hence it has moved from /usr/lib/systemd to
1210 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
1213 * The coredump collection logic has been reworked: when a coredump is
1214 collected it is now written to disk, compressed and processed
1215 (including stacktrace extraction) from a new instantiated service
1216 systemd-coredump@.service, instead of directly from the
1217 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
1218 processing large coredumps can take up a substantial amount of
1219 resources and time, and this previously happened entirely outside of
1220 systemd's service supervision. With the new logic the core_pattern
1221 hook only does minimal metadata collection before passing off control
1222 to the new instantiated service, which is configured with a time
1223 limit, a nice level and other settings to minimize negative impact on
1224 the rest of the system. Also note that the new logic will honour the
1225 RLIMIT_CORE setting of the crashed process, which now allows users
1226 and processes to turn off coredumping for their processes by setting
1229 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
1230 and all forked processes by default. Previously, PID 1 would leave
1231 the setting at "0" for all processes, as set by the kernel. Note that
1232 the resource limit traditionally has no effect on the generated
1233 coredumps on the system if the /proc/sys/kernel/core_pattern hook
1234 logic is used. Since the limit is now honoured (see above) its
1235 default has been changed so that the coredumping logic is enabled by
1236 default for all processes, while allowing specific opt-out.
1238 * When the stacktrace is extracted from processes of system users, this
1239 is now done as "systemd-coredump" user, in order to sandbox this
1240 potentially security sensitive parsing operation. (Note that when
1241 processing coredumps of normal users this is done under the user ID
1242 of process that crashed, as before.) Packagers should take notice
1243 that it is now necessary to create the "systemd-coredump" system user
1244 and group at package installation time.
1246 * The systemd-activate socket activation testing tool gained support
1247 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
1248 and --seqpacket switches. It also has been extended to support both
1249 new-style and inetd-style file descriptor passing. Use the new
1250 --inetd switch to request inetd-style file descriptor passing.
1252 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
1253 variable, which takes a boolean value. If set to false, ANSI color
1254 output is disabled in the tools even when run on a terminal that
1257 * The VXLAN support in networkd now supports two new settings
1258 DestinationPort= and PortRange=.
1260 * A new systemd.machine_id= kernel command line switch has been added,
1261 that may be used to set the machine ID in /etc/machine-id if it is
1262 not initialized yet. This command line option has no effect if the
1263 file is already initialized.
1265 * systemd-nspawn gained a new --as-pid2 switch that invokes any
1266 specified command line as PID 2 rather than PID 1 in the
1267 container. In this mode PID 1 is a minimal stub init process that
1268 implements the special POSIX and Linux semantics of PID 1 regarding
1269 signal and child process management. Note that this stub init process
1270 is implemented in nspawn itself and requires no support from the
1271 container image. This new logic is useful to support running
1272 arbitrary commands in the container, as normal processes are
1273 generally not prepared to run as PID 1.
1275 * systemd-nspawn gained a new --chdir= switch for setting the current
1276 working directory for the process started in the container.
1278 * "journalctl /dev/sda" will now output all kernel log messages for
1279 specified device from the current boot, in addition to all devices
1280 that are parents of it. This should make log output about devices
1281 pretty useful, as long as kernel drivers attach enough metadata to
1282 the log messages. (The usual SATA drivers do.)
1284 * The sd-journal API gained two new calls
1285 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
1286 that report whether log data from /run or /var has been found.
1288 * journalctl gained a new switch "--fields" that prints all journal
1289 record field names currently in use in the journal. This is backed
1290 by two new sd-journal API calls sd_journal_enumerate_fields() and
1291 sd_journal_restart_fields().
1293 * Most configurable timeouts in systemd now expect an argument of
1294 "infinity" to turn them off, instead of "0" as before. The semantics
1295 from now on is that a timeout of "0" means "now", and "infinity"
1296 means "never". To maintain backwards compatibility, "0" continues to
1297 turn off previously existing timeout settings.
1299 * "systemctl reload-or-try-restart" has been renamed to "systemctl
1300 try-reload-or-restart" to clarify what it actually does: the "try"
1301 logic applies to both reloading and restarting, not just restarting.
1302 The old name continues to be accepted for compatibility.
1304 * On boot-up, when PID 1 detects that the system clock is behind the
1305 release date of the systemd version in use, the clock is now set
1306 to the latter. Previously, this was already done in timesyncd, in order
1307 to avoid running with clocks set to the various clock epochs such as
1308 1902, 1938 or 1970. With this change the logic is now done in PID 1
1309 in addition to timesyncd during early boot-up, so that it is enforced
1310 before the first process is spawned by systemd. Note that the logic
1311 in timesyncd remains, as it is more comprehensive and ensures
1312 clock monotonicity by maintaining a persistent timestamp file in
1313 /var. Since /var is generally not available in earliest boot or the
1314 initrd, this part of the logic remains in timesyncd, and is not done
1317 * Support for tweaking details in net_cls.class_id through the
1318 NetClass= configuration directive has been removed, as the kernel
1319 people have decided to deprecate that controller in cgroup v2.
1320 Userspace tools such as nftables are moving over to setting rules
1321 that are specific to the full cgroup path of a task, which obsoletes
1322 these controllers anyway. The NetClass= directive is kept around for
1323 legacy compatibility reasons. For a more in-depth description of the
1324 kernel change, please refer to the respective upstream commit:
1326 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
1328 * A new service setting RuntimeMaxSec= has been added that may be used
1329 to specify a maximum runtime for a service. If the timeout is hit, the
1330 service is terminated and put into a failure state.
1332 * A new service setting AmbientCapabilities= has been added. It allows
1333 configuration of additional Linux process capabilities that are
1334 passed to the activated processes. This is only available on very
1337 * The process resource limit settings in service units may now be used
1338 to configure hard and soft limits individually.
1340 * The various libsystemd APIs such as sd-bus or sd-event now publicly
1341 expose support for gcc's __attribute__((cleanup())) C extension.
1342 Specifically, for many object destructor functions alternative
1343 versions have been added that have names suffixed with "p" and take a
1344 pointer to a pointer to the object to destroy, instead of just a
1345 pointer to the object itself. This is useful because these destructor
1346 functions may be used directly as parameters to the cleanup
1347 construct. Internally, systemd has been a heavy user of this GCC
1348 extension for a long time, and with this change similar support is
1349 now available to consumers of the library outside of systemd. Note
1350 that by using this extension in your sources compatibility with old
1351 and strictly ANSI compatible C compilers is lost. However, all gcc or
1352 LLVM versions of recent years support this extension.
1354 * Timer units gained support for a new setting RandomizedDelaySec= that
1355 allows configuring some additional randomized delay to the configured
1356 time. This is useful to spread out timer events to avoid load peaks in
1357 clusters or larger setups.
1359 * Calendar time specifications now support sub-second accuracy.
1361 * Socket units now support listening on SCTP and UDP-lite protocol
1364 * The sd-event API now comes with a full set of man pages.
1366 * Older versions of systemd contained experimental support for
1367 compressing journal files and coredumps with the LZ4 compressor that
1368 was not compatible with the lz4 binary (due to API limitations of the
1369 lz4 library). This support has been removed; only support for files
1370 compatible with the lz4 binary remains. This LZ4 logic is now
1371 officially supported and no longer considered experimental.
1373 * The dkr image import logic has been removed again from importd. dkr's
1374 micro-services focus doesn't fit into the machine image focus of
1375 importd, and quickly got out of date with the upstream dkr API.
1377 * Creation of the /run/lock/lockdev/ directory was dropped from
1378 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
1379 been available for many years. If you still need this, you need to
1380 create your own tmpfiles.d config file with:
1382 d /run/lock/lockdev 0775 root lock -
1384 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
1385 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
1386 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
1387 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
1388 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
1389 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
1390 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
1391 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
1392 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
1393 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
1394 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
1395 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
1396 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
1397 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
1398 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
1399 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
1400 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
1401 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
1402 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
1404 — Berlin, 2016-02-11
1408 * A number of properties previously only settable in unit
1409 files are now also available as properties to set when
1410 creating transient units programmatically via the bus, as it
1411 is exposed with systemd-run's --property=
1412 setting. Specifically, these are: SyslogIdentifier=,
1413 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
1414 EnvironmentFile=, ReadWriteDirectories=,
1415 ReadOnlyDirectories=, InaccessibleDirectories=,
1416 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
1418 * When creating transient services via the bus API it is now
1419 possible to pass in a set of file descriptors to use as
1420 STDIN/STDOUT/STDERR for the invoked process.
1422 * Slice units may now be created transiently via the bus APIs,
1423 similar to the way service and scope units may already be
1424 created transiently.
1426 * Wherever systemd expects a calendar timestamp specification
1427 (like in journalctl's --since= and --until= switches) UTC
1428 timestamps are now supported. Timestamps suffixed with "UTC"
1429 are now considered to be in Universal Time Coordinated
1430 instead of the local timezone. Also, timestamps may now
1431 optionally be specified with sub-second accuracy. Both of
1432 these additions also apply to recurring calendar event
1433 specification, such as OnCalendar= in timer units.
1435 * journalctl gained a new "--sync" switch that asks the
1436 journal daemon to write all so far unwritten log messages to
1437 disk and sync the files, before returning.
1439 * systemd-tmpfiles learned two new line types "q" and "Q" that
1440 operate like "v", but also set up a basic btrfs quota
1441 hierarchy when used on a btrfs file system with quota
1444 * tmpfiles' "v", "q" and "Q" will now create a plain directory
1445 instead of a subvolume (even on a btrfs file system) if the
1446 root directory is a plain directory, and not a
1447 subvolume. This should simplify things with certain chroot()
1448 environments which are not aware of the concept of btrfs
1451 * systemd-detect-virt gained a new --chroot switch to detect
1452 whether execution takes place in a chroot() environment.
1454 * CPUAffinity= now takes CPU index ranges in addition to
1457 * The various memory-related resource limit settings (such as
1458 LimitAS=) now understand the usual K, M, G, ... suffixes to
1459 the base of 1024 (IEC). Similar, the time-related resource
1460 limit settings understand the usual min, h, day, ...
1463 * There's a new system.conf setting DefaultTasksMax= to
1464 control the default TasksMax= setting for services and
1465 scopes running on the system. (TasksMax= is the primary
1466 setting that exposes the "pids" cgroup controller on systemd
1467 and was introduced in the previous systemd release.) The
1468 setting now defaults to 512, which means services that are
1469 not explicitly configured otherwise will only be able to
1470 create 512 processes or threads at maximum, from this
1471 version on. Note that this means that thread- or
1472 process-heavy services might need to be reconfigured to set
1473 TasksMax= to a higher value. It is sufficient to set
1474 TasksMax= in these specific unit files to a higher value, or
1475 even "infinity". Similar, there's now a logind.conf setting
1476 UserTasksMax= that defaults to 4096 and limits the total
1477 number of processes or tasks each user may own
1478 concurrently. nspawn containers also have the TasksMax=
1479 value set by default now, to 8192. Note that all of this
1480 only has an effect if the "pids" cgroup controller is
1481 enabled in the kernel. The general benefit of these changes
1482 should be a more robust and safer system, that provides a
1483 certain amount of per-service fork() bomb protection.
1485 * systemd-nspawn gained the new --network-veth-extra= switch
1486 to define additional and arbitrarily-named virtual Ethernet
1487 links between the host and the container.
1489 * A new service execution setting PassEnvironment= has been
1490 added that allows importing select environment variables
1491 from PID1's environment block into the environment block of
1494 * Timer units gained support for a new RemainAfterElapse=
1495 setting which takes a boolean argument. It defaults to on,
1496 exposing behaviour unchanged to previous releases. If set to
1497 off, timer units are unloaded after they elapsed if they
1498 cannot elapse again. This is particularly useful for
1499 transient timer units, which shall not stay around longer
1500 than until they first elapse.
1502 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
1503 default now (the kernel default is 16). This is beneficial
1504 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
1505 allows substantially larger numbers of queued
1506 datagrams. This should increase the capability of systemd to
1507 parallelize boot-up, as logging and sd_notify() are unlikely
1508 to stall execution anymore. If you need to change the value
1509 from the new defaults, use the usual sysctl.d/ snippets.
1511 * The compression framing format used by the journal or
1512 coredump processing has changed to be in line with what the
1513 official LZ4 tools generate. LZ4 compression support in
1514 systemd was considered unsupported previously, as the format
1515 was not compatible with the normal tools. With this release
1516 this has changed now, and it is hence safe for downstream
1517 distributions to turn it on. While not compressing as well
1518 as the XZ, LZ4 is substantially faster, which makes
1519 it a good default choice for the compression logic in the
1520 journal and in coredump handling.
1522 * Any reference to /etc/mtab has been dropped from
1523 systemd. The file has been obsolete since a while, but
1524 systemd refused to work on systems where it was incorrectly
1525 set up (it should be a symlink or non-existent). Please make
1526 sure to update to util-linux 2.27.1 or newer in conjunction
1527 with this systemd release, which also drops any reference to
1528 /etc/mtab. If you maintain a distribution make sure that no
1529 software you package still references it, as this is a
1530 likely source of bugs. There's also a glibc bug pending,
1531 asking for removal of any reference to this obsolete file:
1533 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
1535 Note that only util-linux versions built with
1536 --enable-libmount-force-mountinfo are supported.
1538 * Support for the ".snapshot" unit type has been removed. This
1539 feature turned out to be little useful and little used, and
1540 has now been removed from the core and from systemctl.
1542 * The dependency types RequiresOverridable= and
1543 RequisiteOverridable= have been removed from systemd. They
1544 have been used only very sparingly to our knowledge and
1545 other options that provide a similar effect (such as
1546 systemctl --mode=ignore-dependencies) are much more useful
1547 and commonly used. Moreover, they were only half-way
1548 implemented as the option to control behaviour regarding
1549 these dependencies was never added to systemctl. By removing
1550 these dependency types the execution engine becomes a bit
1551 simpler. Unit files that use these dependencies should be
1552 changed to use the non-Overridable dependency types
1553 instead. In fact, when parsing unit files with these
1554 options, that's what systemd will automatically convert them
1555 too, but it will also warn, asking users to fix the unit
1556 files accordingly. Removal of these dependency types should
1557 only affect a negligible number of unit files in the wild.
1559 * Behaviour of networkd's IPForward= option changed
1560 (again). It will no longer maintain a per-interface setting,
1561 but propagate one way from interfaces where this is enabled
1562 to the global kernel setting. The global setting will be
1563 enabled when requested by a network that is set up, but
1564 never be disabled again. This change was made to make sure
1565 IPv4 and IPv6 behaviour regarding packet forwarding is
1566 similar (as the Linux IPv6 stack does not support
1567 per-interface control of this setting) and to minimize
1570 * In unit files the behaviour of %u, %U, %h, %s has
1571 changed. These specifiers will now unconditionally resolve
1572 to the various user database fields of the user that the
1573 systemd instance is running as, instead of the user
1574 configured in the specific unit via User=. Note that this
1575 effectively doesn't change much, as resolving of these
1576 specifiers was already turned off in the --system instance
1577 of systemd, as we cannot do NSS lookups from PID 1. In the
1578 --user instance of systemd these specifiers where correctly
1579 resolved, but hardly made any sense, since the user instance
1580 lacks privileges to do user switches anyway, and User= is
1581 hence useless. Moreover, even in the --user instance of
1582 systemd behaviour was awkward as it would only take settings
1583 from User= assignment placed before the specifier into
1584 account. In order to unify and simplify the logic around
1585 this the specifiers will now always resolve to the
1586 credentials of the user invoking the manager (which in case
1587 of PID 1 is the root user).
1589 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
1590 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
1591 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
1592 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
1593 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
1594 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
1595 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
1596 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
1597 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
1598 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
1599 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
1600 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
1601 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
1602 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
1605 — Berlin, 2015-11-18
1609 * systemd now depends on util-linux v2.27. More specifically,
1610 the newly added mount monitor feature in libmount now
1611 replaces systemd's former own implementation.
1613 * libmount mandates /etc/mtab not to be regular file, and
1614 systemd now enforces this condition at early boot.
1615 /etc/mtab has been deprecated and warned about for a very
1616 long time, so systems running systemd should already have
1617 stopped having this file around as anything else than a
1618 symlink to /proc/self/mounts.
1620 * Support for the "pids" cgroup controller has been added. It
1621 allows accounting the number of tasks in a cgroup and
1622 enforcing limits on it. This adds two new setting
1623 TasksAccounting= and TasksMax= to each unit, as well as a
1624 global option DefaultTasksAccounting=.
1626 * Support for the "net_cls" cgroup controller has been added.
1627 It allows assigning a net class ID to each task in the
1628 cgroup, which can then be used in firewall rules and traffic
1629 shaping configurations. Note that the kernel netfilter net
1630 class code does not currently work reliably for ingress
1631 packets on unestablished sockets.
1633 This adds a new config directive called NetClass= to CGroup
1634 enabled units. Allowed values are positive numbers for fixed
1635 assignments and "auto" for picking a free value
1638 * 'systemctl is-system-running' now returns 'offline' if the
1639 system is not booted with systemd. This command can now be
1640 used as a substitute for 'systemd-notify --booted'.
1642 * Watchdog timeouts have been increased to 3 minutes for all
1643 in-tree service files. Apparently, disk IO issues are more
1644 frequent than we hoped, and user reported >1 minute waiting
1647 * 'machine-id-commit' functionality has been merged into
1648 'machine-id-setup --commit'. The separate binary has been
1651 * The WorkingDirectory= directive in unit files may now be set
1652 to the special value '~'. In this case, the working
1653 directory is set to the home directory of the user
1654 configured in User=.
1656 * "machinectl shell" will now open the shell in the home
1657 directory of the selected user by default.
1659 * The CrashChVT= configuration file setting is renamed to
1660 CrashChangeVT=, following our usual logic of not
1661 abbreviating unnecessarily. The old directive is still
1662 supported for compat reasons. Also, this directive now takes
1663 an integer value between 1 and 63, or a boolean value. The
1664 formerly supported '-1' value for disabling stays around for
1667 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
1668 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
1669 RootDirectory= properties can now be set for transient
1672 * The systemd-analyze tool gained a new "set-log-target" verb
1673 to change the logging target the system manager logs to
1674 dynamically during runtime. This is similar to how
1675 "systemd-analyze set-log-level" already changes the log
1678 * In nspawn /sys is now mounted as tmpfs, with only a selected
1679 set of subdirectories mounted in from the real sysfs. This
1680 enhances security slightly, and is useful for ensuring user
1681 namespaces work correctly.
1683 * Support for USB FunctionFS activation has been added. This
1684 allows implementation of USB gadget services that are
1685 activated as soon as they are requested, so that they don't
1686 have to run continuously, similar to classic socket
1689 * The "systemctl exit" command now optionally takes an
1690 additional parameter that sets the exit code to return from
1691 the systemd manager when exiting. This is only relevant when
1692 running the systemd user instance, or when running the
1693 system instance in a container.
1695 * sd-bus gained the new API calls sd_bus_path_encode_many()
1696 and sd_bus_path_decode_many() that allow easy encoding and
1697 decoding of multiple identifier strings inside a D-Bus
1698 object path. Another new call sd_bus_default_flush_close()
1699 has been added to flush and close per-thread default
1702 * systemd-cgtop gained support for a -M/--machine= switch to
1703 show the control groups within a certain container only.
1705 * "systemctl kill" gained support for an optional --fail
1706 switch. If specified the requested operation will fail of no
1707 processes have been killed, because the unit had no
1708 processes attached, or similar.
1710 * A new systemd.crash_reboot=1 kernel command line option has
1711 been added that triggers a reboot after crashing. This can
1712 also be set through CrashReboot= in systemd.conf.
1714 * The RuntimeDirectory= setting now understands unit
1715 specifiers like %i or %f.
1717 * A new (still internal) library API sd-ipv4acd has been added,
1718 that implements address conflict detection for IPv4. It's
1719 based on code from sd-ipv4ll, and will be useful for
1720 detecting DHCP address conflicts.
1722 * File descriptors passed during socket activation may now be
1723 named. A new API sd_listen_fds_with_names() is added to
1724 access the names. The default names may be overridden,
1725 either in the .socket file using the FileDescriptorName=
1726 parameter, or by passing FDNAME= when storing the file
1727 descriptors using sd_notify().
1729 * systemd-networkd gained support for:
1731 - Setting the IPv6 Router Advertisement settings via
1732 IPv6AcceptRouterAdvertisements= in .network files.
1734 - Configuring the HelloTimeSec=, MaxAgeSec= and
1735 ForwardDelaySec= bridge parameters in .netdev files.
1737 - Configuring PreferredSource= for static routes in
1740 * The "ask-password" framework used to query for LUKS harddisk
1741 passwords or SSL passwords during boot gained support for
1742 caching passwords in the kernel keyring, if it is
1743 available. This makes sure that the user only has to type in
1744 a passphrase once if there are multiple objects to unlock
1745 with the same one. Previously, such password caching was
1746 available only when Plymouth was used; this moves the
1747 caching logic into the systemd codebase itself. The
1748 "systemd-ask-password" utility gained a new --keyname=
1749 switch to control which kernel keyring key to use for
1750 caching a password in. This functionality is also useful for
1751 enabling display managers such as gdm to automatically
1752 unlock the user's GNOME keyring if its passphrase, the
1753 user's password and the harddisk password are the same, if
1754 gdm-autologin is used.
1756 * When downloading tar or raw images using "machinectl
1757 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
1758 file is now also downloaded, if it is available and stored
1759 next to the image file.
1761 * Units of type ".socket" gained a new boolean setting
1762 Writable= which is only useful in conjunction with
1763 ListenSpecial=. If true, enables opening the specified
1764 special file in O_RDWR mode rather than O_RDONLY mode.
1766 * systemd-rfkill has been reworked to become a singleton
1767 service that is activated through /dev/rfkill on each rfkill
1768 state change and saves the settings to disk. This way,
1769 systemd-rfkill is now compatible with devices that exist
1770 only intermittendly, and even restores state if the previous
1771 system shutdown was abrupt rather than clean.
1773 * The journal daemon gained support for vacuuming old journal
1774 files controlled by the number of files that shall remain,
1775 in addition to the already existing control by size and by
1776 date. This is useful as journal interleaving performance
1777 degrades with too many separate journal files, and allows
1778 putting an effective limit on them. The new setting defaults
1779 to 100, but this may be changed by setting SystemMaxFiles=
1780 and RuntimeMaxFiles= in journald.conf. Also, the
1781 "journalctl" tool gained the new --vacuum-files= switch to
1782 manually vacuum journal files to leave only the specified
1783 number of files in place.
1785 * udev will now create /dev/disk/by-path links for ATA devices
1786 on kernels where that is supported.
1788 * Galician, Serbian, Turkish and Korean translations were added.
1790 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
1791 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
1792 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
1793 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
1794 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
1795 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
1796 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
1797 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
1798 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
1799 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
1800 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
1801 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
1802 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
1803 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
1804 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
1805 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
1806 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
1807 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
1809 — Berlin, 2015-10-07
1813 * The DHCP implementation of systemd-networkd gained a set of
1816 - The DHCP server now supports emitting DNS and NTP
1817 information. It may be enabled and configured via
1818 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
1819 and NTP information is enabled, but no servers are
1820 configured, the corresponding uplink information (if there
1821 is any) is propagated.
1823 - Server and client now support transmission and reception
1824 of timezone information. It can be configured via the
1825 newly introduced network options UseTimezone=,
1826 EmitTimezone=, and Timezone=. Transmission of timezone
1827 information is enabled between host and containers by
1828 default now: the container will change its local timezone
1829 to what the host has set.
1831 - Lease timeouts can now be configured via
1832 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
1834 - The DHCP server improved on the stability of
1835 leases. Clients are more likely to get the same lease
1836 information back, even if the server loses state.
1838 - The DHCP server supports two new configuration options to
1839 control the lease address pool metrics, PoolOffset= and
1842 * The encapsulation limit of tunnels in systemd-networkd may
1843 now be configured via 'EncapsulationLimit='. It allows
1844 modifying the maximum additional levels of encapsulation
1845 that are permitted to be prepended to a packet.
1847 * systemd now supports the concept of user buses replacing
1848 session buses, if used with dbus-1.10 (and enabled via dbus
1849 --enable-user-session). It previously only supported this on
1850 kdbus-enabled systems, and this release expands this to
1851 'dbus-daemon' systems.
1853 * systemd-networkd now supports predictable interface names
1856 * systemd now optionally supports the new Linux kernel
1857 "unified" control group hierarchy. If enabled via the kernel
1858 command-line option 'systemd.unified_cgroup_hierarchy=1',
1859 systemd will try to mount the unified cgroup hierarchy
1860 directly on /sys/fs/cgroup. If not enabled, or not
1861 available, systemd will fall back to the legacy cgroup
1862 hierarchy setup, as before. Host system and containers can
1863 mix and match legacy and unified hierarchies as they
1864 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
1865 environment variable to individually select the hierarchy to
1866 use for executed containers. By default, nspawn will use the
1867 unified hierarchy for the containers if the host uses the
1868 unified hierarchy, and the legacy hierarchy otherwise.
1869 Please note that at this point the unified hierarchy is an
1870 experimental kernel feature and is likely to change in one
1871 of the next kernel releases. Therefore, it should not be
1872 enabled by default in downstream distributions yet. The
1873 minimum required kernel version for the unified hierarchy to
1874 work is 4.2. Note that when the unified hierarchy is used
1875 for the first time delegated access to controllers is
1876 safe. Because of this systemd-nspawn containers will get
1877 access to controllers now, as will systemd user
1878 sessions. This means containers and user sessions may now
1879 manage their own resources, partitioning up what the system
1882 * A new special scope unit "init.scope" has been introduced
1883 that encapsulates PID 1 of the system. It may be used to
1884 determine resource usage and enforce resource limits on PID
1885 1 itself. PID 1 hence moved out of the root of the control
1888 * The cgtop tool gained support for filtering out kernel
1889 threads when counting tasks in a control group. Also, the
1890 count of processes is now recursively summed up by
1891 default. Two options -k and --recursive= have been added to
1892 revert to old behaviour. The tool has also been updated to
1893 work correctly in containers now.
1895 * systemd-nspawn's --bind= and --bind-ro= options have been
1896 extended to allow creation of non-recursive bind mounts.
1898 * libsystemd gained two new calls sd_pid_get_cgroup() and
1899 sd_peer_get_cgroup() which return the control group path of
1900 a process or peer of a connected AF_UNIX socket. This
1901 function call is particularly useful when implementing
1902 delegated subtrees support in the control group hierarchy.
1904 * The "sd-event" event loop API of libsystemd now supports
1905 correct dequeuing of real-time signals, without losing
1908 * When systemd requests a PolicyKit decision when managing
1909 units it will now add additional fields to the request,
1910 including unit name and desired operation. This enables more
1911 powerful PolicyKit policies, that make decisions depending
1912 on these parameters.
1914 * nspawn learnt support for .nspawn settings files, that may
1915 accompany the image files or directories of containers, and
1916 may contain additional settings for the container. This is
1917 an alternative to configuring container parameters via the
1918 nspawn command line.
1920 Contributions from: Cristian Rodríguez, Daniel Mack, David
1921 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
1922 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
1923 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
1924 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
1925 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
1926 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
1927 Andersen, Tom Gundersen, Torstein Husebø
1929 — Berlin, 2015-09-08
1933 * machinectl gained a new verb 'shell' which opens a fresh
1934 shell on the target container or the host. It is similar to
1935 the existing 'login' command of machinectl, but spawns the
1936 shell directly without prompting for username or
1937 password. The pseudo machine '.host' now refers to the local
1938 host and is used by default. Hence, 'machinectl shell' can
1939 be used as replacement for 'su -' which spawns a session as
1940 a fresh systemd unit in a way that is fully isolated from
1941 the originating session.
1943 * systemd-networkd learned to cope with private-zone DHCP
1944 options and allows other programs to query the values.
1946 * SELinux access control when enabling/disabling units is no
1947 longer enforced with this release. The previous
1948 implementation was incorrect, and a new corrected
1949 implementation is not yet available. As unit file operations
1950 are still protected via PolicyKit and D-Bus policy this is
1951 not a security problem. Yet, distributions which care about
1952 optimal SELinux support should probably not stabilize on
1955 * sd-bus gained support for matches of type "arg0has=", that
1956 test for membership of strings in string arrays sent in bus
1959 * systemd-resolved now dumps the contents of its DNS and LLMNR
1960 caches to the logs on reception of the SIGUSR1 signal. This
1961 is useful to debug DNS behaviour.
1963 * The coredumpctl tool gained a new --directory= option to
1964 operate on journal files in a specific directory.
1966 * "systemctl reboot" and related commands gained a new
1967 "--message=" option which may be used to set a free-text
1968 wall message when shutting down or rebooting the
1969 system. This message is also logged, which is useful for
1970 figuring out the reason for a reboot or shutdown a
1973 * The "systemd-resolve-host" tool's -i switch now takes
1974 network interface numbers as alternative to interface names.
1976 * A new unit file setting for services has been introduced:
1977 UtmpMode= allows configuration of how precisely systemd
1978 handles utmp and wtmp entries for the service if this is
1979 enabled. This allows writing services that appear similar to
1980 user sessions in the output of the "w", "who", "last" and
1983 * systemd-resolved will now locally synthesize DNS resource
1984 records for the "localhost" and "gateway" domains as well as
1985 the local hostname. This should ensure that clients querying
1986 RRs via resolved will get similar results as those going via
1987 NSS, if nss-myhostname is enabled.
1989 Contributions from: Alastair Hughes, Alex Crawford, Daniel
1990 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
1991 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
1992 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
1993 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
1994 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
1995 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
1996 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
1997 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
1998 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
1999 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
2000 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2002 — Berlin, 2015-08-27
2006 * The systemd-efi-boot-generator functionality was merged into
2007 systemd-gpt-auto-generator.
2009 * systemd-networkd now supports Group Policy for vxlan
2010 devices. It can be enabled via the new boolean configuration
2011 option called 'GroupPolicyExtension='.
2013 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
2014 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
2015 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
2017 — Berlin, 2015-07-31
2021 * The python-systemd code has been removed from the systemd repository.
2022 A new repository has been created which accommodates the code from
2023 now on, and we kindly ask distributions to create a separate package
2024 for this: https://github.com/systemd/python-systemd
2026 * The systemd daemon will now reload its main configuration
2027 (/etc/systemd/system.conf) on daemon-reload.
2029 * sd-dhcp now exposes vendor specific extensions via
2030 sd_dhcp_lease_get_vendor_specific().
2032 * systemd-networkd gained a number of new configuration options.
2034 - A new boolean configuration option for TAP devices called
2035 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
2036 device, thus allowing to send and receive GSO packets.
2038 - A new tunnel configuration option called 'CopyDSCP='.
2039 If enabled, the DSCP field of ip6 tunnels is copied into the
2040 decapsulated packet.
2042 - A set of boolean bridge configuration options were added.
2043 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
2044 and 'UnicastFlood=' are now parsed by networkd and applied to the
2045 respective bridge link device via the respective IFLA_BRPORT_*
2048 - A new string configuration option to override the hostname sent
2049 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
2050 is true, networkd will use the configured hostname instead of the
2051 system hostname when sending DHCP requests.
2053 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
2054 networkd will configure the IPv6 flow-label of the tunnel device
2055 according to RFC2460.
2057 - The 'macvtap' virtual network devices are now supported, similar to
2058 the already supported 'macvlan' devices.
2060 * systemd-resolved now implements RFC5452 to improve resilience against
2061 cache poisoning. Additionally, source port randomization is enabled
2062 by default to further protect against DNS spoofing attacks.
2064 * nss-mymachines now supports translating UIDs and GIDs of running
2065 containers with user-namespaces enabled. If a container 'foo'
2066 translates a host uid 'UID' to the container uid 'TUID', then
2067 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
2068 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
2069 mapped as 'vg-foo-TGID'.
2071 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
2072 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
2073 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
2074 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
2075 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
2076 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
2077 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
2078 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
2079 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
2080 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
2082 — Berlin, 2015-07-29
2086 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
2087 There are no known issues with current sysfs, and udev does not need
2088 or should be used to work around such bugs.
2090 * udev does no longer enable USB HID power management. Several reports
2091 indicate, that some devices cannot handle that setting.
2093 * The udev accelerometer helper was removed. The functionality
2094 is now fully included in iio-sensor-proxy. But this means,
2095 older iio-sensor-proxy versions will no longer provide
2096 accelerometer/orientation data with this systemd version.
2097 Please upgrade iio-sensor-proxy to version 1.0.
2099 * networkd gained a new configuration option IPv6PrivacyExtensions=
2100 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
2101 for Stateless Address") on selected networks.
2103 * For the sake of fewer build-time dependencies and less code in the
2104 main repository, the python bindings are about to be removed in the
2105 next release. A new repository has been created which accommodates
2106 the code from now on, and we kindly ask distributions to create a
2107 separate package for this. The removal will take place in v223.
2109 https://github.com/systemd/python-systemd
2111 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
2112 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
2113 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
2114 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
2115 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
2116 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
2117 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
2118 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
2119 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
2120 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2122 — Berlin, 2015-07-07
2126 * The sd-bus.h and sd-event.h APIs have now been declared
2127 stable and have been added to the official interface of
2128 libsystemd.so. sd-bus implements an alternative D-Bus client
2129 library, that is relatively easy to use, very efficient and
2130 supports both classic D-Bus as well as kdbus as transport
2131 backend. sd-event is a generic event loop abstraction that
2132 is built around Linux epoll, but adds features such as event
2133 prioritization or efficient timer handling. Both APIs are good
2134 choices for C programs looking for a bus and/or event loop
2135 implementation that is minimal and does not have to be
2136 portable to other kernels.
2138 * kdbus support is no longer compile-time optional. It is now
2139 always built-in. However, it can still be disabled at
2140 runtime using the kdbus=0 kernel command line setting, and
2141 that setting may be changed to default to off, by specifying
2142 --disable-kdbus at build-time. Note though that the kernel
2143 command line setting has no effect if the kdbus.ko kernel
2144 module is not installed, in which case kdbus is (obviously)
2145 also disabled. We encourage all downstream distributions to
2146 begin testing kdbus by adding it to the kernel images in the
2147 development distributions, and leaving kdbus support in
2150 * The minimal required util-linux version has been bumped to
2153 * Support for chkconfig (--enable-chkconfig) was removed in
2154 favor of calling an abstraction tool
2155 /lib/systemd/systemd-sysv-install. This needs to be
2156 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
2157 in README for details.
2159 * If there's a systemd unit and a SysV init script for the
2160 same service name, and the user executes "systemctl enable"
2161 for it (or a related call), then this will now enable both
2162 (or execute the related operation on both), not just the
2165 * The libudev API documentation has been converted from gtkdoc
2168 * gudev has been removed from the systemd tree, it is now an
2171 * The systemd-cgtop tool learnt a new --raw switch to generate
2172 "raw" (machine parsable) output.
2174 * networkd's IPForwarding= .network file setting learnt the
2175 new setting "kernel", which ensures that networkd does not
2176 change the IP forwarding sysctl from the default kernel
2179 * The systemd-logind bus API now exposes a new boolean
2180 property "Docked" that reports whether logind considers the
2181 system "docked", i.e. connected to a docking station or not.
2183 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
2184 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
2185 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
2186 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
2187 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
2188 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
2189 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
2190 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
2191 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
2192 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
2193 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
2194 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
2195 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
2196 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
2197 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
2198 Fink, Zbigniew Jędrzejewski-Szmek
2200 — Berlin, 2015-06-19
2204 * The gudev library has been extracted into a separate repository
2205 available at: https://git.gnome.org/browse/libgudev/
2206 It is now managed as part of the Gnome project. Distributions
2207 are recommended to pass --disable-gudev to systemd and use
2208 gudev from the Gnome project instead. gudev is still included
2209 in systemd, for now. It will be removed soon, though. Please
2210 also see the announcement-thread on systemd-devel:
2211 https://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
2213 * systemd now exposes a CPUUsageNSec= property for each
2214 service unit on the bus, that contains the overall consumed
2215 CPU time of a service (the sum of what each process of the
2216 service consumed). This value is only available if
2217 CPUAccounting= is turned on for a service, and is then shown
2218 in the "systemctl status" output.
2220 * Support for configuring alternative mappings of the old SysV
2221 runlevels to systemd targets has been removed. They are now
2222 hardcoded in a way that runlevels 2, 3, 4 all map to
2223 multi-user.target and 5 to graphical.target (which
2224 previously was already the default behaviour).
2226 * The auto-mounter logic gained support for mount point
2227 expiry, using a new TimeoutIdleSec= setting in .automount
2228 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
2230 * The EFI System Partition (ESP) as mounted to /boot by
2231 systemd-efi-boot-generator will now be unmounted
2232 automatically after 2 minutes of not being used. This should
2233 minimize the risk of ESP corruptions.
2235 * New /etc/fstab options x-systemd.requires= and
2236 x-systemd.requires-mounts-for= are now supported to express
2237 additional dependencies for mounts. This is useful for
2238 journalling file systems that support external journal
2239 devices or overlay file systems that require underlying file
2240 systems to be mounted.
2242 * systemd does not support direct live-upgrades (via systemctl
2243 daemon-reexec) from versions older than v44 anymore. As no
2244 distribution we are aware of shipped such old versions in a
2245 stable release this should not be problematic.
2247 * When systemd forks off a new per-connection service instance
2248 it will now set the $REMOTE_ADDR environment variable to the
2249 remote IP address, and $REMOTE_PORT environment variable to
2250 the remote IP port. This behaviour is similar to the
2251 corresponding environment variables defined by CGI.
2253 * systemd-networkd gained support for uplink failure
2254 detection. The BindCarrier= option allows binding interface
2255 configuration dynamically to the link sense of other
2256 interfaces. This is useful to achieve behaviour like in
2259 * systemd-networkd gained support for configuring the DHCP
2260 client identifier to use when requesting leases.
2262 * systemd-networkd now has a per-network UseNTP= option to
2263 configure whether NTP server information acquired via DHCP
2264 is passed on to services like systemd-timesyncd.
2266 * systemd-networkd gained support for vti6 tunnels.
2268 * Note that systemd-networkd manages the sysctl variable
2269 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
2270 it is configured for since v219. The variable controls IP
2271 forwarding, and is a per-interface alternative to the global
2272 /proc/sys/net/ipv[46]/ip_forward. This setting is
2273 configurable in the IPForward= option, which defaults to
2274 "no". This means if networkd is used for an interface it is
2275 no longer sufficient to set the global sysctl option to turn
2276 on IP forwarding! Instead, the .network file option
2277 IPForward= needs to be turned on! Note that the
2278 implementation of this behaviour was broken in v219 and has
2281 * Many bonding and vxlan options are now configurable in
2284 * systemd-nspawn gained a new --property= setting to set unit
2285 properties for the container scope. This is useful for
2286 setting resource parameters (e.g. "CPUShares=500") on
2287 containers started from the command line.
2289 * systemd-nspawn gained a new --private-users= switch to make
2290 use of user namespacing available on recent Linux kernels.
2292 * systemd-nspawn may now be called as part of a shell pipeline
2293 in which case the pipes used for stdin and stdout are passed
2294 directly to the process invoked in the container, without
2295 indirection via a pseudo tty.
2297 * systemd-nspawn gained a new switch to control the UNIX
2298 signal to use when killing the init process of the container
2301 * systemd-nspawn gained a new --overlay= switch for mounting
2302 overlay file systems into the container using the new kernel
2305 * When a container image is imported via systemd-importd and
2306 the host file system is not btrfs, a loopback block device
2307 file is created in /var/lib/machines.raw with a btrfs file
2308 system inside. It is then mounted to /var/lib/machines to
2309 enable btrfs features for container management. The loopback
2310 file and btrfs file system is grown as needed when container
2311 images are imported via systemd-importd.
2313 * systemd-machined/systemd-importd gained support for btrfs
2314 quota, to enforce container disk space limits on disk. This
2315 is exposed in "machinectl set-limit".
2317 * systemd-importd now can import containers from local .tar,
2318 .raw and .qcow2 images, and export them to .tar and .raw. It
2319 can also import dkr v2 images now from the network (on top
2322 * systemd-importd gained support for verifying downloaded
2323 images with gpg2 (previously only gpg1 was supported).
2325 * systemd-machined, systemd-logind, systemd: most bus calls
2326 are now accessible to unprivileged processes via
2327 PolicyKit. Also, systemd-logind will now allow users to kill
2328 their own sessions without further privileges or
2331 * systemd-shutdownd has been removed. This service was
2332 previously responsible for implementing scheduled shutdowns
2333 as exposed in /usr/bin/shutdown's time parameter. This
2334 functionality has now been moved into systemd-logind and is
2335 accessible via a bus interface.
2337 * "systemctl reboot" gained a new switch --firmware-setup that
2338 can be used to reboot into the EFI firmware setup, if that
2339 is available. systemd-logind now exposes an API on the bus
2340 to trigger such reboots, in case graphical desktop UIs want
2341 to cover this functionality.
2343 * "systemctl enable", "systemctl disable" and "systemctl mask"
2344 now support a new "--now" switch. If specified the units
2345 that are enabled will also be started, and the ones
2346 disabled/masked also stopped.
2348 * The Gummiboot EFI boot loader tool has been merged into
2349 systemd, and renamed to "systemd-boot". The bootctl tool has been
2350 updated to support systemd-boot.
2352 * An EFI kernel stub has been added that may be used to create
2353 kernel EFI binaries that contain not only the actual kernel,
2354 but also an initrd, boot splash, command line and OS release
2355 information. This combined binary can then be signed as a
2356 single image, so that the firmware can verify it all in one
2357 step. systemd-boot has special support for EFI binaries created
2358 like this and can extract OS release information from them
2359 and show them in the boot menu. This functionality is useful
2360 to implement cryptographically verified boot schemes.
2362 * Optional support has been added to systemd-fsck to pass
2363 fsck's progress report to an AF_UNIX socket in the file
2366 * udev will no longer create device symlinks for all block
2367 devices by default. A blacklist for excluding special block
2368 devices from this logic has been turned into a whitelist
2369 that requires picking block devices explicitly that require
2372 * A new (currently still internal) API sd-device.h has been
2373 added to libsystemd. This modernized API is supposed to
2374 replace libudev eventually. In fact, already much of libudev
2375 is now just a wrapper around sd-device.h.
2377 * A new hwdb database for storing metadata about pointing
2378 stick devices has been added.
2380 * systemd-tmpfiles gained support for setting file attributes
2381 similar to the "chattr" tool with new 'h' and 'H' lines.
2383 * systemd-journald will no longer unconditionally set the
2384 btrfs NOCOW flag on new journal files. This is instead done
2385 with tmpfiles snippet using the new 'h' line type. This
2386 allows easy disabling of this logic, by masking the
2387 journal-nocow.conf tmpfiles file.
2389 * systemd-journald will now translate audit message types to
2390 human readable identifiers when writing them to the
2391 journal. This should improve readability of audit messages.
2393 * The LUKS logic gained support for the offset= and skip=
2394 options in /etc/crypttab, as previously implemented by
2397 * /usr/lib/os-release gained a new optional field VARIANT= for
2398 distributions that support multiple variants (such as a
2399 desktop edition, a server edition, ...)
2401 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
2402 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
2403 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
2404 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
2405 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
2406 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
2407 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
2408 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
2409 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
2410 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
2411 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
2412 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
2413 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
2414 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
2415 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
2416 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
2417 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
2418 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
2419 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
2420 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
2421 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
2422 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
2423 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
2424 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
2425 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
2426 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
2427 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
2429 — Berlin, 2015-05-22
2433 * Introduce a new API "sd-hwdb.h" for querying the hardware
2434 metadata database. With this minimal interface one can query
2435 and enumerate the udev hwdb, decoupled from the old libudev
2436 library. libudev's interface for this is now only a wrapper
2437 around sd-hwdb. A new tool systemd-hwdb has been added to
2438 interface with and update the database.
2440 * When any of systemd's tools copies files (for example due to
2441 tmpfiles' C lines) a btrfs reflink will attempted first,
2442 before bytewise copying is done.
2444 * systemd-nspawn gained a new --ephemeral switch. When
2445 specified a btrfs snapshot is taken of the container's root
2446 directory, and immediately removed when the container
2447 terminates again. Thus, a container can be started whose
2448 changes never alter the container's root directory, and are
2449 lost on container termination. This switch can also be used
2450 for starting a container off the root file system of the
2451 host without affecting the host OS. This switch is only
2452 available on btrfs file systems.
2454 * systemd-nspawn gained a new --template= switch. It takes the
2455 path to a container tree to use as template for the tree
2456 specified via --directory=, should that directory be
2457 missing. This allows instantiating containers dynamically,
2458 on first run. This switch is only available on btrfs file
2461 * When a .mount unit refers to a mount point on which multiple
2462 mounts are stacked, and the .mount unit is stopped all of
2463 the stacked mount points will now be unmounted until no
2464 mount point remains.
2466 * systemd now has an explicit notion of supported and
2467 unsupported unit types. Jobs enqueued for unsupported unit
2468 types will now fail with an "unsupported" error code. More
2469 specifically .swap, .automount and .device units are not
2470 supported in containers, .busname units are not supported on
2471 non-kdbus systems. .swap and .automount are also not
2472 supported if their respective kernel compile time options
2475 * machinectl gained support for two new "copy-from" and
2476 "copy-to" commands for copying files from a running
2477 container to the host or vice versa.
2479 * machinectl gained support for a new "bind" command to bind
2480 mount host directories into local containers. This is
2481 currently only supported for nspawn containers.
2483 * networkd gained support for configuring bridge forwarding
2484 database entries (fdb) from .network files.
2486 * A new tiny daemon "systemd-importd" has been added that can
2487 download container images in tar, raw, qcow2 or dkr formats,
2488 and make them available locally in /var/lib/machines, so
2489 that they can run as nspawn containers. The daemon can GPG
2490 verify the downloads (not supported for dkr, since it has no
2491 provisions for verifying downloads). It will transparently
2492 decompress bz2, xz, gzip compressed downloads if necessary,
2493 and restore sparse files on disk. The daemon uses privilege
2494 separation to ensure the actual download logic runs with
2495 fewer privileges than the daemon itself. machinectl has
2496 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
2497 make the functionality of importd available to the
2498 user. With this in place the Fedora and Ubuntu "Cloud"
2499 images can be downloaded and booted as containers unmodified
2500 (the Fedora images lack the appropriate GPG signature files
2501 currently, so they cannot be verified, but this will change
2502 soon, hopefully). Note that downloading images is currently
2503 only fully supported on btrfs.
2505 * machinectl is now able to list container images found in
2506 /var/lib/machines, along with some metadata about sizes of
2507 disk and similar. If the directory is located on btrfs and
2508 quota is enabled, this includes quota display. A new command
2509 "image-status" has been added that shows additional
2510 information about images.
2512 * machinectl is now able to clone container images
2513 efficiently, if the underlying file system (btrfs) supports
2514 it, with the new "machinectl clone" command. It also
2515 gained commands for renaming and removing images, as well as
2516 marking them read-only or read-write (supported also on
2517 legacy file systems).
2519 * networkd gained support for collecting LLDP network
2520 announcements, from hardware that supports this. This is
2521 shown in networkctl output.
2523 * systemd-run gained support for a new -t (--pty) switch for
2524 invoking a binary on a pty whose input and output is
2525 connected to the invoking terminal. This allows executing
2526 processes as system services while interactively
2527 communicating with them via the terminal. Most interestingly
2528 this is supported across container boundaries. Invoking
2529 "systemd-run -t /bin/bash" is an alternative to running a
2530 full login session, the difference being that the former
2531 will not register a session, nor go through the PAM session
2534 * tmpfiles gained support for a new "v" line type for creating
2535 btrfs subvolumes. If the underlying file system is a legacy
2536 file system, this automatically degrades to creating a
2537 normal directory. Among others /var/lib/machines is now
2538 created like this at boot, should it be missing.
2540 * The directory /var/lib/containers/ has been deprecated and
2541 been replaced by /var/lib/machines. The term "machines" has
2542 been used in the systemd context as generic term for both
2543 VMs and containers, and hence appears more appropriate for
2544 this, as the directory can also contain raw images bootable
2547 * systemd-nspawn when invoked with -M but without --directory=
2548 or --image= is now capable of searching for the container
2549 root directory, subvolume or disk image automatically, in
2550 /var/lib/machines. systemd-nspawn@.service has been updated
2551 to make use of this, thus allowing it to be used for raw
2554 * A new machines.target unit has been introduced that is
2555 supposed to group all containers/VMs invoked as services on
2556 the system. systemd-nspawn@.service has been updated to
2557 integrate with that.
2559 * machinectl gained a new "start" command, for invoking a
2560 container as a service. "machinectl start foo" is mostly
2561 equivalent to "systemctl start systemd-nspawn@foo.service",
2562 but handles escaping in a nicer way.
2564 * systemd-nspawn will now mount most of the cgroupfs tree
2565 read-only into each container, with the exception of the
2566 container's own subtree in the name=systemd hierarchy.
2568 * journald now sets the special FS_NOCOW file flag for its
2569 journal files. This should improve performance on btrfs, by
2570 avoiding heavy fragmentation when journald's write-pattern
2571 is used on COW file systems. It degrades btrfs' data
2572 integrity guarantees for the files to the same levels as for
2573 ext3/ext4 however. This should be OK though as journald does
2574 its own data integrity checks and all its objects are
2575 checksummed on disk. Also, journald should handle btrfs disk
2576 full events a lot more gracefully now, by processing SIGBUS
2577 errors, and not relying on fallocate() anymore.
2579 * When journald detects that journal files it is writing to
2580 have been deleted it will immediately start new journal
2583 * systemd now provides a way to store file descriptors
2584 per-service in PID 1. This is useful for daemons to ensure
2585 that fds they require are not lost during a daemon
2586 restart. The fds are passed to the daemon on the next
2587 invocation in the same way socket activation fds are
2588 passed. This is now used by journald to ensure that the
2589 various sockets connected to all the system's stdout/stderr
2590 are not lost when journald is restarted. File descriptors
2591 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
2592 an extension to sd_notify(). Note that a limit is enforced
2593 on the number of fds a service can store in PID 1, and it
2594 defaults to 0, so that no fds may be stored, unless this is
2595 explicitly turned on.
2597 * The default TERM variable to use for units connected to a
2598 terminal, when no other value is explicitly is set is now
2599 vt220 rather than vt102. This should be fairly safe still,
2600 but allows PgUp/PgDn work.
2602 * The /etc/crypttab option header= as known from Debian is now
2605 * "loginctl user-status" and "loginctl session-status" will
2606 now show the last 10 lines of log messages of the
2607 user/session following the status output. Similar,
2608 "machinectl status" will show the last 10 log lines
2609 associated with a virtual machine or container
2610 service. (Note that this is usually not the log messages
2611 done in the VM/container itself, but simply what the
2612 container manager logs. For nspawn this includes all console
2615 * "loginctl session-status" without further argument will now
2616 show the status of the session of the caller. Similar,
2617 "lock-session", "unlock-session", "activate",
2618 "enable-linger", "disable-linger" may now be called without
2619 session/user parameter in which case they apply to the
2620 caller's session/user.
2622 * An X11 session scriptlet is now shipped that uploads
2623 $DISPLAY and $XAUTHORITY into the environment of the systemd
2624 --user daemon if a session begins. This should improve
2625 compatibility with X11 enabled applications run as systemd
2628 * Generators are now subject to masking via /etc and /run, the
2629 same way as unit files.
2631 * networkd .network files gained support for configuring
2632 per-link IPv4/IPv6 packet forwarding as well as IPv4
2633 masquerading. This is by default turned on for veth links to
2634 containers, as registered by systemd-nspawn. This means that
2635 nspawn containers run with --network-veth will now get
2636 automatic routed access to the host's networks without any
2637 further configuration or setup, as long as networkd runs on
2640 * systemd-nspawn gained the --port= (-p) switch to expose TCP
2641 or UDP posts of a container on the host. With this in place
2642 it is possible to run containers with private veth links
2643 (--network-veth), and have their functionality exposed on
2644 the host as if their services were running directly on the
2647 * systemd-nspawn's --network-veth switch now gained a short
2648 version "-n", since with the changes above it is now truly
2649 useful out-of-the-box. The systemd-nspawn@.service has been
2650 updated to make use of it too by default.
2652 * systemd-nspawn will now maintain a per-image R/W lock, to
2653 ensure that the same image is not started more than once
2654 writable. (It's OK to run an image multiple times
2655 simultaneously in read-only mode.)
2657 * systemd-nspawn's --image= option is now capable of
2658 dissecting and booting MBR and GPT disk images that contain
2659 only a single active Linux partition. Previously it
2660 supported only GPT disk images with proper GPT type
2661 IDs. This allows running cloud images from major
2662 distributions directly with systemd-nspawn, without
2665 * In addition to collecting mouse dpi data in the udev
2666 hardware database, there's now support for collecting angle
2667 information for mouse scroll wheels. The database is
2668 supposed to guarantee similar scrolling behavior on mice
2669 that it knows about. There's also support for collecting
2670 information about Touchpad types.
2672 * udev's input_id built-in will now also collect touch screen
2673 dimension data and attach it to probed devices.
2675 * /etc/os-release gained support for a Distribution Privacy
2678 * networkd gained support for creating "ipvlan", "gretap",
2679 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
2681 * systemd-tmpfiles gained support for "a" lines for setting
2684 * systemd-nspawn will now mount /tmp in the container to
2685 tmpfs, automatically.
2687 * systemd now exposes the memory.usage_in_bytes cgroup
2688 attribute and shows it for each service in the "systemctl
2689 status" output, if available.
2691 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
2692 immediate reboot is triggered. This useful if shutdown is
2693 hung and is unable to complete, to expedite the
2694 operation. Note that this kind of reboot will still unmount
2695 all file systems, and hence should not result in fsck being
2698 * A .device unit for an optical block device will now be
2699 considered active only when a medium is in the drive. Also,
2700 mount units are now bound to their backing devices thus
2701 triggering automatic unmounting when devices become
2702 unavailable. With this in place systemd will now
2703 automatically unmount left-over mounts when a CD-ROM is
2704 ejected or an USB stick is yanked from the system.
2706 * networkd-wait-online now has support for waiting for
2707 specific interfaces only (with globbing), and for giving up
2708 after a configurable timeout.
2710 * networkd now exits when idle. It will be automatically
2711 restarted as soon as interfaces show up, are removed or
2712 change state. networkd will stay around as long as there is
2713 at least one DHCP state machine or similar around, that keep
2716 * networkd may now configure IPv6 link-local addressing in
2717 addition to IPv4 link-local addressing.
2719 * The IPv6 "token" for use in SLAAC may now be configured for
2720 each .network interface in networkd.
2722 * Routes configured with networkd may now be assigned a scope
2725 * networkd's [Match] sections now support globbing and lists
2726 of multiple space-separated matches per item.
2728 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
2729 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
2730 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
2731 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
2732 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
2733 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
2734 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
2735 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
2736 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
2737 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
2738 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
2739 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
2740 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
2741 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
2742 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
2743 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
2744 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
2745 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
2746 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
2747 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
2748 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
2749 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
2750 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
2751 Hoffmann, Zbigniew Jędrzejewski-Szmek
2753 — Berlin, 2015-02-16
2757 * When querying unit file enablement status (for example via
2758 "systemctl is-enabled"), a new state "indirect" is now known
2759 which indicates that a unit might not be enabled itself, but
2760 another unit listed in its Also= setting might be.
2762 * Similar to the various existing ConditionXYZ= settings for
2763 units, there are now matching AssertXYZ= settings. While
2764 failing conditions cause a unit to be skipped, but its job
2765 to succeed, failing assertions declared like this will cause
2766 a unit start operation and its job to fail.
2768 * hostnamed now knows a new chassis type "embedded".
2770 * systemctl gained a new "edit" command. When used on a unit
2771 file, this allows extending unit files with .d/ drop-in
2772 configuration snippets or editing the full file (after
2773 copying it from /usr/lib to /etc). This will invoke the
2774 user's editor (as configured with $EDITOR), and reload the
2775 modified configuration after editing.
2777 * "systemctl status" now shows the suggested enablement state
2778 for a unit, as declared in the (usually vendor-supplied)
2779 system preset files.
2781 * nss-myhostname will now resolve the single-label host name
2782 "gateway" to the locally configured default IP routing
2783 gateways, ordered by their metrics. This assigns a stable
2784 name to the used gateways, regardless which ones are
2785 currently configured. Note that the name will only be
2786 resolved after all other name sources (if nss-myhostname is
2787 configured properly) and should hence not negatively impact
2788 systems that use the single-label host name "gateway" in
2791 * systemd-inhibit now allows filtering by mode when listing
2794 * Scope and service units gained a new "Delegate" boolean
2795 property, which, when set, allows processes running inside the
2796 unit to further partition resources. This is primarily
2797 useful for systemd user instances as well as container
2800 * journald will now pick up audit messages directly from
2801 the kernel, and log them like any other log message. The
2802 audit fields are split up and fully indexed. This means that
2803 journalctl in many ways is now a (nicer!) alternative to
2804 ausearch, the traditional audit client. Note that this
2805 implements only a minimal audit client. If you want the
2806 special audit modes like reboot-on-log-overflow, please use
2807 the traditional auditd instead, which can be used in
2808 parallel to journald.
2810 * The ConditionSecurity= unit file option now understands the
2811 special string "audit" to check whether auditing is
2814 * journalctl gained two new commands --vacuum-size= and
2815 --vacuum-time= to delete old journal files until the
2816 remaining ones take up no more than the specified size on disk,
2817 or are not older than the specified time.
2819 * A new, native PPPoE library has been added to sd-network,
2820 systemd's library of light-weight networking protocols. This
2821 library will be used in a future version of networkd to
2822 enable PPPoE communication without an external pppd daemon.
2824 * The busctl tool now understands a new "capture" verb that
2825 works similar to "monitor", but writes a packet capture
2826 trace to STDOUT that can be redirected to a file which is
2827 compatible with libcap's capture file format. This can then
2828 be loaded in Wireshark and similar tools to inspect bus
2831 * The busctl tool now understands a new "tree" verb that shows
2832 the object trees of a specific service on the bus, or of all
2835 * The busctl tool now understands a new "introspect" verb that
2836 shows all interfaces and members of objects on the bus,
2837 including their signature and values. This is particularly
2838 useful to get more information about bus objects shown by
2839 the new "busctl tree" command.
2841 * The busctl tool now understands new verbs "call",
2842 "set-property" and "get-property" for invoking bus method
2843 calls, setting and getting bus object properties in a
2846 * busctl gained a new --augment-creds= argument that controls
2847 whether the tool shall augment credential information it
2848 gets from the bus with data from /proc, in a possibly
2851 * nspawn's --link-journal= switch gained two new values
2852 "try-guest" and "try-host" that work like "guest" and
2853 "host", but do not fail if the host has no persistent
2854 journalling enabled. -j is now equivalent to
2855 --link-journal=try-guest.
2857 * macvlan network devices created by nspawn will now have
2858 stable MAC addresses.
2860 * A new SmackProcessLabel= unit setting has been added, which
2861 controls the SMACK security label processes forked off by
2862 the respective unit shall use.
2864 * If compiled with --enable-xkbcommon, systemd-localed will
2865 verify x11 keymap settings by compiling the given keymap. It
2866 will spew out warnings if the compilation fails. This
2867 requires libxkbcommon to be installed.
2869 * When a coredump is collected, a larger number of metadata
2870 fields is now collected and included in the journal records
2871 created for it. More specifically, control group membership,
2872 environment variables, memory maps, working directory,
2873 chroot directory, /proc/$PID/status, and a list of open file
2874 descriptors is now stored in the log entry.
2876 * The udev hwdb now contains DPI information for mice. For
2879 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
2881 * All systemd programs that read standalone configuration
2882 files in /etc now also support a corresponding series of
2883 .conf.d configuration directories in /etc/, /run/,
2884 /usr/local/lib/, /usr/lib/, and (if configured with
2885 --enable-split-usr) /lib/. In particular, the following
2886 configuration files now have corresponding configuration
2887 directories: system.conf user.conf, logind.conf,
2888 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
2889 resolved.conf, timesyncd.conf, journal-remote.conf, and
2890 journal-upload.conf. Note that distributions should use the
2891 configuration directories in /usr/lib/; the directories in
2892 /etc/ are reserved for the system administrator.
2894 * systemd-rfkill will no longer take the rfkill device name
2895 into account when storing rfkill state on disk, as the name
2896 might be dynamically assigned and not stable. Instead, the
2897 ID_PATH udev variable combined with the rfkill type (wlan,
2898 bluetooth, ...) is used.
2900 * A new service systemd-machine-id-commit.service has been
2901 added. When used on systems where /etc is read-only during
2902 boot, and /etc/machine-id is not initialized (but an empty
2903 file), this service will copy the temporary machine ID
2904 created as replacement into /etc after the system is fully
2905 booted up. This is useful for systems that are freshly
2906 installed with a non-initialized machine ID, but should get
2907 a fixed machine ID for subsequent boots.
2909 * networkd's .netdev files now provide a large set of
2910 configuration parameters for VXLAN devices. Similarly, the
2911 bridge port cost parameter is now configurable in .network
2912 files. There's also new support for configuring IP source
2913 routing. networkd .link files gained support for a new
2914 OriginalName= match that is useful to match against the
2915 original interface name the kernel assigned. .network files
2916 may include MTU= and MACAddress= fields for altering the MTU
2917 and MAC address while being connected to a specific network
2920 * The LUKS logic gained supported for configuring
2921 UUID-specific key files. There's also new support for naming
2922 LUKS device from the kernel command line, using the new
2923 luks.name= argument.
2925 * Timer units may now be transiently created via the bus API
2926 (this was previously already available for scope and service
2927 units). In addition it is now possible to create multiple
2928 transient units at the same time with a single bus call. The
2929 "systemd-run" tool has been updated to make use of this for
2930 running commands on a specified time, in at(1)-style.
2932 * tmpfiles gained support for "t" lines, for assigning
2933 extended attributes to files. Among other uses this may be
2934 used to assign SMACK labels to files.
2936 Contributions from: Alin Rauta, Alison Chaiken, Andrej
2937 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
2938 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
2939 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
2940 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
2941 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
2942 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
2943 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
2944 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
2945 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
2946 Michael Biebl, Michael Chapman, Michael Marineau, Michal
2947 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
2948 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
2949 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
2950 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
2951 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
2952 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
2953 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
2955 — Berlin, 2014-12-10
2959 * journalctl gained the new options -t/--identifier= to match
2960 on the syslog identifier (aka "tag"), as well as --utc to
2961 show log timestamps in the UTC timezone. journalctl now also
2962 accepts -n/--lines=all to disable line capping in a pager.
2964 * journalctl gained a new switch, --flush, that synchronously
2965 flushes logs from /run/log/journal to /var/log/journal if
2966 persistent storage is enabled. systemd-journal-flush.service
2967 now waits until the operation is complete.
2969 * Services can notify the manager before they start a reload
2970 (by sending RELOADING=1) or shutdown (by sending
2971 STOPPING=1). This allows the manager to track and show the
2972 internal state of daemons and closes a race condition when
2973 the process is still running but has closed its D-Bus
2976 * Services with Type=oneshot do not have to have any ExecStart
2979 * User units are now loaded also from
2980 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
2981 /run/systemd/user directory that was already previously
2982 supported, but is under the control of the user.
2984 * Job timeouts (i.e. time-outs on the time a job that is
2985 queued stays in the run queue) can now optionally result in
2986 immediate reboot or power-off actions (JobTimeoutAction= and
2987 JobTimeoutRebootArgument=). This is useful on ".target"
2988 units, to limit the maximum time a target remains
2989 undispatched in the run queue, and to trigger an emergency
2990 operation in such a case. This is now used by default to
2991 turn off the system if boot-up (as defined by everything in
2992 basic.target) hangs and does not complete for at least
2993 15min. Also, if power-off or reboot hang for at least 30min
2994 an immediate power-off/reboot operation is triggered. This
2995 functionality is particularly useful to increase reliability
2996 on embedded devices, but also on laptops which might
2997 accidentally get powered on when carried in a backpack and
2998 whose boot stays stuck in a hard disk encryption passphrase
3001 * systemd-logind can be configured to also handle lid switch
3002 events even when the machine is docked or multiple displays
3003 are attached (HandleLidSwitchDocked= option).
3005 * A helper binary and a service have been added which can be
3006 used to resume from hibernation in the initramfs. A
3007 generator will parse the resume= option on the kernel
3008 command line to trigger resume.
3010 * A user console daemon systemd-consoled has been
3011 added. Currently, it is a preview, and will so far open a
3012 single terminal on each session of the user marked as
3013 Desktop=systemd-console.
3015 * Route metrics can be specified for DHCP routes added by
3018 * The SELinux context of socket-activated services can be set
3019 from the information provided by the networking stack
3020 (SELinuxContextFromNet= option).
3022 * Userspace firmware loading support has been removed and
3023 the minimum supported kernel version is thus bumped to 3.7.
3025 * Timeout for udev workers has been increased from 1 to 3
3026 minutes, but a warning will be printed after 1 minute to
3027 help diagnose kernel modules that take a long time to load.
3029 * Udev rules can now remove tags on devices with TAG-="foobar".
3031 * systemd's readahead implementation has been removed. In many
3032 circumstances it didn't give expected benefits even for
3033 rotational disk drives and was becoming less relevant in the
3034 age of SSDs. As none of the developers has been using
3035 rotating media anymore, and nobody stepped up to actively
3036 maintain this component of systemd it has now been removed.
3038 * Swap units can use Options= to specify discard options.
3039 Discard options specified for swaps in /etc/fstab are now
3042 * Docker containers are now detected as a separate type of
3045 * The Password Agent protocol gained support for queries where
3046 the user input is shown, useful e.g. for user names.
3047 systemd-ask-password gained a new --echo option to turn that
3050 * The default sysctl.d/ snippets will now set:
3052 net.core.default_qdisc = fq_codel
3054 This selects Fair Queuing Controlled Delay as the default
3055 queuing discipline for network interfaces. fq_codel helps
3056 fight the network bufferbloat problem. It is believed to be
3057 a good default with no tuning required for most workloads.
3058 Downstream distributions may override this choice. On 10Gbit
3059 servers that do not do forwarding, "fq" may perform better.
3060 Systems without a good clocksource should use "pfifo_fast".
3062 * If kdbus is enabled during build a new option BusPolicy= is
3063 available for service units, that allows locking all service
3064 processes into a stricter bus policy, in order to limit
3065 access to various bus services, or even hide most of them
3066 from the service's view entirely.
3068 * networkctl will now show the .network and .link file
3069 networkd has applied to a specific interface.
3071 * sd-login gained a new API call sd_session_get_desktop() to
3072 query which desktop environment has been selected for a
3075 * UNIX utmp support is now compile-time optional to support
3076 legacy-free systems.
3078 * systemctl gained two new commands "add-wants" and
3079 "add-requires" for pulling in units from specific targets
3082 * If the word "rescue" is specified on the kernel command line
3083 the system will now boot into rescue mode (aka
3084 rescue.target), which was previously available only by
3085 specifying "1" or "systemd.unit=rescue.target" on the kernel
3086 command line. This new kernel command line option nicely
3087 mirrors the already existing "emergency" kernel command line
3090 * New kernel command line options mount.usr=, mount.usrflags=,
3091 mount.usrfstype= have been added that match root=, rootflags=,
3092 rootfstype= but allow mounting a specific file system to
3095 * The $NOTIFY_SOCKET is now also passed to control processes of
3096 services, not only the main process.
3098 * This version reenables support for fsck's -l switch. This
3099 means at least version v2.25 of util-linux is required for
3100 operation, otherwise dead-locks on device nodes may
3101 occur. Again: you need to update util-linux to at least
3102 v2.25 when updating systemd to v217.
3104 * The "multi-seat-x" tool has been removed from systemd, as
3105 its functionality has been integrated into X servers 1.16,
3106 and the tool is hence redundant. It is recommended to update
3107 display managers invoking this tool to simply invoke X
3108 directly from now on, again.
3110 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
3111 message flag has been added for all of systemd's PolicyKit
3112 authenticated method calls has been added. In particular
3113 this now allows optional interactive authorization via
3114 PolicyKit for many of PID1's privileged operations such as
3115 unit file enabling and disabling.
3117 * "udevadm hwdb --update" learnt a new switch "--usr" for
3118 placing the rebuilt hardware database in /usr instead of
3119 /etc. When used only hardware database entries stored in
3120 /usr will be used, and any user database entries in /etc are
3121 ignored. This functionality is useful for vendors to ship a
3122 pre-built database on systems where local configuration is
3123 unnecessary or unlikely.
3125 * Calendar time specifications in .timer units now also
3126 understand the strings "semi-annually", "quarterly" and
3127 "minutely" as shortcuts (in addition to the preexisting
3128 "anually", "hourly", ...).
3130 * systemd-tmpfiles will now correctly create files in /dev
3131 at boot which are marked for creation only at boot. It is
3132 recommended to always create static device nodes with 'c!'
3133 and 'b!', so that they are created only at boot and not
3134 overwritten at runtime.
3136 * When the watchdog logic is used for a service (WatchdogSec=)
3137 and the watchdog timeout is hit the service will now be
3138 terminated with SIGABRT (instead of just SIGTERM), in order
3139 to make sure a proper coredump and backtrace is
3140 generated. This ensures that hanging services will result in
3141 similar coredump/backtrace behaviour as services that hit a
3144 Contributions from: Andreas Henriksson, Andrei Borzenkov,
3145 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
3146 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
3147 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
3148 Herrmann, David Sommerseth, David Strauss, Emil Renner
3149 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
3150 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
3151 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
3152 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
3153 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
3154 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
3155 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
3156 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
3157 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
3158 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
3159 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
3160 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
3161 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
3162 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
3163 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
3164 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
3167 — Berlin, 2014-10-28
3171 * timedated no longer reads NTP implementation unit names from
3172 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
3173 implementations should add a
3175 Conflicts=systemd-timesyncd.service
3177 to their unit files to take over and replace systemd's NTP
3178 default functionality.
3180 * systemd-sysusers gained a new line type "r" for configuring
3181 which UID/GID ranges to allocate system users/groups
3182 from. Lines of type "u" may now add an additional column
3183 that specifies the home directory for the system user to be
3184 created. Also, systemd-sysusers may now optionally read user
3185 information from STDIN instead of a file. This is useful for
3186 invoking it from RPM preinst scriptlets that need to create
3187 users before the first RPM file is installed since these
3188 files might need to be owned by them. A new
3189 %sysusers_create_inline RPM macro has been introduced to do
3190 just that. systemd-sysusers now updates the shadow files as
3191 well as the user/group databases, which should enhance
3192 compatibility with certain tools like grpck.
3194 * A number of bus APIs of PID 1 now optionally consult
3195 PolicyKit to permit access for otherwise unprivileged
3196 clients under certain conditions. Note that this currently
3197 doesn't support interactive authentication yet, but this is
3198 expected to be added eventually, too.
3200 * /etc/machine-info now has new fields for configuring the
3201 deployment environment of the machine, as well as the
3202 location of the machine. hostnamectl has been updated with
3203 new command to update these fields.
3205 * systemd-timesyncd has been updated to automatically acquire
3206 NTP server information from systemd-networkd, which might
3207 have been discovered via DHCP.
3209 * systemd-resolved now includes a caching DNS stub resolver
3210 and a complete LLMNR name resolution implementation. A new
3211 NSS module "nss-resolve" has been added which can be used
3212 instead of glibc's own "nss-dns" to resolve hostnames via
3213 systemd-resolved. Hostnames, addresses and arbitrary RRs may
3214 be resolved via systemd-resolved D-Bus APIs. In contrast to
3215 the glibc internal resolver systemd-resolved is aware of
3216 multi-homed system, and keeps DNS server and caches separate
3217 and per-interface. Queries are sent simultaneously on all
3218 interfaces that have DNS servers configured, in order to
3219 properly handle VPNs and local LANs which might resolve
3220 separate sets of domain names. systemd-resolved may acquire
3221 DNS server information from systemd-networkd automatically,
3222 which in turn might have discovered them via DHCP. A tool
3223 "systemd-resolve-host" has been added that may be used to
3224 query the DNS logic in resolved. systemd-resolved implements
3225 IDNA and automatically uses IDNA or UTF-8 encoding depending
3226 on whether classic DNS or LLMNR is used as transport. In the
3227 next releases we intend to add a DNSSEC and mDNS/DNS-SD
3228 implementation to systemd-resolved.
3230 * A new NSS module nss-mymachines has been added, that
3231 automatically resolves the names of all local registered
3232 containers to their respective IP addresses.
3234 * A new client tool "networkctl" for systemd-networkd has been
3235 added. It currently is entirely passive and will query
3236 networking configuration from udev, rtnetlink and networkd,
3237 and present it to the user in a very friendly
3238 way. Eventually, we hope to extend it to become a full
3239 control utility for networkd.
3241 * .socket units gained a new DeferAcceptSec= setting that
3242 controls the kernels' TCP_DEFER_ACCEPT sockopt for
3243 TCP. Similarly, support for controlling TCP keep-alive
3244 settings has been added (KeepAliveTimeSec=,
3245 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
3246 turning off Nagle's algorithm on TCP has been added
3249 * logind learned a new session type "web", for use in projects
3250 like Cockpit which register web clients as PAM sessions.
3252 * timer units with at least one OnCalendar= setting will now
3253 be started only after timer-sync.target has been
3254 reached. This way they will not elapse before the system
3255 clock has been corrected by a local NTP client or
3256 similar. This is particular useful on RTC-less embedded
3257 machines, that come up with an invalid system clock.
3259 * systemd-nspawn's --network-veth= switch should now result in
3260 stable MAC addresses for both the outer and the inner side
3263 * systemd-nspawn gained a new --volatile= switch for running
3264 container instances with /etc or /var unpopulated.
3266 * The kdbus client code has been updated to use the new Linux
3267 3.17 memfd subsystem instead of the old kdbus-specific one.
3269 * systemd-networkd's DHCP client and server now support
3270 FORCERENEW. There are also new configuration options to
3271 configure the vendor client identifier and broadcast mode
3274 * systemd will no longer inform the kernel about the current
3275 timezone, as this is necessarily incorrect and racy as the
3276 kernel has no understanding of DST and similar
3277 concepts. This hence means FAT timestamps will be always
3278 considered UTC, similar to what Android is already
3279 doing. Also, when the RTC is configured to the local time
3280 (rather than UTC) systemd will never synchronize back to it,
3281 as this might confuse Windows at a later boot.
3283 * systemd-analyze gained a new command "verify" for offline
3284 validation of unit files.
3286 * systemd-networkd gained support for a couple of additional
3287 settings for bonding networking setups. Also, the metric for
3288 statically configured routes may now be configured. For
3289 network interfaces where this is appropriate the peer IP
3290 address may now be configured.
3292 * systemd-networkd's DHCP client will no longer request
3293 broadcasting by default, as this tripped up some networks.
3294 For hardware where broadcast is required the feature should
3295 be switched back on using RequestBroadcast=yes.
3297 * systemd-networkd will now set up IPv4LL addresses (when
3298 enabled) even if DHCP is configured successfully.
3300 * udev will now default to respect network device names given
3301 by the kernel when the kernel indicates that these are
3302 predictable. This behavior can be tweaked by changing
3303 NamePolicy= in the relevant .link file.
3305 * A new library systemd-terminal has been added that
3306 implements full TTY stream parsing and rendering. This
3307 library is supposed to be used later on for implementing a
3308 full userspace VT subsystem, replacing the current kernel
3311 * A new tool systemd-journal-upload has been added to push
3312 journal data to a remote system running
3313 systemd-journal-remote.
3315 * journald will no longer forward all local data to another
3316 running syslog daemon. This change has been made because
3317 rsyslog (which appears to be the most commonly used syslog
3318 implementation these days) no longer makes use of this, and
3319 instead pulls the data out of the journal on its own. Since
3320 forwarding the messages to a non-existent syslog server is
3321 more expensive than we assumed we have now turned this
3322 off. If you run a syslog server that is not a recent rsyslog
3323 version, you have to turn this option on again
3324 (ForwardToSyslog= in journald.conf).
3326 * journald now optionally supports the LZ4 compressor for
3327 larger journal fields. This compressor should perform much
3328 better than XZ which was the previous default.
3330 * machinectl now shows the IP addresses of local containers,
3331 if it knows them, plus the interface name of the container.
3333 * A new tool "systemd-escape" has been added that makes it
3334 easy to escape strings to build unit names and similar.
3336 * sd_notify() messages may now include a new ERRNO= field
3337 which is parsed and collected by systemd and shown among the
3338 "systemctl status" output for a service.
3340 * A new component "systemd-firstboot" has been added that
3341 queries the most basic systemd information (timezone,
3342 hostname, root password) interactively on first
3343 boot. Alternatively it may also be used to provision these
3344 things offline on OS images installed into directories.
3346 * The default sysctl.d/ snippets will now set
3348 net.ipv4.conf.default.promote_secondaries=1
3350 This has the benefit of no flushing secondary IP addresses
3351 when primary addresses are removed.
3353 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
3354 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
3355 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
3356 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
3357 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
3358 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
3359 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
3360 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
3361 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
3362 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
3363 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
3364 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
3365 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
3366 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
3367 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
3369 — Berlin, 2014-08-19
3373 * A new tool systemd-sysusers has been added. This tool
3374 creates system users and groups in /etc/passwd and
3375 /etc/group, based on static declarative system user/group
3376 definitions in /usr/lib/sysusers.d/. This is useful to
3377 enable factory resets and volatile systems that boot up with
3378 an empty /etc directory, and thus need system users and
3379 groups created during early boot. systemd now also ships
3380 with two default sysusers.d/ files for the most basic
3381 users and groups systemd and the core operating system
3384 * A new tmpfiles snippet has been added that rebuilds the
3385 essential files in /etc on boot, should they be missing.
3387 * A directive for ensuring automatic clean-up of
3388 /var/cache/man/ has been removed from the default
3389 configuration. This line should now be shipped by the man
3390 implementation. The necessary change has been made to the
3391 man-db implementation. Note that you need to update your man
3392 implementation to one that ships this line, otherwise no
3393 automatic clean-up of /var/cache/man will take place.
3395 * A new condition ConditionNeedsUpdate= has been added that
3396 may conditionalize services to only run when /etc or /var
3397 are "older" than the vendor operating system resources in
3398 /usr. This is useful for reconstructing or updating /etc
3399 after an offline update of /usr or a factory reset, on the
3400 next reboot. Services that want to run once after such an
3401 update or reset should use this condition and order
3402 themselves before the new systemd-update-done.service, which
3403 will mark the two directories as fully updated. A number of
3404 service files have been added making use of this, to rebuild
3405 the udev hardware database, the journald message catalog and
3406 dynamic loader cache (ldconfig). The systemd-sysusers tool
3407 described above also makes use of this now. With this in
3408 place it is now possible to start up a minimal operating
3409 system with /etc empty cleanly. For more information on the
3410 concepts involved see this recent blog story:
3412 http://0pointer.de/blog/projects/stateless.html
3414 * A new system group "input" has been introduced, and all
3415 input device nodes get this group assigned. This is useful
3416 for system-level software to get access to input devices. It
3417 complements what is already done for "audio" and "video".
3419 * systemd-networkd learnt minimal DHCPv4 server support in
3420 addition to the existing DHCPv4 client support. It also
3421 learnt DHCPv6 client and IPv6 Router Solicitation client
3422 support. The DHCPv4 client gained support for static routes
3423 passed in from the server. Note that the [DHCPv4] section
3424 known in older systemd-networkd versions has been renamed to
3425 [DHCP] and is now also used by the DHCPv6 client. Existing
3426 .network files using settings of this section should be
3427 updated, though compatibility is maintained. Optionally, the
3428 client hostname may now be sent to the DHCP server.
3430 * networkd gained support for vxlan virtual networks as well
3431 as tun/tap and dummy devices.
3433 * networkd gained support for automatic allocation of address
3434 ranges for interfaces from a system-wide pool of
3435 addresses. This is useful for dynamically managing a large
3436 number of interfaces with a single network configuration
3437 file. In particular this is useful to easily assign
3438 appropriate IP addresses to the veth links of a large number
3439 of nspawn instances.
3441 * RPM macros for processing sysusers, sysctl and binfmt
3442 drop-in snippets at package installation time have been
3445 * The /etc/os-release file should now be placed in
3446 /usr/lib/os-release. The old location is automatically
3447 created as symlink. /usr/lib is the more appropriate
3448 location of this file, since it shall actually describe the
3449 vendor operating system shipped in /usr, and not the
3450 configuration stored in /etc.
3452 * .mount units gained a new boolean SloppyOptions= setting
3453 that maps to mount(8)'s -s option which enables permissive
3454 parsing of unknown mount options.
3456 * tmpfiles learnt a new "L+" directive which creates a symlink
3457 but (unlike "L") deletes a pre-existing file first, should
3458 it already exist and not already be the correct
3459 symlink. Similarly, "b+", "c+" and "p+" directives have been
3460 added as well, which create block and character devices, as
3461 well as fifos in the filesystem, possibly removing any
3462 pre-existing files of different types.
3464 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
3465 'argument' field (which so far specified the source to
3466 symlink/copy the files from) is now optional. If omitted the
3467 same file os copied from /usr/share/factory/ suffixed by the
3468 full destination path. This is useful for populating /etc
3469 with essential files, by copying them from vendor defaults
3470 shipped in /usr/share/factory/etc.
3472 * A new command "systemctl preset-all" has been added that
3473 applies the service preset settings to all installed unit
3474 files. A new switch --preset-mode= has been added that
3475 controls whether only enable or only disable operations
3478 * A new command "systemctl is-system-running" has been added
3479 that allows checking the overall state of the system, for
3480 example whether it is fully up and running.
3482 * When the system boots up with an empty /etc, the equivalent
3483 to "systemctl preset-all" is executed during early boot, to
3484 make sure all default services are enabled after a factory
3487 * systemd now contains a minimal preset file that enables the
3488 most basic services systemd ships by default.
3490 * Unit files' [Install] section gained a new DefaultInstance=
3491 field for defining the default instance to create if a
3492 template unit is enabled with no instance specified.
3494 * A new passive target cryptsetup-pre.target has been added
3495 that may be used by services that need to make they run and
3496 finish before the first LUKS cryptographic device is set up.
3498 * The /dev/loop-control and /dev/btrfs-control device nodes
3499 are now owned by the "disk" group by default, opening up
3500 access to this group.
3502 * systemd-coredump will now automatically generate a
3503 stack trace of all core dumps taking place on the system,
3504 based on elfutils' libdw library. This stack trace is logged
3507 * systemd-coredump may now optionally store coredumps directly
3508 on disk (in /var/lib/systemd/coredump, possibly compressed),
3509 instead of storing them unconditionally in the journal. This
3510 mode is the new default. A new configuration file
3511 /etc/systemd/coredump.conf has been added to configure this
3512 and other parameters of systemd-coredump.
3514 * coredumpctl gained a new "info" verb to show details about a
3515 specific coredump. A new switch "-1" has also been added
3516 that makes sure to only show information about the most
3517 recent entry instead of all entries. Also, as the tool is
3518 generally useful now the "systemd-" prefix of the binary
3519 name has been removed. Distributions that want to maintain
3520 compatibility with the old name should add a symlink from
3521 the old name to the new name.
3523 * journald's SplitMode= now defaults to "uid". This makes sure
3524 that unprivileged users can access their own coredumps with
3525 coredumpctl without restrictions.
3527 * New kernel command line options "systemd.wants=" (for
3528 pulling an additional unit during boot), "systemd.mask="
3529 (for masking a specific unit for the boot), and
3530 "systemd.debug-shell" (for enabling the debug shell on tty9)
3531 have been added. This is implemented in the new generator
3532 "systemd-debug-generator".
3534 * systemd-nspawn will now by default filter a couple of
3535 syscalls for containers, among them those required for
3536 kernel module loading, direct x86 IO port access, swap
3537 management, and kexec. Most importantly though
3538 open_by_handle_at() is now prohibited for containers,
3539 closing a hole similar to a recently discussed vulnerability
3540 in docker regarding access to files on file hierarchies the
3541 container should normally not have access to. Note that, for
3542 nspawn, we generally make no security claims anyway (and
3543 this is explicitly documented in the man page), so this is
3544 just a fix for one of the most obvious problems.
3546 * A new man page file-hierarchy(7) has been added that
3547 contains a minimized, modernized version of the file system
3548 layout systemd expects, similar in style to the FHS
3549 specification or hier(5). A new tool systemd-path(1) has
3550 been added to query many of these paths for the local
3553 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
3554 longer done. Since the directory now has a per-user size
3555 limit, and is cleaned on logout this appears unnecessary,
3556 in particular since this now brings the lifecycle of this
3557 directory closer in line with how IPC objects are handled.
3559 * systemd.pc now exports a number of additional directories,
3560 including $libdir (which is useful to identify the library
3561 path for the primary architecture of the system), and a
3562 couple of drop-in directories.
3564 * udev's predictable network interface names now use the dev_port
3565 sysfs attribute, introduced in linux 3.15 instead of dev_id to
3566 distinguish between ports of the same PCI function. dev_id should
3567 only be used for ports using the same HW address, hence the need
3570 * machined has been updated to export the OS version of a
3571 container (read from /etc/os-release and
3572 /usr/lib/os-release) on the bus. This is now shown in
3573 "machinectl status" for a machine.
3575 * A new service setting RestartForceExitStatus= has been
3576 added. If configured to a set of exit signals or process
3577 return values, the service will be restarted when the main
3578 daemon process exits with any of them, regardless of the
3581 * systemctl's -H switch for connecting to remote systemd
3582 machines has been extended so that it may be used to
3583 directly connect to a specific container on the
3584 host. "systemctl -H root@foobar:waldi" will now connect as
3585 user "root" to host "foobar", and then proceed directly to
3586 the container named "waldi". Note that currently you have to
3587 authenticate as user "root" for this to work, as entering
3588 containers is a privileged operation.
3590 Contributions from: Andreas Henriksson, Benjamin Steinwender,
3591 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
3592 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
3593 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
3594 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
3595 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
3596 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
3597 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
3598 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
3599 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
3600 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
3601 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
3603 — Berlin, 2014-07-03
3607 * As an experimental feature, udev now tries to lock the
3608 disk device node (flock(LOCK_SH|LOCK_NB)) while it
3609 executes events for the disk or any of its partitions.
3610 Applications like partitioning programs can lock the
3611 disk device node (flock(LOCK_EX)) and claim temporary
3612 device ownership that way; udev will entirely skip all event
3613 handling for this disk and its partitions. If the disk
3614 was opened for writing, the close will trigger a partition
3615 table rescan in udev's "watch" facility, and if needed
3616 synthesize "change" events for the disk and all its partitions.
3617 This is now unconditionally enabled, and if it turns out to
3618 cause major problems, we might turn it on only for specific
3619 devices, or might need to disable it entirely. Device Mapper
3620 devices are excluded from this logic.
3622 * We temporarily dropped the "-l" switch for fsck invocations,
3623 since they collide with the flock() logic above. util-linux
3624 upstream has been changed already to avoid this conflict,
3625 and we will readd "-l" as soon as util-linux with this
3626 change has been released.
3628 * The dependency on libattr has been removed. Since a long
3629 time, the extended attribute calls have moved to glibc, and
3630 libattr is thus unnecessary.
3632 * Virtualization detection works without privileges now. This
3633 means the systemd-detect-virt binary no longer requires
3634 CAP_SYS_PTRACE file capabilities, and our daemons can run
3635 with fewer privileges.
3637 * systemd-networkd now runs under its own "systemd-network"
3638 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
3639 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
3640 loses the ability to write to files owned by root this way.
3642 * Similarly, systemd-resolved now runs under its own
3643 "systemd-resolve" user with no capabilities remaining.
3645 * Similarly, systemd-bus-proxyd now runs under its own
3646 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
3648 * systemd-networkd gained support for setting up "veth"
3649 virtual Ethernet devices for container connectivity, as well
3650 as GRE and VTI tunnels.
3652 * systemd-networkd will no longer automatically attempt to
3653 manually load kernel modules necessary for certain tunnel
3654 transports. Instead, it is assumed the kernel loads them
3655 automatically when required. This only works correctly on
3656 very new kernels. On older kernels, please consider adding
3657 the kernel modules to /etc/modules-load.d/ as a work-around.
3659 * The resolv.conf file systemd-resolved generates has been
3660 moved to /run/systemd/resolve/. If you have a symlink from
3661 /etc/resolv.conf, it might be necessary to correct it.
3663 * Two new service settings, ProtectHome= and ProtectSystem=,
3664 have been added. When enabled, they will make the user data
3665 (such as /home) inaccessible or read-only and the system
3666 (such as /usr) read-only, for specific services. This allows
3667 very light-weight per-service sandboxing to avoid
3668 modifications of user data or system files from
3669 services. These two new switches have been enabled for all
3670 of systemd's long-running services, where appropriate.
3672 * Socket units gained new SocketUser= and SocketGroup=
3673 settings to set the owner user and group of AF_UNIX sockets
3674 and FIFOs in the file system.
3676 * Socket units gained a new RemoveOnStop= setting. If enabled,
3677 all FIFOS and sockets in the file system will be removed
3678 when the specific socket unit is stopped.
3680 * Socket units gained a new Symlinks= setting. It takes a list
3681 of symlinks to create to file system sockets or FIFOs
3682 created by the specific Unix sockets. This is useful to
3683 manage symlinks to socket nodes with the same life-cycle as
3686 * The /dev/log socket and /dev/initctl FIFO have been moved to
3687 /run, and have been replaced by symlinks. This allows
3688 connecting to these facilities even if PrivateDevices=yes is
3689 used for a service (which makes /dev/log itself unavailable,
3690 but /run is left). This also has the benefit of ensuring
3691 that /dev only contains device nodes, directories and
3692 symlinks, and nothing else.
3694 * sd-daemon gained two new calls sd_pid_notify() and
3695 sd_pid_notifyf(). They are similar to sd_notify() and
3696 sd_notifyf(), but allow overriding of the source PID of
3697 notification messages if permissions permit this. This is
3698 useful to send notify messages on behalf of a different
3699 process (for example, the parent process). The
3700 systemd-notify tool has been updated to make use of this
3701 when sending messages (so that notification messages now
3702 originate from the shell script invoking systemd-notify and
3703 not the systemd-notify process itself. This should minimize
3704 a race where systemd fails to associate notification
3705 messages to services when the originating process already
3708 * A new "on-abnormal" setting for Restart= has been added. If
3709 set, it will result in automatic restarts on all "abnormal"
3710 reasons for a process to exit, which includes unclean
3711 signals, core dumps, timeouts and watchdog timeouts, but
3712 does not include clean and unclean exit codes or clean
3713 signals. Restart=on-abnormal is an alternative for
3714 Restart=on-failure for services that shall be able to
3715 terminate and avoid restarts on certain errors, by
3716 indicating so with an unclean exit code. Restart=on-failure
3717 or Restart=on-abnormal is now the recommended setting for
3718 all long-running services.
3720 * If the InaccessibleDirectories= service setting points to a
3721 mount point (or if there are any submounts contained within
3722 it), it is now attempted to completely unmount it, to make
3723 the file systems truly unavailable for the respective
3726 * The ReadOnlyDirectories= service setting and
3727 systemd-nspawn's --read-only parameter are now recursively
3728 applied to all submounts, too.
3730 * Mount units may now be created transiently via the bus APIs.
3732 * The support for SysV and LSB init scripts has been removed
3733 from the systemd daemon itself. Instead, it is now
3734 implemented as a generator that creates native systemd units
3735 from these scripts when needed. This enables us to remove a
3736 substantial amount of legacy code from PID 1, following the
3737 fact that many distributions only ship a very small number
3738 of LSB/SysV init scripts nowadays.
3740 * Privileged Xen (dom0) domains are not considered
3741 virtualization anymore by the virtualization detection
3742 logic. After all, they generally have unrestricted access to
3743 the hardware and usually are used to manage the unprivileged
3746 * systemd-tmpfiles gained a new "C" line type, for copying
3747 files or entire directories.
3749 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
3750 lines. So far, they have been non-globbing versions of the
3751 latter, and have thus been redundant. In future, it is
3752 recommended to only use "z". "m" has hence been removed
3753 from the documentation, even though it stays supported.
3755 * A tmpfiles snippet to recreate the most basic structure in
3756 /var has been added. This is enough to create the /var/run →
3757 /run symlink and create a couple of structural
3758 directories. This allows systems to boot up with an empty or
3759 volatile /var. Of course, while with this change, the core OS
3760 now is capable with dealing with a volatile /var, not all
3761 user services are ready for it. However, we hope that sooner
3762 or later, many service daemons will be changed upstream so
3763 that they are able to automatically create their necessary
3764 directories in /var at boot, should they be missing. This is
3765 the first step to allow state-less systems that only require
3766 the vendor image for /usr to boot.
3768 * systemd-nspawn has gained a new --tmpfs= switch to mount an
3769 empty tmpfs instance to a specific directory. This is
3770 particularly useful for making use of the automatic
3771 reconstruction of /var (see above), by passing --tmpfs=/var.
3773 * Access modes specified in tmpfiles snippets may now be
3774 prefixed with "~", which indicates that they shall be masked
3775 by whether the existing file or directory is currently
3776 writable, readable or executable at all. Also, if specified,
3777 the sgid/suid/sticky bits will be masked for all
3780 * A new passive target unit "network-pre.target" has been
3781 added which is useful for services that shall run before any
3782 network is configured, for example firewall scripts.
3784 * The "floppy" group that previously owned the /dev/fd*
3785 devices is no longer used. The "disk" group is now used
3786 instead. Distributions should probably deprecate usage of
3789 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
3790 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
3791 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
3792 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
3793 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
3794 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
3797 — Berlin, 2014-06-11
3801 * A new "systemd-timesyncd" daemon has been added for
3802 synchronizing the system clock across the network. It
3803 implements an SNTP client. In contrast to NTP
3804 implementations such as chrony or the NTP reference server,
3805 this only implements a client side, and does not bother with
3806 the full NTP complexity, focusing only on querying time from
3807 one remote server and synchronizing the local clock to
3808 it. Unless you intend to serve NTP to networked clients or
3809 want to connect to local hardware clocks, this simple NTP
3810 client should be more than appropriate for most
3811 installations. The daemon runs with minimal privileges, and
3812 has been hooked up with networkd to only operate when
3813 network connectivity is available. The daemon saves the
3814 current clock to disk every time a new NTP sync has been
3815 acquired, and uses this to possibly correct the system clock
3816 early at bootup, in order to accommodate for systems that
3817 lack an RTC such as the Raspberry Pi and embedded devices,
3818 and to make sure that time monotonically progresses on these
3819 systems, even if it is not always correct. To make use of
3820 this daemon, a new system user and group "systemd-timesync"
3821 needs to be created on installation of systemd.
3823 * The queue "seqnum" interface of libudev has been disabled, as
3824 it was generally incompatible with device namespacing as
3825 sequence numbers of devices go "missing" if the devices are
3826 part of a different namespace.
3828 * "systemctl list-timers" and "systemctl list-sockets" gained
3829 a --recursive switch for showing units of these types also
3830 for all local containers, similar in style to the already
3831 supported --recursive switch for "systemctl list-units".
3833 * A new RebootArgument= setting has been added for service
3834 units, which may be used to specify a kernel reboot argument
3835 to use when triggering reboots with StartLimitAction=.
3837 * A new FailureAction= setting has been added for service
3838 units which may be used to specify an operation to trigger
3839 when a service fails. This works similarly to
3840 StartLimitAction=, but unlike it, controls what is done
3841 immediately rather than only after several attempts to
3842 restart the service in question.
3844 * hostnamed got updated to also expose the kernel name,
3845 release, and version on the bus. This is useful for
3846 executing commands like hostnamectl with the -H switch.
3847 systemd-analyze makes use of this to properly display
3848 details when running non-locally.
3850 * The bootchart tool can now show cgroup information in the
3851 graphs it generates.
3853 * The CFS CPU quota cgroup attribute is now exposed for
3854 services. The new CPUQuota= switch has been added for this
3855 which takes a percentage value. Setting this will have the
3856 result that a service may never get more CPU time than the
3857 specified percentage, even if the machine is otherwise idle.
3859 * systemd-networkd learned IPIP and SIT tunnel support.
3861 * LSB init scripts exposing a dependency on $network will now
3862 get a dependency on network-online.target rather than simply
3863 network.target. This should bring LSB handling closer to
3864 what it was on SysV systems.
3866 * A new fsck.repair= kernel option has been added to control
3867 how fsck shall deal with unclean file systems at boot.
3869 * The (.ini) configuration file parser will now silently
3870 ignore sections whose name begins with "X-". This may be
3871 used to maintain application-specific extension sections in unit
3874 * machined gained a new API to query the IP addresses of
3875 registered containers. "machinectl status" has been updated
3876 to show these addresses in its output.
3878 * A new call sd_uid_get_display() has been added to the
3879 sd-login APIs for querying the "primary" session of a
3880 user. The "primary" session of the user is elected from the
3881 user's sessions and generally a graphical session is
3882 preferred over a text one.
3884 * A minimal systemd-resolved daemon has been added. It
3885 currently simply acts as a companion to systemd-networkd and
3886 manages resolv.conf based on per-interface DNS
3887 configuration, possibly supplied via DHCP. In the long run
3888 we hope to extend this into a local DNSSEC enabled DNS and
3891 * The systemd-networkd-wait-online tool is now enabled by
3892 default. It will delay network-online.target until a network
3893 connection has been configured. The tool primarily integrates
3894 with networkd, but will also make a best effort to make sense
3895 of network configuration performed in some other way.
3897 * Two new service options StartupCPUShares= and
3898 StartupBlockIOWeight= have been added that work similarly to
3899 CPUShares= and BlockIOWeight= however only apply during
3900 system startup. This is useful to prioritize certain services
3901 differently during bootup than during normal runtime.
3903 * hostnamed has been changed to prefer the statically
3904 configured hostname in /etc/hostname (unless set to
3905 'localhost' or empty) over any dynamic one supplied by
3906 dhcp. With this change, the rules for picking the hostname
3907 match more closely the rules of other configuration settings
3908 where the local administrator's configuration in /etc always
3909 overrides any other settings.
3911 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
3912 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
3913 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
3914 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
3915 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
3916 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
3917 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
3918 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
3919 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
3920 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
3921 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
3922 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
3923 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
3924 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
3925 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
3926 Lindskog, WaLyong Cho, Will Woods, Zbigniew
3929 — Beijing, 2014-05-28
3933 * When restoring the screen brightness at boot, stay away from
3934 the darkest setting or from the lowest 5% of the available
3935 range, depending on which is the larger value of both. This
3936 should effectively protect the user from rebooting into a
3937 black screen, should the brightness have been set to minimum
3940 * sd-login gained a new sd_machine_get_class() call to
3941 determine the class ("vm" or "container") of a machine
3942 registered with machined.
3944 * sd-login gained new calls
3945 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
3946 to query the identity of the peer of a local AF_UNIX
3947 connection. They operate similarly to their sd_pid_get_xyz()
3950 * PID 1 will now maintain a system-wide system state engine
3951 with the states "starting", "running", "degraded",
3952 "maintenance", "stopping". These states are bound to system
3953 startup, normal runtime, runtime with at least one failed
3954 service, rescue/emergency mode and system shutdown. This
3955 state is shown in the "systemctl status" output when no unit
3956 name is passed. It is useful to determine system state, in
3957 particularly when doing so for many systems or containers at
3960 * A new command "list-machines" has been added to "systemctl"
3961 that lists all local OS containers and shows their system
3962 state (see above), if systemd runs inside of them.
3964 * systemctl gained a new "-r" switch to recursively enumerate
3965 units on all local containers, when used with the
3966 "list-unit" command (which is the default one that is
3967 executed when no parameters are specified).
3969 * The GPT automatic partition discovery logic will now honour
3970 two GPT partition flags: one may be set on a partition to
3971 cause it to be mounted read-only, and the other may be set
3972 on a partition to ignore it during automatic discovery.
3974 * Two new GPT type UUIDs have been added for automatic root
3975 partition discovery, for 32-bit and 64-bit ARM. This is not
3976 particularly useful for discovering the root directory on
3977 these architectures during bare-metal boots (since UEFI is
3978 not common there), but still very useful to allow booting of
3979 ARM disk images in nspawn with the -i option.
3981 * MAC addresses of interfaces created with nspawn's
3982 --network-interface= switch will now be generated from the
3983 machine name, and thus be stable between multiple invocations
3986 * logind will now automatically remove all IPC objects owned
3987 by a user if she or he fully logs out. This makes sure that
3988 users who are logged out cannot continue to consume IPC
3989 resources. This covers SysV memory, semaphores and message
3990 queues as well as POSIX shared memory and message
3991 queues. Traditionally, SysV and POSIX IPC had no life-cycle
3992 limits. With this functionality, that is corrected. This may
3993 be turned off by using the RemoveIPC= switch of logind.conf.
3995 * The systemd-machine-id-setup and tmpfiles tools gained a
3996 --root= switch to operate on a specific root directory,
3999 * journald can now forward logged messages to the TTYs of all
4000 logged in users ("wall"). This is the default for all
4001 emergency messages now.
4003 * A new tool systemd-journal-remote has been added to stream
4004 journal log messages across the network.
4006 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
4007 controller trees are mounted into it. Note that the
4008 directories mounted beneath it are not read-only. This is a
4009 security measure and is particularly useful because glibc
4010 actually includes a search logic to pick any tmpfs it can
4011 find to implement shm_open() if /dev/shm is not available
4012 (which it might very well be in namespaced setups).
4014 * machinectl gained a new "poweroff" command to cleanly power
4015 down a local OS container.
4017 * The PrivateDevices= unit file setting will now also drop the
4018 CAP_MKNOD capability from the capability bound set, and
4019 imply DevicePolicy=closed.
4021 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
4022 comprehensively on all long-running systemd services where
4023 this is appropriate.
4025 * systemd-udevd will now run in a disassociated mount
4026 namespace. To mount directories from udev rules, make sure to
4027 pull in mount units via SYSTEMD_WANTS properties.
4029 * The kdbus support gained support for uploading policy into
4030 the kernel. sd-bus gained support for creating "monitoring"
4031 connections that can eavesdrop into all bus communication
4032 for debugging purposes.
4034 * Timestamps may now be specified in seconds since the UNIX
4035 epoch Jan 1st, 1970 by specifying "@" followed by the value
4038 * Native tcpwrap support in systemd has been removed. tcpwrap
4039 is old code, not really maintained anymore and has serious
4040 shortcomings, and better options such as firewalls
4041 exist. For setups that require tcpwrap usage, please
4042 consider invoking your socket-activated service via tcpd,
4043 like on traditional inetd.
4045 * A new system.conf configuration option
4046 DefaultTimerAccuracySec= has been added that controls the
4047 default AccuracySec= setting of .timer units.
4049 * Timer units gained a new WakeSystem= switch. If enabled,
4050 timers configured this way will cause the system to resume
4051 from system suspend (if the system supports that, which most
4054 * Timer units gained a new Persistent= switch. If enabled,
4055 timers configured this way will save to disk when they have
4056 been last triggered. This information is then used on next
4057 reboot to possible execute overdue timer events, that
4058 could not take place because the system was powered off.
4059 This enables simple anacron-like behaviour for timer units.
4061 * systemctl's "list-timers" will now also list the time a
4062 timer unit was last triggered in addition to the next time
4063 it will be triggered.
4065 * systemd-networkd will now assign predictable IPv4LL
4066 addresses to its local interfaces.
4068 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
4069 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
4070 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
4071 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
4072 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
4073 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
4074 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
4075 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
4078 — Berlin, 2014-03-25
4082 * A new unit file setting RestrictAddressFamilies= has been
4083 added to restrict which socket address families unit
4084 processes gain access to. This takes address family names
4085 like "AF_INET" or "AF_UNIX", and is useful to minimize the
4086 attack surface of services via exotic protocol stacks. This
4087 is built on seccomp system call filters.
4089 * Two new unit file settings RuntimeDirectory= and
4090 RuntimeDirectoryMode= have been added that may be used to
4091 manage a per-daemon runtime directories below /run. This is
4092 an alternative for setting up directory permissions with
4093 tmpfiles snippets, and has the advantage that the runtime
4094 directory's lifetime is bound to the daemon runtime and that
4095 the daemon starts up with an empty directory each time. This
4096 is particularly useful when writing services that drop
4097 privileges using the User= or Group= setting.
4099 * The DeviceAllow= unit setting now supports globbing for
4100 matching against device group names.
4102 * The systemd configuration file system.conf gained new
4103 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
4104 DefaultMemoryAccounting= to globally turn on/off accounting
4105 for specific resources (cgroups) for all units. These
4106 settings may still be overridden individually in each unit
4109 * systemd-gpt-auto-generator is now able to discover /srv and
4110 root partitions in addition to /home and swap partitions. It
4111 also supports LUKS-encrypted partitions now. With this in
4112 place, automatic discovery of partitions to mount following
4113 the Discoverable Partitions Specification
4114 (https://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
4115 is now a lot more complete. This allows booting without
4116 /etc/fstab and without root= on the kernel command line on
4117 systems prepared appropriately.
4119 * systemd-nspawn gained a new --image= switch which allows
4120 booting up disk images and Linux installations on any block
4121 device that follow the Discoverable Partitions Specification
4122 (see above). This means that installations made with
4123 appropriately updated installers may now be started and
4124 deployed using container managers, completely
4125 unmodified. (We hope that libvirt-lxc will add support for
4126 this feature soon, too.)
4128 * systemd-nspawn gained a new --network-macvlan= setting to
4129 set up a private macvlan interface for the
4130 container. Similarly, systemd-networkd gained a new
4131 Kind=macvlan setting in .netdev files.
4133 * systemd-networkd now supports configuring local addresses
4136 * A new tool systemd-network-wait-online has been added to
4137 synchronously wait for network connectivity using
4140 * The sd-bus.h bus API gained a new sd_bus_track object for
4141 tracking the life-cycle of bus peers. Note that sd-bus.h is
4142 still not a public API though (unless you specify
4143 --enable-kdbus on the configure command line, which however
4144 voids your warranty and you get no API stability guarantee).
4146 * The $XDG_RUNTIME_DIR runtime directories for each user are
4147 now individual tmpfs instances, which has the benefit of
4148 introducing separate pools for each user, with individual
4149 size limits, and thus making sure that unprivileged clients
4150 can no longer negatively impact the system or other users by
4151 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
4152 RuntimeDirectorySize= has been introduced that allows
4153 controlling the default size limit for all users. It
4154 defaults to 10% of the available physical memory. This is no
4155 replacement for quotas on tmpfs though (which the kernel
4156 still does not support), as /dev/shm and /tmp are still
4157 shared resources used by both the system and unprivileged
4160 * logind will now automatically turn off automatic suspending
4161 on laptop lid close when more than one display is
4162 connected. This was previously expected to be implemented
4163 individually in desktop environments (such as GNOME),
4164 however has been added to logind now, in order to fix a
4165 boot-time race where a desktop environment might not have
4166 been started yet and thus not been able to take an inhibitor
4167 lock at the time where logind already suspends the system
4168 due to a closed lid.
4170 * logind will now wait at least 30s after each system
4171 suspend/resume cycle, and 3min after system boot before
4172 suspending the system due to a closed laptop lid. This
4173 should give USB docking stations and similar enough time to
4174 be probed and configured after system resume and boot in
4175 order to then act as suspend blocker.
4177 * systemd-run gained a new --property= setting which allows
4178 initialization of resource control properties (and others)
4179 for the created scope or service unit. Example: "systemd-run
4180 --property=BlockIOWeight=10 updatedb" may be used to run
4181 updatedb at a low block IO scheduling weight.
4183 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
4184 now also work in --scope mode.
4186 * When systemd is compiled with kdbus support, basic support
4187 for enforced policies is now in place. (Note that enabling
4188 kdbus still voids your warranty and no API compatibility
4191 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
4192 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
4193 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
4194 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
4195 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
4196 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
4197 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
4198 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
4199 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
4200 Zbigniew Jędrzejewski-Szmek
4202 — Berlin, 2014-03-12
4206 * systemd will now relabel /dev after loading the SMACK policy
4207 according to SMACK rules.
4209 * A new unit file option AppArmorProfile= has been added to
4210 set the AppArmor profile for the processes of a unit.
4212 * A new condition check ConditionArchitecture= has been added
4213 to conditionalize units based on the system architecture, as
4214 reported by uname()'s "machine" field.
4216 * systemd-networkd now supports matching on the system
4217 virtualization, architecture, kernel command line, host name
4220 * logind is now a lot more aggressive when suspending the
4221 machine due to a closed laptop lid. Instead of acting only
4222 on the lid close action, it will continuously watch the lid
4223 status and act on it. This is useful for laptops where the
4224 power button is on the outside of the chassis so that it can
4225 be reached without opening the lid (such as the Lenovo
4226 Yoga). On those machines, logind will now immediately
4227 re-suspend the machine if the power button has been
4228 accidentally pressed while the laptop was suspended and in a
4229 backpack or similar.
4231 * logind will now watch SW_DOCK switches and inhibit reaction
4232 to the lid switch if it is pressed. This means that logind
4233 will not suspend the machine anymore if the lid is closed
4234 and the system is docked, if the laptop supports SW_DOCK
4235 notifications via the input layer. Note that ACPI docking
4236 stations do not generate this currently. Also note that this
4237 logic is usually not fully sufficient and Desktop
4238 Environments should take a lid switch inhibitor lock when an
4239 external display is connected, as systemd will not watch
4242 * nspawn will now make use of the devices cgroup controller by
4243 default, and only permit creation of and access to the usual
4244 API device nodes like /dev/null or /dev/random, as well as
4245 access to (but not creation of) the pty devices.
4247 * We will now ship a default .network file for
4248 systemd-networkd that automatically configures DHCP for
4249 network interfaces created by nspawn's --network-veth or
4250 --network-bridge= switches.
4252 * systemd will now understand the usual M, K, G, T suffixes
4253 according to SI conventions (i.e. to the base 1000) when
4254 referring to throughput and hardware metrics. It will stay
4255 with IEC conventions (i.e. to the base 1024) for software
4256 metrics, according to what is customary according to
4257 Wikipedia. We explicitly document which base applies for
4258 each configuration option.
4260 * The DeviceAllow= setting in unit files now supports a syntax
4261 to whitelist an entire group of devices node majors at once,
4262 based on the /proc/devices listing. For example, with the
4263 string "char-pts", it is now possible to whitelist all
4264 current and future pseudo-TTYs at once.
4266 * sd-event learned a new "post" event source. Event sources of
4267 this type are triggered by the dispatching of any event
4268 source of a type that is not "post". This is useful for
4269 implementing clean-up and check event sources that are
4270 triggered by other work being done in the program.
4272 * systemd-networkd is no longer statically enabled, but uses
4273 the usual [Install] sections so that it can be
4274 enabled/disabled using systemctl. It still is enabled by
4277 * When creating a veth interface pair with systemd-nspawn, the
4278 host side will now be prefixed with "vb-" if
4279 --network-bridge= is used, and with "ve-" if --network-veth
4280 is used. This way, it is easy to distinguish these cases on
4281 the host, for example to apply different configuration to
4282 them with systemd-networkd.
4284 * The compatibility libraries for libsystemd-journal.so,
4285 libsystem-id128.so, libsystemd-login.so and
4286 libsystemd-daemon.so do not make use of IFUNC
4287 anymore. Instead, we now build libsystemd.so multiple times
4288 under these alternative names. This means that the footprint
4289 is drastically increased, but given that these are
4290 transitional compatibility libraries, this should not matter
4291 much. This change has been made necessary to support the ARM
4292 platform for these compatibility libraries, as the ARM
4293 toolchain is not really at the same level as the toolchain
4294 for other architectures like x86 and does not support
4295 IFUNC. Please make sure to use --enable-compat-libs only
4296 during a transitional period!
4298 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
4299 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
4300 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
4301 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
4302 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
4303 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
4304 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
4305 Zbigniew Jędrzejewski-Szmek
4307 — Berlin, 2014-02-24
4311 * A new component "systemd-networkd" has been added that can
4312 be used to configure local network interfaces statically or
4313 via DHCP. It is capable of bringing up bridges, VLANs, and
4314 bonding. Currently, no hook-ups for interactive network
4315 configuration are provided. Use this for your initrd,
4316 container, embedded, or server setup if you need a simple,
4317 yet powerful, network configuration solution. This
4318 configuration subsystem is quite nifty, as it allows wildcard
4319 hotplug matching in interfaces. For example, with a single
4320 configuration snippet, you can configure that all Ethernet
4321 interfaces showing up are automatically added to a bridge,
4322 or similar. It supports link-sensing and more.
4324 * A new tool "systemd-socket-proxyd" has been added which can
4325 act as a bidirectional proxy for TCP sockets. This is
4326 useful for adding socket activation support to services that
4327 do not actually support socket activation, including virtual
4328 machines and the like.
4330 * Add a new tool to save/restore rfkill state on
4333 * Save/restore state of keyboard backlights in addition to
4334 display backlights on shutdown/boot.
4336 * udev learned a new SECLABEL{} construct to label device
4337 nodes with a specific security label when they appear. For
4338 now, only SECLABEL{selinux} is supported, but the syntax is
4339 prepared for additional security frameworks.
4341 * udev gained a new scheme to configure link-level attributes
4342 from files in /etc/systemd/network/*.link. These files can
4343 match against MAC address, device path, driver name and type,
4344 and will apply attributes like the naming policy, link speed,
4345 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
4346 address assignment policy (randomized, ...).
4348 * The configuration of network interface naming rules for
4349 "permanent interface names" has changed: a new NamePolicy=
4350 setting in the [Link] section of .link files determines the
4351 priority of possible naming schemes (onboard, slot, MAC,
4352 path). The default value of this setting is determined by
4353 /usr/lib/net/links/99-default.link. Old
4354 80-net-name-slot.rules udev configuration file has been
4355 removed, so local configuration overriding this file should
4356 be adapted to override 99-default.link instead.
4358 * When the User= switch is used in a unit file, also
4359 initialize $SHELL= based on the user database entry.
4361 * systemd no longer depends on libdbus. All communication is
4362 now done with sd-bus, systemd's low-level bus library
4365 * kdbus support has been added to PID 1 itself. When kdbus is
4366 enabled, this causes PID 1 to set up the system bus and
4367 enable support for a new ".busname" unit type that
4368 encapsulates bus name activation on kdbus. It works a little
4369 bit like ".socket" units, except for bus names. A new
4370 generator has been added that converts classic dbus1 service
4371 activation files automatically into native systemd .busname
4374 * sd-bus: add a light-weight vtable implementation that allows
4375 defining objects on the bus with a simple static const
4376 vtable array of its methods, signals and properties.
4378 * systemd will not generate or install static dbus
4379 introspection data anymore to /usr/share/dbus-1/interfaces,
4380 as the precise format of these files is unclear, and
4381 nothing makes use of it.
4383 * A proxy daemon is now provided to proxy clients connecting
4384 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
4385 compatibility with classic D-Bus.
4387 * A bus driver implementation has been added that supports the
4388 classic D-Bus bus driver calls on kdbus, also for
4389 compatibility purposes.
4391 * A new API "sd-event.h" has been added that implements a
4392 minimal event loop API built around epoll. It provides a
4393 couple of features that direct epoll usage is lacking:
4394 prioritization of events, scales to large numbers of timer
4395 events, per-event timer slack (accuracy), system-wide
4396 coalescing of timer events, exit handlers, watchdog
4397 supervision support using systemd's sd_notify() API, child
4400 * A new API "sd-rntl.h" has been added that provides an API
4401 around the route netlink interface of the kernel, similar in
4402 style to "sd-bus.h".
4404 * A new API "sd-dhcp-client.h" has been added that provides a
4405 small DHCPv4 client-side implementation. This is used by
4408 * There is a new kernel command line option
4409 "systemd.restore_state=0|1". When set to "0", none of the
4410 systemd tools will restore saved runtime state to hardware
4411 devices. More specifically, the rfkill and backlight states
4414 * The FsckPassNo= compatibility option in mount/service units
4415 has been removed. The fstab generator will now add the
4416 necessary dependencies automatically, and does not require
4417 PID1's support for that anymore.
4419 * journalctl gained a new switch, --list-boots, that lists
4420 recent boots with their times and boot IDs.
4422 * The various tools like systemctl, loginctl, timedatectl,
4423 busctl, systemd-run, ... have gained a new switch "-M" to
4424 connect to a specific, local OS container (as direct
4425 connection, without requiring SSH). This works on any
4426 container that is registered with machined, such as those
4427 created by libvirt-lxc or nspawn.
4429 * systemd-run and systemd-analyze also gained support for "-H"
4430 to connect to remote hosts via SSH. This is particularly
4431 useful for systemd-run because it enables queuing of jobs
4432 onto remote systems.
4434 * machinectl gained a new command "login" to open a getty
4435 login in any local container. This works with any container
4436 that is registered with machined (such as those created by
4437 libvirt-lxc or nspawn), and which runs systemd inside.
4439 * machinectl gained a new "reboot" command that may be used to
4440 trigger a reboot on a specific container that is registered
4441 with machined. This works on any container that runs an init
4442 system of some kind.
4444 * systemctl gained a new "list-timers" command to print a nice
4445 listing of installed timer units with the times they elapse
4448 * Alternative reboot() parameters may now be specified on the
4449 "systemctl reboot" command line and are passed to the
4450 reboot() system call.
4452 * systemctl gained a new --job-mode= switch to configure the
4453 mode to queue a job with. This is a more generic version of
4454 --fail, --irreversible, and --ignore-dependencies, which are
4455 still available but not advertised anymore.
4457 * /etc/systemd/system.conf gained new settings to configure
4458 various default timeouts of units, as well as the default
4459 start limit interval and burst. These may still be overridden
4462 * PID1 will now export on the bus profile data of the security
4463 policy upload process (such as the SELinux policy upload to
4466 * journald: when forwarding logs to the console, include
4467 timestamps (following the setting in
4468 /sys/module/printk/parameters/time).
4470 * OnCalendar= in timer units now understands the special
4471 strings "yearly" and "annually". (Both are equivalent)
4473 * The accuracy of timer units is now configurable with the new
4474 AccuracySec= setting. It defaults to 1min.
4476 * A new dependency type JoinsNamespaceOf= has been added that
4477 allows running two services within the same /tmp and network
4478 namespace, if PrivateNetwork= or PrivateTmp= are used.
4480 * A new command "cat" has been added to systemctl. It outputs
4481 the original unit file of a unit, and concatenates the
4482 contents of additional "drop-in" unit file snippets, so that
4483 the full configuration is shown.
4485 * systemctl now supports globbing on the various "list-xyz"
4486 commands, like "list-units" or "list-sockets", as well as on
4487 those commands which take multiple unit names.
4489 * journalctl's --unit= switch gained support for globbing.
4491 * All systemd daemons now make use of the watchdog logic so
4492 that systemd automatically notices when they hang.
4494 * If the $container_ttys environment variable is set,
4495 getty-generator will automatically spawn a getty for each
4496 listed tty. This is useful for container managers to request
4497 login gettys to be spawned on as many ttys as needed.
4499 * %h, %s, %U specifier support is not available anymore when
4500 used in unit files for PID 1. This is because NSS calls are
4501 not safe from PID 1. They stay available for --user
4502 instances of systemd, and as special case for the root user.
4504 * loginctl gained a new "--no-legend" switch to turn off output
4507 * The "sd-login.h" API gained three new calls:
4508 sd_session_is_remote(), sd_session_get_remote_user(),
4509 sd_session_get_remote_host() to query information about
4512 * The udev hardware database now also carries vendor/product
4513 information of SDIO devices.
4515 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
4516 determine whether watchdog notifications are requested by
4519 * Socket-activated per-connection services now include a
4520 short description of the connection parameters in the
4523 * tmpfiles gained a new "--boot" option. When this is not used,
4524 only lines where the command character is not suffixed with
4525 "!" are executed. When this option is specified, those
4526 options are executed too. This partitions tmpfiles
4527 directives into those that can be safely executed at any
4528 time, and those which should be run only at boot (for
4529 example, a line that creates /run/nologin).
4531 * A new API "sd-resolve.h" has been added which provides a simple
4532 asynchronous wrapper around glibc NSS host name resolution
4533 calls, such as getaddrinfo(). In contrast to glibc's
4534 getaddrinfo_a(), it does not use signals. In contrast to most
4535 other asynchronous name resolution libraries, this one does
4536 not reimplement DNS, but reuses NSS, so that alternate
4537 host name resolution systems continue to work, such as mDNS,
4538 LDAP, etc. This API is based on libasyncns, but it has been
4539 cleaned up for inclusion in systemd.
4541 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
4542 "sd-daemon.h" are no longer found in individual libraries
4543 libsystemd-journal.so, libsystemd-login.so,
4544 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
4545 merged them into a single library, libsystemd.so, which
4546 provides all symbols. The reason for this is cyclic
4547 dependencies, as these libraries tend to use each other's
4548 symbols. So far, we have managed to workaround that by linking
4549 a copy of a good part of our code into each of these
4550 libraries again and again, which, however, makes certain
4551 things hard to do, like sharing static variables. Also, it
4552 substantially increases footprint. With this change, there
4553 is only one library for the basic APIs systemd
4554 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
4555 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
4556 library as well, however are subject to the --enable-kdbus
4557 switch (see below). Note that "sd-dhcp-client.h" is not part
4558 of this library (this is because it only consumes, never
4559 provides, services of/to other APIs). To make the transition
4560 easy from the separate libraries to the unified one, we
4561 provide the --enable-compat-libs compile-time switch which
4562 will generate stub libraries that are compatible with the
4563 old ones but redirect all calls to the new one.
4565 * All of the kdbus logic and the new APIs "sd-bus.h",
4566 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
4567 and "sd-utf8.h" are compile-time optional via the
4568 "--enable-kdbus" switch, and they are not compiled in by
4569 default. To make use of kdbus, you have to explicitly enable
4570 the switch. Note however, that neither the kernel nor the
4571 userspace API for all of this is considered stable yet. We
4572 want to maintain the freedom to still change the APIs for
4573 now. By specifying this build-time switch, you acknowledge
4574 that you are aware of the instability of the current
4577 * Also, note that while kdbus is pretty much complete,
4578 it lacks one thing: proper policy support. This means you
4579 can build a fully working system with all features; however,
4580 it will be highly insecure. Policy support will be added in
4581 one of the next releases, at the same time that we will
4582 declare the APIs stable.
4584 * When the kernel command line argument "kdbus" is specified,
4585 systemd will automatically load the kdbus.ko kernel module. At
4586 this stage of development, it is only useful for testing kdbus
4587 and should not be used in production. Note: if "--enable-kdbus"
4588 is specified, and the kdbus.ko kernel module is available, and
4589 "kdbus" is added to the kernel command line, the entire system
4590 runs with kdbus instead of dbus-daemon, with the above mentioned
4591 problem of missing the system policy enforcement. Also a future
4592 version of kdbus.ko or a newer systemd will not be compatible with
4593 each other, and will unlikely be able to boot the machine if only
4594 one of them is updated.
4596 * systemctl gained a new "import-environment" command which
4597 uploads the caller's environment (or parts thereof) into the
4598 service manager so that it is inherited by services started
4599 by the manager. This is useful to upload variables like
4600 $DISPLAY into the user service manager.
4602 * A new PrivateDevices= switch has been added to service units
4603 which allows running a service with a namespaced /dev
4604 directory that does not contain any device nodes for
4605 physical devices. More specifically, it only includes devices
4606 such as /dev/null, /dev/urandom, and /dev/zero which are API
4609 * logind has been extended to support behaviour like VT
4610 switching on seats that do not support a VT. This makes
4611 multi-session available on seats that are not the first seat
4612 (seat0), and on systems where kernel support for VTs has
4613 been disabled at compile-time.
4615 * If a process holds a delay lock for system sleep or shutdown
4616 and fails to release it in time, we will now log its
4617 identity. This makes it easier to identify processes that
4618 cause slow suspends or power-offs.
4620 * When parsing /etc/crypttab, support for a new key-slot=
4621 option as supported by Debian is added. It allows indicating
4622 which LUKS slot to use on disk, speeding up key loading.
4624 * The sd_journald_sendv() API call has been checked and
4625 officially declared to be async-signal-safe so that it may
4626 be invoked from signal handlers for logging purposes.
4628 * Boot-time status output is now enabled automatically after a
4629 short timeout if boot does not progress, in order to give
4630 the user an indication what she or he is waiting for.
4632 * The boot-time output has been improved to show how much time
4633 remains until jobs expire.
4635 * The KillMode= switch in service units gained a new possible
4636 value "mixed". If set, and the unit is shut down, then the
4637 initial SIGTERM signal is sent only to the main daemon
4638 process, while the following SIGKILL signal is sent to
4639 all remaining processes of the service.
4641 * When a scope unit is registered, a new property "Controller"
4642 may be set. If set to a valid bus name, systemd will send a
4643 RequestStop() signal to this name when it would like to shut
4644 down the scope. This may be used to hook manager logic into
4645 the shutdown logic of scope units. Also, scope units may now
4646 be put in a special "abandoned" state, in which case the
4647 manager process which created them takes no further
4648 responsibilities for it.
4650 * When reading unit files, systemd will now verify
4651 the access mode of these files, and warn about certain
4652 suspicious combinations. This has been added to make it
4653 easier to track down packaging bugs where unit files are
4654 marked executable or world-writable.
4656 * systemd-nspawn gained a new "--setenv=" switch to set
4657 container-wide environment variables. The similar option in
4658 systemd-activate was renamed from "--environment=" to
4659 "--setenv=" for consistency.
4661 * systemd-nspawn has been updated to create a new kdbus domain
4662 for each container that is invoked, thus allowing each
4663 container to have its own set of system and user buses,
4664 independent of the host.
4666 * systemd-nspawn gained a new --drop-capability= switch to run
4667 the container with less capabilities than the default. Both
4668 --drop-capability= and --capability= now take the special
4669 string "all" for dropping or keeping all capabilities.
4671 * systemd-nspawn gained new switches for executing containers
4672 with specific SELinux labels set.
4674 * systemd-nspawn gained a new --quiet switch to not generate
4675 any additional output but the container's own console
4678 * systemd-nspawn gained a new --share-system switch to run a
4679 container without PID namespacing enabled.
4681 * systemd-nspawn gained a new --register= switch to control
4682 whether the container is registered with systemd-machined or
4683 not. This is useful for containers that do not run full
4684 OS images, but only specific apps.
4686 * systemd-nspawn gained a new --keep-unit which may be used
4687 when invoked as the only program from a service unit, and
4688 results in registration of the unit service itself in
4689 systemd-machined, instead of a newly opened scope unit.
4691 * systemd-nspawn gained a new --network-interface= switch for
4692 moving arbitrary interfaces to the container. The new
4693 --network-veth switch creates a virtual Ethernet connection
4694 between host and container. The new --network-bridge=
4695 switch then allows assigning the host side of this virtual
4696 Ethernet connection to a bridge device.
4698 * systemd-nspawn gained a new --personality= switch for
4699 setting the kernel personality for the container. This is
4700 useful when running a 32-bit container on a 64-bit host. A
4701 similar option Personality= is now also available for service
4704 * logind will now also track a "Desktop" identifier for each
4705 session which encodes the desktop environment of it. This is
4706 useful for desktop environments that want to identify
4707 multiple running sessions of itself easily.
4709 * A new SELinuxContext= setting for service units has been
4710 added that allows setting a specific SELinux execution
4711 context for a service.
4713 * Most systemd client tools will now honour $SYSTEMD_LESS for
4714 settings of the "less" pager. By default, these tools will
4715 override $LESS to allow certain operations to work, such as
4716 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
4717 influence this logic.
4719 * systemd's "seccomp" hook-up has been changed to make use of
4720 the libseccomp library instead of using its own
4721 implementation. This has benefits for portability among
4724 * For usage together with SystemCallFilter=, a new
4725 SystemCallErrorNumber= setting has been introduced that
4726 allows configuration of a system error number to be returned
4727 on filtered system calls, instead of immediately killing the
4728 process. Also, SystemCallArchitectures= has been added to
4729 limit access to system calls of a particular architecture
4730 (in order to turn off support for unused secondary
4731 architectures). There is also a global
4732 SystemCallArchitectures= setting in system.conf now to turn
4733 off support for non-native system calls system-wide.
4735 * systemd requires a kernel with a working name_to_handle_at(),
4736 please see the kernel config requirements in the README file.
4738 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
4739 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
4740 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
4741 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
4742 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
4743 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
4744 Elia Pinto, Florian Weimer, George McCollister, Goffredo
4745 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
4746 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
4747 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
4748 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
4749 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
4750 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
4751 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
4752 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
4753 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
4754 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
4755 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
4756 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
4757 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
4758 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
4759 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
4760 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
4761 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
4763 — Berlin, 2014-02-20
4767 * logind has gained support for facilitating privileged input
4768 and drm device access for unprivileged clients. This work is
4769 useful to allow Wayland display servers (and similar
4770 programs, such as kmscon) to run under the user's ID and
4771 access input and drm devices which are normally
4772 protected. When this is used (and the kernel is new enough)
4773 logind will "mute" IO on the file descriptors passed to
4774 Wayland as long as it is in the background and "unmute" it
4775 if it returns into the foreground. This allows secure
4776 session switching without allowing background sessions to
4777 eavesdrop on input and display data. This also introduces
4778 session switching support if VT support is turned off in the
4779 kernel, and on seats that are not seat0.
4781 * A new kernel command line option luks.options= is understood
4782 now which allows specifying LUKS options for usage for LUKS
4783 encrypted partitions specified with luks.uuid=.
4785 * tmpfiles.d(5) snippets may now use specifier expansion in
4786 path names. More specifically %m, %b, %H, %v, are now
4787 replaced by the local machine id, boot id, hostname, and
4788 kernel version number.
4790 * A new tmpfiles.d(5) command "m" has been introduced which
4791 may be used to change the owner/group/access mode of a file
4792 or directory if it exists, but do nothing if it does not.
4794 * This release removes high-level support for the
4795 MemorySoftLimit= cgroup setting. The underlying kernel
4796 cgroup attribute memory.soft_limit= is currently badly
4797 designed and likely to be removed from the kernel API in its
4798 current form, hence we should not expose it for now.
4800 * The memory.use_hierarchy cgroup attribute is now enabled for
4801 all cgroups systemd creates in the memory cgroup
4802 hierarchy. This option is likely to be come the built-in
4803 default in the kernel anyway, and the non-hierarchical mode
4804 never made much sense in the intrinsically hierarchical
4807 * A new field _SYSTEMD_SLICE= is logged along with all journal
4808 messages containing the slice a message was generated
4809 from. This is useful to allow easy per-customer filtering of
4810 logs among other things.
4812 * systemd-journald will no longer adjust the group of journal
4813 files it creates to the "systemd-journal" group. Instead we
4814 rely on the journal directory to be owned by the
4815 "systemd-journal" group, and its setgid bit set, so that the
4816 kernel file system layer will automatically enforce that
4817 journal files inherit this group assignment. The reason for
4818 this change is that we cannot allow NSS look-ups from
4819 journald which would be necessary to resolve
4820 "systemd-journal" to a numeric GID, because this might
4821 create deadlocks if NSS involves synchronous queries to
4822 other daemons (such as nscd, or sssd) which in turn are
4823 logging clients of journald and might block on it, which
4824 would then dead lock. A tmpfiles.d(5) snippet included in
4825 systemd will make sure the setgid bit and group are
4826 properly set on the journal directory if it exists on every
4827 boot. However, we recommend adjusting it manually after
4828 upgrades too (or from RPM scriptlets), so that the change is
4829 not delayed until next reboot.
4831 * Backlight and random seed files in /var/lib/ have moved into
4832 the /var/lib/systemd/ directory, in order to centralize all
4833 systemd generated files in one directory.
4835 * Boot time performance measurements (as displayed by
4836 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
4837 performance information if that's available to determine how
4838 much time BIOS and boot loader initialization required. With
4839 a sufficiently new BIOS you hence no longer need to boot
4840 with Gummiboot to get access to such information.
4842 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
4843 Cristian Rodríguez, Dave Reisner, David Herrmann, David
4844 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
4845 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
4846 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
4847 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
4848 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
4850 — Berlin, 2013-10-02
4854 * The Restart= option for services now understands a new
4855 on-watchdog setting, which will restart the service
4856 automatically if the service stops sending out watchdog keep
4857 alive messages (as configured with WatchdogSec=).
4859 * The getty generator (which is responsible for bringing up a
4860 getty on configured serial consoles) will no longer only
4861 start a getty on the primary kernel console but on all
4862 others, too. This makes the order in which console= is
4863 specified on the kernel command line less important.
4865 * libsystemd-logind gained a new sd_session_get_vt() call to
4866 retrieve the VT number of a session.
4868 * If the option "tries=0" is set for an entry of /etc/crypttab
4869 its passphrase is queried indefinitely instead of any
4870 maximum number of tries.
4872 * If a service with a configure PID file terminates its PID
4873 file will now be removed automatically if it still exists
4874 afterwards. This should put an end to stale PID files.
4876 * systemd-run will now also take relative binary path names
4877 for execution and no longer insists on absolute paths.
4879 * InaccessibleDirectories= and ReadOnlyDirectories= now take
4880 paths that are optionally prefixed with "-" to indicate that
4881 it should not be considered a failure if they do not exist.
4883 * journalctl -o (and similar commands) now understands a new
4884 output mode "short-precise", it is similar to "short" but
4885 shows timestamps with usec accuracy.
4887 * The option "discard" (as known from Debian) is now
4888 synonymous to "allow-discards" in /etc/crypttab. In fact,
4889 "discard" is preferred now (since it is easier to remember
4892 * Some licensing clean-ups were made, so that more code is now
4893 LGPL-2.1 licensed than before.
4895 * A minimal tool to save/restore the display backlight
4896 brightness across reboots has been added. It will store the
4897 backlight setting as late as possible at shutdown, and
4898 restore it as early as possible during reboot.
4900 * A logic to automatically discover and enable home and swap
4901 partitions on GPT disks has been added. With this in place
4902 /etc/fstab becomes optional for many setups as systemd can
4903 discover certain partitions located on the root disk
4904 automatically. Home partitions are recognized under their
4905 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
4906 partitions are recognized under their GPT type ID
4907 0657fd6da4ab43c484e50933c84b4f4f.
4909 * systemd will no longer pass any environment from the kernel
4910 or initrd to system services. If you want to set an
4911 environment for all services, do so via the kernel command
4912 line systemd.setenv= assignment.
4914 * The systemd-sysctl tool no longer natively reads the file
4915 /etc/sysctl.conf. If desired, the file should be symlinked
4916 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
4917 legacy support by a symlink rather than built-in code, it
4918 also makes the otherwise hidden order of application of the
4919 different files visible. (Note that this partly reverts to a
4920 pre-198 application order of sysctl knobs!)
4922 * The "systemctl set-log-level" and "systemctl dump" commands
4923 have been moved to systemd-analyze.
4925 * systemd-run learned the new --remain-after-exit switch,
4926 which causes the scope unit not to be cleaned up
4927 automatically after the process terminated.
4929 * tmpfiles learned a new --exclude-prefix= switch to exclude
4930 certain paths from operation.
4932 * journald will now automatically flush all messages to disk
4933 as soon as a message at the log level CRIT, ALERT or EMERG
4936 Contributions from: Andrew Cook, Brandon Philips, Christian
4937 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
4938 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
4939 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
4940 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
4941 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
4942 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
4943 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
4944 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
4945 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
4946 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
4947 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
4948 William Giokas, Zbigniew Jędrzejewski-Szmek
4950 — Berlin, 2013-09-13
4954 * The documentation has been updated to cover the various new
4955 concepts introduced with 205.
4957 * Unit files now understand the new %v specifier which
4958 resolves to the kernel version string as returned by "uname
4961 * systemctl now supports filtering the unit list output by
4962 load state, active state and sub state, using the new
4965 * "systemctl status" will now show the results of the
4966 condition checks (like ConditionPathExists= and similar) of
4967 the last start attempts of the unit. They are also logged to
4970 * "journalctl -b" may now be used to look for boot output of a
4971 specific boot. Try "journalctl -b -1" for the previous boot,
4972 but the syntax is substantially more powerful.
4974 * "journalctl --show-cursor" has been added which prints the
4975 cursor string the last shown log line. This may then be used
4976 with the new "journalctl --after-cursor=" switch to continue
4977 browsing logs from that point on.
4979 * "journalctl --force" may now be used to force regeneration
4982 * Creation of "dead" device nodes has been moved from udev
4983 into kmod and tmpfiles. Previously, udev would read the kmod
4984 databases to pre-generate dead device nodes based on meta
4985 information contained in kernel modules, so that these would
4986 be auto-loaded on access rather then at boot. As this
4987 does not really have much to do with the exposing actual
4988 kernel devices to userspace this has always been slightly
4989 alien in the udev codebase. Following the new scheme kmod
4990 will now generate a runtime snippet for tmpfiles from the
4991 module meta information and it now is tmpfiles' job to the
4992 create the nodes. This also allows overriding access and
4993 other parameters for the nodes using the usual tmpfiles
4994 facilities. As side effect this allows us to remove the
4995 CAP_SYS_MKNOD capability bit from udevd entirely.
4997 * logind's device ACLs may now be applied to these "dead"
4998 devices nodes too, thus finally allowing managed access to
4999 devices such as /dev/snd/sequencer without loading the
5000 backing module right-away.
5002 * A new RPM macro has been added that may be used to apply
5003 tmpfiles configuration during package installation.
5005 * systemd-detect-virt and ConditionVirtualization= now can
5006 detect User-Mode-Linux machines (UML).
5008 * journald will now implicitly log the effective capabilities
5009 set of processes in the message metadata.
5011 * systemd-cryptsetup has gained support for TrueCrypt volumes.
5013 * The initrd interface has been simplified (more specifically,
5014 support for passing performance data via environment
5015 variables and fsck results via files in /run has been
5016 removed). These features were non-essential, and are
5017 nowadays available in a much nicer way by having systemd in
5018 the initrd serialize its state and have the hosts systemd
5019 deserialize it again.
5021 * The udev "keymap" data files and tools to apply keyboard
5022 specific mappings of scan to key codes, and force-release
5023 scan code lists have been entirely replaced by a udev
5024 "keyboard" builtin and a hwdb data file.
5026 * systemd will now honour the kernel's "quiet" command line
5027 argument also during late shutdown, resulting in a
5028 completely silent shutdown when used.
5030 * There's now an option to control the SO_REUSEPORT socket
5031 option in .socket units.
5033 * Instance units will now automatically get a per-template
5034 subslice of system.slice unless something else is explicitly
5035 configured. For example, instances of sshd@.service will now
5036 implicitly be placed in system-sshd.slice rather than
5037 system.slice as before.
5039 * Test coverage support may now be enabled at build time.
5041 Contributions from: Dave Reisner, Frederic Crozat, Harald
5042 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
5043 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
5044 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
5045 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
5046 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
5047 Giokas, Zbigniew Jędrzejewski-Szmek
5049 — Berlin, 2013-07-23
5053 * Two new unit types have been introduced:
5055 Scope units are very similar to service units, however, are
5056 created out of pre-existing processes — instead of PID 1
5057 forking off the processes. By using scope units it is
5058 possible for system services and applications to group their
5059 own child processes (worker processes) in a powerful way
5060 which then maybe used to organize them, or kill them
5061 together, or apply resource limits on them.
5063 Slice units may be used to partition system resources in an
5064 hierarchical fashion and then assign other units to them. By
5065 default there are now three slices: system.slice (for all
5066 system services), user.slice (for all user sessions),
5067 machine.slice (for VMs and containers).
5069 Slices and scopes have been introduced primarily in
5070 context of the work to move cgroup handling to a
5071 single-writer scheme, where only PID 1
5072 creates/removes/manages cgroups.
5074 * There's a new concept of "transient" units. In contrast to
5075 normal units these units are created via an API at runtime,
5076 not from configuration from disk. More specifically this
5077 means it is now possible to run arbitrary programs as
5078 independent services, with all execution parameters passed
5079 in via bus APIs rather than read from disk. Transient units
5080 make systemd substantially more dynamic then it ever was,
5081 and useful as a general batch manager.
5083 * logind has been updated to make use of scope and slice units
5084 for managing user sessions. As a user logs in he will get
5085 his own private slice unit, to which all sessions are added
5086 as scope units. We also added support for automatically
5087 adding an instance of user@.service for the user into the
5088 slice. Effectively logind will no longer create cgroup
5089 hierarchies on its own now, it will defer entirely to PID 1
5090 for this by means of scope, service and slice units. Since
5091 user sessions this way become entities managed by PID 1
5092 the output of "systemctl" is now a lot more comprehensive.
5094 * A new mini-daemon "systemd-machined" has been added which
5095 may be used by virtualization managers to register local
5096 VMs/containers. nspawn has been updated accordingly, and
5097 libvirt will be updated shortly. machined will collect a bit
5098 of meta information about the VMs/containers, and assign
5099 them their own scope unit (see above). The collected
5100 meta-data is then made available via the "machinectl" tool,
5101 and exposed in "ps" and similar tools. machined/machinectl
5102 is compile-time optional.
5104 * As discussed earlier, the low-level cgroup configuration
5105 options ControlGroup=, ControlGroupModify=,
5106 ControlGroupPersistent=, ControlGroupAttribute= have been
5107 removed. Please use high-level attribute settings instead as
5108 well as slice units.
5110 * A new bus call SetUnitProperties() has been added to alter
5111 various runtime parameters of a unit. This is primarily
5112 useful to alter cgroup parameters dynamically in a nice way,
5113 but will be extended later on to make more properties
5114 modifiable at runtime. systemctl gained a new set-properties
5115 command that wraps this call.
5117 * A new tool "systemd-run" has been added which can be used to
5118 run arbitrary command lines as transient services or scopes,
5119 while configuring a number of settings via the command
5120 line. This tool is currently very basic, however already
5121 very useful. We plan to extend this tool to even allow
5122 queuing of execution jobs with time triggers from the
5123 command line, similar in fashion to "at".
5125 * nspawn will now inform the user explicitly that kernels with
5126 audit enabled break containers, and suggest the user to turn
5129 * Support for detecting the IMA and AppArmor security
5130 frameworks with ConditionSecurity= has been added.
5132 * journalctl gained a new "-k" switch for showing only kernel
5133 messages, mimicking dmesg output; in addition to "--user"
5134 and "--system" switches for showing only user's own logs
5137 * systemd-delta can now show information about drop-in
5138 snippets extending unit files.
5140 * libsystemd-bus has been substantially updated but is still
5141 not available as public API.
5143 * systemd will now look for the "debug" argument on the kernel
5144 command line and enable debug logging, similar to what
5145 "systemd.log_level=debug" already did before.
5147 * "systemctl set-default", "systemctl get-default" has been
5148 added to configure the default.target symlink, which
5149 controls what to boot into by default.
5151 * "systemctl set-log-level" has been added as a convenient
5152 way to raise and lower systemd logging threshold.
5154 * "systemd-analyze plot" will now show the time the various
5155 generators needed for execution, as well as information
5156 about the unit file loading.
5158 * libsystemd-journal gained a new sd_journal_open_files() call
5159 for opening specific journal files. journactl also gained a
5160 new switch to expose this new functionality. Previously we
5161 only supported opening all files from a directory, or all
5162 files from the system, as opening individual files only is
5163 racy due to journal file rotation.
5165 * systemd gained the new DefaultEnvironment= setting in
5166 /etc/systemd/system.conf to set environment variables for
5169 * If a privileged process logs a journal message with the
5170 OBJECT_PID= field set, then journald will automatically
5171 augment this with additional OBJECT_UID=, OBJECT_GID=,
5172 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
5173 system services want to log events about specific client
5174 processes. journactl/systemctl has been updated to make use
5175 of this information if all log messages regarding a specific
5178 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
5179 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
5180 Reisner, David Coppa, David King, David Strauss, Eelco
5181 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
5182 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
5183 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
5184 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
5185 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
5186 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
5187 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
5188 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
5189 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
5190 Łukasz Stelmach, 장동준
5194 * The Python bindings gained some minimal support for the APIs
5195 exposed by libsystemd-logind.
5197 * ConditionSecurity= gained support for detecting SMACK. Since
5198 this condition already supports SELinux and AppArmor we only
5199 miss IMA for this. Patches welcome!
5201 Contributions from: Karol Lewandowski, Lennart Poettering,
5202 Zbigniew Jędrzejewski-Szmek
5206 * systemd-nspawn will now create /etc/resolv.conf if
5207 necessary, before bind-mounting the host's file onto it.
5209 * systemd-nspawn will now store meta information about a
5210 container on the container's cgroup as extended attribute
5211 fields, including the root directory.
5213 * The cgroup hierarchy has been reworked in many ways. All
5214 objects any of the components systemd creates in the cgroup
5215 tree are now suffixed. More specifically, user sessions are
5216 now placed in cgroups suffixed with ".session", users in
5217 cgroups suffixed with ".user", and nspawn containers in
5218 cgroups suffixed with ".nspawn". Furthermore, all cgroup
5219 names are now escaped in a simple scheme to avoid collision
5220 of userspace object names with kernel filenames. This work
5221 is preparation for making these objects relocatable in the
5222 cgroup tree, in order to allow easy resource partitioning of
5223 these objects without causing naming conflicts.
5225 * systemctl list-dependencies gained the new switches
5226 --plain, --reverse, --after and --before.
5228 * systemd-inhibit now shows the process name of processes that
5229 have taken an inhibitor lock.
5231 * nss-myhostname will now also resolve "localhost"
5232 implicitly. This makes /etc/hosts an optional file and
5233 nicely handles that on IPv6 ::1 maps to both "localhost" and
5236 * libsystemd-logind.so gained a new call
5237 sd_get_machine_names() to enumerate running containers and
5238 VMs (currently only supported by very new libvirt and
5239 nspawn). sd_login_monitor can now be used to watch
5240 VMs/containers coming and going.
5242 * .include is not allowed recursively anymore, and only in
5243 unit files. Usually it is better to use drop-in snippets in
5244 .d/*.conf anyway, as introduced with systemd 198.
5246 * systemd-analyze gained a new "critical-chain" command that
5247 determines the slowest chain of units run during system
5248 boot-up. It is very useful for tracking down where
5249 optimizing boot time is the most beneficial.
5251 * systemd will no longer allow manipulating service paths in
5252 the name=systemd:/system cgroup tree using ControlGroup= in
5253 units. (But is still fine with it in all other dirs.)
5255 * There's a new systemd-nspawn@.service service file that may
5256 be used to easily run nspawn containers as system
5257 services. With the container's root directory in
5258 /var/lib/container/foobar it is now sufficient to run
5259 "systemctl start systemd-nspawn@foobar.service" to boot it.
5261 * systemd-cgls gained a new parameter "--machine" to list only
5262 the processes within a certain container.
5264 * ConditionSecurity= now can check for "apparmor". We still
5265 are lacking checks for SMACK and IMA for this condition
5266 check though. Patches welcome!
5268 * A new configuration file /etc/systemd/sleep.conf has been
5269 added that may be used to configure which kernel operation
5270 systemd is supposed to execute when "suspend", "hibernate"
5271 or "hybrid-sleep" is requested. This makes the new kernel
5272 "freeze" state accessible to the user.
5274 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
5275 the passed argument if applicable.
5277 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
5278 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
5279 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
5280 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
5281 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
5282 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
5283 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
5288 * The output of 'systemctl list-jobs' got some polishing. The
5289 '--type=' argument may now be passed more than once. A new
5290 command 'systemctl list-sockets' has been added which shows
5291 a list of kernel sockets systemd is listening on with the
5292 socket units they belong to, plus the units these socket
5295 * The experimental libsystemd-bus library got substantial
5296 updates to work in conjunction with the (also experimental)
5297 kdbus kernel project. It works well enough to exchange
5298 messages with some sophistication. Note that kdbus is not
5299 ready yet, and the library is mostly an elaborate test case
5300 for now, and not installable.
5302 * systemd gained a new unit 'systemd-static-nodes.service'
5303 that generates static device nodes earlier during boot, and
5304 can run in conjunction with udev.
5306 * libsystemd-login gained a new call sd_pid_get_user_unit()
5307 to retrieve the user systemd unit a process is running
5308 in. This is useful for systems where systemd is used as
5311 * systemd-nspawn now places all containers in the new /machine
5312 top-level cgroup directory in the name=systemd
5313 hierarchy. libvirt will soon do the same, so that we get a
5314 uniform separation of /system, /user and /machine for system
5315 services, user processes and containers/virtual
5316 machines. This new cgroup hierarchy is also useful to stick
5317 stable names to specific container instances, which can be
5318 recognized later this way (this name may be controlled
5319 via systemd-nspawn's new -M switch). libsystemd-login also
5320 gained a new call sd_pid_get_machine_name() to retrieve the
5321 name of the container/VM a specific process belongs to.
5323 * bootchart can now store its data in the journal.
5325 * libsystemd-journal gained a new call
5326 sd_journal_add_conjunction() for AND expressions to the
5327 matching logic. This can be used to express more complex
5328 logical expressions.
5330 * journactl can now take multiple --unit= and --user-unit=
5333 * The cryptsetup logic now understands the "luks.key=" kernel
5334 command line switch for specifying a file to read the
5335 decryption key from. Also, if a configured key file is not
5336 found the tool will now automatically fall back to prompting
5339 * Python systemd.journal module was updated to wrap recently
5340 added functions from libsystemd-journal. The interface was
5341 changed to bring the low level interface in s.j._Reader
5342 closer to the C API, and the high level interface in
5343 s.j.Reader was updated to wrap and convert all data about
5346 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
5347 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
5348 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
5349 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
5350 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
5351 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5355 * journalctl --update-catalog now understands a new --root=
5356 option to operate on catalogs found in a different root
5359 * During shutdown after systemd has terminated all running
5360 services a final killing loop kills all remaining left-over
5361 processes. We will now print the name of these processes
5362 when we send SIGKILL to them, since this usually indicates a
5365 * If /etc/crypttab refers to password files stored on
5366 configured mount points automatic dependencies will now be
5367 generated to ensure the specific mount is established first
5368 before the key file is attempted to be read.
5370 * 'systemctl status' will now show information about the
5371 network sockets a socket unit is listening on.
5373 * 'systemctl status' will also shown information about any
5374 drop-in configuration file for units. (Drop-In configuration
5375 files in this context are files such as
5376 /etc/systemd/systemd/foobar.service.d/*.conf)
5378 * systemd-cgtop now optionally shows summed up CPU times of
5379 cgroups. Press '%' while running cgtop to switch between
5380 percentage and absolute mode. This is useful to determine
5381 which cgroups use up the most CPU time over the entire
5382 runtime of the system. systemd-cgtop has also been updated
5383 to be 'pipeable' for processing with further shell tools.
5385 * 'hostnamectl set-hostname' will now allow setting of FQDN
5388 * The formatting and parsing of time span values has been
5389 changed. The parser now understands fractional expressions
5390 such as "5.5h". The formatter will now output fractional
5391 expressions for all time spans under 1min, i.e. "5.123456s"
5392 rather than "5s 123ms 456us". For time spans under 1s
5393 millisecond values are shown, for those under 1ms
5394 microsecond values are shown. This should greatly improve
5395 all time-related output of systemd.
5397 * libsystemd-login and libsystemd-journal gained new
5398 functions for querying the poll() events mask and poll()
5399 timeout value for integration into arbitrary event
5402 * localectl gained the ability to list available X11 keymaps
5403 (models, layouts, variants, options).
5405 * 'systemd-analyze dot' gained the ability to filter for
5406 specific units via shell-style globs, to create smaller,
5407 more useful graphs. I.e. it is now possible to create simple
5408 graphs of all the dependencies between only target units, or
5409 of all units that Avahi has dependencies with.
5411 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
5412 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
5413 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
5414 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
5415 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
5416 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
5417 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
5421 * The boot-time readahead implementation for rotating media
5422 will now read the read-ahead data in multiple passes which
5423 consist of all read requests made in equidistant time
5424 intervals. This means instead of strictly reading read-ahead
5425 data in its physical order on disk we now try to find a
5426 middle ground between physical and access time order.
5428 * /etc/os-release files gained a new BUILD_ID= field for usage
5429 on operating systems that provide continuous builds of OS
5432 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
5433 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
5434 William Douglas, Zbigniew Jędrzejewski-Szmek
5438 * systemd-python gained an API exposing libsystemd-daemon.
5440 * The SMACK setup logic gained support for uploading CIPSO
5443 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
5444 ReadOnlyDirectories= and InaccessibleDirectories= has
5445 changed. The private /tmp and /var/tmp directories are now
5446 shared by all processes of a service (which means
5447 ExecStartPre= may now leave data in /tmp that ExecStart= of
5448 the same service can still access). When a service is
5449 stopped its temporary directories are immediately deleted
5450 (normal clean-up with tmpfiles is still done in addition to
5453 * By default, systemd will now set a couple of sysctl
5454 variables in the kernel: the safe sysrq options are turned
5455 on, IP route verification is turned on, and source routing
5456 disabled. The recently added hardlink and softlink
5457 protection of the kernel is turned on. These settings should
5458 be reasonably safe, and good defaults for all new systems.
5460 * The predictable network naming logic may now be turned off
5461 with a new kernel command line switch: net.ifnames=0.
5463 * A new libsystemd-bus module has been added that implements a
5464 pretty complete D-Bus client library. For details see:
5466 https://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
5468 * journald will now explicitly flush the journal files to disk
5469 at the latest 5min after each write. The file will then also
5470 be marked offline until the next write. This should increase
5471 reliability in case of a crash. The synchronization delay
5472 can be configured via SyncIntervalSec= in journald.conf.
5474 * There's a new remote-fs-setup.target unit that can be used
5475 to pull in specific services when at least one remote file
5476 system is to be mounted.
5478 * There are new targets timers.target and paths.target as
5479 canonical targets to pull user timer and path units in
5480 from. This complements sockets.target with a similar
5481 purpose for socket units.
5483 * libudev gained a new call udev_device_set_attribute_value()
5484 to set sysfs attributes of a device.
5486 * The udev daemon now sets the default number of worker
5487 processes executed in parallel based on the number of available
5488 CPUs instead of the amount of available RAM. This is supposed
5489 to provide a more reliable default and limit a too aggressive
5490 parallelism for setups with 1000s of devices connected.
5492 Contributions from: Auke Kok, Colin Walters, Cristian
5493 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
5494 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
5495 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
5496 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
5497 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
5498 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
5499 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
5500 Zbigniew Jędrzejewski-Szmek
5504 * Configuration of unit files may now be extended via drop-in
5505 files without having to edit/override the unit files
5506 themselves. More specifically, if the administrator wants to
5507 change one value for a service file foobar.service he can
5508 now do so by dropping in a configuration snippet into
5509 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
5510 will load all these snippets and apply them on top of the
5511 main unit configuration file, possibly extending or
5512 overriding its settings. Using these drop-in snippets is
5513 generally nicer than the two earlier options for changing
5514 unit files locally: copying the files from
5515 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
5516 them there; or creating a new file in /etc/systemd/system/
5517 that incorporates the original one via ".include". Drop-in
5518 snippets into these .d/ directories can be placed in any
5519 directory systemd looks for units in, and the usual
5520 overriding semantics between /usr/lib, /etc and /run apply
5523 * Most unit file settings which take lists of items can now be
5524 reset by assigning the empty string to them. For example,
5525 normally, settings such as Environment=FOO=BAR append a new
5526 environment variable assignment to the environment block,
5527 each time they are used. By assigning Environment= the empty
5528 string the environment block can be reset to empty. This is
5529 particularly useful with the .d/*.conf drop-in snippets
5530 mentioned above, since this adds the ability to reset list
5531 settings from vendor unit files via these drop-ins.
5533 * systemctl gained a new "list-dependencies" command for
5534 listing the dependencies of a unit recursively.
5536 * Inhibitors are now honored and listed by "systemctl
5537 suspend", "systemctl poweroff" (and similar) too, not only
5538 GNOME. These commands will also list active sessions by
5541 * Resource limits (as exposed by the various control group
5542 controllers) can now be controlled dynamically at runtime
5543 for all units. More specifically, you can now use a command
5544 like "systemctl set-cgroup-attr foobar.service cpu.shares
5545 2000" to alter the CPU shares a specific service gets. These
5546 settings are stored persistently on disk, and thus allow the
5547 administrator to easily adjust the resource usage of
5548 services with a few simple commands. This dynamic resource
5549 management logic is also available to other programs via the
5550 bus. Almost any kernel cgroup attribute and controller is
5553 * systemd-vconsole-setup will now copy all font settings to
5554 all allocated VTs, where it previously applied them only to
5557 * libsystemd-login gained the new sd_session_get_tty() API
5560 * This release drops support for a few legacy or
5561 distribution-specific LSB facility names when parsing init
5562 scripts: $x-display-manager, $mail-transfer-agent,
5563 $mail-transport-agent, $mail-transfer-agent, $smtp,
5564 $null. Also, the mail-transfer-agent.target unit backing
5565 this has been removed. Distributions which want to retain
5566 compatibility with this should carry the burden for
5567 supporting this themselves and patch support for these back
5568 in, if they really need to. Also, the facilities $syslog and
5569 $local_fs are now ignored, since systemd does not support
5570 early-boot LSB init scripts anymore, and these facilities
5571 are implied anyway for normal services. syslog.target has
5574 * There are new bus calls on PID1's Manager object for
5575 cancelling jobs, and removing snapshot units. Previously,
5576 both calls were only available on the Job and Snapshot
5579 * systemd-journal-gatewayd gained SSL support.
5581 * The various "environment" files, such as /etc/locale.conf
5582 now support continuation lines with a backslash ("\") as
5583 last character in the line, similarly in style (but different)
5584 to how this is supported in shells.
5586 * For normal user processes the _SYSTEMD_USER_UNIT= field is
5587 now implicitly appended to every log entry logged. systemctl
5588 has been updated to filter by this field when operating on a
5589 user systemd instance.
5591 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
5592 CAP_AUDIT_CONTROL capabilities to the capabilities set for
5593 the container. This makes it easier to boot unmodified
5594 Fedora systems in a container, which however still requires
5595 audit=0 to be passed on the kernel command line. Auditing in
5596 kernel and userspace is unfortunately still too broken in
5597 context of containers, hence we recommend compiling it out
5598 of the kernel or using audit=0. Hopefully this will be fixed
5599 one day for good in the kernel.
5601 * nspawn gained the new --bind= and --bind-ro= parameters to
5602 bind mount specific directories from the host into the
5605 * nspawn will now mount its own devpts file system instance
5606 into the container, in order not to leak pty devices from
5607 the host into the container.
5609 * systemd will now read the firmware boot time performance
5610 information from the EFI variables, if the used boot loader
5611 supports this, and takes it into account for boot performance
5612 analysis via "systemd-analyze". This is currently supported
5613 only in conjunction with Gummiboot, but could be supported
5614 by other boot loaders too. For details see:
5616 https://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
5618 * A new generator has been added that automatically mounts the
5619 EFI System Partition (ESP) to /boot, if that directory
5620 exists, is empty, and no other file system has been
5621 configured to be mounted there.
5623 * logind will now send out PrepareForSleep(false) out
5624 unconditionally, after coming back from suspend. This may be
5625 used by applications as asynchronous notification for
5626 system resume events.
5628 * "systemctl unlock-sessions" has been added, that allows
5629 unlocking the screens of all user sessions at once, similar
5630 to how "systemctl lock-sessions" already locked all users
5631 sessions. This is backed by a new D-Bus call UnlockSessions().
5633 * "loginctl seat-status" will now show the master device of a
5634 seat. (i.e. the device of a seat that needs to be around for
5635 the seat to be considered available, usually the graphics
5638 * tmpfiles gained a new "X" line type, that allows
5639 configuration of files and directories (with wildcards) that
5640 shall be excluded from automatic cleanup ("aging").
5642 * udev default rules set the device node permissions now only
5643 at "add" events, and do not change them any longer with a
5644 later "change" event.
5646 * The log messages for lid events and power/sleep keypresses
5647 now carry a message ID.
5649 * We now have a substantially larger unit test suite, but this
5650 continues to be work in progress.
5652 * udevadm hwdb gained a new --root= parameter to change the
5653 root directory to operate relative to.
5655 * logind will now issue a background sync() request to the kernel
5656 early at shutdown, so that dirty buffers are flushed to disk early
5657 instead of at the last moment, in order to optimize shutdown
5660 * A new bootctl tool has been added that is an interface for
5661 certain boot loader operations. This is currently a preview
5662 and is likely to be extended into a small mechanism daemon
5663 like timedated, localed, hostnamed, and can be used by
5664 graphical UIs to enumerate available boot options, and
5665 request boot into firmware operations.
5667 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
5668 the rest of the package. It also has been updated to work
5669 correctly in initrds.
5671 * Policykit previously has been runtime optional, and is now
5672 also compile time optional via a configure switch.
5674 * systemd-analyze has been reimplemented in C. Also "systemctl
5675 dot" has moved into systemd-analyze.
5677 * "systemctl status" with no further parameters will now print
5678 the status of all active or failed units.
5680 * Operations such as "systemctl start" can now be executed
5681 with a new mode "--irreversible" which may be used to queue
5682 operations that cannot accidentally be reversed by a later
5683 job queuing. This is by default used to make shutdown
5684 requests more robust.
5686 * The Python API of systemd now gained a new module for
5687 reading journal files.
5689 * A new tool kernel-install has been added that can install
5690 kernel images according to the Boot Loader Specification:
5692 https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
5694 * Boot time console output has been improved to provide
5695 animated boot time output for hanging jobs.
5697 * A new tool systemd-activate has been added which can be used
5698 to test socket activation with, directly from the command
5699 line. This should make it much easier to test and debug
5700 socket activation in daemons.
5702 * journalctl gained a new "--reverse" (or -r) option to show
5703 journal output in reverse order (i.e. newest line first).
5705 * journalctl gained a new "--pager-end" (or -e) option to jump
5706 to immediately jump to the end of the journal in the
5707 pager. This is only supported in conjunction with "less".
5709 * journalctl gained a new "--user-unit=" option, that works
5710 similarly to "--unit=" but filters for user units rather than
5713 * A number of unit files to ease adoption of systemd in
5714 initrds has been added. This moves some minimal logic from
5715 the various initrd implementations into systemd proper.
5717 * The journal files are now owned by a new group
5718 "systemd-journal", which exists specifically to allow access
5719 to the journal, and nothing else. Previously, we used the
5720 "adm" group for that, which however possibly covers more
5721 than just journal/log file access. This new group is now
5722 already used by systemd-journal-gatewayd to ensure this
5723 daemon gets access to the journal files and as little else
5724 as possible. Note that "make install" will also set FS ACLs
5725 up for /var/log/journal to give "adm" and "wheel" read
5726 access to it, in addition to "systemd-journal" which owns
5727 the journal files. We recommend that packaging scripts also
5728 add read access to "adm" + "wheel" to /var/log/journal, and
5729 all existing/future journal files. To normal users and
5730 administrators little changes, however packagers need to
5731 ensure to create the "systemd-journal" system group at
5732 package installation time.
5734 * The systemd-journal-gatewayd now runs as unprivileged user
5735 systemd-journal-gateway:systemd-journal-gateway. Packaging
5736 scripts need to create these system user/group at
5739 * timedated now exposes a new boolean property CanNTP that
5740 indicates whether a local NTP service is available or not.
5742 * systemd-detect-virt will now also detect xen PVs
5744 * The pstore file system is now mounted by default, if it is
5747 * In addition to the SELinux and IMA policies we will now also
5748 load SMACK policies at early boot.
5750 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
5751 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
5752 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
5753 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
5754 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
5755 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
5756 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
5757 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
5758 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
5759 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
5760 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
5761 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
5762 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
5763 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
5767 * Timer units now support calendar time events in addition to
5768 monotonic time events. That means you can now trigger a unit
5769 based on a calendar time specification such as "Thu,Fri
5770 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
5771 or fifth day of any month of the year 2013, given that it is
5772 a thursday or friday. This brings timer event support
5773 considerably closer to cron's capabilities. For details on
5774 the supported calendar time specification language see
5777 * udev now supports a number of different naming policies for
5778 network interfaces for predictable names, and a combination
5779 of these policies is now the default. Please see this wiki
5780 document for details:
5782 https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
5784 * Auke Kok's bootchart implementation has been added to the
5785 systemd tree. It is an optional component that can graph the
5786 boot in quite some detail. It is one of the best bootchart
5787 implementations around and minimal in its code and
5790 * nss-myhostname has been integrated into the systemd source
5791 tree. nss-myhostname guarantees that the local hostname
5792 always stays resolvable via NSS. It has been a weak
5793 requirement of systemd-hostnamed since a long time, and
5794 since its code is actually trivial we decided to just
5795 include it in systemd's source tree. It can be turned off
5796 with a configure switch.
5798 * The read-ahead logic is now capable of properly detecting
5799 whether a btrfs file system is on SSD or rotating media, in
5800 order to optimize the read-ahead scheme. Previously, it was
5801 only capable of detecting this on traditional file systems
5804 * In udev, additional device properties are now read from the
5805 IAB in addition to the OUI database. Also, Bluetooth company
5806 identities are attached to the devices as well.
5808 * In service files %U may be used as specifier that is
5809 replaced by the configured user name of the service.
5811 * nspawn may now be invoked without a controlling TTY. This
5812 makes it suitable for invocation as its own service. This
5813 may be used to set up a simple containerized server system
5814 using only core OS tools.
5816 * systemd and nspawn can now accept socket file descriptors
5817 when they are started for socket activation. This enables
5818 implementation of socket activated nspawn
5819 containers. i.e. think about autospawning an entire OS image
5820 when the first SSH or HTTP connection is received. We expect
5821 that similar functionality will also be added to libvirt-lxc
5824 * journalctl will now suppress ANSI color codes when
5825 presenting log data.
5827 * systemctl will no longer show control group information for
5828 a unit if the control group is empty anyway.
5830 * logind can now automatically suspend/hibernate/shutdown the
5833 * /etc/machine-info and hostnamed now also expose the chassis
5834 type of the system. This can be used to determine whether
5835 the local system is a laptop, desktop, handset or
5836 tablet. This information may either be configured by the
5837 user/vendor or is automatically determined from ACPI and DMI
5838 information if possible.
5840 * A number of PolicyKit actions are now bound together with
5841 "imply" rules. This should simplify creating UIs because
5842 many actions will now authenticate similar ones as well.
5844 * Unit files learnt a new condition ConditionACPower= which
5845 may be used to conditionalize a unit depending on whether an
5846 AC power source is connected or not, of whether the system
5847 is running on battery power.
5849 * systemctl gained a new "is-failed" verb that may be used in
5850 shell scripts and suchlike to check whether a specific unit
5851 is in the "failed" state.
5853 * The EnvironmentFile= setting in unit files now supports file
5854 globbing, and can hence be used to easily read a number of
5855 environment files at once.
5857 * systemd will no longer detect and recognize specific
5858 distributions. All distribution-specific #ifdeffery has been
5859 removed, systemd is now fully generic and
5860 distribution-agnostic. Effectively, not too much is lost as
5861 a lot of the code is still accessible via explicit configure
5862 switches. However, support for some distribution specific
5863 legacy configuration file formats has been dropped. We
5864 recommend distributions to simply adopt the configuration
5865 files everybody else uses now and convert the old
5866 configuration from packaging scripts. Most distributions
5867 already did that. If that's not possible or desirable,
5868 distributions are welcome to forward port the specific
5869 pieces of code locally from the git history.
5871 * When logging a message about a unit systemd will now always
5872 log the unit name in the message meta data.
5874 * localectl will now also discover system locale data that is
5875 not stored in locale archives, but directly unpacked.
5877 * logind will no longer unconditionally use framebuffer
5878 devices as seat masters, i.e. as devices that are required
5879 to be existing before a seat is considered preset. Instead,
5880 it will now look for all devices that are tagged as
5881 "seat-master" in udev. By default, framebuffer devices will
5882 be marked as such, but depending on local systems, other
5883 devices might be marked as well. This may be used to
5884 integrate graphics cards using closed source drivers (such
5885 as NVidia ones) more nicely into logind. Note however, that
5886 we recommend using the open source NVidia drivers instead,
5887 and no udev rules for the closed-source drivers will be
5888 shipped from us upstream.
5890 Contributions from: Adam Williamson, Alessandro Crismani, Auke
5891 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
5892 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
5893 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
5894 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
5895 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
5896 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
5897 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
5898 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
5899 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
5900 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
5901 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
5906 * udev gained support for loading additional device properties
5907 from an indexed database that is keyed by vendor/product IDs
5908 and similar device identifiers. For the beginning this
5909 "hwdb" is populated with data from the well-known PCI and
5910 USB database, but also includes PNP, ACPI and OID data. In
5911 the longer run this indexed database shall grow into
5912 becoming the one central database for non-essential
5913 userspace device metadata. Previously, data from the PCI/USB
5914 database was only attached to select devices, since the
5915 lookup was a relatively expensive operation due to O(n) time
5916 complexity (with n being the number of entries in the
5917 database). Since this is now O(1), we decided to add in this
5918 data for all devices where this is available, by
5919 default. Note that the indexed database needs to be rebuilt
5920 when new data files are installed. To achieve this you need
5921 to update your packaging scripts to invoke "udevadm hwdb
5922 --update" after installation of hwdb data files. For
5923 RPM-based distributions we introduced the new
5924 %udev_hwdb_update macro for this purpose.
5926 * The Journal gained support for the "Message Catalog", an
5927 indexed database to link up additional information with
5928 journal entries. For further details please check:
5930 https://www.freedesktop.org/wiki/Software/systemd/catalog
5932 The indexed message catalog database also needs to be
5933 rebuilt after installation of message catalog files. Use
5934 "journalctl --update-catalog" for this. For RPM-based
5935 distributions we introduced the %journal_catalog_update
5936 macro for this purpose.
5938 * The Python Journal bindings gained support for the standard
5939 Python logging framework.
5941 * The Journal API gained new functions for checking whether
5942 the underlying file system of a journal file is capable of
5943 properly reporting file change notifications, or whether
5944 applications that want to reflect journal changes "live"
5945 need to recheck journal files continuously in appropriate
5948 * It is now possible to set the "age" field for tmpfiles
5949 entries to 0, indicating that files matching this entry
5950 shall always be removed when the directories are cleaned up.
5952 * coredumpctl gained a new "gdb" verb which invokes gdb
5953 right-away on the selected coredump.
5955 * There's now support for "hybrid sleep" on kernels that
5956 support this, in addition to "suspend" and "hibernate". Use
5957 "systemctl hybrid-sleep" to make use of this.
5959 * logind's HandleSuspendKey= setting (and related settings)
5960 now gained support for a new "lock" setting to simply
5961 request the screen lock on all local sessions, instead of
5962 actually executing a suspend or hibernation.
5964 * systemd will now mount the EFI variables file system by
5967 * Socket units now gained support for configuration of the
5968 SMACK security label.
5970 * timedatectl will now output the time of the last and next
5971 daylight saving change.
5973 * We dropped support for various legacy and distro-specific
5974 concepts, such as insserv, early-boot SysV services
5975 (i.e. those for non-standard runlevels such as 'b' or 'S')
5976 or ArchLinux /etc/rc.conf support. We recommend the
5977 distributions who still need support this to either continue
5978 to maintain the necessary patches downstream, or find a
5979 different solution. (Talk to us if you have questions!)
5981 * Various systemd components will now bypass PolicyKit checks
5982 for root and otherwise handle properly if PolicyKit is not
5983 found to be around. This should fix most issues for
5984 PolicyKit-less systems. Quite frankly this should have been
5985 this way since day one. It is absolutely our intention to
5986 make systemd work fine on PolicyKit-less systems, and we
5987 consider it a bug if something does not work as it should if
5988 PolicyKit is not around.
5990 * For embedded systems it is now possible to build udev and
5991 systemd without blkid and/or kmod support.
5993 * "systemctl switch-root" is now capable of switching root
5994 more than once. I.e. in addition to transitions from the
5995 initrd to the host OS it is now possible to transition to
5996 further OS images from the host. This is useful to implement
5997 offline updating tools.
5999 * Various other additions have been made to the RPM macros
6000 shipped with systemd. Use %udev_rules_update() after
6001 installing new udev rules files. %_udevhwdbdir,
6002 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
6003 %_sysctldir are now available which resolve to the right
6004 directories for packages to place various data files in.
6006 * journalctl gained the new --full switch (in addition to
6007 --all, to disable ellipsation for long messages.
6009 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
6010 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
6011 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
6012 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
6013 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
6014 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
6015 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
6016 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
6017 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
6021 * journalctl gained new --since= and --until= switches to
6022 filter by time. It also now supports nice filtering for
6023 units via --unit=/-u.
6025 * Type=oneshot services may use ExecReload= and do the
6028 * The journal daemon now supports time-based rotation and
6029 vacuuming, in addition to the usual disk-space based
6032 * The journal will now index the available field values for
6033 each field name. This enables clients to show pretty drop
6034 downs of available match values when filtering. The bash
6035 completion of journalctl has been updated
6036 accordingly. journalctl gained a new switch -F to list all
6037 values a certain field takes in the journal database.
6039 * More service events are now written as structured messages
6040 to the journal, and made recognizable via message IDs.
6042 * The timedated, localed and hostnamed mini-services which
6043 previously only provided support for changing time, locale
6044 and hostname settings from graphical DEs such as GNOME now
6045 also have a minimal (but very useful) text-based client
6046 utility each. This is probably the nicest way to changing
6047 these settings from the command line now, especially since
6048 it lists available options and is fully integrated with bash
6051 * There's now a new tool "systemd-coredumpctl" to list and
6052 extract coredumps from the journal.
6054 * We now install a README each in /var/log/ and
6055 /etc/rc.d/init.d explaining where the system logs and init
6056 scripts went. This hopefully should help folks who go to
6057 that dirs and look into the otherwise now empty void and
6058 scratch their heads.
6060 * When user-services are invoked (by systemd --user) the
6061 $MANAGERPID env var is set to the PID of systemd.
6063 * SIGRTMIN+24 when sent to a --user instance will now result
6064 in immediate termination of systemd.
6066 * gatewayd received numerous feature additions such as a
6067 "follow" mode, for live syncing and filtering.
6069 * browse.html now allows filtering and showing detailed
6070 information on specific entries. Keyboard navigation and
6071 mouse screen support has been added.
6073 * gatewayd/journalctl now supports HTML5/JSON
6074 Server-Sent-Events as output.
6076 * The SysV init script compatibility logic will now
6077 heuristically determine whether a script supports the
6078 "reload" verb, and only then make this available as
6081 * "systemctl status --follow" has been removed, use "journalctl
6084 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
6085 have been removed since they are hardly useful to be
6088 * And I'd like to take the opportunity to specifically mention
6089 Zbigniew for his great contributions. Zbigniew, you rock!
6091 Contributions from: Andrew Eikum, Christian Hesse, Colin
6092 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
6093 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
6094 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
6095 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
6096 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
6097 Jędrzejewski-Szmek, Сковорода Никита Андреевич
6101 * If /etc/vconsole.conf is non-existent or empty we will no
6102 longer load any console font or key map at boot by
6103 default. Instead the kernel defaults will be left
6104 intact. This is definitely the right thing to do, as no
6105 configuration should mean no configuration, and hard-coding
6106 font names that are different on all archs is probably a bad
6107 idea. Also, the kernel default key map and font should be
6108 good enough for most cases anyway, and mostly identical to
6109 the userspace fonts/key maps we previously overloaded them
6110 with. If distributions want to continue to default to a
6111 non-kernel font or key map they should ship a default
6112 /etc/vconsole.conf with the appropriate contents.
6114 Contributions from: Colin Walters, Daniel J Walsh, Dave
6115 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
6116 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6120 * journalctl gained a new --cursor= switch to show entries
6121 starting from the specified location in the journal.
6123 * We now enforce a size limit on journal entry fields exported
6124 with "-o json" in journalctl. Fields larger than 4K will be
6125 assigned null. This can be turned off with --all.
6127 * An (optional) journal gateway daemon is now available as
6128 "systemd-journal-gatewayd.service". This service provides
6129 access to the journal via HTTP and JSON. This functionality
6130 will be used to implement live log synchronization in both
6131 pull and push modes, but has various other users too, such
6132 as easy log access for debugging of embedded devices. Right
6133 now it is already useful to retrieve the journal via HTTP:
6135 # systemctl start systemd-journal-gatewayd.service
6136 # wget http://localhost:19531/entries
6138 This will download the journal contents in a
6139 /var/log/messages compatible format. The same as JSON:
6141 # curl -H"Accept: application/json" http://localhost:19531/entries
6143 This service is also accessible via a web browser where a
6144 single static HTML5 app is served that uses the JSON logic
6145 to enable the user to do some basic browsing of the
6146 journal. This will be extended later on. Here's an example
6147 screenshot of this app in its current state:
6149 http://0pointer.de/public/journal-gatewayd
6151 Contributions from: Kay Sievers, Lennart Poettering, Robert
6152 Milasan, Tom Gundersen
6156 * The bash completion logic is now available for journalctl
6159 * We do not mount the "cpuset" controller anymore together with
6160 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
6161 started if no parameters are assigned to it. "cpuset" hence
6162 broke code that assumed it could create "cpu" groups and
6165 * journalctl -f will now subscribe to terminal size changes,
6166 and line break accordingly.
6168 Contributions from: Dave Reisner, Kay Sievers, Lennart
6169 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
6173 * nspawn will now create a symlink /etc/localtime in the
6174 container environment, copying the host's timezone
6175 setting. Previously this has been done via a bind mount, but
6176 since symlinks cannot be bind mounted this has now been
6177 changed to create/update the appropriate symlink.
6179 * journalctl -n's line number argument is now optional, and
6180 will default to 10 if omitted.
6182 * journald will now log the maximum size the journal files may
6183 take up on disk. This is particularly useful if the default
6184 built-in logic of determining this parameter from the file
6185 system size is used. Use "systemctl status
6186 systemd-journald.service" to see this information.
6188 * The multi-seat X wrapper tool has been stripped down. As X
6189 is now capable of enumerating graphics devices via udev in a
6190 seat-aware way the wrapper is not strictly necessary
6191 anymore. A stripped down temporary stop-gap is still shipped
6192 until the upstream display managers have been updated to
6193 fully support the new X logic. Expect this wrapper to be
6194 removed entirely in one of the next releases.
6196 * HandleSleepKey= in logind.conf has been split up into
6197 HandleSuspendKey= and HandleHibernateKey=. The old setting
6198 is not available anymore. X11 and the kernel are
6199 distinguishing between these keys and we should too. This
6200 also means the inhibition lock for these keys has been split
6203 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
6204 Poettering, Lukas Nykryn, Václav Pavlín
6208 * Whenever a unit changes state we will now log this to the
6209 journal and show along the unit's own log output in
6212 * ConditionPathIsMountPoint= can now properly detect bind
6213 mount points too. (Previously, a bind mount of one file
6214 system to another place in the same file system could not be
6215 detected as mount, since they shared struct stat's st_dev
6218 * We will now mount the cgroup controllers cpu, cpuacct,
6219 cpuset and the controllers net_cls, net_prio together by
6222 * nspawn containers will now have a virtualized boot
6223 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
6224 over with a randomized ID at container initialization). This
6225 has the effect of making "journalctl -b" do the right thing
6228 * The JSON output journal serialization has been updated not
6229 to generate "endless" list objects anymore, but rather one
6230 JSON object per line. This is more in line how most JSON
6231 parsers expect JSON objects. The new output mode
6232 "json-pretty" has been added to provide similar output, but
6233 neatly aligned for readability by humans.
6235 * We dropped all explicit sync() invocations in the shutdown
6236 code. The kernel does this implicitly anyway in the kernel
6237 reboot() syscall. halt(8)'s -n option is now a compatibility
6240 * We now support virtualized reboot() in containers, as
6241 supported by newer kernels. We will fall back to exit() if
6242 CAP_SYS_REBOOT is not available to the container. Also,
6243 nspawn makes use of this now and will actually reboot the
6244 container if the containerized OS asks for that.
6246 * journalctl will only show local log output by default
6247 now. Use --merge (-m) to show remote log output, too.
6249 * libsystemd-journal gained the new sd_journal_get_usage()
6250 call to determine the current disk usage of all journal
6251 files. This is exposed in the new "journalctl --disk-usage"
6254 * journald gained a new configuration setting SplitMode= in
6255 journald.conf which may be used to control how user journals
6256 are split off. See journald.conf(5) for details.
6258 * A new condition type ConditionFileNotEmpty= has been added.
6260 * tmpfiles' "w" lines now support file globbing, to write
6261 multiple files at once.
6263 * We added Python bindings for the journal submission
6264 APIs. More Python APIs for a number of selected APIs will
6265 likely follow. Note that we intend to add native bindings
6266 only for the Python language, as we consider it common
6267 enough to deserve bindings shipped within systemd. There are
6268 various projects outside of systemd that provide bindings
6269 for languages such as PHP or Lua.
6271 * Many conditions will now resolve specifiers such as %i. In
6272 addition, PathChanged= and related directives of .path units
6273 now support specifiers as well.
6275 * There's now a new RPM macro definition for the system preset
6278 * journald will now warn if it ca not forward a message to the
6279 syslog daemon because its socket is full.
6281 * timedated will no longer write or process /etc/timezone,
6282 except on Debian. As we do not support late mounted /usr
6283 anymore /etc/localtime always being a symlink is now safe,
6284 and hence the information in /etc/timezone is not necessary
6287 * logind will now always reserve one VT for a text getty (VT6
6288 by default). Previously if more than 6 X sessions where
6289 started they took up all the VTs with auto-spawned gettys,
6290 so that no text gettys were available anymore.
6292 * udev will now automatically inform the btrfs kernel logic
6293 about btrfs RAID components showing up. This should make
6294 simple hotplug based btrfs RAID assembly work.
6296 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
6297 (but not for its children which will stay at the kernel
6298 default). This should allow setups with a lot more listening
6301 * systemd will now always pass the configured timezone to the
6302 kernel at boot. timedated will do the same when the timezone
6305 * logind's inhibition logic has been updated. By default,
6306 logind will now handle the lid switch, the power and sleep
6307 keys all the time, even in graphical sessions. If DEs want
6308 to handle these events on their own they should take the new
6309 handle-power-key, handle-sleep-key and handle-lid-switch
6310 inhibitors during their runtime. A simple way to achieve
6311 that is to invoke the DE wrapped in an invocation of:
6313 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
6315 * Access to unit operations is now checked via SELinux taking
6316 the unit file label and client process label into account.
6318 * systemd will now notify the administrator in the journal
6319 when he over-mounts a non-empty directory.
6321 * There are new specifiers that are resolved in unit files,
6322 for the host name (%H), the machine ID (%m) and the boot ID
6325 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
6326 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
6327 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
6328 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
6329 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
6330 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
6331 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
6335 * Support for reading structured kernel messages from
6336 /dev/kmsg has now been added and is enabled by default.
6338 * Support for reading kernel messages from /proc/kmsg has now
6339 been removed. If you want kernel messages in the journal
6340 make sure to run a recent kernel (>= 3.5) that supports
6341 reading structured messages from /dev/kmsg (see
6342 above). /proc/kmsg is now exclusive property of classic
6343 syslog daemons again.
6345 * The libudev API gained the new
6346 udev_device_new_from_device_id() call.
6348 * The logic for file system namespace (ReadOnlyDirectory=,
6349 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
6350 require pivot_root() anymore. This means fewer temporary
6351 directories are created below /tmp for this feature.
6353 * nspawn containers will now see and receive all submounts
6354 made on the host OS below the root file system of the
6357 * Forward Secure Sealing is now supported for Journal files,
6358 which provide cryptographical sealing of journal files so
6359 that attackers cannot alter log history anymore without this
6360 being detectable. Lennart will soon post a blog story about
6361 this explaining it in more detail.
6363 * There are two new service settings RestartPreventExitStatus=
6364 and SuccessExitStatus= which allow configuration of exit
6365 status (exit code or signal) which will be excepted from the
6366 restart logic, resp. consider successful.
6368 * journalctl gained the new --verify switch that can be used
6369 to check the integrity of the structure of journal files and
6370 (if Forward Secure Sealing is enabled) the contents of
6373 * nspawn containers will now be run with /dev/stdin, /dev/fd/
6374 and similar symlinks pre-created. This makes running shells
6375 as container init process a lot more fun.
6377 * The fstab support can now handle PARTUUID= and PARTLABEL=
6380 * A new ConditionHost= condition has been added to match
6381 against the hostname (with globs) and machine ID. This is
6382 useful for clusters where a single OS image is used to
6383 provision a large number of hosts which shall run slightly
6384 different sets of services.
6386 * Services which hit the restart limit will now be placed in a
6389 Contributions from: Bertram Poettering, Dave Reisner, Huang
6390 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
6391 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
6395 * When running in --user mode systemd will now become a
6396 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
6397 tree a lot more organized.
6399 * A new PartOf= unit dependency type has been introduced that
6400 may be used to group services in a natural way.
6402 * "systemctl enable" may now be used to enable instances of
6405 * journalctl now prints error log levels in red, and
6406 warning/notice log levels in bright white. It also supports
6407 filtering by log level now.
6409 * cgtop gained a new -n switch (similar to top), to configure
6410 the maximum number of iterations to run for. It also gained
6411 -b, to run in batch mode (accepting no input).
6413 * The suffix ".service" may now be omitted on most systemctl
6414 command lines involving service unit names.
6416 * There's a new bus call in logind to lock all sessions, as
6417 well as a loginctl verb for it "lock-sessions".
6419 * libsystemd-logind.so gained a new call sd_journal_perror()
6420 that works similar to libc perror() but logs to the journal
6421 and encodes structured information about the error number.
6423 * /etc/crypttab entries now understand the new keyfile-size=
6426 * shutdown(8) now can send a (configurable) wall message when
6427 a shutdown is cancelled.
6429 * The mount propagation mode for the root file system will now
6430 default to "shared", which is useful to make containers work
6431 nicely out-of-the-box so that they receive new mounts from
6432 the host. This can be undone locally by running "mount
6433 --make-rprivate /" if needed.
6435 * The prefdm.service file has been removed. Distributions
6436 should maintain this unit downstream if they intend to keep
6437 it around. However, we recommend writing normal unit files
6438 for display managers instead.
6440 * Since systemd is a crucial part of the OS we will now
6441 default to a number of compiler switches that improve
6442 security (hardening) such as read-only relocations, stack
6443 protection, and suchlike.
6445 * The TimeoutSec= setting for services is now split into
6446 TimeoutStartSec= and TimeoutStopSec= to allow configuration
6447 of individual time outs for the start and the stop phase of
6450 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
6451 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
6452 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
6453 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
6454 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
6455 Gundersen, Zbigniew Jędrzejewski-Szmek
6459 * The journal and id128 C APIs are now fully documented as man
6462 * Extra safety checks have been added when transitioning from
6463 the initial RAM disk to the main system to avoid accidental
6466 * /etc/crypttab entries now understand the new keyfile-offset=
6469 * systemctl -t can now be used to filter by unit load state.
6471 * The journal C API gained the new sd_journal_wait() call to
6472 make writing synchronous journal clients easier.
6474 * journalctl gained the new -D switch to show journals from a
6477 * journalctl now displays a special marker between log
6478 messages of two different boots.
6480 * The journal is now explicitly flushed to /var via a service
6481 systemd-journal-flush.service, rather than implicitly simply
6482 by seeing /var/log/journal to be writable.
6484 * journalctl (and the journal C APIs) can now match for much
6485 more complex expressions, with alternatives and
6488 * When transitioning from the initial RAM disk to the main
6489 system we will now kill all processes in a killing spree to
6490 ensure no processes stay around by accident.
6492 * Three new specifiers may be used in unit files: %u, %h, %s
6493 resolve to the user name, user home directory resp. user
6494 shell. This is useful for running systemd user instances.
6496 * We now automatically rotate journal files if their data
6497 object hash table gets a fill level > 75%. We also size the
6498 hash table based on the configured maximum file size. This
6499 together should lower hash collisions drastically and thus
6500 speed things up a bit.
6502 * journalctl gained the new "--header" switch to introspect
6503 header data of journal files.
6505 * A new setting SystemCallFilters= has been added to services
6506 which may be used to apply blacklists or whitelists to
6507 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
6509 * nspawn gained a new --link-journal= switch (and quicker: -j)
6510 to link the container journal with the host. This makes it
6511 very easy to centralize log viewing on the host for all
6512 guests while still keeping the journal files separated.
6514 * Many bugfixes and optimizations
6516 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
6517 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
6518 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
6523 * Several tools now understand kernel command line arguments,
6524 which are only read when run in an initial RAM disk. They
6525 usually follow closely their normal counterparts, but are
6528 * There's a new tool to analyze the readahead files that are
6529 automatically generated at boot. Use:
6531 /usr/lib/systemd/systemd-readahead analyze /.readahead
6533 * We now provide an early debug shell on tty9 if this enabled. Use:
6535 systemctl enable debug-shell.service
6537 * All plymouth related units have been moved into the Plymouth
6538 package. Please make sure to upgrade your Plymouth version
6541 * systemd-tmpfiles now supports getting passed the basename of
6542 a configuration file only, in which case it will look for it
6543 in all appropriate directories automatically.
6545 * udevadm info now takes a /dev or /sys path as argument, and
6546 does the right thing. Example:
6548 udevadm info /dev/sda
6549 udevadm info /sys/class/block/sda
6551 * systemctl now prints a warning if a unit is stopped but a
6552 unit that might trigger it continues to run. Example: a
6553 service is stopped but the socket that activates it is left
6556 * "systemctl status" will now mention if the log output was
6557 shortened due to rotation since a service has been started.
6559 * The journal API now exposes functions to determine the
6560 "cutoff" times due to rotation.
6562 * journald now understands SIGUSR1 and SIGUSR2 for triggering
6563 immediately flushing of runtime logs to /var if possible,
6564 resp. for triggering immediate rotation of the journal
6567 * It is now considered an error if a service is attempted to
6568 be stopped that is not loaded.
6570 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
6572 * systemd-analyze now supports Python 3
6574 * tmpfiles now supports cleaning up directories via aging
6575 where the first level dirs are always kept around but
6576 directories beneath it automatically aged. This is enabled
6577 by prefixing the age field with '~'.
6579 * Seat objects now expose CanGraphical, CanTTY properties
6580 which is required to deal with very fast bootups where the
6581 display manager might be running before the graphics drivers
6582 completed initialization.
6584 * Seat objects now expose a State property.
6586 * We now include RPM macros for service enabling/disabling
6587 based on the preset logic. We recommend RPM based
6588 distributions to make use of these macros if possible. This
6589 makes it simpler to reuse RPM spec files across
6592 * We now make sure that the collected systemd unit name is
6593 always valid when services log to the journal via
6596 * There's a new man page kernel-command-line(7) detailing all
6597 command line options we understand.
6599 * The fstab generator may now be disabled at boot by passing
6600 fstab=0 on the kernel command line.
6602 * A new kernel command line option modules-load= is now understood
6603 to load a specific kernel module statically, early at boot.
6605 * Unit names specified on the systemctl command line are now
6606 automatically escaped as needed. Also, if file system or
6607 device paths are specified they are automatically turned
6608 into the appropriate mount or device unit names. Example:
6610 systemctl status /home
6611 systemctl status /dev/sda
6613 * The SysVConsole= configuration option has been removed from
6614 system.conf parsing.
6616 * The SysV search path is no longer exported on the D-Bus
6619 * The Names= option has been removed from unit file parsing.
6621 * There's a new man page bootup(7) detailing the boot process.
6623 * Every unit and every generator we ship with systemd now
6624 comes with full documentation. The self-explanatory boot is
6627 * A couple of services gained "systemd-" prefixes in their
6628 name if they wrap systemd code, rather than only external
6629 code. Among them fsck@.service which is now
6630 systemd-fsck@.service.
6632 * The HaveWatchdog property has been removed from the D-Bus
6635 * systemd.confirm_spawn= on the kernel command line should now
6638 * There's a new man page crypttab(5) which details all options
6639 we actually understand.
6641 * systemd-nspawn gained a new --capability= switch to pass
6642 additional capabilities to the container.
6644 * timedated will now read known NTP implementation unit names
6645 from /usr/lib/systemd/ntp-units.d/*.list,
6646 systemd-timedated-ntp.target has been removed.
6648 * journalctl gained a new switch "-b" that lists log data of
6649 the current boot only.
6651 * The notify socket is in the abstract namespace again, in
6652 order to support daemons which chroot() at start-up.
6654 * There is a new Storage= configuration option for journald
6655 which allows configuration of where log data should go. This
6656 also provides a way to disable journal logging entirely, so
6657 that data collected is only forwarded to the console, the
6658 kernel log buffer or another syslog implementation.
6660 * Many bugfixes and optimizations
6662 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
6663 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
6664 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
6665 Shawn Landden, Tom Gundersen
6669 * "systemctl help <unit>" now shows the man page if one is
6672 * Several new man pages have been added.
6674 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
6675 MaxLevelConsole= can now be specified in
6676 journald.conf. These options allow reducing the amount of
6677 data stored on disk or forwarded by the log level.
6679 * TimerSlackNSec= can now be specified in system.conf for
6680 PID1. This allows system-wide power savings.
6682 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
6683 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
6688 * logind is now capable of (optionally) handling power and
6689 sleep keys as well as the lid switch.
6691 * journalctl now understands the syntax "journalctl
6692 /usr/bin/avahi-daemon" to get all log output of a specific
6695 * CapabilityBoundingSet= in system.conf now also influences
6696 the capability bound set of usermode helpers of the kernel.
6698 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
6699 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
6700 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
6701 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
6705 * Note that we skipped 139 releases here in order to set the
6706 new version to something that is greater than both udev's
6707 and systemd's most recent version number.
6709 * udev: all udev sources are merged into the systemd source tree now.
6710 All future udev development will happen in the systemd tree. It
6711 is still fully supported to use the udev daemon and tools without
6712 systemd running, like in initramfs or other init systems. Building
6713 udev though, will require the *build* of the systemd tree, but
6714 udev can be properly *run* without systemd.
6716 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
6717 should be used to create dead device nodes as workarounds for broken
6720 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
6721 no longer supported. udev_monitor_new_from_netlink() needs to be
6722 used to subscribe to events.
6724 * udev: when udevd is started by systemd, processes which are left
6725 behind by forking them off of udev rules, are unconditionally cleaned
6726 up and killed now after the event handling has finished. Services or
6727 daemons must be started as systemd services. Services can be
6728 pulled-in by udev to get started, but they can no longer be directly
6729 forked by udev rules.
6731 * udev: the daemon binary is called systemd-udevd now and installed
6732 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
6733 to adapt to that, create symlink, or rename the binary after building
6736 * libudev no longer provides these symbols:
6737 udev_monitor_from_socket()
6738 udev_queue_get_failed_list_entry()
6739 udev_get_{dev,sys,run}_path()
6740 The versions number was bumped and symbol versioning introduced.
6742 * systemd-loginctl and systemd-journalctl have been renamed
6743 to loginctl and journalctl to match systemctl.
6745 * The config files: /etc/systemd/systemd-logind.conf and
6746 /etc/systemd/systemd-journald.conf have been renamed to
6747 logind.conf and journald.conf. Package updates should rename
6748 the files to the new names on upgrade.
6750 * For almost all files the license is now LGPL2.1+, changed
6751 from the previous GPL2.0+. Exceptions are some minor stuff
6752 of udev (which will be changed to LGPL2.1 eventually, too),
6753 and the MIT licensed sd-daemon.[ch] library that is suitable
6754 to be used as drop-in files.
6756 * systemd and logind now handle system sleep states, in
6757 particular suspending and hibernating.
6759 * logind now implements a sleep/shutdown/idle inhibiting logic
6760 suitable for a variety of uses. Soonishly Lennart will blog
6761 about this in more detail.
6763 * var-run.mount and var-lock.mount are no longer provided
6764 (which previously bind mounted these directories to their new
6765 places). Distributions which have not converted these
6766 directories to symlinks should consider stealing these files
6767 from git history and add them downstream.
6769 * We introduced the Documentation= field for units and added
6770 this to all our shipped units. This is useful to make it
6771 easier to explore the boot and the purpose of the various
6774 * All smaller setup units (such as
6775 systemd-vconsole-setup.service) now detect properly if they
6776 are run in a container and are skipped when
6777 appropriate. This guarantees an entirely noise-free boot in
6778 Linux container environments such as systemd-nspawn.
6780 * A framework for implementing offline system updates is now
6781 integrated, for details see:
6782 https://www.freedesktop.org/wiki/Software/systemd/SystemUpdates
6784 * A new service type Type=idle is available now which helps us
6785 avoiding ugly interleaving of getty output and boot status
6788 * There's now a system-wide CapabilityBoundingSet= option to
6789 globally reduce the set of capabilities for the
6790 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
6791 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
6792 even CAP_NET_ADMIN system-wide for secure systems.
6794 * There are now system-wide DefaultLimitXXX= options to
6795 globally change the defaults of the various resource limits
6796 for all units started by PID 1.
6798 * Harald Hoyer's systemd test suite has been integrated into
6799 systemd which allows easy testing of systemd builds in qemu
6800 and nspawn. (This is really awesome! Ask us for details!)
6802 * The fstab parser is now implemented as generator, not inside
6805 * systemctl will now warn you if .mount units generated from
6806 /etc/fstab are out of date due to changes in fstab that
6807 have not been read by systemd yet.
6809 * systemd is now suitable for usage in initrds. Dracut has
6810 already been updated to make use of this. With this in place
6811 initrds get a slight bit faster but primarily are much
6812 easier to introspect and debug since "systemctl status" in
6813 the host system can be used to introspect initrd services,
6814 and the journal from the initrd is kept around too.
6816 * systemd-delta has been added, a tool to explore differences
6817 between user/admin configuration and vendor defaults.
6819 * PrivateTmp= now affects both /tmp and /var/tmp.
6821 * Boot time status messages are now much prettier and feature
6822 proper english language. Booting up systemd has never been
6825 * Read-ahead pack files now include the inode number of all
6826 files to pre-cache. When the inode changes the pre-caching
6827 is not attempted. This should be nicer to deal with updated
6828 packages which might result in changes of read-ahead
6831 * We now temporaritly lower the kernel's read_ahead_kb variable
6832 when collecting read-ahead data to ensure the kernel's
6833 built-in read-ahead does not add noise to our measurements
6834 of necessary blocks to pre-cache.
6836 * There's now RequiresMountsFor= to add automatic dependencies
6837 for all mounts necessary for a specific file system path.
6839 * MountAuto= and SwapAuto= have been removed from
6840 system.conf. Mounting file systems at boot has to take place
6843 * nspawn now learned a new switch --uuid= to set the machine
6844 ID on the command line.
6846 * nspawn now learned the -b switch to automatically search
6849 * vt102 is now the default TERM for serial TTYs, upgraded from
6852 * systemd-logind now works on VT-less systems.
6854 * The build tree has been reorganized. The individual
6855 components now have directories of their own.
6857 * A new condition type ConditionPathIsReadWrite= is now available.
6859 * nspawn learned the new -C switch to create cgroups for the
6860 container in other hierarchies.
6862 * We now have support for hardware watchdogs, configurable in
6865 * The scheduled shutdown logic now has a public API.
6867 * We now mount /tmp as tmpfs by default, but this can be
6868 masked and /etc/fstab can override it.
6870 * Since udisks does not make use of /media anymore we are not
6871 mounting a tmpfs on it anymore.
6873 * journalctl gained a new --local switch to only interleave
6874 locally generated journal files.
6876 * We can now load the IMA policy at boot automatically.
6878 * The GTK tools have been split off into a systemd-ui.
6880 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
6881 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
6882 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
6883 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
6884 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
6885 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
6886 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
6887 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
6888 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
6893 * This is mostly a bugfix release
6895 * Support optional initialization of the machine ID from the
6896 KVM or container configured UUID.
6898 * Support immediate reboots with "systemctl reboot -ff"
6900 * Show /etc/os-release data in systemd-analyze output
6902 * Many bugfixes for the journal, including endianness fixes and
6903 ensuring that disk space enforcement works
6905 * sd-login.h is C++ compatible again
6907 * Extend the /etc/os-release format on request of the Debian
6910 * We now refuse non-UTF8 strings used in various configuration
6911 and unit files. This is done to ensure we do not pass invalid
6912 data over D-Bus or expose it elsewhere.
6914 * Register Mimo USB Screens as suitable for automatic seat
6917 * Read SELinux client context from journal clients in a race
6920 * Reorder configuration file lookup order. /etc now always
6921 overrides /run in order to allow the administrator to always
6922 and unconditionally override vendor-supplied or
6923 automatically generated data.
6925 * The various user visible bits of the journal now have man
6926 pages. We still lack man pages for the journal API calls
6929 * We now ship all man pages in HTML format again in the
6932 Contributions from: Dave Reisner, Dirk Eibach, Frederic
6933 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
6934 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
6939 * This is mostly a bugfix release
6941 * systems lacking /etc/os-release are no longer supported.
6943 * Various functionality updates to libsystemd-login.so
6945 * Track class of PAM logins to distinguish greeters from
6948 Contributions from: Kay Sievers, Lennart Poettering, Michael
6953 * This is an important bugfix release for v41.
6955 * Building man pages is now optional which should be useful
6956 for those building systemd from git but unwilling to install
6959 * Watchdog support for supervising services is now usable. In
6960 a future release support for hardware watchdogs
6961 (i.e. /dev/watchdog) will be added building on this.
6963 * Service start rate limiting is now configurable and can be
6964 turned off per service. When a start rate limit is hit a
6965 reboot can automatically be triggered.
6967 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
6969 Contributions from: Benjamin Franzke, Bill Nottingham,
6970 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
6971 Schmidt, Michał Górny, Piotr Drąg
6975 * The systemd binary is installed /usr/lib/systemd/systemd now;
6976 An existing /sbin/init symlink needs to be adapted with the
6979 * The code that loads kernel modules has been ported to invoke
6980 libkmod directly, instead of modprobe. This means we do not
6981 support systems with module-init-tools anymore.
6983 * Watchdog support is now already useful, but still not
6986 * A new kernel command line option systemd.setenv= is
6987 understood to set system wide environment variables
6988 dynamically at boot.
6990 * We now limit the set of capabilities of systemd-journald.
6992 * We now set SIGPIPE to ignore by default, since it only is
6993 useful in shell pipelines, and has little use in general
6994 code. This can be disabled with IgnoreSIPIPE=no in unit
6997 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
6998 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
7003 * This is mostly a bugfix release
7005 * We now expose the reason why a service failed in the
7006 "Result" D-Bus property.
7008 * Rudimentary service watchdog support (will be completed over
7009 the next few releases.)
7011 * When systemd forks off in order execute some service we will
7012 now immediately changes its argv[0] to reflect which process
7013 it will execute. This is useful to minimize the time window
7014 with a generic argv[0], which makes bootcharts more useful
7016 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
7017 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
7018 Mike Kazantsev, Ray Strode
7022 * This is mostly a test release, but incorporates many
7025 * New systemd-cgtop tool to show control groups by their
7028 * Linking against libacl for ACLs is optional again. If
7029 disabled, support tracking device access for active logins
7030 goes becomes unavailable, and so does access to the user
7031 journals by the respective users.
7033 * If a group "adm" exists, journal files are automatically
7034 owned by them, thus allow members of this group full access
7035 to the system journal as well as all user journals.
7037 * The journal now stores the SELinux context of the logging
7038 client for all entries.
7040 * Add C++ inclusion guards to all public headers
7042 * New output mode "cat" in the journal to print only text
7043 messages, without any meta data like date or time.
7045 * Include tiny X server wrapper as a temporary stop-gap to
7046 teach XOrg udev display enumeration. This is used by display
7047 managers such as gdm, and will go away as soon as XOrg
7048 learned native udev hotplugging for display devices.
7050 * Add new systemd-cat tool for executing arbitrary programs
7051 with STDERR/STDOUT connected to the journal. Can also act as
7052 BSD logger replacement, and does so by default.
7054 * Optionally store all locally generated coredumps in the
7055 journal along with meta data.
7057 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
7058 writing short strings to files (for usage for /sys), and for
7059 creating symlinks, character and block device nodes.
7061 * New unit file option ControlGroupPersistent= to make cgroups
7062 persistent, following the mechanisms outlined in
7063 https://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
7065 * Support multiple local RTCs in a sane way
7067 * No longer monopolize IO when replaying readahead data on
7068 rotating disks, since we might starve non-file-system IO to
7069 death, since fanotify() will not see accesses done by blkid,
7072 * Do not show kernel threads in systemd-cgls anymore, unless
7073 requested with new -k switch.
7075 Contributions from: Dan Horák, Kay Sievers, Lennart
7076 Poettering, Michal Schmidt
7080 * This is mostly a test release, but incorporates many
7083 * The git repository moved to:
7084 git://anongit.freedesktop.org/systemd/systemd
7085 ssh://git.freedesktop.org/git/systemd/systemd
7087 * First release with the journal
7088 http://0pointer.de/blog/projects/the-journal.html
7090 * The journal replaces both systemd-kmsg-syslogd and
7091 systemd-stdout-bridge.
7093 * New sd_pid_get_unit() API call in libsystemd-logind
7095 * Many systemadm clean-ups
7097 * Introduce remote-fs-pre.target which is ordered before all
7098 remote mounts and may be used to start services before all
7101 * Added Mageia support
7103 * Add bash completion for systemd-loginctl
7105 * Actively monitor PID file creation for daemons which exit in
7106 the parent process before having finished writing the PID
7107 file in the daemon process. Daemons which do this need to be
7108 fixed (i.e. PID file creation must have finished before the
7109 parent exits), but we now react a bit more gracefully to them.
7111 * Add colourful boot output, mimicking the well-known output
7112 of existing distributions.
7114 * New option PassCredentials= for socket units, for
7115 compatibility with a recent kernel ABI breakage.
7117 * /etc/rc.local is now hooked in via a generator binary, and
7118 thus will no longer act as synchronization point during
7121 * systemctl list-unit-files now supports --root=.
7123 * systemd-tmpfiles now understands two new commands: z, Z for
7124 relabelling files according to the SELinux database. This is
7125 useful to apply SELinux labels to specific files in /sys,
7128 * Output of SysV services is now forwarded to both the console
7129 and the journal by default, not only just the console.
7131 * New man pages for all APIs from libsystemd-login.
7133 * The build tree got reorganized and the build system is a
7134 lot more modular allowing embedded setups to specifically
7135 select the components of systemd they are interested in.
7137 * Support for Linux systems lacking the kernel VT subsystem is
7140 * configure's --with-rootdir= got renamed to
7141 --with-rootprefix= to follow the naming used by udev and
7144 * Unless specified otherwise we will now install to /usr instead
7145 of /usr/local by default.
7147 * Processes with '@' in argv[0][0] are now excluded from the
7148 final shut-down killing spree, following the logic explained
7150 https://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
7152 * All processes remaining in a service cgroup when we enter
7153 the START or START_PRE states are now killed with
7154 SIGKILL. That means it is no longer possible to spawn
7155 background processes from ExecStart= lines (which was never
7156 supported anyway, and bad style).
7158 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
7159 reloading of units together.
7161 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
7162 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
7163 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
7164 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
7165 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek