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 * The option MulticastDNS= of network configuration files has acquired
39 an actual implementation. With MulticastDNS=yes a host can resolve
40 names of remote hosts and to reply to mDNS's A and AAAA requests from
43 * When units are about to be started an additional check is now done to
44 ensure that all dependencies of type BindsTo= (when used in
45 combination with After=) have been started.
47 * systemd-analyze gained a new verb "syscall-filter" which shows which
48 system call groups are defined for the SystemCallFilter= unit file
49 setting, and which system calls they precisely contain.
51 * A new system call filter group "@filesystem" has been added,
52 consisting of various file system related system calls. A group
53 "@reboot" has been added, covering reboot, kexec and shutdown related
54 calls. Finally, a group "@swap" has been added covering swap
55 configuration related calls.
57 * A new unit file option RestrictNamespaces= has been added that may be
58 used to restrict access to the various process namespace types the
59 Linux kernel provides. Specifically, it may be used to take away the
60 right for specific service units to create additional file system,
61 networking, user, and other namespaces. This sandboxing option is
62 particularly relevant due to the high amount of recently discovered
63 namespacing related vulnerabilities in the kernel.
65 * .link gained support for a new AutoNegotiation= setting for
66 configuring Ethernet auto-negotiation.
68 * systemd-networkd's .network files gained support for a new
69 ListenPort= setting in the [DHCP] section to explicitly configure the
70 UDP client port the DHCP client shall listen on.
72 * New systemd-specific mount options are now understood in /etc/fstab:
74 x-systemd.mount-timeout= may be used to configure the maximum
75 permitted runtime of the mount command.
77 x-systemd.device-bound may be set to bind a mount point to its
78 backing device unit, in order to automatically remove a mount point
79 if its backing device is unplugged. This option may also be
80 configured through the new SYSTEMD_MOUNT_DEVICE_BOUND udev property
81 on the block device, which is now automatically set for all CDROM
82 drives, so that mounted CDs are automatically unmounted when they are
83 removed from the drive.
85 x-systemd.after= and x-systemd.before= may be use to explicitly order
86 a mount after or before another unit or mount point.
88 * Enqueued start jobs for device units are now automatically garbage
89 collected if there are no jobs waiting for them anymore.
91 * systemctl list-jobs gained two new switches: With --after, every
92 queued job shows which other queued job is waiting for it; with
93 --before it shows which other jobs every job is waiting for.
95 * systemd-nspawn gained support for ephemeral boots from disk images
96 (or in other words: --ephemeral and --image= may now be
97 combined). Moreover, ephemeral boots are now supported for normal
98 directories, even if the backing file system is not btrfs. Of course,
99 if the file system does not support file system snapshots or
100 reflinks, the initial copy operation will be relatively expensive, but
101 this should still be suitable for many use cases.
103 * Calendar time specifications in .timer units now support
104 specifications relative to the end of a month by using "~" instead of
105 "-" as separator between month and day. For example, "*-02~03" means
106 "The third last day in February". In addition a new syntax for
107 repeated events has been added using the "/" character. For example,
108 "9..17/2:00" means "every two hours from 9am to 5pm".
110 * systemd-socket-proxyd gained a new parameter --connections-max= for
111 configuring the maximum number of concurrent connections.
113 * All python scripts shipped with systemd (specifically: the various
114 tests written in Python) now require Python 3.
116 * sd-id128 gained a new API for generating unique IDs for the host
117 that does not leak the machine ID. Specifically,
118 sd_id128_get_machine_app_specific() derives an ID based on the
119 machine ID a in well-defined, non-reversible, stable way. This is
120 useful whenever an identifier for the host is needed but where the
121 identifier shall not be useful to identify the system beyond the
122 scope of the application itself. (Internally this uses HMAC-SHA256 as
123 keyed hash function using the machine ID as input.)
125 * NotifyAccess= gained a new supported value "exec". When set
126 notifications are accepted from all processes systemd itself invoked,
127 including all control processes.
129 * .nspawn files gained support for defining overlay mounts using the
130 Overlay= and OverlayReadOnly= options. Previously this functionality
131 was only available on the systemd-nspawn command line.
133 * systemd-nspawn's --bind= and --overlay= options gained support for
134 bind/overlay mounts whose source lies within the container tree by
135 prefixing the source path with "+".
137 * systemd-nspawn's --bind= and --overlay= options gained support for
138 automatically allocating a temporary source directory in /var/tmp
139 that is removed when the container dies. Specifically, if the source
140 directory is specified as empty string this mechanism is selected. An
141 example usage is --overlay=+/var::/var, which creates an overlay
142 mount based on the original /var contained in the image, overlayed
143 with a temporary directory in the host's /var/tmp. This way changes
144 to /var are automatically flushed when the container shuts down.
146 * .network files gained a new Unmanaged= boolean setting for explicitly
147 excluding one or more interfaces from management by systemd-networkd.
149 * systemd-nspawn --image= option does now permit raw file system block
150 devices (in addition to images containing partition tables, as
153 * The disk image dissection logic in systemd-nspawn gained support for
154 automatically setting up LUKS encrypted as well as Verity protected
155 partitions. When a container is booted from an encrypted image the
156 passphrase is queried at start-up time. When a container with Verity
157 data is started, the root hash is search in a ".roothash" file
158 accompanying the disk image (alternatively, pass the root hash via
159 the new --root-hash= command line option).
161 * A new tool /usr/lib/systemd/systemd-dissect has been added that may
162 be used to dissect disk images the same way as systemd-nspawn does
163 it, following the Bootable Partition Specification. It may even be
164 used to mount disk images with complex partition setups (including
165 LUKS and Verity partitions) to a local host directory, in order to
166 inspect them. This tool is not considered public API (yet), and is
167 thus not installed into /usr/bin. Please do not rely on its
168 existance, since it might go away or be changed in later systemd
171 * A new generator "systemd-verity-generator" has been added, similar in
172 style to "systemd-cryptsetup-generator", permitting automatic setup of
173 Verity root partitions when systemd boots up. In order to make use of
174 this your partition setup should follow the Discoverable Partitions
175 Specification, and the GPT partition ID of the root file system
176 partition should be identical to the upper 128bit of the Verity root
177 hash. The GPT partition ID of the Verity partition protecting it
178 should be the lower 128bit of the Verity root hash. If the partition
179 image follows this model it is sufficient to specify a single
180 "roothash=" kernel command line argument to both configure which root
181 image and verity partition to use as well as the root hash for
182 it. Note that systemd-nspawn's Verity support follows the same
183 semantics, meaning that disk images with proper Verity data in place
184 may be booted in containers with systemd-nspawn as well as on
185 physical systems via the verity generator. Also note that the "mkosi"
186 tool available at https://github.com/systemd/mkosi has been updated
187 to generate Verity protected disk images following this scheme. In
188 fact, it has been updated to generate disk images that optionally
189 implement a complete UEFI SecureBoot trust chain, involving a signed
190 kernel and initrd image that incorporates such a root hash as well as
191 a Verity-enabled root partition.
193 * Support for the %c, %r, %R specifiers in unit files has been
194 removed. Specifiers are not supposed to be dependent on configuration
195 of unit files themselves (so that they resolve to the same regardless
196 where used in the unit files), but these options were due to the
197 existence of the Slice= option.
199 * The various options in the [Match] section of .network files gained
200 support for negative matching.
202 * The hardware database (hwdb) udev supports has been updated to carry
203 accelerometer quirks.
205 * All system services are now run with a fresh kernel keyring set up
206 for them. The invocation ID is stored by default in it, thus
207 providing a safe, non-overridable way to determine the invocation
210 * Service unit files gained new BindPaths= and BindReadOnlyPaths=
211 options for bind mounting arbitrary paths in a service-specific
212 way. When these options are used, arbitrary host or service files and
213 directories may be mounted to arbitrary locations in the service's
216 * Documentation has been added that lists all of systemd's low-level
217 environment variables:
219 https://github.com/systemd/systemd/blob/master/ENVIRONMENT.md
221 * sd-daemon gained a new API sd_is_socket_sockaddr() for determining
222 whether a specific socket file descriptor matches a specified socket
225 * systemd-firstboot has been updated to check for the
226 systemd.firstboot= kernel command line option. It accepts a boolean
227 and when set to false the first boot questions are skipped.
229 * The systemd-networkd ProxyARP= option has been renamed to
230 IPV4ProxyARP=. Similar, VXLAN-specific option ARPProxy= has been
231 renamed to ReduceARPProxy=. The old names continue to be available
234 * systemd-networkd's bonding device support gained support for two new
235 configuration options ActiveSlave= and PrimarySlave=.
237 * systemd-fstab-generator has been updated to check for the
238 systemd.volatile= kernel command line option, which either takes a
239 boolean parameter or the special value "state". If used the system
240 may be booted in a "volatile" boot mode. Specifically,
241 systemd.volatile=yes is used, the root directory will be mounted as
242 tmpfs, and only /usr is mounted from the actual root file system. If
243 systemd.volatile=state is used, the root directory will be mounted as
244 usual, but /var is mounted as tmpfs. This concept provides similar
245 functionality as systemd-nspawn's --volatile= option, but provides it
246 on physical boots. Use this option for implementing stateless
247 systems, or testing systems with all state and/or configuration reset
248 to the defaults. (Note though that many distributions are not
249 prepared to boot up without a populated /etc or /var, though)
251 * systemd-gpt-auto-generator gained support for LUKS encrypted root
252 partitions. Previously it only supported LUKS encrypted partitions
253 for all other uses, except for the root partition itself.
255 * Socket units gained support for listening on AF_VSOCK sockets for
256 communication in virtualized QEMU environments.
258 * The "configure" script gained a new option --with-fallback-hostname=
259 for specifying the fallback hostname to use if none is configured in
260 /etc/hostname. For example, by specifying
261 --with-fallback-hostname=fedora it is possible to default to a
262 hostname of "fedora" when the user didn't specify anything
265 * systemd-cgls gained support for a new --unit= switch for listing only
266 the control groups of a specific unit. Similar --user-unit= has been
267 added for listing only the control groups of a specific user unit.
269 * systemd-mount gained a new --umount switch for unmounting a mount or
270 automount point (and all mount/automount points below it).
272 * systemd will now refuse full configuration reloads (via systemctl
273 daemon-reload and related calls) unless at least 16MiB of free space
274 are available in /run. This is a safety precaution in order to ensure
275 that generators can safely operate after the reload completed.
277 * A new unit file option RootImage= has been added, which has a similar
278 effect as RootDirectory= but mounts the service's root directory from
279 a disk image instead of plain directory. This logic reuses the same
280 image dissection and mount logic that systemd-nspawn already uses,
281 and hence supports any disk images systemd-nspawn supports, including
282 those following the Discoverable Partition Specification, as well as
283 Verity enabled images. This option enables systemd to run system
284 services directly off disk images acting as resource bundles,
285 possibly even including full integrity data.
287 * A new MountAPIVFS= unit file option has been added, taking a boolean
288 argument. If enabled /proc, /sys and /dev (collectively called the
289 "API VFS") will be mounted for the service. This is only relevant if
290 RootDirectory= or RootImage= is used for the service, as these mounts
291 are of course in place in the host mount namespace anyway.
293 * systemd-nspawn gained support for a new --pivot-root= switch. If
294 specified the root directory within the container image is pivoted to
295 the specified mount point, while the original root disk is moved to a
296 different place. This option enables booting of ostree images
297 directly with systemd-nspawn.
299 * systemd-networkd gained support for configuring IPv6 Proxy NDP
300 addresses via the new IPv6ProxyNDPAddress= .network file setting.
302 * The systemd build scripts will no longer complain if the NTP server
303 addresses are not changed from the defaults. Google is now supporting
304 these NTP servers officially. We still recommend downstreams to
305 properly register an NTP pool with the NTP pool project though.
307 * coredumpctl gained new new "--reverse" option for printing the list
308 of coredumps in reverse order.
310 * The systemd-coredump logic has been improved so that it may be reused
311 for collecting backtraces in non-compiled languages, for example in
312 scripting languages such as Python.
314 * machinectl will now show the UID shift of local containers, if user
315 namespacing is enabled for them.
317 * systemd will now optionally run "environment generator" binaries at
318 configuration load time. They may be used to add environment
319 variables to the environment block passed to services invoked. One
320 user environment generator is shipped by default that sets up
321 environment variables based on files dropped into
322 ~/.config/environment.d/.
324 Contributions from: Adrián López, Alexander Galanin, Alexander
325 Kochetkov, Alexandros Frantzis, Andrey Ulanov, Antoine Eiche, Baruch
326 Siach, Bastien Nocera, Benjamin Robin, Björn, Brandon Philips, Cédric
327 Schieli, Charles (Chas) Williams, Christian Hesse, Daniele Medri,
328 Daniel Drake, Daniel Rusek, Daniel Wagner, Dan Streetman, Dave Reisner,
329 David Glasser, David Herrmann, David Michael, Djalal Harouni, Dmitry
330 Khlebnikov, Dmitry Rozhkov, Dongsu Park, Douglas Christman, Earnestly,
331 Emil Soleyman, Eric Cook, Evgeny Vereshchagin, Felipe Sateler, Fionn
332 Cleary, Florian Klink, Francesco Brozzu, Franck Bui, Gabriel Rauter,
333 Gianluca Boiano, Graeme Lawes, Hans de Goede, Harald Hoyer, Ian
334 Kelling, Ivan Shapovalov, Jakub Wilk, Janne Heß, Jan Synacek, Jason
335 Reeder, Jonathan Boulle, Jörg Thalheim, Jouke Witteveen, Karl Kraus,
336 Kees Cook, Keith Busch, Kieran Colford, kilian-k, Lennart Poettering,
337 Lubomir Rintel, Lucas Werkmeister, Lukas Rusak, Maarten de Vries, Maks
338 Naumov, Mantas Mikulėnas, Marc-Andre Lureau, Marcin Bachry, Mark
339 Stosberg, Martin Ejdestig, Martin Pitt, micah, Michael Biebl, Michael
340 Shields, Michal Schmidt, Michal Sekletar, Michel Kraus, Mike Gilbert,
341 Mirza Krak, Namhyung Kim, nikolaof, peoronoob, Peter Hutterer, Peter
342 Körner, Philip Withnall, Piotr Drąg, Ray Strode, Reverend Homer,
343 Rike-Benjamin Schuppner, Robert Kreuzer, Ronny Chevalier, Ruslan
344 Bilovol, sammynx, Sergey Ptashnick, Sergiusz Urbaniak, Stefan Berger,
345 Stefan Hajnoczi, Stefan Schweter, Susant Sahani, Sylvain Plantefève,
346 Taylor Smock, Thomas Blume, Thomas H. P. Andersen, Tobias Stoeckmann,
347 Tom Gundersen, Torstein Husebø, Viktar Vaŭčkievič, Viktor Mihajlovski,
348 Waldemar Brodkorb, Walter Garcia-Fontes, Wim de With, Yassine
349 Imounachen, Yi EungJun, Yu Watanabe, Zbigniew Jędrzejewski-Szmek,
356 * The new RemoveIPC= option can be used to remove IPC objects owned by
357 the user or group of a service when that service exits.
359 * The new ProtectKernelModules= option can be used to disable explicit
360 load and unload operations of kernel modules by a service. In
361 addition access to /usr/lib/modules is removed if this option is set.
363 * ProtectSystem= option gained a new value "strict", which causes the
364 whole file system tree with the exception of /dev, /proc, and /sys,
365 to be remounted read-only for a service.
367 * The new ProtectKernelTunables= option can be used to disable
368 modification of configuration files in /sys and /proc by a service.
369 Various directories and files are remounted read-only, so access is
370 restricted even if the file permissions would allow it.
372 * The new ProtectControlGroups= option can be used to disable write
373 access by a service to /sys/fs/cgroup.
375 * Various systemd services have been hardened with
376 ProtectKernelTunables=yes, ProtectControlGroups=yes,
377 RestrictAddressFamilies=.
379 * Support for dynamically creating users for the lifetime of a service
380 has been added. If DynamicUser=yes is specified, user and group IDs
381 will be allocated from the range 61184..65519 for the lifetime of the
382 service. They can be resolved using the new nss-systemd.so NSS
383 module. The module must be enabled in /etc/nsswitch.conf. Services
384 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
385 any resources allocated by the service will be cleaned up when the
386 service exits. They also have ProtectHome=read-only and
387 ProtectSystem=strict enabled, so they are not able to make any
388 permanent modifications to the system.
390 * The nss-systemd module also always resolves root and nobody, making
391 it possible to have no /etc/passwd or /etc/group files in minimal
392 container or chroot environments.
394 * Services may be started with their own user namespace using the new
395 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
396 under which the service is running are mapped. All other users are
399 * Support for the cgroup namespace has been added to systemd-nspawn. If
400 supported by kernel, the container system started by systemd-nspawn
401 will have its own view of the cgroup hierarchy. This new behaviour
402 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
404 * The new MemorySwapMax= option can be used to limit the maximum swap
405 usage under the unified cgroup hierarchy.
407 * Support for the CPU controller in the unified cgroup hierarchy has
408 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
409 options. This controller requires out-of-tree patches for the kernel
410 and the support is provisional.
412 * Mount and automount units may now be created transiently
413 (i.e. dynamically at runtime via the bus API, instead of requiring
414 unit files in the file system).
416 * systemd-mount is a new tool which may mount file systems – much like
417 mount(8), optionally pulling in additional dependencies through
418 transient .mount and .automount units. For example, this tool
419 automatically runs fsck on a backing block device before mounting,
420 and allows the automount logic to be used dynamically from the
421 command line for establishing mount points. This tool is particularly
422 useful when dealing with removable media, as it will ensure fsck is
423 run – if necessary – before the first access and that the file system
424 is quickly unmounted after each access by utilizing the automount
425 logic. This maximizes the chance that the file system on the
426 removable media stays in a clean state, and if it isn't in a clean
427 state is fixed automatically.
429 * LazyUnmount=yes option for mount units has been added to expose the
430 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
433 * /efi will be used as the mount point of the EFI boot partition, if
434 the directory is present, and the mount point was not configured
435 through other means (e.g. fstab). If /efi directory does not exist,
436 /boot will be used as before. This makes it easier to automatically
437 mount the EFI partition on systems where /boot is used for something
440 * When operating on GPT disk images for containers, systemd-nspawn will
441 now mount the ESP to /boot or /efi according to the same rules as PID
442 1 running on a host. This allows tools like "bootctl" to operate
443 correctly within such containers, in order to make container images
444 bootable on physical systems.
446 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
448 * Two new user session targets have been added to support running
449 graphical sessions under the systemd --user instance:
450 graphical-session.target and graphical-session-pre.target. See
451 systemd.special(7) for a description of how those targets should be
454 * The vconsole initialization code has been significantly reworked to
455 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
456 support unicode keymaps. Font and keymap configuration will now be
457 copied to all allocated virtual consoles.
459 * FreeBSD's bhyve virtualization is now detected.
461 * Information recorded in the journal for core dumps now includes the
462 contents of /proc/mountinfo and the command line of the process at
463 the top of the process hierarchy (which is usually the init process
466 * systemd-journal-gatewayd learned the --directory= option to serve
467 files from the specified location.
469 * journalctl --root=… can be used to peruse the journal in the
470 /var/log/ directories inside of a container tree. This is similar to
471 the existing --machine= option, but does not require the container to
474 * The hardware database has been extended to support
475 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
478 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
479 specify the click rate for mice which include a horizontal wheel with
480 a click rate that is different than the one for the vertical wheel.
482 * systemd-run gained a new --wait option that makes service execution
483 synchronous. (Specifically, the command will not return until the
484 specified service binary exited.)
486 * systemctl gained a new --wait option that causes the start command to
487 wait until the units being started have terminated again.
489 * A new journal output mode "short-full" has been added which displays
490 timestamps with abbreviated English day names and adds a timezone
491 suffix. Those timestamps include more information than the default
492 "short" output mode, and can be passed directly to journalctl's
493 --since= and --until= options.
495 * /etc/resolv.conf will be bind-mounted into containers started by
496 systemd-nspawn, if possible, so any changes to resolv.conf contents
497 are automatically propagated to the container.
499 * The number of instances for socket-activated services originating
500 from a single IP address can be limited with
501 MaxConnectionsPerSource=, extending the existing setting of
504 * systemd-networkd gained support for vcan ("Virtual CAN") interface
507 * .netdev and .network configuration can now be extended through
510 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
511 Segmentation Offload, Generic Receive Offload, Large Receive Offload
512 can be enabled and disabled using the new UDPSegmentationOffload=,
513 TCPSegmentationOffload=, GenericSegmentationOffload=,
514 GenericReceiveOffload=, LargeReceiveOffload= options in the
515 [Link] section of .link files.
517 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
518 Port VLAN ID can be configured for bridge devices using the new STP=,
519 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
520 section of .netdev files.
522 * The route table to which routes received over DHCP or RA should be
523 added can be configured with the new RouteTable= option in the [DHCP]
524 and [IPv6AcceptRA] sections of .network files.
526 * The Address Resolution Protocol can be disabled on links managed by
527 systemd-networkd using the ARP=no setting in the [Link] section of
530 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
531 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
532 encode information about the result and exit codes of the current
533 service runtime cycle.
535 * systemd-sysctl will now configure kernel parameters in the order
536 they occur in the configuration files. This matches what sysctl
537 has been traditionally doing.
539 * kernel-install "plugins" that are executed to perform various
540 tasks after a new kernel is added and before an old one is removed
541 can now return a special value to terminate the procedure and
542 prevent any later plugins from running.
544 * Journald's SplitMode=login setting has been deprecated. It has been
545 removed from documentation, and its use is discouraged. In a future
546 release it will be completely removed, and made equivalent to current
547 default of SplitMode=uid.
549 * Storage=both option setting in /etc/systemd/coredump.conf has been
550 removed. With fast LZ4 compression storing the core dump twice is not
553 * The --share-system systemd-nspawn option has been replaced with an
554 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
555 this functionality is discouraged. In addition the variables
556 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
557 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
558 individual namespaces.
560 * "machinectl list" now shows the IP address of running containers in
561 the output, as well as OS release information.
563 * "loginctl list" now shows the TTY of each session in the output.
565 * sd-bus gained new API calls sd_bus_track_set_recursive(),
566 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
567 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
568 tracking objects in a "recursive" mode, where a single client can be
569 counted multiple times, if it takes multiple references.
571 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
572 sd_bus_get_exit_on_disconnect(). They may be used to to make a
573 process using sd-bus automatically exit if the bus connection is
576 * Bus clients of the service manager may now "pin" loaded units into
577 memory, by taking an explicit reference on them. This is useful to
578 ensure the client can retrieve runtime data about the service even
579 after the service completed execution. Taking such a reference is
580 available only for privileged clients and should be helpful to watch
581 running services in a race-free manner, and in particular collect
582 information about exit statuses and results.
584 * The nss-resolve module has been changed to strictly return UNAVAIL
585 when communication via D-Bus with resolved failed, and NOTFOUND when
586 a lookup completed but was negative. This means it is now possible to
587 neatly configure fallbacks using nsswitch.conf result checking
588 expressions. Taking benefit of this, the new recommended
589 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
591 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
593 * A new setting CtrlAltDelBurstAction= has been added to
594 /etc/systemd/system.conf which may be used to configure the precise
595 behaviour if the user on the console presses Ctrl-Alt-Del more often
596 than 7 times in 2s. Previously this would unconditionally result in
597 an expedited, immediate reboot. With this new setting the precise
598 operation may be configured in more detail, and also turned off
601 * In .netdev files two new settings RemoteChecksumTx= and
602 RemoteChecksumRx= are now understood that permit configuring the
603 remote checksumming logic for VXLAN networks.
605 * The service manager learnt a new "invocation ID" concept for invoked
606 services. Each runtime cycle of a service will get a new invocation
607 ID (a 128bit random UUID) assigned that identifies the current
608 run of the service uniquely and globally. A new invocation ID
609 is generated each time a service starts up. The journal will store
610 the invocation ID of a service along with any logged messages, thus
611 making the invocation ID useful for matching the online runtime of a
612 service with the offline log data it generated in a safe way without
613 relying on synchronized timestamps. In many ways this new service
614 invocation ID concept is similar to the kernel's boot ID concept that
615 uniquely and globally identifies the runtime of each boot. The
616 invocation ID of a service is passed to the service itself via an
617 environment variable ($INVOCATION_ID). A new bus call
618 GetUnitByInvocationID() has been added that is similar to GetUnit()
619 but instead of retrieving the bus path for a unit by its name
620 retrieves it by its invocation ID. The returned path is valid only as
621 long as the passed invocation ID is current.
623 * systemd-resolved gained a new "DNSStubListener" setting in
624 resolved.conf. It either takes a boolean value or the special values
625 "udp" and "tcp", and configures whether to enable the stub DNS
626 listener on 127.0.0.53:53.
628 * IP addresses configured via networkd may now carry additional
629 configuration settings supported by the kernel. New options include:
630 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
631 PrefixRoute=, AutoJoin=.
633 * The PAM configuration fragment file for "user@.service" shipped with
634 systemd (i.e. the --user instance of systemd) has been stripped to
635 the minimum necessary to make the system boot. Previously, it
636 contained Fedora-specific stanzas that did not apply to other
637 distributions. It is expected that downstream distributions add
638 additional configuration lines, matching their needs to this file,
639 using it only as rough template of what systemd itself needs. Note
640 that this reduced fragment does not even include an invocation of
641 pam_limits which most distributions probably want to add, even though
642 systemd itself does not need it. (There's also the new build time
643 option --with-pamconfdir=no to disable installation of the PAM
646 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
647 capability is now also dropped from its set (in addition to
648 CAP_SYS_MKNOD as before).
650 * In service unit files it is now possible to connect a specific named
651 file descriptor with stdin/stdout/stdout of an executed service. The
652 name may be specified in matching .socket units using the
653 FileDescriptorName= setting.
655 * A number of journal settings may now be configured on the kernel
656 command line. Specifically, the following options are now understood:
657 systemd.journald.max_level_console=,
658 systemd.journald.max_level_store=,
659 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
660 systemd.journald.max_level_wall=.
662 * "systemctl is-enabled --full" will now show by which symlinks a unit
663 file is enabled in the unit dependency tree.
665 * Support for VeraCrypt encrypted partitions has been added to the
666 "cryptsetup" logic and /etc/crypttab.
668 * systemd-detect-virt gained support for a new --private-users switch
669 that checks whether the invoking processes are running inside a user
670 namespace. Similar, a new special value "private-users" for the
671 existing ConditionVirtualization= setting has been added, permitting
672 skipping of specific units in user namespace environments.
674 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
675 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
676 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
677 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
678 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
679 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
680 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
681 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
682 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
683 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
684 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
685 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
686 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
687 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
688 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
689 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
690 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
691 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
692 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
693 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
694 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
695 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
696 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
697 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
698 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
699 Jędrzejewski-Szmek, Zeal Jagannatha
701 — Santa Fe, 2016-11-03
705 * In service units the various ExecXYZ= settings have been extended
706 with an additional special character as first argument of the
707 assigned value: if the character '+' is used the specified command
708 line it will be run with full privileges, regardless of User=,
709 Group=, CapabilityBoundingSet= and similar options. The effect is
710 similar to the existing PermissionsStartOnly= option, but allows
711 configuration of this concept for each executed command line
714 * Services may now alter the service watchdog timeout at runtime by
715 sending a WATCHDOG_USEC= message via sd_notify().
717 * MemoryLimit= and related unit settings now optionally take percentage
718 specifications. The percentage is taken relative to the amount of
719 physical memory in the system (or in case of containers, the assigned
720 amount of memory). This allows scaling service resources neatly with
721 the amount of RAM available on the system. Similarly, systemd-logind's
722 RuntimeDirectorySize= option now also optionally takes percentage
725 * In similar fashion TasksMax= takes percentage values now, too. The
726 value is taken relative to the configured maximum number of processes
727 on the system. The per-service task maximum has been changed to 15%
728 using this functionality. (Effectively this is an increase of 512 →
729 4915 for service units, given the kernel's default pid_max setting.)
731 * Calendar time specifications in .timer units now understand a ".."
732 syntax for time ranges. Example: "4..7:10" may now be used for
733 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
736 * The InaccessableDirectories=, ReadOnlyDirectories= and
737 ReadWriteDirectories= unit file settings have been renamed to
738 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
739 applied to all kinds of file nodes, and not just directories, with
740 the exception of symlinks. Specifically these settings may now be
741 used on block and character device nodes, UNIX sockets and FIFOS as
742 well as regular files. The old names of these settings remain
743 available for compatibility.
745 * systemd will now log about all service processes it kills forcibly
746 (using SIGKILL) because they remained after the clean shutdown phase
747 of the service completed. This should help identifying services that
748 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
749 systemd-logind's configuration a similar log message is generated for
750 processes killed at the end of each session due to this setting.
752 * systemd will now set the $JOURNAL_STREAM environment variable for all
753 services whose stdout/stderr are connected to the Journal (which
754 effectively means by default: all services). The variable contains
755 the device and inode number of the file descriptor used for
756 stdout/stderr. This may be used by invoked programs to detect whether
757 their stdout/stderr is connected to the Journal, in which case they
758 can switch over to direct Journal communication, thus being able to
759 pass extended, structured metadata along with their log messages. As
760 one example, this is now used by glib's logging primitives.
762 * When using systemd's default tmp.mount unit for /tmp, the mount point
763 will now be established with the "nosuid" and "nodev" options. This
764 avoids privilege escalation attacks that put traps and exploits into
765 /tmp. However, this might cause problems if you e. g. put container
766 images or overlays into /tmp; if you need this, override tmp.mount's
767 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
770 * systemd now supports the "memory" cgroup controller also on
773 * The systemd-cgtop tool now optionally takes a control group path as
774 command line argument. If specified, the control group list shown is
775 limited to subgroups of that group.
777 * The SystemCallFilter= unit file setting gained support for
778 pre-defined, named system call filter sets. For example
779 SystemCallFilter=@clock is now an effective way to make all clock
780 changing-related system calls unavailable to a service. A number of
781 similar pre-defined groups are defined. Writing system call filters
782 for system services is simplified substantially with this new
783 concept. Accordingly, all of systemd's own, long-running services now
784 enable system call filtering based on this, by default.
786 * A new service setting MemoryDenyWriteExecute= has been added, taking
787 a boolean value. If turned on, a service may no longer create memory
788 mappings that are writable and executable at the same time. This
789 enhances security for services where this is enabled as it becomes
790 harder to dynamically write and then execute memory in exploited
791 service processes. This option has been enabled for all of systemd's
792 own long-running services.
794 * A new RestrictRealtime= service setting has been added, taking a
795 boolean argument. If set the service's processes may no longer
796 acquire realtime scheduling. This improves security as realtime
797 scheduling may otherwise be used to easily freeze the system.
799 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
800 value. This may be used for requesting that the system manager inside
801 of the container reports start-up completion to nspawn which then
802 propagates this notification further to the service manager
803 supervising nspawn itself. A related option NotifyReady= in .nspawn
804 files has been added too. This functionality allows ordering of the
805 start-up of multiple containers using the usual systemd ordering
808 * machinectl gained a new command "stop" that is an alias for
811 * systemd-resolved gained support for contacting DNS servers on
812 link-local IPv6 addresses.
814 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
815 its caches. A method call for requesting the same operation has been
816 added to the bus API too, and is made available via "systemd-resolve
819 * systemd-resolve gained a new --status switch. If passed a brief
820 summary of the used DNS configuration with per-interface information
823 * resolved.conf gained a new Cache= boolean option, defaulting to
824 on. If turned off local DNS caching is disabled. This comes with a
825 performance penalty in particular when DNSSEC is enabled. Note that
826 resolved disables its internal caching implicitly anyway, when the
827 configured DNS server is on a host-local IP address such as ::1 or
828 127.0.0.1, thus automatically avoiding double local caching.
830 * systemd-resolved now listens on the local IP address 127.0.0.53:53
831 for DNS requests. This improves compatibility with local programs
832 that do not use the libc NSS or systemd-resolved's bus APIs for name
833 resolution. This minimal DNS service is only available to local
834 programs and does not implement the full DNS protocol, but enough to
835 cover local DNS clients. A new, static resolv.conf file, listing just
836 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
837 now recommended to make /etc/resolv.conf a symlink to this file in
838 order to route all DNS lookups to systemd-resolved, regardless if
839 done via NSS, the bus API or raw DNS packets. Note that this local
840 DNS service is not as fully featured as the libc NSS or
841 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
842 used to deliver link-local address information (as this implies
843 sending a local interface index along), LLMNR/mDNS support via this
844 interface is severely restricted. It is thus strongly recommended for
845 all applications to use the libc NSS API or native systemd-resolved
848 * systemd-networkd's bridge support learned a new setting
849 VLANFiltering= for controlling VLAN filtering. Moreover a new section
850 in .network files has been added for configuring VLAN bridging in
851 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
853 * systemd-networkd's IPv6 Router Advertisement code now makes use of
854 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
855 now be acquired without relying on DHCPv6. Two new options
856 UseDomains= and UseDNS= have been added to configure this behaviour.
858 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
859 renamed IPv6AcceptRA=, without altering its behaviour. The old
860 setting name remains available for compatibility reasons.
862 * The systemd-networkd VTI/VTI6 tunneling support gained new options
863 Key=, InputKey= and OutputKey=.
865 * systemd-networkd gained support for VRF ("Virtual Routing Function")
866 interface configuration.
868 * "systemctl edit" may now be used to create new unit files by
869 specifying the --force switch.
871 * sd-event gained a new function sd_event_get_iteration() for
872 requesting the current iteration counter of the event loop. It starts
873 at zero and is increased by one with each event loop iteration.
875 * A new rpm macro %systemd_ordering is provided by the macros.systemd
876 file. It can be used in lieu of %systemd_requires in packages which
877 don't use any systemd functionality and are intended to be installed
878 in minimal containers without systemd present. This macro provides
879 ordering dependencies to ensure that if the package is installed in
880 the same rpm transaction as systemd, systemd will be installed before
881 the scriptlets for the package are executed, allowing unit presets
884 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
885 been added to simplify packaging of generators.
887 * The os-release file gained VERSION_CODENAME field for the
888 distribution nickname (e.g. VERSION_CODENAME=woody).
890 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
891 can be set to disable parsing of metadata and the creation
892 of persistent symlinks for that device.
894 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
895 to make them available to logged-in users has been reverted.
897 * Much of the common code of the various systemd components is now
898 built into an internal shared library libsystemd-shared-231.so
899 (incorporating the systemd version number in the name, to be updated
900 with future releases) that the components link to. This should
901 decrease systemd footprint both in memory during runtime and on
902 disk. Note that the shared library is not for public use, and is
903 neither API not ABI stable, but is likely to change with every new
904 released update. Packagers need to make sure that binaries
905 linking to libsystemd-shared.so are updated in step with the
908 * Configuration for "mkosi" is now part of the systemd
909 repository. mkosi is a tool to easily build legacy-free OS images,
910 and is available on github: https://github.com/systemd/mkosi. If
911 "mkosi" is invoked in the build tree a new raw OS image is generated
912 incorporating the systemd sources currently being worked on and a
913 clean, fresh distribution installation. The generated OS image may be
914 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
915 UEFI PC. This functionality is particularly useful to easily test
916 local changes made to systemd in a pristine, defined environment. See
919 * configure learned the --with-support-url= option to specify the
920 distribution's bugtracker.
922 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
923 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
924 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
925 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
926 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
927 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
928 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
929 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
930 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
931 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
932 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
933 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
934 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
935 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
936 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
937 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
938 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
939 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
940 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
946 * DNSSEC is now turned on by default in systemd-resolved (in
947 "allow-downgrade" mode), but may be turned off during compile time by
948 passing "--with-default-dnssec=no" to "configure" (and of course,
949 during runtime with DNSSEC= in resolved.conf). We recommend
950 downstreams to leave this on at least during development cycles and
951 report any issues with the DNSSEC logic upstream. We are very
952 interested in collecting feedback about the DNSSEC validator and its
953 limitations in the wild. Note however, that DNSSEC support is
954 probably nothing downstreams should turn on in stable distros just
955 yet, as it might create incompatibilities with a few DNS servers and
956 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
957 automatically whenever we detect such incompatible setups, but there
958 might be systems we do not cover yet. Hence: please help us testing
959 the DNSSEC code, leave this on where you can, report back, but then
960 again don't consider turning this on in your stable, LTS or
961 production release just yet. (Note that you have to enable
962 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
963 and its DNSSEC mode for host name resolution from local
966 * systemd-resolve conveniently resolves DANE records with the --tlsa
967 option and OPENPGPKEY records with the --openpgp option. It also
968 supports dumping raw DNS record data via the new --raw= switch.
970 * systemd-logind will now by default terminate user processes that are
971 part of the user session scope unit (session-XX.scope) when the user
972 logs out. This behavior is controlled by the KillUserProcesses=
973 setting in logind.conf, and the previous default of "no" is now
974 changed to "yes". This means that user sessions will be properly
975 cleaned up after, but additional steps are necessary to allow
976 intentionally long-running processes to survive logout.
978 While the user is logged in at least once, user@.service is running,
979 and any service that should survive the end of any individual login
980 session can be started at a user service or scope using systemd-run.
981 systemd-run(1) man page has been extended with an example which shows
982 how to run screen in a scope unit underneath user@.service. The same
983 command works for tmux.
985 After the user logs out of all sessions, user@.service will be
986 terminated too, by default, unless the user has "lingering" enabled.
987 To effectively allow users to run long-term tasks even if they are
988 logged out, lingering must be enabled for them. See loginctl(1) for
989 details. The default polkit policy was modified to allow users to
990 set lingering for themselves without authentication.
992 Previous defaults can be restored at compile time by the
993 --without-kill-user-processes option to "configure".
995 * systemd-logind gained new configuration settings SessionsMax= and
996 InhibitorsMax=, both with a default of 8192. It will not register new
997 user sessions or inhibitors above this limit.
999 * systemd-logind will now reload configuration on SIGHUP.
1001 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
1002 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
1003 enable. Also, support for the "io" cgroup controller in the unified
1004 hierarchy has been added, so that the "memory", "pids" and "io" are
1005 now the controllers that are supported on the unified hierarchy.
1007 WARNING: it is not possible to use previous systemd versions with
1008 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
1009 is necessary to also update systemd in the initramfs if using the
1010 unified hierarchy. An updated SELinux policy is also required.
1012 * LLDP support has been extended, and both passive (receive-only) and
1013 active (sender) modes are supported. Passive mode ("routers-only") is
1014 enabled by default in systemd-networkd. Active LLDP mode is enabled
1015 by default for containers on the internal network. The "networkctl
1016 lldp" command may be used to list information gathered. "networkctl
1017 status" will also show basic LLDP information on connected peers now.
1019 * The IAID and DUID unique identifier sent in DHCP requests may now be
1020 configured for the system and each .network file managed by
1021 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
1023 * systemd-networkd gained support for configuring proxy ARP support for
1024 each interface, via the ProxyArp= setting in .network files. It also
1025 gained support for configuring the multicast querier feature of
1026 bridge devices, via the new MulticastQuerier= setting in .netdev
1027 files. Similarly, snooping on the IGMP traffic can be controlled
1028 via the new setting MulticastSnooping=.
1030 A new setting PreferredLifetime= has been added for addresses
1031 configured in .network file to configure the lifetime intended for an
1034 The systemd-networkd DHCP server gained the option EmitRouter=, which
1035 defaults to yes, to configure whether the DHCP Option 3 (Router)
1038 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
1039 systemd-socket-activate and installed into /usr/bin. It is now fully
1042 * systemd-journald now uses separate threads to flush changes to disk
1043 when closing journal files, thus reducing impact of slow disk I/O on
1044 logging performance.
1046 * The sd-journal API gained two new calls
1047 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
1048 can be used to open journal files using file descriptors instead of
1049 file or directory paths. sd_journal_open_container() has been
1050 deprecated, sd_journal_open_directory_fd() should be used instead
1051 with the flag SD_JOURNAL_OS_ROOT.
1053 * journalctl learned a new output mode "-o short-unix" that outputs log
1054 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
1055 UTC). It also gained support for a new --no-hostname setting to
1056 suppress the hostname column in the family of "short" output modes.
1058 * systemd-ask-password now optionally skips printing of the password to
1059 stdout with --no-output which can be useful in scripts.
1061 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
1062 (devices tagged with ID_MAKER_TOOL) are now tagged with
1063 "uaccess" and are available to logged in users.
1065 * The DeviceAllow= unit setting now supports specifiers (with "%").
1067 * "systemctl show" gained a new --value switch, which allows print a
1068 only the contents of a specific unit property, without also printing
1069 the property's name. Similar support was added to "show*" verbs
1070 of loginctl and machinectl that output "key=value" lists.
1072 * A new unit type "generated" was added for files dynamically generated
1073 by generator tools. Similarly, a new unit type "transient" is used
1074 for unit files created using the runtime API. "systemctl enable" will
1075 refuse to operate on such files.
1077 * A new command "systemctl revert" has been added that may be used to
1078 revert to the vendor version of a unit file, in case local changes
1079 have been made by adding drop-ins or overriding the unit file.
1081 * "machinectl clean" gained a new verb to automatically remove all or
1082 just hidden container images.
1084 * systemd-tmpfiles gained support for a new line type "e" for emptying
1085 directories, if they exist, without creating them if they don't.
1087 * systemd-nspawn gained support for automatically patching the UID/GIDs
1088 of the owners and the ACLs of all files and directories in a
1089 container tree to match the UID/GID user namespacing range selected
1090 for the container invocation. This mode is enabled via the new
1091 --private-users-chown switch. It also gained support for
1092 automatically choosing a free, previously unused UID/GID range when
1093 starting a container, via the new --private-users=pick setting (which
1094 implies --private-users-chown). Together, these options for the first
1095 time make user namespacing for nspawn containers fully automatic and
1096 thus deployable. The systemd-nspawn@.service template unit file has
1097 been changed to use this functionality by default.
1099 * systemd-nspawn gained a new --network-zone= switch, that allows
1100 creating ad-hoc virtual Ethernet links between multiple containers,
1101 that only exist as long as at least one container referencing them is
1102 running. This allows easy connecting of multiple containers with a
1103 common link that implements an Ethernet broadcast domain. Each of
1104 these network "zones" may be named relatively freely by the user, and
1105 may be referenced by any number of containers, but each container may
1106 only reference one of these "zones". On the lower level, this is
1107 implemented by an automatically managed bridge network interface for
1108 each zone, that is created when the first container referencing its
1109 zone is created and removed when the last one referencing its zone
1112 * The default start timeout may now be configured on the kernel command
1113 line via systemd.default_timeout_start_sec=. It was already
1114 configurable via the DefaultTimeoutStartSec= option in
1115 /etc/systemd/system.conf.
1117 * Socket units gained a new TriggerLimitIntervalSec= and
1118 TriggerLimitBurst= setting to configure a limit on the activation
1119 rate of the socket unit.
1121 * The LimitNICE= setting now optionally takes normal UNIX nice values
1122 in addition to the raw integer limit value. If the specified
1123 parameter is prefixed with "+" or "-" and is in the range -20..19 the
1124 value is understood as UNIX nice value. If not prefixed like this it
1125 is understood as raw RLIMIT_NICE limit.
1127 * Note that the effect of the PrivateDevices= unit file setting changed
1128 slightly with this release: the per-device /dev file system will be
1129 mounted read-only from this version on, and will have "noexec"
1130 set. This (minor) change of behavior might cause some (exceptional)
1131 legacy software to break, when PrivateDevices=yes is set for its
1132 service. Please leave PrivateDevices= off if you run into problems
1135 * systemd-bootchart has been split out to a separate repository:
1136 https://github.com/systemd/systemd-bootchart
1138 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
1139 merged into the kernel in its current form.
1141 * The compatibility libraries libsystemd-daemon.so,
1142 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
1143 which have been deprecated since systemd-209 have been removed along
1144 with the corresponding pkg-config files. All symbols provided by
1145 those libraries are provided by libsystemd.so.
1147 * The Capabilities= unit file setting has been removed (it is ignored
1148 for backwards compatibility). AmbientCapabilities= and
1149 CapabilityBoundingSet= should be used instead.
1151 * A new special target has been added, initrd-root-device.target,
1152 which creates a synchronization point for dependencies of the root
1153 device in early userspace. Initramfs builders must ensure that this
1154 target is now included in early userspace.
1156 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
1157 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
1158 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
1159 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
1160 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
1161 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
1162 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
1163 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
1164 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
1165 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
1166 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
1167 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
1168 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
1169 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
1170 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
1171 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
1172 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
1173 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
1174 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
1175 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
1176 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
1177 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
1178 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
1179 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
1180 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
1183 — Fairfax, 2016-05-21
1187 * The systemd-resolved DNS resolver service has gained a substantial
1188 set of new features, most prominently it may now act as a DNSSEC
1189 validating stub resolver. DNSSEC mode is currently turned off by
1190 default, but is expected to be turned on by default in one of the
1191 next releases. For now, we invite everybody to test the DNSSEC logic
1192 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
1193 service also gained a full set of D-Bus interfaces, including calls
1194 to configure DNS and DNSSEC settings per link (for use by external
1195 network management software). systemd-resolved and systemd-networkd
1196 now distinguish between "search" and "routing" domains. The former
1197 are used to qualify single-label names, the latter are used purely
1198 for routing lookups within certain domains to specific links.
1199 resolved now also synthesizes RRs for all entries from /etc/hosts.
1201 * The systemd-resolve tool (which is a client utility for
1202 systemd-resolved) has been improved considerably and is now fully
1203 supported and documented. Hence it has moved from /usr/lib/systemd to
1206 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
1209 * The coredump collection logic has been reworked: when a coredump is
1210 collected it is now written to disk, compressed and processed
1211 (including stacktrace extraction) from a new instantiated service
1212 systemd-coredump@.service, instead of directly from the
1213 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
1214 processing large coredumps can take up a substantial amount of
1215 resources and time, and this previously happened entirely outside of
1216 systemd's service supervision. With the new logic the core_pattern
1217 hook only does minimal metadata collection before passing off control
1218 to the new instantiated service, which is configured with a time
1219 limit, a nice level and other settings to minimize negative impact on
1220 the rest of the system. Also note that the new logic will honour the
1221 RLIMIT_CORE setting of the crashed process, which now allows users
1222 and processes to turn off coredumping for their processes by setting
1225 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
1226 and all forked processes by default. Previously, PID 1 would leave
1227 the setting at "0" for all processes, as set by the kernel. Note that
1228 the resource limit traditionally has no effect on the generated
1229 coredumps on the system if the /proc/sys/kernel/core_pattern hook
1230 logic is used. Since the limit is now honoured (see above) its
1231 default has been changed so that the coredumping logic is enabled by
1232 default for all processes, while allowing specific opt-out.
1234 * When the stacktrace is extracted from processes of system users, this
1235 is now done as "systemd-coredump" user, in order to sandbox this
1236 potentially security sensitive parsing operation. (Note that when
1237 processing coredumps of normal users this is done under the user ID
1238 of process that crashed, as before.) Packagers should take notice
1239 that it is now necessary to create the "systemd-coredump" system user
1240 and group at package installation time.
1242 * The systemd-activate socket activation testing tool gained support
1243 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
1244 and --seqpacket switches. It also has been extended to support both
1245 new-style and inetd-style file descriptor passing. Use the new
1246 --inetd switch to request inetd-style file descriptor passing.
1248 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
1249 variable, which takes a boolean value. If set to false, ANSI color
1250 output is disabled in the tools even when run on a terminal that
1253 * The VXLAN support in networkd now supports two new settings
1254 DestinationPort= and PortRange=.
1256 * A new systemd.machine_id= kernel command line switch has been added,
1257 that may be used to set the machine ID in /etc/machine-id if it is
1258 not initialized yet. This command line option has no effect if the
1259 file is already initialized.
1261 * systemd-nspawn gained a new --as-pid2 switch that invokes any
1262 specified command line as PID 2 rather than PID 1 in the
1263 container. In this mode PID 1 is a minimal stub init process that
1264 implements the special POSIX and Linux semantics of PID 1 regarding
1265 signal and child process management. Note that this stub init process
1266 is implemented in nspawn itself and requires no support from the
1267 container image. This new logic is useful to support running
1268 arbitrary commands in the container, as normal processes are
1269 generally not prepared to run as PID 1.
1271 * systemd-nspawn gained a new --chdir= switch for setting the current
1272 working directory for the process started in the container.
1274 * "journalctl /dev/sda" will now output all kernel log messages for
1275 specified device from the current boot, in addition to all devices
1276 that are parents of it. This should make log output about devices
1277 pretty useful, as long as kernel drivers attach enough metadata to
1278 the log messages. (The usual SATA drivers do.)
1280 * The sd-journal API gained two new calls
1281 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
1282 that report whether log data from /run or /var has been found.
1284 * journalctl gained a new switch "--fields" that prints all journal
1285 record field names currently in use in the journal. This is backed
1286 by two new sd-journal API calls sd_journal_enumerate_fields() and
1287 sd_journal_restart_fields().
1289 * Most configurable timeouts in systemd now expect an argument of
1290 "infinity" to turn them off, instead of "0" as before. The semantics
1291 from now on is that a timeout of "0" means "now", and "infinity"
1292 means "never". To maintain backwards compatibility, "0" continues to
1293 turn off previously existing timeout settings.
1295 * "systemctl reload-or-try-restart" has been renamed to "systemctl
1296 try-reload-or-restart" to clarify what it actually does: the "try"
1297 logic applies to both reloading and restarting, not just restarting.
1298 The old name continues to be accepted for compatibility.
1300 * On boot-up, when PID 1 detects that the system clock is behind the
1301 release date of the systemd version in use, the clock is now set
1302 to the latter. Previously, this was already done in timesyncd, in order
1303 to avoid running with clocks set to the various clock epochs such as
1304 1902, 1938 or 1970. With this change the logic is now done in PID 1
1305 in addition to timesyncd during early boot-up, so that it is enforced
1306 before the first process is spawned by systemd. Note that the logic
1307 in timesyncd remains, as it is more comprehensive and ensures
1308 clock monotonicity by maintaining a persistent timestamp file in
1309 /var. Since /var is generally not available in earliest boot or the
1310 initrd, this part of the logic remains in timesyncd, and is not done
1313 * Support for tweaking details in net_cls.class_id through the
1314 NetClass= configuration directive has been removed, as the kernel
1315 people have decided to deprecate that controller in cgroup v2.
1316 Userspace tools such as nftables are moving over to setting rules
1317 that are specific to the full cgroup path of a task, which obsoletes
1318 these controllers anyway. The NetClass= directive is kept around for
1319 legacy compatibility reasons. For a more in-depth description of the
1320 kernel change, please refer to the respective upstream commit:
1322 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
1324 * A new service setting RuntimeMaxSec= has been added that may be used
1325 to specify a maximum runtime for a service. If the timeout is hit, the
1326 service is terminated and put into a failure state.
1328 * A new service setting AmbientCapabilities= has been added. It allows
1329 configuration of additional Linux process capabilities that are
1330 passed to the activated processes. This is only available on very
1333 * The process resource limit settings in service units may now be used
1334 to configure hard and soft limits individually.
1336 * The various libsystemd APIs such as sd-bus or sd-event now publicly
1337 expose support for gcc's __attribute__((cleanup())) C extension.
1338 Specifically, for many object destructor functions alternative
1339 versions have been added that have names suffixed with "p" and take a
1340 pointer to a pointer to the object to destroy, instead of just a
1341 pointer to the object itself. This is useful because these destructor
1342 functions may be used directly as parameters to the cleanup
1343 construct. Internally, systemd has been a heavy user of this GCC
1344 extension for a long time, and with this change similar support is
1345 now available to consumers of the library outside of systemd. Note
1346 that by using this extension in your sources compatibility with old
1347 and strictly ANSI compatible C compilers is lost. However, all gcc or
1348 LLVM versions of recent years support this extension.
1350 * Timer units gained support for a new setting RandomizedDelaySec= that
1351 allows configuring some additional randomized delay to the configured
1352 time. This is useful to spread out timer events to avoid load peaks in
1353 clusters or larger setups.
1355 * Calendar time specifications now support sub-second accuracy.
1357 * Socket units now support listening on SCTP and UDP-lite protocol
1360 * The sd-event API now comes with a full set of man pages.
1362 * Older versions of systemd contained experimental support for
1363 compressing journal files and coredumps with the LZ4 compressor that
1364 was not compatible with the lz4 binary (due to API limitations of the
1365 lz4 library). This support has been removed; only support for files
1366 compatible with the lz4 binary remains. This LZ4 logic is now
1367 officially supported and no longer considered experimental.
1369 * The dkr image import logic has been removed again from importd. dkr's
1370 micro-services focus doesn't fit into the machine image focus of
1371 importd, and quickly got out of date with the upstream dkr API.
1373 * Creation of the /run/lock/lockdev/ directory was dropped from
1374 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
1375 been available for many years. If you still need this, you need to
1376 create your own tmpfiles.d config file with:
1378 d /run/lock/lockdev 0775 root lock -
1380 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
1381 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
1382 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
1383 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
1384 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
1385 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
1386 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
1387 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
1388 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
1389 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
1390 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
1391 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
1392 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
1393 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
1394 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
1395 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
1396 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
1397 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
1398 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
1400 — Berlin, 2016-02-11
1404 * A number of properties previously only settable in unit
1405 files are now also available as properties to set when
1406 creating transient units programmatically via the bus, as it
1407 is exposed with systemd-run's --property=
1408 setting. Specifically, these are: SyslogIdentifier=,
1409 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
1410 EnvironmentFile=, ReadWriteDirectories=,
1411 ReadOnlyDirectories=, InaccessibleDirectories=,
1412 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
1414 * When creating transient services via the bus API it is now
1415 possible to pass in a set of file descriptors to use as
1416 STDIN/STDOUT/STDERR for the invoked process.
1418 * Slice units may now be created transiently via the bus APIs,
1419 similar to the way service and scope units may already be
1420 created transiently.
1422 * Wherever systemd expects a calendar timestamp specification
1423 (like in journalctl's --since= and --until= switches) UTC
1424 timestamps are now supported. Timestamps suffixed with "UTC"
1425 are now considered to be in Universal Time Coordinated
1426 instead of the local timezone. Also, timestamps may now
1427 optionally be specified with sub-second accuracy. Both of
1428 these additions also apply to recurring calendar event
1429 specification, such as OnCalendar= in timer units.
1431 * journalctl gained a new "--sync" switch that asks the
1432 journal daemon to write all so far unwritten log messages to
1433 disk and sync the files, before returning.
1435 * systemd-tmpfiles learned two new line types "q" and "Q" that
1436 operate like "v", but also set up a basic btrfs quota
1437 hierarchy when used on a btrfs file system with quota
1440 * tmpfiles' "v", "q" and "Q" will now create a plain directory
1441 instead of a subvolume (even on a btrfs file system) if the
1442 root directory is a plain directory, and not a
1443 subvolume. This should simplify things with certain chroot()
1444 environments which are not aware of the concept of btrfs
1447 * systemd-detect-virt gained a new --chroot switch to detect
1448 whether execution takes place in a chroot() environment.
1450 * CPUAffinity= now takes CPU index ranges in addition to
1453 * The various memory-related resource limit settings (such as
1454 LimitAS=) now understand the usual K, M, G, ... suffixes to
1455 the base of 1024 (IEC). Similar, the time-related resource
1456 limit settings understand the usual min, h, day, ...
1459 * There's a new system.conf setting DefaultTasksMax= to
1460 control the default TasksMax= setting for services and
1461 scopes running on the system. (TasksMax= is the primary
1462 setting that exposes the "pids" cgroup controller on systemd
1463 and was introduced in the previous systemd release.) The
1464 setting now defaults to 512, which means services that are
1465 not explicitly configured otherwise will only be able to
1466 create 512 processes or threads at maximum, from this
1467 version on. Note that this means that thread- or
1468 process-heavy services might need to be reconfigured to set
1469 TasksMax= to a higher value. It is sufficient to set
1470 TasksMax= in these specific unit files to a higher value, or
1471 even "infinity". Similar, there's now a logind.conf setting
1472 UserTasksMax= that defaults to 4096 and limits the total
1473 number of processes or tasks each user may own
1474 concurrently. nspawn containers also have the TasksMax=
1475 value set by default now, to 8192. Note that all of this
1476 only has an effect if the "pids" cgroup controller is
1477 enabled in the kernel. The general benefit of these changes
1478 should be a more robust and safer system, that provides a
1479 certain amount of per-service fork() bomb protection.
1481 * systemd-nspawn gained the new --network-veth-extra= switch
1482 to define additional and arbitrarily-named virtual Ethernet
1483 links between the host and the container.
1485 * A new service execution setting PassEnvironment= has been
1486 added that allows importing select environment variables
1487 from PID1's environment block into the environment block of
1490 * Timer units gained support for a new RemainAfterElapse=
1491 setting which takes a boolean argument. It defaults to on,
1492 exposing behaviour unchanged to previous releases. If set to
1493 off, timer units are unloaded after they elapsed if they
1494 cannot elapse again. This is particularly useful for
1495 transient timer units, which shall not stay around longer
1496 than until they first elapse.
1498 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
1499 default now (the kernel default is 16). This is beneficial
1500 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
1501 allows substantially larger numbers of queued
1502 datagrams. This should increase the capability of systemd to
1503 parallelize boot-up, as logging and sd_notify() are unlikely
1504 to stall execution anymore. If you need to change the value
1505 from the new defaults, use the usual sysctl.d/ snippets.
1507 * The compression framing format used by the journal or
1508 coredump processing has changed to be in line with what the
1509 official LZ4 tools generate. LZ4 compression support in
1510 systemd was considered unsupported previously, as the format
1511 was not compatible with the normal tools. With this release
1512 this has changed now, and it is hence safe for downstream
1513 distributions to turn it on. While not compressing as well
1514 as the XZ, LZ4 is substantially faster, which makes
1515 it a good default choice for the compression logic in the
1516 journal and in coredump handling.
1518 * Any reference to /etc/mtab has been dropped from
1519 systemd. The file has been obsolete since a while, but
1520 systemd refused to work on systems where it was incorrectly
1521 set up (it should be a symlink or non-existent). Please make
1522 sure to update to util-linux 2.27.1 or newer in conjunction
1523 with this systemd release, which also drops any reference to
1524 /etc/mtab. If you maintain a distribution make sure that no
1525 software you package still references it, as this is a
1526 likely source of bugs. There's also a glibc bug pending,
1527 asking for removal of any reference to this obsolete file:
1529 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
1531 Note that only util-linux versions built with
1532 --enable-libmount-force-mountinfo are supported.
1534 * Support for the ".snapshot" unit type has been removed. This
1535 feature turned out to be little useful and little used, and
1536 has now been removed from the core and from systemctl.
1538 * The dependency types RequiresOverridable= and
1539 RequisiteOverridable= have been removed from systemd. They
1540 have been used only very sparingly to our knowledge and
1541 other options that provide a similar effect (such as
1542 systemctl --mode=ignore-dependencies) are much more useful
1543 and commonly used. Moreover, they were only half-way
1544 implemented as the option to control behaviour regarding
1545 these dependencies was never added to systemctl. By removing
1546 these dependency types the execution engine becomes a bit
1547 simpler. Unit files that use these dependencies should be
1548 changed to use the non-Overridable dependency types
1549 instead. In fact, when parsing unit files with these
1550 options, that's what systemd will automatically convert them
1551 too, but it will also warn, asking users to fix the unit
1552 files accordingly. Removal of these dependency types should
1553 only affect a negligible number of unit files in the wild.
1555 * Behaviour of networkd's IPForward= option changed
1556 (again). It will no longer maintain a per-interface setting,
1557 but propagate one way from interfaces where this is enabled
1558 to the global kernel setting. The global setting will be
1559 enabled when requested by a network that is set up, but
1560 never be disabled again. This change was made to make sure
1561 IPv4 and IPv6 behaviour regarding packet forwarding is
1562 similar (as the Linux IPv6 stack does not support
1563 per-interface control of this setting) and to minimize
1566 * In unit files the behaviour of %u, %U, %h, %s has
1567 changed. These specifiers will now unconditionally resolve
1568 to the various user database fields of the user that the
1569 systemd instance is running as, instead of the user
1570 configured in the specific unit via User=. Note that this
1571 effectively doesn't change much, as resolving of these
1572 specifiers was already turned off in the --system instance
1573 of systemd, as we cannot do NSS lookups from PID 1. In the
1574 --user instance of systemd these specifiers where correctly
1575 resolved, but hardly made any sense, since the user instance
1576 lacks privileges to do user switches anyway, and User= is
1577 hence useless. Moreover, even in the --user instance of
1578 systemd behaviour was awkward as it would only take settings
1579 from User= assignment placed before the specifier into
1580 account. In order to unify and simplify the logic around
1581 this the specifiers will now always resolve to the
1582 credentials of the user invoking the manager (which in case
1583 of PID 1 is the root user).
1585 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
1586 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
1587 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
1588 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
1589 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
1590 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
1591 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
1592 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
1593 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
1594 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
1595 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
1596 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
1597 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
1598 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
1601 — Berlin, 2015-11-18
1605 * systemd now depends on util-linux v2.27. More specifically,
1606 the newly added mount monitor feature in libmount now
1607 replaces systemd's former own implementation.
1609 * libmount mandates /etc/mtab not to be regular file, and
1610 systemd now enforces this condition at early boot.
1611 /etc/mtab has been deprecated and warned about for a very
1612 long time, so systems running systemd should already have
1613 stopped having this file around as anything else than a
1614 symlink to /proc/self/mounts.
1616 * Support for the "pids" cgroup controller has been added. It
1617 allows accounting the number of tasks in a cgroup and
1618 enforcing limits on it. This adds two new setting
1619 TasksAccounting= and TasksMax= to each unit, as well as a
1620 global option DefaultTasksAccounting=.
1622 * Support for the "net_cls" cgroup controller has been added.
1623 It allows assigning a net class ID to each task in the
1624 cgroup, which can then be used in firewall rules and traffic
1625 shaping configurations. Note that the kernel netfilter net
1626 class code does not currently work reliably for ingress
1627 packets on unestablished sockets.
1629 This adds a new config directive called NetClass= to CGroup
1630 enabled units. Allowed values are positive numbers for fixed
1631 assignments and "auto" for picking a free value
1634 * 'systemctl is-system-running' now returns 'offline' if the
1635 system is not booted with systemd. This command can now be
1636 used as a substitute for 'systemd-notify --booted'.
1638 * Watchdog timeouts have been increased to 3 minutes for all
1639 in-tree service files. Apparently, disk IO issues are more
1640 frequent than we hoped, and user reported >1 minute waiting
1643 * 'machine-id-commit' functionality has been merged into
1644 'machine-id-setup --commit'. The separate binary has been
1647 * The WorkingDirectory= directive in unit files may now be set
1648 to the special value '~'. In this case, the working
1649 directory is set to the home directory of the user
1650 configured in User=.
1652 * "machinectl shell" will now open the shell in the home
1653 directory of the selected user by default.
1655 * The CrashChVT= configuration file setting is renamed to
1656 CrashChangeVT=, following our usual logic of not
1657 abbreviating unnecessarily. The old directive is still
1658 supported for compat reasons. Also, this directive now takes
1659 an integer value between 1 and 63, or a boolean value. The
1660 formerly supported '-1' value for disabling stays around for
1663 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
1664 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
1665 RootDirectory= properties can now be set for transient
1668 * The systemd-analyze tool gained a new "set-log-target" verb
1669 to change the logging target the system manager logs to
1670 dynamically during runtime. This is similar to how
1671 "systemd-analyze set-log-level" already changes the log
1674 * In nspawn /sys is now mounted as tmpfs, with only a selected
1675 set of subdirectories mounted in from the real sysfs. This
1676 enhances security slightly, and is useful for ensuring user
1677 namespaces work correctly.
1679 * Support for USB FunctionFS activation has been added. This
1680 allows implementation of USB gadget services that are
1681 activated as soon as they are requested, so that they don't
1682 have to run continuously, similar to classic socket
1685 * The "systemctl exit" command now optionally takes an
1686 additional parameter that sets the exit code to return from
1687 the systemd manager when exiting. This is only relevant when
1688 running the systemd user instance, or when running the
1689 system instance in a container.
1691 * sd-bus gained the new API calls sd_bus_path_encode_many()
1692 and sd_bus_path_decode_many() that allow easy encoding and
1693 decoding of multiple identifier strings inside a D-Bus
1694 object path. Another new call sd_bus_default_flush_close()
1695 has been added to flush and close per-thread default
1698 * systemd-cgtop gained support for a -M/--machine= switch to
1699 show the control groups within a certain container only.
1701 * "systemctl kill" gained support for an optional --fail
1702 switch. If specified the requested operation will fail of no
1703 processes have been killed, because the unit had no
1704 processes attached, or similar.
1706 * A new systemd.crash_reboot=1 kernel command line option has
1707 been added that triggers a reboot after crashing. This can
1708 also be set through CrashReboot= in systemd.conf.
1710 * The RuntimeDirectory= setting now understands unit
1711 specifiers like %i or %f.
1713 * A new (still internal) library API sd-ipv4acd has been added,
1714 that implements address conflict detection for IPv4. It's
1715 based on code from sd-ipv4ll, and will be useful for
1716 detecting DHCP address conflicts.
1718 * File descriptors passed during socket activation may now be
1719 named. A new API sd_listen_fds_with_names() is added to
1720 access the names. The default names may be overridden,
1721 either in the .socket file using the FileDescriptorName=
1722 parameter, or by passing FDNAME= when storing the file
1723 descriptors using sd_notify().
1725 * systemd-networkd gained support for:
1727 - Setting the IPv6 Router Advertisement settings via
1728 IPv6AcceptRouterAdvertisements= in .network files.
1730 - Configuring the HelloTimeSec=, MaxAgeSec= and
1731 ForwardDelaySec= bridge parameters in .netdev files.
1733 - Configuring PreferredSource= for static routes in
1736 * The "ask-password" framework used to query for LUKS harddisk
1737 passwords or SSL passwords during boot gained support for
1738 caching passwords in the kernel keyring, if it is
1739 available. This makes sure that the user only has to type in
1740 a passphrase once if there are multiple objects to unlock
1741 with the same one. Previously, such password caching was
1742 available only when Plymouth was used; this moves the
1743 caching logic into the systemd codebase itself. The
1744 "systemd-ask-password" utility gained a new --keyname=
1745 switch to control which kernel keyring key to use for
1746 caching a password in. This functionality is also useful for
1747 enabling display managers such as gdm to automatically
1748 unlock the user's GNOME keyring if its passphrase, the
1749 user's password and the harddisk password are the same, if
1750 gdm-autologin is used.
1752 * When downloading tar or raw images using "machinectl
1753 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
1754 file is now also downloaded, if it is available and stored
1755 next to the image file.
1757 * Units of type ".socket" gained a new boolean setting
1758 Writable= which is only useful in conjunction with
1759 ListenSpecial=. If true, enables opening the specified
1760 special file in O_RDWR mode rather than O_RDONLY mode.
1762 * systemd-rfkill has been reworked to become a singleton
1763 service that is activated through /dev/rfkill on each rfkill
1764 state change and saves the settings to disk. This way,
1765 systemd-rfkill is now compatible with devices that exist
1766 only intermittendly, and even restores state if the previous
1767 system shutdown was abrupt rather than clean.
1769 * The journal daemon gained support for vacuuming old journal
1770 files controlled by the number of files that shall remain,
1771 in addition to the already existing control by size and by
1772 date. This is useful as journal interleaving performance
1773 degrades with too many separate journal files, and allows
1774 putting an effective limit on them. The new setting defaults
1775 to 100, but this may be changed by setting SystemMaxFiles=
1776 and RuntimeMaxFiles= in journald.conf. Also, the
1777 "journalctl" tool gained the new --vacuum-files= switch to
1778 manually vacuum journal files to leave only the specified
1779 number of files in place.
1781 * udev will now create /dev/disk/by-path links for ATA devices
1782 on kernels where that is supported.
1784 * Galician, Serbian, Turkish and Korean translations were added.
1786 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
1787 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
1788 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
1789 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
1790 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
1791 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
1792 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
1793 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
1794 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
1795 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
1796 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
1797 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
1798 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
1799 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
1800 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
1801 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
1802 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
1803 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
1805 — Berlin, 2015-10-07
1809 * The DHCP implementation of systemd-networkd gained a set of
1812 - The DHCP server now supports emitting DNS and NTP
1813 information. It may be enabled and configured via
1814 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
1815 and NTP information is enabled, but no servers are
1816 configured, the corresponding uplink information (if there
1817 is any) is propagated.
1819 - Server and client now support transmission and reception
1820 of timezone information. It can be configured via the
1821 newly introduced network options UseTimezone=,
1822 EmitTimezone=, and Timezone=. Transmission of timezone
1823 information is enabled between host and containers by
1824 default now: the container will change its local timezone
1825 to what the host has set.
1827 - Lease timeouts can now be configured via
1828 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
1830 - The DHCP server improved on the stability of
1831 leases. Clients are more likely to get the same lease
1832 information back, even if the server loses state.
1834 - The DHCP server supports two new configuration options to
1835 control the lease address pool metrics, PoolOffset= and
1838 * The encapsulation limit of tunnels in systemd-networkd may
1839 now be configured via 'EncapsulationLimit='. It allows
1840 modifying the maximum additional levels of encapsulation
1841 that are permitted to be prepended to a packet.
1843 * systemd now supports the concept of user buses replacing
1844 session buses, if used with dbus-1.10 (and enabled via dbus
1845 --enable-user-session). It previously only supported this on
1846 kdbus-enabled systems, and this release expands this to
1847 'dbus-daemon' systems.
1849 * systemd-networkd now supports predictable interface names
1852 * systemd now optionally supports the new Linux kernel
1853 "unified" control group hierarchy. If enabled via the kernel
1854 command-line option 'systemd.unified_cgroup_hierarchy=1',
1855 systemd will try to mount the unified cgroup hierarchy
1856 directly on /sys/fs/cgroup. If not enabled, or not
1857 available, systemd will fall back to the legacy cgroup
1858 hierarchy setup, as before. Host system and containers can
1859 mix and match legacy and unified hierarchies as they
1860 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
1861 environment variable to individually select the hierarchy to
1862 use for executed containers. By default, nspawn will use the
1863 unified hierarchy for the containers if the host uses the
1864 unified hierarchy, and the legacy hierarchy otherwise.
1865 Please note that at this point the unified hierarchy is an
1866 experimental kernel feature and is likely to change in one
1867 of the next kernel releases. Therefore, it should not be
1868 enabled by default in downstream distributions yet. The
1869 minimum required kernel version for the unified hierarchy to
1870 work is 4.2. Note that when the unified hierarchy is used
1871 for the first time delegated access to controllers is
1872 safe. Because of this systemd-nspawn containers will get
1873 access to controllers now, as will systemd user
1874 sessions. This means containers and user sessions may now
1875 manage their own resources, partitioning up what the system
1878 * A new special scope unit "init.scope" has been introduced
1879 that encapsulates PID 1 of the system. It may be used to
1880 determine resource usage and enforce resource limits on PID
1881 1 itself. PID 1 hence moved out of the root of the control
1884 * The cgtop tool gained support for filtering out kernel
1885 threads when counting tasks in a control group. Also, the
1886 count of processes is now recursively summed up by
1887 default. Two options -k and --recursive= have been added to
1888 revert to old behaviour. The tool has also been updated to
1889 work correctly in containers now.
1891 * systemd-nspawn's --bind= and --bind-ro= options have been
1892 extended to allow creation of non-recursive bind mounts.
1894 * libsystemd gained two new calls sd_pid_get_cgroup() and
1895 sd_peer_get_cgroup() which return the control group path of
1896 a process or peer of a connected AF_UNIX socket. This
1897 function call is particularly useful when implementing
1898 delegated subtrees support in the control group hierarchy.
1900 * The "sd-event" event loop API of libsystemd now supports
1901 correct dequeuing of real-time signals, without losing
1904 * When systemd requests a PolicyKit decision when managing
1905 units it will now add additional fields to the request,
1906 including unit name and desired operation. This enables more
1907 powerful PolicyKit policies, that make decisions depending
1908 on these parameters.
1910 * nspawn learnt support for .nspawn settings files, that may
1911 accompany the image files or directories of containers, and
1912 may contain additional settings for the container. This is
1913 an alternative to configuring container parameters via the
1914 nspawn command line.
1916 Contributions from: Cristian Rodríguez, Daniel Mack, David
1917 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
1918 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
1919 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
1920 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
1921 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
1922 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
1923 Andersen, Tom Gundersen, Torstein Husebø
1925 — Berlin, 2015-09-08
1929 * machinectl gained a new verb 'shell' which opens a fresh
1930 shell on the target container or the host. It is similar to
1931 the existing 'login' command of machinectl, but spawns the
1932 shell directly without prompting for username or
1933 password. The pseudo machine '.host' now refers to the local
1934 host and is used by default. Hence, 'machinectl shell' can
1935 be used as replacement for 'su -' which spawns a session as
1936 a fresh systemd unit in a way that is fully isolated from
1937 the originating session.
1939 * systemd-networkd learned to cope with private-zone DHCP
1940 options and allows other programs to query the values.
1942 * SELinux access control when enabling/disabling units is no
1943 longer enforced with this release. The previous
1944 implementation was incorrect, and a new corrected
1945 implementation is not yet available. As unit file operations
1946 are still protected via PolicyKit and D-Bus policy this is
1947 not a security problem. Yet, distributions which care about
1948 optimal SELinux support should probably not stabilize on
1951 * sd-bus gained support for matches of type "arg0has=", that
1952 test for membership of strings in string arrays sent in bus
1955 * systemd-resolved now dumps the contents of its DNS and LLMNR
1956 caches to the logs on reception of the SIGUSR1 signal. This
1957 is useful to debug DNS behaviour.
1959 * The coredumpctl tool gained a new --directory= option to
1960 operate on journal files in a specific directory.
1962 * "systemctl reboot" and related commands gained a new
1963 "--message=" option which may be used to set a free-text
1964 wall message when shutting down or rebooting the
1965 system. This message is also logged, which is useful for
1966 figuring out the reason for a reboot or shutdown a
1969 * The "systemd-resolve-host" tool's -i switch now takes
1970 network interface numbers as alternative to interface names.
1972 * A new unit file setting for services has been introduced:
1973 UtmpMode= allows configuration of how precisely systemd
1974 handles utmp and wtmp entries for the service if this is
1975 enabled. This allows writing services that appear similar to
1976 user sessions in the output of the "w", "who", "last" and
1979 * systemd-resolved will now locally synthesize DNS resource
1980 records for the "localhost" and "gateway" domains as well as
1981 the local hostname. This should ensure that clients querying
1982 RRs via resolved will get similar results as those going via
1983 NSS, if nss-myhostname is enabled.
1985 Contributions from: Alastair Hughes, Alex Crawford, Daniel
1986 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
1987 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
1988 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
1989 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
1990 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
1991 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
1992 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
1993 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
1994 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
1995 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
1996 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
1998 — Berlin, 2015-08-27
2002 * The systemd-efi-boot-generator functionality was merged into
2003 systemd-gpt-auto-generator.
2005 * systemd-networkd now supports Group Policy for vxlan
2006 devices. It can be enabled via the new boolean configuration
2007 option called 'GroupPolicyExtension='.
2009 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
2010 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
2011 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
2013 — Berlin, 2015-07-31
2017 * The python-systemd code has been removed from the systemd repository.
2018 A new repository has been created which accommodates the code from
2019 now on, and we kindly ask distributions to create a separate package
2020 for this: https://github.com/systemd/python-systemd
2022 * The systemd daemon will now reload its main configuration
2023 (/etc/systemd/system.conf) on daemon-reload.
2025 * sd-dhcp now exposes vendor specific extensions via
2026 sd_dhcp_lease_get_vendor_specific().
2028 * systemd-networkd gained a number of new configuration options.
2030 - A new boolean configuration option for TAP devices called
2031 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
2032 device, thus allowing to send and receive GSO packets.
2034 - A new tunnel configuration option called 'CopyDSCP='.
2035 If enabled, the DSCP field of ip6 tunnels is copied into the
2036 decapsulated packet.
2038 - A set of boolean bridge configuration options were added.
2039 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
2040 and 'UnicastFlood=' are now parsed by networkd and applied to the
2041 respective bridge link device via the respective IFLA_BRPORT_*
2044 - A new string configuration option to override the hostname sent
2045 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
2046 is true, networkd will use the configured hostname instead of the
2047 system hostname when sending DHCP requests.
2049 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
2050 networkd will configure the IPv6 flow-label of the tunnel device
2051 according to RFC2460.
2053 - The 'macvtap' virtual network devices are now supported, similar to
2054 the already supported 'macvlan' devices.
2056 * systemd-resolved now implements RFC5452 to improve resilience against
2057 cache poisoning. Additionally, source port randomization is enabled
2058 by default to further protect against DNS spoofing attacks.
2060 * nss-mymachines now supports translating UIDs and GIDs of running
2061 containers with user-namespaces enabled. If a container 'foo'
2062 translates a host uid 'UID' to the container uid 'TUID', then
2063 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
2064 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
2065 mapped as 'vg-foo-TGID'.
2067 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
2068 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
2069 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
2070 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
2071 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
2072 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
2073 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
2074 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
2075 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
2076 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
2078 — Berlin, 2015-07-29
2082 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
2083 There are no known issues with current sysfs, and udev does not need
2084 or should be used to work around such bugs.
2086 * udev does no longer enable USB HID power management. Several reports
2087 indicate, that some devices cannot handle that setting.
2089 * The udev accelerometer helper was removed. The functionality
2090 is now fully included in iio-sensor-proxy. But this means,
2091 older iio-sensor-proxy versions will no longer provide
2092 accelerometer/orientation data with this systemd version.
2093 Please upgrade iio-sensor-proxy to version 1.0.
2095 * networkd gained a new configuration option IPv6PrivacyExtensions=
2096 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
2097 for Stateless Address") on selected networks.
2099 * For the sake of fewer build-time dependencies and less code in the
2100 main repository, the python bindings are about to be removed in the
2101 next release. A new repository has been created which accommodates
2102 the code from now on, and we kindly ask distributions to create a
2103 separate package for this. The removal will take place in v223.
2105 https://github.com/systemd/python-systemd
2107 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
2108 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
2109 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
2110 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
2111 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
2112 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
2113 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
2114 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
2115 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
2116 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
2118 — Berlin, 2015-07-07
2122 * The sd-bus.h and sd-event.h APIs have now been declared
2123 stable and have been added to the official interface of
2124 libsystemd.so. sd-bus implements an alternative D-Bus client
2125 library, that is relatively easy to use, very efficient and
2126 supports both classic D-Bus as well as kdbus as transport
2127 backend. sd-event is a generic event loop abstraction that
2128 is built around Linux epoll, but adds features such as event
2129 prioritization or efficient timer handling. Both APIs are good
2130 choices for C programs looking for a bus and/or event loop
2131 implementation that is minimal and does not have to be
2132 portable to other kernels.
2134 * kdbus support is no longer compile-time optional. It is now
2135 always built-in. However, it can still be disabled at
2136 runtime using the kdbus=0 kernel command line setting, and
2137 that setting may be changed to default to off, by specifying
2138 --disable-kdbus at build-time. Note though that the kernel
2139 command line setting has no effect if the kdbus.ko kernel
2140 module is not installed, in which case kdbus is (obviously)
2141 also disabled. We encourage all downstream distributions to
2142 begin testing kdbus by adding it to the kernel images in the
2143 development distributions, and leaving kdbus support in
2146 * The minimal required util-linux version has been bumped to
2149 * Support for chkconfig (--enable-chkconfig) was removed in
2150 favor of calling an abstraction tool
2151 /lib/systemd/systemd-sysv-install. This needs to be
2152 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
2153 in README for details.
2155 * If there's a systemd unit and a SysV init script for the
2156 same service name, and the user executes "systemctl enable"
2157 for it (or a related call), then this will now enable both
2158 (or execute the related operation on both), not just the
2161 * The libudev API documentation has been converted from gtkdoc
2164 * gudev has been removed from the systemd tree, it is now an
2167 * The systemd-cgtop tool learnt a new --raw switch to generate
2168 "raw" (machine parsable) output.
2170 * networkd's IPForwarding= .network file setting learnt the
2171 new setting "kernel", which ensures that networkd does not
2172 change the IP forwarding sysctl from the default kernel
2175 * The systemd-logind bus API now exposes a new boolean
2176 property "Docked" that reports whether logind considers the
2177 system "docked", i.e. connected to a docking station or not.
2179 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
2180 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
2181 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
2182 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
2183 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
2184 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
2185 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
2186 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
2187 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
2188 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
2189 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
2190 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
2191 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
2192 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
2193 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
2194 Fink, Zbigniew Jędrzejewski-Szmek
2196 — Berlin, 2015-06-19
2200 * The gudev library has been extracted into a separate repository
2201 available at: https://git.gnome.org/browse/libgudev/
2202 It is now managed as part of the Gnome project. Distributions
2203 are recommended to pass --disable-gudev to systemd and use
2204 gudev from the Gnome project instead. gudev is still included
2205 in systemd, for now. It will be removed soon, though. Please
2206 also see the announcement-thread on systemd-devel:
2207 http://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
2209 * systemd now exposes a CPUUsageNSec= property for each
2210 service unit on the bus, that contains the overall consumed
2211 CPU time of a service (the sum of what each process of the
2212 service consumed). This value is only available if
2213 CPUAccounting= is turned on for a service, and is then shown
2214 in the "systemctl status" output.
2216 * Support for configuring alternative mappings of the old SysV
2217 runlevels to systemd targets has been removed. They are now
2218 hardcoded in a way that runlevels 2, 3, 4 all map to
2219 multi-user.target and 5 to graphical.target (which
2220 previously was already the default behaviour).
2222 * The auto-mounter logic gained support for mount point
2223 expiry, using a new TimeoutIdleSec= setting in .automount
2224 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
2226 * The EFI System Partition (ESP) as mounted to /boot by
2227 systemd-efi-boot-generator will now be unmounted
2228 automatically after 2 minutes of not being used. This should
2229 minimize the risk of ESP corruptions.
2231 * New /etc/fstab options x-systemd.requires= and
2232 x-systemd.requires-mounts-for= are now supported to express
2233 additional dependencies for mounts. This is useful for
2234 journalling file systems that support external journal
2235 devices or overlay file systems that require underlying file
2236 systems to be mounted.
2238 * systemd does not support direct live-upgrades (via systemctl
2239 daemon-reexec) from versions older than v44 anymore. As no
2240 distribution we are aware of shipped such old versions in a
2241 stable release this should not be problematic.
2243 * When systemd forks off a new per-connection service instance
2244 it will now set the $REMOTE_ADDR environment variable to the
2245 remote IP address, and $REMOTE_PORT environment variable to
2246 the remote IP port. This behaviour is similar to the
2247 corresponding environment variables defined by CGI.
2249 * systemd-networkd gained support for uplink failure
2250 detection. The BindCarrier= option allows binding interface
2251 configuration dynamically to the link sense of other
2252 interfaces. This is useful to achieve behaviour like in
2255 * systemd-networkd gained support for configuring the DHCP
2256 client identifier to use when requesting leases.
2258 * systemd-networkd now has a per-network UseNTP= option to
2259 configure whether NTP server information acquired via DHCP
2260 is passed on to services like systemd-timesyncd.
2262 * systemd-networkd gained support for vti6 tunnels.
2264 * Note that systemd-networkd manages the sysctl variable
2265 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
2266 it is configured for since v219. The variable controls IP
2267 forwarding, and is a per-interface alternative to the global
2268 /proc/sys/net/ipv[46]/ip_forward. This setting is
2269 configurable in the IPForward= option, which defaults to
2270 "no". This means if networkd is used for an interface it is
2271 no longer sufficient to set the global sysctl option to turn
2272 on IP forwarding! Instead, the .network file option
2273 IPForward= needs to be turned on! Note that the
2274 implementation of this behaviour was broken in v219 and has
2277 * Many bonding and vxlan options are now configurable in
2280 * systemd-nspawn gained a new --property= setting to set unit
2281 properties for the container scope. This is useful for
2282 setting resource parameters (e.g. "CPUShares=500") on
2283 containers started from the command line.
2285 * systemd-nspawn gained a new --private-users= switch to make
2286 use of user namespacing available on recent Linux kernels.
2288 * systemd-nspawn may now be called as part of a shell pipeline
2289 in which case the pipes used for stdin and stdout are passed
2290 directly to the process invoked in the container, without
2291 indirection via a pseudo tty.
2293 * systemd-nspawn gained a new switch to control the UNIX
2294 signal to use when killing the init process of the container
2297 * systemd-nspawn gained a new --overlay= switch for mounting
2298 overlay file systems into the container using the new kernel
2301 * When a container image is imported via systemd-importd and
2302 the host file system is not btrfs, a loopback block device
2303 file is created in /var/lib/machines.raw with a btrfs file
2304 system inside. It is then mounted to /var/lib/machines to
2305 enable btrfs features for container management. The loopback
2306 file and btrfs file system is grown as needed when container
2307 images are imported via systemd-importd.
2309 * systemd-machined/systemd-importd gained support for btrfs
2310 quota, to enforce container disk space limits on disk. This
2311 is exposed in "machinectl set-limit".
2313 * systemd-importd now can import containers from local .tar,
2314 .raw and .qcow2 images, and export them to .tar and .raw. It
2315 can also import dkr v2 images now from the network (on top
2318 * systemd-importd gained support for verifying downloaded
2319 images with gpg2 (previously only gpg1 was supported).
2321 * systemd-machined, systemd-logind, systemd: most bus calls
2322 are now accessible to unprivileged processes via
2323 PolicyKit. Also, systemd-logind will now allow users to kill
2324 their own sessions without further privileges or
2327 * systemd-shutdownd has been removed. This service was
2328 previously responsible for implementing scheduled shutdowns
2329 as exposed in /usr/bin/shutdown's time parameter. This
2330 functionality has now been moved into systemd-logind and is
2331 accessible via a bus interface.
2333 * "systemctl reboot" gained a new switch --firmware-setup that
2334 can be used to reboot into the EFI firmware setup, if that
2335 is available. systemd-logind now exposes an API on the bus
2336 to trigger such reboots, in case graphical desktop UIs want
2337 to cover this functionality.
2339 * "systemctl enable", "systemctl disable" and "systemctl mask"
2340 now support a new "--now" switch. If specified the units
2341 that are enabled will also be started, and the ones
2342 disabled/masked also stopped.
2344 * The Gummiboot EFI boot loader tool has been merged into
2345 systemd, and renamed to "systemd-boot". The bootctl tool has been
2346 updated to support systemd-boot.
2348 * An EFI kernel stub has been added that may be used to create
2349 kernel EFI binaries that contain not only the actual kernel,
2350 but also an initrd, boot splash, command line and OS release
2351 information. This combined binary can then be signed as a
2352 single image, so that the firmware can verify it all in one
2353 step. systemd-boot has special support for EFI binaries created
2354 like this and can extract OS release information from them
2355 and show them in the boot menu. This functionality is useful
2356 to implement cryptographically verified boot schemes.
2358 * Optional support has been added to systemd-fsck to pass
2359 fsck's progress report to an AF_UNIX socket in the file
2362 * udev will no longer create device symlinks for all block
2363 devices by default. A blacklist for excluding special block
2364 devices from this logic has been turned into a whitelist
2365 that requires picking block devices explicitly that require
2368 * A new (currently still internal) API sd-device.h has been
2369 added to libsystemd. This modernized API is supposed to
2370 replace libudev eventually. In fact, already much of libudev
2371 is now just a wrapper around sd-device.h.
2373 * A new hwdb database for storing metadata about pointing
2374 stick devices has been added.
2376 * systemd-tmpfiles gained support for setting file attributes
2377 similar to the "chattr" tool with new 'h' and 'H' lines.
2379 * systemd-journald will no longer unconditionally set the
2380 btrfs NOCOW flag on new journal files. This is instead done
2381 with tmpfiles snippet using the new 'h' line type. This
2382 allows easy disabling of this logic, by masking the
2383 journal-nocow.conf tmpfiles file.
2385 * systemd-journald will now translate audit message types to
2386 human readable identifiers when writing them to the
2387 journal. This should improve readability of audit messages.
2389 * The LUKS logic gained support for the offset= and skip=
2390 options in /etc/crypttab, as previously implemented by
2393 * /usr/lib/os-release gained a new optional field VARIANT= for
2394 distributions that support multiple variants (such as a
2395 desktop edition, a server edition, ...)
2397 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
2398 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
2399 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
2400 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
2401 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
2402 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
2403 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
2404 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
2405 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
2406 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
2407 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
2408 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
2409 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
2410 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
2411 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
2412 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
2413 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
2414 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
2415 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
2416 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
2417 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
2418 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
2419 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
2420 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
2421 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
2422 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
2423 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
2425 — Berlin, 2015-05-22
2429 * Introduce a new API "sd-hwdb.h" for querying the hardware
2430 metadata database. With this minimal interface one can query
2431 and enumerate the udev hwdb, decoupled from the old libudev
2432 library. libudev's interface for this is now only a wrapper
2433 around sd-hwdb. A new tool systemd-hwdb has been added to
2434 interface with and update the database.
2436 * When any of systemd's tools copies files (for example due to
2437 tmpfiles' C lines) a btrfs reflink will attempted first,
2438 before bytewise copying is done.
2440 * systemd-nspawn gained a new --ephemeral switch. When
2441 specified a btrfs snapshot is taken of the container's root
2442 directory, and immediately removed when the container
2443 terminates again. Thus, a container can be started whose
2444 changes never alter the container's root directory, and are
2445 lost on container termination. This switch can also be used
2446 for starting a container off the root file system of the
2447 host without affecting the host OS. This switch is only
2448 available on btrfs file systems.
2450 * systemd-nspawn gained a new --template= switch. It takes the
2451 path to a container tree to use as template for the tree
2452 specified via --directory=, should that directory be
2453 missing. This allows instantiating containers dynamically,
2454 on first run. This switch is only available on btrfs file
2457 * When a .mount unit refers to a mount point on which multiple
2458 mounts are stacked, and the .mount unit is stopped all of
2459 the stacked mount points will now be unmounted until no
2460 mount point remains.
2462 * systemd now has an explicit notion of supported and
2463 unsupported unit types. Jobs enqueued for unsupported unit
2464 types will now fail with an "unsupported" error code. More
2465 specifically .swap, .automount and .device units are not
2466 supported in containers, .busname units are not supported on
2467 non-kdbus systems. .swap and .automount are also not
2468 supported if their respective kernel compile time options
2471 * machinectl gained support for two new "copy-from" and
2472 "copy-to" commands for copying files from a running
2473 container to the host or vice versa.
2475 * machinectl gained support for a new "bind" command to bind
2476 mount host directories into local containers. This is
2477 currently only supported for nspawn containers.
2479 * networkd gained support for configuring bridge forwarding
2480 database entries (fdb) from .network files.
2482 * A new tiny daemon "systemd-importd" has been added that can
2483 download container images in tar, raw, qcow2 or dkr formats,
2484 and make them available locally in /var/lib/machines, so
2485 that they can run as nspawn containers. The daemon can GPG
2486 verify the downloads (not supported for dkr, since it has no
2487 provisions for verifying downloads). It will transparently
2488 decompress bz2, xz, gzip compressed downloads if necessary,
2489 and restore sparse files on disk. The daemon uses privilege
2490 separation to ensure the actual download logic runs with
2491 fewer privileges than the daemon itself. machinectl has
2492 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
2493 make the functionality of importd available to the
2494 user. With this in place the Fedora and Ubuntu "Cloud"
2495 images can be downloaded and booted as containers unmodified
2496 (the Fedora images lack the appropriate GPG signature files
2497 currently, so they cannot be verified, but this will change
2498 soon, hopefully). Note that downloading images is currently
2499 only fully supported on btrfs.
2501 * machinectl is now able to list container images found in
2502 /var/lib/machines, along with some metadata about sizes of
2503 disk and similar. If the directory is located on btrfs and
2504 quota is enabled, this includes quota display. A new command
2505 "image-status" has been added that shows additional
2506 information about images.
2508 * machinectl is now able to clone container images
2509 efficiently, if the underlying file system (btrfs) supports
2510 it, with the new "machinectl clone" command. It also
2511 gained commands for renaming and removing images, as well as
2512 marking them read-only or read-write (supported also on
2513 legacy file systems).
2515 * networkd gained support for collecting LLDP network
2516 announcements, from hardware that supports this. This is
2517 shown in networkctl output.
2519 * systemd-run gained support for a new -t (--pty) switch for
2520 invoking a binary on a pty whose input and output is
2521 connected to the invoking terminal. This allows executing
2522 processes as system services while interactively
2523 communicating with them via the terminal. Most interestingly
2524 this is supported across container boundaries. Invoking
2525 "systemd-run -t /bin/bash" is an alternative to running a
2526 full login session, the difference being that the former
2527 will not register a session, nor go through the PAM session
2530 * tmpfiles gained support for a new "v" line type for creating
2531 btrfs subvolumes. If the underlying file system is a legacy
2532 file system, this automatically degrades to creating a
2533 normal directory. Among others /var/lib/machines is now
2534 created like this at boot, should it be missing.
2536 * The directory /var/lib/containers/ has been deprecated and
2537 been replaced by /var/lib/machines. The term "machines" has
2538 been used in the systemd context as generic term for both
2539 VMs and containers, and hence appears more appropriate for
2540 this, as the directory can also contain raw images bootable
2543 * systemd-nspawn when invoked with -M but without --directory=
2544 or --image= is now capable of searching for the container
2545 root directory, subvolume or disk image automatically, in
2546 /var/lib/machines. systemd-nspawn@.service has been updated
2547 to make use of this, thus allowing it to be used for raw
2550 * A new machines.target unit has been introduced that is
2551 supposed to group all containers/VMs invoked as services on
2552 the system. systemd-nspawn@.service has been updated to
2553 integrate with that.
2555 * machinectl gained a new "start" command, for invoking a
2556 container as a service. "machinectl start foo" is mostly
2557 equivalent to "systemctl start systemd-nspawn@foo.service",
2558 but handles escaping in a nicer way.
2560 * systemd-nspawn will now mount most of the cgroupfs tree
2561 read-only into each container, with the exception of the
2562 container's own subtree in the name=systemd hierarchy.
2564 * journald now sets the special FS_NOCOW file flag for its
2565 journal files. This should improve performance on btrfs, by
2566 avoiding heavy fragmentation when journald's write-pattern
2567 is used on COW file systems. It degrades btrfs' data
2568 integrity guarantees for the files to the same levels as for
2569 ext3/ext4 however. This should be OK though as journald does
2570 its own data integrity checks and all its objects are
2571 checksummed on disk. Also, journald should handle btrfs disk
2572 full events a lot more gracefully now, by processing SIGBUS
2573 errors, and not relying on fallocate() anymore.
2575 * When journald detects that journal files it is writing to
2576 have been deleted it will immediately start new journal
2579 * systemd now provides a way to store file descriptors
2580 per-service in PID 1. This is useful for daemons to ensure
2581 that fds they require are not lost during a daemon
2582 restart. The fds are passed to the daemon on the next
2583 invocation in the same way socket activation fds are
2584 passed. This is now used by journald to ensure that the
2585 various sockets connected to all the system's stdout/stderr
2586 are not lost when journald is restarted. File descriptors
2587 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
2588 an extension to sd_notify(). Note that a limit is enforced
2589 on the number of fds a service can store in PID 1, and it
2590 defaults to 0, so that no fds may be stored, unless this is
2591 explicitly turned on.
2593 * The default TERM variable to use for units connected to a
2594 terminal, when no other value is explicitly is set is now
2595 vt220 rather than vt102. This should be fairly safe still,
2596 but allows PgUp/PgDn work.
2598 * The /etc/crypttab option header= as known from Debian is now
2601 * "loginctl user-status" and "loginctl session-status" will
2602 now show the last 10 lines of log messages of the
2603 user/session following the status output. Similar,
2604 "machinectl status" will show the last 10 log lines
2605 associated with a virtual machine or container
2606 service. (Note that this is usually not the log messages
2607 done in the VM/container itself, but simply what the
2608 container manager logs. For nspawn this includes all console
2611 * "loginctl session-status" without further argument will now
2612 show the status of the session of the caller. Similar,
2613 "lock-session", "unlock-session", "activate",
2614 "enable-linger", "disable-linger" may now be called without
2615 session/user parameter in which case they apply to the
2616 caller's session/user.
2618 * An X11 session scriptlet is now shipped that uploads
2619 $DISPLAY and $XAUTHORITY into the environment of the systemd
2620 --user daemon if a session begins. This should improve
2621 compatibility with X11 enabled applications run as systemd
2624 * Generators are now subject to masking via /etc and /run, the
2625 same way as unit files.
2627 * networkd .network files gained support for configuring
2628 per-link IPv4/IPv6 packet forwarding as well as IPv4
2629 masquerading. This is by default turned on for veth links to
2630 containers, as registered by systemd-nspawn. This means that
2631 nspawn containers run with --network-veth will now get
2632 automatic routed access to the host's networks without any
2633 further configuration or setup, as long as networkd runs on
2636 * systemd-nspawn gained the --port= (-p) switch to expose TCP
2637 or UDP posts of a container on the host. With this in place
2638 it is possible to run containers with private veth links
2639 (--network-veth), and have their functionality exposed on
2640 the host as if their services were running directly on the
2643 * systemd-nspawn's --network-veth switch now gained a short
2644 version "-n", since with the changes above it is now truly
2645 useful out-of-the-box. The systemd-nspawn@.service has been
2646 updated to make use of it too by default.
2648 * systemd-nspawn will now maintain a per-image R/W lock, to
2649 ensure that the same image is not started more than once
2650 writable. (It's OK to run an image multiple times
2651 simultaneously in read-only mode.)
2653 * systemd-nspawn's --image= option is now capable of
2654 dissecting and booting MBR and GPT disk images that contain
2655 only a single active Linux partition. Previously it
2656 supported only GPT disk images with proper GPT type
2657 IDs. This allows running cloud images from major
2658 distributions directly with systemd-nspawn, without
2661 * In addition to collecting mouse dpi data in the udev
2662 hardware database, there's now support for collecting angle
2663 information for mouse scroll wheels. The database is
2664 supposed to guarantee similar scrolling behavior on mice
2665 that it knows about. There's also support for collecting
2666 information about Touchpad types.
2668 * udev's input_id built-in will now also collect touch screen
2669 dimension data and attach it to probed devices.
2671 * /etc/os-release gained support for a Distribution Privacy
2674 * networkd gained support for creating "ipvlan", "gretap",
2675 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
2677 * systemd-tmpfiles gained support for "a" lines for setting
2680 * systemd-nspawn will now mount /tmp in the container to
2681 tmpfs, automatically.
2683 * systemd now exposes the memory.usage_in_bytes cgroup
2684 attribute and shows it for each service in the "systemctl
2685 status" output, if available.
2687 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
2688 immediate reboot is triggered. This useful if shutdown is
2689 hung and is unable to complete, to expedite the
2690 operation. Note that this kind of reboot will still unmount
2691 all file systems, and hence should not result in fsck being
2694 * A .device unit for an optical block device will now be
2695 considered active only when a medium is in the drive. Also,
2696 mount units are now bound to their backing devices thus
2697 triggering automatic unmounting when devices become
2698 unavailable. With this in place systemd will now
2699 automatically unmount left-over mounts when a CD-ROM is
2700 ejected or an USB stick is yanked from the system.
2702 * networkd-wait-online now has support for waiting for
2703 specific interfaces only (with globbing), and for giving up
2704 after a configurable timeout.
2706 * networkd now exits when idle. It will be automatically
2707 restarted as soon as interfaces show up, are removed or
2708 change state. networkd will stay around as long as there is
2709 at least one DHCP state machine or similar around, that keep
2712 * networkd may now configure IPv6 link-local addressing in
2713 addition to IPv4 link-local addressing.
2715 * The IPv6 "token" for use in SLAAC may now be configured for
2716 each .network interface in networkd.
2718 * Routes configured with networkd may now be assigned a scope
2721 * networkd's [Match] sections now support globbing and lists
2722 of multiple space-separated matches per item.
2724 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
2725 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
2726 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
2727 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
2728 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
2729 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
2730 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
2731 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
2732 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
2733 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
2734 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
2735 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
2736 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
2737 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
2738 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
2739 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
2740 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
2741 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
2742 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
2743 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
2744 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
2745 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
2746 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
2747 Hoffmann, Zbigniew Jędrzejewski-Szmek
2749 — Berlin, 2015-02-16
2753 * When querying unit file enablement status (for example via
2754 "systemctl is-enabled"), a new state "indirect" is now known
2755 which indicates that a unit might not be enabled itself, but
2756 another unit listed in its Also= setting might be.
2758 * Similar to the various existing ConditionXYZ= settings for
2759 units, there are now matching AssertXYZ= settings. While
2760 failing conditions cause a unit to be skipped, but its job
2761 to succeed, failing assertions declared like this will cause
2762 a unit start operation and its job to fail.
2764 * hostnamed now knows a new chassis type "embedded".
2766 * systemctl gained a new "edit" command. When used on a unit
2767 file, this allows extending unit files with .d/ drop-in
2768 configuration snippets or editing the full file (after
2769 copying it from /usr/lib to /etc). This will invoke the
2770 user's editor (as configured with $EDITOR), and reload the
2771 modified configuration after editing.
2773 * "systemctl status" now shows the suggested enablement state
2774 for a unit, as declared in the (usually vendor-supplied)
2775 system preset files.
2777 * nss-myhostname will now resolve the single-label host name
2778 "gateway" to the locally configured default IP routing
2779 gateways, ordered by their metrics. This assigns a stable
2780 name to the used gateways, regardless which ones are
2781 currently configured. Note that the name will only be
2782 resolved after all other name sources (if nss-myhostname is
2783 configured properly) and should hence not negatively impact
2784 systems that use the single-label host name "gateway" in
2787 * systemd-inhibit now allows filtering by mode when listing
2790 * Scope and service units gained a new "Delegate" boolean
2791 property, which, when set, allows processes running inside the
2792 unit to further partition resources. This is primarily
2793 useful for systemd user instances as well as container
2796 * journald will now pick up audit messages directly from
2797 the kernel, and log them like any other log message. The
2798 audit fields are split up and fully indexed. This means that
2799 journalctl in many ways is now a (nicer!) alternative to
2800 ausearch, the traditional audit client. Note that this
2801 implements only a minimal audit client. If you want the
2802 special audit modes like reboot-on-log-overflow, please use
2803 the traditional auditd instead, which can be used in
2804 parallel to journald.
2806 * The ConditionSecurity= unit file option now understands the
2807 special string "audit" to check whether auditing is
2810 * journalctl gained two new commands --vacuum-size= and
2811 --vacuum-time= to delete old journal files until the
2812 remaining ones take up no more than the specified size on disk,
2813 or are not older than the specified time.
2815 * A new, native PPPoE library has been added to sd-network,
2816 systemd's library of light-weight networking protocols. This
2817 library will be used in a future version of networkd to
2818 enable PPPoE communication without an external pppd daemon.
2820 * The busctl tool now understands a new "capture" verb that
2821 works similar to "monitor", but writes a packet capture
2822 trace to STDOUT that can be redirected to a file which is
2823 compatible with libcap's capture file format. This can then
2824 be loaded in Wireshark and similar tools to inspect bus
2827 * The busctl tool now understands a new "tree" verb that shows
2828 the object trees of a specific service on the bus, or of all
2831 * The busctl tool now understands a new "introspect" verb that
2832 shows all interfaces and members of objects on the bus,
2833 including their signature and values. This is particularly
2834 useful to get more information about bus objects shown by
2835 the new "busctl tree" command.
2837 * The busctl tool now understands new verbs "call",
2838 "set-property" and "get-property" for invoking bus method
2839 calls, setting and getting bus object properties in a
2842 * busctl gained a new --augment-creds= argument that controls
2843 whether the tool shall augment credential information it
2844 gets from the bus with data from /proc, in a possibly
2847 * nspawn's --link-journal= switch gained two new values
2848 "try-guest" and "try-host" that work like "guest" and
2849 "host", but do not fail if the host has no persistent
2850 journalling enabled. -j is now equivalent to
2851 --link-journal=try-guest.
2853 * macvlan network devices created by nspawn will now have
2854 stable MAC addresses.
2856 * A new SmackProcessLabel= unit setting has been added, which
2857 controls the SMACK security label processes forked off by
2858 the respective unit shall use.
2860 * If compiled with --enable-xkbcommon, systemd-localed will
2861 verify x11 keymap settings by compiling the given keymap. It
2862 will spew out warnings if the compilation fails. This
2863 requires libxkbcommon to be installed.
2865 * When a coredump is collected, a larger number of metadata
2866 fields is now collected and included in the journal records
2867 created for it. More specifically, control group membership,
2868 environment variables, memory maps, working directory,
2869 chroot directory, /proc/$PID/status, and a list of open file
2870 descriptors is now stored in the log entry.
2872 * The udev hwdb now contains DPI information for mice. For
2875 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
2877 * All systemd programs that read standalone configuration
2878 files in /etc now also support a corresponding series of
2879 .conf.d configuration directories in /etc/, /run/,
2880 /usr/local/lib/, /usr/lib/, and (if configured with
2881 --enable-split-usr) /lib/. In particular, the following
2882 configuration files now have corresponding configuration
2883 directories: system.conf user.conf, logind.conf,
2884 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
2885 resolved.conf, timesyncd.conf, journal-remote.conf, and
2886 journal-upload.conf. Note that distributions should use the
2887 configuration directories in /usr/lib/; the directories in
2888 /etc/ are reserved for the system administrator.
2890 * systemd-rfkill will no longer take the rfkill device name
2891 into account when storing rfkill state on disk, as the name
2892 might be dynamically assigned and not stable. Instead, the
2893 ID_PATH udev variable combined with the rfkill type (wlan,
2894 bluetooth, ...) is used.
2896 * A new service systemd-machine-id-commit.service has been
2897 added. When used on systems where /etc is read-only during
2898 boot, and /etc/machine-id is not initialized (but an empty
2899 file), this service will copy the temporary machine ID
2900 created as replacement into /etc after the system is fully
2901 booted up. This is useful for systems that are freshly
2902 installed with a non-initialized machine ID, but should get
2903 a fixed machine ID for subsequent boots.
2905 * networkd's .netdev files now provide a large set of
2906 configuration parameters for VXLAN devices. Similarly, the
2907 bridge port cost parameter is now configurable in .network
2908 files. There's also new support for configuring IP source
2909 routing. networkd .link files gained support for a new
2910 OriginalName= match that is useful to match against the
2911 original interface name the kernel assigned. .network files
2912 may include MTU= and MACAddress= fields for altering the MTU
2913 and MAC address while being connected to a specific network
2916 * The LUKS logic gained supported for configuring
2917 UUID-specific key files. There's also new support for naming
2918 LUKS device from the kernel command line, using the new
2919 luks.name= argument.
2921 * Timer units may now be transiently created via the bus API
2922 (this was previously already available for scope and service
2923 units). In addition it is now possible to create multiple
2924 transient units at the same time with a single bus call. The
2925 "systemd-run" tool has been updated to make use of this for
2926 running commands on a specified time, in at(1)-style.
2928 * tmpfiles gained support for "t" lines, for assigning
2929 extended attributes to files. Among other uses this may be
2930 used to assign SMACK labels to files.
2932 Contributions from: Alin Rauta, Alison Chaiken, Andrej
2933 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
2934 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
2935 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
2936 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
2937 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
2938 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
2939 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
2940 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
2941 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
2942 Michael Biebl, Michael Chapman, Michael Marineau, Michal
2943 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
2944 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
2945 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
2946 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
2947 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
2948 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
2949 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
2951 — Berlin, 2014-12-10
2955 * journalctl gained the new options -t/--identifier= to match
2956 on the syslog identifier (aka "tag"), as well as --utc to
2957 show log timestamps in the UTC timezone. journalctl now also
2958 accepts -n/--lines=all to disable line capping in a pager.
2960 * journalctl gained a new switch, --flush, that synchronously
2961 flushes logs from /run/log/journal to /var/log/journal if
2962 persistent storage is enabled. systemd-journal-flush.service
2963 now waits until the operation is complete.
2965 * Services can notify the manager before they start a reload
2966 (by sending RELOADING=1) or shutdown (by sending
2967 STOPPING=1). This allows the manager to track and show the
2968 internal state of daemons and closes a race condition when
2969 the process is still running but has closed its D-Bus
2972 * Services with Type=oneshot do not have to have any ExecStart
2975 * User units are now loaded also from
2976 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
2977 /run/systemd/user directory that was already previously
2978 supported, but is under the control of the user.
2980 * Job timeouts (i.e. time-outs on the time a job that is
2981 queued stays in the run queue) can now optionally result in
2982 immediate reboot or power-off actions (JobTimeoutAction= and
2983 JobTimeoutRebootArgument=). This is useful on ".target"
2984 units, to limit the maximum time a target remains
2985 undispatched in the run queue, and to trigger an emergency
2986 operation in such a case. This is now used by default to
2987 turn off the system if boot-up (as defined by everything in
2988 basic.target) hangs and does not complete for at least
2989 15min. Also, if power-off or reboot hang for at least 30min
2990 an immediate power-off/reboot operation is triggered. This
2991 functionality is particularly useful to increase reliability
2992 on embedded devices, but also on laptops which might
2993 accidentally get powered on when carried in a backpack and
2994 whose boot stays stuck in a hard disk encryption passphrase
2997 * systemd-logind can be configured to also handle lid switch
2998 events even when the machine is docked or multiple displays
2999 are attached (HandleLidSwitchDocked= option).
3001 * A helper binary and a service have been added which can be
3002 used to resume from hibernation in the initramfs. A
3003 generator will parse the resume= option on the kernel
3004 command line to trigger resume.
3006 * A user console daemon systemd-consoled has been
3007 added. Currently, it is a preview, and will so far open a
3008 single terminal on each session of the user marked as
3009 Desktop=systemd-console.
3011 * Route metrics can be specified for DHCP routes added by
3014 * The SELinux context of socket-activated services can be set
3015 from the information provided by the networking stack
3016 (SELinuxContextFromNet= option).
3018 * Userspace firmware loading support has been removed and
3019 the minimum supported kernel version is thus bumped to 3.7.
3021 * Timeout for udev workers has been increased from 1 to 3
3022 minutes, but a warning will be printed after 1 minute to
3023 help diagnose kernel modules that take a long time to load.
3025 * Udev rules can now remove tags on devices with TAG-="foobar".
3027 * systemd's readahead implementation has been removed. In many
3028 circumstances it didn't give expected benefits even for
3029 rotational disk drives and was becoming less relevant in the
3030 age of SSDs. As none of the developers has been using
3031 rotating media anymore, and nobody stepped up to actively
3032 maintain this component of systemd it has now been removed.
3034 * Swap units can use Options= to specify discard options.
3035 Discard options specified for swaps in /etc/fstab are now
3038 * Docker containers are now detected as a separate type of
3041 * The Password Agent protocol gained support for queries where
3042 the user input is shown, useful e.g. for user names.
3043 systemd-ask-password gained a new --echo option to turn that
3046 * The default sysctl.d/ snippets will now set:
3048 net.core.default_qdisc = fq_codel
3050 This selects Fair Queuing Controlled Delay as the default
3051 queuing discipline for network interfaces. fq_codel helps
3052 fight the network bufferbloat problem. It is believed to be
3053 a good default with no tuning required for most workloads.
3054 Downstream distributions may override this choice. On 10Gbit
3055 servers that do not do forwarding, "fq" may perform better.
3056 Systems without a good clocksource should use "pfifo_fast".
3058 * If kdbus is enabled during build a new option BusPolicy= is
3059 available for service units, that allows locking all service
3060 processes into a stricter bus policy, in order to limit
3061 access to various bus services, or even hide most of them
3062 from the service's view entirely.
3064 * networkctl will now show the .network and .link file
3065 networkd has applied to a specific interface.
3067 * sd-login gained a new API call sd_session_get_desktop() to
3068 query which desktop environment has been selected for a
3071 * UNIX utmp support is now compile-time optional to support
3072 legacy-free systems.
3074 * systemctl gained two new commands "add-wants" and
3075 "add-requires" for pulling in units from specific targets
3078 * If the word "rescue" is specified on the kernel command line
3079 the system will now boot into rescue mode (aka
3080 rescue.target), which was previously available only by
3081 specifying "1" or "systemd.unit=rescue.target" on the kernel
3082 command line. This new kernel command line option nicely
3083 mirrors the already existing "emergency" kernel command line
3086 * New kernel command line options mount.usr=, mount.usrflags=,
3087 mount.usrfstype= have been added that match root=, rootflags=,
3088 rootfstype= but allow mounting a specific file system to
3091 * The $NOTIFY_SOCKET is now also passed to control processes of
3092 services, not only the main process.
3094 * This version reenables support for fsck's -l switch. This
3095 means at least version v2.25 of util-linux is required for
3096 operation, otherwise dead-locks on device nodes may
3097 occur. Again: you need to update util-linux to at least
3098 v2.25 when updating systemd to v217.
3100 * The "multi-seat-x" tool has been removed from systemd, as
3101 its functionality has been integrated into X servers 1.16,
3102 and the tool is hence redundant. It is recommended to update
3103 display managers invoking this tool to simply invoke X
3104 directly from now on, again.
3106 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
3107 message flag has been added for all of systemd's PolicyKit
3108 authenticated method calls has been added. In particular
3109 this now allows optional interactive authorization via
3110 PolicyKit for many of PID1's privileged operations such as
3111 unit file enabling and disabling.
3113 * "udevadm hwdb --update" learnt a new switch "--usr" for
3114 placing the rebuilt hardware database in /usr instead of
3115 /etc. When used only hardware database entries stored in
3116 /usr will be used, and any user database entries in /etc are
3117 ignored. This functionality is useful for vendors to ship a
3118 pre-built database on systems where local configuration is
3119 unnecessary or unlikely.
3121 * Calendar time specifications in .timer units now also
3122 understand the strings "semi-annually", "quarterly" and
3123 "minutely" as shortcuts (in addition to the preexisting
3124 "anually", "hourly", ...).
3126 * systemd-tmpfiles will now correctly create files in /dev
3127 at boot which are marked for creation only at boot. It is
3128 recommended to always create static device nodes with 'c!'
3129 and 'b!', so that they are created only at boot and not
3130 overwritten at runtime.
3132 * When the watchdog logic is used for a service (WatchdogSec=)
3133 and the watchdog timeout is hit the service will now be
3134 terminated with SIGABRT (instead of just SIGTERM), in order
3135 to make sure a proper coredump and backtrace is
3136 generated. This ensures that hanging services will result in
3137 similar coredump/backtrace behaviour as services that hit a
3140 Contributions from: Andreas Henriksson, Andrei Borzenkov,
3141 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
3142 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
3143 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
3144 Herrmann, David Sommerseth, David Strauss, Emil Renner
3145 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
3146 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
3147 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
3148 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
3149 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
3150 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
3151 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
3152 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
3153 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
3154 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
3155 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
3156 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
3157 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
3158 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
3159 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
3160 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
3163 — Berlin, 2014-10-28
3167 * timedated no longer reads NTP implementation unit names from
3168 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
3169 implementations should add a
3171 Conflicts=systemd-timesyncd.service
3173 to their unit files to take over and replace systemd's NTP
3174 default functionality.
3176 * systemd-sysusers gained a new line type "r" for configuring
3177 which UID/GID ranges to allocate system users/groups
3178 from. Lines of type "u" may now add an additional column
3179 that specifies the home directory for the system user to be
3180 created. Also, systemd-sysusers may now optionally read user
3181 information from STDIN instead of a file. This is useful for
3182 invoking it from RPM preinst scriptlets that need to create
3183 users before the first RPM file is installed since these
3184 files might need to be owned by them. A new
3185 %sysusers_create_inline RPM macro has been introduced to do
3186 just that. systemd-sysusers now updates the shadow files as
3187 well as the user/group databases, which should enhance
3188 compatibility with certain tools like grpck.
3190 * A number of bus APIs of PID 1 now optionally consult
3191 PolicyKit to permit access for otherwise unprivileged
3192 clients under certain conditions. Note that this currently
3193 doesn't support interactive authentication yet, but this is
3194 expected to be added eventually, too.
3196 * /etc/machine-info now has new fields for configuring the
3197 deployment environment of the machine, as well as the
3198 location of the machine. hostnamectl has been updated with
3199 new command to update these fields.
3201 * systemd-timesyncd has been updated to automatically acquire
3202 NTP server information from systemd-networkd, which might
3203 have been discovered via DHCP.
3205 * systemd-resolved now includes a caching DNS stub resolver
3206 and a complete LLMNR name resolution implementation. A new
3207 NSS module "nss-resolve" has been added which can be used
3208 instead of glibc's own "nss-dns" to resolve hostnames via
3209 systemd-resolved. Hostnames, addresses and arbitrary RRs may
3210 be resolved via systemd-resolved D-Bus APIs. In contrast to
3211 the glibc internal resolver systemd-resolved is aware of
3212 multi-homed system, and keeps DNS server and caches separate
3213 and per-interface. Queries are sent simultaneously on all
3214 interfaces that have DNS servers configured, in order to
3215 properly handle VPNs and local LANs which might resolve
3216 separate sets of domain names. systemd-resolved may acquire
3217 DNS server information from systemd-networkd automatically,
3218 which in turn might have discovered them via DHCP. A tool
3219 "systemd-resolve-host" has been added that may be used to
3220 query the DNS logic in resolved. systemd-resolved implements
3221 IDNA and automatically uses IDNA or UTF-8 encoding depending
3222 on whether classic DNS or LLMNR is used as transport. In the
3223 next releases we intend to add a DNSSEC and mDNS/DNS-SD
3224 implementation to systemd-resolved.
3226 * A new NSS module nss-mymachines has been added, that
3227 automatically resolves the names of all local registered
3228 containers to their respective IP addresses.
3230 * A new client tool "networkctl" for systemd-networkd has been
3231 added. It currently is entirely passive and will query
3232 networking configuration from udev, rtnetlink and networkd,
3233 and present it to the user in a very friendly
3234 way. Eventually, we hope to extend it to become a full
3235 control utility for networkd.
3237 * .socket units gained a new DeferAcceptSec= setting that
3238 controls the kernels' TCP_DEFER_ACCEPT sockopt for
3239 TCP. Similarly, support for controlling TCP keep-alive
3240 settings has been added (KeepAliveTimeSec=,
3241 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
3242 turning off Nagle's algorithm on TCP has been added
3245 * logind learned a new session type "web", for use in projects
3246 like Cockpit which register web clients as PAM sessions.
3248 * timer units with at least one OnCalendar= setting will now
3249 be started only after timer-sync.target has been
3250 reached. This way they will not elapse before the system
3251 clock has been corrected by a local NTP client or
3252 similar. This is particular useful on RTC-less embedded
3253 machines, that come up with an invalid system clock.
3255 * systemd-nspawn's --network-veth= switch should now result in
3256 stable MAC addresses for both the outer and the inner side
3259 * systemd-nspawn gained a new --volatile= switch for running
3260 container instances with /etc or /var unpopulated.
3262 * The kdbus client code has been updated to use the new Linux
3263 3.17 memfd subsystem instead of the old kdbus-specific one.
3265 * systemd-networkd's DHCP client and server now support
3266 FORCERENEW. There are also new configuration options to
3267 configure the vendor client identifier and broadcast mode
3270 * systemd will no longer inform the kernel about the current
3271 timezone, as this is necessarily incorrect and racy as the
3272 kernel has no understanding of DST and similar
3273 concepts. This hence means FAT timestamps will be always
3274 considered UTC, similar to what Android is already
3275 doing. Also, when the RTC is configured to the local time
3276 (rather than UTC) systemd will never synchronize back to it,
3277 as this might confuse Windows at a later boot.
3279 * systemd-analyze gained a new command "verify" for offline
3280 validation of unit files.
3282 * systemd-networkd gained support for a couple of additional
3283 settings for bonding networking setups. Also, the metric for
3284 statically configured routes may now be configured. For
3285 network interfaces where this is appropriate the peer IP
3286 address may now be configured.
3288 * systemd-networkd's DHCP client will no longer request
3289 broadcasting by default, as this tripped up some networks.
3290 For hardware where broadcast is required the feature should
3291 be switched back on using RequestBroadcast=yes.
3293 * systemd-networkd will now set up IPv4LL addresses (when
3294 enabled) even if DHCP is configured successfully.
3296 * udev will now default to respect network device names given
3297 by the kernel when the kernel indicates that these are
3298 predictable. This behavior can be tweaked by changing
3299 NamePolicy= in the relevant .link file.
3301 * A new library systemd-terminal has been added that
3302 implements full TTY stream parsing and rendering. This
3303 library is supposed to be used later on for implementing a
3304 full userspace VT subsystem, replacing the current kernel
3307 * A new tool systemd-journal-upload has been added to push
3308 journal data to a remote system running
3309 systemd-journal-remote.
3311 * journald will no longer forward all local data to another
3312 running syslog daemon. This change has been made because
3313 rsyslog (which appears to be the most commonly used syslog
3314 implementation these days) no longer makes use of this, and
3315 instead pulls the data out of the journal on its own. Since
3316 forwarding the messages to a non-existent syslog server is
3317 more expensive than we assumed we have now turned this
3318 off. If you run a syslog server that is not a recent rsyslog
3319 version, you have to turn this option on again
3320 (ForwardToSyslog= in journald.conf).
3322 * journald now optionally supports the LZ4 compressor for
3323 larger journal fields. This compressor should perform much
3324 better than XZ which was the previous default.
3326 * machinectl now shows the IP addresses of local containers,
3327 if it knows them, plus the interface name of the container.
3329 * A new tool "systemd-escape" has been added that makes it
3330 easy to escape strings to build unit names and similar.
3332 * sd_notify() messages may now include a new ERRNO= field
3333 which is parsed and collected by systemd and shown among the
3334 "systemctl status" output for a service.
3336 * A new component "systemd-firstboot" has been added that
3337 queries the most basic systemd information (timezone,
3338 hostname, root password) interactively on first
3339 boot. Alternatively it may also be used to provision these
3340 things offline on OS images installed into directories.
3342 * The default sysctl.d/ snippets will now set
3344 net.ipv4.conf.default.promote_secondaries=1
3346 This has the benefit of no flushing secondary IP addresses
3347 when primary addresses are removed.
3349 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
3350 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
3351 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
3352 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
3353 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
3354 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
3355 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
3356 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
3357 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
3358 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
3359 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
3360 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
3361 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
3362 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
3363 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
3365 — Berlin, 2014-08-19
3369 * A new tool systemd-sysusers has been added. This tool
3370 creates system users and groups in /etc/passwd and
3371 /etc/group, based on static declarative system user/group
3372 definitions in /usr/lib/sysusers.d/. This is useful to
3373 enable factory resets and volatile systems that boot up with
3374 an empty /etc directory, and thus need system users and
3375 groups created during early boot. systemd now also ships
3376 with two default sysusers.d/ files for the most basic
3377 users and groups systemd and the core operating system
3380 * A new tmpfiles snippet has been added that rebuilds the
3381 essential files in /etc on boot, should they be missing.
3383 * A directive for ensuring automatic clean-up of
3384 /var/cache/man/ has been removed from the default
3385 configuration. This line should now be shipped by the man
3386 implementation. The necessary change has been made to the
3387 man-db implementation. Note that you need to update your man
3388 implementation to one that ships this line, otherwise no
3389 automatic clean-up of /var/cache/man will take place.
3391 * A new condition ConditionNeedsUpdate= has been added that
3392 may conditionalize services to only run when /etc or /var
3393 are "older" than the vendor operating system resources in
3394 /usr. This is useful for reconstructing or updating /etc
3395 after an offline update of /usr or a factory reset, on the
3396 next reboot. Services that want to run once after such an
3397 update or reset should use this condition and order
3398 themselves before the new systemd-update-done.service, which
3399 will mark the two directories as fully updated. A number of
3400 service files have been added making use of this, to rebuild
3401 the udev hardware database, the journald message catalog and
3402 dynamic loader cache (ldconfig). The systemd-sysusers tool
3403 described above also makes use of this now. With this in
3404 place it is now possible to start up a minimal operating
3405 system with /etc empty cleanly. For more information on the
3406 concepts involved see this recent blog story:
3408 http://0pointer.de/blog/projects/stateless.html
3410 * A new system group "input" has been introduced, and all
3411 input device nodes get this group assigned. This is useful
3412 for system-level software to get access to input devices. It
3413 complements what is already done for "audio" and "video".
3415 * systemd-networkd learnt minimal DHCPv4 server support in
3416 addition to the existing DHCPv4 client support. It also
3417 learnt DHCPv6 client and IPv6 Router Solicitation client
3418 support. The DHCPv4 client gained support for static routes
3419 passed in from the server. Note that the [DHCPv4] section
3420 known in older systemd-networkd versions has been renamed to
3421 [DHCP] and is now also used by the DHCPv6 client. Existing
3422 .network files using settings of this section should be
3423 updated, though compatibility is maintained. Optionally, the
3424 client hostname may now be sent to the DHCP server.
3426 * networkd gained support for vxlan virtual networks as well
3427 as tun/tap and dummy devices.
3429 * networkd gained support for automatic allocation of address
3430 ranges for interfaces from a system-wide pool of
3431 addresses. This is useful for dynamically managing a large
3432 number of interfaces with a single network configuration
3433 file. In particular this is useful to easily assign
3434 appropriate IP addresses to the veth links of a large number
3435 of nspawn instances.
3437 * RPM macros for processing sysusers, sysctl and binfmt
3438 drop-in snippets at package installation time have been
3441 * The /etc/os-release file should now be placed in
3442 /usr/lib/os-release. The old location is automatically
3443 created as symlink. /usr/lib is the more appropriate
3444 location of this file, since it shall actually describe the
3445 vendor operating system shipped in /usr, and not the
3446 configuration stored in /etc.
3448 * .mount units gained a new boolean SloppyOptions= setting
3449 that maps to mount(8)'s -s option which enables permissive
3450 parsing of unknown mount options.
3452 * tmpfiles learnt a new "L+" directive which creates a symlink
3453 but (unlike "L") deletes a pre-existing file first, should
3454 it already exist and not already be the correct
3455 symlink. Similarly, "b+", "c+" and "p+" directives have been
3456 added as well, which create block and character devices, as
3457 well as fifos in the filesystem, possibly removing any
3458 pre-existing files of different types.
3460 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
3461 'argument' field (which so far specified the source to
3462 symlink/copy the files from) is now optional. If omitted the
3463 same file os copied from /usr/share/factory/ suffixed by the
3464 full destination path. This is useful for populating /etc
3465 with essential files, by copying them from vendor defaults
3466 shipped in /usr/share/factory/etc.
3468 * A new command "systemctl preset-all" has been added that
3469 applies the service preset settings to all installed unit
3470 files. A new switch --preset-mode= has been added that
3471 controls whether only enable or only disable operations
3474 * A new command "systemctl is-system-running" has been added
3475 that allows checking the overall state of the system, for
3476 example whether it is fully up and running.
3478 * When the system boots up with an empty /etc, the equivalent
3479 to "systemctl preset-all" is executed during early boot, to
3480 make sure all default services are enabled after a factory
3483 * systemd now contains a minimal preset file that enables the
3484 most basic services systemd ships by default.
3486 * Unit files' [Install] section gained a new DefaultInstance=
3487 field for defining the default instance to create if a
3488 template unit is enabled with no instance specified.
3490 * A new passive target cryptsetup-pre.target has been added
3491 that may be used by services that need to make they run and
3492 finish before the first LUKS cryptographic device is set up.
3494 * The /dev/loop-control and /dev/btrfs-control device nodes
3495 are now owned by the "disk" group by default, opening up
3496 access to this group.
3498 * systemd-coredump will now automatically generate a
3499 stack trace of all core dumps taking place on the system,
3500 based on elfutils' libdw library. This stack trace is logged
3503 * systemd-coredump may now optionally store coredumps directly
3504 on disk (in /var/lib/systemd/coredump, possibly compressed),
3505 instead of storing them unconditionally in the journal. This
3506 mode is the new default. A new configuration file
3507 /etc/systemd/coredump.conf has been added to configure this
3508 and other parameters of systemd-coredump.
3510 * coredumpctl gained a new "info" verb to show details about a
3511 specific coredump. A new switch "-1" has also been added
3512 that makes sure to only show information about the most
3513 recent entry instead of all entries. Also, as the tool is
3514 generally useful now the "systemd-" prefix of the binary
3515 name has been removed. Distributions that want to maintain
3516 compatibility with the old name should add a symlink from
3517 the old name to the new name.
3519 * journald's SplitMode= now defaults to "uid". This makes sure
3520 that unprivileged users can access their own coredumps with
3521 coredumpctl without restrictions.
3523 * New kernel command line options "systemd.wants=" (for
3524 pulling an additional unit during boot), "systemd.mask="
3525 (for masking a specific unit for the boot), and
3526 "systemd.debug-shell" (for enabling the debug shell on tty9)
3527 have been added. This is implemented in the new generator
3528 "systemd-debug-generator".
3530 * systemd-nspawn will now by default filter a couple of
3531 syscalls for containers, among them those required for
3532 kernel module loading, direct x86 IO port access, swap
3533 management, and kexec. Most importantly though
3534 open_by_handle_at() is now prohibited for containers,
3535 closing a hole similar to a recently discussed vulnerability
3536 in docker regarding access to files on file hierarchies the
3537 container should normally not have access to. Note that, for
3538 nspawn, we generally make no security claims anyway (and
3539 this is explicitly documented in the man page), so this is
3540 just a fix for one of the most obvious problems.
3542 * A new man page file-hierarchy(7) has been added that
3543 contains a minimized, modernized version of the file system
3544 layout systemd expects, similar in style to the FHS
3545 specification or hier(5). A new tool systemd-path(1) has
3546 been added to query many of these paths for the local
3549 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
3550 longer done. Since the directory now has a per-user size
3551 limit, and is cleaned on logout this appears unnecessary,
3552 in particular since this now brings the lifecycle of this
3553 directory closer in line with how IPC objects are handled.
3555 * systemd.pc now exports a number of additional directories,
3556 including $libdir (which is useful to identify the library
3557 path for the primary architecture of the system), and a
3558 couple of drop-in directories.
3560 * udev's predictable network interface names now use the dev_port
3561 sysfs attribute, introduced in linux 3.15 instead of dev_id to
3562 distinguish between ports of the same PCI function. dev_id should
3563 only be used for ports using the same HW address, hence the need
3566 * machined has been updated to export the OS version of a
3567 container (read from /etc/os-release and
3568 /usr/lib/os-release) on the bus. This is now shown in
3569 "machinectl status" for a machine.
3571 * A new service setting RestartForceExitStatus= has been
3572 added. If configured to a set of exit signals or process
3573 return values, the service will be restarted when the main
3574 daemon process exits with any of them, regardless of the
3577 * systemctl's -H switch for connecting to remote systemd
3578 machines has been extended so that it may be used to
3579 directly connect to a specific container on the
3580 host. "systemctl -H root@foobar:waldi" will now connect as
3581 user "root" to host "foobar", and then proceed directly to
3582 the container named "waldi". Note that currently you have to
3583 authenticate as user "root" for this to work, as entering
3584 containers is a privileged operation.
3586 Contributions from: Andreas Henriksson, Benjamin Steinwender,
3587 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
3588 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
3589 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
3590 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
3591 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
3592 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
3593 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
3594 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
3595 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
3596 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
3597 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
3599 — Berlin, 2014-07-03
3603 * As an experimental feature, udev now tries to lock the
3604 disk device node (flock(LOCK_SH|LOCK_NB)) while it
3605 executes events for the disk or any of its partitions.
3606 Applications like partitioning programs can lock the
3607 disk device node (flock(LOCK_EX)) and claim temporary
3608 device ownership that way; udev will entirely skip all event
3609 handling for this disk and its partitions. If the disk
3610 was opened for writing, the close will trigger a partition
3611 table rescan in udev's "watch" facility, and if needed
3612 synthesize "change" events for the disk and all its partitions.
3613 This is now unconditionally enabled, and if it turns out to
3614 cause major problems, we might turn it on only for specific
3615 devices, or might need to disable it entirely. Device Mapper
3616 devices are excluded from this logic.
3618 * We temporarily dropped the "-l" switch for fsck invocations,
3619 since they collide with the flock() logic above. util-linux
3620 upstream has been changed already to avoid this conflict,
3621 and we will readd "-l" as soon as util-linux with this
3622 change has been released.
3624 * The dependency on libattr has been removed. Since a long
3625 time, the extended attribute calls have moved to glibc, and
3626 libattr is thus unnecessary.
3628 * Virtualization detection works without privileges now. This
3629 means the systemd-detect-virt binary no longer requires
3630 CAP_SYS_PTRACE file capabilities, and our daemons can run
3631 with fewer privileges.
3633 * systemd-networkd now runs under its own "systemd-network"
3634 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
3635 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
3636 loses the ability to write to files owned by root this way.
3638 * Similarly, systemd-resolved now runs under its own
3639 "systemd-resolve" user with no capabilities remaining.
3641 * Similarly, systemd-bus-proxyd now runs under its own
3642 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
3644 * systemd-networkd gained support for setting up "veth"
3645 virtual Ethernet devices for container connectivity, as well
3646 as GRE and VTI tunnels.
3648 * systemd-networkd will no longer automatically attempt to
3649 manually load kernel modules necessary for certain tunnel
3650 transports. Instead, it is assumed the kernel loads them
3651 automatically when required. This only works correctly on
3652 very new kernels. On older kernels, please consider adding
3653 the kernel modules to /etc/modules-load.d/ as a work-around.
3655 * The resolv.conf file systemd-resolved generates has been
3656 moved to /run/systemd/resolve/. If you have a symlink from
3657 /etc/resolv.conf, it might be necessary to correct it.
3659 * Two new service settings, ProtectHome= and ProtectSystem=,
3660 have been added. When enabled, they will make the user data
3661 (such as /home) inaccessible or read-only and the system
3662 (such as /usr) read-only, for specific services. This allows
3663 very light-weight per-service sandboxing to avoid
3664 modifications of user data or system files from
3665 services. These two new switches have been enabled for all
3666 of systemd's long-running services, where appropriate.
3668 * Socket units gained new SocketUser= and SocketGroup=
3669 settings to set the owner user and group of AF_UNIX sockets
3670 and FIFOs in the file system.
3672 * Socket units gained a new RemoveOnStop= setting. If enabled,
3673 all FIFOS and sockets in the file system will be removed
3674 when the specific socket unit is stopped.
3676 * Socket units gained a new Symlinks= setting. It takes a list
3677 of symlinks to create to file system sockets or FIFOs
3678 created by the specific Unix sockets. This is useful to
3679 manage symlinks to socket nodes with the same life-cycle as
3682 * The /dev/log socket and /dev/initctl FIFO have been moved to
3683 /run, and have been replaced by symlinks. This allows
3684 connecting to these facilities even if PrivateDevices=yes is
3685 used for a service (which makes /dev/log itself unavailable,
3686 but /run is left). This also has the benefit of ensuring
3687 that /dev only contains device nodes, directories and
3688 symlinks, and nothing else.
3690 * sd-daemon gained two new calls sd_pid_notify() and
3691 sd_pid_notifyf(). They are similar to sd_notify() and
3692 sd_notifyf(), but allow overriding of the source PID of
3693 notification messages if permissions permit this. This is
3694 useful to send notify messages on behalf of a different
3695 process (for example, the parent process). The
3696 systemd-notify tool has been updated to make use of this
3697 when sending messages (so that notification messages now
3698 originate from the shell script invoking systemd-notify and
3699 not the systemd-notify process itself. This should minimize
3700 a race where systemd fails to associate notification
3701 messages to services when the originating process already
3704 * A new "on-abnormal" setting for Restart= has been added. If
3705 set, it will result in automatic restarts on all "abnormal"
3706 reasons for a process to exit, which includes unclean
3707 signals, core dumps, timeouts and watchdog timeouts, but
3708 does not include clean and unclean exit codes or clean
3709 signals. Restart=on-abnormal is an alternative for
3710 Restart=on-failure for services that shall be able to
3711 terminate and avoid restarts on certain errors, by
3712 indicating so with an unclean exit code. Restart=on-failure
3713 or Restart=on-abnormal is now the recommended setting for
3714 all long-running services.
3716 * If the InaccessibleDirectories= service setting points to a
3717 mount point (or if there are any submounts contained within
3718 it), it is now attempted to completely unmount it, to make
3719 the file systems truly unavailable for the respective
3722 * The ReadOnlyDirectories= service setting and
3723 systemd-nspawn's --read-only parameter are now recursively
3724 applied to all submounts, too.
3726 * Mount units may now be created transiently via the bus APIs.
3728 * The support for SysV and LSB init scripts has been removed
3729 from the systemd daemon itself. Instead, it is now
3730 implemented as a generator that creates native systemd units
3731 from these scripts when needed. This enables us to remove a
3732 substantial amount of legacy code from PID 1, following the
3733 fact that many distributions only ship a very small number
3734 of LSB/SysV init scripts nowadays.
3736 * Privileged Xen (dom0) domains are not considered
3737 virtualization anymore by the virtualization detection
3738 logic. After all, they generally have unrestricted access to
3739 the hardware and usually are used to manage the unprivileged
3742 * systemd-tmpfiles gained a new "C" line type, for copying
3743 files or entire directories.
3745 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
3746 lines. So far, they have been non-globbing versions of the
3747 latter, and have thus been redundant. In future, it is
3748 recommended to only use "z". "m" has hence been removed
3749 from the documentation, even though it stays supported.
3751 * A tmpfiles snippet to recreate the most basic structure in
3752 /var has been added. This is enough to create the /var/run →
3753 /run symlink and create a couple of structural
3754 directories. This allows systems to boot up with an empty or
3755 volatile /var. Of course, while with this change, the core OS
3756 now is capable with dealing with a volatile /var, not all
3757 user services are ready for it. However, we hope that sooner
3758 or later, many service daemons will be changed upstream so
3759 that they are able to automatically create their necessary
3760 directories in /var at boot, should they be missing. This is
3761 the first step to allow state-less systems that only require
3762 the vendor image for /usr to boot.
3764 * systemd-nspawn has gained a new --tmpfs= switch to mount an
3765 empty tmpfs instance to a specific directory. This is
3766 particularly useful for making use of the automatic
3767 reconstruction of /var (see above), by passing --tmpfs=/var.
3769 * Access modes specified in tmpfiles snippets may now be
3770 prefixed with "~", which indicates that they shall be masked
3771 by whether the existing file or directory is currently
3772 writable, readable or executable at all. Also, if specified,
3773 the sgid/suid/sticky bits will be masked for all
3776 * A new passive target unit "network-pre.target" has been
3777 added which is useful for services that shall run before any
3778 network is configured, for example firewall scripts.
3780 * The "floppy" group that previously owned the /dev/fd*
3781 devices is no longer used. The "disk" group is now used
3782 instead. Distributions should probably deprecate usage of
3785 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
3786 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
3787 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
3788 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
3789 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
3790 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
3793 — Berlin, 2014-06-11
3797 * A new "systemd-timesyncd" daemon has been added for
3798 synchronizing the system clock across the network. It
3799 implements an SNTP client. In contrast to NTP
3800 implementations such as chrony or the NTP reference server,
3801 this only implements a client side, and does not bother with
3802 the full NTP complexity, focusing only on querying time from
3803 one remote server and synchronizing the local clock to
3804 it. Unless you intend to serve NTP to networked clients or
3805 want to connect to local hardware clocks, this simple NTP
3806 client should be more than appropriate for most
3807 installations. The daemon runs with minimal privileges, and
3808 has been hooked up with networkd to only operate when
3809 network connectivity is available. The daemon saves the
3810 current clock to disk every time a new NTP sync has been
3811 acquired, and uses this to possibly correct the system clock
3812 early at bootup, in order to accommodate for systems that
3813 lack an RTC such as the Raspberry Pi and embedded devices,
3814 and to make sure that time monotonically progresses on these
3815 systems, even if it is not always correct. To make use of
3816 this daemon, a new system user and group "systemd-timesync"
3817 needs to be created on installation of systemd.
3819 * The queue "seqnum" interface of libudev has been disabled, as
3820 it was generally incompatible with device namespacing as
3821 sequence numbers of devices go "missing" if the devices are
3822 part of a different namespace.
3824 * "systemctl list-timers" and "systemctl list-sockets" gained
3825 a --recursive switch for showing units of these types also
3826 for all local containers, similar in style to the already
3827 supported --recursive switch for "systemctl list-units".
3829 * A new RebootArgument= setting has been added for service
3830 units, which may be used to specify a kernel reboot argument
3831 to use when triggering reboots with StartLimitAction=.
3833 * A new FailureAction= setting has been added for service
3834 units which may be used to specify an operation to trigger
3835 when a service fails. This works similarly to
3836 StartLimitAction=, but unlike it, controls what is done
3837 immediately rather than only after several attempts to
3838 restart the service in question.
3840 * hostnamed got updated to also expose the kernel name,
3841 release, and version on the bus. This is useful for
3842 executing commands like hostnamectl with the -H switch.
3843 systemd-analyze makes use of this to properly display
3844 details when running non-locally.
3846 * The bootchart tool can now show cgroup information in the
3847 graphs it generates.
3849 * The CFS CPU quota cgroup attribute is now exposed for
3850 services. The new CPUQuota= switch has been added for this
3851 which takes a percentage value. Setting this will have the
3852 result that a service may never get more CPU time than the
3853 specified percentage, even if the machine is otherwise idle.
3855 * systemd-networkd learned IPIP and SIT tunnel support.
3857 * LSB init scripts exposing a dependency on $network will now
3858 get a dependency on network-online.target rather than simply
3859 network.target. This should bring LSB handling closer to
3860 what it was on SysV systems.
3862 * A new fsck.repair= kernel option has been added to control
3863 how fsck shall deal with unclean file systems at boot.
3865 * The (.ini) configuration file parser will now silently
3866 ignore sections whose name begins with "X-". This may be
3867 used to maintain application-specific extension sections in unit
3870 * machined gained a new API to query the IP addresses of
3871 registered containers. "machinectl status" has been updated
3872 to show these addresses in its output.
3874 * A new call sd_uid_get_display() has been added to the
3875 sd-login APIs for querying the "primary" session of a
3876 user. The "primary" session of the user is elected from the
3877 user's sessions and generally a graphical session is
3878 preferred over a text one.
3880 * A minimal systemd-resolved daemon has been added. It
3881 currently simply acts as a companion to systemd-networkd and
3882 manages resolv.conf based on per-interface DNS
3883 configuration, possibly supplied via DHCP. In the long run
3884 we hope to extend this into a local DNSSEC enabled DNS and
3887 * The systemd-networkd-wait-online tool is now enabled by
3888 default. It will delay network-online.target until a network
3889 connection has been configured. The tool primarily integrates
3890 with networkd, but will also make a best effort to make sense
3891 of network configuration performed in some other way.
3893 * Two new service options StartupCPUShares= and
3894 StartupBlockIOWeight= have been added that work similarly to
3895 CPUShares= and BlockIOWeight= however only apply during
3896 system startup. This is useful to prioritize certain services
3897 differently during bootup than during normal runtime.
3899 * hostnamed has been changed to prefer the statically
3900 configured hostname in /etc/hostname (unless set to
3901 'localhost' or empty) over any dynamic one supplied by
3902 dhcp. With this change, the rules for picking the hostname
3903 match more closely the rules of other configuration settings
3904 where the local administrator's configuration in /etc always
3905 overrides any other settings.
3907 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
3908 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
3909 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
3910 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
3911 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
3912 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
3913 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
3914 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
3915 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
3916 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
3917 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
3918 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
3919 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
3920 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
3921 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
3922 Lindskog, WaLyong Cho, Will Woods, Zbigniew
3925 — Beijing, 2014-05-28
3929 * When restoring the screen brightness at boot, stay away from
3930 the darkest setting or from the lowest 5% of the available
3931 range, depending on which is the larger value of both. This
3932 should effectively protect the user from rebooting into a
3933 black screen, should the brightness have been set to minimum
3936 * sd-login gained a new sd_machine_get_class() call to
3937 determine the class ("vm" or "container") of a machine
3938 registered with machined.
3940 * sd-login gained new calls
3941 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
3942 to query the identity of the peer of a local AF_UNIX
3943 connection. They operate similarly to their sd_pid_get_xyz()
3946 * PID 1 will now maintain a system-wide system state engine
3947 with the states "starting", "running", "degraded",
3948 "maintenance", "stopping". These states are bound to system
3949 startup, normal runtime, runtime with at least one failed
3950 service, rescue/emergency mode and system shutdown. This
3951 state is shown in the "systemctl status" output when no unit
3952 name is passed. It is useful to determine system state, in
3953 particularly when doing so for many systems or containers at
3956 * A new command "list-machines" has been added to "systemctl"
3957 that lists all local OS containers and shows their system
3958 state (see above), if systemd runs inside of them.
3960 * systemctl gained a new "-r" switch to recursively enumerate
3961 units on all local containers, when used with the
3962 "list-unit" command (which is the default one that is
3963 executed when no parameters are specified).
3965 * The GPT automatic partition discovery logic will now honour
3966 two GPT partition flags: one may be set on a partition to
3967 cause it to be mounted read-only, and the other may be set
3968 on a partition to ignore it during automatic discovery.
3970 * Two new GPT type UUIDs have been added for automatic root
3971 partition discovery, for 32-bit and 64-bit ARM. This is not
3972 particularly useful for discovering the root directory on
3973 these architectures during bare-metal boots (since UEFI is
3974 not common there), but still very useful to allow booting of
3975 ARM disk images in nspawn with the -i option.
3977 * MAC addresses of interfaces created with nspawn's
3978 --network-interface= switch will now be generated from the
3979 machine name, and thus be stable between multiple invocations
3982 * logind will now automatically remove all IPC objects owned
3983 by a user if she or he fully logs out. This makes sure that
3984 users who are logged out cannot continue to consume IPC
3985 resources. This covers SysV memory, semaphores and message
3986 queues as well as POSIX shared memory and message
3987 queues. Traditionally, SysV and POSIX IPC had no life-cycle
3988 limits. With this functionality, that is corrected. This may
3989 be turned off by using the RemoveIPC= switch of logind.conf.
3991 * The systemd-machine-id-setup and tmpfiles tools gained a
3992 --root= switch to operate on a specific root directory,
3995 * journald can now forward logged messages to the TTYs of all
3996 logged in users ("wall"). This is the default for all
3997 emergency messages now.
3999 * A new tool systemd-journal-remote has been added to stream
4000 journal log messages across the network.
4002 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
4003 controller trees are mounted into it. Note that the
4004 directories mounted beneath it are not read-only. This is a
4005 security measure and is particularly useful because glibc
4006 actually includes a search logic to pick any tmpfs it can
4007 find to implement shm_open() if /dev/shm is not available
4008 (which it might very well be in namespaced setups).
4010 * machinectl gained a new "poweroff" command to cleanly power
4011 down a local OS container.
4013 * The PrivateDevices= unit file setting will now also drop the
4014 CAP_MKNOD capability from the capability bound set, and
4015 imply DevicePolicy=closed.
4017 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
4018 comprehensively on all long-running systemd services where
4019 this is appropriate.
4021 * systemd-udevd will now run in a disassociated mount
4022 namespace. To mount directories from udev rules, make sure to
4023 pull in mount units via SYSTEMD_WANTS properties.
4025 * The kdbus support gained support for uploading policy into
4026 the kernel. sd-bus gained support for creating "monitoring"
4027 connections that can eavesdrop into all bus communication
4028 for debugging purposes.
4030 * Timestamps may now be specified in seconds since the UNIX
4031 epoch Jan 1st, 1970 by specifying "@" followed by the value
4034 * Native tcpwrap support in systemd has been removed. tcpwrap
4035 is old code, not really maintained anymore and has serious
4036 shortcomings, and better options such as firewalls
4037 exist. For setups that require tcpwrap usage, please
4038 consider invoking your socket-activated service via tcpd,
4039 like on traditional inetd.
4041 * A new system.conf configuration option
4042 DefaultTimerAccuracySec= has been added that controls the
4043 default AccuracySec= setting of .timer units.
4045 * Timer units gained a new WakeSystem= switch. If enabled,
4046 timers configured this way will cause the system to resume
4047 from system suspend (if the system supports that, which most
4050 * Timer units gained a new Persistent= switch. If enabled,
4051 timers configured this way will save to disk when they have
4052 been last triggered. This information is then used on next
4053 reboot to possible execute overdue timer events, that
4054 could not take place because the system was powered off.
4055 This enables simple anacron-like behaviour for timer units.
4057 * systemctl's "list-timers" will now also list the time a
4058 timer unit was last triggered in addition to the next time
4059 it will be triggered.
4061 * systemd-networkd will now assign predictable IPv4LL
4062 addresses to its local interfaces.
4064 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
4065 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
4066 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
4067 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
4068 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
4069 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
4070 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
4071 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
4074 — Berlin, 2014-03-25
4078 * A new unit file setting RestrictAddressFamilies= has been
4079 added to restrict which socket address families unit
4080 processes gain access to. This takes address family names
4081 like "AF_INET" or "AF_UNIX", and is useful to minimize the
4082 attack surface of services via exotic protocol stacks. This
4083 is built on seccomp system call filters.
4085 * Two new unit file settings RuntimeDirectory= and
4086 RuntimeDirectoryMode= have been added that may be used to
4087 manage a per-daemon runtime directories below /run. This is
4088 an alternative for setting up directory permissions with
4089 tmpfiles snippets, and has the advantage that the runtime
4090 directory's lifetime is bound to the daemon runtime and that
4091 the daemon starts up with an empty directory each time. This
4092 is particularly useful when writing services that drop
4093 privileges using the User= or Group= setting.
4095 * The DeviceAllow= unit setting now supports globbing for
4096 matching against device group names.
4098 * The systemd configuration file system.conf gained new
4099 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
4100 DefaultMemoryAccounting= to globally turn on/off accounting
4101 for specific resources (cgroups) for all units. These
4102 settings may still be overridden individually in each unit
4105 * systemd-gpt-auto-generator is now able to discover /srv and
4106 root partitions in addition to /home and swap partitions. It
4107 also supports LUKS-encrypted partitions now. With this in
4108 place, automatic discovery of partitions to mount following
4109 the Discoverable Partitions Specification
4110 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
4111 is now a lot more complete. This allows booting without
4112 /etc/fstab and without root= on the kernel command line on
4113 systems prepared appropriately.
4115 * systemd-nspawn gained a new --image= switch which allows
4116 booting up disk images and Linux installations on any block
4117 device that follow the Discoverable Partitions Specification
4118 (see above). This means that installations made with
4119 appropriately updated installers may now be started and
4120 deployed using container managers, completely
4121 unmodified. (We hope that libvirt-lxc will add support for
4122 this feature soon, too.)
4124 * systemd-nspawn gained a new --network-macvlan= setting to
4125 set up a private macvlan interface for the
4126 container. Similarly, systemd-networkd gained a new
4127 Kind=macvlan setting in .netdev files.
4129 * systemd-networkd now supports configuring local addresses
4132 * A new tool systemd-network-wait-online has been added to
4133 synchronously wait for network connectivity using
4136 * The sd-bus.h bus API gained a new sd_bus_track object for
4137 tracking the life-cycle of bus peers. Note that sd-bus.h is
4138 still not a public API though (unless you specify
4139 --enable-kdbus on the configure command line, which however
4140 voids your warranty and you get no API stability guarantee).
4142 * The $XDG_RUNTIME_DIR runtime directories for each user are
4143 now individual tmpfs instances, which has the benefit of
4144 introducing separate pools for each user, with individual
4145 size limits, and thus making sure that unprivileged clients
4146 can no longer negatively impact the system or other users by
4147 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
4148 RuntimeDirectorySize= has been introduced that allows
4149 controlling the default size limit for all users. It
4150 defaults to 10% of the available physical memory. This is no
4151 replacement for quotas on tmpfs though (which the kernel
4152 still does not support), as /dev/shm and /tmp are still
4153 shared resources used by both the system and unprivileged
4156 * logind will now automatically turn off automatic suspending
4157 on laptop lid close when more than one display is
4158 connected. This was previously expected to be implemented
4159 individually in desktop environments (such as GNOME),
4160 however has been added to logind now, in order to fix a
4161 boot-time race where a desktop environment might not have
4162 been started yet and thus not been able to take an inhibitor
4163 lock at the time where logind already suspends the system
4164 due to a closed lid.
4166 * logind will now wait at least 30s after each system
4167 suspend/resume cycle, and 3min after system boot before
4168 suspending the system due to a closed laptop lid. This
4169 should give USB docking stations and similar enough time to
4170 be probed and configured after system resume and boot in
4171 order to then act as suspend blocker.
4173 * systemd-run gained a new --property= setting which allows
4174 initialization of resource control properties (and others)
4175 for the created scope or service unit. Example: "systemd-run
4176 --property=BlockIOWeight=10 updatedb" may be used to run
4177 updatedb at a low block IO scheduling weight.
4179 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
4180 now also work in --scope mode.
4182 * When systemd is compiled with kdbus support, basic support
4183 for enforced policies is now in place. (Note that enabling
4184 kdbus still voids your warranty and no API compatibility
4187 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
4188 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
4189 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
4190 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
4191 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
4192 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
4193 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
4194 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
4195 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
4196 Zbigniew Jędrzejewski-Szmek
4198 — Berlin, 2014-03-12
4202 * systemd will now relabel /dev after loading the SMACK policy
4203 according to SMACK rules.
4205 * A new unit file option AppArmorProfile= has been added to
4206 set the AppArmor profile for the processes of a unit.
4208 * A new condition check ConditionArchitecture= has been added
4209 to conditionalize units based on the system architecture, as
4210 reported by uname()'s "machine" field.
4212 * systemd-networkd now supports matching on the system
4213 virtualization, architecture, kernel command line, host name
4216 * logind is now a lot more aggressive when suspending the
4217 machine due to a closed laptop lid. Instead of acting only
4218 on the lid close action, it will continuously watch the lid
4219 status and act on it. This is useful for laptops where the
4220 power button is on the outside of the chassis so that it can
4221 be reached without opening the lid (such as the Lenovo
4222 Yoga). On those machines, logind will now immediately
4223 re-suspend the machine if the power button has been
4224 accidentally pressed while the laptop was suspended and in a
4225 backpack or similar.
4227 * logind will now watch SW_DOCK switches and inhibit reaction
4228 to the lid switch if it is pressed. This means that logind
4229 will not suspend the machine anymore if the lid is closed
4230 and the system is docked, if the laptop supports SW_DOCK
4231 notifications via the input layer. Note that ACPI docking
4232 stations do not generate this currently. Also note that this
4233 logic is usually not fully sufficient and Desktop
4234 Environments should take a lid switch inhibitor lock when an
4235 external display is connected, as systemd will not watch
4238 * nspawn will now make use of the devices cgroup controller by
4239 default, and only permit creation of and access to the usual
4240 API device nodes like /dev/null or /dev/random, as well as
4241 access to (but not creation of) the pty devices.
4243 * We will now ship a default .network file for
4244 systemd-networkd that automatically configures DHCP for
4245 network interfaces created by nspawn's --network-veth or
4246 --network-bridge= switches.
4248 * systemd will now understand the usual M, K, G, T suffixes
4249 according to SI conventions (i.e. to the base 1000) when
4250 referring to throughput and hardware metrics. It will stay
4251 with IEC conventions (i.e. to the base 1024) for software
4252 metrics, according to what is customary according to
4253 Wikipedia. We explicitly document which base applies for
4254 each configuration option.
4256 * The DeviceAllow= setting in unit files now supports a syntax
4257 to whitelist an entire group of devices node majors at once,
4258 based on the /proc/devices listing. For example, with the
4259 string "char-pts", it is now possible to whitelist all
4260 current and future pseudo-TTYs at once.
4262 * sd-event learned a new "post" event source. Event sources of
4263 this type are triggered by the dispatching of any event
4264 source of a type that is not "post". This is useful for
4265 implementing clean-up and check event sources that are
4266 triggered by other work being done in the program.
4268 * systemd-networkd is no longer statically enabled, but uses
4269 the usual [Install] sections so that it can be
4270 enabled/disabled using systemctl. It still is enabled by
4273 * When creating a veth interface pair with systemd-nspawn, the
4274 host side will now be prefixed with "vb-" if
4275 --network-bridge= is used, and with "ve-" if --network-veth
4276 is used. This way, it is easy to distinguish these cases on
4277 the host, for example to apply different configuration to
4278 them with systemd-networkd.
4280 * The compatibility libraries for libsystemd-journal.so,
4281 libsystem-id128.so, libsystemd-login.so and
4282 libsystemd-daemon.so do not make use of IFUNC
4283 anymore. Instead, we now build libsystemd.so multiple times
4284 under these alternative names. This means that the footprint
4285 is drastically increased, but given that these are
4286 transitional compatibility libraries, this should not matter
4287 much. This change has been made necessary to support the ARM
4288 platform for these compatibility libraries, as the ARM
4289 toolchain is not really at the same level as the toolchain
4290 for other architectures like x86 and does not support
4291 IFUNC. Please make sure to use --enable-compat-libs only
4292 during a transitional period!
4294 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
4295 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
4296 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
4297 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
4298 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
4299 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
4300 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
4301 Zbigniew Jędrzejewski-Szmek
4303 — Berlin, 2014-02-24
4307 * A new component "systemd-networkd" has been added that can
4308 be used to configure local network interfaces statically or
4309 via DHCP. It is capable of bringing up bridges, VLANs, and
4310 bonding. Currently, no hook-ups for interactive network
4311 configuration are provided. Use this for your initrd,
4312 container, embedded, or server setup if you need a simple,
4313 yet powerful, network configuration solution. This
4314 configuration subsystem is quite nifty, as it allows wildcard
4315 hotplug matching in interfaces. For example, with a single
4316 configuration snippet, you can configure that all Ethernet
4317 interfaces showing up are automatically added to a bridge,
4318 or similar. It supports link-sensing and more.
4320 * A new tool "systemd-socket-proxyd" has been added which can
4321 act as a bidirectional proxy for TCP sockets. This is
4322 useful for adding socket activation support to services that
4323 do not actually support socket activation, including virtual
4324 machines and the like.
4326 * Add a new tool to save/restore rfkill state on
4329 * Save/restore state of keyboard backlights in addition to
4330 display backlights on shutdown/boot.
4332 * udev learned a new SECLABEL{} construct to label device
4333 nodes with a specific security label when they appear. For
4334 now, only SECLABEL{selinux} is supported, but the syntax is
4335 prepared for additional security frameworks.
4337 * udev gained a new scheme to configure link-level attributes
4338 from files in /etc/systemd/network/*.link. These files can
4339 match against MAC address, device path, driver name and type,
4340 and will apply attributes like the naming policy, link speed,
4341 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
4342 address assignment policy (randomized, ...).
4344 * The configuration of network interface naming rules for
4345 "permanent interface names" has changed: a new NamePolicy=
4346 setting in the [Link] section of .link files determines the
4347 priority of possible naming schemes (onboard, slot, MAC,
4348 path). The default value of this setting is determined by
4349 /usr/lib/net/links/99-default.link. Old
4350 80-net-name-slot.rules udev configuration file has been
4351 removed, so local configuration overriding this file should
4352 be adapted to override 99-default.link instead.
4354 * When the User= switch is used in a unit file, also
4355 initialize $SHELL= based on the user database entry.
4357 * systemd no longer depends on libdbus. All communication is
4358 now done with sd-bus, systemd's low-level bus library
4361 * kdbus support has been added to PID 1 itself. When kdbus is
4362 enabled, this causes PID 1 to set up the system bus and
4363 enable support for a new ".busname" unit type that
4364 encapsulates bus name activation on kdbus. It works a little
4365 bit like ".socket" units, except for bus names. A new
4366 generator has been added that converts classic dbus1 service
4367 activation files automatically into native systemd .busname
4370 * sd-bus: add a light-weight vtable implementation that allows
4371 defining objects on the bus with a simple static const
4372 vtable array of its methods, signals and properties.
4374 * systemd will not generate or install static dbus
4375 introspection data anymore to /usr/share/dbus-1/interfaces,
4376 as the precise format of these files is unclear, and
4377 nothing makes use of it.
4379 * A proxy daemon is now provided to proxy clients connecting
4380 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
4381 compatibility with classic D-Bus.
4383 * A bus driver implementation has been added that supports the
4384 classic D-Bus bus driver calls on kdbus, also for
4385 compatibility purposes.
4387 * A new API "sd-event.h" has been added that implements a
4388 minimal event loop API built around epoll. It provides a
4389 couple of features that direct epoll usage is lacking:
4390 prioritization of events, scales to large numbers of timer
4391 events, per-event timer slack (accuracy), system-wide
4392 coalescing of timer events, exit handlers, watchdog
4393 supervision support using systemd's sd_notify() API, child
4396 * A new API "sd-rntl.h" has been added that provides an API
4397 around the route netlink interface of the kernel, similar in
4398 style to "sd-bus.h".
4400 * A new API "sd-dhcp-client.h" has been added that provides a
4401 small DHCPv4 client-side implementation. This is used by
4404 * There is a new kernel command line option
4405 "systemd.restore_state=0|1". When set to "0", none of the
4406 systemd tools will restore saved runtime state to hardware
4407 devices. More specifically, the rfkill and backlight states
4410 * The FsckPassNo= compatibility option in mount/service units
4411 has been removed. The fstab generator will now add the
4412 necessary dependencies automatically, and does not require
4413 PID1's support for that anymore.
4415 * journalctl gained a new switch, --list-boots, that lists
4416 recent boots with their times and boot IDs.
4418 * The various tools like systemctl, loginctl, timedatectl,
4419 busctl, systemd-run, ... have gained a new switch "-M" to
4420 connect to a specific, local OS container (as direct
4421 connection, without requiring SSH). This works on any
4422 container that is registered with machined, such as those
4423 created by libvirt-lxc or nspawn.
4425 * systemd-run and systemd-analyze also gained support for "-H"
4426 to connect to remote hosts via SSH. This is particularly
4427 useful for systemd-run because it enables queuing of jobs
4428 onto remote systems.
4430 * machinectl gained a new command "login" to open a getty
4431 login in any local container. This works with any container
4432 that is registered with machined (such as those created by
4433 libvirt-lxc or nspawn), and which runs systemd inside.
4435 * machinectl gained a new "reboot" command that may be used to
4436 trigger a reboot on a specific container that is registered
4437 with machined. This works on any container that runs an init
4438 system of some kind.
4440 * systemctl gained a new "list-timers" command to print a nice
4441 listing of installed timer units with the times they elapse
4444 * Alternative reboot() parameters may now be specified on the
4445 "systemctl reboot" command line and are passed to the
4446 reboot() system call.
4448 * systemctl gained a new --job-mode= switch to configure the
4449 mode to queue a job with. This is a more generic version of
4450 --fail, --irreversible, and --ignore-dependencies, which are
4451 still available but not advertised anymore.
4453 * /etc/systemd/system.conf gained new settings to configure
4454 various default timeouts of units, as well as the default
4455 start limit interval and burst. These may still be overridden
4458 * PID1 will now export on the bus profile data of the security
4459 policy upload process (such as the SELinux policy upload to
4462 * journald: when forwarding logs to the console, include
4463 timestamps (following the setting in
4464 /sys/module/printk/parameters/time).
4466 * OnCalendar= in timer units now understands the special
4467 strings "yearly" and "annually". (Both are equivalent)
4469 * The accuracy of timer units is now configurable with the new
4470 AccuracySec= setting. It defaults to 1min.
4472 * A new dependency type JoinsNamespaceOf= has been added that
4473 allows running two services within the same /tmp and network
4474 namespace, if PrivateNetwork= or PrivateTmp= are used.
4476 * A new command "cat" has been added to systemctl. It outputs
4477 the original unit file of a unit, and concatenates the
4478 contents of additional "drop-in" unit file snippets, so that
4479 the full configuration is shown.
4481 * systemctl now supports globbing on the various "list-xyz"
4482 commands, like "list-units" or "list-sockets", as well as on
4483 those commands which take multiple unit names.
4485 * journalctl's --unit= switch gained support for globbing.
4487 * All systemd daemons now make use of the watchdog logic so
4488 that systemd automatically notices when they hang.
4490 * If the $container_ttys environment variable is set,
4491 getty-generator will automatically spawn a getty for each
4492 listed tty. This is useful for container managers to request
4493 login gettys to be spawned on as many ttys as needed.
4495 * %h, %s, %U specifier support is not available anymore when
4496 used in unit files for PID 1. This is because NSS calls are
4497 not safe from PID 1. They stay available for --user
4498 instances of systemd, and as special case for the root user.
4500 * loginctl gained a new "--no-legend" switch to turn off output
4503 * The "sd-login.h" API gained three new calls:
4504 sd_session_is_remote(), sd_session_get_remote_user(),
4505 sd_session_get_remote_host() to query information about
4508 * The udev hardware database now also carries vendor/product
4509 information of SDIO devices.
4511 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
4512 determine whether watchdog notifications are requested by
4515 * Socket-activated per-connection services now include a
4516 short description of the connection parameters in the
4519 * tmpfiles gained a new "--boot" option. When this is not used,
4520 only lines where the command character is not suffixed with
4521 "!" are executed. When this option is specified, those
4522 options are executed too. This partitions tmpfiles
4523 directives into those that can be safely executed at any
4524 time, and those which should be run only at boot (for
4525 example, a line that creates /run/nologin).
4527 * A new API "sd-resolve.h" has been added which provides a simple
4528 asynchronous wrapper around glibc NSS host name resolution
4529 calls, such as getaddrinfo(). In contrast to glibc's
4530 getaddrinfo_a(), it does not use signals. In contrast to most
4531 other asynchronous name resolution libraries, this one does
4532 not reimplement DNS, but reuses NSS, so that alternate
4533 host name resolution systems continue to work, such as mDNS,
4534 LDAP, etc. This API is based on libasyncns, but it has been
4535 cleaned up for inclusion in systemd.
4537 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
4538 "sd-daemon.h" are no longer found in individual libraries
4539 libsystemd-journal.so, libsystemd-login.so,
4540 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
4541 merged them into a single library, libsystemd.so, which
4542 provides all symbols. The reason for this is cyclic
4543 dependencies, as these libraries tend to use each other's
4544 symbols. So far, we have managed to workaround that by linking
4545 a copy of a good part of our code into each of these
4546 libraries again and again, which, however, makes certain
4547 things hard to do, like sharing static variables. Also, it
4548 substantially increases footprint. With this change, there
4549 is only one library for the basic APIs systemd
4550 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
4551 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
4552 library as well, however are subject to the --enable-kdbus
4553 switch (see below). Note that "sd-dhcp-client.h" is not part
4554 of this library (this is because it only consumes, never
4555 provides, services of/to other APIs). To make the transition
4556 easy from the separate libraries to the unified one, we
4557 provide the --enable-compat-libs compile-time switch which
4558 will generate stub libraries that are compatible with the
4559 old ones but redirect all calls to the new one.
4561 * All of the kdbus logic and the new APIs "sd-bus.h",
4562 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
4563 and "sd-utf8.h" are compile-time optional via the
4564 "--enable-kdbus" switch, and they are not compiled in by
4565 default. To make use of kdbus, you have to explicitly enable
4566 the switch. Note however, that neither the kernel nor the
4567 userspace API for all of this is considered stable yet. We
4568 want to maintain the freedom to still change the APIs for
4569 now. By specifying this build-time switch, you acknowledge
4570 that you are aware of the instability of the current
4573 * Also, note that while kdbus is pretty much complete,
4574 it lacks one thing: proper policy support. This means you
4575 can build a fully working system with all features; however,
4576 it will be highly insecure. Policy support will be added in
4577 one of the next releases, at the same time that we will
4578 declare the APIs stable.
4580 * When the kernel command line argument "kdbus" is specified,
4581 systemd will automatically load the kdbus.ko kernel module. At
4582 this stage of development, it is only useful for testing kdbus
4583 and should not be used in production. Note: if "--enable-kdbus"
4584 is specified, and the kdbus.ko kernel module is available, and
4585 "kdbus" is added to the kernel command line, the entire system
4586 runs with kdbus instead of dbus-daemon, with the above mentioned
4587 problem of missing the system policy enforcement. Also a future
4588 version of kdbus.ko or a newer systemd will not be compatible with
4589 each other, and will unlikely be able to boot the machine if only
4590 one of them is updated.
4592 * systemctl gained a new "import-environment" command which
4593 uploads the caller's environment (or parts thereof) into the
4594 service manager so that it is inherited by services started
4595 by the manager. This is useful to upload variables like
4596 $DISPLAY into the user service manager.
4598 * A new PrivateDevices= switch has been added to service units
4599 which allows running a service with a namespaced /dev
4600 directory that does not contain any device nodes for
4601 physical devices. More specifically, it only includes devices
4602 such as /dev/null, /dev/urandom, and /dev/zero which are API
4605 * logind has been extended to support behaviour like VT
4606 switching on seats that do not support a VT. This makes
4607 multi-session available on seats that are not the first seat
4608 (seat0), and on systems where kernel support for VTs has
4609 been disabled at compile-time.
4611 * If a process holds a delay lock for system sleep or shutdown
4612 and fails to release it in time, we will now log its
4613 identity. This makes it easier to identify processes that
4614 cause slow suspends or power-offs.
4616 * When parsing /etc/crypttab, support for a new key-slot=
4617 option as supported by Debian is added. It allows indicating
4618 which LUKS slot to use on disk, speeding up key loading.
4620 * The sd_journald_sendv() API call has been checked and
4621 officially declared to be async-signal-safe so that it may
4622 be invoked from signal handlers for logging purposes.
4624 * Boot-time status output is now enabled automatically after a
4625 short timeout if boot does not progress, in order to give
4626 the user an indication what she or he is waiting for.
4628 * The boot-time output has been improved to show how much time
4629 remains until jobs expire.
4631 * The KillMode= switch in service units gained a new possible
4632 value "mixed". If set, and the unit is shut down, then the
4633 initial SIGTERM signal is sent only to the main daemon
4634 process, while the following SIGKILL signal is sent to
4635 all remaining processes of the service.
4637 * When a scope unit is registered, a new property "Controller"
4638 may be set. If set to a valid bus name, systemd will send a
4639 RequestStop() signal to this name when it would like to shut
4640 down the scope. This may be used to hook manager logic into
4641 the shutdown logic of scope units. Also, scope units may now
4642 be put in a special "abandoned" state, in which case the
4643 manager process which created them takes no further
4644 responsibilities for it.
4646 * When reading unit files, systemd will now verify
4647 the access mode of these files, and warn about certain
4648 suspicious combinations. This has been added to make it
4649 easier to track down packaging bugs where unit files are
4650 marked executable or world-writable.
4652 * systemd-nspawn gained a new "--setenv=" switch to set
4653 container-wide environment variables. The similar option in
4654 systemd-activate was renamed from "--environment=" to
4655 "--setenv=" for consistency.
4657 * systemd-nspawn has been updated to create a new kdbus domain
4658 for each container that is invoked, thus allowing each
4659 container to have its own set of system and user buses,
4660 independent of the host.
4662 * systemd-nspawn gained a new --drop-capability= switch to run
4663 the container with less capabilities than the default. Both
4664 --drop-capability= and --capability= now take the special
4665 string "all" for dropping or keeping all capabilities.
4667 * systemd-nspawn gained new switches for executing containers
4668 with specific SELinux labels set.
4670 * systemd-nspawn gained a new --quiet switch to not generate
4671 any additional output but the container's own console
4674 * systemd-nspawn gained a new --share-system switch to run a
4675 container without PID namespacing enabled.
4677 * systemd-nspawn gained a new --register= switch to control
4678 whether the container is registered with systemd-machined or
4679 not. This is useful for containers that do not run full
4680 OS images, but only specific apps.
4682 * systemd-nspawn gained a new --keep-unit which may be used
4683 when invoked as the only program from a service unit, and
4684 results in registration of the unit service itself in
4685 systemd-machined, instead of a newly opened scope unit.
4687 * systemd-nspawn gained a new --network-interface= switch for
4688 moving arbitrary interfaces to the container. The new
4689 --network-veth switch creates a virtual Ethernet connection
4690 between host and container. The new --network-bridge=
4691 switch then allows assigning the host side of this virtual
4692 Ethernet connection to a bridge device.
4694 * systemd-nspawn gained a new --personality= switch for
4695 setting the kernel personality for the container. This is
4696 useful when running a 32-bit container on a 64-bit host. A
4697 similar option Personality= is now also available for service
4700 * logind will now also track a "Desktop" identifier for each
4701 session which encodes the desktop environment of it. This is
4702 useful for desktop environments that want to identify
4703 multiple running sessions of itself easily.
4705 * A new SELinuxContext= setting for service units has been
4706 added that allows setting a specific SELinux execution
4707 context for a service.
4709 * Most systemd client tools will now honour $SYSTEMD_LESS for
4710 settings of the "less" pager. By default, these tools will
4711 override $LESS to allow certain operations to work, such as
4712 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
4713 influence this logic.
4715 * systemd's "seccomp" hook-up has been changed to make use of
4716 the libseccomp library instead of using its own
4717 implementation. This has benefits for portability among
4720 * For usage together with SystemCallFilter=, a new
4721 SystemCallErrorNumber= setting has been introduced that
4722 allows configuration of a system error number to be returned
4723 on filtered system calls, instead of immediately killing the
4724 process. Also, SystemCallArchitectures= has been added to
4725 limit access to system calls of a particular architecture
4726 (in order to turn off support for unused secondary
4727 architectures). There is also a global
4728 SystemCallArchitectures= setting in system.conf now to turn
4729 off support for non-native system calls system-wide.
4731 * systemd requires a kernel with a working name_to_handle_at(),
4732 please see the kernel config requirements in the README file.
4734 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
4735 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
4736 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
4737 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
4738 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
4739 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
4740 Elia Pinto, Florian Weimer, George McCollister, Goffredo
4741 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
4742 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
4743 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
4744 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
4745 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
4746 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
4747 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
4748 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
4749 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
4750 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
4751 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
4752 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
4753 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
4754 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
4755 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
4756 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
4757 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
4759 — Berlin, 2014-02-20
4763 * logind has gained support for facilitating privileged input
4764 and drm device access for unprivileged clients. This work is
4765 useful to allow Wayland display servers (and similar
4766 programs, such as kmscon) to run under the user's ID and
4767 access input and drm devices which are normally
4768 protected. When this is used (and the kernel is new enough)
4769 logind will "mute" IO on the file descriptors passed to
4770 Wayland as long as it is in the background and "unmute" it
4771 if it returns into the foreground. This allows secure
4772 session switching without allowing background sessions to
4773 eavesdrop on input and display data. This also introduces
4774 session switching support if VT support is turned off in the
4775 kernel, and on seats that are not seat0.
4777 * A new kernel command line option luks.options= is understood
4778 now which allows specifying LUKS options for usage for LUKS
4779 encrypted partitions specified with luks.uuid=.
4781 * tmpfiles.d(5) snippets may now use specifier expansion in
4782 path names. More specifically %m, %b, %H, %v, are now
4783 replaced by the local machine id, boot id, hostname, and
4784 kernel version number.
4786 * A new tmpfiles.d(5) command "m" has been introduced which
4787 may be used to change the owner/group/access mode of a file
4788 or directory if it exists, but do nothing if it does not.
4790 * This release removes high-level support for the
4791 MemorySoftLimit= cgroup setting. The underlying kernel
4792 cgroup attribute memory.soft_limit= is currently badly
4793 designed and likely to be removed from the kernel API in its
4794 current form, hence we should not expose it for now.
4796 * The memory.use_hierarchy cgroup attribute is now enabled for
4797 all cgroups systemd creates in the memory cgroup
4798 hierarchy. This option is likely to be come the built-in
4799 default in the kernel anyway, and the non-hierarchical mode
4800 never made much sense in the intrinsically hierarchical
4803 * A new field _SYSTEMD_SLICE= is logged along with all journal
4804 messages containing the slice a message was generated
4805 from. This is useful to allow easy per-customer filtering of
4806 logs among other things.
4808 * systemd-journald will no longer adjust the group of journal
4809 files it creates to the "systemd-journal" group. Instead we
4810 rely on the journal directory to be owned by the
4811 "systemd-journal" group, and its setgid bit set, so that the
4812 kernel file system layer will automatically enforce that
4813 journal files inherit this group assignment. The reason for
4814 this change is that we cannot allow NSS look-ups from
4815 journald which would be necessary to resolve
4816 "systemd-journal" to a numeric GID, because this might
4817 create deadlocks if NSS involves synchronous queries to
4818 other daemons (such as nscd, or sssd) which in turn are
4819 logging clients of journald and might block on it, which
4820 would then dead lock. A tmpfiles.d(5) snippet included in
4821 systemd will make sure the setgid bit and group are
4822 properly set on the journal directory if it exists on every
4823 boot. However, we recommend adjusting it manually after
4824 upgrades too (or from RPM scriptlets), so that the change is
4825 not delayed until next reboot.
4827 * Backlight and random seed files in /var/lib/ have moved into
4828 the /var/lib/systemd/ directory, in order to centralize all
4829 systemd generated files in one directory.
4831 * Boot time performance measurements (as displayed by
4832 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
4833 performance information if that's available to determine how
4834 much time BIOS and boot loader initialization required. With
4835 a sufficiently new BIOS you hence no longer need to boot
4836 with Gummiboot to get access to such information.
4838 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
4839 Cristian Rodríguez, Dave Reisner, David Herrmann, David
4840 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
4841 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
4842 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
4843 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
4844 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
4846 — Berlin, 2013-10-02
4850 * The Restart= option for services now understands a new
4851 on-watchdog setting, which will restart the service
4852 automatically if the service stops sending out watchdog keep
4853 alive messages (as configured with WatchdogSec=).
4855 * The getty generator (which is responsible for bringing up a
4856 getty on configured serial consoles) will no longer only
4857 start a getty on the primary kernel console but on all
4858 others, too. This makes the order in which console= is
4859 specified on the kernel command line less important.
4861 * libsystemd-logind gained a new sd_session_get_vt() call to
4862 retrieve the VT number of a session.
4864 * If the option "tries=0" is set for an entry of /etc/crypttab
4865 its passphrase is queried indefinitely instead of any
4866 maximum number of tries.
4868 * If a service with a configure PID file terminates its PID
4869 file will now be removed automatically if it still exists
4870 afterwards. This should put an end to stale PID files.
4872 * systemd-run will now also take relative binary path names
4873 for execution and no longer insists on absolute paths.
4875 * InaccessibleDirectories= and ReadOnlyDirectories= now take
4876 paths that are optionally prefixed with "-" to indicate that
4877 it should not be considered a failure if they do not exist.
4879 * journalctl -o (and similar commands) now understands a new
4880 output mode "short-precise", it is similar to "short" but
4881 shows timestamps with usec accuracy.
4883 * The option "discard" (as known from Debian) is now
4884 synonymous to "allow-discards" in /etc/crypttab. In fact,
4885 "discard" is preferred now (since it is easier to remember
4888 * Some licensing clean-ups were made, so that more code is now
4889 LGPL-2.1 licensed than before.
4891 * A minimal tool to save/restore the display backlight
4892 brightness across reboots has been added. It will store the
4893 backlight setting as late as possible at shutdown, and
4894 restore it as early as possible during reboot.
4896 * A logic to automatically discover and enable home and swap
4897 partitions on GPT disks has been added. With this in place
4898 /etc/fstab becomes optional for many setups as systemd can
4899 discover certain partitions located on the root disk
4900 automatically. Home partitions are recognized under their
4901 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
4902 partitions are recognized under their GPT type ID
4903 0657fd6da4ab43c484e50933c84b4f4f.
4905 * systemd will no longer pass any environment from the kernel
4906 or initrd to system services. If you want to set an
4907 environment for all services, do so via the kernel command
4908 line systemd.setenv= assignment.
4910 * The systemd-sysctl tool no longer natively reads the file
4911 /etc/sysctl.conf. If desired, the file should be symlinked
4912 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
4913 legacy support by a symlink rather than built-in code, it
4914 also makes the otherwise hidden order of application of the
4915 different files visible. (Note that this partly reverts to a
4916 pre-198 application order of sysctl knobs!)
4918 * The "systemctl set-log-level" and "systemctl dump" commands
4919 have been moved to systemd-analyze.
4921 * systemd-run learned the new --remain-after-exit switch,
4922 which causes the scope unit not to be cleaned up
4923 automatically after the process terminated.
4925 * tmpfiles learned a new --exclude-prefix= switch to exclude
4926 certain paths from operation.
4928 * journald will now automatically flush all messages to disk
4929 as soon as a message at the log level CRIT, ALERT or EMERG
4932 Contributions from: Andrew Cook, Brandon Philips, Christian
4933 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
4934 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
4935 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
4936 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
4937 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
4938 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
4939 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
4940 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
4941 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
4942 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
4943 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
4944 William Giokas, Zbigniew Jędrzejewski-Szmek
4946 — Berlin, 2013-09-13
4950 * The documentation has been updated to cover the various new
4951 concepts introduced with 205.
4953 * Unit files now understand the new %v specifier which
4954 resolves to the kernel version string as returned by "uname
4957 * systemctl now supports filtering the unit list output by
4958 load state, active state and sub state, using the new
4961 * "systemctl status" will now show the results of the
4962 condition checks (like ConditionPathExists= and similar) of
4963 the last start attempts of the unit. They are also logged to
4966 * "journalctl -b" may now be used to look for boot output of a
4967 specific boot. Try "journalctl -b -1" for the previous boot,
4968 but the syntax is substantially more powerful.
4970 * "journalctl --show-cursor" has been added which prints the
4971 cursor string the last shown log line. This may then be used
4972 with the new "journalctl --after-cursor=" switch to continue
4973 browsing logs from that point on.
4975 * "journalctl --force" may now be used to force regeneration
4978 * Creation of "dead" device nodes has been moved from udev
4979 into kmod and tmpfiles. Previously, udev would read the kmod
4980 databases to pre-generate dead device nodes based on meta
4981 information contained in kernel modules, so that these would
4982 be auto-loaded on access rather then at boot. As this
4983 does not really have much to do with the exposing actual
4984 kernel devices to userspace this has always been slightly
4985 alien in the udev codebase. Following the new scheme kmod
4986 will now generate a runtime snippet for tmpfiles from the
4987 module meta information and it now is tmpfiles' job to the
4988 create the nodes. This also allows overriding access and
4989 other parameters for the nodes using the usual tmpfiles
4990 facilities. As side effect this allows us to remove the
4991 CAP_SYS_MKNOD capability bit from udevd entirely.
4993 * logind's device ACLs may now be applied to these "dead"
4994 devices nodes too, thus finally allowing managed access to
4995 devices such as /dev/snd/sequencer without loading the
4996 backing module right-away.
4998 * A new RPM macro has been added that may be used to apply
4999 tmpfiles configuration during package installation.
5001 * systemd-detect-virt and ConditionVirtualization= now can
5002 detect User-Mode-Linux machines (UML).
5004 * journald will now implicitly log the effective capabilities
5005 set of processes in the message metadata.
5007 * systemd-cryptsetup has gained support for TrueCrypt volumes.
5009 * The initrd interface has been simplified (more specifically,
5010 support for passing performance data via environment
5011 variables and fsck results via files in /run has been
5012 removed). These features were non-essential, and are
5013 nowadays available in a much nicer way by having systemd in
5014 the initrd serialize its state and have the hosts systemd
5015 deserialize it again.
5017 * The udev "keymap" data files and tools to apply keyboard
5018 specific mappings of scan to key codes, and force-release
5019 scan code lists have been entirely replaced by a udev
5020 "keyboard" builtin and a hwdb data file.
5022 * systemd will now honour the kernel's "quiet" command line
5023 argument also during late shutdown, resulting in a
5024 completely silent shutdown when used.
5026 * There's now an option to control the SO_REUSEPORT socket
5027 option in .socket units.
5029 * Instance units will now automatically get a per-template
5030 subslice of system.slice unless something else is explicitly
5031 configured. For example, instances of sshd@.service will now
5032 implicitly be placed in system-sshd.slice rather than
5033 system.slice as before.
5035 * Test coverage support may now be enabled at build time.
5037 Contributions from: Dave Reisner, Frederic Crozat, Harald
5038 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
5039 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
5040 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
5041 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
5042 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
5043 Giokas, Zbigniew Jędrzejewski-Szmek
5045 — Berlin, 2013-07-23
5049 * Two new unit types have been introduced:
5051 Scope units are very similar to service units, however, are
5052 created out of pre-existing processes — instead of PID 1
5053 forking off the processes. By using scope units it is
5054 possible for system services and applications to group their
5055 own child processes (worker processes) in a powerful way
5056 which then maybe used to organize them, or kill them
5057 together, or apply resource limits on them.
5059 Slice units may be used to partition system resources in an
5060 hierarchical fashion and then assign other units to them. By
5061 default there are now three slices: system.slice (for all
5062 system services), user.slice (for all user sessions),
5063 machine.slice (for VMs and containers).
5065 Slices and scopes have been introduced primarily in
5066 context of the work to move cgroup handling to a
5067 single-writer scheme, where only PID 1
5068 creates/removes/manages cgroups.
5070 * There's a new concept of "transient" units. In contrast to
5071 normal units these units are created via an API at runtime,
5072 not from configuration from disk. More specifically this
5073 means it is now possible to run arbitrary programs as
5074 independent services, with all execution parameters passed
5075 in via bus APIs rather than read from disk. Transient units
5076 make systemd substantially more dynamic then it ever was,
5077 and useful as a general batch manager.
5079 * logind has been updated to make use of scope and slice units
5080 for managing user sessions. As a user logs in he will get
5081 his own private slice unit, to which all sessions are added
5082 as scope units. We also added support for automatically
5083 adding an instance of user@.service for the user into the
5084 slice. Effectively logind will no longer create cgroup
5085 hierarchies on its own now, it will defer entirely to PID 1
5086 for this by means of scope, service and slice units. Since
5087 user sessions this way become entities managed by PID 1
5088 the output of "systemctl" is now a lot more comprehensive.
5090 * A new mini-daemon "systemd-machined" has been added which
5091 may be used by virtualization managers to register local
5092 VMs/containers. nspawn has been updated accordingly, and
5093 libvirt will be updated shortly. machined will collect a bit
5094 of meta information about the VMs/containers, and assign
5095 them their own scope unit (see above). The collected
5096 meta-data is then made available via the "machinectl" tool,
5097 and exposed in "ps" and similar tools. machined/machinectl
5098 is compile-time optional.
5100 * As discussed earlier, the low-level cgroup configuration
5101 options ControlGroup=, ControlGroupModify=,
5102 ControlGroupPersistent=, ControlGroupAttribute= have been
5103 removed. Please use high-level attribute settings instead as
5104 well as slice units.
5106 * A new bus call SetUnitProperties() has been added to alter
5107 various runtime parameters of a unit. This is primarily
5108 useful to alter cgroup parameters dynamically in a nice way,
5109 but will be extended later on to make more properties
5110 modifiable at runtime. systemctl gained a new set-properties
5111 command that wraps this call.
5113 * A new tool "systemd-run" has been added which can be used to
5114 run arbitrary command lines as transient services or scopes,
5115 while configuring a number of settings via the command
5116 line. This tool is currently very basic, however already
5117 very useful. We plan to extend this tool to even allow
5118 queuing of execution jobs with time triggers from the
5119 command line, similar in fashion to "at".
5121 * nspawn will now inform the user explicitly that kernels with
5122 audit enabled break containers, and suggest the user to turn
5125 * Support for detecting the IMA and AppArmor security
5126 frameworks with ConditionSecurity= has been added.
5128 * journalctl gained a new "-k" switch for showing only kernel
5129 messages, mimicking dmesg output; in addition to "--user"
5130 and "--system" switches for showing only user's own logs
5133 * systemd-delta can now show information about drop-in
5134 snippets extending unit files.
5136 * libsystemd-bus has been substantially updated but is still
5137 not available as public API.
5139 * systemd will now look for the "debug" argument on the kernel
5140 command line and enable debug logging, similar to what
5141 "systemd.log_level=debug" already did before.
5143 * "systemctl set-default", "systemctl get-default" has been
5144 added to configure the default.target symlink, which
5145 controls what to boot into by default.
5147 * "systemctl set-log-level" has been added as a convenient
5148 way to raise and lower systemd logging threshold.
5150 * "systemd-analyze plot" will now show the time the various
5151 generators needed for execution, as well as information
5152 about the unit file loading.
5154 * libsystemd-journal gained a new sd_journal_open_files() call
5155 for opening specific journal files. journactl also gained a
5156 new switch to expose this new functionality. Previously we
5157 only supported opening all files from a directory, or all
5158 files from the system, as opening individual files only is
5159 racy due to journal file rotation.
5161 * systemd gained the new DefaultEnvironment= setting in
5162 /etc/systemd/system.conf to set environment variables for
5165 * If a privileged process logs a journal message with the
5166 OBJECT_PID= field set, then journald will automatically
5167 augment this with additional OBJECT_UID=, OBJECT_GID=,
5168 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
5169 system services want to log events about specific client
5170 processes. journactl/systemctl has been updated to make use
5171 of this information if all log messages regarding a specific
5174 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
5175 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
5176 Reisner, David Coppa, David King, David Strauss, Eelco
5177 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
5178 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
5179 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
5180 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
5181 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
5182 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
5183 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
5184 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
5185 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
5186 Łukasz Stelmach, 장동준
5190 * The Python bindings gained some minimal support for the APIs
5191 exposed by libsystemd-logind.
5193 * ConditionSecurity= gained support for detecting SMACK. Since
5194 this condition already supports SELinux and AppArmor we only
5195 miss IMA for this. Patches welcome!
5197 Contributions from: Karol Lewandowski, Lennart Poettering,
5198 Zbigniew Jędrzejewski-Szmek
5202 * systemd-nspawn will now create /etc/resolv.conf if
5203 necessary, before bind-mounting the host's file onto it.
5205 * systemd-nspawn will now store meta information about a
5206 container on the container's cgroup as extended attribute
5207 fields, including the root directory.
5209 * The cgroup hierarchy has been reworked in many ways. All
5210 objects any of the components systemd creates in the cgroup
5211 tree are now suffixed. More specifically, user sessions are
5212 now placed in cgroups suffixed with ".session", users in
5213 cgroups suffixed with ".user", and nspawn containers in
5214 cgroups suffixed with ".nspawn". Furthermore, all cgroup
5215 names are now escaped in a simple scheme to avoid collision
5216 of userspace object names with kernel filenames. This work
5217 is preparation for making these objects relocatable in the
5218 cgroup tree, in order to allow easy resource partitioning of
5219 these objects without causing naming conflicts.
5221 * systemctl list-dependencies gained the new switches
5222 --plain, --reverse, --after and --before.
5224 * systemd-inhibit now shows the process name of processes that
5225 have taken an inhibitor lock.
5227 * nss-myhostname will now also resolve "localhost"
5228 implicitly. This makes /etc/hosts an optional file and
5229 nicely handles that on IPv6 ::1 maps to both "localhost" and
5232 * libsystemd-logind.so gained a new call
5233 sd_get_machine_names() to enumerate running containers and
5234 VMs (currently only supported by very new libvirt and
5235 nspawn). sd_login_monitor can now be used to watch
5236 VMs/containers coming and going.
5238 * .include is not allowed recursively anymore, and only in
5239 unit files. Usually it is better to use drop-in snippets in
5240 .d/*.conf anyway, as introduced with systemd 198.
5242 * systemd-analyze gained a new "critical-chain" command that
5243 determines the slowest chain of units run during system
5244 boot-up. It is very useful for tracking down where
5245 optimizing boot time is the most beneficial.
5247 * systemd will no longer allow manipulating service paths in
5248 the name=systemd:/system cgroup tree using ControlGroup= in
5249 units. (But is still fine with it in all other dirs.)
5251 * There's a new systemd-nspawn@.service service file that may
5252 be used to easily run nspawn containers as system
5253 services. With the container's root directory in
5254 /var/lib/container/foobar it is now sufficient to run
5255 "systemctl start systemd-nspawn@foobar.service" to boot it.
5257 * systemd-cgls gained a new parameter "--machine" to list only
5258 the processes within a certain container.
5260 * ConditionSecurity= now can check for "apparmor". We still
5261 are lacking checks for SMACK and IMA for this condition
5262 check though. Patches welcome!
5264 * A new configuration file /etc/systemd/sleep.conf has been
5265 added that may be used to configure which kernel operation
5266 systemd is supposed to execute when "suspend", "hibernate"
5267 or "hybrid-sleep" is requested. This makes the new kernel
5268 "freeze" state accessible to the user.
5270 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
5271 the passed argument if applicable.
5273 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
5274 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
5275 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
5276 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
5277 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
5278 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
5279 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
5284 * The output of 'systemctl list-jobs' got some polishing. The
5285 '--type=' argument may now be passed more than once. A new
5286 command 'systemctl list-sockets' has been added which shows
5287 a list of kernel sockets systemd is listening on with the
5288 socket units they belong to, plus the units these socket
5291 * The experimental libsystemd-bus library got substantial
5292 updates to work in conjunction with the (also experimental)
5293 kdbus kernel project. It works well enough to exchange
5294 messages with some sophistication. Note that kdbus is not
5295 ready yet, and the library is mostly an elaborate test case
5296 for now, and not installable.
5298 * systemd gained a new unit 'systemd-static-nodes.service'
5299 that generates static device nodes earlier during boot, and
5300 can run in conjunction with udev.
5302 * libsystemd-login gained a new call sd_pid_get_user_unit()
5303 to retrieve the user systemd unit a process is running
5304 in. This is useful for systems where systemd is used as
5307 * systemd-nspawn now places all containers in the new /machine
5308 top-level cgroup directory in the name=systemd
5309 hierarchy. libvirt will soon do the same, so that we get a
5310 uniform separation of /system, /user and /machine for system
5311 services, user processes and containers/virtual
5312 machines. This new cgroup hierarchy is also useful to stick
5313 stable names to specific container instances, which can be
5314 recognized later this way (this name may be controlled
5315 via systemd-nspawn's new -M switch). libsystemd-login also
5316 gained a new call sd_pid_get_machine_name() to retrieve the
5317 name of the container/VM a specific process belongs to.
5319 * bootchart can now store its data in the journal.
5321 * libsystemd-journal gained a new call
5322 sd_journal_add_conjunction() for AND expressions to the
5323 matching logic. This can be used to express more complex
5324 logical expressions.
5326 * journactl can now take multiple --unit= and --user-unit=
5329 * The cryptsetup logic now understands the "luks.key=" kernel
5330 command line switch for specifying a file to read the
5331 decryption key from. Also, if a configured key file is not
5332 found the tool will now automatically fall back to prompting
5335 * Python systemd.journal module was updated to wrap recently
5336 added functions from libsystemd-journal. The interface was
5337 changed to bring the low level interface in s.j._Reader
5338 closer to the C API, and the high level interface in
5339 s.j.Reader was updated to wrap and convert all data about
5342 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
5343 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
5344 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
5345 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
5346 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
5347 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5351 * journalctl --update-catalog now understands a new --root=
5352 option to operate on catalogs found in a different root
5355 * During shutdown after systemd has terminated all running
5356 services a final killing loop kills all remaining left-over
5357 processes. We will now print the name of these processes
5358 when we send SIGKILL to them, since this usually indicates a
5361 * If /etc/crypttab refers to password files stored on
5362 configured mount points automatic dependencies will now be
5363 generated to ensure the specific mount is established first
5364 before the key file is attempted to be read.
5366 * 'systemctl status' will now show information about the
5367 network sockets a socket unit is listening on.
5369 * 'systemctl status' will also shown information about any
5370 drop-in configuration file for units. (Drop-In configuration
5371 files in this context are files such as
5372 /etc/systemd/systemd/foobar.service.d/*.conf)
5374 * systemd-cgtop now optionally shows summed up CPU times of
5375 cgroups. Press '%' while running cgtop to switch between
5376 percentage and absolute mode. This is useful to determine
5377 which cgroups use up the most CPU time over the entire
5378 runtime of the system. systemd-cgtop has also been updated
5379 to be 'pipeable' for processing with further shell tools.
5381 * 'hostnamectl set-hostname' will now allow setting of FQDN
5384 * The formatting and parsing of time span values has been
5385 changed. The parser now understands fractional expressions
5386 such as "5.5h". The formatter will now output fractional
5387 expressions for all time spans under 1min, i.e. "5.123456s"
5388 rather than "5s 123ms 456us". For time spans under 1s
5389 millisecond values are shown, for those under 1ms
5390 microsecond values are shown. This should greatly improve
5391 all time-related output of systemd.
5393 * libsystemd-login and libsystemd-journal gained new
5394 functions for querying the poll() events mask and poll()
5395 timeout value for integration into arbitrary event
5398 * localectl gained the ability to list available X11 keymaps
5399 (models, layouts, variants, options).
5401 * 'systemd-analyze dot' gained the ability to filter for
5402 specific units via shell-style globs, to create smaller,
5403 more useful graphs. I.e. it is now possible to create simple
5404 graphs of all the dependencies between only target units, or
5405 of all units that Avahi has dependencies with.
5407 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
5408 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
5409 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
5410 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
5411 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
5412 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
5413 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
5417 * The boot-time readahead implementation for rotating media
5418 will now read the read-ahead data in multiple passes which
5419 consist of all read requests made in equidistant time
5420 intervals. This means instead of strictly reading read-ahead
5421 data in its physical order on disk we now try to find a
5422 middle ground between physical and access time order.
5424 * /etc/os-release files gained a new BUILD_ID= field for usage
5425 on operating systems that provide continuous builds of OS
5428 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
5429 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
5430 William Douglas, Zbigniew Jędrzejewski-Szmek
5434 * systemd-python gained an API exposing libsystemd-daemon.
5436 * The SMACK setup logic gained support for uploading CIPSO
5439 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
5440 ReadOnlyDirectories= and InaccessibleDirectories= has
5441 changed. The private /tmp and /var/tmp directories are now
5442 shared by all processes of a service (which means
5443 ExecStartPre= may now leave data in /tmp that ExecStart= of
5444 the same service can still access). When a service is
5445 stopped its temporary directories are immediately deleted
5446 (normal clean-up with tmpfiles is still done in addition to
5449 * By default, systemd will now set a couple of sysctl
5450 variables in the kernel: the safe sysrq options are turned
5451 on, IP route verification is turned on, and source routing
5452 disabled. The recently added hardlink and softlink
5453 protection of the kernel is turned on. These settings should
5454 be reasonably safe, and good defaults for all new systems.
5456 * The predictable network naming logic may now be turned off
5457 with a new kernel command line switch: net.ifnames=0.
5459 * A new libsystemd-bus module has been added that implements a
5460 pretty complete D-Bus client library. For details see:
5462 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
5464 * journald will now explicitly flush the journal files to disk
5465 at the latest 5min after each write. The file will then also
5466 be marked offline until the next write. This should increase
5467 reliability in case of a crash. The synchronization delay
5468 can be configured via SyncIntervalSec= in journald.conf.
5470 * There's a new remote-fs-setup.target unit that can be used
5471 to pull in specific services when at least one remote file
5472 system is to be mounted.
5474 * There are new targets timers.target and paths.target as
5475 canonical targets to pull user timer and path units in
5476 from. This complements sockets.target with a similar
5477 purpose for socket units.
5479 * libudev gained a new call udev_device_set_attribute_value()
5480 to set sysfs attributes of a device.
5482 * The udev daemon now sets the default number of worker
5483 processes executed in parallel based on the number of available
5484 CPUs instead of the amount of available RAM. This is supposed
5485 to provide a more reliable default and limit a too aggressive
5486 parallelism for setups with 1000s of devices connected.
5488 Contributions from: Auke Kok, Colin Walters, Cristian
5489 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
5490 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
5491 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
5492 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
5493 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
5494 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
5495 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
5496 Zbigniew Jędrzejewski-Szmek
5500 * Configuration of unit files may now be extended via drop-in
5501 files without having to edit/override the unit files
5502 themselves. More specifically, if the administrator wants to
5503 change one value for a service file foobar.service he can
5504 now do so by dropping in a configuration snippet into
5505 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
5506 will load all these snippets and apply them on top of the
5507 main unit configuration file, possibly extending or
5508 overriding its settings. Using these drop-in snippets is
5509 generally nicer than the two earlier options for changing
5510 unit files locally: copying the files from
5511 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
5512 them there; or creating a new file in /etc/systemd/system/
5513 that incorporates the original one via ".include". Drop-in
5514 snippets into these .d/ directories can be placed in any
5515 directory systemd looks for units in, and the usual
5516 overriding semantics between /usr/lib, /etc and /run apply
5519 * Most unit file settings which take lists of items can now be
5520 reset by assigning the empty string to them. For example,
5521 normally, settings such as Environment=FOO=BAR append a new
5522 environment variable assignment to the environment block,
5523 each time they are used. By assigning Environment= the empty
5524 string the environment block can be reset to empty. This is
5525 particularly useful with the .d/*.conf drop-in snippets
5526 mentioned above, since this adds the ability to reset list
5527 settings from vendor unit files via these drop-ins.
5529 * systemctl gained a new "list-dependencies" command for
5530 listing the dependencies of a unit recursively.
5532 * Inhibitors are now honored and listed by "systemctl
5533 suspend", "systemctl poweroff" (and similar) too, not only
5534 GNOME. These commands will also list active sessions by
5537 * Resource limits (as exposed by the various control group
5538 controllers) can now be controlled dynamically at runtime
5539 for all units. More specifically, you can now use a command
5540 like "systemctl set-cgroup-attr foobar.service cpu.shares
5541 2000" to alter the CPU shares a specific service gets. These
5542 settings are stored persistently on disk, and thus allow the
5543 administrator to easily adjust the resource usage of
5544 services with a few simple commands. This dynamic resource
5545 management logic is also available to other programs via the
5546 bus. Almost any kernel cgroup attribute and controller is
5549 * systemd-vconsole-setup will now copy all font settings to
5550 all allocated VTs, where it previously applied them only to
5553 * libsystemd-login gained the new sd_session_get_tty() API
5556 * This release drops support for a few legacy or
5557 distribution-specific LSB facility names when parsing init
5558 scripts: $x-display-manager, $mail-transfer-agent,
5559 $mail-transport-agent, $mail-transfer-agent, $smtp,
5560 $null. Also, the mail-transfer-agent.target unit backing
5561 this has been removed. Distributions which want to retain
5562 compatibility with this should carry the burden for
5563 supporting this themselves and patch support for these back
5564 in, if they really need to. Also, the facilities $syslog and
5565 $local_fs are now ignored, since systemd does not support
5566 early-boot LSB init scripts anymore, and these facilities
5567 are implied anyway for normal services. syslog.target has
5570 * There are new bus calls on PID1's Manager object for
5571 cancelling jobs, and removing snapshot units. Previously,
5572 both calls were only available on the Job and Snapshot
5575 * systemd-journal-gatewayd gained SSL support.
5577 * The various "environment" files, such as /etc/locale.conf
5578 now support continuation lines with a backslash ("\") as
5579 last character in the line, similarly in style (but different)
5580 to how this is supported in shells.
5582 * For normal user processes the _SYSTEMD_USER_UNIT= field is
5583 now implicitly appended to every log entry logged. systemctl
5584 has been updated to filter by this field when operating on a
5585 user systemd instance.
5587 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
5588 CAP_AUDIT_CONTROL capabilities to the capabilities set for
5589 the container. This makes it easier to boot unmodified
5590 Fedora systems in a container, which however still requires
5591 audit=0 to be passed on the kernel command line. Auditing in
5592 kernel and userspace is unfortunately still too broken in
5593 context of containers, hence we recommend compiling it out
5594 of the kernel or using audit=0. Hopefully this will be fixed
5595 one day for good in the kernel.
5597 * nspawn gained the new --bind= and --bind-ro= parameters to
5598 bind mount specific directories from the host into the
5601 * nspawn will now mount its own devpts file system instance
5602 into the container, in order not to leak pty devices from
5603 the host into the container.
5605 * systemd will now read the firmware boot time performance
5606 information from the EFI variables, if the used boot loader
5607 supports this, and takes it into account for boot performance
5608 analysis via "systemd-analyze". This is currently supported
5609 only in conjunction with Gummiboot, but could be supported
5610 by other boot loaders too. For details see:
5612 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
5614 * A new generator has been added that automatically mounts the
5615 EFI System Partition (ESP) to /boot, if that directory
5616 exists, is empty, and no other file system has been
5617 configured to be mounted there.
5619 * logind will now send out PrepareForSleep(false) out
5620 unconditionally, after coming back from suspend. This may be
5621 used by applications as asynchronous notification for
5622 system resume events.
5624 * "systemctl unlock-sessions" has been added, that allows
5625 unlocking the screens of all user sessions at once, similar
5626 to how "systemctl lock-sessions" already locked all users
5627 sessions. This is backed by a new D-Bus call UnlockSessions().
5629 * "loginctl seat-status" will now show the master device of a
5630 seat. (i.e. the device of a seat that needs to be around for
5631 the seat to be considered available, usually the graphics
5634 * tmpfiles gained a new "X" line type, that allows
5635 configuration of files and directories (with wildcards) that
5636 shall be excluded from automatic cleanup ("aging").
5638 * udev default rules set the device node permissions now only
5639 at "add" events, and do not change them any longer with a
5640 later "change" event.
5642 * The log messages for lid events and power/sleep keypresses
5643 now carry a message ID.
5645 * We now have a substantially larger unit test suite, but this
5646 continues to be work in progress.
5648 * udevadm hwdb gained a new --root= parameter to change the
5649 root directory to operate relative to.
5651 * logind will now issue a background sync() request to the kernel
5652 early at shutdown, so that dirty buffers are flushed to disk early
5653 instead of at the last moment, in order to optimize shutdown
5656 * A new bootctl tool has been added that is an interface for
5657 certain boot loader operations. This is currently a preview
5658 and is likely to be extended into a small mechanism daemon
5659 like timedated, localed, hostnamed, and can be used by
5660 graphical UIs to enumerate available boot options, and
5661 request boot into firmware operations.
5663 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
5664 the rest of the package. It also has been updated to work
5665 correctly in initrds.
5667 * Policykit previously has been runtime optional, and is now
5668 also compile time optional via a configure switch.
5670 * systemd-analyze has been reimplemented in C. Also "systemctl
5671 dot" has moved into systemd-analyze.
5673 * "systemctl status" with no further parameters will now print
5674 the status of all active or failed units.
5676 * Operations such as "systemctl start" can now be executed
5677 with a new mode "--irreversible" which may be used to queue
5678 operations that cannot accidentally be reversed by a later
5679 job queuing. This is by default used to make shutdown
5680 requests more robust.
5682 * The Python API of systemd now gained a new module for
5683 reading journal files.
5685 * A new tool kernel-install has been added that can install
5686 kernel images according to the Boot Loader Specification:
5688 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
5690 * Boot time console output has been improved to provide
5691 animated boot time output for hanging jobs.
5693 * A new tool systemd-activate has been added which can be used
5694 to test socket activation with, directly from the command
5695 line. This should make it much easier to test and debug
5696 socket activation in daemons.
5698 * journalctl gained a new "--reverse" (or -r) option to show
5699 journal output in reverse order (i.e. newest line first).
5701 * journalctl gained a new "--pager-end" (or -e) option to jump
5702 to immediately jump to the end of the journal in the
5703 pager. This is only supported in conjunction with "less".
5705 * journalctl gained a new "--user-unit=" option, that works
5706 similarly to "--unit=" but filters for user units rather than
5709 * A number of unit files to ease adoption of systemd in
5710 initrds has been added. This moves some minimal logic from
5711 the various initrd implementations into systemd proper.
5713 * The journal files are now owned by a new group
5714 "systemd-journal", which exists specifically to allow access
5715 to the journal, and nothing else. Previously, we used the
5716 "adm" group for that, which however possibly covers more
5717 than just journal/log file access. This new group is now
5718 already used by systemd-journal-gatewayd to ensure this
5719 daemon gets access to the journal files and as little else
5720 as possible. Note that "make install" will also set FS ACLs
5721 up for /var/log/journal to give "adm" and "wheel" read
5722 access to it, in addition to "systemd-journal" which owns
5723 the journal files. We recommend that packaging scripts also
5724 add read access to "adm" + "wheel" to /var/log/journal, and
5725 all existing/future journal files. To normal users and
5726 administrators little changes, however packagers need to
5727 ensure to create the "systemd-journal" system group at
5728 package installation time.
5730 * The systemd-journal-gatewayd now runs as unprivileged user
5731 systemd-journal-gateway:systemd-journal-gateway. Packaging
5732 scripts need to create these system user/group at
5735 * timedated now exposes a new boolean property CanNTP that
5736 indicates whether a local NTP service is available or not.
5738 * systemd-detect-virt will now also detect xen PVs
5740 * The pstore file system is now mounted by default, if it is
5743 * In addition to the SELinux and IMA policies we will now also
5744 load SMACK policies at early boot.
5746 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
5747 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
5748 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
5749 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
5750 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
5751 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
5752 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
5753 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
5754 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
5755 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
5756 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
5757 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
5758 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
5759 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
5763 * Timer units now support calendar time events in addition to
5764 monotonic time events. That means you can now trigger a unit
5765 based on a calendar time specification such as "Thu,Fri
5766 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
5767 or fifth day of any month of the year 2013, given that it is
5768 a thursday or friday. This brings timer event support
5769 considerably closer to cron's capabilities. For details on
5770 the supported calendar time specification language see
5773 * udev now supports a number of different naming policies for
5774 network interfaces for predictable names, and a combination
5775 of these policies is now the default. Please see this wiki
5776 document for details:
5778 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
5780 * Auke Kok's bootchart implementation has been added to the
5781 systemd tree. It is an optional component that can graph the
5782 boot in quite some detail. It is one of the best bootchart
5783 implementations around and minimal in its code and
5786 * nss-myhostname has been integrated into the systemd source
5787 tree. nss-myhostname guarantees that the local hostname
5788 always stays resolvable via NSS. It has been a weak
5789 requirement of systemd-hostnamed since a long time, and
5790 since its code is actually trivial we decided to just
5791 include it in systemd's source tree. It can be turned off
5792 with a configure switch.
5794 * The read-ahead logic is now capable of properly detecting
5795 whether a btrfs file system is on SSD or rotating media, in
5796 order to optimize the read-ahead scheme. Previously, it was
5797 only capable of detecting this on traditional file systems
5800 * In udev, additional device properties are now read from the
5801 IAB in addition to the OUI database. Also, Bluetooth company
5802 identities are attached to the devices as well.
5804 * In service files %U may be used as specifier that is
5805 replaced by the configured user name of the service.
5807 * nspawn may now be invoked without a controlling TTY. This
5808 makes it suitable for invocation as its own service. This
5809 may be used to set up a simple containerized server system
5810 using only core OS tools.
5812 * systemd and nspawn can now accept socket file descriptors
5813 when they are started for socket activation. This enables
5814 implementation of socket activated nspawn
5815 containers. i.e. think about autospawning an entire OS image
5816 when the first SSH or HTTP connection is received. We expect
5817 that similar functionality will also be added to libvirt-lxc
5820 * journalctl will now suppress ANSI color codes when
5821 presenting log data.
5823 * systemctl will no longer show control group information for
5824 a unit if the control group is empty anyway.
5826 * logind can now automatically suspend/hibernate/shutdown the
5829 * /etc/machine-info and hostnamed now also expose the chassis
5830 type of the system. This can be used to determine whether
5831 the local system is a laptop, desktop, handset or
5832 tablet. This information may either be configured by the
5833 user/vendor or is automatically determined from ACPI and DMI
5834 information if possible.
5836 * A number of PolicyKit actions are now bound together with
5837 "imply" rules. This should simplify creating UIs because
5838 many actions will now authenticate similar ones as well.
5840 * Unit files learnt a new condition ConditionACPower= which
5841 may be used to conditionalize a unit depending on whether an
5842 AC power source is connected or not, of whether the system
5843 is running on battery power.
5845 * systemctl gained a new "is-failed" verb that may be used in
5846 shell scripts and suchlike to check whether a specific unit
5847 is in the "failed" state.
5849 * The EnvironmentFile= setting in unit files now supports file
5850 globbing, and can hence be used to easily read a number of
5851 environment files at once.
5853 * systemd will no longer detect and recognize specific
5854 distributions. All distribution-specific #ifdeffery has been
5855 removed, systemd is now fully generic and
5856 distribution-agnostic. Effectively, not too much is lost as
5857 a lot of the code is still accessible via explicit configure
5858 switches. However, support for some distribution specific
5859 legacy configuration file formats has been dropped. We
5860 recommend distributions to simply adopt the configuration
5861 files everybody else uses now and convert the old
5862 configuration from packaging scripts. Most distributions
5863 already did that. If that's not possible or desirable,
5864 distributions are welcome to forward port the specific
5865 pieces of code locally from the git history.
5867 * When logging a message about a unit systemd will now always
5868 log the unit name in the message meta data.
5870 * localectl will now also discover system locale data that is
5871 not stored in locale archives, but directly unpacked.
5873 * logind will no longer unconditionally use framebuffer
5874 devices as seat masters, i.e. as devices that are required
5875 to be existing before a seat is considered preset. Instead,
5876 it will now look for all devices that are tagged as
5877 "seat-master" in udev. By default, framebuffer devices will
5878 be marked as such, but depending on local systems, other
5879 devices might be marked as well. This may be used to
5880 integrate graphics cards using closed source drivers (such
5881 as NVidia ones) more nicely into logind. Note however, that
5882 we recommend using the open source NVidia drivers instead,
5883 and no udev rules for the closed-source drivers will be
5884 shipped from us upstream.
5886 Contributions from: Adam Williamson, Alessandro Crismani, Auke
5887 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
5888 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
5889 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
5890 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
5891 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
5892 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
5893 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
5894 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
5895 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
5896 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
5897 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
5902 * udev gained support for loading additional device properties
5903 from an indexed database that is keyed by vendor/product IDs
5904 and similar device identifiers. For the beginning this
5905 "hwdb" is populated with data from the well-known PCI and
5906 USB database, but also includes PNP, ACPI and OID data. In
5907 the longer run this indexed database shall grow into
5908 becoming the one central database for non-essential
5909 userspace device metadata. Previously, data from the PCI/USB
5910 database was only attached to select devices, since the
5911 lookup was a relatively expensive operation due to O(n) time
5912 complexity (with n being the number of entries in the
5913 database). Since this is now O(1), we decided to add in this
5914 data for all devices where this is available, by
5915 default. Note that the indexed database needs to be rebuilt
5916 when new data files are installed. To achieve this you need
5917 to update your packaging scripts to invoke "udevadm hwdb
5918 --update" after installation of hwdb data files. For
5919 RPM-based distributions we introduced the new
5920 %udev_hwdb_update macro for this purpose.
5922 * The Journal gained support for the "Message Catalog", an
5923 indexed database to link up additional information with
5924 journal entries. For further details please check:
5926 http://www.freedesktop.org/wiki/Software/systemd/catalog
5928 The indexed message catalog database also needs to be
5929 rebuilt after installation of message catalog files. Use
5930 "journalctl --update-catalog" for this. For RPM-based
5931 distributions we introduced the %journal_catalog_update
5932 macro for this purpose.
5934 * The Python Journal bindings gained support for the standard
5935 Python logging framework.
5937 * The Journal API gained new functions for checking whether
5938 the underlying file system of a journal file is capable of
5939 properly reporting file change notifications, or whether
5940 applications that want to reflect journal changes "live"
5941 need to recheck journal files continuously in appropriate
5944 * It is now possible to set the "age" field for tmpfiles
5945 entries to 0, indicating that files matching this entry
5946 shall always be removed when the directories are cleaned up.
5948 * coredumpctl gained a new "gdb" verb which invokes gdb
5949 right-away on the selected coredump.
5951 * There's now support for "hybrid sleep" on kernels that
5952 support this, in addition to "suspend" and "hibernate". Use
5953 "systemctl hybrid-sleep" to make use of this.
5955 * logind's HandleSuspendKey= setting (and related settings)
5956 now gained support for a new "lock" setting to simply
5957 request the screen lock on all local sessions, instead of
5958 actually executing a suspend or hibernation.
5960 * systemd will now mount the EFI variables file system by
5963 * Socket units now gained support for configuration of the
5964 SMACK security label.
5966 * timedatectl will now output the time of the last and next
5967 daylight saving change.
5969 * We dropped support for various legacy and distro-specific
5970 concepts, such as insserv, early-boot SysV services
5971 (i.e. those for non-standard runlevels such as 'b' or 'S')
5972 or ArchLinux /etc/rc.conf support. We recommend the
5973 distributions who still need support this to either continue
5974 to maintain the necessary patches downstream, or find a
5975 different solution. (Talk to us if you have questions!)
5977 * Various systemd components will now bypass PolicyKit checks
5978 for root and otherwise handle properly if PolicyKit is not
5979 found to be around. This should fix most issues for
5980 PolicyKit-less systems. Quite frankly this should have been
5981 this way since day one. It is absolutely our intention to
5982 make systemd work fine on PolicyKit-less systems, and we
5983 consider it a bug if something does not work as it should if
5984 PolicyKit is not around.
5986 * For embedded systems it is now possible to build udev and
5987 systemd without blkid and/or kmod support.
5989 * "systemctl switch-root" is now capable of switching root
5990 more than once. I.e. in addition to transitions from the
5991 initrd to the host OS it is now possible to transition to
5992 further OS images from the host. This is useful to implement
5993 offline updating tools.
5995 * Various other additions have been made to the RPM macros
5996 shipped with systemd. Use %udev_rules_update() after
5997 installing new udev rules files. %_udevhwdbdir,
5998 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
5999 %_sysctldir are now available which resolve to the right
6000 directories for packages to place various data files in.
6002 * journalctl gained the new --full switch (in addition to
6003 --all, to disable ellipsation for long messages.
6005 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
6006 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
6007 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
6008 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
6009 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
6010 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
6011 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
6012 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
6013 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
6017 * journalctl gained new --since= and --until= switches to
6018 filter by time. It also now supports nice filtering for
6019 units via --unit=/-u.
6021 * Type=oneshot services may use ExecReload= and do the
6024 * The journal daemon now supports time-based rotation and
6025 vacuuming, in addition to the usual disk-space based
6028 * The journal will now index the available field values for
6029 each field name. This enables clients to show pretty drop
6030 downs of available match values when filtering. The bash
6031 completion of journalctl has been updated
6032 accordingly. journalctl gained a new switch -F to list all
6033 values a certain field takes in the journal database.
6035 * More service events are now written as structured messages
6036 to the journal, and made recognizable via message IDs.
6038 * The timedated, localed and hostnamed mini-services which
6039 previously only provided support for changing time, locale
6040 and hostname settings from graphical DEs such as GNOME now
6041 also have a minimal (but very useful) text-based client
6042 utility each. This is probably the nicest way to changing
6043 these settings from the command line now, especially since
6044 it lists available options and is fully integrated with bash
6047 * There's now a new tool "systemd-coredumpctl" to list and
6048 extract coredumps from the journal.
6050 * We now install a README each in /var/log/ and
6051 /etc/rc.d/init.d explaining where the system logs and init
6052 scripts went. This hopefully should help folks who go to
6053 that dirs and look into the otherwise now empty void and
6054 scratch their heads.
6056 * When user-services are invoked (by systemd --user) the
6057 $MANAGERPID env var is set to the PID of systemd.
6059 * SIGRTMIN+24 when sent to a --user instance will now result
6060 in immediate termination of systemd.
6062 * gatewayd received numerous feature additions such as a
6063 "follow" mode, for live syncing and filtering.
6065 * browse.html now allows filtering and showing detailed
6066 information on specific entries. Keyboard navigation and
6067 mouse screen support has been added.
6069 * gatewayd/journalctl now supports HTML5/JSON
6070 Server-Sent-Events as output.
6072 * The SysV init script compatibility logic will now
6073 heuristically determine whether a script supports the
6074 "reload" verb, and only then make this available as
6077 * "systemctl status --follow" has been removed, use "journalctl
6080 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
6081 have been removed since they are hardly useful to be
6084 * And I'd like to take the opportunity to specifically mention
6085 Zbigniew for his great contributions. Zbigniew, you rock!
6087 Contributions from: Andrew Eikum, Christian Hesse, Colin
6088 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
6089 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
6090 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
6091 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
6092 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
6093 Jędrzejewski-Szmek, Сковорода Никита Андреевич
6097 * If /etc/vconsole.conf is non-existent or empty we will no
6098 longer load any console font or key map at boot by
6099 default. Instead the kernel defaults will be left
6100 intact. This is definitely the right thing to do, as no
6101 configuration should mean no configuration, and hard-coding
6102 font names that are different on all archs is probably a bad
6103 idea. Also, the kernel default key map and font should be
6104 good enough for most cases anyway, and mostly identical to
6105 the userspace fonts/key maps we previously overloaded them
6106 with. If distributions want to continue to default to a
6107 non-kernel font or key map they should ship a default
6108 /etc/vconsole.conf with the appropriate contents.
6110 Contributions from: Colin Walters, Daniel J Walsh, Dave
6111 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
6112 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
6116 * journalctl gained a new --cursor= switch to show entries
6117 starting from the specified location in the journal.
6119 * We now enforce a size limit on journal entry fields exported
6120 with "-o json" in journalctl. Fields larger than 4K will be
6121 assigned null. This can be turned off with --all.
6123 * An (optional) journal gateway daemon is now available as
6124 "systemd-journal-gatewayd.service". This service provides
6125 access to the journal via HTTP and JSON. This functionality
6126 will be used to implement live log synchronization in both
6127 pull and push modes, but has various other users too, such
6128 as easy log access for debugging of embedded devices. Right
6129 now it is already useful to retrieve the journal via HTTP:
6131 # systemctl start systemd-journal-gatewayd.service
6132 # wget http://localhost:19531/entries
6134 This will download the journal contents in a
6135 /var/log/messages compatible format. The same as JSON:
6137 # curl -H"Accept: application/json" http://localhost:19531/entries
6139 This service is also accessible via a web browser where a
6140 single static HTML5 app is served that uses the JSON logic
6141 to enable the user to do some basic browsing of the
6142 journal. This will be extended later on. Here's an example
6143 screenshot of this app in its current state:
6145 http://0pointer.de/public/journal-gatewayd
6147 Contributions from: Kay Sievers, Lennart Poettering, Robert
6148 Milasan, Tom Gundersen
6152 * The bash completion logic is now available for journalctl
6155 * We do not mount the "cpuset" controller anymore together with
6156 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
6157 started if no parameters are assigned to it. "cpuset" hence
6158 broke code that assumed it could create "cpu" groups and
6161 * journalctl -f will now subscribe to terminal size changes,
6162 and line break accordingly.
6164 Contributions from: Dave Reisner, Kay Sievers, Lennart
6165 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
6169 * nspawn will now create a symlink /etc/localtime in the
6170 container environment, copying the host's timezone
6171 setting. Previously this has been done via a bind mount, but
6172 since symlinks cannot be bind mounted this has now been
6173 changed to create/update the appropriate symlink.
6175 * journalctl -n's line number argument is now optional, and
6176 will default to 10 if omitted.
6178 * journald will now log the maximum size the journal files may
6179 take up on disk. This is particularly useful if the default
6180 built-in logic of determining this parameter from the file
6181 system size is used. Use "systemctl status
6182 systemd-journald.service" to see this information.
6184 * The multi-seat X wrapper tool has been stripped down. As X
6185 is now capable of enumerating graphics devices via udev in a
6186 seat-aware way the wrapper is not strictly necessary
6187 anymore. A stripped down temporary stop-gap is still shipped
6188 until the upstream display managers have been updated to
6189 fully support the new X logic. Expect this wrapper to be
6190 removed entirely in one of the next releases.
6192 * HandleSleepKey= in logind.conf has been split up into
6193 HandleSuspendKey= and HandleHibernateKey=. The old setting
6194 is not available anymore. X11 and the kernel are
6195 distinguishing between these keys and we should too. This
6196 also means the inhibition lock for these keys has been split
6199 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
6200 Poettering, Lukas Nykryn, Václav Pavlín
6204 * Whenever a unit changes state we will now log this to the
6205 journal and show along the unit's own log output in
6208 * ConditionPathIsMountPoint= can now properly detect bind
6209 mount points too. (Previously, a bind mount of one file
6210 system to another place in the same file system could not be
6211 detected as mount, since they shared struct stat's st_dev
6214 * We will now mount the cgroup controllers cpu, cpuacct,
6215 cpuset and the controllers net_cls, net_prio together by
6218 * nspawn containers will now have a virtualized boot
6219 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
6220 over with a randomized ID at container initialization). This
6221 has the effect of making "journalctl -b" do the right thing
6224 * The JSON output journal serialization has been updated not
6225 to generate "endless" list objects anymore, but rather one
6226 JSON object per line. This is more in line how most JSON
6227 parsers expect JSON objects. The new output mode
6228 "json-pretty" has been added to provide similar output, but
6229 neatly aligned for readability by humans.
6231 * We dropped all explicit sync() invocations in the shutdown
6232 code. The kernel does this implicitly anyway in the kernel
6233 reboot() syscall. halt(8)'s -n option is now a compatibility
6236 * We now support virtualized reboot() in containers, as
6237 supported by newer kernels. We will fall back to exit() if
6238 CAP_SYS_REBOOT is not available to the container. Also,
6239 nspawn makes use of this now and will actually reboot the
6240 container if the containerized OS asks for that.
6242 * journalctl will only show local log output by default
6243 now. Use --merge (-m) to show remote log output, too.
6245 * libsystemd-journal gained the new sd_journal_get_usage()
6246 call to determine the current disk usage of all journal
6247 files. This is exposed in the new "journalctl --disk-usage"
6250 * journald gained a new configuration setting SplitMode= in
6251 journald.conf which may be used to control how user journals
6252 are split off. See journald.conf(5) for details.
6254 * A new condition type ConditionFileNotEmpty= has been added.
6256 * tmpfiles' "w" lines now support file globbing, to write
6257 multiple files at once.
6259 * We added Python bindings for the journal submission
6260 APIs. More Python APIs for a number of selected APIs will
6261 likely follow. Note that we intend to add native bindings
6262 only for the Python language, as we consider it common
6263 enough to deserve bindings shipped within systemd. There are
6264 various projects outside of systemd that provide bindings
6265 for languages such as PHP or Lua.
6267 * Many conditions will now resolve specifiers such as %i. In
6268 addition, PathChanged= and related directives of .path units
6269 now support specifiers as well.
6271 * There's now a new RPM macro definition for the system preset
6274 * journald will now warn if it ca not forward a message to the
6275 syslog daemon because its socket is full.
6277 * timedated will no longer write or process /etc/timezone,
6278 except on Debian. As we do not support late mounted /usr
6279 anymore /etc/localtime always being a symlink is now safe,
6280 and hence the information in /etc/timezone is not necessary
6283 * logind will now always reserve one VT for a text getty (VT6
6284 by default). Previously if more than 6 X sessions where
6285 started they took up all the VTs with auto-spawned gettys,
6286 so that no text gettys were available anymore.
6288 * udev will now automatically inform the btrfs kernel logic
6289 about btrfs RAID components showing up. This should make
6290 simple hotplug based btrfs RAID assembly work.
6292 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
6293 (but not for its children which will stay at the kernel
6294 default). This should allow setups with a lot more listening
6297 * systemd will now always pass the configured timezone to the
6298 kernel at boot. timedated will do the same when the timezone
6301 * logind's inhibition logic has been updated. By default,
6302 logind will now handle the lid switch, the power and sleep
6303 keys all the time, even in graphical sessions. If DEs want
6304 to handle these events on their own they should take the new
6305 handle-power-key, handle-sleep-key and handle-lid-switch
6306 inhibitors during their runtime. A simple way to achieve
6307 that is to invoke the DE wrapped in an invocation of:
6309 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
6311 * Access to unit operations is now checked via SELinux taking
6312 the unit file label and client process label into account.
6314 * systemd will now notify the administrator in the journal
6315 when he over-mounts a non-empty directory.
6317 * There are new specifiers that are resolved in unit files,
6318 for the host name (%H), the machine ID (%m) and the boot ID
6321 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
6322 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
6323 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
6324 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
6325 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
6326 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
6327 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
6331 * Support for reading structured kernel messages from
6332 /dev/kmsg has now been added and is enabled by default.
6334 * Support for reading kernel messages from /proc/kmsg has now
6335 been removed. If you want kernel messages in the journal
6336 make sure to run a recent kernel (>= 3.5) that supports
6337 reading structured messages from /dev/kmsg (see
6338 above). /proc/kmsg is now exclusive property of classic
6339 syslog daemons again.
6341 * The libudev API gained the new
6342 udev_device_new_from_device_id() call.
6344 * The logic for file system namespace (ReadOnlyDirectory=,
6345 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
6346 require pivot_root() anymore. This means fewer temporary
6347 directories are created below /tmp for this feature.
6349 * nspawn containers will now see and receive all submounts
6350 made on the host OS below the root file system of the
6353 * Forward Secure Sealing is now supported for Journal files,
6354 which provide cryptographical sealing of journal files so
6355 that attackers cannot alter log history anymore without this
6356 being detectable. Lennart will soon post a blog story about
6357 this explaining it in more detail.
6359 * There are two new service settings RestartPreventExitStatus=
6360 and SuccessExitStatus= which allow configuration of exit
6361 status (exit code or signal) which will be excepted from the
6362 restart logic, resp. consider successful.
6364 * journalctl gained the new --verify switch that can be used
6365 to check the integrity of the structure of journal files and
6366 (if Forward Secure Sealing is enabled) the contents of
6369 * nspawn containers will now be run with /dev/stdin, /dev/fd/
6370 and similar symlinks pre-created. This makes running shells
6371 as container init process a lot more fun.
6373 * The fstab support can now handle PARTUUID= and PARTLABEL=
6376 * A new ConditionHost= condition has been added to match
6377 against the hostname (with globs) and machine ID. This is
6378 useful for clusters where a single OS image is used to
6379 provision a large number of hosts which shall run slightly
6380 different sets of services.
6382 * Services which hit the restart limit will now be placed in a
6385 Contributions from: Bertram Poettering, Dave Reisner, Huang
6386 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
6387 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
6391 * When running in --user mode systemd will now become a
6392 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
6393 tree a lot more organized.
6395 * A new PartOf= unit dependency type has been introduced that
6396 may be used to group services in a natural way.
6398 * "systemctl enable" may now be used to enable instances of
6401 * journalctl now prints error log levels in red, and
6402 warning/notice log levels in bright white. It also supports
6403 filtering by log level now.
6405 * cgtop gained a new -n switch (similar to top), to configure
6406 the maximum number of iterations to run for. It also gained
6407 -b, to run in batch mode (accepting no input).
6409 * The suffix ".service" may now be omitted on most systemctl
6410 command lines involving service unit names.
6412 * There's a new bus call in logind to lock all sessions, as
6413 well as a loginctl verb for it "lock-sessions".
6415 * libsystemd-logind.so gained a new call sd_journal_perror()
6416 that works similar to libc perror() but logs to the journal
6417 and encodes structured information about the error number.
6419 * /etc/crypttab entries now understand the new keyfile-size=
6422 * shutdown(8) now can send a (configurable) wall message when
6423 a shutdown is cancelled.
6425 * The mount propagation mode for the root file system will now
6426 default to "shared", which is useful to make containers work
6427 nicely out-of-the-box so that they receive new mounts from
6428 the host. This can be undone locally by running "mount
6429 --make-rprivate /" if needed.
6431 * The prefdm.service file has been removed. Distributions
6432 should maintain this unit downstream if they intend to keep
6433 it around. However, we recommend writing normal unit files
6434 for display managers instead.
6436 * Since systemd is a crucial part of the OS we will now
6437 default to a number of compiler switches that improve
6438 security (hardening) such as read-only relocations, stack
6439 protection, and suchlike.
6441 * The TimeoutSec= setting for services is now split into
6442 TimeoutStartSec= and TimeoutStopSec= to allow configuration
6443 of individual time outs for the start and the stop phase of
6446 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
6447 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
6448 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
6449 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
6450 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
6451 Gundersen, Zbigniew Jędrzejewski-Szmek
6455 * The journal and id128 C APIs are now fully documented as man
6458 * Extra safety checks have been added when transitioning from
6459 the initial RAM disk to the main system to avoid accidental
6462 * /etc/crypttab entries now understand the new keyfile-offset=
6465 * systemctl -t can now be used to filter by unit load state.
6467 * The journal C API gained the new sd_journal_wait() call to
6468 make writing synchronous journal clients easier.
6470 * journalctl gained the new -D switch to show journals from a
6473 * journalctl now displays a special marker between log
6474 messages of two different boots.
6476 * The journal is now explicitly flushed to /var via a service
6477 systemd-journal-flush.service, rather than implicitly simply
6478 by seeing /var/log/journal to be writable.
6480 * journalctl (and the journal C APIs) can now match for much
6481 more complex expressions, with alternatives and
6484 * When transitioning from the initial RAM disk to the main
6485 system we will now kill all processes in a killing spree to
6486 ensure no processes stay around by accident.
6488 * Three new specifiers may be used in unit files: %u, %h, %s
6489 resolve to the user name, user home directory resp. user
6490 shell. This is useful for running systemd user instances.
6492 * We now automatically rotate journal files if their data
6493 object hash table gets a fill level > 75%. We also size the
6494 hash table based on the configured maximum file size. This
6495 together should lower hash collisions drastically and thus
6496 speed things up a bit.
6498 * journalctl gained the new "--header" switch to introspect
6499 header data of journal files.
6501 * A new setting SystemCallFilters= has been added to services
6502 which may be used to apply blacklists or whitelists to
6503 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
6505 * nspawn gained a new --link-journal= switch (and quicker: -j)
6506 to link the container journal with the host. This makes it
6507 very easy to centralize log viewing on the host for all
6508 guests while still keeping the journal files separated.
6510 * Many bugfixes and optimizations
6512 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
6513 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
6514 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
6519 * Several tools now understand kernel command line arguments,
6520 which are only read when run in an initial RAM disk. They
6521 usually follow closely their normal counterparts, but are
6524 * There's a new tool to analyze the readahead files that are
6525 automatically generated at boot. Use:
6527 /usr/lib/systemd/systemd-readahead analyze /.readahead
6529 * We now provide an early debug shell on tty9 if this enabled. Use:
6531 systemctl enable debug-shell.service
6533 * All plymouth related units have been moved into the Plymouth
6534 package. Please make sure to upgrade your Plymouth version
6537 * systemd-tmpfiles now supports getting passed the basename of
6538 a configuration file only, in which case it will look for it
6539 in all appropriate directories automatically.
6541 * udevadm info now takes a /dev or /sys path as argument, and
6542 does the right thing. Example:
6544 udevadm info /dev/sda
6545 udevadm info /sys/class/block/sda
6547 * systemctl now prints a warning if a unit is stopped but a
6548 unit that might trigger it continues to run. Example: a
6549 service is stopped but the socket that activates it is left
6552 * "systemctl status" will now mention if the log output was
6553 shortened due to rotation since a service has been started.
6555 * The journal API now exposes functions to determine the
6556 "cutoff" times due to rotation.
6558 * journald now understands SIGUSR1 and SIGUSR2 for triggering
6559 immediately flushing of runtime logs to /var if possible,
6560 resp. for triggering immediate rotation of the journal
6563 * It is now considered an error if a service is attempted to
6564 be stopped that is not loaded.
6566 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
6568 * systemd-analyze now supports Python 3
6570 * tmpfiles now supports cleaning up directories via aging
6571 where the first level dirs are always kept around but
6572 directories beneath it automatically aged. This is enabled
6573 by prefixing the age field with '~'.
6575 * Seat objects now expose CanGraphical, CanTTY properties
6576 which is required to deal with very fast bootups where the
6577 display manager might be running before the graphics drivers
6578 completed initialization.
6580 * Seat objects now expose a State property.
6582 * We now include RPM macros for service enabling/disabling
6583 based on the preset logic. We recommend RPM based
6584 distributions to make use of these macros if possible. This
6585 makes it simpler to reuse RPM spec files across
6588 * We now make sure that the collected systemd unit name is
6589 always valid when services log to the journal via
6592 * There's a new man page kernel-command-line(7) detailing all
6593 command line options we understand.
6595 * The fstab generator may now be disabled at boot by passing
6596 fstab=0 on the kernel command line.
6598 * A new kernel command line option modules-load= is now understood
6599 to load a specific kernel module statically, early at boot.
6601 * Unit names specified on the systemctl command line are now
6602 automatically escaped as needed. Also, if file system or
6603 device paths are specified they are automatically turned
6604 into the appropriate mount or device unit names. Example:
6606 systemctl status /home
6607 systemctl status /dev/sda
6609 * The SysVConsole= configuration option has been removed from
6610 system.conf parsing.
6612 * The SysV search path is no longer exported on the D-Bus
6615 * The Names= option has been removed from unit file parsing.
6617 * There's a new man page bootup(7) detailing the boot process.
6619 * Every unit and every generator we ship with systemd now
6620 comes with full documentation. The self-explanatory boot is
6623 * A couple of services gained "systemd-" prefixes in their
6624 name if they wrap systemd code, rather than only external
6625 code. Among them fsck@.service which is now
6626 systemd-fsck@.service.
6628 * The HaveWatchdog property has been removed from the D-Bus
6631 * systemd.confirm_spawn= on the kernel command line should now
6634 * There's a new man page crypttab(5) which details all options
6635 we actually understand.
6637 * systemd-nspawn gained a new --capability= switch to pass
6638 additional capabilities to the container.
6640 * timedated will now read known NTP implementation unit names
6641 from /usr/lib/systemd/ntp-units.d/*.list,
6642 systemd-timedated-ntp.target has been removed.
6644 * journalctl gained a new switch "-b" that lists log data of
6645 the current boot only.
6647 * The notify socket is in the abstract namespace again, in
6648 order to support daemons which chroot() at start-up.
6650 * There is a new Storage= configuration option for journald
6651 which allows configuration of where log data should go. This
6652 also provides a way to disable journal logging entirely, so
6653 that data collected is only forwarded to the console, the
6654 kernel log buffer or another syslog implementation.
6656 * Many bugfixes and optimizations
6658 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
6659 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
6660 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
6661 Shawn Landden, Tom Gundersen
6665 * "systemctl help <unit>" now shows the man page if one is
6668 * Several new man pages have been added.
6670 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
6671 MaxLevelConsole= can now be specified in
6672 journald.conf. These options allow reducing the amount of
6673 data stored on disk or forwarded by the log level.
6675 * TimerSlackNSec= can now be specified in system.conf for
6676 PID1. This allows system-wide power savings.
6678 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
6679 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
6684 * logind is now capable of (optionally) handling power and
6685 sleep keys as well as the lid switch.
6687 * journalctl now understands the syntax "journalctl
6688 /usr/bin/avahi-daemon" to get all log output of a specific
6691 * CapabilityBoundingSet= in system.conf now also influences
6692 the capability bound set of usermode helpers of the kernel.
6694 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
6695 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
6696 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
6697 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
6701 * Note that we skipped 139 releases here in order to set the
6702 new version to something that is greater than both udev's
6703 and systemd's most recent version number.
6705 * udev: all udev sources are merged into the systemd source tree now.
6706 All future udev development will happen in the systemd tree. It
6707 is still fully supported to use the udev daemon and tools without
6708 systemd running, like in initramfs or other init systems. Building
6709 udev though, will require the *build* of the systemd tree, but
6710 udev can be properly *run* without systemd.
6712 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
6713 should be used to create dead device nodes as workarounds for broken
6716 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
6717 no longer supported. udev_monitor_new_from_netlink() needs to be
6718 used to subscribe to events.
6720 * udev: when udevd is started by systemd, processes which are left
6721 behind by forking them off of udev rules, are unconditionally cleaned
6722 up and killed now after the event handling has finished. Services or
6723 daemons must be started as systemd services. Services can be
6724 pulled-in by udev to get started, but they can no longer be directly
6725 forked by udev rules.
6727 * udev: the daemon binary is called systemd-udevd now and installed
6728 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
6729 to adapt to that, create symlink, or rename the binary after building
6732 * libudev no longer provides these symbols:
6733 udev_monitor_from_socket()
6734 udev_queue_get_failed_list_entry()
6735 udev_get_{dev,sys,run}_path()
6736 The versions number was bumped and symbol versioning introduced.
6738 * systemd-loginctl and systemd-journalctl have been renamed
6739 to loginctl and journalctl to match systemctl.
6741 * The config files: /etc/systemd/systemd-logind.conf and
6742 /etc/systemd/systemd-journald.conf have been renamed to
6743 logind.conf and journald.conf. Package updates should rename
6744 the files to the new names on upgrade.
6746 * For almost all files the license is now LGPL2.1+, changed
6747 from the previous GPL2.0+. Exceptions are some minor stuff
6748 of udev (which will be changed to LGPL2.1 eventually, too),
6749 and the MIT licensed sd-daemon.[ch] library that is suitable
6750 to be used as drop-in files.
6752 * systemd and logind now handle system sleep states, in
6753 particular suspending and hibernating.
6755 * logind now implements a sleep/shutdown/idle inhibiting logic
6756 suitable for a variety of uses. Soonishly Lennart will blog
6757 about this in more detail.
6759 * var-run.mount and var-lock.mount are no longer provided
6760 (which previously bind mounted these directories to their new
6761 places). Distributions which have not converted these
6762 directories to symlinks should consider stealing these files
6763 from git history and add them downstream.
6765 * We introduced the Documentation= field for units and added
6766 this to all our shipped units. This is useful to make it
6767 easier to explore the boot and the purpose of the various
6770 * All smaller setup units (such as
6771 systemd-vconsole-setup.service) now detect properly if they
6772 are run in a container and are skipped when
6773 appropriate. This guarantees an entirely noise-free boot in
6774 Linux container environments such as systemd-nspawn.
6776 * A framework for implementing offline system updates is now
6777 integrated, for details see:
6778 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
6780 * A new service type Type=idle is available now which helps us
6781 avoiding ugly interleaving of getty output and boot status
6784 * There's now a system-wide CapabilityBoundingSet= option to
6785 globally reduce the set of capabilities for the
6786 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
6787 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
6788 even CAP_NET_ADMIN system-wide for secure systems.
6790 * There are now system-wide DefaultLimitXXX= options to
6791 globally change the defaults of the various resource limits
6792 for all units started by PID 1.
6794 * Harald Hoyer's systemd test suite has been integrated into
6795 systemd which allows easy testing of systemd builds in qemu
6796 and nspawn. (This is really awesome! Ask us for details!)
6798 * The fstab parser is now implemented as generator, not inside
6801 * systemctl will now warn you if .mount units generated from
6802 /etc/fstab are out of date due to changes in fstab that
6803 have not been read by systemd yet.
6805 * systemd is now suitable for usage in initrds. Dracut has
6806 already been updated to make use of this. With this in place
6807 initrds get a slight bit faster but primarily are much
6808 easier to introspect and debug since "systemctl status" in
6809 the host system can be used to introspect initrd services,
6810 and the journal from the initrd is kept around too.
6812 * systemd-delta has been added, a tool to explore differences
6813 between user/admin configuration and vendor defaults.
6815 * PrivateTmp= now affects both /tmp and /var/tmp.
6817 * Boot time status messages are now much prettier and feature
6818 proper english language. Booting up systemd has never been
6821 * Read-ahead pack files now include the inode number of all
6822 files to pre-cache. When the inode changes the pre-caching
6823 is not attempted. This should be nicer to deal with updated
6824 packages which might result in changes of read-ahead
6827 * We now temporaritly lower the kernel's read_ahead_kb variable
6828 when collecting read-ahead data to ensure the kernel's
6829 built-in read-ahead does not add noise to our measurements
6830 of necessary blocks to pre-cache.
6832 * There's now RequiresMountsFor= to add automatic dependencies
6833 for all mounts necessary for a specific file system path.
6835 * MountAuto= and SwapAuto= have been removed from
6836 system.conf. Mounting file systems at boot has to take place
6839 * nspawn now learned a new switch --uuid= to set the machine
6840 ID on the command line.
6842 * nspawn now learned the -b switch to automatically search
6845 * vt102 is now the default TERM for serial TTYs, upgraded from
6848 * systemd-logind now works on VT-less systems.
6850 * The build tree has been reorganized. The individual
6851 components now have directories of their own.
6853 * A new condition type ConditionPathIsReadWrite= is now available.
6855 * nspawn learned the new -C switch to create cgroups for the
6856 container in other hierarchies.
6858 * We now have support for hardware watchdogs, configurable in
6861 * The scheduled shutdown logic now has a public API.
6863 * We now mount /tmp as tmpfs by default, but this can be
6864 masked and /etc/fstab can override it.
6866 * Since udisks does not make use of /media anymore we are not
6867 mounting a tmpfs on it anymore.
6869 * journalctl gained a new --local switch to only interleave
6870 locally generated journal files.
6872 * We can now load the IMA policy at boot automatically.
6874 * The GTK tools have been split off into a systemd-ui.
6876 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
6877 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
6878 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
6879 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
6880 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
6881 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
6882 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
6883 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
6884 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
6889 * This is mostly a bugfix release
6891 * Support optional initialization of the machine ID from the
6892 KVM or container configured UUID.
6894 * Support immediate reboots with "systemctl reboot -ff"
6896 * Show /etc/os-release data in systemd-analyze output
6898 * Many bugfixes for the journal, including endianness fixes and
6899 ensuring that disk space enforcement works
6901 * sd-login.h is C++ compatible again
6903 * Extend the /etc/os-release format on request of the Debian
6906 * We now refuse non-UTF8 strings used in various configuration
6907 and unit files. This is done to ensure we do not pass invalid
6908 data over D-Bus or expose it elsewhere.
6910 * Register Mimo USB Screens as suitable for automatic seat
6913 * Read SELinux client context from journal clients in a race
6916 * Reorder configuration file lookup order. /etc now always
6917 overrides /run in order to allow the administrator to always
6918 and unconditionally override vendor-supplied or
6919 automatically generated data.
6921 * The various user visible bits of the journal now have man
6922 pages. We still lack man pages for the journal API calls
6925 * We now ship all man pages in HTML format again in the
6928 Contributions from: Dave Reisner, Dirk Eibach, Frederic
6929 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
6930 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
6935 * This is mostly a bugfix release
6937 * systems lacking /etc/os-release are no longer supported.
6939 * Various functionality updates to libsystemd-login.so
6941 * Track class of PAM logins to distinguish greeters from
6944 Contributions from: Kay Sievers, Lennart Poettering, Michael
6949 * This is an important bugfix release for v41.
6951 * Building man pages is now optional which should be useful
6952 for those building systemd from git but unwilling to install
6955 * Watchdog support for supervising services is now usable. In
6956 a future release support for hardware watchdogs
6957 (i.e. /dev/watchdog) will be added building on this.
6959 * Service start rate limiting is now configurable and can be
6960 turned off per service. When a start rate limit is hit a
6961 reboot can automatically be triggered.
6963 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
6965 Contributions from: Benjamin Franzke, Bill Nottingham,
6966 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
6967 Schmidt, Michał Górny, Piotr Drąg
6971 * The systemd binary is installed /usr/lib/systemd/systemd now;
6972 An existing /sbin/init symlink needs to be adapted with the
6975 * The code that loads kernel modules has been ported to invoke
6976 libkmod directly, instead of modprobe. This means we do not
6977 support systems with module-init-tools anymore.
6979 * Watchdog support is now already useful, but still not
6982 * A new kernel command line option systemd.setenv= is
6983 understood to set system wide environment variables
6984 dynamically at boot.
6986 * We now limit the set of capabilities of systemd-journald.
6988 * We now set SIGPIPE to ignore by default, since it only is
6989 useful in shell pipelines, and has little use in general
6990 code. This can be disabled with IgnoreSIPIPE=no in unit
6993 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
6994 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
6999 * This is mostly a bugfix release
7001 * We now expose the reason why a service failed in the
7002 "Result" D-Bus property.
7004 * Rudimentary service watchdog support (will be completed over
7005 the next few releases.)
7007 * When systemd forks off in order execute some service we will
7008 now immediately changes its argv[0] to reflect which process
7009 it will execute. This is useful to minimize the time window
7010 with a generic argv[0], which makes bootcharts more useful
7012 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
7013 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
7014 Mike Kazantsev, Ray Strode
7018 * This is mostly a test release, but incorporates many
7021 * New systemd-cgtop tool to show control groups by their
7024 * Linking against libacl for ACLs is optional again. If
7025 disabled, support tracking device access for active logins
7026 goes becomes unavailable, and so does access to the user
7027 journals by the respective users.
7029 * If a group "adm" exists, journal files are automatically
7030 owned by them, thus allow members of this group full access
7031 to the system journal as well as all user journals.
7033 * The journal now stores the SELinux context of the logging
7034 client for all entries.
7036 * Add C++ inclusion guards to all public headers
7038 * New output mode "cat" in the journal to print only text
7039 messages, without any meta data like date or time.
7041 * Include tiny X server wrapper as a temporary stop-gap to
7042 teach XOrg udev display enumeration. This is used by display
7043 managers such as gdm, and will go away as soon as XOrg
7044 learned native udev hotplugging for display devices.
7046 * Add new systemd-cat tool for executing arbitrary programs
7047 with STDERR/STDOUT connected to the journal. Can also act as
7048 BSD logger replacement, and does so by default.
7050 * Optionally store all locally generated coredumps in the
7051 journal along with meta data.
7053 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
7054 writing short strings to files (for usage for /sys), and for
7055 creating symlinks, character and block device nodes.
7057 * New unit file option ControlGroupPersistent= to make cgroups
7058 persistent, following the mechanisms outlined in
7059 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
7061 * Support multiple local RTCs in a sane way
7063 * No longer monopolize IO when replaying readahead data on
7064 rotating disks, since we might starve non-file-system IO to
7065 death, since fanotify() will not see accesses done by blkid,
7068 * Do not show kernel threads in systemd-cgls anymore, unless
7069 requested with new -k switch.
7071 Contributions from: Dan Horák, Kay Sievers, Lennart
7072 Poettering, Michal Schmidt
7076 * This is mostly a test release, but incorporates many
7079 * The git repository moved to:
7080 git://anongit.freedesktop.org/systemd/systemd
7081 ssh://git.freedesktop.org/git/systemd/systemd
7083 * First release with the journal
7084 http://0pointer.de/blog/projects/the-journal.html
7086 * The journal replaces both systemd-kmsg-syslogd and
7087 systemd-stdout-bridge.
7089 * New sd_pid_get_unit() API call in libsystemd-logind
7091 * Many systemadm clean-ups
7093 * Introduce remote-fs-pre.target which is ordered before all
7094 remote mounts and may be used to start services before all
7097 * Added Mageia support
7099 * Add bash completion for systemd-loginctl
7101 * Actively monitor PID file creation for daemons which exit in
7102 the parent process before having finished writing the PID
7103 file in the daemon process. Daemons which do this need to be
7104 fixed (i.e. PID file creation must have finished before the
7105 parent exits), but we now react a bit more gracefully to them.
7107 * Add colourful boot output, mimicking the well-known output
7108 of existing distributions.
7110 * New option PassCredentials= for socket units, for
7111 compatibility with a recent kernel ABI breakage.
7113 * /etc/rc.local is now hooked in via a generator binary, and
7114 thus will no longer act as synchronization point during
7117 * systemctl list-unit-files now supports --root=.
7119 * systemd-tmpfiles now understands two new commands: z, Z for
7120 relabelling files according to the SELinux database. This is
7121 useful to apply SELinux labels to specific files in /sys,
7124 * Output of SysV services is now forwarded to both the console
7125 and the journal by default, not only just the console.
7127 * New man pages for all APIs from libsystemd-login.
7129 * The build tree got reorganized and the build system is a
7130 lot more modular allowing embedded setups to specifically
7131 select the components of systemd they are interested in.
7133 * Support for Linux systems lacking the kernel VT subsystem is
7136 * configure's --with-rootdir= got renamed to
7137 --with-rootprefix= to follow the naming used by udev and
7140 * Unless specified otherwise we will now install to /usr instead
7141 of /usr/local by default.
7143 * Processes with '@' in argv[0][0] are now excluded from the
7144 final shut-down killing spree, following the logic explained
7146 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
7148 * All processes remaining in a service cgroup when we enter
7149 the START or START_PRE states are now killed with
7150 SIGKILL. That means it is no longer possible to spawn
7151 background processes from ExecStart= lines (which was never
7152 supported anyway, and bad style).
7154 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
7155 reloading of units together.
7157 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
7158 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
7159 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
7160 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
7161 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek