1 systemd System and Service Manager
3 CHANGES WITH 233 in spe
5 * The confirmation spawn prompt has been reworked to offer the
8 (c)ontinue, proceed without asking anymore
9 (D)ump, show the state of the unit
10 (f)ail, don't execute the command and pretend it failed
12 (i)nfo, show a short summary of the unit
13 (j)obs, show jobs that are in progress
14 (s)kip, don't execute the command and pretend it succeeded
15 (y)es, execute the command
17 The 'n' choice for the confirmation spawn prompt has been removed,
18 because its meaning was confusing.
20 * Services of Type=notify require a READY=1 notification to be sent
21 during startup. If no such message is sent, the service now fails,
22 even if the main process exited with a successful exit code.
26 * The new RemoveIPC= option can be used to remove IPC objects owned by
27 the user or group of a service when that service exits.
29 * The new ProtectKernelModules= option can be used to disable explicit
30 load and unload operations of kernel modules by a service. In
31 addition access to /usr/lib/modules is removed if this option is set.
33 * ProtectSystem= option gained a new value "strict", which causes the
34 whole file system tree with the exception of /dev, /proc, and /sys,
35 to be remounted read-only for a service.
37 * The new ProtectKernelTunables= option can be used to disable
38 modification of configuration files in /sys and /proc by a service.
39 Various directories and files are remounted read-only, so access is
40 restricted even if the file permissions would allow it.
42 * The new ProtectControlGroups= option can be used to disable write
43 access by a service to /sys/fs/cgroup.
45 * Various systemd services have been hardened with
46 ProtectKernelTunables=yes, ProtectControlGroups=yes,
47 RestrictAddressFamilies=.
49 * Support for dynamically creating users for the lifetime of a service
50 has been added. If DynamicUser=yes is specified, user and group IDs
51 will be allocated from the range 61184..65519 for the lifetime of the
52 service. They can be resolved using the new nss-systemd.so NSS
53 module. The module must be enabled in /etc/nsswitch.conf. Services
54 started in this way have PrivateTmp= and RemoveIPC= enabled, so that
55 any resources allocated by the service will be cleaned up when the
56 service exits. They also have ProtectHome=read-only and
57 ProtectSystem=strict enabled, so they are not able to make any
58 permanent modifications to the system.
60 * The nss-systemd module also always resolves root and nobody, making
61 it possible to have no /etc/passwd or /etc/group files in minimal
62 container or chroot environments.
64 * Services may be started with their own user namespace using the new
65 boolean PrivateUsers= option. Only root, nobody, and the uid/gid
66 under which the service is running are mapped. All other users are
69 * Support for the cgroup namespace has been added to systemd-nspawn. If
70 supported by kernel, the container system started by systemd-nspawn
71 will have its own view of the cgroup hierarchy. This new behaviour
72 can be disabled using $SYSTEMD_NSPAWN_USE_CGNS environment variable.
74 * The new MemorySwapMax= option can be used to limit the maximum swap
75 usage under the unified cgroup hierarchy.
77 * Support for the CPU controller in the unified cgroup hierarchy has
78 been added, via the CPUWeight=, CPUStartupWeight=, CPUAccounting=
79 options. This controller requires out-of-tree patches for the kernel
80 and the support is provisional.
82 * Mount and automount units may now be created transiently
83 (i.e. dynamically at runtime via the bus API, instead of requiring
84 unit files in the file system).
86 * systemd-mount is a new tool which may mount file systems – much like
87 mount(8), optionally pulling in additional dependencies through
88 transient .mount and .automount units. For example, this tool
89 automatically runs fsck on a backing block device before mounting,
90 and allows the automount logic to be used dynamically from the
91 command line for establishing mount points. This tool is particularly
92 useful when dealing with removable media, as it will ensure fsck is
93 run – if necessary – before the first access and that the file system
94 is quickly unmounted after each access by utilizing the automount
95 logic. This maximizes the chance that the file system on the
96 removable media stays in a clean state, and if it isn't in a clean
97 state is fixed automatically.
99 * LazyUnmount=yes option for mount units has been added to expose the
100 umount --lazy option. Similarly, ForceUnmount=yes exposes the --force
103 * /efi will be used as the mount point of the EFI boot partition, if
104 the directory is present, and the mount point was not configured
105 through other means (e.g. fstab). If /efi directory does not exist,
106 /boot will be used as before. This makes it easier to automatically
107 mount the EFI partition on systems where /boot is used for something
110 * When operating on GPT disk images for containers, systemd-nspawn will
111 now mount the ESP to /boot or /efi according to the same rules as PID
112 1 running on a host. This allows tools like "bootctl" to operate
113 correctly within such containers, in order to make container images
114 bootable on physical systems.
116 * disk/by-id and disk/by-path symlinks are now created for NVMe drives.
118 * Two new user session targets have been added to support running
119 graphical sessions under the systemd --user instance:
120 graphical-session.target and graphical-session-pre.target. See
121 systemd.special(7) for a description of how those targets should be
124 * The vconsole initialization code has been significantly reworked to
125 use KD_FONT_OP_GET/SET ioctls instead of KD_FONT_OP_COPY and better
126 support unicode keymaps. Font and keymap configuration will now be
127 copied to all allocated virtual consoles.
129 * FreeBSD's bhyve virtualization is now detected.
131 * Information recorded in the journal for core dumps now includes the
132 contents of /proc/mountinfo and the command line of the process at
133 the top of the process hierarchy (which is usually the init process
136 * systemd-journal-gatewayd learned the --directory= option to serve
137 files from the specified location.
139 * journalctl --root=… can be used to peruse the journal in the
140 /var/log/ directories inside of a container tree. This is similar to
141 the existing --machine= option, but does not require the container to
144 * The hardware database has been extended to support
145 ID_INPUT_TRACKBALL, used in addition to ID_INPUT_MOUSE to identify
148 MOUSE_WHEEL_CLICK_ANGLE_HORIZONTAL hwdb property has been added to
149 specify the click rate for mice which include a horizontal wheel with
150 a click rate that is different than the one for the vertical wheel.
152 * systemd-run gained a new --wait option that makes service execution
153 synchronous. (Specifically, the command will not return until the
154 specified service binary exited.)
156 * systemctl gained a new --wait option that causes the start command to
157 wait until the units being started have terminated again.
159 * A new journal output mode "short-full" has been added which displays
160 timestamps with abbreviated English day names and adds a timezone
161 suffix. Those timestamps include more information than the default
162 "short" output mode, and can be passed directly to journalctl's
163 --since= and --until= options.
165 * /etc/resolv.conf will be bind-mounted into containers started by
166 systemd-nspawn, if possible, so any changes to resolv.conf contents
167 are automatically propagated to the container.
169 * The number of instances for socket-activated services originating
170 from a single IP address can be limited with
171 MaxConnectionsPerSource=, extending the existing setting of
174 * systemd-networkd gained support for vcan ("Virtual CAN") interface
177 * .netdev and .network configuration can now be extended through
180 * UDP Segmentation Offload, TCP Segmentation Offload, Generic
181 Segmentation Offload, Generic Receive Offload, Large Receive Offload
182 can be enabled and disabled using the new UDPSegmentationOffload=,
183 TCPSegmentationOffload=, GenericSegmentationOffload=,
184 GenericReceiveOffload=, LargeReceiveOffload= options in the
185 [Link] section of .link files.
187 * The Spanning Tree Protocol, Priority, Aging Time, and the Default
188 Port VLAN ID can be configured for bridge devices using the new STP=,
189 Priority=, AgeingTimeSec=, and DefaultPVID= settings in the [Bridge]
190 section of .netdev files.
192 * The route table to which routes received over DHCP or RA should be
193 added can be configured with the new RouteTable= option in the [DHCP]
194 and [IPv6AcceptRA] sections of .network files.
196 * The Address Resolution Protocol can be disabled on links managed by
197 systemd-networkd using the ARP=no setting in the [Link] section of
200 * New environment variables $SERVICE_RESULT, $EXIT_CODE and
201 $EXIT_STATUS are set for ExecStop= and ExecStopPost= commands, and
202 encode information about the result and exit codes of the current
203 service runtime cycle.
205 * systemd-sysctl will now configure kernel parameters in the order
206 they occur in the configuration files. This matches what sysctl
207 has been traditionally doing.
209 * kernel-install "plugins" that are executed to perform various
210 tasks after a new kernel is added and before an old one is removed
211 can now return a special value to terminate the procedure and
212 prevent any later plugins from running.
214 * Journald's SplitMode=login setting has been deprecated. It has been
215 removed from documentation, and its use is discouraged. In a future
216 release it will be completely removed, and made equivalent to current
217 default of SplitMode=uid.
219 * Storage=both option setting in /etc/systemd/coredump.conf has been
220 removed. With fast LZ4 compression storing the core dump twice is not
223 * The --share-system systemd-nspawn option has been replaced with an
224 (undocumented) variable $SYSTEMD_NSPAWN_SHARE_SYSTEM, but the use of
225 this functionality is discouraged. In addition the variables
226 $SYSTEMD_NSPAWN_SHARE_NS_IPC, $SYSTEMD_NSPAWN_SHARE_NS_PID,
227 $SYSTEMD_NSPAWN_SHARE_NS_UTS may be used to control the unsharing of
228 individual namespaces.
230 * "machinectl list" now shows the IP address of running containers in
231 the output, as well as OS release information.
233 * "loginctl list" now shows the TTY of each session in the output.
235 * sd-bus gained new API calls sd_bus_track_set_recursive(),
236 sd_bus_track_get_recursive(), sd_bus_track_count_name(),
237 sd_bus_track_count_sender(). They permit usage of sd_bus_track peer
238 tracking objects in a "recursive" mode, where a single client can be
239 counted multiple times, if it takes multiple references.
241 * sd-bus gained new API calls sd_bus_set_exit_on_disconnect() and
242 sd_bus_get_exit_on_disconnect(). They may be used to to make a
243 process using sd-bus automatically exit if the bus connection is
246 * Bus clients of the service manager may now "pin" loaded units into
247 memory, by taking an explicit reference on them. This is useful to
248 ensure the client can retrieve runtime data about the service even
249 after the service completed execution. Taking such a reference is
250 available only for privileged clients and should be helpful to watch
251 running services in a race-free manner, and in particular collect
252 information about exit statuses and results.
254 * The nss-resolve module has been changed to strictly return UNAVAIL
255 when communication via D-Bus with resolved failed, and NOTFOUND when
256 a lookup completed but was negative. This means it is now possible to
257 neatly configure fallbacks using nsswitch.conf result checking
258 expressions. Taking benefit of this, the new recommended
259 configuration line for the "hosts" entry in /etc/nsswitch.conf is:
261 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
263 * A new setting CtrlAltDelBurstAction= has been added to
264 /etc/systemd/system.conf which may be used to configure the precise
265 behaviour if the user on the console presses Ctrl-Alt-Del more often
266 than 7 times in 2s. Previously this would unconditionally result in
267 an expedited, immediate reboot. With this new setting the precise
268 operation may be configured in more detail, and also turned off
271 * In .netdev files two new settings RemoteChecksumTx= and
272 RemoteChecksumRx= are now understood that permit configuring the
273 remote checksumming logic for VXLAN networks.
275 * The service manager learnt a new "invocation ID" concept for invoked
276 services. Each runtime cycle of a service will get a new invocation
277 ID (a 128bit random UUID) assigned that identifies the current
278 run of the service uniquely and globally. A new invocation ID
279 is generated each time a service starts up. The journal will store
280 the invocation ID of a service along with any logged messages, thus
281 making the invocation ID useful for matching the online runtime of a
282 service with the offline log data it generated in a safe way without
283 relying on synchronized timestamps. In many ways this new service
284 invocation ID concept is similar to the kernel's boot ID concept that
285 uniquely and globally identifies the runtime of each boot. The
286 invocation ID of a service is passed to the service itself via an
287 environment variable ($INVOCATION_ID). A new bus call
288 GetUnitByInvocationID() has been added that is similar to GetUnit()
289 but instead of retrieving the bus path for a unit by its name
290 retrieves it by its invocation ID. The returned path is valid only as
291 long as the passed invocation ID is current.
293 * systemd-resolved gained a new "DNSStubListener" setting in
294 resolved.conf. It either takes a boolean value or the special values
295 "udp" and "tcp", and configures whether to enable the stub DNS
296 listener on 127.0.0.53:53.
298 * IP addresses configured via networkd may now carry additional
299 configuration settings supported by the kernel. New options include:
300 HomeAddress=, DuplicateAddressDetection=, ManageTemporaryAddress=,
301 PrefixRoute=, AutoJoin=.
303 * The PAM configuration fragment file for "user@.service" shipped with
304 systemd (i.e. the --user instance of systemd) has been stripped to
305 the minimum necessary to make the system boot. Previously, it
306 contained Fedora-specific stanzas that did not apply to other
307 distributions. It is expected that downstream distributions add
308 additional configuration lines, matching their needs to this file,
309 using it only as rough template of what systemd itself needs. Note
310 that this reduced fragment does not even include an invocation of
311 pam_limits which most distributions probably want to add, even though
312 systemd itself does not need it. (There's also the new build time
313 option --with-pamconfdir=no to disable installation of the PAM
316 * If PrivateDevices=yes is set for a service the CAP_SYS_RAWIO
317 capability is now also dropped from its set (in addition to
318 CAP_SYS_MKNOD as before).
320 * In service unit files it is now possible to connect a specific named
321 file descriptor with stdin/stdout/stdout of an executed service. The
322 name may be specified in matching .socket units using the
323 FileDescriptorName= setting.
325 * A number of journal settings may now be configured on the kernel
326 command line. Specifically, the following options are now understood:
327 systemd.journald.max_level_console=,
328 systemd.journald.max_level_store=,
329 systemd.journald.max_level_syslog=, systemd.journald.max_level_kmsg=,
330 systemd.journald.max_level_wall=.
332 * "systemctl is-enabled --full" will now show by which symlinks a unit
333 file is enabled in the unit dependency tree.
335 * Support for VeraCrypt encrypted partitions has been added to the
336 "cryptsetup" logic and /etc/crypttab.
338 * systemd-detect-virt gained support for a new --private-users switch
339 that checks whether the invoking processes are running inside a user
340 namespace. Similar, a new special value "private-users" for the
341 existing ConditionVirtualization= setting has been added, permitting
342 skipping of specific units in user namespace environments.
344 Contributions from: Alban Crequy, Alexander Kuleshov, Alfie John,
345 Andreas Henriksson, Andrew Jeddeloh, Balázs Úr, Bart Rulon, Benjamin
346 Richter, Ben Gamari, Ben Harris, Brian J. Murrell, Christian Brauner,
347 Christian Rebischke, Clinton Roy, Colin Walters, Cristian Rodríguez,
348 Daniel Hahler, Daniel Mack, Daniel Maixner, Daniel Rusek, Dan Dedrick,
349 Davide Cavalca, David Herrmann, David Michael, Dennis Wassenberg,
350 Djalal Harouni, Dongsu Park, Douglas Christman, Elias Probst, Eric
351 Cook, Erik Karlsson, Evgeny Vereshchagin, Felipe Sateler, Felix Zhang,
352 Franck Bui, George Hilliard, Giuseppe Scrivano, HATAYAMA Daisuke,
353 Heikki Kemppainen, Hendrik Brueckner, hi117, Ismo Puustinen, Ivan
354 Shapovalov, Jakub Filak, Jakub Wilk, Jan Synacek, Jason Kölker,
355 Jean-Sébastien Bour, Jiří Pírko, Jonathan Boulle, Jorge Niedbalski,
356 Keith Busch, kristbaum, Kyle Russell, Lans Zhang, Lennart Poettering,
357 Leonardo Brondani Schenkel, Lucas Werkmeister, Luca Bruno, Lukáš
358 Nykrýn, Maciek Borzecki, Mantas Mikulėnas, Marc-Antoine Perennou,
359 Marcel Holtmann, Marcos Mello, Martin Ejdestig, Martin Pitt, Matej
360 Habrnal, Maxime de Roucy, Michael Biebl, Michael Chapman, Michael Hoy,
361 Michael Olbrich, Michael Pope, Michal Sekletar, Michal Soltys, Mike
362 Gilbert, Nick Owens, Patrik Flykt, Paweł Szewczyk, Peter Hutterer,
363 Piotr Drąg, Reid Price, Richard W.M. Jones, Roman Stingler, Ronny
364 Chevalier, Seraphime Kirkovski, Stefan Schweter, Steve Muir, Susant
365 Sahani, Tejun Heo, Thomas Blume, Thomas H. P. Andersen, Tiago Levit,
366 Tobias Jungel, Tomáš Janoušek, Topi Miettinen, Torstein Husebø, Umut
367 Tezduyar Lindskog, Vito Caputo, WaLyong Cho, Wilhelm Schuster, Yann
368 E. MORIN, Yi EungJun, Yuki Inoguchi, Yu Watanabe, Zbigniew
369 Jędrzejewski-Szmek, Zeal Jagannatha
371 — Santa Fe, 2016-11-03
375 * In service units the various ExecXYZ= settings have been extended
376 with an additional special character as first argument of the
377 assigned value: if the character '+' is used the specified command
378 line it will be run with full privileges, regardless of User=,
379 Group=, CapabilityBoundingSet= and similar options. The effect is
380 similar to the existing PermissionsStartOnly= option, but allows
381 configuration of this concept for each executed command line
384 * Services may now alter the service watchdog timeout at runtime by
385 sending a WATCHDOG_USEC= message via sd_notify().
387 * MemoryLimit= and related unit settings now optionally take percentage
388 specifications. The percentage is taken relative to the amount of
389 physical memory in the system (or in case of containers, the assigned
390 amount of memory). This allows scaling service resources neatly with
391 the amount of RAM available on the system. Similarly, systemd-logind's
392 RuntimeDirectorySize= option now also optionally takes percentage
395 * In similar fashion TasksMax= takes percentage values now, too. The
396 value is taken relative to the configured maximum number of processes
397 on the system. The per-service task maximum has been changed to 15%
398 using this functionality. (Effectively this is an increase of 512 →
399 4915 for service units, given the kernel's default pid_max setting.)
401 * Calendar time specifications in .timer units now understand a ".."
402 syntax for time ranges. Example: "4..7:10" may now be used for
403 defining a timer that is triggered at 4:10am, 5:10am, 6:10am and
406 * The InaccessableDirectories=, ReadOnlyDirectories= and
407 ReadWriteDirectories= unit file settings have been renamed to
408 InaccessablePaths=, ReadOnlyPaths= and ReadWritePaths= and may now be
409 applied to all kinds of file nodes, and not just directories, with
410 the exception of symlinks. Specifically these settings may now be
411 used on block and character device nodes, UNIX sockets and FIFOS as
412 well as regular files. The old names of these settings remain
413 available for compatibility.
415 * systemd will now log about all service processes it kills forcibly
416 (using SIGKILL) because they remained after the clean shutdown phase
417 of the service completed. This should help identifying services that
418 shut down uncleanly. Moreover if KillUserProcesses= is enabled in
419 systemd-logind's configuration a similar log message is generated for
420 processes killed at the end of each session due to this setting.
422 * systemd will now set the $JOURNAL_STREAM environment variable for all
423 services whose stdout/stderr are connected to the Journal (which
424 effectively means by default: all services). The variable contains
425 the device and inode number of the file descriptor used for
426 stdout/stderr. This may be used by invoked programs to detect whether
427 their stdout/stderr is connected to the Journal, in which case they
428 can switch over to direct Journal communication, thus being able to
429 pass extended, structured metadata along with their log messages. As
430 one example, this is now used by glib's logging primitives.
432 * When using systemd's default tmp.mount unit for /tmp, the mount point
433 will now be established with the "nosuid" and "nodev" options. This
434 avoids privilege escalation attacks that put traps and exploits into
435 /tmp. However, this might cause problems if you e. g. put container
436 images or overlays into /tmp; if you need this, override tmp.mount's
437 "Options=" with a drop-in, or mount /tmp from /etc/fstab with your
440 * systemd now supports the "memory" cgroup controller also on
443 * The systemd-cgtop tool now optionally takes a control group path as
444 command line argument. If specified, the control group list shown is
445 limited to subgroups of that group.
447 * The SystemCallFilter= unit file setting gained support for
448 pre-defined, named system call filter sets. For example
449 SystemCallFilter=@clock is now an effective way to make all clock
450 changing-related system calls unavailable to a service. A number of
451 similar pre-defined groups are defined. Writing system call filters
452 for system services is simplified substantially with this new
453 concept. Accordingly, all of systemd's own, long-running services now
454 enable system call filtering based on this, by default.
456 * A new service setting MemoryDenyWriteExecute= has been added, taking
457 a boolean value. If turned on, a service may no longer create memory
458 mappings that are writable and executable at the same time. This
459 enhances security for services where this is enabled as it becomes
460 harder to dynamically write and then execute memory in exploited
461 service processes. This option has been enabled for all of systemd's
462 own long-running services.
464 * A new RestrictRealtime= service setting has been added, taking a
465 boolean argument. If set the service's processes may no longer
466 acquire realtime scheduling. This improves security as realtime
467 scheduling may otherwise be used to easily freeze the system.
469 * systemd-nspawn gained a new switch --notify-ready= taking a boolean
470 value. This may be used for requesting that the system manager inside
471 of the container reports start-up completion to nspawn which then
472 propagates this notification further to the service manager
473 supervising nspawn itself. A related option NotifyReady= in .nspawn
474 files has been added too. This functionality allows ordering of the
475 start-up of multiple containers using the usual systemd ordering
478 * machinectl gained a new command "stop" that is an alias for
481 * systemd-resolved gained support for contacting DNS servers on
482 link-local IPv6 addresses.
484 * If systemd-resolved receives the SIGUSR2 signal it will now flush all
485 its caches. A method call for requesting the same operation has been
486 added to the bus API too, and is made available via "systemd-resolve
489 * systemd-resolve gained a new --status switch. If passed a brief
490 summary of the used DNS configuration with per-interface information
493 * resolved.conf gained a new Cache= boolean option, defaulting to
494 on. If turned off local DNS caching is disabled. This comes with a
495 performance penalty in particular when DNSSEC is enabled. Note that
496 resolved disables its internal caching implicitly anyway, when the
497 configured DNS server is on a host-local IP address such as ::1 or
498 127.0.0.1, thus automatically avoiding double local caching.
500 * systemd-resolved now listens on the local IP address 127.0.0.53:53
501 for DNS requests. This improves compatibility with local programs
502 that do not use the libc NSS or systemd-resolved's bus APIs for name
503 resolution. This minimal DNS service is only available to local
504 programs and does not implement the full DNS protocol, but enough to
505 cover local DNS clients. A new, static resolv.conf file, listing just
506 this DNS server is now shipped in /usr/lib/systemd/resolv.conf. It is
507 now recommended to make /etc/resolv.conf a symlink to this file in
508 order to route all DNS lookups to systemd-resolved, regardless if
509 done via NSS, the bus API or raw DNS packets. Note that this local
510 DNS service is not as fully featured as the libc NSS or
511 systemd-resolved's bus APIs. For example, as unicast DNS cannot be
512 used to deliver link-local address information (as this implies
513 sending a local interface index along), LLMNR/mDNS support via this
514 interface is severely restricted. It is thus strongly recommended for
515 all applications to use the libc NSS API or native systemd-resolved
518 * systemd-networkd's bridge support learned a new setting
519 VLANFiltering= for controlling VLAN filtering. Moreover a new section
520 in .network files has been added for configuring VLAN bridging in
521 more detail: VLAN=, EgressUntagged=, PVID= in [BridgeVLAN].
523 * systemd-networkd's IPv6 Router Advertisement code now makes use of
524 the DNSSL and RDNSS options. This means IPv6 DNS configuration may
525 now be acquired without relying on DHCPv6. Two new options
526 UseDomains= and UseDNS= have been added to configure this behaviour.
528 * systemd-networkd's IPv6AcceptRouterAdvertisements= option has been
529 renamed IPv6AcceptRA=, without altering its behaviour. The old
530 setting name remains available for compatibility reasons.
532 * The systemd-networkd VTI/VTI6 tunneling support gained new options
533 Key=, InputKey= and OutputKey=.
535 * systemd-networkd gained support for VRF ("Virtual Routing Function")
536 interface configuration.
538 * "systemctl edit" may now be used to create new unit files by
539 specifying the --force switch.
541 * sd-event gained a new function sd_event_get_iteration() for
542 requesting the current iteration counter of the event loop. It starts
543 at zero and is increased by one with each event loop iteration.
545 * A new rpm macro %systemd_ordering is provided by the macros.systemd
546 file. It can be used in lieu of %systemd_requires in packages which
547 don't use any systemd functionality and are intended to be installed
548 in minimal containers without systemd present. This macro provides
549 ordering dependencies to ensure that if the package is installed in
550 the same rpm transaction as systemd, systemd will be installed before
551 the scriptlets for the package are executed, allowing unit presets
554 New macros %_systemdgeneratordir and %_systemdusergeneratordir have
555 been added to simplify packaging of generators.
557 * The os-release file gained VERSION_CODENAME field for the
558 distribution nickname (e.g. VERSION_CODENAME=woody).
560 * New udev property UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG=1
561 can be set to disable parsing of metadata and the creation
562 of persistent symlinks for that device.
564 * The v230 change to tag framebuffer devices (/dev/fb*) with "uaccess"
565 to make them available to logged-in users has been reverted.
567 * Much of the common code of the various systemd components is now
568 built into an internal shared library libsystemd-shared-231.so
569 (incorporating the systemd version number in the name, to be updated
570 with future releases) that the components link to. This should
571 decrease systemd footprint both in memory during runtime and on
572 disk. Note that the shared library is not for public use, and is
573 neither API not ABI stable, but is likely to change with every new
574 released update. Packagers need to make sure that binaries
575 linking to libsystemd-shared.so are updated in step with the
578 * Configuration for "mkosi" is now part of the systemd
579 repository. mkosi is a tool to easily build legacy-free OS images,
580 and is available on github: https://github.com/systemd/mkosi. If
581 "mkosi" is invoked in the build tree a new raw OS image is generated
582 incorporating the systemd sources currently being worked on and a
583 clean, fresh distribution installation. The generated OS image may be
584 booted up with "systemd-nspawn -b -i", qemu-kvm or on any physical
585 UEFI PC. This functionality is particularly useful to easily test
586 local changes made to systemd in a pristine, defined environment. See
589 * configure learned the --with-support-url= option to specify the
590 distribution's bugtracker.
592 Contributions from: Alban Crequy, Alessandro Puccetti, Alessio Igor
593 Bogani, Alexander Kuleshov, Alexander Kurtz, Alex Gaynor, Andika
594 Triwidada, Andreas Pokorny, Andreas Rammhold, Andrew Jeddeloh, Ansgar
595 Burchardt, Atrotors, Benjamin Drung, Brian Boylston, Christian Hesse,
596 Christian Rebischke, Daniele Medri, Daniel Mack, Dave Reisner, David
597 Herrmann, David Michael, Djalal Harouni, Douglas Christman, Elias
598 Probst, Evgeny Vereshchagin, Federico Mena Quintero, Felipe Sateler,
599 Franck Bui, Harald Hoyer, Ian Lee, Ivan Shapovalov, Jakub Wilk, Jan
600 Janssen, Jean-Sébastien Bour, John Paul Adrian Glaubitz, Jouke
601 Witteveen, Kai Ruhnau, kpengboy, Kyle Walker, Lénaïc Huard, Lennart
602 Poettering, Luca Bruno, Lukas Lösche, Lukáš Nykrýn, mahkoh, Marcel
603 Holtmann, Martin Pitt, Marty Plummer, Matthieu Codron, Max Prokhorov,
604 Michael Biebl, Michael Karcher, Michael Olbrich, Michał Bartoszkiewicz,
605 Michal Sekletar, Michal Soltys, Minkyung, Muhammet Kara, mulkieran,
606 Otto Wallenius, Pablo Lezaeta Reyes, Peter Hutterer, Ronny Chevalier,
607 Rusty Bird, Stef Walter, Susant Sahani, Tejun Heo, Thomas Blume, Thomas
608 Haller, Thomas H. P. Andersen, Tobias Jungel, Tom Gundersen, Tom Yan,
609 Topi Miettinen, Torstein Husebø, Valentin Vidić, Viktar Vaŭčkievič,
610 WaLyong Cho, Weng Xuetian, Werner Fink, Zbigniew Jędrzejewski-Szmek
616 * DNSSEC is now turned on by default in systemd-resolved (in
617 "allow-downgrade" mode), but may be turned off during compile time by
618 passing "--with-default-dnssec=no" to "configure" (and of course,
619 during runtime with DNSSEC= in resolved.conf). We recommend
620 downstreams to leave this on at least during development cycles and
621 report any issues with the DNSSEC logic upstream. We are very
622 interested in collecting feedback about the DNSSEC validator and its
623 limitations in the wild. Note however, that DNSSEC support is
624 probably nothing downstreams should turn on in stable distros just
625 yet, as it might create incompatibilities with a few DNS servers and
626 networks. We tried hard to make sure we downgrade to non-DNSSEC mode
627 automatically whenever we detect such incompatible setups, but there
628 might be systems we do not cover yet. Hence: please help us testing
629 the DNSSEC code, leave this on where you can, report back, but then
630 again don't consider turning this on in your stable, LTS or
631 production release just yet. (Note that you have to enable
632 nss-resolve in /etc/nsswitch.conf, to actually use systemd-resolved
633 and its DNSSEC mode for host name resolution from local
636 * systemd-resolve conveniently resolves DANE records with the --tlsa
637 option and OPENPGPKEY records with the --openpgp option. It also
638 supports dumping raw DNS record data via the new --raw= switch.
640 * systemd-logind will now by default terminate user processes that are
641 part of the user session scope unit (session-XX.scope) when the user
642 logs out. This behavior is controlled by the KillUserProcesses=
643 setting in logind.conf, and the previous default of "no" is now
644 changed to "yes". This means that user sessions will be properly
645 cleaned up after, but additional steps are necessary to allow
646 intentionally long-running processes to survive logout.
648 While the user is logged in at least once, user@.service is running,
649 and any service that should survive the end of any individual login
650 session can be started at a user service or scope using systemd-run.
651 systemd-run(1) man page has been extended with an example which shows
652 how to run screen in a scope unit underneath user@.service. The same
653 command works for tmux.
655 After the user logs out of all sessions, user@.service will be
656 terminated too, by default, unless the user has "lingering" enabled.
657 To effectively allow users to run long-term tasks even if they are
658 logged out, lingering must be enabled for them. See loginctl(1) for
659 details. The default polkit policy was modified to allow users to
660 set lingering for themselves without authentication.
662 Previous defaults can be restored at compile time by the
663 --without-kill-user-processes option to "configure".
665 * systemd-logind gained new configuration settings SessionsMax= and
666 InhibitorsMax=, both with a default of 8192. It will not register new
667 user sessions or inhibitors above this limit.
669 * systemd-logind will now reload configuration on SIGHUP.
671 * The unified cgroup hierarchy added in Linux 4.5 is now supported.
672 Use systemd.unified_cgroup_hierarchy=1 on the kernel command line to
673 enable. Also, support for the "io" cgroup controller in the unified
674 hierarchy has been added, so that the "memory", "pids" and "io" are
675 now the controllers that are supported on the unified hierarchy.
677 WARNING: it is not possible to use previous systemd versions with
678 systemd.unified_cgroup_hierarchy=1 and the new kernel. Therefore it
679 is necessary to also update systemd in the initramfs if using the
680 unified hierarchy. An updated SELinux policy is also required.
682 * LLDP support has been extended, and both passive (receive-only) and
683 active (sender) modes are supported. Passive mode ("routers-only") is
684 enabled by default in systemd-networkd. Active LLDP mode is enabled
685 by default for containers on the internal network. The "networkctl
686 lldp" command may be used to list information gathered. "networkctl
687 status" will also show basic LLDP information on connected peers now.
689 * The IAID and DUID unique identifier sent in DHCP requests may now be
690 configured for the system and each .network file managed by
691 systemd-networkd using the DUIDType=, DUIDRawData=, IAID= options.
693 * systemd-networkd gained support for configuring proxy ARP support for
694 each interface, via the ProxyArp= setting in .network files. It also
695 gained support for configuring the multicast querier feature of
696 bridge devices, via the new MulticastQuerier= setting in .netdev
697 files. Similarly, snooping on the IGMP traffic can be controlled
698 via the new setting MulticastSnooping=.
700 A new setting PreferredLifetime= has been added for addresses
701 configured in .network file to configure the lifetime intended for an
704 The systemd-networkd DHCP server gained the option EmitRouter=, which
705 defaults to yes, to configure whether the DHCP Option 3 (Router)
708 * The testing tool /usr/lib/systemd/systemd-activate is renamed to
709 systemd-socket-activate and installed into /usr/bin. It is now fully
712 * systemd-journald now uses separate threads to flush changes to disk
713 when closing journal files, thus reducing impact of slow disk I/O on
716 * The sd-journal API gained two new calls
717 sd_journal_open_directory_fd() and sd_journal_open_files_fd() which
718 can be used to open journal files using file descriptors instead of
719 file or directory paths. sd_journal_open_container() has been
720 deprecated, sd_journal_open_directory_fd() should be used instead
721 with the flag SD_JOURNAL_OS_ROOT.
723 * journalctl learned a new output mode "-o short-unix" that outputs log
724 lines prefixed by their UNIX time (i.e. seconds since Jan 1st, 1970
725 UTC). It also gained support for a new --no-hostname setting to
726 suppress the hostname column in the family of "short" output modes.
728 * systemd-ask-password now optionally skips printing of the password to
729 stdout with --no-output which can be useful in scripts.
731 * Framebuffer devices (/dev/fb*) and 3D printers and scanners
732 (devices tagged with ID_MAKER_TOOL) are now tagged with
733 "uaccess" and are available to logged in users.
735 * The DeviceAllow= unit setting now supports specifiers (with "%").
737 * "systemctl show" gained a new --value switch, which allows print a
738 only the contents of a specific unit property, without also printing
739 the property's name. Similar support was added to "show*" verbs
740 of loginctl and machinectl that output "key=value" lists.
742 * A new unit type "generated" was added for files dynamically generated
743 by generator tools. Similarly, a new unit type "transient" is used
744 for unit files created using the runtime API. "systemctl enable" will
745 refuse to operate on such files.
747 * A new command "systemctl revert" has been added that may be used to
748 revert to the vendor version of a unit file, in case local changes
749 have been made by adding drop-ins or overriding the unit file.
751 * "machinectl clean" gained a new verb to automatically remove all or
752 just hidden container images.
754 * systemd-tmpfiles gained support for a new line type "e" for emptying
755 directories, if they exist, without creating them if they don't.
757 * systemd-nspawn gained support for automatically patching the UID/GIDs
758 of the owners and the ACLs of all files and directories in a
759 container tree to match the UID/GID user namespacing range selected
760 for the container invocation. This mode is enabled via the new
761 --private-users-chown switch. It also gained support for
762 automatically choosing a free, previously unused UID/GID range when
763 starting a container, via the new --private-users=pick setting (which
764 implies --private-users-chown). Together, these options for the first
765 time make user namespacing for nspawn containers fully automatic and
766 thus deployable. The systemd-nspawn@.service template unit file has
767 been changed to use this functionality by default.
769 * systemd-nspawn gained a new --network-zone= switch, that allows
770 creating ad-hoc virtual Ethernet links between multiple containers,
771 that only exist as long as at least one container referencing them is
772 running. This allows easy connecting of multiple containers with a
773 common link that implements an Ethernet broadcast domain. Each of
774 these network "zones" may be named relatively freely by the user, and
775 may be referenced by any number of containers, but each container may
776 only reference one of these "zones". On the lower level, this is
777 implemented by an automatically managed bridge network interface for
778 each zone, that is created when the first container referencing its
779 zone is created and removed when the last one referencing its zone
782 * The default start timeout may now be configured on the kernel command
783 line via systemd.default_timeout_start_sec=. It was already
784 configurable via the DefaultTimeoutStartSec= option in
785 /etc/systemd/system.conf.
787 * Socket units gained a new TriggerLimitIntervalSec= and
788 TriggerLimitBurst= setting to configure a limit on the activation
789 rate of the socket unit.
791 * The LimitNICE= setting now optionally takes normal UNIX nice values
792 in addition to the raw integer limit value. If the specified
793 parameter is prefixed with "+" or "-" and is in the range -20..19 the
794 value is understood as UNIX nice value. If not prefixed like this it
795 is understood as raw RLIMIT_NICE limit.
797 * Note that the effect of the PrivateDevices= unit file setting changed
798 slightly with this release: the per-device /dev file system will be
799 mounted read-only from this version on, and will have "noexec"
800 set. This (minor) change of behavior might cause some (exceptional)
801 legacy software to break, when PrivateDevices=yes is set for its
802 service. Please leave PrivateDevices= off if you run into problems
805 * systemd-bootchart has been split out to a separate repository:
806 https://github.com/systemd/systemd-bootchart
808 * systemd-bus-proxyd has been removed, as kdbus is unlikely to still be
809 merged into the kernel in its current form.
811 * The compatibility libraries libsystemd-daemon.so,
812 libsystemd-journal.so, libsystemd-id128.so, and libsystemd-login.so
813 which have been deprecated since systemd-209 have been removed along
814 with the corresponding pkg-config files. All symbols provided by
815 those libraries are provided by libsystemd.so.
817 * The Capabilities= unit file setting has been removed (it is ignored
818 for backwards compatibility). AmbientCapabilities= and
819 CapabilityBoundingSet= should be used instead.
821 * A new special target has been added, initrd-root-device.target,
822 which creates a synchronization point for dependencies of the root
823 device in early userspace. Initramfs builders must ensure that this
824 target is now included in early userspace.
826 Contributions from: Alban Crequy, Alexander Kuleshov, Alexander Shopov,
827 Alex Crawford, Andre Klärner, Andrew Eikum, Beniamino Galvani, Benjamin
828 Robin, Biao Lu, Bjørnar Ness, Calvin Owens, Christian Hesse, Clemens
829 Gruber, Colin Guthrie, Daniel Drake, Daniele Medri, Daniel J Walsh,
830 Daniel Mack, Dan Nicholson, daurnimator, David Herrmann, David
831 R. Hedges, Elias Probst, Emmanuel Gil Peyrot, EMOziko, Evgeny
832 Vereshchagin, Federico, Felipe Sateler, Filipe Brandenburger, Franck
833 Bui, frankheckenbach, gdamjan, Georgia Brikis, Harald Hoyer, Hendrik
834 Brueckner, Hristo Venev, Iago López Galeiras, Ian Kelling, Ismo
835 Puustinen, Jakub Wilk, Jaroslav Škarvada, Jeff Huang, Joel Holdsworth,
836 John Paul Adrian Glaubitz, Jonathan Boulle, kayrus, Klearchos
837 Chaloulos, Kyle Russell, Lars Uebernickel, Lennart Poettering, Lubomir
838 Rintel, Lukáš Nykrýn, Mantas Mikulėnas, Marcel Holtmann, Martin Pitt,
839 Michael Biebl, michaelolbrich, Michał Bartoszkiewicz, Michal Koutný,
840 Michal Sekletar, Mike Frysinger, Mike Gilbert, Mingcong Bai, Ming Lin,
841 mulkieran, muzena, Nalin Dahyabhai, Naohiro Aota, Nathan McSween,
842 Nicolas Braud-Santoni, Patrik Flykt, Peter Hutterer, Peter Mattern,
843 Petr Lautrbach, Petros Angelatos, Piotr Drąg, Rabin Vincent, Robert
844 Węcławski, Ronny Chevalier, Samuel Tardieu, Stefan Saraev, Stefan
845 Schallenberg aka nafets227, Steven Siloti, Susant Sahani, Sylvain
846 Plantefève, Taylor Smock, Tejun Heo, Thomas Blume, Thomas Haller,
847 Thomas H. P. Andersen, Tobias Klauser, Tom Gundersen, topimiettinen,
848 Torstein Husebø, Umut Tezduyar Lindskog, Uwe Kleine-König, Victor Toso,
849 Vinay Kulkarni, Vito Caputo, Vittorio G (VittGam), Vladimir Panteleev,
850 Wieland Hoffmann, Wouter Verhelst, Yu Watanabe, Zbigniew
853 — Fairfax, 2016-05-21
857 * The systemd-resolved DNS resolver service has gained a substantial
858 set of new features, most prominently it may now act as a DNSSEC
859 validating stub resolver. DNSSEC mode is currently turned off by
860 default, but is expected to be turned on by default in one of the
861 next releases. For now, we invite everybody to test the DNSSEC logic
862 by setting DNSSEC=allow-downgrade in /etc/systemd/resolved.conf. The
863 service also gained a full set of D-Bus interfaces, including calls
864 to configure DNS and DNSSEC settings per link (for use by external
865 network management software). systemd-resolved and systemd-networkd
866 now distinguish between "search" and "routing" domains. The former
867 are used to qualify single-label names, the latter are used purely
868 for routing lookups within certain domains to specific links.
869 resolved now also synthesizes RRs for all entries from /etc/hosts.
871 * The systemd-resolve tool (which is a client utility for
872 systemd-resolved) has been improved considerably and is now fully
873 supported and documented. Hence it has moved from /usr/lib/systemd to
876 * /dev/disk/by-path/ symlink support has been (re-)added for virtio
879 * The coredump collection logic has been reworked: when a coredump is
880 collected it is now written to disk, compressed and processed
881 (including stacktrace extraction) from a new instantiated service
882 systemd-coredump@.service, instead of directly from the
883 /proc/sys/kernel/core_pattern hook we provide. This is beneficial as
884 processing large coredumps can take up a substantial amount of
885 resources and time, and this previously happened entirely outside of
886 systemd's service supervision. With the new logic the core_pattern
887 hook only does minimal metadata collection before passing off control
888 to the new instantiated service, which is configured with a time
889 limit, a nice level and other settings to minimize negative impact on
890 the rest of the system. Also note that the new logic will honour the
891 RLIMIT_CORE setting of the crashed process, which now allows users
892 and processes to turn off coredumping for their processes by setting
895 * The RLIMIT_CORE resource limit now defaults to "unlimited" for PID 1
896 and all forked processes by default. Previously, PID 1 would leave
897 the setting at "0" for all processes, as set by the kernel. Note that
898 the resource limit traditionally has no effect on the generated
899 coredumps on the system if the /proc/sys/kernel/core_pattern hook
900 logic is used. Since the limit is now honoured (see above) its
901 default has been changed so that the coredumping logic is enabled by
902 default for all processes, while allowing specific opt-out.
904 * When the stacktrace is extracted from processes of system users, this
905 is now done as "systemd-coredump" user, in order to sandbox this
906 potentially security sensitive parsing operation. (Note that when
907 processing coredumps of normal users this is done under the user ID
908 of process that crashed, as before.) Packagers should take notice
909 that it is now necessary to create the "systemd-coredump" system user
910 and group at package installation time.
912 * The systemd-activate socket activation testing tool gained support
913 for SOCK_DGRAM and SOCK_SEQPACKET sockets using the new --datagram
914 and --seqpacket switches. It also has been extended to support both
915 new-style and inetd-style file descriptor passing. Use the new
916 --inetd switch to request inetd-style file descriptor passing.
918 * Most systemd tools now honor a new $SYSTEMD_COLORS environment
919 variable, which takes a boolean value. If set to false, ANSI color
920 output is disabled in the tools even when run on a terminal that
923 * The VXLAN support in networkd now supports two new settings
924 DestinationPort= and PortRange=.
926 * A new systemd.machine_id= kernel command line switch has been added,
927 that may be used to set the machine ID in /etc/machine-id if it is
928 not initialized yet. This command line option has no effect if the
929 file is already initialized.
931 * systemd-nspawn gained a new --as-pid2 switch that invokes any
932 specified command line as PID 2 rather than PID 1 in the
933 container. In this mode PID 1 is a minimal stub init process that
934 implements the special POSIX and Linux semantics of PID 1 regarding
935 signal and child process management. Note that this stub init process
936 is implemented in nspawn itself and requires no support from the
937 container image. This new logic is useful to support running
938 arbitrary commands in the container, as normal processes are
939 generally not prepared to run as PID 1.
941 * systemd-nspawn gained a new --chdir= switch for setting the current
942 working directory for the process started in the container.
944 * "journalctl /dev/sda" will now output all kernel log messages for
945 specified device from the current boot, in addition to all devices
946 that are parents of it. This should make log output about devices
947 pretty useful, as long as kernel drivers attach enough metadata to
948 the log messages. (The usual SATA drivers do.)
950 * The sd-journal API gained two new calls
951 sd_journal_has_runtime_files() and sd_journal_has_persistent_files()
952 that report whether log data from /run or /var has been found.
954 * journalctl gained a new switch "--fields" that prints all journal
955 record field names currently in use in the journal. This is backed
956 by two new sd-journal API calls sd_journal_enumerate_fields() and
957 sd_journal_restart_fields().
959 * Most configurable timeouts in systemd now expect an argument of
960 "infinity" to turn them off, instead of "0" as before. The semantics
961 from now on is that a timeout of "0" means "now", and "infinity"
962 means "never". To maintain backwards compatibility, "0" continues to
963 turn off previously existing timeout settings.
965 * "systemctl reload-or-try-restart" has been renamed to "systemctl
966 try-reload-or-restart" to clarify what it actually does: the "try"
967 logic applies to both reloading and restarting, not just restarting.
968 The old name continues to be accepted for compatibility.
970 * On boot-up, when PID 1 detects that the system clock is behind the
971 release date of the systemd version in use, the clock is now set
972 to the latter. Previously, this was already done in timesyncd, in order
973 to avoid running with clocks set to the various clock epochs such as
974 1902, 1938 or 1970. With this change the logic is now done in PID 1
975 in addition to timesyncd during early boot-up, so that it is enforced
976 before the first process is spawned by systemd. Note that the logic
977 in timesyncd remains, as it is more comprehensive and ensures
978 clock monotonicity by maintaining a persistent timestamp file in
979 /var. Since /var is generally not available in earliest boot or the
980 initrd, this part of the logic remains in timesyncd, and is not done
983 * Support for tweaking details in net_cls.class_id through the
984 NetClass= configuration directive has been removed, as the kernel
985 people have decided to deprecate that controller in cgroup v2.
986 Userspace tools such as nftables are moving over to setting rules
987 that are specific to the full cgroup path of a task, which obsoletes
988 these controllers anyway. The NetClass= directive is kept around for
989 legacy compatibility reasons. For a more in-depth description of the
990 kernel change, please refer to the respective upstream commit:
992 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd1060a1d671
994 * A new service setting RuntimeMaxSec= has been added that may be used
995 to specify a maximum runtime for a service. If the timeout is hit, the
996 service is terminated and put into a failure state.
998 * A new service setting AmbientCapabilities= has been added. It allows
999 configuration of additional Linux process capabilities that are
1000 passed to the activated processes. This is only available on very
1003 * The process resource limit settings in service units may now be used
1004 to configure hard and soft limits individually.
1006 * The various libsystemd APIs such as sd-bus or sd-event now publicly
1007 expose support for gcc's __attribute__((cleanup())) C extension.
1008 Specifically, for many object destructor functions alternative
1009 versions have been added that have names suffixed with "p" and take a
1010 pointer to a pointer to the object to destroy, instead of just a
1011 pointer to the object itself. This is useful because these destructor
1012 functions may be used directly as parameters to the cleanup
1013 construct. Internally, systemd has been a heavy user of this GCC
1014 extension for a long time, and with this change similar support is
1015 now available to consumers of the library outside of systemd. Note
1016 that by using this extension in your sources compatibility with old
1017 and strictly ANSI compatible C compilers is lost. However, all gcc or
1018 LLVM versions of recent years support this extension.
1020 * Timer units gained support for a new setting RandomizedDelaySec= that
1021 allows configuring some additional randomized delay to the configured
1022 time. This is useful to spread out timer events to avoid load peaks in
1023 clusters or larger setups.
1025 * Calendar time specifications now support sub-second accuracy.
1027 * Socket units now support listening on SCTP and UDP-lite protocol
1030 * The sd-event API now comes with a full set of man pages.
1032 * Older versions of systemd contained experimental support for
1033 compressing journal files and coredumps with the LZ4 compressor that
1034 was not compatible with the lz4 binary (due to API limitations of the
1035 lz4 library). This support has been removed; only support for files
1036 compatible with the lz4 binary remains. This LZ4 logic is now
1037 officially supported and no longer considered experimental.
1039 * The dkr image import logic has been removed again from importd. dkr's
1040 micro-services focus doesn't fit into the machine image focus of
1041 importd, and quickly got out of date with the upstream dkr API.
1043 * Creation of the /run/lock/lockdev/ directory was dropped from
1044 tmpfiles.d/legacy.conf. Better locking mechanisms like flock() have
1045 been available for many years. If you still need this, you need to
1046 create your own tmpfiles.d config file with:
1048 d /run/lock/lockdev 0775 root lock -
1050 Contributions from: Abdo Roig-Maranges, Alban Crequy, Aleksander
1051 Adamowski, Alexander Kuleshov, Andreas Pokorny, Andrei Borzenkov,
1052 Andrew Wilcox, Arthur Clement, Beniamino Galvani, Casey Schaufler,
1053 Chris Atkinson, Chris Mayo, Christian Hesse, Damjan Georgievski, Dan
1054 Dedrick, Daniele Medri, Daniel J Walsh, Daniel Korostil, Daniel Mack,
1055 David Herrmann, Dimitri John Ledkov, Dominik Hannen, Douglas Christman,
1056 Evgeny Vereshchagin, Filipe Brandenburger, Franck Bui, Gabor Kelemen,
1057 Harald Hoyer, Hayden Walles, Helmut Grohne, Henrik Kaare Poulsen,
1058 Hristo Venev, Hui Wang, Indrajit Raychaudhuri, Ismo Puustinen, Jakub
1059 Wilk, Jan Alexander Steffens (heftig), Jan Engelhardt, Jan Synacek,
1060 Joost Bremmer, Jorgen Schaefer, Karel Zak, Klearchos Chaloulos,
1061 lc85446, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel
1062 Holtmann, Martin Pitt, Michael Biebl, Michael Olbrich, Michael Scherer,
1063 Michał Górny, Michal Sekletar, Nicolas Cornu, Nicolas Iooss, Nils
1064 Carlson, nmartensen, nnz1024, Patrick Ohly, Peter Hutterer, Phillip Sz,
1065 Ronny Chevalier, Samu Kallio, Shawn Landden, Stef Walter, Susant
1066 Sahani, Sylvain Plantefève, Tadej Janež, Thomas Hindoe Paaboel
1067 Andersen, Tom Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito
1068 Caputo, WaLyong Cho, Yu Watanabe, Zbigniew Jędrzejewski-Szmek
1070 — Berlin, 2016-02-11
1074 * A number of properties previously only settable in unit
1075 files are now also available as properties to set when
1076 creating transient units programmatically via the bus, as it
1077 is exposed with systemd-run's --property=
1078 setting. Specifically, these are: SyslogIdentifier=,
1079 SyslogLevelPrefix=, TimerSlackNSec=, OOMScoreAdjust=,
1080 EnvironmentFile=, ReadWriteDirectories=,
1081 ReadOnlyDirectories=, InaccessibleDirectories=,
1082 ProtectSystem=, ProtectHome=, RuntimeDirectory=.
1084 * When creating transient services via the bus API it is now
1085 possible to pass in a set of file descriptors to use as
1086 STDIN/STDOUT/STDERR for the invoked process.
1088 * Slice units may now be created transiently via the bus APIs,
1089 similar to the way service and scope units may already be
1090 created transiently.
1092 * Wherever systemd expects a calendar timestamp specification
1093 (like in journalctl's --since= and --until= switches) UTC
1094 timestamps are now supported. Timestamps suffixed with "UTC"
1095 are now considered to be in Universal Time Coordinated
1096 instead of the local timezone. Also, timestamps may now
1097 optionally be specified with sub-second accuracy. Both of
1098 these additions also apply to recurring calendar event
1099 specification, such as OnCalendar= in timer units.
1101 * journalctl gained a new "--sync" switch that asks the
1102 journal daemon to write all so far unwritten log messages to
1103 disk and sync the files, before returning.
1105 * systemd-tmpfiles learned two new line types "q" and "Q" that
1106 operate like "v", but also set up a basic btrfs quota
1107 hierarchy when used on a btrfs file system with quota
1110 * tmpfiles' "v", "q" and "Q" will now create a plain directory
1111 instead of a subvolume (even on a btrfs file system) if the
1112 root directory is a plain directory, and not a
1113 subvolume. This should simplify things with certain chroot()
1114 environments which are not aware of the concept of btrfs
1117 * systemd-detect-virt gained a new --chroot switch to detect
1118 whether execution takes place in a chroot() environment.
1120 * CPUAffinity= now takes CPU index ranges in addition to
1123 * The various memory-related resource limit settings (such as
1124 LimitAS=) now understand the usual K, M, G, ... suffixes to
1125 the base of 1024 (IEC). Similar, the time-related resource
1126 limit settings understand the usual min, h, day, ...
1129 * There's a new system.conf setting DefaultTasksMax= to
1130 control the default TasksMax= setting for services and
1131 scopes running on the system. (TasksMax= is the primary
1132 setting that exposes the "pids" cgroup controller on systemd
1133 and was introduced in the previous systemd release.) The
1134 setting now defaults to 512, which means services that are
1135 not explicitly configured otherwise will only be able to
1136 create 512 processes or threads at maximum, from this
1137 version on. Note that this means that thread- or
1138 process-heavy services might need to be reconfigured to set
1139 TasksMax= to a higher value. It is sufficient to set
1140 TasksMax= in these specific unit files to a higher value, or
1141 even "infinity". Similar, there's now a logind.conf setting
1142 UserTasksMax= that defaults to 4096 and limits the total
1143 number of processes or tasks each user may own
1144 concurrently. nspawn containers also have the TasksMax=
1145 value set by default now, to 8192. Note that all of this
1146 only has an effect if the "pids" cgroup controller is
1147 enabled in the kernel. The general benefit of these changes
1148 should be a more robust and safer system, that provides a
1149 certain amount of per-service fork() bomb protection.
1151 * systemd-nspawn gained the new --network-veth-extra= switch
1152 to define additional and arbitrarily-named virtual Ethernet
1153 links between the host and the container.
1155 * A new service execution setting PassEnvironment= has been
1156 added that allows importing select environment variables
1157 from PID1's environment block into the environment block of
1160 * Timer units gained support for a new RemainAfterElapse=
1161 setting which takes a boolean argument. It defaults to on,
1162 exposing behaviour unchanged to previous releases. If set to
1163 off, timer units are unloaded after they elapsed if they
1164 cannot elapse again. This is particularly useful for
1165 transient timer units, which shall not stay around longer
1166 than until they first elapse.
1168 * systemd will now bump the net.unix.max_dgram_qlen to 512 by
1169 default now (the kernel default is 16). This is beneficial
1170 for avoiding blocking on AF_UNIX/SOCK_DGRAM sockets since it
1171 allows substantially larger numbers of queued
1172 datagrams. This should increase the capability of systemd to
1173 parallelize boot-up, as logging and sd_notify() are unlikely
1174 to stall execution anymore. If you need to change the value
1175 from the new defaults, use the usual sysctl.d/ snippets.
1177 * The compression framing format used by the journal or
1178 coredump processing has changed to be in line with what the
1179 official LZ4 tools generate. LZ4 compression support in
1180 systemd was considered unsupported previously, as the format
1181 was not compatible with the normal tools. With this release
1182 this has changed now, and it is hence safe for downstream
1183 distributions to turn it on. While not compressing as well
1184 as the XZ, LZ4 is substantially faster, which makes
1185 it a good default choice for the compression logic in the
1186 journal and in coredump handling.
1188 * Any reference to /etc/mtab has been dropped from
1189 systemd. The file has been obsolete since a while, but
1190 systemd refused to work on systems where it was incorrectly
1191 set up (it should be a symlink or non-existent). Please make
1192 sure to update to util-linux 2.27.1 or newer in conjunction
1193 with this systemd release, which also drops any reference to
1194 /etc/mtab. If you maintain a distribution make sure that no
1195 software you package still references it, as this is a
1196 likely source of bugs. There's also a glibc bug pending,
1197 asking for removal of any reference to this obsolete file:
1199 https://sourceware.org/bugzilla/show_bug.cgi?id=19108
1201 Note that only util-linux versions built with
1202 --enable-libmount-force-mountinfo are supported.
1204 * Support for the ".snapshot" unit type has been removed. This
1205 feature turned out to be little useful and little used, and
1206 has now been removed from the core and from systemctl.
1208 * The dependency types RequiresOverridable= and
1209 RequisiteOverridable= have been removed from systemd. They
1210 have been used only very sparingly to our knowledge and
1211 other options that provide a similar effect (such as
1212 systemctl --mode=ignore-dependencies) are much more useful
1213 and commonly used. Moreover, they were only half-way
1214 implemented as the option to control behaviour regarding
1215 these dependencies was never added to systemctl. By removing
1216 these dependency types the execution engine becomes a bit
1217 simpler. Unit files that use these dependencies should be
1218 changed to use the non-Overridable dependency types
1219 instead. In fact, when parsing unit files with these
1220 options, that's what systemd will automatically convert them
1221 too, but it will also warn, asking users to fix the unit
1222 files accordingly. Removal of these dependency types should
1223 only affect a negligible number of unit files in the wild.
1225 * Behaviour of networkd's IPForward= option changed
1226 (again). It will no longer maintain a per-interface setting,
1227 but propagate one way from interfaces where this is enabled
1228 to the global kernel setting. The global setting will be
1229 enabled when requested by a network that is set up, but
1230 never be disabled again. This change was made to make sure
1231 IPv4 and IPv6 behaviour regarding packet forwarding is
1232 similar (as the Linux IPv6 stack does not support
1233 per-interface control of this setting) and to minimize
1236 * In unit files the behaviour of %u, %U, %h, %s has
1237 changed. These specifiers will now unconditionally resolve
1238 to the various user database fields of the user that the
1239 systemd instance is running as, instead of the user
1240 configured in the specific unit via User=. Note that this
1241 effectively doesn't change much, as resolving of these
1242 specifiers was already turned off in the --system instance
1243 of systemd, as we cannot do NSS lookups from PID 1. In the
1244 --user instance of systemd these specifiers where correctly
1245 resolved, but hardly made any sense, since the user instance
1246 lacks privileges to do user switches anyway, and User= is
1247 hence useless. Moreover, even in the --user instance of
1248 systemd behaviour was awkward as it would only take settings
1249 from User= assignment placed before the specifier into
1250 account. In order to unify and simplify the logic around
1251 this the specifiers will now always resolve to the
1252 credentials of the user invoking the manager (which in case
1253 of PID 1 is the root user).
1255 Contributions from: Andrew Jones, Beniamino Galvani, Boyuan
1256 Yang, Daniel Machon, Daniel Mack, David Herrmann, David
1257 Reynolds, David Strauss, Dongsu Park, Evgeny Vereshchagin,
1258 Felipe Sateler, Filipe Brandenburger, Franck Bui, Hristo
1259 Venev, Iago López Galeiras, Jan Engelhardt, Jan Janssen, Jan
1260 Synacek, Jesus Ornelas Aguayo, Karel Zak, kayrus, Kay Sievers,
1261 Lennart Poettering, Liu Yuan Yuan, Mantas Mikulėnas, Marcel
1262 Holtmann, Marcin Bachry, Marcos Alano, Marcos Mello, Mark
1263 Theunissen, Martin Pitt, Michael Marineau, Michael Olbrich,
1264 Michal Schmidt, Michal Sekletar, Mirco Tischler, Nick Owens,
1265 Nicolas Cornu, Patrik Flykt, Peter Hutterer, reverendhomer,
1266 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Shawn Landden,
1267 Susant Sahani, Thomas Haller, Thomas Hindoe Paaboel Andersen,
1268 Tom Gundersen, Torstein Husebø, Vito Caputo, Zbigniew
1271 — Berlin, 2015-11-18
1275 * systemd now depends on util-linux v2.27. More specifically,
1276 the newly added mount monitor feature in libmount now
1277 replaces systemd's former own implementation.
1279 * libmount mandates /etc/mtab not to be regular file, and
1280 systemd now enforces this condition at early boot.
1281 /etc/mtab has been deprecated and warned about for a very
1282 long time, so systems running systemd should already have
1283 stopped having this file around as anything else than a
1284 symlink to /proc/self/mounts.
1286 * Support for the "pids" cgroup controller has been added. It
1287 allows accounting the number of tasks in a cgroup and
1288 enforcing limits on it. This adds two new setting
1289 TasksAccounting= and TasksMax= to each unit, as well as a
1290 global option DefaultTasksAccounting=.
1292 * Support for the "net_cls" cgroup controller has been added.
1293 It allows assigning a net class ID to each task in the
1294 cgroup, which can then be used in firewall rules and traffic
1295 shaping configurations. Note that the kernel netfilter net
1296 class code does not currently work reliably for ingress
1297 packets on unestablished sockets.
1299 This adds a new config directive called NetClass= to CGroup
1300 enabled units. Allowed values are positive numbers for fixed
1301 assignments and "auto" for picking a free value
1304 * 'systemctl is-system-running' now returns 'offline' if the
1305 system is not booted with systemd. This command can now be
1306 used as a substitute for 'systemd-notify --booted'.
1308 * Watchdog timeouts have been increased to 3 minutes for all
1309 in-tree service files. Apparently, disk IO issues are more
1310 frequent than we hoped, and user reported >1 minute waiting
1313 * 'machine-id-commit' functionality has been merged into
1314 'machine-id-setup --commit'. The separate binary has been
1317 * The WorkingDirectory= directive in unit files may now be set
1318 to the special value '~'. In this case, the working
1319 directory is set to the home directory of the user
1320 configured in User=.
1322 * "machinectl shell" will now open the shell in the home
1323 directory of the selected user by default.
1325 * The CrashChVT= configuration file setting is renamed to
1326 CrashChangeVT=, following our usual logic of not
1327 abbreviating unnecessarily. The old directive is still
1328 supported for compat reasons. Also, this directive now takes
1329 an integer value between 1 and 63, or a boolean value. The
1330 formerly supported '-1' value for disabling stays around for
1333 * The PrivateTmp=, PrivateDevices=, PrivateNetwork=,
1334 NoNewPrivileges=, TTYPath=, WorkingDirectory= and
1335 RootDirectory= properties can now be set for transient
1338 * The systemd-analyze tool gained a new "set-log-target" verb
1339 to change the logging target the system manager logs to
1340 dynamically during runtime. This is similar to how
1341 "systemd-analyze set-log-level" already changes the log
1344 * In nspawn /sys is now mounted as tmpfs, with only a selected
1345 set of subdirectories mounted in from the real sysfs. This
1346 enhances security slightly, and is useful for ensuring user
1347 namespaces work correctly.
1349 * Support for USB FunctionFS activation has been added. This
1350 allows implementation of USB gadget services that are
1351 activated as soon as they are requested, so that they don't
1352 have to run continuously, similar to classic socket
1355 * The "systemctl exit" command now optionally takes an
1356 additional parameter that sets the exit code to return from
1357 the systemd manager when exiting. This is only relevant when
1358 running the systemd user instance, or when running the
1359 system instance in a container.
1361 * sd-bus gained the new API calls sd_bus_path_encode_many()
1362 and sd_bus_path_decode_many() that allow easy encoding and
1363 decoding of multiple identifier strings inside a D-Bus
1364 object path. Another new call sd_bus_default_flush_close()
1365 has been added to flush and close per-thread default
1368 * systemd-cgtop gained support for a -M/--machine= switch to
1369 show the control groups within a certain container only.
1371 * "systemctl kill" gained support for an optional --fail
1372 switch. If specified the requested operation will fail of no
1373 processes have been killed, because the unit had no
1374 processes attached, or similar.
1376 * A new systemd.crash_reboot=1 kernel command line option has
1377 been added that triggers a reboot after crashing. This can
1378 also be set through CrashReboot= in systemd.conf.
1380 * The RuntimeDirectory= setting now understands unit
1381 specifiers like %i or %f.
1383 * A new (still internal) library API sd-ipv4acd has been added,
1384 that implements address conflict detection for IPv4. It's
1385 based on code from sd-ipv4ll, and will be useful for
1386 detecting DHCP address conflicts.
1388 * File descriptors passed during socket activation may now be
1389 named. A new API sd_listen_fds_with_names() is added to
1390 access the names. The default names may be overridden,
1391 either in the .socket file using the FileDescriptorName=
1392 parameter, or by passing FDNAME= when storing the file
1393 descriptors using sd_notify().
1395 * systemd-networkd gained support for:
1397 - Setting the IPv6 Router Advertisement settings via
1398 IPv6AcceptRouterAdvertisements= in .network files.
1400 - Configuring the HelloTimeSec=, MaxAgeSec= and
1401 ForwardDelaySec= bridge parameters in .netdev files.
1403 - Configuring PreferredSource= for static routes in
1406 * The "ask-password" framework used to query for LUKS harddisk
1407 passwords or SSL passwords during boot gained support for
1408 caching passwords in the kernel keyring, if it is
1409 available. This makes sure that the user only has to type in
1410 a passphrase once if there are multiple objects to unlock
1411 with the same one. Previously, such password caching was
1412 available only when Plymouth was used; this moves the
1413 caching logic into the systemd codebase itself. The
1414 "systemd-ask-password" utility gained a new --keyname=
1415 switch to control which kernel keyring key to use for
1416 caching a password in. This functionality is also useful for
1417 enabling display managers such as gdm to automatically
1418 unlock the user's GNOME keyring if its passphrase, the
1419 user's password and the harddisk password are the same, if
1420 gdm-autologin is used.
1422 * When downloading tar or raw images using "machinectl
1423 pull-tar" or "machinectl pull-raw", a matching ".nspawn"
1424 file is now also downloaded, if it is available and stored
1425 next to the image file.
1427 * Units of type ".socket" gained a new boolean setting
1428 Writable= which is only useful in conjunction with
1429 ListenSpecial=. If true, enables opening the specified
1430 special file in O_RDWR mode rather than O_RDONLY mode.
1432 * systemd-rfkill has been reworked to become a singleton
1433 service that is activated through /dev/rfkill on each rfkill
1434 state change and saves the settings to disk. This way,
1435 systemd-rfkill is now compatible with devices that exist
1436 only intermittendly, and even restores state if the previous
1437 system shutdown was abrupt rather than clean.
1439 * The journal daemon gained support for vacuuming old journal
1440 files controlled by the number of files that shall remain,
1441 in addition to the already existing control by size and by
1442 date. This is useful as journal interleaving performance
1443 degrades with too many separate journal files, and allows
1444 putting an effective limit on them. The new setting defaults
1445 to 100, but this may be changed by setting SystemMaxFiles=
1446 and RuntimeMaxFiles= in journald.conf. Also, the
1447 "journalctl" tool gained the new --vacuum-files= switch to
1448 manually vacuum journal files to leave only the specified
1449 number of files in place.
1451 * udev will now create /dev/disk/by-path links for ATA devices
1452 on kernels where that is supported.
1454 * Galician, Serbian, Turkish and Korean translations were added.
1456 Contributions from: Aaro Koskinen, Alban Crequy, Beniamino
1457 Galvani, Benjamin Robin, Branislav Blaskovic, Chen-Han Hsiao
1458 (Stanley), Daniel Buch, Daniel Machon, Daniel Mack, David
1459 Herrmann, David Milburn, doubleodoug, Evgeny Vereshchagin,
1460 Felipe Franciosi, Filipe Brandenburger, Fran Dieguez, Gabriel
1461 de Perthuis, Georg Müller, Hans de Goede, Hendrik Brueckner,
1462 Ivan Shapovalov, Jacob Keller, Jan Engelhardt, Jan Janssen,
1463 Jan Synacek, Jens Kuske, Karel Zak, Kay Sievers, Krzesimir
1464 Nowak, Krzysztof Kotlenga, Lars Uebernickel, Lennart
1465 Poettering, Lukas Nykryn, Łukasz Stelmach, Maciej Wereski,
1466 Marcel Holtmann, Marius Thesing, Martin Pitt, Michael Biebl,
1467 Michael Gebetsroither, Michal Schmidt, Michal Sekletar, Mike
1468 Gilbert, Muhammet Kara, nazgul77, Nicolas Cornu, NoXPhasma,
1469 Olof Johansson, Patrik Flykt, Pawel Szewczyk, reverendhomer,
1470 Ronny Chevalier, Sangjung Woo, Seong-ho Cho, Susant Sahani,
1471 Sylvain Plantefève, Thomas Haller, Thomas Hindoe Paaboel
1472 Andersen, Tom Gundersen, Tom Lyon, Viktar Vauchkevich,
1473 Zbigniew Jędrzejewski-Szmek, Марко М. Костић
1475 — Berlin, 2015-10-07
1479 * The DHCP implementation of systemd-networkd gained a set of
1482 - The DHCP server now supports emitting DNS and NTP
1483 information. It may be enabled and configured via
1484 EmitDNS=, DNS=, EmitNTP=, and NTP=. If transmission of DNS
1485 and NTP information is enabled, but no servers are
1486 configured, the corresponding uplink information (if there
1487 is any) is propagated.
1489 - Server and client now support transmission and reception
1490 of timezone information. It can be configured via the
1491 newly introduced network options UseTimezone=,
1492 EmitTimezone=, and Timezone=. Transmission of timezone
1493 information is enabled between host and containers by
1494 default now: the container will change its local timezone
1495 to what the host has set.
1497 - Lease timeouts can now be configured via
1498 MaxLeaseTimeSec= and DefaultLeaseTimeSec=.
1500 - The DHCP server improved on the stability of
1501 leases. Clients are more likely to get the same lease
1502 information back, even if the server loses state.
1504 - The DHCP server supports two new configuration options to
1505 control the lease address pool metrics, PoolOffset= and
1508 * The encapsulation limit of tunnels in systemd-networkd may
1509 now be configured via 'EncapsulationLimit='. It allows
1510 modifying the maximum additional levels of encapsulation
1511 that are permitted to be prepended to a packet.
1513 * systemd now supports the concept of user buses replacing
1514 session buses, if used with dbus-1.10 (and enabled via dbus
1515 --enable-user-session). It previously only supported this on
1516 kdbus-enabled systems, and this release expands this to
1517 'dbus-daemon' systems.
1519 * systemd-networkd now supports predictable interface names
1522 * systemd now optionally supports the new Linux kernel
1523 "unified" control group hierarchy. If enabled via the kernel
1524 command-line option 'systemd.unified_cgroup_hierarchy=1',
1525 systemd will try to mount the unified cgroup hierarchy
1526 directly on /sys/fs/cgroup. If not enabled, or not
1527 available, systemd will fall back to the legacy cgroup
1528 hierarchy setup, as before. Host system and containers can
1529 mix and match legacy and unified hierarchies as they
1530 wish. nspawn understands the $UNIFIED_CGROUP_HIERARCHY
1531 environment variable to individually select the hierarchy to
1532 use for executed containers. By default, nspawn will use the
1533 unified hierarchy for the containers if the host uses the
1534 unified hierarchy, and the legacy hierarchy otherwise.
1535 Please note that at this point the unified hierarchy is an
1536 experimental kernel feature and is likely to change in one
1537 of the next kernel releases. Therefore, it should not be
1538 enabled by default in downstream distributions yet. The
1539 minimum required kernel version for the unified hierarchy to
1540 work is 4.2. Note that when the unified hierarchy is used
1541 for the first time delegated access to controllers is
1542 safe. Because of this systemd-nspawn containers will get
1543 access to controllers now, as will systemd user
1544 sessions. This means containers and user sessions may now
1545 manage their own resources, partitioning up what the system
1548 * A new special scope unit "init.scope" has been introduced
1549 that encapsulates PID 1 of the system. It may be used to
1550 determine resource usage and enforce resource limits on PID
1551 1 itself. PID 1 hence moved out of the root of the control
1554 * The cgtop tool gained support for filtering out kernel
1555 threads when counting tasks in a control group. Also, the
1556 count of processes is now recursively summed up by
1557 default. Two options -k and --recursive= have been added to
1558 revert to old behaviour. The tool has also been updated to
1559 work correctly in containers now.
1561 * systemd-nspawn's --bind= and --bind-ro= options have been
1562 extended to allow creation of non-recursive bind mounts.
1564 * libsystemd gained two new calls sd_pid_get_cgroup() and
1565 sd_peer_get_cgroup() which return the control group path of
1566 a process or peer of a connected AF_UNIX socket. This
1567 function call is particularly useful when implementing
1568 delegated subtrees support in the control group hierarchy.
1570 * The "sd-event" event loop API of libsystemd now supports
1571 correct dequeuing of real-time signals, without losing
1574 * When systemd requests a PolicyKit decision when managing
1575 units it will now add additional fields to the request,
1576 including unit name and desired operation. This enables more
1577 powerful PolicyKit policies, that make decisions depending
1578 on these parameters.
1580 * nspawn learnt support for .nspawn settings files, that may
1581 accompany the image files or directories of containers, and
1582 may contain additional settings for the container. This is
1583 an alternative to configuring container parameters via the
1584 nspawn command line.
1586 Contributions from: Cristian Rodríguez, Daniel Mack, David
1587 Herrmann, Eugene Yakubovich, Evgeny Vereshchagin, Filipe
1588 Brandenburger, Hans de Goede, Jan Alexander Steffens, Jan
1589 Synacek, Kay Sievers, Lennart Poettering, Mangix, Marcel
1590 Holtmann, Martin Pitt, Michael Biebl, Michael Chapman, Michal
1591 Sekletar, Peter Hutterer, Piotr Drąg, reverendhomer, Robin
1592 Hack, Susant Sahani, Sylvain Pasche, Thomas Hindoe Paaboel
1593 Andersen, Tom Gundersen, Torstein Husebø
1595 — Berlin, 2015-09-08
1599 * machinectl gained a new verb 'shell' which opens a fresh
1600 shell on the target container or the host. It is similar to
1601 the existing 'login' command of machinectl, but spawns the
1602 shell directly without prompting for username or
1603 password. The pseudo machine '.host' now refers to the local
1604 host and is used by default. Hence, 'machinectl shell' can
1605 be used as replacement for 'su -' which spawns a session as
1606 a fresh systemd unit in a way that is fully isolated from
1607 the originating session.
1609 * systemd-networkd learned to cope with private-zone DHCP
1610 options and allows other programs to query the values.
1612 * SELinux access control when enabling/disabling units is no
1613 longer enforced with this release. The previous
1614 implementation was incorrect, and a new corrected
1615 implementation is not yet available. As unit file operations
1616 are still protected via PolicyKit and D-Bus policy this is
1617 not a security problem. Yet, distributions which care about
1618 optimal SELinux support should probably not stabilize on
1621 * sd-bus gained support for matches of type "arg0has=", that
1622 test for membership of strings in string arrays sent in bus
1625 * systemd-resolved now dumps the contents of its DNS and LLMNR
1626 caches to the logs on reception of the SIGUSR1 signal. This
1627 is useful to debug DNS behaviour.
1629 * The coredumpctl tool gained a new --directory= option to
1630 operate on journal files in a specific directory.
1632 * "systemctl reboot" and related commands gained a new
1633 "--message=" option which may be used to set a free-text
1634 wall message when shutting down or rebooting the
1635 system. This message is also logged, which is useful for
1636 figuring out the reason for a reboot or shutdown a
1639 * The "systemd-resolve-host" tool's -i switch now takes
1640 network interface numbers as alternative to interface names.
1642 * A new unit file setting for services has been introduced:
1643 UtmpMode= allows configuration of how precisely systemd
1644 handles utmp and wtmp entries for the service if this is
1645 enabled. This allows writing services that appear similar to
1646 user sessions in the output of the "w", "who", "last" and
1649 * systemd-resolved will now locally synthesize DNS resource
1650 records for the "localhost" and "gateway" domains as well as
1651 the local hostname. This should ensure that clients querying
1652 RRs via resolved will get similar results as those going via
1653 NSS, if nss-myhostname is enabled.
1655 Contributions from: Alastair Hughes, Alex Crawford, Daniel
1656 Mack, David Herrmann, Dimitri John Ledkov, Eric Kostrowski,
1657 Evgeny Vereshchagin, Felipe Sateler, HATAYAMA Daisuke, Jan
1658 Pokorný, Jan Synacek, Johnny Robeson, Karel Zak, Kay Sievers,
1659 Kefeng Wang, Lennart Poettering, Major Hayden, Marcel
1660 Holtmann, Markus Elfring, Martin Mikkelsen, Martin Pitt, Matt
1661 Turner, Maxim Mikityanskiy, Michael Biebl, Namhyung Kim,
1662 Nicolas Cornu, Owen W. Taylor, Patrik Flykt, Peter Hutterer,
1663 reverendhomer, Richard Maw, Ronny Chevalier, Seth Jennings,
1664 Stef Walter, Susant Sahani, Thomas Blume, Thomas Hindoe
1665 Paaboel Andersen, Thomas Meyer, Tom Gundersen, Vincent Batts,
1666 WaLyong Cho, Zbigniew Jędrzejewski-Szmek
1668 — Berlin, 2015-08-27
1672 * The systemd-efi-boot-generator functionality was merged into
1673 systemd-gpt-auto-generator.
1675 * systemd-networkd now supports Group Policy for vxlan
1676 devices. It can be enabled via the new boolean configuration
1677 option called 'GroupPolicyExtension='.
1679 Contributions from: Andreas Kempf, Christian Hesse, Daniel Mack, David
1680 Herrmann, Herman Fries, Johannes Nixdorf, Kay Sievers, Lennart
1681 Poettering, Peter Hutterer, Susant Sahani, Tom Gundersen
1683 — Berlin, 2015-07-31
1687 * The python-systemd code has been removed from the systemd repository.
1688 A new repository has been created which accommodates the code from
1689 now on, and we kindly ask distributions to create a separate package
1690 for this: https://github.com/systemd/python-systemd
1692 * The systemd daemon will now reload its main configuration
1693 (/etc/systemd/system.conf) on daemon-reload.
1695 * sd-dhcp now exposes vendor specific extensions via
1696 sd_dhcp_lease_get_vendor_specific().
1698 * systemd-networkd gained a number of new configuration options.
1700 - A new boolean configuration option for TAP devices called
1701 'VNetHeader='. If set, the IFF_VNET_HDR flag is set for the
1702 device, thus allowing to send and receive GSO packets.
1704 - A new tunnel configuration option called 'CopyDSCP='.
1705 If enabled, the DSCP field of ip6 tunnels is copied into the
1706 decapsulated packet.
1708 - A set of boolean bridge configuration options were added.
1709 'UseBPDU=', 'HairPin=', 'FastLeave=', 'AllowPortToBeRoot=',
1710 and 'UnicastFlood=' are now parsed by networkd and applied to the
1711 respective bridge link device via the respective IFLA_BRPORT_*
1714 - A new string configuration option to override the hostname sent
1715 to a DHCP server, called 'Hostname='. If set and 'SendHostname='
1716 is true, networkd will use the configured hostname instead of the
1717 system hostname when sending DHCP requests.
1719 - A new tunnel configuration option called 'IPv6FlowLabel='. If set,
1720 networkd will configure the IPv6 flow-label of the tunnel device
1721 according to RFC2460.
1723 - The 'macvtap' virtual network devices are now supported, similar to
1724 the already supported 'macvlan' devices.
1726 * systemd-resolved now implements RFC5452 to improve resilience against
1727 cache poisoning. Additionally, source port randomization is enabled
1728 by default to further protect against DNS spoofing attacks.
1730 * nss-mymachines now supports translating UIDs and GIDs of running
1731 containers with user-namespaces enabled. If a container 'foo'
1732 translates a host uid 'UID' to the container uid 'TUID', then
1733 nss-mymachines will also map uid 'UID' to/from username 'vu-foo-TUID'
1734 (with 'foo' and 'TUID' replaced accordingly). Similarly, groups are
1735 mapped as 'vg-foo-TGID'.
1737 Contributions from: Beniamino Galvani, cee1, Christian Hesse, Daniel
1738 Buch, Daniel Mack, daurnimator, David Herrmann, Dimitri John Ledkov,
1739 HATAYAMA Daisuke, Ivan Shapovalov, Jan Alexander Steffens (heftig),
1740 Johan Ouwerkerk, Jose Carlos Venegas Munoz, Karel Zak, Kay Sievers,
1741 Lennart Poettering, Lidong Zhong, Martin Pitt, Michael Biebl, Michael
1742 Olbrich, Michal Schmidt, Michal Sekletar, Mike Gilbert, Namhyung Kim,
1743 Nick Owens, Peter Hutterer, Richard Maw, Steven Allen, Sungbae Yoo,
1744 Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel Andersen, Tom
1745 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Vito Caputo,
1746 Vivenzio Pagliari, Zbigniew Jędrzejewski-Szmek
1748 — Berlin, 2015-07-29
1752 * udev does not longer support the WAIT_FOR_SYSFS= key in udev rules.
1753 There are no known issues with current sysfs, and udev does not need
1754 or should be used to work around such bugs.
1756 * udev does no longer enable USB HID power management. Several reports
1757 indicate, that some devices cannot handle that setting.
1759 * The udev accelerometer helper was removed. The functionality
1760 is now fully included in iio-sensor-proxy. But this means,
1761 older iio-sensor-proxy versions will no longer provide
1762 accelerometer/orientation data with this systemd version.
1763 Please upgrade iio-sensor-proxy to version 1.0.
1765 * networkd gained a new configuration option IPv6PrivacyExtensions=
1766 which enables IPv6 privacy extensions (RFC 4941, "Privacy Extensions
1767 for Stateless Address") on selected networks.
1769 * For the sake of fewer build-time dependencies and less code in the
1770 main repository, the python bindings are about to be removed in the
1771 next release. A new repository has been created which accommodates
1772 the code from now on, and we kindly ask distributions to create a
1773 separate package for this. The removal will take place in v223.
1775 https://github.com/systemd/python-systemd
1777 Contributions from: Abdo Roig-Maranges, Andrew Eikum, Bastien Nocera,
1778 Cédric Delmas, Christian Hesse, Christos Trochalakis, Daniel Mack,
1779 daurnimator, David Herrmann, Dimitri John Ledkov, Eric Biggers, Eric
1780 Cook, Felipe Sateler, Geert Jansen, Gerd Hoffmann, Gianpaolo Macario,
1781 Greg Kroah-Hartman, Iago López Galeiras, Jan Alexander Steffens
1782 (heftig), Jan Engelhardt, Jay Strict, Kay Sievers, Lennart Poettering,
1783 Markus Knetschke, Martin Pitt, Michael Biebl, Michael Marineau, Michal
1784 Sekletar, Miguel Bernal Marin, Peter Hutterer, Richard Maw, rinrinne,
1785 Susant Sahani, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
1786 Husebø, Vedran Miletić, WaLyong Cho, Zbigniew Jędrzejewski-Szmek
1788 — Berlin, 2015-07-07
1792 * The sd-bus.h and sd-event.h APIs have now been declared
1793 stable and have been added to the official interface of
1794 libsystemd.so. sd-bus implements an alternative D-Bus client
1795 library, that is relatively easy to use, very efficient and
1796 supports both classic D-Bus as well as kdbus as transport
1797 backend. sd-event is a generic event loop abstraction that
1798 is built around Linux epoll, but adds features such as event
1799 prioritization or efficient timer handling. Both APIs are good
1800 choices for C programs looking for a bus and/or event loop
1801 implementation that is minimal and does not have to be
1802 portable to other kernels.
1804 * kdbus support is no longer compile-time optional. It is now
1805 always built-in. However, it can still be disabled at
1806 runtime using the kdbus=0 kernel command line setting, and
1807 that setting may be changed to default to off, by specifying
1808 --disable-kdbus at build-time. Note though that the kernel
1809 command line setting has no effect if the kdbus.ko kernel
1810 module is not installed, in which case kdbus is (obviously)
1811 also disabled. We encourage all downstream distributions to
1812 begin testing kdbus by adding it to the kernel images in the
1813 development distributions, and leaving kdbus support in
1816 * The minimal required util-linux version has been bumped to
1819 * Support for chkconfig (--enable-chkconfig) was removed in
1820 favor of calling an abstraction tool
1821 /lib/systemd/systemd-sysv-install. This needs to be
1822 implemented for your distribution. See "SYSV INIT.D SCRIPTS"
1823 in README for details.
1825 * If there's a systemd unit and a SysV init script for the
1826 same service name, and the user executes "systemctl enable"
1827 for it (or a related call), then this will now enable both
1828 (or execute the related operation on both), not just the
1831 * The libudev API documentation has been converted from gtkdoc
1834 * gudev has been removed from the systemd tree, it is now an
1837 * The systemd-cgtop tool learnt a new --raw switch to generate
1838 "raw" (machine parsable) output.
1840 * networkd's IPForwarding= .network file setting learnt the
1841 new setting "kernel", which ensures that networkd does not
1842 change the IP forwarding sysctl from the default kernel
1845 * The systemd-logind bus API now exposes a new boolean
1846 property "Docked" that reports whether logind considers the
1847 system "docked", i.e. connected to a docking station or not.
1849 Contributions from: Alex Crawford, Andreas Pokorny, Andrei
1850 Borzenkov, Charles Duffy, Colin Guthrie, Cristian Rodríguez,
1851 Daniele Medri, Daniel Hahler, Daniel Mack, David Herrmann,
1852 David Mohr, Dimitri John Ledkov, Djalal Harouni, dslul, Ed
1853 Swierk, Eric Cook, Filipe Brandenburger, Gianpaolo Macario,
1854 Harald Hoyer, Iago López Galeiras, Igor Vuk, Jan Synacek,
1855 Jason Pleau, Jason S. McMullan, Jean Delvare, Jeff Huang,
1856 Jonathan Boulle, Karel Zak, Kay Sievers, kloun, Lennart
1857 Poettering, Marc-Antoine Perennou, Marcel Holtmann, Mario
1858 Limonciello, Martin Pitt, Michael Biebl, Michael Olbrich,
1859 Michal Schmidt, Mike Gilbert, Nick Owens, Pablo Lezaeta Reyes,
1860 Patrick Donnelly, Pavel Odvody, Peter Hutterer, Philip
1861 Withnall, Ronny Chevalier, Simon McVittie, Susant Sahani,
1862 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Torstein
1863 Husebø, Umut Tezduyar Lindskog, Viktar Vauchkevich, Werner
1864 Fink, Zbigniew Jędrzejewski-Szmek
1866 — Berlin, 2015-06-19
1870 * The gudev library has been extracted into a separate repository
1871 available at: https://git.gnome.org/browse/libgudev/
1872 It is now managed as part of the Gnome project. Distributions
1873 are recommended to pass --disable-gudev to systemd and use
1874 gudev from the Gnome project instead. gudev is still included
1875 in systemd, for now. It will be removed soon, though. Please
1876 also see the announcement-thread on systemd-devel:
1877 http://lists.freedesktop.org/archives/systemd-devel/2015-May/032070.html
1879 * systemd now exposes a CPUUsageNSec= property for each
1880 service unit on the bus, that contains the overall consumed
1881 CPU time of a service (the sum of what each process of the
1882 service consumed). This value is only available if
1883 CPUAccounting= is turned on for a service, and is then shown
1884 in the "systemctl status" output.
1886 * Support for configuring alternative mappings of the old SysV
1887 runlevels to systemd targets has been removed. They are now
1888 hardcoded in a way that runlevels 2, 3, 4 all map to
1889 multi-user.target and 5 to graphical.target (which
1890 previously was already the default behaviour).
1892 * The auto-mounter logic gained support for mount point
1893 expiry, using a new TimeoutIdleSec= setting in .automount
1894 units. (Also available as x-systemd.idle-timeout= in /etc/fstab).
1896 * The EFI System Partition (ESP) as mounted to /boot by
1897 systemd-efi-boot-generator will now be unmounted
1898 automatically after 2 minutes of not being used. This should
1899 minimize the risk of ESP corruptions.
1901 * New /etc/fstab options x-systemd.requires= and
1902 x-systemd.requires-mounts-for= are now supported to express
1903 additional dependencies for mounts. This is useful for
1904 journalling file systems that support external journal
1905 devices or overlay file systems that require underlying file
1906 systems to be mounted.
1908 * systemd does not support direct live-upgrades (via systemctl
1909 daemon-reexec) from versions older than v44 anymore. As no
1910 distribution we are aware of shipped such old versions in a
1911 stable release this should not be problematic.
1913 * When systemd forks off a new per-connection service instance
1914 it will now set the $REMOTE_ADDR environment variable to the
1915 remote IP address, and $REMOTE_PORT environment variable to
1916 the remote IP port. This behaviour is similar to the
1917 corresponding environment variables defined by CGI.
1919 * systemd-networkd gained support for uplink failure
1920 detection. The BindCarrier= option allows binding interface
1921 configuration dynamically to the link sense of other
1922 interfaces. This is useful to achieve behaviour like in
1925 * systemd-networkd gained support for configuring the DHCP
1926 client identifier to use when requesting leases.
1928 * systemd-networkd now has a per-network UseNTP= option to
1929 configure whether NTP server information acquired via DHCP
1930 is passed on to services like systemd-timesyncd.
1932 * systemd-networkd gained support for vti6 tunnels.
1934 * Note that systemd-networkd manages the sysctl variable
1935 /proc/sys/net/ipv[46]/conf/*/forwarding for each interface
1936 it is configured for since v219. The variable controls IP
1937 forwarding, and is a per-interface alternative to the global
1938 /proc/sys/net/ipv[46]/ip_forward. This setting is
1939 configurable in the IPForward= option, which defaults to
1940 "no". This means if networkd is used for an interface it is
1941 no longer sufficient to set the global sysctl option to turn
1942 on IP forwarding! Instead, the .network file option
1943 IPForward= needs to be turned on! Note that the
1944 implementation of this behaviour was broken in v219 and has
1947 * Many bonding and vxlan options are now configurable in
1950 * systemd-nspawn gained a new --property= setting to set unit
1951 properties for the container scope. This is useful for
1952 setting resource parameters (e.g. "CPUShares=500") on
1953 containers started from the command line.
1955 * systemd-nspawn gained a new --private-users= switch to make
1956 use of user namespacing available on recent Linux kernels.
1958 * systemd-nspawn may now be called as part of a shell pipeline
1959 in which case the pipes used for stdin and stdout are passed
1960 directly to the process invoked in the container, without
1961 indirection via a pseudo tty.
1963 * systemd-nspawn gained a new switch to control the UNIX
1964 signal to use when killing the init process of the container
1967 * systemd-nspawn gained a new --overlay= switch for mounting
1968 overlay file systems into the container using the new kernel
1971 * When a container image is imported via systemd-importd and
1972 the host file system is not btrfs, a loopback block device
1973 file is created in /var/lib/machines.raw with a btrfs file
1974 system inside. It is then mounted to /var/lib/machines to
1975 enable btrfs features for container management. The loopback
1976 file and btrfs file system is grown as needed when container
1977 images are imported via systemd-importd.
1979 * systemd-machined/systemd-importd gained support for btrfs
1980 quota, to enforce container disk space limits on disk. This
1981 is exposed in "machinectl set-limit".
1983 * systemd-importd now can import containers from local .tar,
1984 .raw and .qcow2 images, and export them to .tar and .raw. It
1985 can also import dkr v2 images now from the network (on top
1988 * systemd-importd gained support for verifying downloaded
1989 images with gpg2 (previously only gpg1 was supported).
1991 * systemd-machined, systemd-logind, systemd: most bus calls
1992 are now accessible to unprivileged processes via
1993 PolicyKit. Also, systemd-logind will now allow users to kill
1994 their own sessions without further privileges or
1997 * systemd-shutdownd has been removed. This service was
1998 previously responsible for implementing scheduled shutdowns
1999 as exposed in /usr/bin/shutdown's time parameter. This
2000 functionality has now been moved into systemd-logind and is
2001 accessible via a bus interface.
2003 * "systemctl reboot" gained a new switch --firmware-setup that
2004 can be used to reboot into the EFI firmware setup, if that
2005 is available. systemd-logind now exposes an API on the bus
2006 to trigger such reboots, in case graphical desktop UIs want
2007 to cover this functionality.
2009 * "systemctl enable", "systemctl disable" and "systemctl mask"
2010 now support a new "--now" switch. If specified the units
2011 that are enabled will also be started, and the ones
2012 disabled/masked also stopped.
2014 * The Gummiboot EFI boot loader tool has been merged into
2015 systemd, and renamed to "systemd-boot". The bootctl tool has been
2016 updated to support systemd-boot.
2018 * An EFI kernel stub has been added that may be used to create
2019 kernel EFI binaries that contain not only the actual kernel,
2020 but also an initrd, boot splash, command line and OS release
2021 information. This combined binary can then be signed as a
2022 single image, so that the firmware can verify it all in one
2023 step. systemd-boot has special support for EFI binaries created
2024 like this and can extract OS release information from them
2025 and show them in the boot menu. This functionality is useful
2026 to implement cryptographically verified boot schemes.
2028 * Optional support has been added to systemd-fsck to pass
2029 fsck's progress report to an AF_UNIX socket in the file
2032 * udev will no longer create device symlinks for all block
2033 devices by default. A blacklist for excluding special block
2034 devices from this logic has been turned into a whitelist
2035 that requires picking block devices explicitly that require
2038 * A new (currently still internal) API sd-device.h has been
2039 added to libsystemd. This modernized API is supposed to
2040 replace libudev eventually. In fact, already much of libudev
2041 is now just a wrapper around sd-device.h.
2043 * A new hwdb database for storing metadata about pointing
2044 stick devices has been added.
2046 * systemd-tmpfiles gained support for setting file attributes
2047 similar to the "chattr" tool with new 'h' and 'H' lines.
2049 * systemd-journald will no longer unconditionally set the
2050 btrfs NOCOW flag on new journal files. This is instead done
2051 with tmpfiles snippet using the new 'h' line type. This
2052 allows easy disabling of this logic, by masking the
2053 journal-nocow.conf tmpfiles file.
2055 * systemd-journald will now translate audit message types to
2056 human readable identifiers when writing them to the
2057 journal. This should improve readability of audit messages.
2059 * The LUKS logic gained support for the offset= and skip=
2060 options in /etc/crypttab, as previously implemented by
2063 * /usr/lib/os-release gained a new optional field VARIANT= for
2064 distributions that support multiple variants (such as a
2065 desktop edition, a server edition, ...)
2067 Contributions from: Aaro Koskinen, Adam Goode, Alban Crequy,
2068 Alberto Fanjul Alonso, Alexander Sverdlin, Alex Puchades, Alin
2069 Rauta, Alison Chaiken, Andrew Jones, Arend van Spriel,
2070 Benedikt Morbach, Benjamin Franzke, Benjamin Tissoires, Blaž
2071 Tomažič, Chris Morgan, Chris Morin, Colin Walters, Cristian
2072 Rodríguez, Daniel Buch, Daniel Drake, Daniele Medri, Daniel
2073 Mack, Daniel Mustieles, daurnimator, Davide Bettio, David
2074 Herrmann, David Strauss, Didier Roche, Dimitri John Ledkov,
2075 Eric Cook, Gavin Li, Goffredo Baroncelli, Hannes Reinecke,
2076 Hans de Goede, Hans-Peter Deifel, Harald Hoyer, Iago López
2077 Galeiras, Ivan Shapovalov, Jan Engelhardt, Jan Janssen, Jan
2078 Pazdziora, Jan Synacek, Jasper St. Pierre, Jay Faulkner, John
2079 Paul Adrian Glaubitz, Jonathon Gilbert, Karel Zak, Kay
2080 Sievers, Koen Kooi, Lennart Poettering, Lubomir Rintel, Lucas
2081 De Marchi, Lukas Nykryn, Lukas Rusak, Lukasz Skalski, Łukasz
2082 Stelmach, Mantas Mikulėnas, Marc-Antoine Perennou, Marcel
2083 Holtmann, Martin Pitt, Mathieu Chevrier, Matthew Garrett,
2084 Michael Biebl, Michael Marineau, Michael Olbrich, Michal
2085 Schmidt, Michal Sekletar, Mirco Tischler, Nir Soffer, Patrik
2086 Flykt, Pavel Odvody, Peter Hutterer, Peter Lemenkov, Peter
2087 Waller, Piotr Drąg, Raul Gutierrez S, Richard Maw, Ronny
2088 Chevalier, Ross Burton, Sebastian Rasmussen, Sergey Ptashnick,
2089 Seth Jennings, Shawn Landden, Simon Farnsworth, Stefan Junker,
2090 Stephen Gallagher, Susant Sahani, Sylvain Plantefève, Thomas
2091 Haller, Thomas Hindoe Paaboel Andersen, Tobias Hunger, Tom
2092 Gundersen, Torstein Husebø, Umut Tezduyar Lindskog, Will
2093 Woods, Zachary Cook, Zbigniew Jędrzejewski-Szmek
2095 — Berlin, 2015-05-22
2099 * Introduce a new API "sd-hwdb.h" for querying the hardware
2100 metadata database. With this minimal interface one can query
2101 and enumerate the udev hwdb, decoupled from the old libudev
2102 library. libudev's interface for this is now only a wrapper
2103 around sd-hwdb. A new tool systemd-hwdb has been added to
2104 interface with and update the database.
2106 * When any of systemd's tools copies files (for example due to
2107 tmpfiles' C lines) a btrfs reflink will attempted first,
2108 before bytewise copying is done.
2110 * systemd-nspawn gained a new --ephemeral switch. When
2111 specified a btrfs snapshot is taken of the container's root
2112 directory, and immediately removed when the container
2113 terminates again. Thus, a container can be started whose
2114 changes never alter the container's root directory, and are
2115 lost on container termination. This switch can also be used
2116 for starting a container off the root file system of the
2117 host without affecting the host OS. This switch is only
2118 available on btrfs file systems.
2120 * systemd-nspawn gained a new --template= switch. It takes the
2121 path to a container tree to use as template for the tree
2122 specified via --directory=, should that directory be
2123 missing. This allows instantiating containers dynamically,
2124 on first run. This switch is only available on btrfs file
2127 * When a .mount unit refers to a mount point on which multiple
2128 mounts are stacked, and the .mount unit is stopped all of
2129 the stacked mount points will now be unmounted until no
2130 mount point remains.
2132 * systemd now has an explicit notion of supported and
2133 unsupported unit types. Jobs enqueued for unsupported unit
2134 types will now fail with an "unsupported" error code. More
2135 specifically .swap, .automount and .device units are not
2136 supported in containers, .busname units are not supported on
2137 non-kdbus systems. .swap and .automount are also not
2138 supported if their respective kernel compile time options
2141 * machinectl gained support for two new "copy-from" and
2142 "copy-to" commands for copying files from a running
2143 container to the host or vice versa.
2145 * machinectl gained support for a new "bind" command to bind
2146 mount host directories into local containers. This is
2147 currently only supported for nspawn containers.
2149 * networkd gained support for configuring bridge forwarding
2150 database entries (fdb) from .network files.
2152 * A new tiny daemon "systemd-importd" has been added that can
2153 download container images in tar, raw, qcow2 or dkr formats,
2154 and make them available locally in /var/lib/machines, so
2155 that they can run as nspawn containers. The daemon can GPG
2156 verify the downloads (not supported for dkr, since it has no
2157 provisions for verifying downloads). It will transparently
2158 decompress bz2, xz, gzip compressed downloads if necessary,
2159 and restore sparse files on disk. The daemon uses privilege
2160 separation to ensure the actual download logic runs with
2161 fewer privileges than the daemon itself. machinectl has
2162 gained new commands "pull-tar", "pull-raw" and "pull-dkr" to
2163 make the functionality of importd available to the
2164 user. With this in place the Fedora and Ubuntu "Cloud"
2165 images can be downloaded and booted as containers unmodified
2166 (the Fedora images lack the appropriate GPG signature files
2167 currently, so they cannot be verified, but this will change
2168 soon, hopefully). Note that downloading images is currently
2169 only fully supported on btrfs.
2171 * machinectl is now able to list container images found in
2172 /var/lib/machines, along with some metadata about sizes of
2173 disk and similar. If the directory is located on btrfs and
2174 quota is enabled, this includes quota display. A new command
2175 "image-status" has been added that shows additional
2176 information about images.
2178 * machinectl is now able to clone container images
2179 efficiently, if the underlying file system (btrfs) supports
2180 it, with the new "machinectl clone" command. It also
2181 gained commands for renaming and removing images, as well as
2182 marking them read-only or read-write (supported also on
2183 legacy file systems).
2185 * networkd gained support for collecting LLDP network
2186 announcements, from hardware that supports this. This is
2187 shown in networkctl output.
2189 * systemd-run gained support for a new -t (--pty) switch for
2190 invoking a binary on a pty whose input and output is
2191 connected to the invoking terminal. This allows executing
2192 processes as system services while interactively
2193 communicating with them via the terminal. Most interestingly
2194 this is supported across container boundaries. Invoking
2195 "systemd-run -t /bin/bash" is an alternative to running a
2196 full login session, the difference being that the former
2197 will not register a session, nor go through the PAM session
2200 * tmpfiles gained support for a new "v" line type for creating
2201 btrfs subvolumes. If the underlying file system is a legacy
2202 file system, this automatically degrades to creating a
2203 normal directory. Among others /var/lib/machines is now
2204 created like this at boot, should it be missing.
2206 * The directory /var/lib/containers/ has been deprecated and
2207 been replaced by /var/lib/machines. The term "machines" has
2208 been used in the systemd context as generic term for both
2209 VMs and containers, and hence appears more appropriate for
2210 this, as the directory can also contain raw images bootable
2213 * systemd-nspawn when invoked with -M but without --directory=
2214 or --image= is now capable of searching for the container
2215 root directory, subvolume or disk image automatically, in
2216 /var/lib/machines. systemd-nspawn@.service has been updated
2217 to make use of this, thus allowing it to be used for raw
2220 * A new machines.target unit has been introduced that is
2221 supposed to group all containers/VMs invoked as services on
2222 the system. systemd-nspawn@.service has been updated to
2223 integrate with that.
2225 * machinectl gained a new "start" command, for invoking a
2226 container as a service. "machinectl start foo" is mostly
2227 equivalent to "systemctl start systemd-nspawn@foo.service",
2228 but handles escaping in a nicer way.
2230 * systemd-nspawn will now mount most of the cgroupfs tree
2231 read-only into each container, with the exception of the
2232 container's own subtree in the name=systemd hierarchy.
2234 * journald now sets the special FS_NOCOW file flag for its
2235 journal files. This should improve performance on btrfs, by
2236 avoiding heavy fragmentation when journald's write-pattern
2237 is used on COW file systems. It degrades btrfs' data
2238 integrity guarantees for the files to the same levels as for
2239 ext3/ext4 however. This should be OK though as journald does
2240 its own data integrity checks and all its objects are
2241 checksummed on disk. Also, journald should handle btrfs disk
2242 full events a lot more gracefully now, by processing SIGBUS
2243 errors, and not relying on fallocate() anymore.
2245 * When journald detects that journal files it is writing to
2246 have been deleted it will immediately start new journal
2249 * systemd now provides a way to store file descriptors
2250 per-service in PID 1. This is useful for daemons to ensure
2251 that fds they require are not lost during a daemon
2252 restart. The fds are passed to the daemon on the next
2253 invocation in the same way socket activation fds are
2254 passed. This is now used by journald to ensure that the
2255 various sockets connected to all the system's stdout/stderr
2256 are not lost when journald is restarted. File descriptors
2257 may be stored in PID 1 via the sd_pid_notify_with_fds() API,
2258 an extension to sd_notify(). Note that a limit is enforced
2259 on the number of fds a service can store in PID 1, and it
2260 defaults to 0, so that no fds may be stored, unless this is
2261 explicitly turned on.
2263 * The default TERM variable to use for units connected to a
2264 terminal, when no other value is explicitly is set is now
2265 vt220 rather than vt102. This should be fairly safe still,
2266 but allows PgUp/PgDn work.
2268 * The /etc/crypttab option header= as known from Debian is now
2271 * "loginctl user-status" and "loginctl session-status" will
2272 now show the last 10 lines of log messages of the
2273 user/session following the status output. Similar,
2274 "machinectl status" will show the last 10 log lines
2275 associated with a virtual machine or container
2276 service. (Note that this is usually not the log messages
2277 done in the VM/container itself, but simply what the
2278 container manager logs. For nspawn this includes all console
2281 * "loginctl session-status" without further argument will now
2282 show the status of the session of the caller. Similar,
2283 "lock-session", "unlock-session", "activate",
2284 "enable-linger", "disable-linger" may now be called without
2285 session/user parameter in which case they apply to the
2286 caller's session/user.
2288 * An X11 session scriptlet is now shipped that uploads
2289 $DISPLAY and $XAUTHORITY into the environment of the systemd
2290 --user daemon if a session begins. This should improve
2291 compatibility with X11 enabled applications run as systemd
2294 * Generators are now subject to masking via /etc and /run, the
2295 same way as unit files.
2297 * networkd .network files gained support for configuring
2298 per-link IPv4/IPv6 packet forwarding as well as IPv4
2299 masquerading. This is by default turned on for veth links to
2300 containers, as registered by systemd-nspawn. This means that
2301 nspawn containers run with --network-veth will now get
2302 automatic routed access to the host's networks without any
2303 further configuration or setup, as long as networkd runs on
2306 * systemd-nspawn gained the --port= (-p) switch to expose TCP
2307 or UDP posts of a container on the host. With this in place
2308 it is possible to run containers with private veth links
2309 (--network-veth), and have their functionality exposed on
2310 the host as if their services were running directly on the
2313 * systemd-nspawn's --network-veth switch now gained a short
2314 version "-n", since with the changes above it is now truly
2315 useful out-of-the-box. The systemd-nspawn@.service has been
2316 updated to make use of it too by default.
2318 * systemd-nspawn will now maintain a per-image R/W lock, to
2319 ensure that the same image is not started more than once
2320 writable. (It's OK to run an image multiple times
2321 simultaneously in read-only mode.)
2323 * systemd-nspawn's --image= option is now capable of
2324 dissecting and booting MBR and GPT disk images that contain
2325 only a single active Linux partition. Previously it
2326 supported only GPT disk images with proper GPT type
2327 IDs. This allows running cloud images from major
2328 distributions directly with systemd-nspawn, without
2331 * In addition to collecting mouse dpi data in the udev
2332 hardware database, there's now support for collecting angle
2333 information for mouse scroll wheels. The database is
2334 supposed to guarantee similar scrolling behavior on mice
2335 that it knows about. There's also support for collecting
2336 information about Touchpad types.
2338 * udev's input_id built-in will now also collect touch screen
2339 dimension data and attach it to probed devices.
2341 * /etc/os-release gained support for a Distribution Privacy
2344 * networkd gained support for creating "ipvlan", "gretap",
2345 "ip6gre", "ip6gretap" and "ip6tnl" network devices.
2347 * systemd-tmpfiles gained support for "a" lines for setting
2350 * systemd-nspawn will now mount /tmp in the container to
2351 tmpfs, automatically.
2353 * systemd now exposes the memory.usage_in_bytes cgroup
2354 attribute and shows it for each service in the "systemctl
2355 status" output, if available.
2357 * When the user presses Ctrl-Alt-Del more than 7x within 2s an
2358 immediate reboot is triggered. This useful if shutdown is
2359 hung and is unable to complete, to expedite the
2360 operation. Note that this kind of reboot will still unmount
2361 all file systems, and hence should not result in fsck being
2364 * A .device unit for an optical block device will now be
2365 considered active only when a medium is in the drive. Also,
2366 mount units are now bound to their backing devices thus
2367 triggering automatic unmounting when devices become
2368 unavailable. With this in place systemd will now
2369 automatically unmount left-over mounts when a CD-ROM is
2370 ejected or an USB stick is yanked from the system.
2372 * networkd-wait-online now has support for waiting for
2373 specific interfaces only (with globbing), and for giving up
2374 after a configurable timeout.
2376 * networkd now exits when idle. It will be automatically
2377 restarted as soon as interfaces show up, are removed or
2378 change state. networkd will stay around as long as there is
2379 at least one DHCP state machine or similar around, that keep
2382 * networkd may now configure IPv6 link-local addressing in
2383 addition to IPv4 link-local addressing.
2385 * The IPv6 "token" for use in SLAAC may now be configured for
2386 each .network interface in networkd.
2388 * Routes configured with networkd may now be assigned a scope
2391 * networkd's [Match] sections now support globbing and lists
2392 of multiple space-separated matches per item.
2394 Contributions from: Alban Crequy, Alin Rauta, Andrey Chaser,
2395 Bastien Nocera, Bruno Bottazzini, Carlos Garnacho, Carlos
2396 Morata Castillo, Chris Atkinson, Chris J. Arges, Christian
2397 Kirbach, Christian Seiler, Christoph Brill, Colin Guthrie,
2398 Colin Walters, Cristian Rodríguez, Daniele Medri, Daniel Mack,
2399 Dave Reisner, David Herrmann, Djalal Harouni, Erik Auerswald,
2400 Filipe Brandenburger, Frank Theile, Gabor Kelemen, Gabriel de
2401 Perthuis, Harald Hoyer, Hui Wang, Ivan Shapovalov, Jan
2402 Engelhardt, Jan Synacek, Jay Faulkner, Johannes Hölzl, Jonas
2403 Ådahl, Jonathan Boulle, Josef Andersson, Kay Sievers, Ken
2404 Werner, Lennart Poettering, Lucas De Marchi, Lukas Märdian,
2405 Lukas Nykryn, Lukasz Skalski, Luke Shumaker, Mantas Mikulėnas,
2406 Manuel Mendez, Marcel Holtmann, Marc Schmitzer, Marko
2407 Myllynen, Martin Pitt, Maxim Mikityanskiy, Michael Biebl,
2408 Michael Marineau, Michael Olbrich, Michal Schmidt, Mindaugas
2409 Baranauskas, Moez Bouhlel, Naveen Kumar, Patrik Flykt, Paul
2410 Martin, Peter Hutterer, Peter Mattern, Philippe De Swert,
2411 Piotr Drąg, Rafael Ferreira, Rami Rosen, Robert Milasan, Ronny
2412 Chevalier, Sangjung Woo, Sebastien Bacher, Sergey Ptashnick,
2413 Shawn Landden, Stéphane Graber, Susant Sahani, Sylvain
2414 Plantefève, Thomas Hindoe Paaboel Andersen, Tim JP, Tom
2415 Gundersen, Topi Miettinen, Torstein Husebø, Umut Tezduyar
2416 Lindskog, Veres Lajos, Vincent Batts, WaLyong Cho, Wieland
2417 Hoffmann, Zbigniew Jędrzejewski-Szmek
2419 — Berlin, 2015-02-16
2423 * When querying unit file enablement status (for example via
2424 "systemctl is-enabled"), a new state "indirect" is now known
2425 which indicates that a unit might not be enabled itself, but
2426 another unit listed in its Also= setting might be.
2428 * Similar to the various existing ConditionXYZ= settings for
2429 units, there are now matching AssertXYZ= settings. While
2430 failing conditions cause a unit to be skipped, but its job
2431 to succeed, failing assertions declared like this will cause
2432 a unit start operation and its job to fail.
2434 * hostnamed now knows a new chassis type "embedded".
2436 * systemctl gained a new "edit" command. When used on a unit
2437 file, this allows extending unit files with .d/ drop-in
2438 configuration snippets or editing the full file (after
2439 copying it from /usr/lib to /etc). This will invoke the
2440 user's editor (as configured with $EDITOR), and reload the
2441 modified configuration after editing.
2443 * "systemctl status" now shows the suggested enablement state
2444 for a unit, as declared in the (usually vendor-supplied)
2445 system preset files.
2447 * nss-myhostname will now resolve the single-label host name
2448 "gateway" to the locally configured default IP routing
2449 gateways, ordered by their metrics. This assigns a stable
2450 name to the used gateways, regardless which ones are
2451 currently configured. Note that the name will only be
2452 resolved after all other name sources (if nss-myhostname is
2453 configured properly) and should hence not negatively impact
2454 systems that use the single-label host name "gateway" in
2457 * systemd-inhibit now allows filtering by mode when listing
2460 * Scope and service units gained a new "Delegate" boolean
2461 property, which, when set, allows processes running inside the
2462 unit to further partition resources. This is primarily
2463 useful for systemd user instances as well as container
2466 * journald will now pick up audit messages directly from
2467 the kernel, and log them like any other log message. The
2468 audit fields are split up and fully indexed. This means that
2469 journalctl in many ways is now a (nicer!) alternative to
2470 ausearch, the traditional audit client. Note that this
2471 implements only a minimal audit client. If you want the
2472 special audit modes like reboot-on-log-overflow, please use
2473 the traditional auditd instead, which can be used in
2474 parallel to journald.
2476 * The ConditionSecurity= unit file option now understands the
2477 special string "audit" to check whether auditing is
2480 * journalctl gained two new commands --vacuum-size= and
2481 --vacuum-time= to delete old journal files until the
2482 remaining ones take up no more than the specified size on disk,
2483 or are not older than the specified time.
2485 * A new, native PPPoE library has been added to sd-network,
2486 systemd's library of light-weight networking protocols. This
2487 library will be used in a future version of networkd to
2488 enable PPPoE communication without an external pppd daemon.
2490 * The busctl tool now understands a new "capture" verb that
2491 works similar to "monitor", but writes a packet capture
2492 trace to STDOUT that can be redirected to a file which is
2493 compatible with libcap's capture file format. This can then
2494 be loaded in Wireshark and similar tools to inspect bus
2497 * The busctl tool now understands a new "tree" verb that shows
2498 the object trees of a specific service on the bus, or of all
2501 * The busctl tool now understands a new "introspect" verb that
2502 shows all interfaces and members of objects on the bus,
2503 including their signature and values. This is particularly
2504 useful to get more information about bus objects shown by
2505 the new "busctl tree" command.
2507 * The busctl tool now understands new verbs "call",
2508 "set-property" and "get-property" for invoking bus method
2509 calls, setting and getting bus object properties in a
2512 * busctl gained a new --augment-creds= argument that controls
2513 whether the tool shall augment credential information it
2514 gets from the bus with data from /proc, in a possibly
2517 * nspawn's --link-journal= switch gained two new values
2518 "try-guest" and "try-host" that work like "guest" and
2519 "host", but do not fail if the host has no persistent
2520 journalling enabled. -j is now equivalent to
2521 --link-journal=try-guest.
2523 * macvlan network devices created by nspawn will now have
2524 stable MAC addresses.
2526 * A new SmackProcessLabel= unit setting has been added, which
2527 controls the SMACK security label processes forked off by
2528 the respective unit shall use.
2530 * If compiled with --enable-xkbcommon, systemd-localed will
2531 verify x11 keymap settings by compiling the given keymap. It
2532 will spew out warnings if the compilation fails. This
2533 requires libxkbcommon to be installed.
2535 * When a coredump is collected, a larger number of metadata
2536 fields is now collected and included in the journal records
2537 created for it. More specifically, control group membership,
2538 environment variables, memory maps, working directory,
2539 chroot directory, /proc/$PID/status, and a list of open file
2540 descriptors is now stored in the log entry.
2542 * The udev hwdb now contains DPI information for mice. For
2545 http://who-t.blogspot.de/2014/12/building-a-dpi-database-for-mice.html
2547 * All systemd programs that read standalone configuration
2548 files in /etc now also support a corresponding series of
2549 .conf.d configuration directories in /etc/, /run/,
2550 /usr/local/lib/, /usr/lib/, and (if configured with
2551 --enable-split-usr) /lib/. In particular, the following
2552 configuration files now have corresponding configuration
2553 directories: system.conf user.conf, logind.conf,
2554 journald.conf, sleep.conf, bootchart.conf, coredump.conf,
2555 resolved.conf, timesyncd.conf, journal-remote.conf, and
2556 journal-upload.conf. Note that distributions should use the
2557 configuration directories in /usr/lib/; the directories in
2558 /etc/ are reserved for the system administrator.
2560 * systemd-rfkill will no longer take the rfkill device name
2561 into account when storing rfkill state on disk, as the name
2562 might be dynamically assigned and not stable. Instead, the
2563 ID_PATH udev variable combined with the rfkill type (wlan,
2564 bluetooth, ...) is used.
2566 * A new service systemd-machine-id-commit.service has been
2567 added. When used on systems where /etc is read-only during
2568 boot, and /etc/machine-id is not initialized (but an empty
2569 file), this service will copy the temporary machine ID
2570 created as replacement into /etc after the system is fully
2571 booted up. This is useful for systems that are freshly
2572 installed with a non-initialized machine ID, but should get
2573 a fixed machine ID for subsequent boots.
2575 * networkd's .netdev files now provide a large set of
2576 configuration parameters for VXLAN devices. Similarly, the
2577 bridge port cost parameter is now configurable in .network
2578 files. There's also new support for configuring IP source
2579 routing. networkd .link files gained support for a new
2580 OriginalName= match that is useful to match against the
2581 original interface name the kernel assigned. .network files
2582 may include MTU= and MACAddress= fields for altering the MTU
2583 and MAC address while being connected to a specific network
2586 * The LUKS logic gained supported for configuring
2587 UUID-specific key files. There's also new support for naming
2588 LUKS device from the kernel command line, using the new
2589 luks.name= argument.
2591 * Timer units may now be transiently created via the bus API
2592 (this was previously already available for scope and service
2593 units). In addition it is now possible to create multiple
2594 transient units at the same time with a single bus call. The
2595 "systemd-run" tool has been updated to make use of this for
2596 running commands on a specified time, in at(1)-style.
2598 * tmpfiles gained support for "t" lines, for assigning
2599 extended attributes to files. Among other uses this may be
2600 used to assign SMACK labels to files.
2602 Contributions from: Alin Rauta, Alison Chaiken, Andrej
2603 Manduch, Bastien Nocera, Chris Atkinson, Chris Leech, Chris
2604 Mayo, Colin Guthrie, Colin Walters, Cristian Rodríguez,
2605 Daniele Medri, Daniel Mack, Dan Williams, Dan Winship, Dave
2606 Reisner, David Herrmann, Didier Roche, Felipe Sateler, Gavin
2607 Li, Hans de Goede, Harald Hoyer, Iago López Galeiras, Ivan
2608 Shapovalov, Jakub Filak, Jan Janssen, Jan Synacek, Joe
2609 Lawrence, Josh Triplett, Kay Sievers, Lennart Poettering,
2610 Lukas Nykryn, Łukasz Stelmach, Maciej Wereski, Mantas
2611 Mikulėnas, Marcel Holtmann, Martin Pitt, Maurizio Lombardi,
2612 Michael Biebl, Michael Chapman, Michael Marineau, Michal
2613 Schmidt, Michal Sekletar, Olivier Brunel, Patrik Flykt, Peter
2614 Hutterer, Przemyslaw Kedzierski, Rami Rosen, Ray Strode,
2615 Richard Schütz, Richard W.M. Jones, Ronny Chevalier, Ross
2616 Lagerwall, Sean Young, Stanisław Pitucha, Susant Sahani,
2617 Thomas Haller, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
2618 Torstein Husebø, Umut Tezduyar Lindskog, Vicente Olivert
2619 Riera, WaLyong Cho, Wesley Dawson, Zbigniew Jędrzejewski-Szmek
2621 — Berlin, 2014-12-10
2625 * journalctl gained the new options -t/--identifier= to match
2626 on the syslog identifier (aka "tag"), as well as --utc to
2627 show log timestamps in the UTC timezone. journalctl now also
2628 accepts -n/--lines=all to disable line capping in a pager.
2630 * journalctl gained a new switch, --flush, that synchronously
2631 flushes logs from /run/log/journal to /var/log/journal if
2632 persistent storage is enabled. systemd-journal-flush.service
2633 now waits until the operation is complete.
2635 * Services can notify the manager before they start a reload
2636 (by sending RELOADING=1) or shutdown (by sending
2637 STOPPING=1). This allows the manager to track and show the
2638 internal state of daemons and closes a race condition when
2639 the process is still running but has closed its D-Bus
2642 * Services with Type=oneshot do not have to have any ExecStart
2645 * User units are now loaded also from
2646 $XDG_RUNTIME_DIR/systemd/user/. This is similar to the
2647 /run/systemd/user directory that was already previously
2648 supported, but is under the control of the user.
2650 * Job timeouts (i.e. time-outs on the time a job that is
2651 queued stays in the run queue) can now optionally result in
2652 immediate reboot or power-off actions (JobTimeoutAction= and
2653 JobTimeoutRebootArgument=). This is useful on ".target"
2654 units, to limit the maximum time a target remains
2655 undispatched in the run queue, and to trigger an emergency
2656 operation in such a case. This is now used by default to
2657 turn off the system if boot-up (as defined by everything in
2658 basic.target) hangs and does not complete for at least
2659 15min. Also, if power-off or reboot hang for at least 30min
2660 an immediate power-off/reboot operation is triggered. This
2661 functionality is particularly useful to increase reliability
2662 on embedded devices, but also on laptops which might
2663 accidentally get powered on when carried in a backpack and
2664 whose boot stays stuck in a hard disk encryption passphrase
2667 * systemd-logind can be configured to also handle lid switch
2668 events even when the machine is docked or multiple displays
2669 are attached (HandleLidSwitchDocked= option).
2671 * A helper binary and a service have been added which can be
2672 used to resume from hibernation in the initramfs. A
2673 generator will parse the resume= option on the kernel
2674 command line to trigger resume.
2676 * A user console daemon systemd-consoled has been
2677 added. Currently, it is a preview, and will so far open a
2678 single terminal on each session of the user marked as
2679 Desktop=systemd-console.
2681 * Route metrics can be specified for DHCP routes added by
2684 * The SELinux context of socket-activated services can be set
2685 from the information provided by the networking stack
2686 (SELinuxContextFromNet= option).
2688 * Userspace firmware loading support has been removed and
2689 the minimum supported kernel version is thus bumped to 3.7.
2691 * Timeout for udev workers has been increased from 1 to 3
2692 minutes, but a warning will be printed after 1 minute to
2693 help diagnose kernel modules that take a long time to load.
2695 * Udev rules can now remove tags on devices with TAG-="foobar".
2697 * systemd's readahead implementation has been removed. In many
2698 circumstances it didn't give expected benefits even for
2699 rotational disk drives and was becoming less relevant in the
2700 age of SSDs. As none of the developers has been using
2701 rotating media anymore, and nobody stepped up to actively
2702 maintain this component of systemd it has now been removed.
2704 * Swap units can use Options= to specify discard options.
2705 Discard options specified for swaps in /etc/fstab are now
2708 * Docker containers are now detected as a separate type of
2711 * The Password Agent protocol gained support for queries where
2712 the user input is shown, useful e.g. for user names.
2713 systemd-ask-password gained a new --echo option to turn that
2716 * The default sysctl.d/ snippets will now set:
2718 net.core.default_qdisc = fq_codel
2720 This selects Fair Queuing Controlled Delay as the default
2721 queuing discipline for network interfaces. fq_codel helps
2722 fight the network bufferbloat problem. It is believed to be
2723 a good default with no tuning required for most workloads.
2724 Downstream distributions may override this choice. On 10Gbit
2725 servers that do not do forwarding, "fq" may perform better.
2726 Systems without a good clocksource should use "pfifo_fast".
2728 * If kdbus is enabled during build a new option BusPolicy= is
2729 available for service units, that allows locking all service
2730 processes into a stricter bus policy, in order to limit
2731 access to various bus services, or even hide most of them
2732 from the service's view entirely.
2734 * networkctl will now show the .network and .link file
2735 networkd has applied to a specific interface.
2737 * sd-login gained a new API call sd_session_get_desktop() to
2738 query which desktop environment has been selected for a
2741 * UNIX utmp support is now compile-time optional to support
2742 legacy-free systems.
2744 * systemctl gained two new commands "add-wants" and
2745 "add-requires" for pulling in units from specific targets
2748 * If the word "rescue" is specified on the kernel command line
2749 the system will now boot into rescue mode (aka
2750 rescue.target), which was previously available only by
2751 specifying "1" or "systemd.unit=rescue.target" on the kernel
2752 command line. This new kernel command line option nicely
2753 mirrors the already existing "emergency" kernel command line
2756 * New kernel command line options mount.usr=, mount.usrflags=,
2757 mount.usrfstype= have been added that match root=, rootflags=,
2758 rootfstype= but allow mounting a specific file system to
2761 * The $NOTIFY_SOCKET is now also passed to control processes of
2762 services, not only the main process.
2764 * This version reenables support for fsck's -l switch. This
2765 means at least version v2.25 of util-linux is required for
2766 operation, otherwise dead-locks on device nodes may
2767 occur. Again: you need to update util-linux to at least
2768 v2.25 when updating systemd to v217.
2770 * The "multi-seat-x" tool has been removed from systemd, as
2771 its functionality has been integrated into X servers 1.16,
2772 and the tool is hence redundant. It is recommended to update
2773 display managers invoking this tool to simply invoke X
2774 directly from now on, again.
2776 * Support for the new ALLOW_INTERACTIVE_AUTHORIZATION D-Bus
2777 message flag has been added for all of systemd's PolicyKit
2778 authenticated method calls has been added. In particular
2779 this now allows optional interactive authorization via
2780 PolicyKit for many of PID1's privileged operations such as
2781 unit file enabling and disabling.
2783 * "udevadm hwdb --update" learnt a new switch "--usr" for
2784 placing the rebuilt hardware database in /usr instead of
2785 /etc. When used only hardware database entries stored in
2786 /usr will be used, and any user database entries in /etc are
2787 ignored. This functionality is useful for vendors to ship a
2788 pre-built database on systems where local configuration is
2789 unnecessary or unlikely.
2791 * Calendar time specifications in .timer units now also
2792 understand the strings "semi-annually", "quarterly" and
2793 "minutely" as shortcuts (in addition to the preexisting
2794 "anually", "hourly", ...).
2796 * systemd-tmpfiles will now correctly create files in /dev
2797 at boot which are marked for creation only at boot. It is
2798 recommended to always create static device nodes with 'c!'
2799 and 'b!', so that they are created only at boot and not
2800 overwritten at runtime.
2802 * When the watchdog logic is used for a service (WatchdogSec=)
2803 and the watchdog timeout is hit the service will now be
2804 terminated with SIGABRT (instead of just SIGTERM), in order
2805 to make sure a proper coredump and backtrace is
2806 generated. This ensures that hanging services will result in
2807 similar coredump/backtrace behaviour as services that hit a
2810 Contributions from: Andreas Henriksson, Andrei Borzenkov,
2811 Angus Gibson, Ansgar Burchardt, Ben Wolsieffer, Brandon L.
2812 Black, Christian Hesse, Cristian Rodríguez, Daniel Buch,
2813 Daniele Medri, Daniel Mack, Dan Williams, Dave Reisner, David
2814 Herrmann, David Sommerseth, David Strauss, Emil Renner
2815 Berthing, Eric Cook, Evangelos Foutras, Filipe Brandenburger,
2816 Gustavo Sverzut Barbieri, Hans de Goede, Harald Hoyer, Hristo
2817 Venev, Hugo Grostabussiat, Ivan Shapovalov, Jan Janssen, Jan
2818 Synacek, Jonathan Liu, Juho Son, Karel Zak, Kay Sievers, Klaus
2819 Purer, Koen Kooi, Lennart Poettering, Lukas Nykryn, Lukasz
2820 Skalski, Łukasz Stelmach, Mantas Mikulėnas, Marcel Holtmann,
2821 Marius Tessmann, Marko Myllynen, Martin Pitt, Michael Biebl,
2822 Michael Marineau, Michael Olbrich, Michael Scherer, Michal
2823 Schmidt, Michal Sekletar, Miroslav Lichvar, Patrik Flykt,
2824 Philippe De Swert, Piotr Drąg, Rahul Sundaram, Richard
2825 Weinberger, Robert Milasan, Ronny Chevalier, Ruben Kerkhof,
2826 Santiago Vila, Sergey Ptashnick, Simon McVittie, Sjoerd
2827 Simons, Stefan Brüns, Steven Allen, Steven Noonan, Susant
2828 Sahani, Sylvain Plantefève, Thomas Hindoe Paaboel Andersen,
2829 Timofey Titovets, Tobias Hunger, Tom Gundersen, Torstein
2830 Husebø, Umut Tezduyar Lindskog, WaLyong Cho, Zbigniew
2833 — Berlin, 2014-10-28
2837 * timedated no longer reads NTP implementation unit names from
2838 /usr/lib/systemd/ntp-units.d/*.list. Alternative NTP
2839 implementations should add a
2841 Conflicts=systemd-timesyncd.service
2843 to their unit files to take over and replace systemd's NTP
2844 default functionality.
2846 * systemd-sysusers gained a new line type "r" for configuring
2847 which UID/GID ranges to allocate system users/groups
2848 from. Lines of type "u" may now add an additional column
2849 that specifies the home directory for the system user to be
2850 created. Also, systemd-sysusers may now optionally read user
2851 information from STDIN instead of a file. This is useful for
2852 invoking it from RPM preinst scriptlets that need to create
2853 users before the first RPM file is installed since these
2854 files might need to be owned by them. A new
2855 %sysusers_create_inline RPM macro has been introduced to do
2856 just that. systemd-sysusers now updates the shadow files as
2857 well as the user/group databases, which should enhance
2858 compatibility with certain tools like grpck.
2860 * A number of bus APIs of PID 1 now optionally consult
2861 PolicyKit to permit access for otherwise unprivileged
2862 clients under certain conditions. Note that this currently
2863 doesn't support interactive authentication yet, but this is
2864 expected to be added eventually, too.
2866 * /etc/machine-info now has new fields for configuring the
2867 deployment environment of the machine, as well as the
2868 location of the machine. hostnamectl has been updated with
2869 new command to update these fields.
2871 * systemd-timesyncd has been updated to automatically acquire
2872 NTP server information from systemd-networkd, which might
2873 have been discovered via DHCP.
2875 * systemd-resolved now includes a caching DNS stub resolver
2876 and a complete LLMNR name resolution implementation. A new
2877 NSS module "nss-resolve" has been added which can be used
2878 instead of glibc's own "nss-dns" to resolve hostnames via
2879 systemd-resolved. Hostnames, addresses and arbitrary RRs may
2880 be resolved via systemd-resolved D-Bus APIs. In contrast to
2881 the glibc internal resolver systemd-resolved is aware of
2882 multi-homed system, and keeps DNS server and caches separate
2883 and per-interface. Queries are sent simultaneously on all
2884 interfaces that have DNS servers configured, in order to
2885 properly handle VPNs and local LANs which might resolve
2886 separate sets of domain names. systemd-resolved may acquire
2887 DNS server information from systemd-networkd automatically,
2888 which in turn might have discovered them via DHCP. A tool
2889 "systemd-resolve-host" has been added that may be used to
2890 query the DNS logic in resolved. systemd-resolved implements
2891 IDNA and automatically uses IDNA or UTF-8 encoding depending
2892 on whether classic DNS or LLMNR is used as transport. In the
2893 next releases we intend to add a DNSSEC and mDNS/DNS-SD
2894 implementation to systemd-resolved.
2896 * A new NSS module nss-mymachines has been added, that
2897 automatically resolves the names of all local registered
2898 containers to their respective IP addresses.
2900 * A new client tool "networkctl" for systemd-networkd has been
2901 added. It currently is entirely passive and will query
2902 networking configuration from udev, rtnetlink and networkd,
2903 and present it to the user in a very friendly
2904 way. Eventually, we hope to extend it to become a full
2905 control utility for networkd.
2907 * .socket units gained a new DeferAcceptSec= setting that
2908 controls the kernels' TCP_DEFER_ACCEPT sockopt for
2909 TCP. Similarly, support for controlling TCP keep-alive
2910 settings has been added (KeepAliveTimeSec=,
2911 KeepAliveIntervalSec=, KeepAliveProbes=). Also, support for
2912 turning off Nagle's algorithm on TCP has been added
2915 * logind learned a new session type "web", for use in projects
2916 like Cockpit which register web clients as PAM sessions.
2918 * timer units with at least one OnCalendar= setting will now
2919 be started only after timer-sync.target has been
2920 reached. This way they will not elapse before the system
2921 clock has been corrected by a local NTP client or
2922 similar. This is particular useful on RTC-less embedded
2923 machines, that come up with an invalid system clock.
2925 * systemd-nspawn's --network-veth= switch should now result in
2926 stable MAC addresses for both the outer and the inner side
2929 * systemd-nspawn gained a new --volatile= switch for running
2930 container instances with /etc or /var unpopulated.
2932 * The kdbus client code has been updated to use the new Linux
2933 3.17 memfd subsystem instead of the old kdbus-specific one.
2935 * systemd-networkd's DHCP client and server now support
2936 FORCERENEW. There are also new configuration options to
2937 configure the vendor client identifier and broadcast mode
2940 * systemd will no longer inform the kernel about the current
2941 timezone, as this is necessarily incorrect and racy as the
2942 kernel has no understanding of DST and similar
2943 concepts. This hence means FAT timestamps will be always
2944 considered UTC, similar to what Android is already
2945 doing. Also, when the RTC is configured to the local time
2946 (rather than UTC) systemd will never synchronize back to it,
2947 as this might confuse Windows at a later boot.
2949 * systemd-analyze gained a new command "verify" for offline
2950 validation of unit files.
2952 * systemd-networkd gained support for a couple of additional
2953 settings for bonding networking setups. Also, the metric for
2954 statically configured routes may now be configured. For
2955 network interfaces where this is appropriate the peer IP
2956 address may now be configured.
2958 * systemd-networkd's DHCP client will no longer request
2959 broadcasting by default, as this tripped up some networks.
2960 For hardware where broadcast is required the feature should
2961 be switched back on using RequestBroadcast=yes.
2963 * systemd-networkd will now set up IPv4LL addresses (when
2964 enabled) even if DHCP is configured successfully.
2966 * udev will now default to respect network device names given
2967 by the kernel when the kernel indicates that these are
2968 predictable. This behavior can be tweaked by changing
2969 NamePolicy= in the relevant .link file.
2971 * A new library systemd-terminal has been added that
2972 implements full TTY stream parsing and rendering. This
2973 library is supposed to be used later on for implementing a
2974 full userspace VT subsystem, replacing the current kernel
2977 * A new tool systemd-journal-upload has been added to push
2978 journal data to a remote system running
2979 systemd-journal-remote.
2981 * journald will no longer forward all local data to another
2982 running syslog daemon. This change has been made because
2983 rsyslog (which appears to be the most commonly used syslog
2984 implementation these days) no longer makes use of this, and
2985 instead pulls the data out of the journal on its own. Since
2986 forwarding the messages to a non-existent syslog server is
2987 more expensive than we assumed we have now turned this
2988 off. If you run a syslog server that is not a recent rsyslog
2989 version, you have to turn this option on again
2990 (ForwardToSyslog= in journald.conf).
2992 * journald now optionally supports the LZ4 compressor for
2993 larger journal fields. This compressor should perform much
2994 better than XZ which was the previous default.
2996 * machinectl now shows the IP addresses of local containers,
2997 if it knows them, plus the interface name of the container.
2999 * A new tool "systemd-escape" has been added that makes it
3000 easy to escape strings to build unit names and similar.
3002 * sd_notify() messages may now include a new ERRNO= field
3003 which is parsed and collected by systemd and shown among the
3004 "systemctl status" output for a service.
3006 * A new component "systemd-firstboot" has been added that
3007 queries the most basic systemd information (timezone,
3008 hostname, root password) interactively on first
3009 boot. Alternatively it may also be used to provision these
3010 things offline on OS images installed into directories.
3012 * The default sysctl.d/ snippets will now set
3014 net.ipv4.conf.default.promote_secondaries=1
3016 This has the benefit of no flushing secondary IP addresses
3017 when primary addresses are removed.
3019 Contributions from: Ansgar Burchardt, Bastien Nocera, Colin
3020 Walters, Dan Dedrick, Daniel Buch, Daniel Korostil, Daniel
3021 Mack, Dan Williams, Dave Reisner, David Herrmann, Denis
3022 Kenzior, Eelco Dolstra, Eric Cook, Hannes Reinecke, Harald
3023 Hoyer, Hong Shick Pak, Hui Wang, Jean-André Santoni, Jóhann
3024 B. Guðmundsson, Jon Severinsson, Karel Zak, Kay Sievers, Kevin
3025 Wells, Lennart Poettering, Lukas Nykryn, Mantas Mikulėnas,
3026 Marc-Antoine Perennou, Martin Pitt, Michael Biebl, Michael
3027 Marineau, Michael Olbrich, Michal Schmidt, Michal Sekletar,
3028 Miguel Angel Ajo, Mike Gilbert, Olivier Brunel, Robert
3029 Schiele, Ronny Chevalier, Simon McVittie, Sjoerd Simons, Stef
3030 Walter, Steven Noonan, Susant Sahani, Tanu Kaskinen, Thomas
3031 Blume, Thomas Hindoe Paaboel Andersen, Timofey Titovets,
3032 Tobias Geerinckx-Rice, Tomasz Torcz, Tom Gundersen, Umut
3033 Tezduyar Lindskog, Zbigniew Jędrzejewski-Szmek
3035 — Berlin, 2014-08-19
3039 * A new tool systemd-sysusers has been added. This tool
3040 creates system users and groups in /etc/passwd and
3041 /etc/group, based on static declarative system user/group
3042 definitions in /usr/lib/sysusers.d/. This is useful to
3043 enable factory resets and volatile systems that boot up with
3044 an empty /etc directory, and thus need system users and
3045 groups created during early boot. systemd now also ships
3046 with two default sysusers.d/ files for the most basic
3047 users and groups systemd and the core operating system
3050 * A new tmpfiles snippet has been added that rebuilds the
3051 essential files in /etc on boot, should they be missing.
3053 * A directive for ensuring automatic clean-up of
3054 /var/cache/man/ has been removed from the default
3055 configuration. This line should now be shipped by the man
3056 implementation. The necessary change has been made to the
3057 man-db implementation. Note that you need to update your man
3058 implementation to one that ships this line, otherwise no
3059 automatic clean-up of /var/cache/man will take place.
3061 * A new condition ConditionNeedsUpdate= has been added that
3062 may conditionalize services to only run when /etc or /var
3063 are "older" than the vendor operating system resources in
3064 /usr. This is useful for reconstructing or updating /etc
3065 after an offline update of /usr or a factory reset, on the
3066 next reboot. Services that want to run once after such an
3067 update or reset should use this condition and order
3068 themselves before the new systemd-update-done.service, which
3069 will mark the two directories as fully updated. A number of
3070 service files have been added making use of this, to rebuild
3071 the udev hardware database, the journald message catalog and
3072 dynamic loader cache (ldconfig). The systemd-sysusers tool
3073 described above also makes use of this now. With this in
3074 place it is now possible to start up a minimal operating
3075 system with /etc empty cleanly. For more information on the
3076 concepts involved see this recent blog story:
3078 http://0pointer.de/blog/projects/stateless.html
3080 * A new system group "input" has been introduced, and all
3081 input device nodes get this group assigned. This is useful
3082 for system-level software to get access to input devices. It
3083 complements what is already done for "audio" and "video".
3085 * systemd-networkd learnt minimal DHCPv4 server support in
3086 addition to the existing DHCPv4 client support. It also
3087 learnt DHCPv6 client and IPv6 Router Solicitation client
3088 support. The DHCPv4 client gained support for static routes
3089 passed in from the server. Note that the [DHCPv4] section
3090 known in older systemd-networkd versions has been renamed to
3091 [DHCP] and is now also used by the DHCPv6 client. Existing
3092 .network files using settings of this section should be
3093 updated, though compatibility is maintained. Optionally, the
3094 client hostname may now be sent to the DHCP server.
3096 * networkd gained support for vxlan virtual networks as well
3097 as tun/tap and dummy devices.
3099 * networkd gained support for automatic allocation of address
3100 ranges for interfaces from a system-wide pool of
3101 addresses. This is useful for dynamically managing a large
3102 number of interfaces with a single network configuration
3103 file. In particular this is useful to easily assign
3104 appropriate IP addresses to the veth links of a large number
3105 of nspawn instances.
3107 * RPM macros for processing sysusers, sysctl and binfmt
3108 drop-in snippets at package installation time have been
3111 * The /etc/os-release file should now be placed in
3112 /usr/lib/os-release. The old location is automatically
3113 created as symlink. /usr/lib is the more appropriate
3114 location of this file, since it shall actually describe the
3115 vendor operating system shipped in /usr, and not the
3116 configuration stored in /etc.
3118 * .mount units gained a new boolean SloppyOptions= setting
3119 that maps to mount(8)'s -s option which enables permissive
3120 parsing of unknown mount options.
3122 * tmpfiles learnt a new "L+" directive which creates a symlink
3123 but (unlike "L") deletes a pre-existing file first, should
3124 it already exist and not already be the correct
3125 symlink. Similarly, "b+", "c+" and "p+" directives have been
3126 added as well, which create block and character devices, as
3127 well as fifos in the filesystem, possibly removing any
3128 pre-existing files of different types.
3130 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
3131 'argument' field (which so far specified the source to
3132 symlink/copy the files from) is now optional. If omitted the
3133 same file os copied from /usr/share/factory/ suffixed by the
3134 full destination path. This is useful for populating /etc
3135 with essential files, by copying them from vendor defaults
3136 shipped in /usr/share/factory/etc.
3138 * A new command "systemctl preset-all" has been added that
3139 applies the service preset settings to all installed unit
3140 files. A new switch --preset-mode= has been added that
3141 controls whether only enable or only disable operations
3144 * A new command "systemctl is-system-running" has been added
3145 that allows checking the overall state of the system, for
3146 example whether it is fully up and running.
3148 * When the system boots up with an empty /etc, the equivalent
3149 to "systemctl preset-all" is executed during early boot, to
3150 make sure all default services are enabled after a factory
3153 * systemd now contains a minimal preset file that enables the
3154 most basic services systemd ships by default.
3156 * Unit files' [Install] section gained a new DefaultInstance=
3157 field for defining the default instance to create if a
3158 template unit is enabled with no instance specified.
3160 * A new passive target cryptsetup-pre.target has been added
3161 that may be used by services that need to make they run and
3162 finish before the first LUKS cryptographic device is set up.
3164 * The /dev/loop-control and /dev/btrfs-control device nodes
3165 are now owned by the "disk" group by default, opening up
3166 access to this group.
3168 * systemd-coredump will now automatically generate a
3169 stack trace of all core dumps taking place on the system,
3170 based on elfutils' libdw library. This stack trace is logged
3173 * systemd-coredump may now optionally store coredumps directly
3174 on disk (in /var/lib/systemd/coredump, possibly compressed),
3175 instead of storing them unconditionally in the journal. This
3176 mode is the new default. A new configuration file
3177 /etc/systemd/coredump.conf has been added to configure this
3178 and other parameters of systemd-coredump.
3180 * coredumpctl gained a new "info" verb to show details about a
3181 specific coredump. A new switch "-1" has also been added
3182 that makes sure to only show information about the most
3183 recent entry instead of all entries. Also, as the tool is
3184 generally useful now the "systemd-" prefix of the binary
3185 name has been removed. Distributions that want to maintain
3186 compatibility with the old name should add a symlink from
3187 the old name to the new name.
3189 * journald's SplitMode= now defaults to "uid". This makes sure
3190 that unprivileged users can access their own coredumps with
3191 coredumpctl without restrictions.
3193 * New kernel command line options "systemd.wants=" (for
3194 pulling an additional unit during boot), "systemd.mask="
3195 (for masking a specific unit for the boot), and
3196 "systemd.debug-shell" (for enabling the debug shell on tty9)
3197 have been added. This is implemented in the new generator
3198 "systemd-debug-generator".
3200 * systemd-nspawn will now by default filter a couple of
3201 syscalls for containers, among them those required for
3202 kernel module loading, direct x86 IO port access, swap
3203 management, and kexec. Most importantly though
3204 open_by_handle_at() is now prohibited for containers,
3205 closing a hole similar to a recently discussed vulnerability
3206 in docker regarding access to files on file hierarchies the
3207 container should normally not have access to. Note that, for
3208 nspawn, we generally make no security claims anyway (and
3209 this is explicitly documented in the man page), so this is
3210 just a fix for one of the most obvious problems.
3212 * A new man page file-hierarchy(7) has been added that
3213 contains a minimized, modernized version of the file system
3214 layout systemd expects, similar in style to the FHS
3215 specification or hier(5). A new tool systemd-path(1) has
3216 been added to query many of these paths for the local
3219 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
3220 longer done. Since the directory now has a per-user size
3221 limit, and is cleaned on logout this appears unnecessary,
3222 in particular since this now brings the lifecycle of this
3223 directory closer in line with how IPC objects are handled.
3225 * systemd.pc now exports a number of additional directories,
3226 including $libdir (which is useful to identify the library
3227 path for the primary architecture of the system), and a
3228 couple of drop-in directories.
3230 * udev's predictable network interface names now use the dev_port
3231 sysfs attribute, introduced in linux 3.15 instead of dev_id to
3232 distinguish between ports of the same PCI function. dev_id should
3233 only be used for ports using the same HW address, hence the need
3236 * machined has been updated to export the OS version of a
3237 container (read from /etc/os-release and
3238 /usr/lib/os-release) on the bus. This is now shown in
3239 "machinectl status" for a machine.
3241 * A new service setting RestartForceExitStatus= has been
3242 added. If configured to a set of exit signals or process
3243 return values, the service will be restarted when the main
3244 daemon process exits with any of them, regardless of the
3247 * systemctl's -H switch for connecting to remote systemd
3248 machines has been extended so that it may be used to
3249 directly connect to a specific container on the
3250 host. "systemctl -H root@foobar:waldi" will now connect as
3251 user "root" to host "foobar", and then proceed directly to
3252 the container named "waldi". Note that currently you have to
3253 authenticate as user "root" for this to work, as entering
3254 containers is a privileged operation.
3256 Contributions from: Andreas Henriksson, Benjamin Steinwender,
3257 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
3258 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
3259 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
3260 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
3261 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
3262 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
3263 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
3264 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
3265 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
3266 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
3267 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
3269 — Berlin, 2014-07-03
3273 * As an experimental feature, udev now tries to lock the
3274 disk device node (flock(LOCK_SH|LOCK_NB)) while it
3275 executes events for the disk or any of its partitions.
3276 Applications like partitioning programs can lock the
3277 disk device node (flock(LOCK_EX)) and claim temporary
3278 device ownership that way; udev will entirely skip all event
3279 handling for this disk and its partitions. If the disk
3280 was opened for writing, the close will trigger a partition
3281 table rescan in udev's "watch" facility, and if needed
3282 synthesize "change" events for the disk and all its partitions.
3283 This is now unconditionally enabled, and if it turns out to
3284 cause major problems, we might turn it on only for specific
3285 devices, or might need to disable it entirely. Device Mapper
3286 devices are excluded from this logic.
3288 * We temporarily dropped the "-l" switch for fsck invocations,
3289 since they collide with the flock() logic above. util-linux
3290 upstream has been changed already to avoid this conflict,
3291 and we will readd "-l" as soon as util-linux with this
3292 change has been released.
3294 * The dependency on libattr has been removed. Since a long
3295 time, the extended attribute calls have moved to glibc, and
3296 libattr is thus unnecessary.
3298 * Virtualization detection works without privileges now. This
3299 means the systemd-detect-virt binary no longer requires
3300 CAP_SYS_PTRACE file capabilities, and our daemons can run
3301 with fewer privileges.
3303 * systemd-networkd now runs under its own "systemd-network"
3304 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
3305 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
3306 loses the ability to write to files owned by root this way.
3308 * Similarly, systemd-resolved now runs under its own
3309 "systemd-resolve" user with no capabilities remaining.
3311 * Similarly, systemd-bus-proxyd now runs under its own
3312 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
3314 * systemd-networkd gained support for setting up "veth"
3315 virtual Ethernet devices for container connectivity, as well
3316 as GRE and VTI tunnels.
3318 * systemd-networkd will no longer automatically attempt to
3319 manually load kernel modules necessary for certain tunnel
3320 transports. Instead, it is assumed the kernel loads them
3321 automatically when required. This only works correctly on
3322 very new kernels. On older kernels, please consider adding
3323 the kernel modules to /etc/modules-load.d/ as a work-around.
3325 * The resolv.conf file systemd-resolved generates has been
3326 moved to /run/systemd/resolve/. If you have a symlink from
3327 /etc/resolv.conf, it might be necessary to correct it.
3329 * Two new service settings, ProtectHome= and ProtectSystem=,
3330 have been added. When enabled, they will make the user data
3331 (such as /home) inaccessible or read-only and the system
3332 (such as /usr) read-only, for specific services. This allows
3333 very light-weight per-service sandboxing to avoid
3334 modifications of user data or system files from
3335 services. These two new switches have been enabled for all
3336 of systemd's long-running services, where appropriate.
3338 * Socket units gained new SocketUser= and SocketGroup=
3339 settings to set the owner user and group of AF_UNIX sockets
3340 and FIFOs in the file system.
3342 * Socket units gained a new RemoveOnStop= setting. If enabled,
3343 all FIFOS and sockets in the file system will be removed
3344 when the specific socket unit is stopped.
3346 * Socket units gained a new Symlinks= setting. It takes a list
3347 of symlinks to create to file system sockets or FIFOs
3348 created by the specific Unix sockets. This is useful to
3349 manage symlinks to socket nodes with the same life-cycle as
3352 * The /dev/log socket and /dev/initctl FIFO have been moved to
3353 /run, and have been replaced by symlinks. This allows
3354 connecting to these facilities even if PrivateDevices=yes is
3355 used for a service (which makes /dev/log itself unavailable,
3356 but /run is left). This also has the benefit of ensuring
3357 that /dev only contains device nodes, directories and
3358 symlinks, and nothing else.
3360 * sd-daemon gained two new calls sd_pid_notify() and
3361 sd_pid_notifyf(). They are similar to sd_notify() and
3362 sd_notifyf(), but allow overriding of the source PID of
3363 notification messages if permissions permit this. This is
3364 useful to send notify messages on behalf of a different
3365 process (for example, the parent process). The
3366 systemd-notify tool has been updated to make use of this
3367 when sending messages (so that notification messages now
3368 originate from the shell script invoking systemd-notify and
3369 not the systemd-notify process itself. This should minimize
3370 a race where systemd fails to associate notification
3371 messages to services when the originating process already
3374 * A new "on-abnormal" setting for Restart= has been added. If
3375 set, it will result in automatic restarts on all "abnormal"
3376 reasons for a process to exit, which includes unclean
3377 signals, core dumps, timeouts and watchdog timeouts, but
3378 does not include clean and unclean exit codes or clean
3379 signals. Restart=on-abnormal is an alternative for
3380 Restart=on-failure for services that shall be able to
3381 terminate and avoid restarts on certain errors, by
3382 indicating so with an unclean exit code. Restart=on-failure
3383 or Restart=on-abnormal is now the recommended setting for
3384 all long-running services.
3386 * If the InaccessibleDirectories= service setting points to a
3387 mount point (or if there are any submounts contained within
3388 it), it is now attempted to completely unmount it, to make
3389 the file systems truly unavailable for the respective
3392 * The ReadOnlyDirectories= service setting and
3393 systemd-nspawn's --read-only parameter are now recursively
3394 applied to all submounts, too.
3396 * Mount units may now be created transiently via the bus APIs.
3398 * The support for SysV and LSB init scripts has been removed
3399 from the systemd daemon itself. Instead, it is now
3400 implemented as a generator that creates native systemd units
3401 from these scripts when needed. This enables us to remove a
3402 substantial amount of legacy code from PID 1, following the
3403 fact that many distributions only ship a very small number
3404 of LSB/SysV init scripts nowadays.
3406 * Privileged Xen (dom0) domains are not considered
3407 virtualization anymore by the virtualization detection
3408 logic. After all, they generally have unrestricted access to
3409 the hardware and usually are used to manage the unprivileged
3412 * systemd-tmpfiles gained a new "C" line type, for copying
3413 files or entire directories.
3415 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
3416 lines. So far, they have been non-globbing versions of the
3417 latter, and have thus been redundant. In future, it is
3418 recommended to only use "z". "m" has hence been removed
3419 from the documentation, even though it stays supported.
3421 * A tmpfiles snippet to recreate the most basic structure in
3422 /var has been added. This is enough to create the /var/run →
3423 /run symlink and create a couple of structural
3424 directories. This allows systems to boot up with an empty or
3425 volatile /var. Of course, while with this change, the core OS
3426 now is capable with dealing with a volatile /var, not all
3427 user services are ready for it. However, we hope that sooner
3428 or later, many service daemons will be changed upstream so
3429 that they are able to automatically create their necessary
3430 directories in /var at boot, should they be missing. This is
3431 the first step to allow state-less systems that only require
3432 the vendor image for /usr to boot.
3434 * systemd-nspawn has gained a new --tmpfs= switch to mount an
3435 empty tmpfs instance to a specific directory. This is
3436 particularly useful for making use of the automatic
3437 reconstruction of /var (see above), by passing --tmpfs=/var.
3439 * Access modes specified in tmpfiles snippets may now be
3440 prefixed with "~", which indicates that they shall be masked
3441 by whether the existing file or directory is currently
3442 writable, readable or executable at all. Also, if specified,
3443 the sgid/suid/sticky bits will be masked for all
3446 * A new passive target unit "network-pre.target" has been
3447 added which is useful for services that shall run before any
3448 network is configured, for example firewall scripts.
3450 * The "floppy" group that previously owned the /dev/fd*
3451 devices is no longer used. The "disk" group is now used
3452 instead. Distributions should probably deprecate usage of
3455 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
3456 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
3457 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
3458 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
3459 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
3460 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
3463 — Berlin, 2014-06-11
3467 * A new "systemd-timesyncd" daemon has been added for
3468 synchronizing the system clock across the network. It
3469 implements an SNTP client. In contrast to NTP
3470 implementations such as chrony or the NTP reference server,
3471 this only implements a client side, and does not bother with
3472 the full NTP complexity, focusing only on querying time from
3473 one remote server and synchronizing the local clock to
3474 it. Unless you intend to serve NTP to networked clients or
3475 want to connect to local hardware clocks, this simple NTP
3476 client should be more than appropriate for most
3477 installations. The daemon runs with minimal privileges, and
3478 has been hooked up with networkd to only operate when
3479 network connectivity is available. The daemon saves the
3480 current clock to disk every time a new NTP sync has been
3481 acquired, and uses this to possibly correct the system clock
3482 early at bootup, in order to accommodate for systems that
3483 lack an RTC such as the Raspberry Pi and embedded devices,
3484 and to make sure that time monotonically progresses on these
3485 systems, even if it is not always correct. To make use of
3486 this daemon, a new system user and group "systemd-timesync"
3487 needs to be created on installation of systemd.
3489 * The queue "seqnum" interface of libudev has been disabled, as
3490 it was generally incompatible with device namespacing as
3491 sequence numbers of devices go "missing" if the devices are
3492 part of a different namespace.
3494 * "systemctl list-timers" and "systemctl list-sockets" gained
3495 a --recursive switch for showing units of these types also
3496 for all local containers, similar in style to the already
3497 supported --recursive switch for "systemctl list-units".
3499 * A new RebootArgument= setting has been added for service
3500 units, which may be used to specify a kernel reboot argument
3501 to use when triggering reboots with StartLimitAction=.
3503 * A new FailureAction= setting has been added for service
3504 units which may be used to specify an operation to trigger
3505 when a service fails. This works similarly to
3506 StartLimitAction=, but unlike it, controls what is done
3507 immediately rather than only after several attempts to
3508 restart the service in question.
3510 * hostnamed got updated to also expose the kernel name,
3511 release, and version on the bus. This is useful for
3512 executing commands like hostnamectl with the -H switch.
3513 systemd-analyze makes use of this to properly display
3514 details when running non-locally.
3516 * The bootchart tool can now show cgroup information in the
3517 graphs it generates.
3519 * The CFS CPU quota cgroup attribute is now exposed for
3520 services. The new CPUQuota= switch has been added for this
3521 which takes a percentage value. Setting this will have the
3522 result that a service may never get more CPU time than the
3523 specified percentage, even if the machine is otherwise idle.
3525 * systemd-networkd learned IPIP and SIT tunnel support.
3527 * LSB init scripts exposing a dependency on $network will now
3528 get a dependency on network-online.target rather than simply
3529 network.target. This should bring LSB handling closer to
3530 what it was on SysV systems.
3532 * A new fsck.repair= kernel option has been added to control
3533 how fsck shall deal with unclean file systems at boot.
3535 * The (.ini) configuration file parser will now silently
3536 ignore sections whose name begins with "X-". This may be
3537 used to maintain application-specific extension sections in unit
3540 * machined gained a new API to query the IP addresses of
3541 registered containers. "machinectl status" has been updated
3542 to show these addresses in its output.
3544 * A new call sd_uid_get_display() has been added to the
3545 sd-login APIs for querying the "primary" session of a
3546 user. The "primary" session of the user is elected from the
3547 user's sessions and generally a graphical session is
3548 preferred over a text one.
3550 * A minimal systemd-resolved daemon has been added. It
3551 currently simply acts as a companion to systemd-networkd and
3552 manages resolv.conf based on per-interface DNS
3553 configuration, possibly supplied via DHCP. In the long run
3554 we hope to extend this into a local DNSSEC enabled DNS and
3557 * The systemd-networkd-wait-online tool is now enabled by
3558 default. It will delay network-online.target until a network
3559 connection has been configured. The tool primarily integrates
3560 with networkd, but will also make a best effort to make sense
3561 of network configuration performed in some other way.
3563 * Two new service options StartupCPUShares= and
3564 StartupBlockIOWeight= have been added that work similarly to
3565 CPUShares= and BlockIOWeight= however only apply during
3566 system startup. This is useful to prioritize certain services
3567 differently during bootup than during normal runtime.
3569 * hostnamed has been changed to prefer the statically
3570 configured hostname in /etc/hostname (unless set to
3571 'localhost' or empty) over any dynamic one supplied by
3572 dhcp. With this change, the rules for picking the hostname
3573 match more closely the rules of other configuration settings
3574 where the local administrator's configuration in /etc always
3575 overrides any other settings.
3577 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
3578 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
3579 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
3580 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
3581 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
3582 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
3583 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
3584 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
3585 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
3586 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
3587 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
3588 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
3589 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
3590 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
3591 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
3592 Lindskog, WaLyong Cho, Will Woods, Zbigniew
3595 — Beijing, 2014-05-28
3599 * When restoring the screen brightness at boot, stay away from
3600 the darkest setting or from the lowest 5% of the available
3601 range, depending on which is the larger value of both. This
3602 should effectively protect the user from rebooting into a
3603 black screen, should the brightness have been set to minimum
3606 * sd-login gained a new sd_machine_get_class() call to
3607 determine the class ("vm" or "container") of a machine
3608 registered with machined.
3610 * sd-login gained new calls
3611 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
3612 to query the identity of the peer of a local AF_UNIX
3613 connection. They operate similarly to their sd_pid_get_xyz()
3616 * PID 1 will now maintain a system-wide system state engine
3617 with the states "starting", "running", "degraded",
3618 "maintenance", "stopping". These states are bound to system
3619 startup, normal runtime, runtime with at least one failed
3620 service, rescue/emergency mode and system shutdown. This
3621 state is shown in the "systemctl status" output when no unit
3622 name is passed. It is useful to determine system state, in
3623 particularly when doing so for many systems or containers at
3626 * A new command "list-machines" has been added to "systemctl"
3627 that lists all local OS containers and shows their system
3628 state (see above), if systemd runs inside of them.
3630 * systemctl gained a new "-r" switch to recursively enumerate
3631 units on all local containers, when used with the
3632 "list-unit" command (which is the default one that is
3633 executed when no parameters are specified).
3635 * The GPT automatic partition discovery logic will now honour
3636 two GPT partition flags: one may be set on a partition to
3637 cause it to be mounted read-only, and the other may be set
3638 on a partition to ignore it during automatic discovery.
3640 * Two new GPT type UUIDs have been added for automatic root
3641 partition discovery, for 32-bit and 64-bit ARM. This is not
3642 particularly useful for discovering the root directory on
3643 these architectures during bare-metal boots (since UEFI is
3644 not common there), but still very useful to allow booting of
3645 ARM disk images in nspawn with the -i option.
3647 * MAC addresses of interfaces created with nspawn's
3648 --network-interface= switch will now be generated from the
3649 machine name, and thus be stable between multiple invocations
3652 * logind will now automatically remove all IPC objects owned
3653 by a user if she or he fully logs out. This makes sure that
3654 users who are logged out cannot continue to consume IPC
3655 resources. This covers SysV memory, semaphores and message
3656 queues as well as POSIX shared memory and message
3657 queues. Traditionally, SysV and POSIX IPC had no life-cycle
3658 limits. With this functionality, that is corrected. This may
3659 be turned off by using the RemoveIPC= switch of logind.conf.
3661 * The systemd-machine-id-setup and tmpfiles tools gained a
3662 --root= switch to operate on a specific root directory,
3665 * journald can now forward logged messages to the TTYs of all
3666 logged in users ("wall"). This is the default for all
3667 emergency messages now.
3669 * A new tool systemd-journal-remote has been added to stream
3670 journal log messages across the network.
3672 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
3673 controller trees are mounted into it. Note that the
3674 directories mounted beneath it are not read-only. This is a
3675 security measure and is particularly useful because glibc
3676 actually includes a search logic to pick any tmpfs it can
3677 find to implement shm_open() if /dev/shm is not available
3678 (which it might very well be in namespaced setups).
3680 * machinectl gained a new "poweroff" command to cleanly power
3681 down a local OS container.
3683 * The PrivateDevices= unit file setting will now also drop the
3684 CAP_MKNOD capability from the capability bound set, and
3685 imply DevicePolicy=closed.
3687 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
3688 comprehensively on all long-running systemd services where
3689 this is appropriate.
3691 * systemd-udevd will now run in a disassociated mount
3692 namespace. To mount directories from udev rules, make sure to
3693 pull in mount units via SYSTEMD_WANTS properties.
3695 * The kdbus support gained support for uploading policy into
3696 the kernel. sd-bus gained support for creating "monitoring"
3697 connections that can eavesdrop into all bus communication
3698 for debugging purposes.
3700 * Timestamps may now be specified in seconds since the UNIX
3701 epoch Jan 1st, 1970 by specifying "@" followed by the value
3704 * Native tcpwrap support in systemd has been removed. tcpwrap
3705 is old code, not really maintained anymore and has serious
3706 shortcomings, and better options such as firewalls
3707 exist. For setups that require tcpwrap usage, please
3708 consider invoking your socket-activated service via tcpd,
3709 like on traditional inetd.
3711 * A new system.conf configuration option
3712 DefaultTimerAccuracySec= has been added that controls the
3713 default AccuracySec= setting of .timer units.
3715 * Timer units gained a new WakeSystem= switch. If enabled,
3716 timers configured this way will cause the system to resume
3717 from system suspend (if the system supports that, which most
3720 * Timer units gained a new Persistent= switch. If enabled,
3721 timers configured this way will save to disk when they have
3722 been last triggered. This information is then used on next
3723 reboot to possible execute overdue timer events, that
3724 could not take place because the system was powered off.
3725 This enables simple anacron-like behaviour for timer units.
3727 * systemctl's "list-timers" will now also list the time a
3728 timer unit was last triggered in addition to the next time
3729 it will be triggered.
3731 * systemd-networkd will now assign predictable IPv4LL
3732 addresses to its local interfaces.
3734 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
3735 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
3736 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
3737 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
3738 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
3739 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
3740 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
3741 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
3744 — Berlin, 2014-03-25
3748 * A new unit file setting RestrictAddressFamilies= has been
3749 added to restrict which socket address families unit
3750 processes gain access to. This takes address family names
3751 like "AF_INET" or "AF_UNIX", and is useful to minimize the
3752 attack surface of services via exotic protocol stacks. This
3753 is built on seccomp system call filters.
3755 * Two new unit file settings RuntimeDirectory= and
3756 RuntimeDirectoryMode= have been added that may be used to
3757 manage a per-daemon runtime directories below /run. This is
3758 an alternative for setting up directory permissions with
3759 tmpfiles snippets, and has the advantage that the runtime
3760 directory's lifetime is bound to the daemon runtime and that
3761 the daemon starts up with an empty directory each time. This
3762 is particularly useful when writing services that drop
3763 privileges using the User= or Group= setting.
3765 * The DeviceAllow= unit setting now supports globbing for
3766 matching against device group names.
3768 * The systemd configuration file system.conf gained new
3769 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
3770 DefaultMemoryAccounting= to globally turn on/off accounting
3771 for specific resources (cgroups) for all units. These
3772 settings may still be overridden individually in each unit
3775 * systemd-gpt-auto-generator is now able to discover /srv and
3776 root partitions in addition to /home and swap partitions. It
3777 also supports LUKS-encrypted partitions now. With this in
3778 place, automatic discovery of partitions to mount following
3779 the Discoverable Partitions Specification
3780 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
3781 is now a lot more complete. This allows booting without
3782 /etc/fstab and without root= on the kernel command line on
3783 systems prepared appropriately.
3785 * systemd-nspawn gained a new --image= switch which allows
3786 booting up disk images and Linux installations on any block
3787 device that follow the Discoverable Partitions Specification
3788 (see above). This means that installations made with
3789 appropriately updated installers may now be started and
3790 deployed using container managers, completely
3791 unmodified. (We hope that libvirt-lxc will add support for
3792 this feature soon, too.)
3794 * systemd-nspawn gained a new --network-macvlan= setting to
3795 set up a private macvlan interface for the
3796 container. Similarly, systemd-networkd gained a new
3797 Kind=macvlan setting in .netdev files.
3799 * systemd-networkd now supports configuring local addresses
3802 * A new tool systemd-network-wait-online has been added to
3803 synchronously wait for network connectivity using
3806 * The sd-bus.h bus API gained a new sd_bus_track object for
3807 tracking the life-cycle of bus peers. Note that sd-bus.h is
3808 still not a public API though (unless you specify
3809 --enable-kdbus on the configure command line, which however
3810 voids your warranty and you get no API stability guarantee).
3812 * The $XDG_RUNTIME_DIR runtime directories for each user are
3813 now individual tmpfs instances, which has the benefit of
3814 introducing separate pools for each user, with individual
3815 size limits, and thus making sure that unprivileged clients
3816 can no longer negatively impact the system or other users by
3817 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
3818 RuntimeDirectorySize= has been introduced that allows
3819 controlling the default size limit for all users. It
3820 defaults to 10% of the available physical memory. This is no
3821 replacement for quotas on tmpfs though (which the kernel
3822 still does not support), as /dev/shm and /tmp are still
3823 shared resources used by both the system and unprivileged
3826 * logind will now automatically turn off automatic suspending
3827 on laptop lid close when more than one display is
3828 connected. This was previously expected to be implemented
3829 individually in desktop environments (such as GNOME),
3830 however has been added to logind now, in order to fix a
3831 boot-time race where a desktop environment might not have
3832 been started yet and thus not been able to take an inhibitor
3833 lock at the time where logind already suspends the system
3834 due to a closed lid.
3836 * logind will now wait at least 30s after each system
3837 suspend/resume cycle, and 3min after system boot before
3838 suspending the system due to a closed laptop lid. This
3839 should give USB docking stations and similar enough time to
3840 be probed and configured after system resume and boot in
3841 order to then act as suspend blocker.
3843 * systemd-run gained a new --property= setting which allows
3844 initialization of resource control properties (and others)
3845 for the created scope or service unit. Example: "systemd-run
3846 --property=BlockIOWeight=10 updatedb" may be used to run
3847 updatedb at a low block IO scheduling weight.
3849 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
3850 now also work in --scope mode.
3852 * When systemd is compiled with kdbus support, basic support
3853 for enforced policies is now in place. (Note that enabling
3854 kdbus still voids your warranty and no API compatibility
3857 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
3858 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
3859 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
3860 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
3861 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
3862 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
3863 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
3864 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
3865 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
3866 Zbigniew Jędrzejewski-Szmek
3868 — Berlin, 2014-03-12
3872 * systemd will now relabel /dev after loading the SMACK policy
3873 according to SMACK rules.
3875 * A new unit file option AppArmorProfile= has been added to
3876 set the AppArmor profile for the processes of a unit.
3878 * A new condition check ConditionArchitecture= has been added
3879 to conditionalize units based on the system architecture, as
3880 reported by uname()'s "machine" field.
3882 * systemd-networkd now supports matching on the system
3883 virtualization, architecture, kernel command line, host name
3886 * logind is now a lot more aggressive when suspending the
3887 machine due to a closed laptop lid. Instead of acting only
3888 on the lid close action, it will continuously watch the lid
3889 status and act on it. This is useful for laptops where the
3890 power button is on the outside of the chassis so that it can
3891 be reached without opening the lid (such as the Lenovo
3892 Yoga). On those machines, logind will now immediately
3893 re-suspend the machine if the power button has been
3894 accidentally pressed while the laptop was suspended and in a
3895 backpack or similar.
3897 * logind will now watch SW_DOCK switches and inhibit reaction
3898 to the lid switch if it is pressed. This means that logind
3899 will not suspend the machine anymore if the lid is closed
3900 and the system is docked, if the laptop supports SW_DOCK
3901 notifications via the input layer. Note that ACPI docking
3902 stations do not generate this currently. Also note that this
3903 logic is usually not fully sufficient and Desktop
3904 Environments should take a lid switch inhibitor lock when an
3905 external display is connected, as systemd will not watch
3908 * nspawn will now make use of the devices cgroup controller by
3909 default, and only permit creation of and access to the usual
3910 API device nodes like /dev/null or /dev/random, as well as
3911 access to (but not creation of) the pty devices.
3913 * We will now ship a default .network file for
3914 systemd-networkd that automatically configures DHCP for
3915 network interfaces created by nspawn's --network-veth or
3916 --network-bridge= switches.
3918 * systemd will now understand the usual M, K, G, T suffixes
3919 according to SI conventions (i.e. to the base 1000) when
3920 referring to throughput and hardware metrics. It will stay
3921 with IEC conventions (i.e. to the base 1024) for software
3922 metrics, according to what is customary according to
3923 Wikipedia. We explicitly document which base applies for
3924 each configuration option.
3926 * The DeviceAllow= setting in unit files now supports a syntax
3927 to whitelist an entire group of devices node majors at once,
3928 based on the /proc/devices listing. For example, with the
3929 string "char-pts", it is now possible to whitelist all
3930 current and future pseudo-TTYs at once.
3932 * sd-event learned a new "post" event source. Event sources of
3933 this type are triggered by the dispatching of any event
3934 source of a type that is not "post". This is useful for
3935 implementing clean-up and check event sources that are
3936 triggered by other work being done in the program.
3938 * systemd-networkd is no longer statically enabled, but uses
3939 the usual [Install] sections so that it can be
3940 enabled/disabled using systemctl. It still is enabled by
3943 * When creating a veth interface pair with systemd-nspawn, the
3944 host side will now be prefixed with "vb-" if
3945 --network-bridge= is used, and with "ve-" if --network-veth
3946 is used. This way, it is easy to distinguish these cases on
3947 the host, for example to apply different configuration to
3948 them with systemd-networkd.
3950 * The compatibility libraries for libsystemd-journal.so,
3951 libsystem-id128.so, libsystemd-login.so and
3952 libsystemd-daemon.so do not make use of IFUNC
3953 anymore. Instead, we now build libsystemd.so multiple times
3954 under these alternative names. This means that the footprint
3955 is drastically increased, but given that these are
3956 transitional compatibility libraries, this should not matter
3957 much. This change has been made necessary to support the ARM
3958 platform for these compatibility libraries, as the ARM
3959 toolchain is not really at the same level as the toolchain
3960 for other architectures like x86 and does not support
3961 IFUNC. Please make sure to use --enable-compat-libs only
3962 during a transitional period!
3964 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
3965 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
3966 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
3967 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
3968 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
3969 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
3970 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
3971 Zbigniew Jędrzejewski-Szmek
3973 — Berlin, 2014-02-24
3977 * A new component "systemd-networkd" has been added that can
3978 be used to configure local network interfaces statically or
3979 via DHCP. It is capable of bringing up bridges, VLANs, and
3980 bonding. Currently, no hook-ups for interactive network
3981 configuration are provided. Use this for your initrd,
3982 container, embedded, or server setup if you need a simple,
3983 yet powerful, network configuration solution. This
3984 configuration subsystem is quite nifty, as it allows wildcard
3985 hotplug matching in interfaces. For example, with a single
3986 configuration snippet, you can configure that all Ethernet
3987 interfaces showing up are automatically added to a bridge,
3988 or similar. It supports link-sensing and more.
3990 * A new tool "systemd-socket-proxyd" has been added which can
3991 act as a bidirectional proxy for TCP sockets. This is
3992 useful for adding socket activation support to services that
3993 do not actually support socket activation, including virtual
3994 machines and the like.
3996 * Add a new tool to save/restore rfkill state on
3999 * Save/restore state of keyboard backlights in addition to
4000 display backlights on shutdown/boot.
4002 * udev learned a new SECLABEL{} construct to label device
4003 nodes with a specific security label when they appear. For
4004 now, only SECLABEL{selinux} is supported, but the syntax is
4005 prepared for additional security frameworks.
4007 * udev gained a new scheme to configure link-level attributes
4008 from files in /etc/systemd/network/*.link. These files can
4009 match against MAC address, device path, driver name and type,
4010 and will apply attributes like the naming policy, link speed,
4011 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
4012 address assignment policy (randomized, ...).
4014 * The configuration of network interface naming rules for
4015 "permanent interface names" has changed: a new NamePolicy=
4016 setting in the [Link] section of .link files determines the
4017 priority of possible naming schemes (onboard, slot, MAC,
4018 path). The default value of this setting is determined by
4019 /usr/lib/net/links/99-default.link. Old
4020 80-net-name-slot.rules udev configuration file has been
4021 removed, so local configuration overriding this file should
4022 be adapted to override 99-default.link instead.
4024 * When the User= switch is used in a unit file, also
4025 initialize $SHELL= based on the user database entry.
4027 * systemd no longer depends on libdbus. All communication is
4028 now done with sd-bus, systemd's low-level bus library
4031 * kdbus support has been added to PID 1 itself. When kdbus is
4032 enabled, this causes PID 1 to set up the system bus and
4033 enable support for a new ".busname" unit type that
4034 encapsulates bus name activation on kdbus. It works a little
4035 bit like ".socket" units, except for bus names. A new
4036 generator has been added that converts classic dbus1 service
4037 activation files automatically into native systemd .busname
4040 * sd-bus: add a light-weight vtable implementation that allows
4041 defining objects on the bus with a simple static const
4042 vtable array of its methods, signals and properties.
4044 * systemd will not generate or install static dbus
4045 introspection data anymore to /usr/share/dbus-1/interfaces,
4046 as the precise format of these files is unclear, and
4047 nothing makes use of it.
4049 * A proxy daemon is now provided to proxy clients connecting
4050 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
4051 compatibility with classic D-Bus.
4053 * A bus driver implementation has been added that supports the
4054 classic D-Bus bus driver calls on kdbus, also for
4055 compatibility purposes.
4057 * A new API "sd-event.h" has been added that implements a
4058 minimal event loop API built around epoll. It provides a
4059 couple of features that direct epoll usage is lacking:
4060 prioritization of events, scales to large numbers of timer
4061 events, per-event timer slack (accuracy), system-wide
4062 coalescing of timer events, exit handlers, watchdog
4063 supervision support using systemd's sd_notify() API, child
4066 * A new API "sd-rntl.h" has been added that provides an API
4067 around the route netlink interface of the kernel, similar in
4068 style to "sd-bus.h".
4070 * A new API "sd-dhcp-client.h" has been added that provides a
4071 small DHCPv4 client-side implementation. This is used by
4074 * There is a new kernel command line option
4075 "systemd.restore_state=0|1". When set to "0", none of the
4076 systemd tools will restore saved runtime state to hardware
4077 devices. More specifically, the rfkill and backlight states
4080 * The FsckPassNo= compatibility option in mount/service units
4081 has been removed. The fstab generator will now add the
4082 necessary dependencies automatically, and does not require
4083 PID1's support for that anymore.
4085 * journalctl gained a new switch, --list-boots, that lists
4086 recent boots with their times and boot IDs.
4088 * The various tools like systemctl, loginctl, timedatectl,
4089 busctl, systemd-run, ... have gained a new switch "-M" to
4090 connect to a specific, local OS container (as direct
4091 connection, without requiring SSH). This works on any
4092 container that is registered with machined, such as those
4093 created by libvirt-lxc or nspawn.
4095 * systemd-run and systemd-analyze also gained support for "-H"
4096 to connect to remote hosts via SSH. This is particularly
4097 useful for systemd-run because it enables queuing of jobs
4098 onto remote systems.
4100 * machinectl gained a new command "login" to open a getty
4101 login in any local container. This works with any container
4102 that is registered with machined (such as those created by
4103 libvirt-lxc or nspawn), and which runs systemd inside.
4105 * machinectl gained a new "reboot" command that may be used to
4106 trigger a reboot on a specific container that is registered
4107 with machined. This works on any container that runs an init
4108 system of some kind.
4110 * systemctl gained a new "list-timers" command to print a nice
4111 listing of installed timer units with the times they elapse
4114 * Alternative reboot() parameters may now be specified on the
4115 "systemctl reboot" command line and are passed to the
4116 reboot() system call.
4118 * systemctl gained a new --job-mode= switch to configure the
4119 mode to queue a job with. This is a more generic version of
4120 --fail, --irreversible, and --ignore-dependencies, which are
4121 still available but not advertised anymore.
4123 * /etc/systemd/system.conf gained new settings to configure
4124 various default timeouts of units, as well as the default
4125 start limit interval and burst. These may still be overridden
4128 * PID1 will now export on the bus profile data of the security
4129 policy upload process (such as the SELinux policy upload to
4132 * journald: when forwarding logs to the console, include
4133 timestamps (following the setting in
4134 /sys/module/printk/parameters/time).
4136 * OnCalendar= in timer units now understands the special
4137 strings "yearly" and "annually". (Both are equivalent)
4139 * The accuracy of timer units is now configurable with the new
4140 AccuracySec= setting. It defaults to 1min.
4142 * A new dependency type JoinsNamespaceOf= has been added that
4143 allows running two services within the same /tmp and network
4144 namespace, if PrivateNetwork= or PrivateTmp= are used.
4146 * A new command "cat" has been added to systemctl. It outputs
4147 the original unit file of a unit, and concatenates the
4148 contents of additional "drop-in" unit file snippets, so that
4149 the full configuration is shown.
4151 * systemctl now supports globbing on the various "list-xyz"
4152 commands, like "list-units" or "list-sockets", as well as on
4153 those commands which take multiple unit names.
4155 * journalctl's --unit= switch gained support for globbing.
4157 * All systemd daemons now make use of the watchdog logic so
4158 that systemd automatically notices when they hang.
4160 * If the $container_ttys environment variable is set,
4161 getty-generator will automatically spawn a getty for each
4162 listed tty. This is useful for container managers to request
4163 login gettys to be spawned on as many ttys as needed.
4165 * %h, %s, %U specifier support is not available anymore when
4166 used in unit files for PID 1. This is because NSS calls are
4167 not safe from PID 1. They stay available for --user
4168 instances of systemd, and as special case for the root user.
4170 * loginctl gained a new "--no-legend" switch to turn off output
4173 * The "sd-login.h" API gained three new calls:
4174 sd_session_is_remote(), sd_session_get_remote_user(),
4175 sd_session_get_remote_host() to query information about
4178 * The udev hardware database now also carries vendor/product
4179 information of SDIO devices.
4181 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
4182 determine whether watchdog notifications are requested by
4185 * Socket-activated per-connection services now include a
4186 short description of the connection parameters in the
4189 * tmpfiles gained a new "--boot" option. When this is not used,
4190 only lines where the command character is not suffixed with
4191 "!" are executed. When this option is specified, those
4192 options are executed too. This partitions tmpfiles
4193 directives into those that can be safely executed at any
4194 time, and those which should be run only at boot (for
4195 example, a line that creates /run/nologin).
4197 * A new API "sd-resolve.h" has been added which provides a simple
4198 asynchronous wrapper around glibc NSS host name resolution
4199 calls, such as getaddrinfo(). In contrast to glibc's
4200 getaddrinfo_a(), it does not use signals. In contrast to most
4201 other asynchronous name resolution libraries, this one does
4202 not reimplement DNS, but reuses NSS, so that alternate
4203 host name resolution systems continue to work, such as mDNS,
4204 LDAP, etc. This API is based on libasyncns, but it has been
4205 cleaned up for inclusion in systemd.
4207 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
4208 "sd-daemon.h" are no longer found in individual libraries
4209 libsystemd-journal.so, libsystemd-login.so,
4210 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
4211 merged them into a single library, libsystemd.so, which
4212 provides all symbols. The reason for this is cyclic
4213 dependencies, as these libraries tend to use each other's
4214 symbols. So far, we have managed to workaround that by linking
4215 a copy of a good part of our code into each of these
4216 libraries again and again, which, however, makes certain
4217 things hard to do, like sharing static variables. Also, it
4218 substantially increases footprint. With this change, there
4219 is only one library for the basic APIs systemd
4220 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
4221 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
4222 library as well, however are subject to the --enable-kdbus
4223 switch (see below). Note that "sd-dhcp-client.h" is not part
4224 of this library (this is because it only consumes, never
4225 provides, services of/to other APIs). To make the transition
4226 easy from the separate libraries to the unified one, we
4227 provide the --enable-compat-libs compile-time switch which
4228 will generate stub libraries that are compatible with the
4229 old ones but redirect all calls to the new one.
4231 * All of the kdbus logic and the new APIs "sd-bus.h",
4232 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
4233 and "sd-utf8.h" are compile-time optional via the
4234 "--enable-kdbus" switch, and they are not compiled in by
4235 default. To make use of kdbus, you have to explicitly enable
4236 the switch. Note however, that neither the kernel nor the
4237 userspace API for all of this is considered stable yet. We
4238 want to maintain the freedom to still change the APIs for
4239 now. By specifying this build-time switch, you acknowledge
4240 that you are aware of the instability of the current
4243 * Also, note that while kdbus is pretty much complete,
4244 it lacks one thing: proper policy support. This means you
4245 can build a fully working system with all features; however,
4246 it will be highly insecure. Policy support will be added in
4247 one of the next releases, at the same time that we will
4248 declare the APIs stable.
4250 * When the kernel command line argument "kdbus" is specified,
4251 systemd will automatically load the kdbus.ko kernel module. At
4252 this stage of development, it is only useful for testing kdbus
4253 and should not be used in production. Note: if "--enable-kdbus"
4254 is specified, and the kdbus.ko kernel module is available, and
4255 "kdbus" is added to the kernel command line, the entire system
4256 runs with kdbus instead of dbus-daemon, with the above mentioned
4257 problem of missing the system policy enforcement. Also a future
4258 version of kdbus.ko or a newer systemd will not be compatible with
4259 each other, and will unlikely be able to boot the machine if only
4260 one of them is updated.
4262 * systemctl gained a new "import-environment" command which
4263 uploads the caller's environment (or parts thereof) into the
4264 service manager so that it is inherited by services started
4265 by the manager. This is useful to upload variables like
4266 $DISPLAY into the user service manager.
4268 * A new PrivateDevices= switch has been added to service units
4269 which allows running a service with a namespaced /dev
4270 directory that does not contain any device nodes for
4271 physical devices. More specifically, it only includes devices
4272 such as /dev/null, /dev/urandom, and /dev/zero which are API
4275 * logind has been extended to support behaviour like VT
4276 switching on seats that do not support a VT. This makes
4277 multi-session available on seats that are not the first seat
4278 (seat0), and on systems where kernel support for VTs has
4279 been disabled at compile-time.
4281 * If a process holds a delay lock for system sleep or shutdown
4282 and fails to release it in time, we will now log its
4283 identity. This makes it easier to identify processes that
4284 cause slow suspends or power-offs.
4286 * When parsing /etc/crypttab, support for a new key-slot=
4287 option as supported by Debian is added. It allows indicating
4288 which LUKS slot to use on disk, speeding up key loading.
4290 * The sd_journald_sendv() API call has been checked and
4291 officially declared to be async-signal-safe so that it may
4292 be invoked from signal handlers for logging purposes.
4294 * Boot-time status output is now enabled automatically after a
4295 short timeout if boot does not progress, in order to give
4296 the user an indication what she or he is waiting for.
4298 * The boot-time output has been improved to show how much time
4299 remains until jobs expire.
4301 * The KillMode= switch in service units gained a new possible
4302 value "mixed". If set, and the unit is shut down, then the
4303 initial SIGTERM signal is sent only to the main daemon
4304 process, while the following SIGKILL signal is sent to
4305 all remaining processes of the service.
4307 * When a scope unit is registered, a new property "Controller"
4308 may be set. If set to a valid bus name, systemd will send a
4309 RequestStop() signal to this name when it would like to shut
4310 down the scope. This may be used to hook manager logic into
4311 the shutdown logic of scope units. Also, scope units may now
4312 be put in a special "abandoned" state, in which case the
4313 manager process which created them takes no further
4314 responsibilities for it.
4316 * When reading unit files, systemd will now verify
4317 the access mode of these files, and warn about certain
4318 suspicious combinations. This has been added to make it
4319 easier to track down packaging bugs where unit files are
4320 marked executable or world-writable.
4322 * systemd-nspawn gained a new "--setenv=" switch to set
4323 container-wide environment variables. The similar option in
4324 systemd-activate was renamed from "--environment=" to
4325 "--setenv=" for consistency.
4327 * systemd-nspawn has been updated to create a new kdbus domain
4328 for each container that is invoked, thus allowing each
4329 container to have its own set of system and user buses,
4330 independent of the host.
4332 * systemd-nspawn gained a new --drop-capability= switch to run
4333 the container with less capabilities than the default. Both
4334 --drop-capability= and --capability= now take the special
4335 string "all" for dropping or keeping all capabilities.
4337 * systemd-nspawn gained new switches for executing containers
4338 with specific SELinux labels set.
4340 * systemd-nspawn gained a new --quiet switch to not generate
4341 any additional output but the container's own console
4344 * systemd-nspawn gained a new --share-system switch to run a
4345 container without PID namespacing enabled.
4347 * systemd-nspawn gained a new --register= switch to control
4348 whether the container is registered with systemd-machined or
4349 not. This is useful for containers that do not run full
4350 OS images, but only specific apps.
4352 * systemd-nspawn gained a new --keep-unit which may be used
4353 when invoked as the only program from a service unit, and
4354 results in registration of the unit service itself in
4355 systemd-machined, instead of a newly opened scope unit.
4357 * systemd-nspawn gained a new --network-interface= switch for
4358 moving arbitrary interfaces to the container. The new
4359 --network-veth switch creates a virtual Ethernet connection
4360 between host and container. The new --network-bridge=
4361 switch then allows assigning the host side of this virtual
4362 Ethernet connection to a bridge device.
4364 * systemd-nspawn gained a new --personality= switch for
4365 setting the kernel personality for the container. This is
4366 useful when running a 32-bit container on a 64-bit host. A
4367 similar option Personality= is now also available for service
4370 * logind will now also track a "Desktop" identifier for each
4371 session which encodes the desktop environment of it. This is
4372 useful for desktop environments that want to identify
4373 multiple running sessions of itself easily.
4375 * A new SELinuxContext= setting for service units has been
4376 added that allows setting a specific SELinux execution
4377 context for a service.
4379 * Most systemd client tools will now honour $SYSTEMD_LESS for
4380 settings of the "less" pager. By default, these tools will
4381 override $LESS to allow certain operations to work, such as
4382 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
4383 influence this logic.
4385 * systemd's "seccomp" hook-up has been changed to make use of
4386 the libseccomp library instead of using its own
4387 implementation. This has benefits for portability among
4390 * For usage together with SystemCallFilter=, a new
4391 SystemCallErrorNumber= setting has been introduced that
4392 allows configuration of a system error number to be returned
4393 on filtered system calls, instead of immediately killing the
4394 process. Also, SystemCallArchitectures= has been added to
4395 limit access to system calls of a particular architecture
4396 (in order to turn off support for unused secondary
4397 architectures). There is also a global
4398 SystemCallArchitectures= setting in system.conf now to turn
4399 off support for non-native system calls system-wide.
4401 * systemd requires a kernel with a working name_to_handle_at(),
4402 please see the kernel config requirements in the README file.
4404 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
4405 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
4406 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
4407 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
4408 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
4409 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
4410 Elia Pinto, Florian Weimer, George McCollister, Goffredo
4411 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
4412 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
4413 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
4414 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
4415 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
4416 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
4417 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
4418 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
4419 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
4420 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
4421 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
4422 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
4423 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
4424 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
4425 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
4426 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
4427 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
4429 — Berlin, 2014-02-20
4433 * logind has gained support for facilitating privileged input
4434 and drm device access for unprivileged clients. This work is
4435 useful to allow Wayland display servers (and similar
4436 programs, such as kmscon) to run under the user's ID and
4437 access input and drm devices which are normally
4438 protected. When this is used (and the kernel is new enough)
4439 logind will "mute" IO on the file descriptors passed to
4440 Wayland as long as it is in the background and "unmute" it
4441 if it returns into the foreground. This allows secure
4442 session switching without allowing background sessions to
4443 eavesdrop on input and display data. This also introduces
4444 session switching support if VT support is turned off in the
4445 kernel, and on seats that are not seat0.
4447 * A new kernel command line option luks.options= is understood
4448 now which allows specifying LUKS options for usage for LUKS
4449 encrypted partitions specified with luks.uuid=.
4451 * tmpfiles.d(5) snippets may now use specifier expansion in
4452 path names. More specifically %m, %b, %H, %v, are now
4453 replaced by the local machine id, boot id, hostname, and
4454 kernel version number.
4456 * A new tmpfiles.d(5) command "m" has been introduced which
4457 may be used to change the owner/group/access mode of a file
4458 or directory if it exists, but do nothing if it does not.
4460 * This release removes high-level support for the
4461 MemorySoftLimit= cgroup setting. The underlying kernel
4462 cgroup attribute memory.soft_limit= is currently badly
4463 designed and likely to be removed from the kernel API in its
4464 current form, hence we should not expose it for now.
4466 * The memory.use_hierarchy cgroup attribute is now enabled for
4467 all cgroups systemd creates in the memory cgroup
4468 hierarchy. This option is likely to be come the built-in
4469 default in the kernel anyway, and the non-hierarchical mode
4470 never made much sense in the intrinsically hierarchical
4473 * A new field _SYSTEMD_SLICE= is logged along with all journal
4474 messages containing the slice a message was generated
4475 from. This is useful to allow easy per-customer filtering of
4476 logs among other things.
4478 * systemd-journald will no longer adjust the group of journal
4479 files it creates to the "systemd-journal" group. Instead we
4480 rely on the journal directory to be owned by the
4481 "systemd-journal" group, and its setgid bit set, so that the
4482 kernel file system layer will automatically enforce that
4483 journal files inherit this group assignment. The reason for
4484 this change is that we cannot allow NSS look-ups from
4485 journald which would be necessary to resolve
4486 "systemd-journal" to a numeric GID, because this might
4487 create deadlocks if NSS involves synchronous queries to
4488 other daemons (such as nscd, or sssd) which in turn are
4489 logging clients of journald and might block on it, which
4490 would then dead lock. A tmpfiles.d(5) snippet included in
4491 systemd will make sure the setgid bit and group are
4492 properly set on the journal directory if it exists on every
4493 boot. However, we recommend adjusting it manually after
4494 upgrades too (or from RPM scriptlets), so that the change is
4495 not delayed until next reboot.
4497 * Backlight and random seed files in /var/lib/ have moved into
4498 the /var/lib/systemd/ directory, in order to centralize all
4499 systemd generated files in one directory.
4501 * Boot time performance measurements (as displayed by
4502 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
4503 performance information if that's available to determine how
4504 much time BIOS and boot loader initialization required. With
4505 a sufficiently new BIOS you hence no longer need to boot
4506 with Gummiboot to get access to such information.
4508 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
4509 Cristian Rodríguez, Dave Reisner, David Herrmann, David
4510 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
4511 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
4512 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
4513 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
4514 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
4516 — Berlin, 2013-10-02
4520 * The Restart= option for services now understands a new
4521 on-watchdog setting, which will restart the service
4522 automatically if the service stops sending out watchdog keep
4523 alive messages (as configured with WatchdogSec=).
4525 * The getty generator (which is responsible for bringing up a
4526 getty on configured serial consoles) will no longer only
4527 start a getty on the primary kernel console but on all
4528 others, too. This makes the order in which console= is
4529 specified on the kernel command line less important.
4531 * libsystemd-logind gained a new sd_session_get_vt() call to
4532 retrieve the VT number of a session.
4534 * If the option "tries=0" is set for an entry of /etc/crypttab
4535 its passphrase is queried indefinitely instead of any
4536 maximum number of tries.
4538 * If a service with a configure PID file terminates its PID
4539 file will now be removed automatically if it still exists
4540 afterwards. This should put an end to stale PID files.
4542 * systemd-run will now also take relative binary path names
4543 for execution and no longer insists on absolute paths.
4545 * InaccessibleDirectories= and ReadOnlyDirectories= now take
4546 paths that are optionally prefixed with "-" to indicate that
4547 it should not be considered a failure if they do not exist.
4549 * journalctl -o (and similar commands) now understands a new
4550 output mode "short-precise", it is similar to "short" but
4551 shows timestamps with usec accuracy.
4553 * The option "discard" (as known from Debian) is now
4554 synonymous to "allow-discards" in /etc/crypttab. In fact,
4555 "discard" is preferred now (since it is easier to remember
4558 * Some licensing clean-ups were made, so that more code is now
4559 LGPL-2.1 licensed than before.
4561 * A minimal tool to save/restore the display backlight
4562 brightness across reboots has been added. It will store the
4563 backlight setting as late as possible at shutdown, and
4564 restore it as early as possible during reboot.
4566 * A logic to automatically discover and enable home and swap
4567 partitions on GPT disks has been added. With this in place
4568 /etc/fstab becomes optional for many setups as systemd can
4569 discover certain partitions located on the root disk
4570 automatically. Home partitions are recognized under their
4571 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
4572 partitions are recognized under their GPT type ID
4573 0657fd6da4ab43c484e50933c84b4f4f.
4575 * systemd will no longer pass any environment from the kernel
4576 or initrd to system services. If you want to set an
4577 environment for all services, do so via the kernel command
4578 line systemd.setenv= assignment.
4580 * The systemd-sysctl tool no longer natively reads the file
4581 /etc/sysctl.conf. If desired, the file should be symlinked
4582 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
4583 legacy support by a symlink rather than built-in code, it
4584 also makes the otherwise hidden order of application of the
4585 different files visible. (Note that this partly reverts to a
4586 pre-198 application order of sysctl knobs!)
4588 * The "systemctl set-log-level" and "systemctl dump" commands
4589 have been moved to systemd-analyze.
4591 * systemd-run learned the new --remain-after-exit switch,
4592 which causes the scope unit not to be cleaned up
4593 automatically after the process terminated.
4595 * tmpfiles learned a new --exclude-prefix= switch to exclude
4596 certain paths from operation.
4598 * journald will now automatically flush all messages to disk
4599 as soon as a message at the log level CRIT, ALERT or EMERG
4602 Contributions from: Andrew Cook, Brandon Philips, Christian
4603 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
4604 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
4605 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
4606 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
4607 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
4608 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
4609 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
4610 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
4611 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
4612 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
4613 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
4614 William Giokas, Zbigniew Jędrzejewski-Szmek
4616 — Berlin, 2013-09-13
4620 * The documentation has been updated to cover the various new
4621 concepts introduced with 205.
4623 * Unit files now understand the new %v specifier which
4624 resolves to the kernel version string as returned by "uname
4627 * systemctl now supports filtering the unit list output by
4628 load state, active state and sub state, using the new
4631 * "systemctl status" will now show the results of the
4632 condition checks (like ConditionPathExists= and similar) of
4633 the last start attempts of the unit. They are also logged to
4636 * "journalctl -b" may now be used to look for boot output of a
4637 specific boot. Try "journalctl -b -1" for the previous boot,
4638 but the syntax is substantially more powerful.
4640 * "journalctl --show-cursor" has been added which prints the
4641 cursor string the last shown log line. This may then be used
4642 with the new "journalctl --after-cursor=" switch to continue
4643 browsing logs from that point on.
4645 * "journalctl --force" may now be used to force regeneration
4648 * Creation of "dead" device nodes has been moved from udev
4649 into kmod and tmpfiles. Previously, udev would read the kmod
4650 databases to pre-generate dead device nodes based on meta
4651 information contained in kernel modules, so that these would
4652 be auto-loaded on access rather then at boot. As this
4653 does not really have much to do with the exposing actual
4654 kernel devices to userspace this has always been slightly
4655 alien in the udev codebase. Following the new scheme kmod
4656 will now generate a runtime snippet for tmpfiles from the
4657 module meta information and it now is tmpfiles' job to the
4658 create the nodes. This also allows overriding access and
4659 other parameters for the nodes using the usual tmpfiles
4660 facilities. As side effect this allows us to remove the
4661 CAP_SYS_MKNOD capability bit from udevd entirely.
4663 * logind's device ACLs may now be applied to these "dead"
4664 devices nodes too, thus finally allowing managed access to
4665 devices such as /dev/snd/sequencer without loading the
4666 backing module right-away.
4668 * A new RPM macro has been added that may be used to apply
4669 tmpfiles configuration during package installation.
4671 * systemd-detect-virt and ConditionVirtualization= now can
4672 detect User-Mode-Linux machines (UML).
4674 * journald will now implicitly log the effective capabilities
4675 set of processes in the message metadata.
4677 * systemd-cryptsetup has gained support for TrueCrypt volumes.
4679 * The initrd interface has been simplified (more specifically,
4680 support for passing performance data via environment
4681 variables and fsck results via files in /run has been
4682 removed). These features were non-essential, and are
4683 nowadays available in a much nicer way by having systemd in
4684 the initrd serialize its state and have the hosts systemd
4685 deserialize it again.
4687 * The udev "keymap" data files and tools to apply keyboard
4688 specific mappings of scan to key codes, and force-release
4689 scan code lists have been entirely replaced by a udev
4690 "keyboard" builtin and a hwdb data file.
4692 * systemd will now honour the kernel's "quiet" command line
4693 argument also during late shutdown, resulting in a
4694 completely silent shutdown when used.
4696 * There's now an option to control the SO_REUSEPORT socket
4697 option in .socket units.
4699 * Instance units will now automatically get a per-template
4700 subslice of system.slice unless something else is explicitly
4701 configured. For example, instances of sshd@.service will now
4702 implicitly be placed in system-sshd.slice rather than
4703 system.slice as before.
4705 * Test coverage support may now be enabled at build time.
4707 Contributions from: Dave Reisner, Frederic Crozat, Harald
4708 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
4709 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
4710 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
4711 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
4712 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
4713 Giokas, Zbigniew Jędrzejewski-Szmek
4715 — Berlin, 2013-07-23
4719 * Two new unit types have been introduced:
4721 Scope units are very similar to service units, however, are
4722 created out of pre-existing processes — instead of PID 1
4723 forking off the processes. By using scope units it is
4724 possible for system services and applications to group their
4725 own child processes (worker processes) in a powerful way
4726 which then maybe used to organize them, or kill them
4727 together, or apply resource limits on them.
4729 Slice units may be used to partition system resources in an
4730 hierarchical fashion and then assign other units to them. By
4731 default there are now three slices: system.slice (for all
4732 system services), user.slice (for all user sessions),
4733 machine.slice (for VMs and containers).
4735 Slices and scopes have been introduced primarily in
4736 context of the work to move cgroup handling to a
4737 single-writer scheme, where only PID 1
4738 creates/removes/manages cgroups.
4740 * There's a new concept of "transient" units. In contrast to
4741 normal units these units are created via an API at runtime,
4742 not from configuration from disk. More specifically this
4743 means it is now possible to run arbitrary programs as
4744 independent services, with all execution parameters passed
4745 in via bus APIs rather than read from disk. Transient units
4746 make systemd substantially more dynamic then it ever was,
4747 and useful as a general batch manager.
4749 * logind has been updated to make use of scope and slice units
4750 for managing user sessions. As a user logs in he will get
4751 his own private slice unit, to which all sessions are added
4752 as scope units. We also added support for automatically
4753 adding an instance of user@.service for the user into the
4754 slice. Effectively logind will no longer create cgroup
4755 hierarchies on its own now, it will defer entirely to PID 1
4756 for this by means of scope, service and slice units. Since
4757 user sessions this way become entities managed by PID 1
4758 the output of "systemctl" is now a lot more comprehensive.
4760 * A new mini-daemon "systemd-machined" has been added which
4761 may be used by virtualization managers to register local
4762 VMs/containers. nspawn has been updated accordingly, and
4763 libvirt will be updated shortly. machined will collect a bit
4764 of meta information about the VMs/containers, and assign
4765 them their own scope unit (see above). The collected
4766 meta-data is then made available via the "machinectl" tool,
4767 and exposed in "ps" and similar tools. machined/machinectl
4768 is compile-time optional.
4770 * As discussed earlier, the low-level cgroup configuration
4771 options ControlGroup=, ControlGroupModify=,
4772 ControlGroupPersistent=, ControlGroupAttribute= have been
4773 removed. Please use high-level attribute settings instead as
4774 well as slice units.
4776 * A new bus call SetUnitProperties() has been added to alter
4777 various runtime parameters of a unit. This is primarily
4778 useful to alter cgroup parameters dynamically in a nice way,
4779 but will be extended later on to make more properties
4780 modifiable at runtime. systemctl gained a new set-properties
4781 command that wraps this call.
4783 * A new tool "systemd-run" has been added which can be used to
4784 run arbitrary command lines as transient services or scopes,
4785 while configuring a number of settings via the command
4786 line. This tool is currently very basic, however already
4787 very useful. We plan to extend this tool to even allow
4788 queuing of execution jobs with time triggers from the
4789 command line, similar in fashion to "at".
4791 * nspawn will now inform the user explicitly that kernels with
4792 audit enabled break containers, and suggest the user to turn
4795 * Support for detecting the IMA and AppArmor security
4796 frameworks with ConditionSecurity= has been added.
4798 * journalctl gained a new "-k" switch for showing only kernel
4799 messages, mimicking dmesg output; in addition to "--user"
4800 and "--system" switches for showing only user's own logs
4803 * systemd-delta can now show information about drop-in
4804 snippets extending unit files.
4806 * libsystemd-bus has been substantially updated but is still
4807 not available as public API.
4809 * systemd will now look for the "debug" argument on the kernel
4810 command line and enable debug logging, similar to what
4811 "systemd.log_level=debug" already did before.
4813 * "systemctl set-default", "systemctl get-default" has been
4814 added to configure the default.target symlink, which
4815 controls what to boot into by default.
4817 * "systemctl set-log-level" has been added as a convenient
4818 way to raise and lower systemd logging threshold.
4820 * "systemd-analyze plot" will now show the time the various
4821 generators needed for execution, as well as information
4822 about the unit file loading.
4824 * libsystemd-journal gained a new sd_journal_open_files() call
4825 for opening specific journal files. journactl also gained a
4826 new switch to expose this new functionality. Previously we
4827 only supported opening all files from a directory, or all
4828 files from the system, as opening individual files only is
4829 racy due to journal file rotation.
4831 * systemd gained the new DefaultEnvironment= setting in
4832 /etc/systemd/system.conf to set environment variables for
4835 * If a privileged process logs a journal message with the
4836 OBJECT_PID= field set, then journald will automatically
4837 augment this with additional OBJECT_UID=, OBJECT_GID=,
4838 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
4839 system services want to log events about specific client
4840 processes. journactl/systemctl has been updated to make use
4841 of this information if all log messages regarding a specific
4844 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
4845 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
4846 Reisner, David Coppa, David King, David Strauss, Eelco
4847 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
4848 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
4849 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
4850 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
4851 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
4852 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
4853 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
4854 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
4855 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
4856 Łukasz Stelmach, 장동준
4860 * The Python bindings gained some minimal support for the APIs
4861 exposed by libsystemd-logind.
4863 * ConditionSecurity= gained support for detecting SMACK. Since
4864 this condition already supports SELinux and AppArmor we only
4865 miss IMA for this. Patches welcome!
4867 Contributions from: Karol Lewandowski, Lennart Poettering,
4868 Zbigniew Jędrzejewski-Szmek
4872 * systemd-nspawn will now create /etc/resolv.conf if
4873 necessary, before bind-mounting the host's file onto it.
4875 * systemd-nspawn will now store meta information about a
4876 container on the container's cgroup as extended attribute
4877 fields, including the root directory.
4879 * The cgroup hierarchy has been reworked in many ways. All
4880 objects any of the components systemd creates in the cgroup
4881 tree are now suffixed. More specifically, user sessions are
4882 now placed in cgroups suffixed with ".session", users in
4883 cgroups suffixed with ".user", and nspawn containers in
4884 cgroups suffixed with ".nspawn". Furthermore, all cgroup
4885 names are now escaped in a simple scheme to avoid collision
4886 of userspace object names with kernel filenames. This work
4887 is preparation for making these objects relocatable in the
4888 cgroup tree, in order to allow easy resource partitioning of
4889 these objects without causing naming conflicts.
4891 * systemctl list-dependencies gained the new switches
4892 --plain, --reverse, --after and --before.
4894 * systemd-inhibit now shows the process name of processes that
4895 have taken an inhibitor lock.
4897 * nss-myhostname will now also resolve "localhost"
4898 implicitly. This makes /etc/hosts an optional file and
4899 nicely handles that on IPv6 ::1 maps to both "localhost" and
4902 * libsystemd-logind.so gained a new call
4903 sd_get_machine_names() to enumerate running containers and
4904 VMs (currently only supported by very new libvirt and
4905 nspawn). sd_login_monitor can now be used to watch
4906 VMs/containers coming and going.
4908 * .include is not allowed recursively anymore, and only in
4909 unit files. Usually it is better to use drop-in snippets in
4910 .d/*.conf anyway, as introduced with systemd 198.
4912 * systemd-analyze gained a new "critical-chain" command that
4913 determines the slowest chain of units run during system
4914 boot-up. It is very useful for tracking down where
4915 optimizing boot time is the most beneficial.
4917 * systemd will no longer allow manipulating service paths in
4918 the name=systemd:/system cgroup tree using ControlGroup= in
4919 units. (But is still fine with it in all other dirs.)
4921 * There's a new systemd-nspawn@.service service file that may
4922 be used to easily run nspawn containers as system
4923 services. With the container's root directory in
4924 /var/lib/container/foobar it is now sufficient to run
4925 "systemctl start systemd-nspawn@foobar.service" to boot it.
4927 * systemd-cgls gained a new parameter "--machine" to list only
4928 the processes within a certain container.
4930 * ConditionSecurity= now can check for "apparmor". We still
4931 are lacking checks for SMACK and IMA for this condition
4932 check though. Patches welcome!
4934 * A new configuration file /etc/systemd/sleep.conf has been
4935 added that may be used to configure which kernel operation
4936 systemd is supposed to execute when "suspend", "hibernate"
4937 or "hybrid-sleep" is requested. This makes the new kernel
4938 "freeze" state accessible to the user.
4940 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
4941 the passed argument if applicable.
4943 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
4944 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
4945 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
4946 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
4947 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
4948 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
4949 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
4954 * The output of 'systemctl list-jobs' got some polishing. The
4955 '--type=' argument may now be passed more than once. A new
4956 command 'systemctl list-sockets' has been added which shows
4957 a list of kernel sockets systemd is listening on with the
4958 socket units they belong to, plus the units these socket
4961 * The experimental libsystemd-bus library got substantial
4962 updates to work in conjunction with the (also experimental)
4963 kdbus kernel project. It works well enough to exchange
4964 messages with some sophistication. Note that kdbus is not
4965 ready yet, and the library is mostly an elaborate test case
4966 for now, and not installable.
4968 * systemd gained a new unit 'systemd-static-nodes.service'
4969 that generates static device nodes earlier during boot, and
4970 can run in conjunction with udev.
4972 * libsystemd-login gained a new call sd_pid_get_user_unit()
4973 to retrieve the user systemd unit a process is running
4974 in. This is useful for systems where systemd is used as
4977 * systemd-nspawn now places all containers in the new /machine
4978 top-level cgroup directory in the name=systemd
4979 hierarchy. libvirt will soon do the same, so that we get a
4980 uniform separation of /system, /user and /machine for system
4981 services, user processes and containers/virtual
4982 machines. This new cgroup hierarchy is also useful to stick
4983 stable names to specific container instances, which can be
4984 recognized later this way (this name may be controlled
4985 via systemd-nspawn's new -M switch). libsystemd-login also
4986 gained a new call sd_pid_get_machine_name() to retrieve the
4987 name of the container/VM a specific process belongs to.
4989 * bootchart can now store its data in the journal.
4991 * libsystemd-journal gained a new call
4992 sd_journal_add_conjunction() for AND expressions to the
4993 matching logic. This can be used to express more complex
4994 logical expressions.
4996 * journactl can now take multiple --unit= and --user-unit=
4999 * The cryptsetup logic now understands the "luks.key=" kernel
5000 command line switch for specifying a file to read the
5001 decryption key from. Also, if a configured key file is not
5002 found the tool will now automatically fall back to prompting
5005 * Python systemd.journal module was updated to wrap recently
5006 added functions from libsystemd-journal. The interface was
5007 changed to bring the low level interface in s.j._Reader
5008 closer to the C API, and the high level interface in
5009 s.j.Reader was updated to wrap and convert all data about
5012 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
5013 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
5014 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
5015 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
5016 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
5017 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5021 * journalctl --update-catalog now understands a new --root=
5022 option to operate on catalogs found in a different root
5025 * During shutdown after systemd has terminated all running
5026 services a final killing loop kills all remaining left-over
5027 processes. We will now print the name of these processes
5028 when we send SIGKILL to them, since this usually indicates a
5031 * If /etc/crypttab refers to password files stored on
5032 configured mount points automatic dependencies will now be
5033 generated to ensure the specific mount is established first
5034 before the key file is attempted to be read.
5036 * 'systemctl status' will now show information about the
5037 network sockets a socket unit is listening on.
5039 * 'systemctl status' will also shown information about any
5040 drop-in configuration file for units. (Drop-In configuration
5041 files in this context are files such as
5042 /etc/systemd/systemd/foobar.service.d/*.conf)
5044 * systemd-cgtop now optionally shows summed up CPU times of
5045 cgroups. Press '%' while running cgtop to switch between
5046 percentage and absolute mode. This is useful to determine
5047 which cgroups use up the most CPU time over the entire
5048 runtime of the system. systemd-cgtop has also been updated
5049 to be 'pipeable' for processing with further shell tools.
5051 * 'hostnamectl set-hostname' will now allow setting of FQDN
5054 * The formatting and parsing of time span values has been
5055 changed. The parser now understands fractional expressions
5056 such as "5.5h". The formatter will now output fractional
5057 expressions for all time spans under 1min, i.e. "5.123456s"
5058 rather than "5s 123ms 456us". For time spans under 1s
5059 millisecond values are shown, for those under 1ms
5060 microsecond values are shown. This should greatly improve
5061 all time-related output of systemd.
5063 * libsystemd-login and libsystemd-journal gained new
5064 functions for querying the poll() events mask and poll()
5065 timeout value for integration into arbitrary event
5068 * localectl gained the ability to list available X11 keymaps
5069 (models, layouts, variants, options).
5071 * 'systemd-analyze dot' gained the ability to filter for
5072 specific units via shell-style globs, to create smaller,
5073 more useful graphs. I.e. it is now possible to create simple
5074 graphs of all the dependencies between only target units, or
5075 of all units that Avahi has dependencies with.
5077 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
5078 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
5079 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
5080 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
5081 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
5082 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
5083 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
5087 * The boot-time readahead implementation for rotating media
5088 will now read the read-ahead data in multiple passes which
5089 consist of all read requests made in equidistant time
5090 intervals. This means instead of strictly reading read-ahead
5091 data in its physical order on disk we now try to find a
5092 middle ground between physical and access time order.
5094 * /etc/os-release files gained a new BUILD_ID= field for usage
5095 on operating systems that provide continuous builds of OS
5098 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
5099 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
5100 William Douglas, Zbigniew Jędrzejewski-Szmek
5104 * systemd-python gained an API exposing libsystemd-daemon.
5106 * The SMACK setup logic gained support for uploading CIPSO
5109 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
5110 ReadOnlyDirectories= and InaccessibleDirectories= has
5111 changed. The private /tmp and /var/tmp directories are now
5112 shared by all processes of a service (which means
5113 ExecStartPre= may now leave data in /tmp that ExecStart= of
5114 the same service can still access). When a service is
5115 stopped its temporary directories are immediately deleted
5116 (normal clean-up with tmpfiles is still done in addition to
5119 * By default, systemd will now set a couple of sysctl
5120 variables in the kernel: the safe sysrq options are turned
5121 on, IP route verification is turned on, and source routing
5122 disabled. The recently added hardlink and softlink
5123 protection of the kernel is turned on. These settings should
5124 be reasonably safe, and good defaults for all new systems.
5126 * The predictable network naming logic may now be turned off
5127 with a new kernel command line switch: net.ifnames=0.
5129 * A new libsystemd-bus module has been added that implements a
5130 pretty complete D-Bus client library. For details see:
5132 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
5134 * journald will now explicitly flush the journal files to disk
5135 at the latest 5min after each write. The file will then also
5136 be marked offline until the next write. This should increase
5137 reliability in case of a crash. The synchronization delay
5138 can be configured via SyncIntervalSec= in journald.conf.
5140 * There's a new remote-fs-setup.target unit that can be used
5141 to pull in specific services when at least one remote file
5142 system is to be mounted.
5144 * There are new targets timers.target and paths.target as
5145 canonical targets to pull user timer and path units in
5146 from. This complements sockets.target with a similar
5147 purpose for socket units.
5149 * libudev gained a new call udev_device_set_attribute_value()
5150 to set sysfs attributes of a device.
5152 * The udev daemon now sets the default number of worker
5153 processes executed in parallel based on the number of available
5154 CPUs instead of the amount of available RAM. This is supposed
5155 to provide a more reliable default and limit a too aggressive
5156 parallelism for setups with 1000s of devices connected.
5158 Contributions from: Auke Kok, Colin Walters, Cristian
5159 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
5160 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
5161 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
5162 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
5163 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
5164 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
5165 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
5166 Zbigniew Jędrzejewski-Szmek
5170 * Configuration of unit files may now be extended via drop-in
5171 files without having to edit/override the unit files
5172 themselves. More specifically, if the administrator wants to
5173 change one value for a service file foobar.service he can
5174 now do so by dropping in a configuration snippet into
5175 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
5176 will load all these snippets and apply them on top of the
5177 main unit configuration file, possibly extending or
5178 overriding its settings. Using these drop-in snippets is
5179 generally nicer than the two earlier options for changing
5180 unit files locally: copying the files from
5181 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
5182 them there; or creating a new file in /etc/systemd/system/
5183 that incorporates the original one via ".include". Drop-in
5184 snippets into these .d/ directories can be placed in any
5185 directory systemd looks for units in, and the usual
5186 overriding semantics between /usr/lib, /etc and /run apply
5189 * Most unit file settings which take lists of items can now be
5190 reset by assigning the empty string to them. For example,
5191 normally, settings such as Environment=FOO=BAR append a new
5192 environment variable assignment to the environment block,
5193 each time they are used. By assigning Environment= the empty
5194 string the environment block can be reset to empty. This is
5195 particularly useful with the .d/*.conf drop-in snippets
5196 mentioned above, since this adds the ability to reset list
5197 settings from vendor unit files via these drop-ins.
5199 * systemctl gained a new "list-dependencies" command for
5200 listing the dependencies of a unit recursively.
5202 * Inhibitors are now honored and listed by "systemctl
5203 suspend", "systemctl poweroff" (and similar) too, not only
5204 GNOME. These commands will also list active sessions by
5207 * Resource limits (as exposed by the various control group
5208 controllers) can now be controlled dynamically at runtime
5209 for all units. More specifically, you can now use a command
5210 like "systemctl set-cgroup-attr foobar.service cpu.shares
5211 2000" to alter the CPU shares a specific service gets. These
5212 settings are stored persistently on disk, and thus allow the
5213 administrator to easily adjust the resource usage of
5214 services with a few simple commands. This dynamic resource
5215 management logic is also available to other programs via the
5216 bus. Almost any kernel cgroup attribute and controller is
5219 * systemd-vconsole-setup will now copy all font settings to
5220 all allocated VTs, where it previously applied them only to
5223 * libsystemd-login gained the new sd_session_get_tty() API
5226 * This release drops support for a few legacy or
5227 distribution-specific LSB facility names when parsing init
5228 scripts: $x-display-manager, $mail-transfer-agent,
5229 $mail-transport-agent, $mail-transfer-agent, $smtp,
5230 $null. Also, the mail-transfer-agent.target unit backing
5231 this has been removed. Distributions which want to retain
5232 compatibility with this should carry the burden for
5233 supporting this themselves and patch support for these back
5234 in, if they really need to. Also, the facilities $syslog and
5235 $local_fs are now ignored, since systemd does not support
5236 early-boot LSB init scripts anymore, and these facilities
5237 are implied anyway for normal services. syslog.target has
5240 * There are new bus calls on PID1's Manager object for
5241 cancelling jobs, and removing snapshot units. Previously,
5242 both calls were only available on the Job and Snapshot
5245 * systemd-journal-gatewayd gained SSL support.
5247 * The various "environment" files, such as /etc/locale.conf
5248 now support continuation lines with a backslash ("\") as
5249 last character in the line, similarly in style (but different)
5250 to how this is supported in shells.
5252 * For normal user processes the _SYSTEMD_USER_UNIT= field is
5253 now implicitly appended to every log entry logged. systemctl
5254 has been updated to filter by this field when operating on a
5255 user systemd instance.
5257 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
5258 CAP_AUDIT_CONTROL capabilities to the capabilities set for
5259 the container. This makes it easier to boot unmodified
5260 Fedora systems in a container, which however still requires
5261 audit=0 to be passed on the kernel command line. Auditing in
5262 kernel and userspace is unfortunately still too broken in
5263 context of containers, hence we recommend compiling it out
5264 of the kernel or using audit=0. Hopefully this will be fixed
5265 one day for good in the kernel.
5267 * nspawn gained the new --bind= and --bind-ro= parameters to
5268 bind mount specific directories from the host into the
5271 * nspawn will now mount its own devpts file system instance
5272 into the container, in order not to leak pty devices from
5273 the host into the container.
5275 * systemd will now read the firmware boot time performance
5276 information from the EFI variables, if the used boot loader
5277 supports this, and takes it into account for boot performance
5278 analysis via "systemd-analyze". This is currently supported
5279 only in conjunction with Gummiboot, but could be supported
5280 by other boot loaders too. For details see:
5282 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
5284 * A new generator has been added that automatically mounts the
5285 EFI System Partition (ESP) to /boot, if that directory
5286 exists, is empty, and no other file system has been
5287 configured to be mounted there.
5289 * logind will now send out PrepareForSleep(false) out
5290 unconditionally, after coming back from suspend. This may be
5291 used by applications as asynchronous notification for
5292 system resume events.
5294 * "systemctl unlock-sessions" has been added, that allows
5295 unlocking the screens of all user sessions at once, similar
5296 to how "systemctl lock-sessions" already locked all users
5297 sessions. This is backed by a new D-Bus call UnlockSessions().
5299 * "loginctl seat-status" will now show the master device of a
5300 seat. (i.e. the device of a seat that needs to be around for
5301 the seat to be considered available, usually the graphics
5304 * tmpfiles gained a new "X" line type, that allows
5305 configuration of files and directories (with wildcards) that
5306 shall be excluded from automatic cleanup ("aging").
5308 * udev default rules set the device node permissions now only
5309 at "add" events, and do not change them any longer with a
5310 later "change" event.
5312 * The log messages for lid events and power/sleep keypresses
5313 now carry a message ID.
5315 * We now have a substantially larger unit test suite, but this
5316 continues to be work in progress.
5318 * udevadm hwdb gained a new --root= parameter to change the
5319 root directory to operate relative to.
5321 * logind will now issue a background sync() request to the kernel
5322 early at shutdown, so that dirty buffers are flushed to disk early
5323 instead of at the last moment, in order to optimize shutdown
5326 * A new bootctl tool has been added that is an interface for
5327 certain boot loader operations. This is currently a preview
5328 and is likely to be extended into a small mechanism daemon
5329 like timedated, localed, hostnamed, and can be used by
5330 graphical UIs to enumerate available boot options, and
5331 request boot into firmware operations.
5333 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
5334 the rest of the package. It also has been updated to work
5335 correctly in initrds.
5337 * Policykit previously has been runtime optional, and is now
5338 also compile time optional via a configure switch.
5340 * systemd-analyze has been reimplemented in C. Also "systemctl
5341 dot" has moved into systemd-analyze.
5343 * "systemctl status" with no further parameters will now print
5344 the status of all active or failed units.
5346 * Operations such as "systemctl start" can now be executed
5347 with a new mode "--irreversible" which may be used to queue
5348 operations that cannot accidentally be reversed by a later
5349 job queuing. This is by default used to make shutdown
5350 requests more robust.
5352 * The Python API of systemd now gained a new module for
5353 reading journal files.
5355 * A new tool kernel-install has been added that can install
5356 kernel images according to the Boot Loader Specification:
5358 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
5360 * Boot time console output has been improved to provide
5361 animated boot time output for hanging jobs.
5363 * A new tool systemd-activate has been added which can be used
5364 to test socket activation with, directly from the command
5365 line. This should make it much easier to test and debug
5366 socket activation in daemons.
5368 * journalctl gained a new "--reverse" (or -r) option to show
5369 journal output in reverse order (i.e. newest line first).
5371 * journalctl gained a new "--pager-end" (or -e) option to jump
5372 to immediately jump to the end of the journal in the
5373 pager. This is only supported in conjunction with "less".
5375 * journalctl gained a new "--user-unit=" option, that works
5376 similarly to "--unit=" but filters for user units rather than
5379 * A number of unit files to ease adoption of systemd in
5380 initrds has been added. This moves some minimal logic from
5381 the various initrd implementations into systemd proper.
5383 * The journal files are now owned by a new group
5384 "systemd-journal", which exists specifically to allow access
5385 to the journal, and nothing else. Previously, we used the
5386 "adm" group for that, which however possibly covers more
5387 than just journal/log file access. This new group is now
5388 already used by systemd-journal-gatewayd to ensure this
5389 daemon gets access to the journal files and as little else
5390 as possible. Note that "make install" will also set FS ACLs
5391 up for /var/log/journal to give "adm" and "wheel" read
5392 access to it, in addition to "systemd-journal" which owns
5393 the journal files. We recommend that packaging scripts also
5394 add read access to "adm" + "wheel" to /var/log/journal, and
5395 all existing/future journal files. To normal users and
5396 administrators little changes, however packagers need to
5397 ensure to create the "systemd-journal" system group at
5398 package installation time.
5400 * The systemd-journal-gatewayd now runs as unprivileged user
5401 systemd-journal-gateway:systemd-journal-gateway. Packaging
5402 scripts need to create these system user/group at
5405 * timedated now exposes a new boolean property CanNTP that
5406 indicates whether a local NTP service is available or not.
5408 * systemd-detect-virt will now also detect xen PVs
5410 * The pstore file system is now mounted by default, if it is
5413 * In addition to the SELinux and IMA policies we will now also
5414 load SMACK policies at early boot.
5416 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
5417 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
5418 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
5419 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
5420 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
5421 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
5422 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
5423 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
5424 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
5425 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
5426 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
5427 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
5428 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
5429 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
5433 * Timer units now support calendar time events in addition to
5434 monotonic time events. That means you can now trigger a unit
5435 based on a calendar time specification such as "Thu,Fri
5436 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
5437 or fifth day of any month of the year 2013, given that it is
5438 a thursday or friday. This brings timer event support
5439 considerably closer to cron's capabilities. For details on
5440 the supported calendar time specification language see
5443 * udev now supports a number of different naming policies for
5444 network interfaces for predictable names, and a combination
5445 of these policies is now the default. Please see this wiki
5446 document for details:
5448 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
5450 * Auke Kok's bootchart implementation has been added to the
5451 systemd tree. It is an optional component that can graph the
5452 boot in quite some detail. It is one of the best bootchart
5453 implementations around and minimal in its code and
5456 * nss-myhostname has been integrated into the systemd source
5457 tree. nss-myhostname guarantees that the local hostname
5458 always stays resolvable via NSS. It has been a weak
5459 requirement of systemd-hostnamed since a long time, and
5460 since its code is actually trivial we decided to just
5461 include it in systemd's source tree. It can be turned off
5462 with a configure switch.
5464 * The read-ahead logic is now capable of properly detecting
5465 whether a btrfs file system is on SSD or rotating media, in
5466 order to optimize the read-ahead scheme. Previously, it was
5467 only capable of detecting this on traditional file systems
5470 * In udev, additional device properties are now read from the
5471 IAB in addition to the OUI database. Also, Bluetooth company
5472 identities are attached to the devices as well.
5474 * In service files %U may be used as specifier that is
5475 replaced by the configured user name of the service.
5477 * nspawn may now be invoked without a controlling TTY. This
5478 makes it suitable for invocation as its own service. This
5479 may be used to set up a simple containerized server system
5480 using only core OS tools.
5482 * systemd and nspawn can now accept socket file descriptors
5483 when they are started for socket activation. This enables
5484 implementation of socket activated nspawn
5485 containers. i.e. think about autospawning an entire OS image
5486 when the first SSH or HTTP connection is received. We expect
5487 that similar functionality will also be added to libvirt-lxc
5490 * journalctl will now suppress ANSI color codes when
5491 presenting log data.
5493 * systemctl will no longer show control group information for
5494 a unit if the control group is empty anyway.
5496 * logind can now automatically suspend/hibernate/shutdown the
5499 * /etc/machine-info and hostnamed now also expose the chassis
5500 type of the system. This can be used to determine whether
5501 the local system is a laptop, desktop, handset or
5502 tablet. This information may either be configured by the
5503 user/vendor or is automatically determined from ACPI and DMI
5504 information if possible.
5506 * A number of PolicyKit actions are now bound together with
5507 "imply" rules. This should simplify creating UIs because
5508 many actions will now authenticate similar ones as well.
5510 * Unit files learnt a new condition ConditionACPower= which
5511 may be used to conditionalize a unit depending on whether an
5512 AC power source is connected or not, of whether the system
5513 is running on battery power.
5515 * systemctl gained a new "is-failed" verb that may be used in
5516 shell scripts and suchlike to check whether a specific unit
5517 is in the "failed" state.
5519 * The EnvironmentFile= setting in unit files now supports file
5520 globbing, and can hence be used to easily read a number of
5521 environment files at once.
5523 * systemd will no longer detect and recognize specific
5524 distributions. All distribution-specific #ifdeffery has been
5525 removed, systemd is now fully generic and
5526 distribution-agnostic. Effectively, not too much is lost as
5527 a lot of the code is still accessible via explicit configure
5528 switches. However, support for some distribution specific
5529 legacy configuration file formats has been dropped. We
5530 recommend distributions to simply adopt the configuration
5531 files everybody else uses now and convert the old
5532 configuration from packaging scripts. Most distributions
5533 already did that. If that's not possible or desirable,
5534 distributions are welcome to forward port the specific
5535 pieces of code locally from the git history.
5537 * When logging a message about a unit systemd will now always
5538 log the unit name in the message meta data.
5540 * localectl will now also discover system locale data that is
5541 not stored in locale archives, but directly unpacked.
5543 * logind will no longer unconditionally use framebuffer
5544 devices as seat masters, i.e. as devices that are required
5545 to be existing before a seat is considered preset. Instead,
5546 it will now look for all devices that are tagged as
5547 "seat-master" in udev. By default, framebuffer devices will
5548 be marked as such, but depending on local systems, other
5549 devices might be marked as well. This may be used to
5550 integrate graphics cards using closed source drivers (such
5551 as NVidia ones) more nicely into logind. Note however, that
5552 we recommend using the open source NVidia drivers instead,
5553 and no udev rules for the closed-source drivers will be
5554 shipped from us upstream.
5556 Contributions from: Adam Williamson, Alessandro Crismani, Auke
5557 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
5558 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
5559 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
5560 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
5561 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
5562 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
5563 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
5564 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
5565 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
5566 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
5567 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
5572 * udev gained support for loading additional device properties
5573 from an indexed database that is keyed by vendor/product IDs
5574 and similar device identifiers. For the beginning this
5575 "hwdb" is populated with data from the well-known PCI and
5576 USB database, but also includes PNP, ACPI and OID data. In
5577 the longer run this indexed database shall grow into
5578 becoming the one central database for non-essential
5579 userspace device metadata. Previously, data from the PCI/USB
5580 database was only attached to select devices, since the
5581 lookup was a relatively expensive operation due to O(n) time
5582 complexity (with n being the number of entries in the
5583 database). Since this is now O(1), we decided to add in this
5584 data for all devices where this is available, by
5585 default. Note that the indexed database needs to be rebuilt
5586 when new data files are installed. To achieve this you need
5587 to update your packaging scripts to invoke "udevadm hwdb
5588 --update" after installation of hwdb data files. For
5589 RPM-based distributions we introduced the new
5590 %udev_hwdb_update macro for this purpose.
5592 * The Journal gained support for the "Message Catalog", an
5593 indexed database to link up additional information with
5594 journal entries. For further details please check:
5596 http://www.freedesktop.org/wiki/Software/systemd/catalog
5598 The indexed message catalog database also needs to be
5599 rebuilt after installation of message catalog files. Use
5600 "journalctl --update-catalog" for this. For RPM-based
5601 distributions we introduced the %journal_catalog_update
5602 macro for this purpose.
5604 * The Python Journal bindings gained support for the standard
5605 Python logging framework.
5607 * The Journal API gained new functions for checking whether
5608 the underlying file system of a journal file is capable of
5609 properly reporting file change notifications, or whether
5610 applications that want to reflect journal changes "live"
5611 need to recheck journal files continuously in appropriate
5614 * It is now possible to set the "age" field for tmpfiles
5615 entries to 0, indicating that files matching this entry
5616 shall always be removed when the directories are cleaned up.
5618 * coredumpctl gained a new "gdb" verb which invokes gdb
5619 right-away on the selected coredump.
5621 * There's now support for "hybrid sleep" on kernels that
5622 support this, in addition to "suspend" and "hibernate". Use
5623 "systemctl hybrid-sleep" to make use of this.
5625 * logind's HandleSuspendKey= setting (and related settings)
5626 now gained support for a new "lock" setting to simply
5627 request the screen lock on all local sessions, instead of
5628 actually executing a suspend or hibernation.
5630 * systemd will now mount the EFI variables file system by
5633 * Socket units now gained support for configuration of the
5634 SMACK security label.
5636 * timedatectl will now output the time of the last and next
5637 daylight saving change.
5639 * We dropped support for various legacy and distro-specific
5640 concepts, such as insserv, early-boot SysV services
5641 (i.e. those for non-standard runlevels such as 'b' or 'S')
5642 or ArchLinux /etc/rc.conf support. We recommend the
5643 distributions who still need support this to either continue
5644 to maintain the necessary patches downstream, or find a
5645 different solution. (Talk to us if you have questions!)
5647 * Various systemd components will now bypass PolicyKit checks
5648 for root and otherwise handle properly if PolicyKit is not
5649 found to be around. This should fix most issues for
5650 PolicyKit-less systems. Quite frankly this should have been
5651 this way since day one. It is absolutely our intention to
5652 make systemd work fine on PolicyKit-less systems, and we
5653 consider it a bug if something does not work as it should if
5654 PolicyKit is not around.
5656 * For embedded systems it is now possible to build udev and
5657 systemd without blkid and/or kmod support.
5659 * "systemctl switch-root" is now capable of switching root
5660 more than once. I.e. in addition to transitions from the
5661 initrd to the host OS it is now possible to transition to
5662 further OS images from the host. This is useful to implement
5663 offline updating tools.
5665 * Various other additions have been made to the RPM macros
5666 shipped with systemd. Use %udev_rules_update() after
5667 installing new udev rules files. %_udevhwdbdir,
5668 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
5669 %_sysctldir are now available which resolve to the right
5670 directories for packages to place various data files in.
5672 * journalctl gained the new --full switch (in addition to
5673 --all, to disable ellipsation for long messages.
5675 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
5676 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
5677 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
5678 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
5679 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
5680 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
5681 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
5682 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
5683 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
5687 * journalctl gained new --since= and --until= switches to
5688 filter by time. It also now supports nice filtering for
5689 units via --unit=/-u.
5691 * Type=oneshot services may use ExecReload= and do the
5694 * The journal daemon now supports time-based rotation and
5695 vacuuming, in addition to the usual disk-space based
5698 * The journal will now index the available field values for
5699 each field name. This enables clients to show pretty drop
5700 downs of available match values when filtering. The bash
5701 completion of journalctl has been updated
5702 accordingly. journalctl gained a new switch -F to list all
5703 values a certain field takes in the journal database.
5705 * More service events are now written as structured messages
5706 to the journal, and made recognizable via message IDs.
5708 * The timedated, localed and hostnamed mini-services which
5709 previously only provided support for changing time, locale
5710 and hostname settings from graphical DEs such as GNOME now
5711 also have a minimal (but very useful) text-based client
5712 utility each. This is probably the nicest way to changing
5713 these settings from the command line now, especially since
5714 it lists available options and is fully integrated with bash
5717 * There's now a new tool "systemd-coredumpctl" to list and
5718 extract coredumps from the journal.
5720 * We now install a README each in /var/log/ and
5721 /etc/rc.d/init.d explaining where the system logs and init
5722 scripts went. This hopefully should help folks who go to
5723 that dirs and look into the otherwise now empty void and
5724 scratch their heads.
5726 * When user-services are invoked (by systemd --user) the
5727 $MANAGERPID env var is set to the PID of systemd.
5729 * SIGRTMIN+24 when sent to a --user instance will now result
5730 in immediate termination of systemd.
5732 * gatewayd received numerous feature additions such as a
5733 "follow" mode, for live syncing and filtering.
5735 * browse.html now allows filtering and showing detailed
5736 information on specific entries. Keyboard navigation and
5737 mouse screen support has been added.
5739 * gatewayd/journalctl now supports HTML5/JSON
5740 Server-Sent-Events as output.
5742 * The SysV init script compatibility logic will now
5743 heuristically determine whether a script supports the
5744 "reload" verb, and only then make this available as
5747 * "systemctl status --follow" has been removed, use "journalctl
5750 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
5751 have been removed since they are hardly useful to be
5754 * And I'd like to take the opportunity to specifically mention
5755 Zbigniew for his great contributions. Zbigniew, you rock!
5757 Contributions from: Andrew Eikum, Christian Hesse, Colin
5758 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
5759 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
5760 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
5761 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
5762 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
5763 Jędrzejewski-Szmek, Сковорода Никита Андреевич
5767 * If /etc/vconsole.conf is non-existent or empty we will no
5768 longer load any console font or key map at boot by
5769 default. Instead the kernel defaults will be left
5770 intact. This is definitely the right thing to do, as no
5771 configuration should mean no configuration, and hard-coding
5772 font names that are different on all archs is probably a bad
5773 idea. Also, the kernel default key map and font should be
5774 good enough for most cases anyway, and mostly identical to
5775 the userspace fonts/key maps we previously overloaded them
5776 with. If distributions want to continue to default to a
5777 non-kernel font or key map they should ship a default
5778 /etc/vconsole.conf with the appropriate contents.
5780 Contributions from: Colin Walters, Daniel J Walsh, Dave
5781 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
5782 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
5786 * journalctl gained a new --cursor= switch to show entries
5787 starting from the specified location in the journal.
5789 * We now enforce a size limit on journal entry fields exported
5790 with "-o json" in journalctl. Fields larger than 4K will be
5791 assigned null. This can be turned off with --all.
5793 * An (optional) journal gateway daemon is now available as
5794 "systemd-journal-gatewayd.service". This service provides
5795 access to the journal via HTTP and JSON. This functionality
5796 will be used to implement live log synchronization in both
5797 pull and push modes, but has various other users too, such
5798 as easy log access for debugging of embedded devices. Right
5799 now it is already useful to retrieve the journal via HTTP:
5801 # systemctl start systemd-journal-gatewayd.service
5802 # wget http://localhost:19531/entries
5804 This will download the journal contents in a
5805 /var/log/messages compatible format. The same as JSON:
5807 # curl -H"Accept: application/json" http://localhost:19531/entries
5809 This service is also accessible via a web browser where a
5810 single static HTML5 app is served that uses the JSON logic
5811 to enable the user to do some basic browsing of the
5812 journal. This will be extended later on. Here's an example
5813 screenshot of this app in its current state:
5815 http://0pointer.de/public/journal-gatewayd
5817 Contributions from: Kay Sievers, Lennart Poettering, Robert
5818 Milasan, Tom Gundersen
5822 * The bash completion logic is now available for journalctl
5825 * We do not mount the "cpuset" controller anymore together with
5826 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
5827 started if no parameters are assigned to it. "cpuset" hence
5828 broke code that assumed it could create "cpu" groups and
5831 * journalctl -f will now subscribe to terminal size changes,
5832 and line break accordingly.
5834 Contributions from: Dave Reisner, Kay Sievers, Lennart
5835 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
5839 * nspawn will now create a symlink /etc/localtime in the
5840 container environment, copying the host's timezone
5841 setting. Previously this has been done via a bind mount, but
5842 since symlinks cannot be bind mounted this has now been
5843 changed to create/update the appropriate symlink.
5845 * journalctl -n's line number argument is now optional, and
5846 will default to 10 if omitted.
5848 * journald will now log the maximum size the journal files may
5849 take up on disk. This is particularly useful if the default
5850 built-in logic of determining this parameter from the file
5851 system size is used. Use "systemctl status
5852 systemd-journald.service" to see this information.
5854 * The multi-seat X wrapper tool has been stripped down. As X
5855 is now capable of enumerating graphics devices via udev in a
5856 seat-aware way the wrapper is not strictly necessary
5857 anymore. A stripped down temporary stop-gap is still shipped
5858 until the upstream display managers have been updated to
5859 fully support the new X logic. Expect this wrapper to be
5860 removed entirely in one of the next releases.
5862 * HandleSleepKey= in logind.conf has been split up into
5863 HandleSuspendKey= and HandleHibernateKey=. The old setting
5864 is not available anymore. X11 and the kernel are
5865 distinguishing between these keys and we should too. This
5866 also means the inhibition lock for these keys has been split
5869 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
5870 Poettering, Lukas Nykryn, Václav Pavlín
5874 * Whenever a unit changes state we will now log this to the
5875 journal and show along the unit's own log output in
5878 * ConditionPathIsMountPoint= can now properly detect bind
5879 mount points too. (Previously, a bind mount of one file
5880 system to another place in the same file system could not be
5881 detected as mount, since they shared struct stat's st_dev
5884 * We will now mount the cgroup controllers cpu, cpuacct,
5885 cpuset and the controllers net_cls, net_prio together by
5888 * nspawn containers will now have a virtualized boot
5889 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
5890 over with a randomized ID at container initialization). This
5891 has the effect of making "journalctl -b" do the right thing
5894 * The JSON output journal serialization has been updated not
5895 to generate "endless" list objects anymore, but rather one
5896 JSON object per line. This is more in line how most JSON
5897 parsers expect JSON objects. The new output mode
5898 "json-pretty" has been added to provide similar output, but
5899 neatly aligned for readability by humans.
5901 * We dropped all explicit sync() invocations in the shutdown
5902 code. The kernel does this implicitly anyway in the kernel
5903 reboot() syscall. halt(8)'s -n option is now a compatibility
5906 * We now support virtualized reboot() in containers, as
5907 supported by newer kernels. We will fall back to exit() if
5908 CAP_SYS_REBOOT is not available to the container. Also,
5909 nspawn makes use of this now and will actually reboot the
5910 container if the containerized OS asks for that.
5912 * journalctl will only show local log output by default
5913 now. Use --merge (-m) to show remote log output, too.
5915 * libsystemd-journal gained the new sd_journal_get_usage()
5916 call to determine the current disk usage of all journal
5917 files. This is exposed in the new "journalctl --disk-usage"
5920 * journald gained a new configuration setting SplitMode= in
5921 journald.conf which may be used to control how user journals
5922 are split off. See journald.conf(5) for details.
5924 * A new condition type ConditionFileNotEmpty= has been added.
5926 * tmpfiles' "w" lines now support file globbing, to write
5927 multiple files at once.
5929 * We added Python bindings for the journal submission
5930 APIs. More Python APIs for a number of selected APIs will
5931 likely follow. Note that we intend to add native bindings
5932 only for the Python language, as we consider it common
5933 enough to deserve bindings shipped within systemd. There are
5934 various projects outside of systemd that provide bindings
5935 for languages such as PHP or Lua.
5937 * Many conditions will now resolve specifiers such as %i. In
5938 addition, PathChanged= and related directives of .path units
5939 now support specifiers as well.
5941 * There's now a new RPM macro definition for the system preset
5944 * journald will now warn if it ca not forward a message to the
5945 syslog daemon because its socket is full.
5947 * timedated will no longer write or process /etc/timezone,
5948 except on Debian. As we do not support late mounted /usr
5949 anymore /etc/localtime always being a symlink is now safe,
5950 and hence the information in /etc/timezone is not necessary
5953 * logind will now always reserve one VT for a text getty (VT6
5954 by default). Previously if more than 6 X sessions where
5955 started they took up all the VTs with auto-spawned gettys,
5956 so that no text gettys were available anymore.
5958 * udev will now automatically inform the btrfs kernel logic
5959 about btrfs RAID components showing up. This should make
5960 simple hotplug based btrfs RAID assembly work.
5962 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
5963 (but not for its children which will stay at the kernel
5964 default). This should allow setups with a lot more listening
5967 * systemd will now always pass the configured timezone to the
5968 kernel at boot. timedated will do the same when the timezone
5971 * logind's inhibition logic has been updated. By default,
5972 logind will now handle the lid switch, the power and sleep
5973 keys all the time, even in graphical sessions. If DEs want
5974 to handle these events on their own they should take the new
5975 handle-power-key, handle-sleep-key and handle-lid-switch
5976 inhibitors during their runtime. A simple way to achieve
5977 that is to invoke the DE wrapped in an invocation of:
5979 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
5981 * Access to unit operations is now checked via SELinux taking
5982 the unit file label and client process label into account.
5984 * systemd will now notify the administrator in the journal
5985 when he over-mounts a non-empty directory.
5987 * There are new specifiers that are resolved in unit files,
5988 for the host name (%H), the machine ID (%m) and the boot ID
5991 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
5992 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
5993 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
5994 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
5995 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
5996 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
5997 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
6001 * Support for reading structured kernel messages from
6002 /dev/kmsg has now been added and is enabled by default.
6004 * Support for reading kernel messages from /proc/kmsg has now
6005 been removed. If you want kernel messages in the journal
6006 make sure to run a recent kernel (>= 3.5) that supports
6007 reading structured messages from /dev/kmsg (see
6008 above). /proc/kmsg is now exclusive property of classic
6009 syslog daemons again.
6011 * The libudev API gained the new
6012 udev_device_new_from_device_id() call.
6014 * The logic for file system namespace (ReadOnlyDirectory=,
6015 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
6016 require pivot_root() anymore. This means fewer temporary
6017 directories are created below /tmp for this feature.
6019 * nspawn containers will now see and receive all submounts
6020 made on the host OS below the root file system of the
6023 * Forward Secure Sealing is now supported for Journal files,
6024 which provide cryptographical sealing of journal files so
6025 that attackers cannot alter log history anymore without this
6026 being detectable. Lennart will soon post a blog story about
6027 this explaining it in more detail.
6029 * There are two new service settings RestartPreventExitStatus=
6030 and SuccessExitStatus= which allow configuration of exit
6031 status (exit code or signal) which will be excepted from the
6032 restart logic, resp. consider successful.
6034 * journalctl gained the new --verify switch that can be used
6035 to check the integrity of the structure of journal files and
6036 (if Forward Secure Sealing is enabled) the contents of
6039 * nspawn containers will now be run with /dev/stdin, /dev/fd/
6040 and similar symlinks pre-created. This makes running shells
6041 as container init process a lot more fun.
6043 * The fstab support can now handle PARTUUID= and PARTLABEL=
6046 * A new ConditionHost= condition has been added to match
6047 against the hostname (with globs) and machine ID. This is
6048 useful for clusters where a single OS image is used to
6049 provision a large number of hosts which shall run slightly
6050 different sets of services.
6052 * Services which hit the restart limit will now be placed in a
6055 Contributions from: Bertram Poettering, Dave Reisner, Huang
6056 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
6057 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
6061 * When running in --user mode systemd will now become a
6062 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
6063 tree a lot more organized.
6065 * A new PartOf= unit dependency type has been introduced that
6066 may be used to group services in a natural way.
6068 * "systemctl enable" may now be used to enable instances of
6071 * journalctl now prints error log levels in red, and
6072 warning/notice log levels in bright white. It also supports
6073 filtering by log level now.
6075 * cgtop gained a new -n switch (similar to top), to configure
6076 the maximum number of iterations to run for. It also gained
6077 -b, to run in batch mode (accepting no input).
6079 * The suffix ".service" may now be omitted on most systemctl
6080 command lines involving service unit names.
6082 * There's a new bus call in logind to lock all sessions, as
6083 well as a loginctl verb for it "lock-sessions".
6085 * libsystemd-logind.so gained a new call sd_journal_perror()
6086 that works similar to libc perror() but logs to the journal
6087 and encodes structured information about the error number.
6089 * /etc/crypttab entries now understand the new keyfile-size=
6092 * shutdown(8) now can send a (configurable) wall message when
6093 a shutdown is cancelled.
6095 * The mount propagation mode for the root file system will now
6096 default to "shared", which is useful to make containers work
6097 nicely out-of-the-box so that they receive new mounts from
6098 the host. This can be undone locally by running "mount
6099 --make-rprivate /" if needed.
6101 * The prefdm.service file has been removed. Distributions
6102 should maintain this unit downstream if they intend to keep
6103 it around. However, we recommend writing normal unit files
6104 for display managers instead.
6106 * Since systemd is a crucial part of the OS we will now
6107 default to a number of compiler switches that improve
6108 security (hardening) such as read-only relocations, stack
6109 protection, and suchlike.
6111 * The TimeoutSec= setting for services is now split into
6112 TimeoutStartSec= and TimeoutStopSec= to allow configuration
6113 of individual time outs for the start and the stop phase of
6116 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
6117 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
6118 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
6119 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
6120 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
6121 Gundersen, Zbigniew Jędrzejewski-Szmek
6125 * The journal and id128 C APIs are now fully documented as man
6128 * Extra safety checks have been added when transitioning from
6129 the initial RAM disk to the main system to avoid accidental
6132 * /etc/crypttab entries now understand the new keyfile-offset=
6135 * systemctl -t can now be used to filter by unit load state.
6137 * The journal C API gained the new sd_journal_wait() call to
6138 make writing synchronous journal clients easier.
6140 * journalctl gained the new -D switch to show journals from a
6143 * journalctl now displays a special marker between log
6144 messages of two different boots.
6146 * The journal is now explicitly flushed to /var via a service
6147 systemd-journal-flush.service, rather than implicitly simply
6148 by seeing /var/log/journal to be writable.
6150 * journalctl (and the journal C APIs) can now match for much
6151 more complex expressions, with alternatives and
6154 * When transitioning from the initial RAM disk to the main
6155 system we will now kill all processes in a killing spree to
6156 ensure no processes stay around by accident.
6158 * Three new specifiers may be used in unit files: %u, %h, %s
6159 resolve to the user name, user home directory resp. user
6160 shell. This is useful for running systemd user instances.
6162 * We now automatically rotate journal files if their data
6163 object hash table gets a fill level > 75%. We also size the
6164 hash table based on the configured maximum file size. This
6165 together should lower hash collisions drastically and thus
6166 speed things up a bit.
6168 * journalctl gained the new "--header" switch to introspect
6169 header data of journal files.
6171 * A new setting SystemCallFilters= has been added to services
6172 which may be used to apply blacklists or whitelists to
6173 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
6175 * nspawn gained a new --link-journal= switch (and quicker: -j)
6176 to link the container journal with the host. This makes it
6177 very easy to centralize log viewing on the host for all
6178 guests while still keeping the journal files separated.
6180 * Many bugfixes and optimizations
6182 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
6183 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
6184 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
6189 * Several tools now understand kernel command line arguments,
6190 which are only read when run in an initial RAM disk. They
6191 usually follow closely their normal counterparts, but are
6194 * There's a new tool to analyze the readahead files that are
6195 automatically generated at boot. Use:
6197 /usr/lib/systemd/systemd-readahead analyze /.readahead
6199 * We now provide an early debug shell on tty9 if this enabled. Use:
6201 systemctl enable debug-shell.service
6203 * All plymouth related units have been moved into the Plymouth
6204 package. Please make sure to upgrade your Plymouth version
6207 * systemd-tmpfiles now supports getting passed the basename of
6208 a configuration file only, in which case it will look for it
6209 in all appropriate directories automatically.
6211 * udevadm info now takes a /dev or /sys path as argument, and
6212 does the right thing. Example:
6214 udevadm info /dev/sda
6215 udevadm info /sys/class/block/sda
6217 * systemctl now prints a warning if a unit is stopped but a
6218 unit that might trigger it continues to run. Example: a
6219 service is stopped but the socket that activates it is left
6222 * "systemctl status" will now mention if the log output was
6223 shortened due to rotation since a service has been started.
6225 * The journal API now exposes functions to determine the
6226 "cutoff" times due to rotation.
6228 * journald now understands SIGUSR1 and SIGUSR2 for triggering
6229 immediately flushing of runtime logs to /var if possible,
6230 resp. for triggering immediate rotation of the journal
6233 * It is now considered an error if a service is attempted to
6234 be stopped that is not loaded.
6236 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
6238 * systemd-analyze now supports Python 3
6240 * tmpfiles now supports cleaning up directories via aging
6241 where the first level dirs are always kept around but
6242 directories beneath it automatically aged. This is enabled
6243 by prefixing the age field with '~'.
6245 * Seat objects now expose CanGraphical, CanTTY properties
6246 which is required to deal with very fast bootups where the
6247 display manager might be running before the graphics drivers
6248 completed initialization.
6250 * Seat objects now expose a State property.
6252 * We now include RPM macros for service enabling/disabling
6253 based on the preset logic. We recommend RPM based
6254 distributions to make use of these macros if possible. This
6255 makes it simpler to reuse RPM spec files across
6258 * We now make sure that the collected systemd unit name is
6259 always valid when services log to the journal via
6262 * There's a new man page kernel-command-line(7) detailing all
6263 command line options we understand.
6265 * The fstab generator may now be disabled at boot by passing
6266 fstab=0 on the kernel command line.
6268 * A new kernel command line option modules-load= is now understood
6269 to load a specific kernel module statically, early at boot.
6271 * Unit names specified on the systemctl command line are now
6272 automatically escaped as needed. Also, if file system or
6273 device paths are specified they are automatically turned
6274 into the appropriate mount or device unit names. Example:
6276 systemctl status /home
6277 systemctl status /dev/sda
6279 * The SysVConsole= configuration option has been removed from
6280 system.conf parsing.
6282 * The SysV search path is no longer exported on the D-Bus
6285 * The Names= option has been removed from unit file parsing.
6287 * There's a new man page bootup(7) detailing the boot process.
6289 * Every unit and every generator we ship with systemd now
6290 comes with full documentation. The self-explanatory boot is
6293 * A couple of services gained "systemd-" prefixes in their
6294 name if they wrap systemd code, rather than only external
6295 code. Among them fsck@.service which is now
6296 systemd-fsck@.service.
6298 * The HaveWatchdog property has been removed from the D-Bus
6301 * systemd.confirm_spawn= on the kernel command line should now
6304 * There's a new man page crypttab(5) which details all options
6305 we actually understand.
6307 * systemd-nspawn gained a new --capability= switch to pass
6308 additional capabilities to the container.
6310 * timedated will now read known NTP implementation unit names
6311 from /usr/lib/systemd/ntp-units.d/*.list,
6312 systemd-timedated-ntp.target has been removed.
6314 * journalctl gained a new switch "-b" that lists log data of
6315 the current boot only.
6317 * The notify socket is in the abstract namespace again, in
6318 order to support daemons which chroot() at start-up.
6320 * There is a new Storage= configuration option for journald
6321 which allows configuration of where log data should go. This
6322 also provides a way to disable journal logging entirely, so
6323 that data collected is only forwarded to the console, the
6324 kernel log buffer or another syslog implementation.
6326 * Many bugfixes and optimizations
6328 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
6329 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
6330 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
6331 Shawn Landden, Tom Gundersen
6335 * "systemctl help <unit>" now shows the man page if one is
6338 * Several new man pages have been added.
6340 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
6341 MaxLevelConsole= can now be specified in
6342 journald.conf. These options allow reducing the amount of
6343 data stored on disk or forwarded by the log level.
6345 * TimerSlackNSec= can now be specified in system.conf for
6346 PID1. This allows system-wide power savings.
6348 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
6349 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
6354 * logind is now capable of (optionally) handling power and
6355 sleep keys as well as the lid switch.
6357 * journalctl now understands the syntax "journalctl
6358 /usr/bin/avahi-daemon" to get all log output of a specific
6361 * CapabilityBoundingSet= in system.conf now also influences
6362 the capability bound set of usermode helpers of the kernel.
6364 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
6365 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
6366 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
6367 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
6371 * Note that we skipped 139 releases here in order to set the
6372 new version to something that is greater than both udev's
6373 and systemd's most recent version number.
6375 * udev: all udev sources are merged into the systemd source tree now.
6376 All future udev development will happen in the systemd tree. It
6377 is still fully supported to use the udev daemon and tools without
6378 systemd running, like in initramfs or other init systems. Building
6379 udev though, will require the *build* of the systemd tree, but
6380 udev can be properly *run* without systemd.
6382 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
6383 should be used to create dead device nodes as workarounds for broken
6386 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
6387 no longer supported. udev_monitor_new_from_netlink() needs to be
6388 used to subscribe to events.
6390 * udev: when udevd is started by systemd, processes which are left
6391 behind by forking them off of udev rules, are unconditionally cleaned
6392 up and killed now after the event handling has finished. Services or
6393 daemons must be started as systemd services. Services can be
6394 pulled-in by udev to get started, but they can no longer be directly
6395 forked by udev rules.
6397 * udev: the daemon binary is called systemd-udevd now and installed
6398 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
6399 to adapt to that, create symlink, or rename the binary after building
6402 * libudev no longer provides these symbols:
6403 udev_monitor_from_socket()
6404 udev_queue_get_failed_list_entry()
6405 udev_get_{dev,sys,run}_path()
6406 The versions number was bumped and symbol versioning introduced.
6408 * systemd-loginctl and systemd-journalctl have been renamed
6409 to loginctl and journalctl to match systemctl.
6411 * The config files: /etc/systemd/systemd-logind.conf and
6412 /etc/systemd/systemd-journald.conf have been renamed to
6413 logind.conf and journald.conf. Package updates should rename
6414 the files to the new names on upgrade.
6416 * For almost all files the license is now LGPL2.1+, changed
6417 from the previous GPL2.0+. Exceptions are some minor stuff
6418 of udev (which will be changed to LGPL2.1 eventually, too),
6419 and the MIT licensed sd-daemon.[ch] library that is suitable
6420 to be used as drop-in files.
6422 * systemd and logind now handle system sleep states, in
6423 particular suspending and hibernating.
6425 * logind now implements a sleep/shutdown/idle inhibiting logic
6426 suitable for a variety of uses. Soonishly Lennart will blog
6427 about this in more detail.
6429 * var-run.mount and var-lock.mount are no longer provided
6430 (which previously bind mounted these directories to their new
6431 places). Distributions which have not converted these
6432 directories to symlinks should consider stealing these files
6433 from git history and add them downstream.
6435 * We introduced the Documentation= field for units and added
6436 this to all our shipped units. This is useful to make it
6437 easier to explore the boot and the purpose of the various
6440 * All smaller setup units (such as
6441 systemd-vconsole-setup.service) now detect properly if they
6442 are run in a container and are skipped when
6443 appropriate. This guarantees an entirely noise-free boot in
6444 Linux container environments such as systemd-nspawn.
6446 * A framework for implementing offline system updates is now
6447 integrated, for details see:
6448 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
6450 * A new service type Type=idle is available now which helps us
6451 avoiding ugly interleaving of getty output and boot status
6454 * There's now a system-wide CapabilityBoundingSet= option to
6455 globally reduce the set of capabilities for the
6456 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
6457 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
6458 even CAP_NET_ADMIN system-wide for secure systems.
6460 * There are now system-wide DefaultLimitXXX= options to
6461 globally change the defaults of the various resource limits
6462 for all units started by PID 1.
6464 * Harald Hoyer's systemd test suite has been integrated into
6465 systemd which allows easy testing of systemd builds in qemu
6466 and nspawn. (This is really awesome! Ask us for details!)
6468 * The fstab parser is now implemented as generator, not inside
6471 * systemctl will now warn you if .mount units generated from
6472 /etc/fstab are out of date due to changes in fstab that
6473 have not been read by systemd yet.
6475 * systemd is now suitable for usage in initrds. Dracut has
6476 already been updated to make use of this. With this in place
6477 initrds get a slight bit faster but primarily are much
6478 easier to introspect and debug since "systemctl status" in
6479 the host system can be used to introspect initrd services,
6480 and the journal from the initrd is kept around too.
6482 * systemd-delta has been added, a tool to explore differences
6483 between user/admin configuration and vendor defaults.
6485 * PrivateTmp= now affects both /tmp and /var/tmp.
6487 * Boot time status messages are now much prettier and feature
6488 proper english language. Booting up systemd has never been
6491 * Read-ahead pack files now include the inode number of all
6492 files to pre-cache. When the inode changes the pre-caching
6493 is not attempted. This should be nicer to deal with updated
6494 packages which might result in changes of read-ahead
6497 * We now temporaritly lower the kernel's read_ahead_kb variable
6498 when collecting read-ahead data to ensure the kernel's
6499 built-in read-ahead does not add noise to our measurements
6500 of necessary blocks to pre-cache.
6502 * There's now RequiresMountsFor= to add automatic dependencies
6503 for all mounts necessary for a specific file system path.
6505 * MountAuto= and SwapAuto= have been removed from
6506 system.conf. Mounting file systems at boot has to take place
6509 * nspawn now learned a new switch --uuid= to set the machine
6510 ID on the command line.
6512 * nspawn now learned the -b switch to automatically search
6515 * vt102 is now the default TERM for serial TTYs, upgraded from
6518 * systemd-logind now works on VT-less systems.
6520 * The build tree has been reorganized. The individual
6521 components now have directories of their own.
6523 * A new condition type ConditionPathIsReadWrite= is now available.
6525 * nspawn learned the new -C switch to create cgroups for the
6526 container in other hierarchies.
6528 * We now have support for hardware watchdogs, configurable in
6531 * The scheduled shutdown logic now has a public API.
6533 * We now mount /tmp as tmpfs by default, but this can be
6534 masked and /etc/fstab can override it.
6536 * Since udisks does not make use of /media anymore we are not
6537 mounting a tmpfs on it anymore.
6539 * journalctl gained a new --local switch to only interleave
6540 locally generated journal files.
6542 * We can now load the IMA policy at boot automatically.
6544 * The GTK tools have been split off into a systemd-ui.
6546 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
6547 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
6548 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
6549 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
6550 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
6551 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
6552 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
6553 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
6554 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
6559 * This is mostly a bugfix release
6561 * Support optional initialization of the machine ID from the
6562 KVM or container configured UUID.
6564 * Support immediate reboots with "systemctl reboot -ff"
6566 * Show /etc/os-release data in systemd-analyze output
6568 * Many bugfixes for the journal, including endianness fixes and
6569 ensuring that disk space enforcement works
6571 * sd-login.h is C++ compatible again
6573 * Extend the /etc/os-release format on request of the Debian
6576 * We now refuse non-UTF8 strings used in various configuration
6577 and unit files. This is done to ensure we do not pass invalid
6578 data over D-Bus or expose it elsewhere.
6580 * Register Mimo USB Screens as suitable for automatic seat
6583 * Read SELinux client context from journal clients in a race
6586 * Reorder configuration file lookup order. /etc now always
6587 overrides /run in order to allow the administrator to always
6588 and unconditionally override vendor-supplied or
6589 automatically generated data.
6591 * The various user visible bits of the journal now have man
6592 pages. We still lack man pages for the journal API calls
6595 * We now ship all man pages in HTML format again in the
6598 Contributions from: Dave Reisner, Dirk Eibach, Frederic
6599 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
6600 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
6605 * This is mostly a bugfix release
6607 * systems lacking /etc/os-release are no longer supported.
6609 * Various functionality updates to libsystemd-login.so
6611 * Track class of PAM logins to distinguish greeters from
6614 Contributions from: Kay Sievers, Lennart Poettering, Michael
6619 * This is an important bugfix release for v41.
6621 * Building man pages is now optional which should be useful
6622 for those building systemd from git but unwilling to install
6625 * Watchdog support for supervising services is now usable. In
6626 a future release support for hardware watchdogs
6627 (i.e. /dev/watchdog) will be added building on this.
6629 * Service start rate limiting is now configurable and can be
6630 turned off per service. When a start rate limit is hit a
6631 reboot can automatically be triggered.
6633 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
6635 Contributions from: Benjamin Franzke, Bill Nottingham,
6636 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
6637 Schmidt, Michał Górny, Piotr Drąg
6641 * The systemd binary is installed /usr/lib/systemd/systemd now;
6642 An existing /sbin/init symlink needs to be adapted with the
6645 * The code that loads kernel modules has been ported to invoke
6646 libkmod directly, instead of modprobe. This means we do not
6647 support systems with module-init-tools anymore.
6649 * Watchdog support is now already useful, but still not
6652 * A new kernel command line option systemd.setenv= is
6653 understood to set system wide environment variables
6654 dynamically at boot.
6656 * We now limit the set of capabilities of systemd-journald.
6658 * We now set SIGPIPE to ignore by default, since it only is
6659 useful in shell pipelines, and has little use in general
6660 code. This can be disabled with IgnoreSIPIPE=no in unit
6663 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
6664 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
6669 * This is mostly a bugfix release
6671 * We now expose the reason why a service failed in the
6672 "Result" D-Bus property.
6674 * Rudimentary service watchdog support (will be completed over
6675 the next few releases.)
6677 * When systemd forks off in order execute some service we will
6678 now immediately changes its argv[0] to reflect which process
6679 it will execute. This is useful to minimize the time window
6680 with a generic argv[0], which makes bootcharts more useful
6682 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
6683 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
6684 Mike Kazantsev, Ray Strode
6688 * This is mostly a test release, but incorporates many
6691 * New systemd-cgtop tool to show control groups by their
6694 * Linking against libacl for ACLs is optional again. If
6695 disabled, support tracking device access for active logins
6696 goes becomes unavailable, and so does access to the user
6697 journals by the respective users.
6699 * If a group "adm" exists, journal files are automatically
6700 owned by them, thus allow members of this group full access
6701 to the system journal as well as all user journals.
6703 * The journal now stores the SELinux context of the logging
6704 client for all entries.
6706 * Add C++ inclusion guards to all public headers
6708 * New output mode "cat" in the journal to print only text
6709 messages, without any meta data like date or time.
6711 * Include tiny X server wrapper as a temporary stop-gap to
6712 teach XOrg udev display enumeration. This is used by display
6713 managers such as gdm, and will go away as soon as XOrg
6714 learned native udev hotplugging for display devices.
6716 * Add new systemd-cat tool for executing arbitrary programs
6717 with STDERR/STDOUT connected to the journal. Can also act as
6718 BSD logger replacement, and does so by default.
6720 * Optionally store all locally generated coredumps in the
6721 journal along with meta data.
6723 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
6724 writing short strings to files (for usage for /sys), and for
6725 creating symlinks, character and block device nodes.
6727 * New unit file option ControlGroupPersistent= to make cgroups
6728 persistent, following the mechanisms outlined in
6729 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
6731 * Support multiple local RTCs in a sane way
6733 * No longer monopolize IO when replaying readahead data on
6734 rotating disks, since we might starve non-file-system IO to
6735 death, since fanotify() will not see accesses done by blkid,
6738 * Do not show kernel threads in systemd-cgls anymore, unless
6739 requested with new -k switch.
6741 Contributions from: Dan Horák, Kay Sievers, Lennart
6742 Poettering, Michal Schmidt
6746 * This is mostly a test release, but incorporates many
6749 * The git repository moved to:
6750 git://anongit.freedesktop.org/systemd/systemd
6751 ssh://git.freedesktop.org/git/systemd/systemd
6753 * First release with the journal
6754 http://0pointer.de/blog/projects/the-journal.html
6756 * The journal replaces both systemd-kmsg-syslogd and
6757 systemd-stdout-bridge.
6759 * New sd_pid_get_unit() API call in libsystemd-logind
6761 * Many systemadm clean-ups
6763 * Introduce remote-fs-pre.target which is ordered before all
6764 remote mounts and may be used to start services before all
6767 * Added Mageia support
6769 * Add bash completion for systemd-loginctl
6771 * Actively monitor PID file creation for daemons which exit in
6772 the parent process before having finished writing the PID
6773 file in the daemon process. Daemons which do this need to be
6774 fixed (i.e. PID file creation must have finished before the
6775 parent exits), but we now react a bit more gracefully to them.
6777 * Add colourful boot output, mimicking the well-known output
6778 of existing distributions.
6780 * New option PassCredentials= for socket units, for
6781 compatibility with a recent kernel ABI breakage.
6783 * /etc/rc.local is now hooked in via a generator binary, and
6784 thus will no longer act as synchronization point during
6787 * systemctl list-unit-files now supports --root=.
6789 * systemd-tmpfiles now understands two new commands: z, Z for
6790 relabelling files according to the SELinux database. This is
6791 useful to apply SELinux labels to specific files in /sys,
6794 * Output of SysV services is now forwarded to both the console
6795 and the journal by default, not only just the console.
6797 * New man pages for all APIs from libsystemd-login.
6799 * The build tree got reorganized and the build system is a
6800 lot more modular allowing embedded setups to specifically
6801 select the components of systemd they are interested in.
6803 * Support for Linux systems lacking the kernel VT subsystem is
6806 * configure's --with-rootdir= got renamed to
6807 --with-rootprefix= to follow the naming used by udev and
6810 * Unless specified otherwise we will now install to /usr instead
6811 of /usr/local by default.
6813 * Processes with '@' in argv[0][0] are now excluded from the
6814 final shut-down killing spree, following the logic explained
6816 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
6818 * All processes remaining in a service cgroup when we enter
6819 the START or START_PRE states are now killed with
6820 SIGKILL. That means it is no longer possible to spawn
6821 background processes from ExecStart= lines (which was never
6822 supported anyway, and bad style).
6824 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
6825 reloading of units together.
6827 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
6828 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
6829 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
6830 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
6831 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek